CHOLERA IN A TIME OF EL NIÑO AND VULNERABILITY IN PIURA, PERU:
A CLIMATE AFFAIRS APPROACH
By
Iván J. Ramírez

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
DOCTOR OF PHILOSOPHY

Geography

2012

ABSTRACT
CHOLERA IN A TIME OF EL NIÑO AND VULNERABILITY IN PIURA, PERU:
A CLIMATE AFFAIRS APPROACH
By
Iván J. Ramírez

The goal of my dissertation research is to reconstruct the temporal and spatial
associations among El Niño-Southern Oscillation (ENSO), social vulnerability and cholera
incidence in Piura, Peru from 1991 to 2001 in order to better understand El Niño’s
impact on the cholera epidemic in Peru during the 1990s. Piura is important to study
because it was one of the first places to report cholera in Peru. It also had one of the
highest incidence rates in the country, and historically, the region is known for El Niño.
My overarching research questions are: (1) What was the impact of ENSO on cholera
incidence in Piura; and (2) How did social vulnerability influence this relationship? My
research hypotheses are: (a) There was a temporal association between ENSO, climate
and cholera cases in Piura in the 1990s. Furthermore, these associations were stronger
after 1992 compared to the onset of the epidemic in 1991; and (b) The spatial variability
of the ENSO-climate-cholera associations in Piura in 1997-98 will be explained by the
spatial distribution of social vulnerability. Moreover, the level of social vulnerability
within districts in Piura will either antagonize or buffer the effects of ENSO and climate
on cholera incidence.

I address my research questions and hypotheses using a climate affairs approach
that is informed by disease ecology and vulnerability theories from the geographic
subfields of medical and human-environment geography. Climate affairs is an
integrating concept in the earth and social sciences used to understand the
interrelationships among climate, environment and society worldwide. Using climate
affairs, I developed a conceptual framework that: 1) examines cholera transmission
within a broader conception of ENSO; 2) links ENSO-cholera associations to social
vulnerability; and 3) considers ENSO-cholera interactions at multiple scales. The key
findings of this research suggest that cholera’s temporal association with ENSO was
transient throughout the 1990s; the strongest association was found during the 1997-98
El Niño. I also found that cholera transmission occurred through the interactions of
global and local sea surface temperatures with rainfall. Furthermore, I demonstrated
that the spatial distribution of social vulnerability can in part explain the associations
between global and local climate and cholera during the 1997-98 El Niño. However,
these associations varied by time lag, district and variable. It also appears that districts
on the west coast of the subregion of Piura were the most vulnerable. Lastly, important
to the understanding of these findings is that interpretation of ENSO and its association
with cholera will highly depend on the Niño definition and region chosen for analysis.
This research contributes to future climate-informed initiatives that enhance societal
capacities, while focusing on population health and the monitoring of populations
during future climate events in Piura, Peru and the Latin American region.

Copyright by
IVÁN J. RAMÍREZ
2012

This thesis is dedicated to my parents Jorge and Nilda Ramírez

v

ACKNOWLEDGEMENTS
I would like to greatly thank my dissertation advisor Dr. Sue C. Grady for
guidance through the years. I really would not have made it without her full-hearted
support. I would also like to thank my committee: Dr. Antoinette WinklerPrins
([Geography] who I credit for bringing me to MSU), Dr. Julie Winkler (Geography), and
Dr. Steve Esquith (Philosophy). I would like to express my deep appreciation to Dr.
Michael H. Glantz (CCB), my external committee member, for the strong support and
Niño wisdom he has bestowed on me since 2006. I am indebted to my collaborators in
Peru: Ing. Norma Ordinola, University of Piura, Ing. Grover Otero (Proyecto Chira-Piura),
Dr. Elsa Galarza and Joanna Noelia Kamiche Zegarra (University of the Pacific) and Dr.
Luis Miguel Castrov V. I would also like to thank: INEI (Piura), the Ministry of Health
(Piura and Lima), General Office of Epidemiology (Piura and Lima), Dr. Ana Gil
(International Institute of Nutrition), Dr. Marcos Cueto (Institute of Peruvian Studies),
Dr. Hildegardo Córdova Aguilar (Applied Geography Research Center at the Pontifical
Catholic University), and Pablo Lagos (IGP). I would also like to thank Dr. Gil Compo
(wavelet script), Mark Finn (ArcGIS genius), and Moises Smart (Matlab wizard) for their
technical support; and a special thank you to Dr. Diane M. Doberneck (UOE) . This study
was supported through grants by the following institutions at Michigan State University:
Department of Geography, Department of Philosophy (Ethics and Development
Specialization), Center for Latin American and Caribbean Studies, and the Graduate
School. Support was also provided by CCB in Boulder, CO. In closing, I am forever
grateful to my wife Jieun Lee for supporting my dreams and aspirations.
vi

TABLE OF CONTENTS

LIST OF TABLES ........................................................................................................................... x
LIST OF FIGURES ........................................................................................................................ xi
CHAPTER 1: INTRODUCTION ....................................................................................................... 1
1.1
1.2

1.3
1.4
1.5

The Cholera Epidemic in Peru.......................................................................................... 6
Explaining the Cholera Epidemic in Peru ......................................................................... 8
1.2.1 Passing Ship Hypothesis ..................................................................................... 8
1.2.2 El Niño-cholera Hypothesis ................................................................................ 8
Challenging the El Niño-cholera Hypothesis .................................................................. 10
Study Significance ......................................................................................................... 13
Dissertation Format ...................................................................................................... 13

CHAPTER 2: LITERATURE REVIEW ............................................................................................. 15
2.1

2.2

2.3
2.4

ENSO Science and Knowledge ....................................................................................... 15
2.1.1 ENSO Events .................................................................................................... 16
2.1.2 ENSO Monitoring and Definition ...................................................................... 17
2.1.3 ENSO Teleconnections in Peru ......................................................................... 21
2.1.4 ENSO Ecosystem and Societal Impacts in Peru ................................................. 23
Climate and Cholera Ecology ......................................................................................... 25
2.2.1 Global Climate Parameters and Cholera ........................................................... 27
2.2.2 Local Sea Surface Temperature ........................................................................ 30
2.2.3 Local Rainfall .................................................................................................... 33
2.2.4 Temporal Lags .................................................................................................. 34
Cholera Transmission and Human Ecology .................................................................... 35
Cholera and Population Vulnerability ............................................................................ 38
2.4.1 Cholera and Determinants of Health ................................................................ 39
2.4.2 Vulnerable Subpopulations .............................................................................. 42

CHAPTER 3: REEXAMINING EL NIÑO AND CHOLERA IN PERU .................................................... 45
3.1
3.2
3.3
3.4
3.5
3.6
3.7
3.8

Recounting El Niño Impacts on Cholera in Peru ............................................................. 45
The Importance of Definition ........................................................................................ 48
The La Niña Factor ........................................................................................................ 53
Geography of El Niño Impacts ....................................................................................... 56
Rainfall Extremes .......................................................................................................... 60
Social Dimensions ......................................................................................................... 61
Summary....................................................................................................................... 62
Research Questions, Hypotheses and Objectives .......................................................... 63
vii

CHAPTER 4: RESEARCH DESIGN................................................................................................. 65
4.1

4.2
4.3

Approach, Concepts and Framework............................................................................. 65
4.1.1 Climate Affairs Approach ................................................................................. 66
4.1.2 Theoretical Concepts ....................................................................................... 67
4.1.3 Conceptual Framework .................................................................................... 70
Study Area and Population ............................................................................................ 76
Data and Methods ........................................................................................................ 83
4.3.1 Objective 1: Develop a conceptual framework that characterizes the
potential ecological pathways and conditions of cholera vulnerability and
transmission in Piura ..................................................................................................... 83
4.3.2 Objective 2: Characterize the temporal associations between ENSO,
climate and cholera cases in Piura from 1991 to 2001 ................................................... 85
4.3.3 Objective 3: Construct a social vulnerability index (SVI) that characterizes
social vulnerability to cholera in the subregion of Piura in 1997-98 ............................... 94
4.3.4 Objective 4: Characterize the spatial and temporal associations between
ENSO, climate and cholera incidence by district in Piura for 1997-98 and estimate
the degree to which social vulnerability influenced this relationship ........................... 100
4.3.5 Objective 5: Reflect on the findings from this study using a climate and
development ethics perspective in order to better formulate recommendations ....... 103

CHAPTER 5: RESULTS .............................................................................................................. 106
5.1
5.2
5.3
5.4

5.5
5.6
5.7

ENSO Events by Niño Index (Objective 2) .................................................................... 106
Cholera Cases by Niño Index and ENSO Event (Objective 2) ........................................ 111
Wavelet Transform Analyses (Objective 2) .................................................................. 113
Wavelet Coherence Analyses (Objective 2) ................................................................. 134
5.4.1 Climate-Cholera Associations ......................................................................... 134
5.4.2 Cross-Wavelet Transform: Climate-Cholera ................................................... 144
5.4.3 Cross-Correlation Analysis: Climate-Cholera .................................................. 152
The Social Vulnerability Index (Objective 3) ................................................................. 154
5.5.1 Mapping the SVI by District ............................................................................ 160
Measuring Associations between the SVI and Cholera Incidence (Objective 4)............ 166
5.6.1 Mapping Associations between SVI and Cholera Incidence ............................ 167
Measuring Climate Associations with Cholera by District (Objective 4) ....................... 176
5.7.1 Mapping Climate Associations with Cholera by District .................................. 190

CHAPTER 6: DISCUSSION......................................................................................................... 204
6.1
6.2

The ENSO Context ....................................................................................................... 204
Temporal Associations between Cholera and ENSO in Piura, Peru: 1991-2001 ............ 205
6.2.1 Cholera and Global Climate ............................................................................ 206
6.2.2 Cholera and Local Climate .............................................................................. 209
6.2.3 Was there a temporal association between cholera incidence in Piura
and ENSO in the 1990s? Was this association stronger after 1992? ............................. 210
viii

6.3

6.4
6.5

ENSO and the Social Vulnerability of Cholera Incidence in Piura, Peru: 1997-98 .......... 211
6.3.1 Cholera and Social Vulnerability by District: 1997-98 ..................................... 213
6.3.2 Cholera and Global and Local Climate by District: 1998 .................................. 215
6.3.3 How did social vulnerability influence the climate-cholera relationships
in Piura? Does the spatial distribution of social vulnerability within districts in
Piura explain the spatial variability of the ENSO-cholera associations in Piura in
1997-98? ..................................................................................................................... 217
Study Limitations ........................................................................................................ 222
Ethical Geographies of the Cholera Epidemic in Piura ................................................. 223
6.5.1 Ethics of Climate and Development ............................................................... 224
6.5.2 Ethical Geographies ....................................................................................... 226
6.5.3 Conclusions .................................................................................................... 234

CHAPTER 7: CONCLUSIONS ..................................................................................................... 235
APPENDICES ........................................................................................................................... 244
APPENDIX 1 ............................................................................................................................ 245
APPENDIX 2 ............................................................................................................................ 247
APPENDIX 3 ............................................................................................................................ 249
APPENDIX 4 ............................................................................................................................ 257
APPENDIX 5 ............................................................................................................................ 259
REFERENCES ........................................................................................................................... 269

ix

LIST OF TABLES

TABLE 4.1 SUMMARY OF STATISTICS FOR CHOLERA AND GLOBAL AND LOCAL CLIMATE PARAMETERS FOR
THE YEARS 1971 TO 2001 ...................................................................................................88
TABLE 4.2 DESCRIPTION OF VARIABLES USED IN PRINCIPAL COMPONENTS ANALYSIS...........................96
TABLE 4.3 SUMMARY OF STATISTICS FOR VARIABLES AND PEARSON’S CORRELATIONS WITH CHOLERA ....97
TABLE 4.4 CORRELATION MATRIX FOR VARIABLES USED IN PRINCIPAL COMPONENTS ANALYSIS. ............99
TABLE 5.1 COMPARISON OF EL NIÑO (RED) AND LA NIÑA (BLUE) EVENTS ACCORDING TO THE NIÑO 3.4
INDEX FROM 1990 TO 2001...............................................................................................108
TABLE 5.2 COMPARISON OF EL NIÑO (RED) AND LA NIÑA (BLUE) EVENTS ACCORDING TO THE NIÑO 1+2
INDEX FROM 1990 TO 2001...............................................................................................108
TABLE 5.3 NIÑO 3.4 INDEX: TIMING OF EL NIÑO AND LA NIÑA EVENTS FROM 1990 TO 2001. ..........109
TABLE 5.4 NIÑO 1+2 INDEX: TIMING OF EL NIÑO AND LA NIÑA EVENTS FROM 1990 TO 2001. .........110
TABLE 5.5 COMPARISON OF ENSO MONTHS AND CHOLERA CASES USING THE NIÑO 3.4 INDEX FROM 1990
TO 2001. ....................................................................................................................... 112
TABLE 5.6 COMPARISON OF ENSO MONTHS AND CHOLERA CASES USING THE NIÑO 1+2 INDEX FROM
1990 TO 2001. ..............................................................................................................112
TABLE 5.7 CROSS-CORRELATIONS BETWEEN CHOLERA CASE ANOMALY AND GLOBAL AND LOCAL CLIMATE
TIME SERIES. ...................................................................................................................153
TABLE 5.8 SUMMARY OF WAVELET COHERENCE ANALYSES BY LAG, DIRECTION, AND PHASE. .............153
TABLE 5.9 SVI - TOTAL VARIANCE EXPLAINED IN THE PRINCIPAL COMPONENTS ANALYSIS (PCA) ........157
TABLE 5.10 ROTATED COMPONENT MATRIX IN THE PRINCIPAL COMPONENTS ANALYSIS (PCA). .........158
TABLE 5.11 SUMMARY STATISTICS FOR ORDINARY LEAST SQUARES REGRESSION (OLS) ....................166
TABLE 5.12 COMPARISON STATISTICS FOR GEOGRAPHICALLY WEIGHTED REGRESSION (GWR) AND
ORDINARY LEAST SQUARES REGRESSION (OLS) .......................................................................167
TABLE 5.13 SELECTED DISTRICTS FROM GEOGRAPHICALLY WEIGHTED REGRESSION (GWR) BASED ON
PREDICTED VALUES AND STANDARDIZED RESIDUALS..................................................................175
TABLE 6.1 COMPARISON OF SOCIAL VULNERABILITY AND CLIMATE ASSOCIATIONS (R2) WITH CHOLERA. 220
x

LIST OF FIGURES

FIGURE 1.1 MAP OF THE DEPARTMENT OF PIURA THAT HIGHLIGHTS THE SUBREGION OF PIURA. FOR
INTERPRETATION OF THE REFERENCES TO COLOR IN THIS AND ALL OTHER FIGURES, THE READER IS REFERRED
TO THE ELECTRONIC VERSION OF THIS DISSERTATION. .....................................................................4
FIGURE 2.1 MAP OF NIÑO REGIONS. ......................................................................................19
FIGURE 3.1 MONTHLY CHOLERA CASES IN PERU AND NIÑO 1+2 AND NIÑO 4 SEA SURFACE TEMPERATURE
ANOMALY (SSTA) FROM 1990 TO 1995. ...............................................................................51
FIGURE 3.2 CHOLERA CASES IN PERU AND EL NIÑOS AND LA NIÑAS FROM 1995 TO 2000 ..................55
FIGURE 3.3 CHOLERA CASES FROM 1993 TO 1998 IN LIMA (COAST)..............................................57
FIGURE 3.4 CHOLERA CASES FROM 1993 TO 1998 IN CAJAMARCA (MOUNTAIN) ..............................58
FIGURE 3.5 CHOLERA CASES FROM 1993 TO 1998 IN LORETO (JUNGLE).........................................59
FIGURE 4.1 CLIMATE AFFAIRS AS AN INTEGRATING APPROACH AND CONCEPT. ..................................67
FIGURE 4.2 CONCEPTUAL FRAMEWORK ADAPTED FROM TURNER ET AL. (2003). ...............................73
FIGURE 4.3 ENSO IMPACTS ON THE CHOLERA TRANSMISSION CYCLE .............................................75
FIGURE 4.4 MONTHLY AVERAGE TEMPERATURE (°C) AT MIRAFLORES STATION, PIURA FOR 1971 TO
2000. .............................................................................................................................78
FIGURE 4.5 MONTHLY AVERAGE RAINFALL (MM) AT MIRAFLORES STATION, PIURA FOR 1971 TO 2000. 79
FIGURE 4.6 MAP OF THE SUBREGION OF PIURA BY DISTRICT (ID)...................................................81
FIGURE 4.7 HISTOGRAMS OF (A) RAINFALL (SQUARE-ROOT TRANSFORMED) (MM) BY MONTH; AND (B)
RAINFALL (SQUARE-ROOT TRANSFORMED) ANOMALY (MM) BY MONTH. ..........................................89
FIGURE 4.8 HISTOGRAMS OF (A) CHOLERA CASES (SQUARE-ROOT TRANSFORMED) BY MONTH AND (B)
CHOLERA CASES (SQUARE-ROOT TRANSFORMED) ANOMALY BY MONTH. ..........................................90
FIGURE 5.1 NIÑO 3.4 INDEX FROM 1971 TO 2001 AT ±0.5 (°C) THRESHOLD (GRAY DOTTED LINES)....106
FIGURE 5.2 NIÑO 1+2 INDEX FROM 1971 TO 2001 AT ±0.5 (°C) THRESHOLD (GRAY DOTTED LINES)...107
FIGURE 5.3 CHOLERA CASES (SQUARE-ROOT TRANSFORMED) ANOMALY FROM 1991 TO 2001. .........114
FIGURE 5.4 NIÑO 3.4. SEA SURFACE TEMPERATURE ANOMALY (SSTA) (°C) FROM 1971 TO 2001. ..115

xi

FIGURE 5.5 NIÑO 1+2 SEA SURFACE TEMPERATURE ANOMALY (SSTA) (°C) FROM 197 TO-2001......116
FIGURE 5.6 PAITA SEA SURFACE TEMPERATURE ANOMALY (SSTA) (°C) FROM 1971 TO 2001. .........117
FIGURE 5.7 TEMPERATURE MAXIMUM ANOMALY (TMAXA) (°C) IN PIURA FROM 1971 TO 2001. ....118
FIGURE 5.8 TEMPERATURE MEAN ANOMALY (TMEANA) (°C) IN PIURA FROM 1971 TO 2001. ........119
FIGURE 5.9 TEMPERATURE MINIMUM ANOMALY (TMINA) (°C) IN PIURA FROM 1971 TO 2001 ......120
FIGURE 5.10 RAINFALL (SQUARE-ROOT TRANSFORMED) ANOMALY (MM) IN PIURA FROM 1971 TO 2001
....................................................................................................................................121
FIGURE 5.11 WAVELET TRANSFORM ANALYSIS OF CHOLERA CASES ANOMALY IN PIURA BY MONTH FROM
1991 TO 2001. ..............................................................................................................123
FIGURE 5.12 WAVELET TRANSFORM ANALYSIS OF NIÑO 3.4 SEA SURFACE TEMPERATURE ANOMALY
(SSTA) (°C) BY MONTH FROM 1971 TO 2001 ........................................................................125
FIGURE 5.13 WAVELET TRANSFORM ANALYSIS OF NIÑO 1+2 SEA SURFACE TEMPERATURE ANOMALY
(SSTA) (°C) BY MONTH FROM 1971 TO 2001 ........................................................................126
FIGURE 5.14 WAVELET TRANSFORM ANALYSES OF PAITA SEA SURFACE TEMPERATURE ANOMALY (SSTA)
(°C) BY MONTH FROM 1971 TO 2001. .................................................................................127
FIGURE 5.15 WAVELET TRANSFORM ANALYSES OF TEMPERATURE MAXIMUM ANOMALY (TMAXA) (°C)
BY MONTH IN PIURA FROM 1971 TO 2001. ........................................................................... 129
FIGURE 5.16 WAVELET TRANSFORM ANALYSES OF TEMPERATURE MEAN ANOMALY (TMEANA) (°C) BY
MONTH IN PIURA FROM 1971 TO 2001. ............................................................................... 130
FIGURE 5.17 WAVELET TRANSFORM ANALYSES OF TEMPERATURE MINIMUM ANOMALY (TMINA) (°C) BY
MONTH IN PIURA FROM 1971 TO 2001. ............................................................................... 131
FIGURE 5.18 WAVELET TRANSFORM ANALYSES OF RAINFALL ANOMALY (MM) BY MONTH IN PIURA FROM
1971 TO 2001. ..............................................................................................................133
FIGURE 5.19 WAVELET COHERENCE OF NIÑO 3.4 SEA SURFACE TEMPERATURE ANOMALY (SSTA) AND
CHOLERA CASES IN PIURA FROM 1991 TO 2001. ....................................................................135
FIGURE 5.20 WAVELET COHERENCE OF NIÑO 1+2 SEA SURFACE TEMPERATURE ANOMALY (SSTA) AND
CHOLERA CASES IN PIURA FROM 1991 TO 2001. ....................................................................136
FIGURE 5.21 WAVELET COHERENCE OF PAITA SEA SURFACE TEMPERATURE ANOMALY (SSTA) AND
CHOLERA CASES IN PIURA FROM 1991 TO 2001. ....................................................................137
FIGURE 5.22 WAVELET COHERENCE OF TEMPERATURE MAXIMUM ANOMALY (TMAXA) AND CHOLERA
CASES IN PIURA FROM 1991 TO 2001. .................................................................................139

xii

FIGURE 5.23 WAVELET COHERENCE OF TEMPERATURE MEAN ANOMALY (TMEANA) AND CHOLERA
CASES IN PIURA FROM 1991 TO 2001. .................................................................................140
FIGURE 5.24 WAVELET COHERENCE OF TEMPERATURE MINIMUM ANOMALY (TMINA) AND CHOLERA
CASES IN PIURA FROM 1991 TO 2001. .................................................................................141
FIGURE 5.25 WAVELET COHERENCE OF RAINFALL ANOMALY AND CHOLERA CASES IN PIURA FROM 1991
TO 2001 ........................................................................................................................ 143
FIGURE 5.26 CROSS-WAVELET OF NIÑO 3.4 SEA SURFACE TEMPERATURE ANOMALY (SSTA) AND
CHOLERA CASES IN PIURA FROM 1991 TO 2001. ....................................................................145
FIGURE 5.27 CROSS-WAVELET OF NIÑO 1+2 SEA SURFACE TEMPERATURE ANOMALY (SSTA) AND
CHOLERA CASES IN PIURA FROM 1991 TO 2001. ....................................................................146
FIGURE 5.28 CROSS-WAVELET OF PAITA SEA SURFACE TEMPERATURE ANOMALY (SSTA) AND CHOLERA
CASES IN PIURA FROM 1991 TO 2001. .................................................................................147
FIGURE 5.29 CROSS-WAVELET OF TEMPERATURE MAXIMUM ANOMALY (TMAXA) AND CHOLERA CASES
IN PIURA FROM 1991 TO 2001 ........................................................................................... 148
FIGURE 5.30 CROSS-WAVELET OF TEMPERATURE MEAN ANOMALY (TMEANA) AND CHOLERA CASES IN
PIURA FROM 1991 TO 2001. .............................................................................................149
FIGURE 5.31 CROSS-WAVELET OF TEMPERATURE MINIMUM ANOMALY (TMINA) AND CHOLERA CASES IN
PIURA. ...........................................................................................................................150
FIGURE 5.32 CROSS-WAVELET OF RAINFALL ANOMALY AND CHOLERA CASES IN PIURA FROM 1991 TO
2001. ...........................................................................................................................151
FIGURE 5.33 TOTAL CHOLERA CASES IN PIURA FOR 1997-98, BASED ON NATURAL BREAKS CLASSIFICATION
....................................................................................................................................155
FIGURE 5.34 CHOLERA INCIDENCE RATE (PER 1000) FOR 1997-98 IN PIURA, BASED ON NATURAL BREAKS
CLASSIFICATION ................................................................................................................ 156
FIGURE 5.35 SOCIAL VULNERABILITY INDEX (SVI), BASED ON STANDARD DEVIATION CLASSIFICATION. ...161
FIGURE 5.36 RURAL AND RIVER WATER (SVIF1), BASED ON STANDARD DEVIATION CLASSIFICATION ....162
FIGURE 5.37 URBAN AND WATER TRUCK (SVIF2), BASED ON STANDARD DEVIATION CLASSIFICATION...163
FIGURE 5.38 URBAN AND WELL WATER (SVIF3), BASED ON STANDARD DEVIATION CLASSIFICATION. ...164
FIGURE 5.39 OTHER WATER SOURCES (SVIF4), BASED ON STANDARD DEVIATION CLASSIFICATION. .....165
FIGURE 5.40 CHOLERA INCIDENCE PREDICTIONS BY SOCIAL VULNERABILITY INDEX (SVI). PREDICTED
VALUES, BASED ON NATURAL BREAKS CLASSIFICATION, FROM GEOGRAPHICALLY WEIGHTED REGRESSION
(GWR) ..........................................................................................................................169
xiii

FIGURE 5.41 CHOLERA INCIDENCE PREDICTIONS BY SOCIAL VULNERABILITY INDEX (SVI). STANDARDIZED
RESIDUALS, BASED ON THE STANDARD DEVIATION CLASSIFICATION, FROM GEOGRAPHICALLY WEIGHTED
REGRESSION (GWR) ......................................................................................................... 170
FIGURE 5.42 CHOLERA INCIDENCE PREDICTIONS BY SOCIAL VULNERABILITY INDEX FACTOR 1 (SVIF1).
PREDICTED VALUES, BASED ON NATURAL BREAKS CLASSIFICATION, FROM GEOGRAPHICALLY WEIGHTED
REGRESSION (GWR) ......................................................................................................... 171
FIGURE 5.43 CHOLERA INCIDENCE PREDICTIONS BY SOCIAL VULNERABILITY INDEX FACTOR 1 (SVIF1).
STANDARDIZED RESIDUALS, BASED ON THE STANDARD DEVIATION CLASSIFICATION, FROM GEOGRAPHICALLY
WEIGHTED REGRESSION (GWR). .......................................................................................... 172
FIGURE 5.44 CHOLERA INCIDENCE PREDICTIONS BY SOCIAL VULNERABILITY INDEX FACTOR 4 (SVIF4).
PREDICTED VALUES, BASED ON NATURAL BREAKS CLASSIFICATION, FROM GEOGRAPHICALLY WEIGHTED
REGRESSION (GWR). ........................................................................................................ 173
FIGURE 5.45 CHOLERA INCIDENCE PREDICTIONS BY SOCIAL VULNERABILITY INDEX FACTOR 4 (SVIF4).
STANDARDIZED RESIDUALS, BASED ON THE STANDARD DEVIATION CLASSIFICATION, FROM GEOGRAPHICALLY
WEIGHTED REGRESSION (GWR). .......................................................................................... 174
FIGURE 5.46 CHOLERA INCIDENCE RATE PER 100,000 (SQUARE-ROOT TRANSFORMED) FOR PIURA IN
1997-98 BY MONTH.........................................................................................................177
FIGURE 5.47 CHOLERA INCIDENCE RATE PER 1000 (SQUARE-ROOT TRANSFORMED) BY DISTRICT AND
MONTH IN 1998. ............................................................................................................. 178
FIGURE 5.48 CHOLERA INCIDENCE RATE PER 1000 (SQUARE-ROOT TRANSFORMED) BY DISTRICT AND
MONTH IN 1998. ............................................................................................................. 179
FIGURE 5.49 CHOLERA INCIDENCE RATE PER 1000 (SQUARE-ROOT TRANSFORMED) BY DISTRICT AND
MONTH IN 1998. ............................................................................................................. 180
FIGURE 5.50 CHOLERA INCIDENCE RATE PER 1000 (SQUARE-ROOT TRANSFORMED) BY DISTRICT AND
MONTH IN 1998. ............................................................................................................. 181
FIGURE 5.51 CHOLERA INCIDENCE RATE PER 1000 (SQUARE-ROOT TRANSFORMED) BY DISTRICT AND
MONTH IN 1998. ............................................................................................................. 182
FIGURE 5.52 CHOLERA INCIDENCE RATE PER 1000 (SQUARE-ROOT TRANSFORMED) BY DISTRICT AND
MONTH IN 1998. ............................................................................................................. 183
FIGURE 5.53 SEA SURFACE TEMPERATURE ANOMALY (SSTA) BY MONTH FOR NIÑO 3.4, NIÑO 1+2, AND
PAITA IN 1997-98. ..........................................................................................................185
FIGURE 5.54 TEMPERATURE MAXIMUM (TMAXA), MEAN (TMEANA), AND MINIMUM (TMINA)
ANOMALIES BY MONTH IN 1997-98. ....................................................................................186
FIGURE 5.55 RAINFALL (SQUARE-ROOT TRANSFORMED) BY MONTH IN 1997-98. ............................187
xiv

FIGURE 5.56 MAP OF SELECTED DISTRICTS FOR CLIMATE-CHOLERA ANALYSIS ................................189
FIGURE 5.57 NIÑO 3.4 SEA SURFACE TEMPERATURE ANOMALY (SSTA) (1 MONTH LAG) ASSOCIATIONS
WITH CHOLERA INCIDENCE RATES (PER 1000), R-SQUARED VALUES BASED ON NATURAL BREAKS
CLASSIFICATION ................................................................................................................ 192
FIGURE 5.58 NIÑO 3.4 SEA SURFACE TEMPERATURE ANOMALY (SSTA) (1 MONTH LAG) ASSOCIATIONS
WITH CHOLERA INCIDENCE RATES (PER 1000), P VALUES BASED ON NATURAL BREAKS CLASSIFICATION. . 193
FIGURE 5.59 PAITA SEA SURFACE TEMPERATURE ANOMALY (SSTA) (1 MONTH LAG) ASSOCIATIONS WITH
CHOLERA INCIDENCE RATES (PER 1000), R-SQUARED VALUES BASED ON NATURAL BREAKS CLASSIFICATION
....................................................................................................................................194
FIGURE 5.60 PAITA SEA SURFACE TEMPERATURE ANOMALY (SSTA) (1 MONTH LAG) ASSOCIATIONS WITH
CHOLERA INCIDENCE RATES (PER 1000), P VALUES BASED ON NATURAL BREAKS CLASSIFICATION .........195
FIGURE 5.61 RAINFALL TOTAL (1 MONTH LAG) ASSOCIATIONS WITH CHOLERA INCIDENCE RATES (PER
1000), R-SQUARED VALUES BASED ON NATURAL BREAKS CLASSIFICATION.......................................196
FIGURE 5.62 RAINFALL TOTAL (1 MONTH LAG) ASSOCIATIONS WITH CHOLERA INCIDENCE RATES (PER
1000), P VALUES BASED ON NATURAL BREAKS CLASSIFICATION. ...................................................197
FIGURE 5.63 RAINFALL TOTAL (2 MONTH LAG) ASSOCIATIONS WITH CHOLERA INCIDENCE RATES (PER
1000), R-SQUARED VALUES BASED ON NATURAL BREAKS CLASSIFICATION.......................................198
FIGURE 5.64 RAINFALL TOTAL (2 MONTH LAG) ASSOCIATIONS WITH CHOLERA INCIDENCE RATES (PER
1000), P VALUES BASED ON NATURAL BREAKS CLASSIFICATION. ...................................................199
FIGURE 5.65 TEMPERATURE MAXIMUM ANOMALY (TMAXA) (6 MONTH LAG) ASSOCIATIONS WITH
CHOLERA INCIDENCE RATES (PER 1000), R-SQUARED VALUES BASED ON NATURAL BREAKS CLASSIFICATION
....................................................................................................................................200
FIGURE 5.66 TEMPERATURE MAXIMUM ANOMALY (TMAXA) (6 MONTH LAG) ASSOCIATIONS WITH
CHOLERA INCIDENCE RATES (PER 1000), P VALUES BASED ON NATURAL BREAKS CLASSIFICATION .........201
FIGURE 5.67 TEMPERATURE MEAN ANOMALY (TMEANA) (6 MONTH LAG) ASSOCIATIONS WITH CHOLERA
INCIDENCE RATES (PER 1000), R-SQUARED VALUES BASED ON NATURAL BREAKS CLASSIFICATION.........202
FIGURE 5.68 TEMPERATURE MEAN ANOMALY (TMEANA) (6 MONTH LAG) ASSOCIATIONS WITH CHOLERA
INCIDENCE RATES (PER 1000), P VALUES BASED ON NATURAL BREAKS CLASSIFICATION. .....................203
FIGURE 6.1 MAP OF SELECTED DISTRICTS FOR COMPARATIVE ANALYSIS ........................................218

xv

1. CHAPTER 1: INTRODUCTION
Understanding climate impacts on infectious diseases is becoming increasingly
important as public health practitioners are concerned that a changing climate will not
only affect the ecology of infectious diseases, but also the basic determinants of health
1

that protect populations from disease transmission. From an ethical perspective, an
imminent concern is that climate-related impacts on infectious diseases will be
disproportionately felt in developing countries (World Health Organization [WHO]
2008), where basic needs are inadequately met in large segments of populations
(Gasper 2005: 1). The untoward effect of poor societal conditions on global health is
already evident by the high prevalence rates of malnutrition and preventable diseases,
such as cholera, in Latin America, Asia and Africa (WHO 2009). According to Harm de
Blij (2009) these human conditions reflect the geography of uneven global development.
The broader implication is that climate change along with existing human vulnerabilities
will increase the risk of infectious disease outbreaks. Despite these concerns, most
studies about climate change and infectious disease have focused on the ecology of
disease with less attention paid to the influence of social vulnerability (Cutter et al.
2003). It is only recently that efforts have begun to address the intersections of climate
and society and their duel potential effects on infectious disease and population health
(Galvao et al. 2009).
1

By basic determinants of health, I am referring to factors, highlighted by the recent
climate change initiative at the WHO (2008). They include clean water and air,
adequate shelter and quality health care, and reflect Gasper’s (1996) concept of basic
needs ethics.
1

In this dissertation research, I will contribute to this emerging body of research
by investigating the cholera epidemic in Peru during the 1990s. The epidemic was a
significant event because cholera, a waterborne disease caused by the bacteria Vibrio
cholerae, reemerged in Latin America after being absent for almost a century. Within
one year of the initial outbreak approximately 400,000 cholera cases were reported in
the region (Ministry of Health, Lima, Peru [MINSA] 1994b). Thereafter, cholera
remained endemic in Peru and throughout Latin America until 2002 (MINSA 2005a).
The emergence of cholera was initially attributed to contaminated waste dispelled from
a passing ship from Asia (Gangarosa and Tauxe 1992: 353); however, this explanation
was later challenged by another hypothesis, which suggested that the epidemic was
precipitated by El Niño-Southern Oscillation (ENSO) (Epstein et al. 1993; Colwell 1996).
ENSO, which includes El Niño (warm phase) and La Niña (cold phase), is an important
source of climate variability in the Latin American region with ecosystem and societal
impacts reported in Peru (Glantz 1981; Caviedes 1984; Lagos and Buizer 1992).
The El Niño-cholera hypothesis is a well accepted explanation for the epidemic
supported by a few studies that found positive relationships between elevated ambient
and sea water temperatures and cholera incidence in Peru, particularly during the 199798 El Niño (MINSA 1994b; 1998d; 2000; Speelmon et al. 2000; Huanca 2004; Gil et al.
2004). Adding credibility to the hypothesis is growing global evidence of climatecholera links in Bangladesh (Pascual et al. 2000; Lobitz et al. 2003), India (Ruiz-Moreno
et al. 2007; Constantin de Magny et al. 2008), Ghana (Constantin de Magny et al. 2007),
and South Africa (Mendelsohn and Dawson 2008). In Peru, however, the El Niño and

2

cholera association remains unclear for several reasons. First, there are questions about
the timing of El Niño and the onset of the cholera epidemic in 1990-91. For example, to
date, it has not been proven that El Niño precipitated the cholera outbreak (SalazarLindo et al. 2008). Second, most studies in Peru have focused on temperature-related
impacts on cholera and not El Niño per se. Third, less is known about other aspects of
ENSO, such as the best definition of the event to use when studying cholera impacts, the
potential impact of the cold phase La Niña, temporal and geographic variations in ENSOcholera impacts, and the potential impact of rainfall extremes. Moreover, social factors
that contributed to population vulnerability, which may have exacerbated ENSO-related
impacts, have not been considered in the El Niño explanation of the cholera epidemic in
Peru.
The goal of my dissertation research is to reconstruct the temporal and spatial
associations among ENSO, social vulnerability and cholera incidence in the health
subregion of Piura, Peru from 1991 to 2001 in order to better understand El Niño’s
impact on the cholera epidemic in Peru. Figure 1.1 is a map of the Department of Piura
that highlights the subregion of Piura and the capital city of Piura. Piura is important to
study because it was one of three places where cholera was first reported during the
initial outbreak in 1991. Subsequently, Piura had one of the highest cholera incidence
rates in the country (MINSA 1994b). Piura is also a region historically known for El Niño
(Woodman 1998), poverty and preventable infectious diseases (Sandoval 1999; Ministry
of Health, Piura, Peru [MINSA Piura] 2005). Therefore, more knowledge of this region

3

will help to improve our understanding of the independent and interactive effects of
climate variability and social factors on cholera transmission.
80°0'0"W

4°0'0"S

Department of Piura

4°0'0"S

81°0'0"W

5°0'0"S

5°0'0"S

Peru

6°0'0"S

6°0'0"S

PIURA
Piura

HealthSubregions
Health Subregions
Luciano Castilla
Luciano Castilla
Piura
Piura
0

81°0'0"W

25

50 Kilometers

80°0'0"W

Figure 1.1 Map of the Department of Piura that highlights the subregion of Piura. For
interpretation of the references to color in this and all other figures, the reader is
referred to the electronic version of this dissertation.

4

My overarching research questions will be: (1) What was the impact of ENSO on
cholera incidence in Piura; and (2) How did social vulnerability influence this
relationship? I will address my research questions using a climate affairs approach
(Glantz 2003; Consortium for Capacity Building [CCB] 2010) that is informed by disease
ecology and social vulnerability theories from the geographic subfields of medical
(Mayer 2000; Meade and Erickson 2005) and human-environment geography (Cutter
2003; Turner et al. 2003; Zimmerer and Bassett 2003). Climate affairs is an integrating
concept in the earth and social sciences used to understand the interrelationships
among climate, environment and society worldwide. A climate affairs approach
encompasses, but is not limited to: 1) climate science and knowledge as a foundation to
understanding atmospheric processes and interactions with natural and human
environments; it also emphasizes the importance of multidisciplinary efforts and local
and regional knowledge to this understanding; 2) climate impacts are the positive and
negative effects on ecosystems and societies, and societal responses to these
interactions; and 3) climate ethics and equity refers to moral issues that arise from
climate-society interactions (e.g., differential vulnerability and impacts) (Glantz 2003).
From a climate affairs orientation, I will suggest a conceptual framework that: 1)
examines the cholera transmission cycle within a broader conception of ENSO; 2) links
ENSO-cholera associations to social vulnerability; and 3) considers ENSO-cholera
interactions at multiple scales. This study is also informed by previous research on
cholera (Cueto 2003; Nelson et al. 2009); climate impacts on cholera ecology
(Constantin de Magny and Colwell 2009); El Niño, climate and society studies (Glantz

5

2001a; Caviedes 2001); and dissertation fieldwork in Lima and Piura, Peru in the
summers of 2008 and 2009. The quantitative and qualitative associations among ENSO,
social vulnerability and cholera incidence will be explored using a mix of time series
approaches, multivariate and geostatistical methods, and a historical review of
documents from the period of the epidemic.
In this introductory chapter, I will present a brief review of the cholera epidemic
in Peru and explanations for the epidemic including the El Niño-cholera hypothesis. I
will then present a set of arguments that motivated me to challenge the current
assumptions about El Niño impacts on cholera in Peru. I will conclude this chapter with
the study significance and format of the dissertation.

1.1

The Cholera Epidemic in Peru
In the early 1990s, Peru experienced the first cholera epidemic in the western

hemisphere since the 1890s (Centers for Disease Control and Prevention [CDC] 1991). It
marked the arrival of the El Tor strain of V. cholerae, which was responsible for the
Seventh Cholera Pandemic that originated in Indonesia in 1961 (Glass et al. 1991). The
first cases of cholera were reported almost simultaneously in January 1991 across three
coastal cities: Chancay, Chimbote and Piura (MINSA 1994b). One year following these
initial outbreaks the epidemic spread throughout Peru and subsequently to other Latin
American countries, infecting almost 400,000 people (Pan American Health Organization
[PAHO] 1991a). The disease was so widespread because populations lacked immunity
and public water and sanitation infrastructures were either in decline or not accessible

6

(PAHO 1991b; Salazar-lindo et al. 1993: 401-413; Tauxe et al. 1995). Over the next
decade, 52.0% (n=703,000) of all reported cholera cases in the Latin American region
were reported in Peru (PAHO 2008).
The emergence of cholera in Peru took public health authorities at MINSA, PAHO
and CDC by surprise. Initially, they could not believe it was cholera because the disease
had not been reported in Latin America for almost a century (Gangarosa and Tauxe
1992). However, a joint effort led by MINSA and CDC in February 1991 confirmed that
cholera was indeed the disease afflicting hundreds of Peruvians (CDC 1991). Health
officials were also puzzled because the disease had not appeared in previous decades
when living conditions in Latin America were reportedly worse (Gangarosa and Tauxe
1992). Authorities across the region were anticipating a major cholera outbreak in the
1970s after an explosive epidemic was reported in West Africa (Glass et al. 1991). In
response, many countries in Latin America including Peru began to foster diarrheal
disease programs with the help of multilateral organizations who supplied them with
resources including oral rehydration solutions (Glass et al. 1991; Gangarosa and Tauxe
1992). Cholera, however, did not appear as expected in the 1970s or 1980s, but instead,
emerged unexpectedly in the 1990s.

7

1.2

Explaining the Cholera Epidemic in Peru

1.2.1 Passing Ship Hypothesis
The source of the cholera epidemic in Peru remains unknown, but several
hypotheses have been put forth to explain the emergence. Two initial explanations
were associated with a passing ship. The first explanation suggested that an infected
person on a ship from Asia docked off the coast of Peru and through unknown activities,
introduced V. cholerae to the mainland’s public water system. The second explanation
suggested that a passing ship dumped V. cholerae contaminated ballast water (waste
and discharge from ballast tanks) into a harbor along the coast infecting marine
organisms, which subsequently came into contact with human populations (Gangarosa
and Tauxe 1992: 353). Both explanations were plausible since cholera outbreaks in the
past had been associated with travelers (Glass et al. 1991) and ballast water (McCarthy
and Khambaty 1994).

1.2.2 El Niño-cholera Hypothesis
Evidence that disputed the passing ship hypothesis was later reported by Seas et
al. (2000) who identified 5 clinical cases of cholera preceding the epidemic (as early as
October 1990) in coastal cities in Peru. The study suggested that V. cholerae was

8

2

already present in the coastal environment of Peru (prior to the arrival of the ship ), and
that cholera transmission was at low levels among the population. The earliest case was
found on 23 October 1990 in Trujillo (approximately 600 km north of Lima). The study
by Seas et al (2000) was important because it supported a second hypothesis that linked
the cholera epidemic to El Niño (Epstein et al. 1993; Colwell 1996; Mourino-Perez 1998).
This alternative explanation suggested instead that El Niño contributed to ecological
changes in the eastern equatorial Pacific Ocean, which in turn promoted the abundance
of plankton (assumed to harbor V. cholerae) off the coastal shores of Peru.
Subsequently, storm surges transported infected plankton inland where transmission
occurred at multiple locations along the coast (Colwell 1996; Seas et al. 2000). Epstein
(1992; 1993) was the first to associate the epidemic with climate but also suggested that
nutrient runoff (due to agricultural activities) contributed to the reproduction of
plankton blooms. It was later suggested by Colwell (1996) and Mourino-Perez (1998)
that contaminated plankton from Asia may have been transported via eastward-flowing
ocean currents induced by El Niño, which was presumed to have lasted from 1990 to
1995 (Colwell 1996). The reasoning underlying the El Niño-cholera hypothesis was that
El Niño seemed the most logical, if not obvious, mechanism for simultaneous outbreaks
on the Peruvian coast spanning a distance of 1100 km (Seas et al. 2000).

2

It should be noted that finding the original report that documents the arrival of the
ship and time is difficult. I have been citing secondary sources that mention the
hypothesis.
9

1.3

Challenging the El Niño-cholera Hypothesis
Although El Niño’s impact on the epidemic in Peru is plausible, there are several

arguments that motivated me to challenge El Niño’s association. These arguments are
based on characteristics about ENSO, which I argue have been overlooked in previous
studies.
The importance of definition
The El Niño-cholera hypothesis is based upon a contentious El Niño period (1990
to 1995). During this time, there was disagreement among experts about the timing of
El Niño events and the number of events that developed (Glantz 2001: 21). As such, the
timing of El Niño and its impact on the epidemic remains questionable. One factor
which may resolve this issue is the definition of an ENSO event, which affects how one
interprets the cycle characteristics of an El Niño (Trenberth 1997). Considering this
factor may shed new light on the El Niño association and the ENSO context during
cholera transmission in Peru.
The La Niña factor
Preceding and following the 1997-98 El Niño were La Niña events during which
cholera cases appeared to decline. La Niña events are also associated with societal
impacts; however, knowledge about its impacts is less known because this phase is less
studied (Glantz 2002b: 8). Perhaps La Niña periods were protective years for cholera
transmission.

10

Geography of El Niño impacts
Previous El Niño-cholera studies in Peru were limited in their geographic scope
of analysis. These studies followed the initial outbreak and examined cross-sections of
the 1990s. No study has ever examined the entire decade when cholera was prevalent
in Peru or investigated the initial time segment when cholera first emerged. Reasons
why the entire period has not been examined despite its significance are perhaps
3

related to the lack of data and funding challenges. Another reason could be that after
cholera subsided in 2004 other urgent health issues became public health priorities.
Furthermore, El Niño-cholera studies were generally focused on Lima. El Niño-cholera
associations in other geographies of the country are less known.
Rainfall extremes
Rainfall extremes are important teleconnections in Peru, which have not been
investigated in relation to the cholera epidemic. In 1997 to 1998, there was an extreme
El Niño, which became historical because of its impacts on economic sectors,
infrastructure and human health including a resurgence of cholera in Peru (Consejo
Nacional del Ambiente [CONAM] 1999: 159-161; Amat y Leon et al. 2008: 18-19). Much
of the reported damages and health impacts were associated with heavy rains, which
suggests that rainfall extremes may have been an important pathway for cholera
transmission in Peru in 1997-98 and perhaps previously.

3

Lack of funding may be a reason for researchers at Peruvian institutions. Dr. Ana I. Gil
at the International Institute of Nutrition in Lima conveyed to me that resources were
limited (personal communication July 2009).
11

Social dimensions
Prior to the emergence of cholera in 1990/91, several El Niños were reported in
Peru including the mega event of 1982-83, and yet cholera outbreaks were not reported
during that time. Therefore, this may suggest that the presence of El Niño alone cannot
solely explain the cholera epidemic in Peru. It has been reported that cholera diffused
rapidly because of widespread infrastructure and socioeconomic deprivation across
Peru (PAHO 1991b; Brooke 1991; Ries et al. 1992; Swerdlow et al. 1992; Besser et al.
1995; Ticker and Gouveia-Vigeant 2005). Furthermore, the country was responding to
humanitarian crises (United Nations Disaster Relief Organization [UNDRO] 1990b;
1990a), while restructuring its economy (Nash 1991) and combating terrorism (Youngers
2000). Integrating this underlying context of population vulnerability will be important
to fully understand the role of ENSO impacts on cholera outbreaks and transmission in
Peru.
In light of these arguments, the cholera epidemic in Peru and the association
with El Niño warrants further inquiry. In particular, the story and explanations about
the cholera epidemic in Peru require rethinking, not only because of the arguments
presented above, but also because the explanations (climate and social) have not been
integrated. Thus, retelling the story about cholera in Peru with combined elements
from existing explanations and new information would yield a better understanding of
the epidemic in Peru and the role of ENSO and climate on cholera incidence and
transmission.

12

1.4

Study Significance
This research will advance knowledge in geography by developing a framework

that integrates theoretical approaches from medical geography and humanenvironment geography to address climate impacts on infectious disease and health, an
emerging area of inquiry among disciplines and policy interest among communities,
governments, and international relations. Specifically, this research links knowledge in
Piura, Peru to the growing research on climate and cholera in Bangladesh, India,
Vietnam, South Africa and Ghana. More broadly, it speaks to the larger body of work
pertaining to climate change, health and society (WHO 2008). It also contributes to the
growing body of literature on development ethics and the newer field of ‘climate’ ethics
and justice by engaging moral concerns and questions about ethics and equity issues
that arise from climate-disease-society interactions. As such, this research is part of and
promotes climate affairs, an approach to climate and society studies that emphasizes
multidisciplinary research that bridges the natural and social sciences with the
humanities. For public health and development policymakers this study provides an
analogy from which to learn lessons about climate variability and extremes, infectious
disease, and human well-being.

1.5

Dissertation Format
I divided my dissertation into seven chapters. Chapter 1 is the introduction.

Chapter 2 presents a literature review of topics pertinent to this study. It includes ENSO
science and knowledge, climate and cholera ecology, cholera transmission and human

13

ecology, and cholera and population vulnerability. Chapter 3 reexamines El Niño’s
association with cholera in Peru, presents several arguments that challenge the
association, and concludes by stating my research objectives and hypotheses. Chapter 4
describes my research design including the research approach, theoretical concepts and
framework, and the data and methods for each research objective. Chapter 5 highlights
key findings. Chapter 6 discusses the findings, addresses my research questions and
hypotheses, and reflects on the ethical geographies of the study. Chapter 7 presents
some concluding thoughts and implications for future research, policy and ethics.

14

2. CHAPTER 2: LITERATURE REVIEW
In this chapter, I present a literature review of topics relevant to this dissertation
research. I begin with ENSO, its cycle and phase characteristics, physical
teleconnections and ecosystem and societal impacts in Peru. In the next section, I
review what we currently know about the relationships among ENSO, climate and
cholera disease ecology in studies around the world. I then discuss cholera transmission
routes and human ecology. In the final section, I describe cholera incidence in relation
to population vulnerability.

2.1

ENSO Science and Knowledge
ENSO is a quasi-periodic natural phenomenon resulting from ocean-atmosphere

interactions in the equatorial Pacific Ocean. It is the second most predictable climatic
fluctuation after the natural flow of the seasons. El Niño, the oceanic component of
ENSO, is known as a warming of waters off the coast of Peru, but also refers to basinwide oceanic changes that extend across the equatorial Pacific Ocean (Bjerknes 1969).
It was first named by Peruvian fishermen (on the northern coast) who noticed a periodic
warming coinciding with Christmas (Carillo 1892). The Southern Oscillation considered
the atmospheric component of ENSO is a seesaw-like pattern of sea-level pressure
measured between Darwin, Australia and Tahiti. Together these components make up
the ENSO cycle, which influences local to global variability and extremes of weather and
climate and impacts ecosystems and human populations within its sphere of influence
(Graham and White 1988; Caviedes 2001; Glantz 2001a; Cane 2004; Philander 2006).
15

The ENSO cycle consists of El Niño (warm) and La Niña (cold) extreme events,
which alter “average” SST conditions in the equatorial Pacific Ocean (National Oceanic
and Atmsopheric Administration [NOAA] 2005). Warm and cold events are
characterized by a multi-phase-process: precursor, the onset (development), growth
and maturity, and decay. Each event is unique in the way it evolves including in
magnitude and duration and the location and intensity of impacts (Glantz 2001a: 100).
Although an event is unique from one time period to another, some common aspects
are generally found, including how often an event occurs (e.g., every 3 to 7 years) and
how long an event lasts (e.g., 12 to 18 months) (Goddard et al. 2001). In the 1990s,
however, it appeared there were more frequent warm episodes, including the strongest
event of the twentieth century (Bell and Halpert 1998). In the next subsections, a
general description is given of ENSO event characteristics and impacts on climate and
society in Peru.

2.1.1 ENSO Events
During “average” conditions in the equatorial Pacific Ocean basin, upwelling
processes off the coast of Peru bring cold and nutrient rich waters from the deep ocean
to the sunlit sea surface zone, which provides an abundance of food for many marine
species including plankton. Easterly winds driven by pressure differences contribute to
upwelling by dragging waters from the eastern Pacific towards the western Pacific,
where warm waters pile up, the oceanic thermocline deepens, and sea level rises
(Philander 1985). Over the western Pacific warm air rises and falls over the eastern
16

Pacific, contributing to low air pressure and wet conditions over Indonesia and N.
Australia, and high air pressure and dry conditions off the coasts of Ecuador and Peru
(Walker 1924).
When El Niño develops, easterly winds weaken and appear to reverse disrupting
upwelling off the coast of Peru and movement of surface waters from the eastern
Pacific to the western Pacific. The interactions and changes in the ocean and
atmosphere are basin-wide and result in a positive feedback: a mass of warm waters
shifts from the west and travel via Kelvin (internal) waves toward the west coast of
South America (i.e., Ecuador, Peru and Chile); convective rainfall activity follows the
warm pool of water, which expands into the central and eastern Pacific, where the
oceanic thermocline deepens; consequently, sea-level heights rise and sea surface
temperatures increase as Kelvin waves carrying warm water reach the coast of Peru
(Philander 1985). Over the western Pacific, air pressure becomes high, suppressing
rainfall over N. Australia and Indonesia; over the eastern Pacific air pressure becomes
low and storms and rainfall occur on a normally arid coast of Peru. During La Niña, the
opposite occurs and “average” conditions are enhanced; easterly winds strengthen and
air pressure rises in the eastern Pacific suppressing rainfall convection and enhancing
coastal upwelling (NOAA 2005).

2.1.2 ENSO Monitoring and Definition
ENSO events are monitored across the equatorial Pacific Ocean basin in 5
delimited areas known as Niño regions: Niño 1 – (5-10°S, 80-90°W); Niño 2 – (0-5°S, 8017

90°W); Niño 3 – (5°N-5°S, 150-90°W); Niño 3.4 – (5°N-5°S, 120-170°W); and Niño 4 –
(5°N-5°S, 160°E-150°W). Figure 2.1 shows a map of these regions. Events are identified
by observing the ocean-atmosphere parameters and conditions described previously
that denote the different stages of the ENSO event life cycle. The most common
indicators of ENSO events are anomalous changes of sea surface temperature (SST),
surface winds, and east-west pressure (NOAA 2005). El Niños and La Niñas are typically
estimated using the aforementioned variables to generate indices, such as Niño Region
Indices (SST anomalies in each region), the Southern Oscillation Index ([SOI], sea level
pressure differences between Darwin and Tahiti) or the Multivariate ENSO Index (a
combination of SST, sea-level pressure and surface winds) (International Research
Institute for Climate and Society 2007; Bureau of Meteorology in Australia [BOM] 2010;
Walter 2010). Other important variables are outgoing longwave radiation (OLR) (e.g.,
to estimate convection) and sea-level height off the coast of Peru (NOAA 2005).

18

Figure 2.1 Map of Niño Regions.
Source: http://www.bom.gov.au/climate/enso/indices/about.shtml.

19

The characterization of an ENSO event and its cycle is determined by the
definition applied. Several definitions of ENSO have emerged since the 1980s that range
from quantitative to qualitative interpretations, as well as regional definitions
4

(Trenberth 1997), such as that utilized by some Peruvian institutions. Quantitatively, an
ENSO event is commonly defined by observing monthly mean SSTs in a selected Niño
region (this depends on the user) over periods where a selected threshold (e.g., +0.4°C)
is exceeded. Trenberth (1997) demonstrated the importance of definition using a
modified operational definition from the Japanese Meteorological Agency, which
requires a minimum of 6 months for a 5-month running means of monthly SST
anomalies to exceed an anomaly threshold. He compared the results of different
anomaly thresholds (i.e., +0.3°C, +0.4°C, and +0.5°C) in Niño 3 and Niño 3.4 regions in
the Equatorial Pacific. The results demonstrated that the anomaly threshold and Niño
region influenced when an event began and ended. Trenberth also found that the Niño
3.4 index was more consistent with historical studies; however, he also recommended
using a definition and criteria that suits the region or needs of the user. Clearly
understanding the ENSO definition that is used is important in health studies because it

4

By Peruvians I am referring to the general understanding and interpretation of El Niño
by the institutions that are responsible for monitoring this phenomenon. The definition
utilized in Peru is based on the Scientific Committee for Ocean Research (SCOR) working
group which states the following criteria signify the appearance of an El Niño along the
coast of Ecuador and Peru as far south as Lima (12°South): 1) normalized sea surface
temperature (SST) anomaly exceeding one standard deviation for at least four
consecutive months; and 2) normalized SST anomaly should occur at least at three of
five Peruvian coastal stations. It differs from NOAA’s operational definition based on
SST anomalies in the Niño Region 3.4 (Lagos et al. 2008); however, I should add that
Peruvians also use the NOAA index for comparison.
20

defines: (a) which indicators will be utilized in the study; (b) how each event is
subsequently identified; and (c) the interpretation of the timing and intensity of events.

2.1.3 ENSO Teleconnections in Peru
During ENSO events, temperature and rainfall patterns are influenced around
the world such that some places may become wetter while others become drier during
El Niño and La Niña (Kousky et al. 1984; Ropeleski and Halpert 1987; 92). These impacts
on climate known as ‘teleconnections’ are based on statistically and physically proven
linkages between ocean-atmosphere interactions in the central and eastern equatorial
Pacific and climate anomalies in distance places (Wyrtki 1973; Flohn and Fleer 1975;
Kiladis and Diaz 1989; Glantz 2001a: 3). Teleconnections are generalized in terms of El
Niño and La Niña associations by season from December to February and June to August
(Ropeleski and Halpert 1987; Diaz et al. 2001). They are strongest during the Southern
Hemisphere (SH) summer (e.g., December to March) when SST is warmest in the
equatorial Pacific (NOAA 2005). Furthermore, the strength of teleconnections depends
on the magnitude and spatial extent of SST anomalies in the Niño regions (Goddard et
al. 2001), as well as the distance from the central and eastern equatorial Pacific Ocean
(Glantz 1998). In this dissertation research the term teleconnections is used more

21

broadly to include: 1) all places that are commonly associated with ENSO; and 2)
impacts based on historical and local references.

5

In Peru, ENSO teleconnections vary by region, but are most notable on the
northern coast, an area which is typically arid, except during El Niños when torrential
rains are reported from December to March (Caviedes 1973; 84; Horel and CornejoGarrido 1986; Woodman 1998; National Meteorology and Hydrology Service of Peru
[SENAMHI] 2004; Rodriguez et al. 2005). In other regions in Peru, El Niño
teleconnections are less well-defined, but can be generalized as follows: warmer than
average conditions from June to August on the central coast (SENAMHI 2004); above
average rains in the Southern Andes in November; below average rains from January to
March in the Southern and Central Andes (Lagos et al. 2008; SENAMHI 2009); and
rainfall deficit in November and December in the Amazon (Marengo 1999; Marengo et
al. 2008). La Niña teleconnections are less known but some reports indicate that colder
than average temperatures and severe drought have been documented in the coastal
north in July and August during cold events (Ordinola 2002). In general ENSO
teleconnections should be interpreted with caution, because of ENSO’s variability in
character from event to event and the influences of other factors on Peru’s climate,
such as the Andes, the Humboldt Current, the InterTropical Convergence Zone, and

5

It is important to broaden the term because using a strict definition of teleconnections
would exclude Piura, the area of study in this research. It would also dismiss impacts
based on local unpublished literature consisting of reports, anecdotes and descriptive
statistics documenting the experiences of Peruvians during ENSO events.
22

conditions in the tropical Atlantic (Lagos and Buizer 1992; Marengo et al. 2008;
SENAMHI 2009).

2.1.4 ENSO Ecosystem and Societal Impacts in Peru
In Peru, ENSO teleconnections and impacts on ecosystems and societies are
documented as far back as the 1500s (Glantz 2001a: 165). Years associated with El
Niños are synonymous with climate-related hazards and disasters (Glantz 2001a: 27;
CONAM 2002). In the last century, the most memorable El Niño years in Peru were
1925, 1972-73, 1982-83 and 1997-98 because of the magnitude of damage to
infrastructure, economy, and people’s lives (Lagos and Buizer 1992; Woodman 1998;
Caviedes 2001). The latter two events were the strongests in the twentieth century.
Impacts are reported during and after events and are sometimes generalized as a preevent, event and post-event in order to compare the effects of ENSO events on different
sectors and population factors. According to a study by Bouma et al. (1997), societies
experience the greatest impacts in post-event years based on the number of persons
affected by El Niño-related disasters. The study suggested that impacts on society are
felt well after El Niño/ La Niña years.
El Niño-related weather and climate can contribute to ecosystem disruption and
damages to economy and infrastructure via temperature and rainfall variability and
extremes. In Peru, some of the most vulnerable sectors to impacts from ENSO extremes
are fisheries, agriculture, energy, and public health (CONAM 2002; SENAMHI 2004).

23

As stated earlier, El Niños disrupt upwelling, which, along with invading warm
waters affect marine ecosystems off the coast of Peru (Barber and Chavez 1983;
Tarazona and Valle 1999; Escribano et al. 2004; Chavez et al. 2008), and ultimately the
livelihoods of coastal communities and the fishing industry. Fish populations respond by
migrating north and south, going to greater ocean depths, or swimming closer to shore.
For some species, such as anchovy, which is an important commercial resource in Peru,
El Niño-related extremes can adversely impact spawning (Pizarro 1999; Carr and Broad
2003). In 1973, the collapse of the anchovy industry was blamed on El Niño, but later it
was recognized that overharvesting was also a major contributing factor (Caviedes and
Fik 1992; Glantz 2001a: 232).
In the agricultural sector, ENSO-related rainfall and temperature extremes can
impact the harvest of crops and livestock (Lagos 1998; Woodman 1998; CONAM 2002).
Two examples are potatoes and rice. In the central highlands of the Andes, it has been
reported that potatoe harvesting is vulnerable to El Niño-related drought (Orlove et al.
2000). While in northern coastal Peru, colder than average conditions during La Niñas
correlate with lower rice yields (Ordinola 2002). ENSO can also impact agriculture
through ecosystem changes which affect the ecology of insects and pathogens that
either feed on or attack crops (e.g., potatoes) (Cisneros and Mujica 1999).
ENSO-related extreme weather can also impact the built environment including
energy infrastructure. Power outages related to storms are particularly damaging since
they can affect water and sanitation systems, electrical service in homes, businesses and
hospitals. During such events, public water supply can be interrupted (e.g., chlorination)

24

and foods can spoil (e.g., no power for refrigeration). In addition, infrastructure damage
because of flooding (e.g., the washing out of bridges, roads and communication lines)
can isolate communities and displace persons who lose their homes (PAHO 1998b).
Furthermore, when collapse of infrastructure (e.g., water and sanitation
systems) co-occurs with ecosystem changes (e.g., changing environmental conditions for
vectors and pathogens) in Peru, human exposure to disease and injury can increase
(Gueri et al. 1984; PAHO 1998a; Valverde 1998; PAHO 1998a; Sandoval 1999; CONAM
2002; Kovats et al. 2003). Overall, ENSO extremes can potentially disrupt societal wellbeing in Peru.

2.2

Climate and Cholera Ecology
Cholera is a diarrheal disease caused by the waterborne bacteria Vibrio cholerae.

There are over 200 serotypes of V. cholerae that exist naturally in aquatic environments,
with two serogroups of public health concern, serogroup 0139 and serogroup 01.
Currently serogroup 0139 is primarily found in Bangladesh, India and Pakistan
(Zuckerman et al. 2007), while serogroup 01, specifically the biotype El Tor, is endemic
in many countries of the world (WHO 2009). The El Tor biotype was responsible for the
Seventh Cholera Pandemic, which began in 1961 in Indonesia and subsequently spread
to South and West Asia, the former Soviet Union, Europe, the Mediterranean, Africa,
and eventually reached the Americas (Glass et al. 1991).
Historically it was thought that humans were the only reservoir for V. cholerae,
but in recent decades studies strongly suggest that aquatic organisms, such as plankton,

25

aquatic plants (Tamplin et al. 1990; Lipp et al. 2002) and biofilm (Alam et al. 2007), can
act as environmental hosts. It is suspected that V. cholerae live in a symbiotic
relationship with marine hosts whereby the bacteria attaches itself to the surface of
organisms that inhabit aquatic ecosystems (Tamplin et al. 1990).
In particular, phytoplankton is an important factor in the climate and
environment association with cholera, and the hypothesis in Peru. When phytoplankton
multiplies it is a prime source of food for organisms called zooplankton, which include
crustaceans (e.g., shrimp and lobster) and copepods (small crustaceans) (Constantin de
Magny et al. 2011). Lipp et al. (2002) has suggested that zooplankton is a disease vector
4

in the cholera transmission cycle. One copepod may contain up to 10 cells of V.
cholerae, an adequate dosage for human infection (Glass and Black 1992: 136).
Exposure to the bacteria is reported to occur when humans drink untreated water or
consume V. cholerae-carrying crustaceans or fish without proper cooking (Huq et al.
2001).
Studies also suggest that V. cholerae is autochthonous to brackish and estuary
waters. That is, these organisms can live naturally in the environment without humans
(Tamplin et al. 1990; Colwell 2004). The suspicion that V. cholerae could reproduce
without human fecal contamination was speculated by Robert Koch who first discovered
the organism in 1883 (Colwell and Spira 1992). Reinforcing this notion were historical
reports of outbreaks in coastal areas without the discovery of a human carrier (Huq et
al. 2001; Pascual et al. 2002). This conjecture along with V. cholerae’s known
association with water, and the evidence of an environmental reservoir has led many
26

researchers to consider climate as an environmental factor in the cholera transmission;
namely, through impacts on the habitat and reproduction of V. cholerae and plankton
(Colwell 1996; Lipp et al. 2002). Climate and environmental factors that are important
to the ecology of V. cholerae and phytoplankton include elevated temperature, pH,
salinity, and iron (Lipp et al. 2002); however, it has also been reported that V. cholerae
can survive in low salinity conditions (Singleton et al. 1982).
Thus far, studies have shown that climate has been associated with cholera in
Bangladesh (Lobitz et al. 2000; Pascual et al. 2000; Emch et al. 2008; Hashizume et al.
2008; Akanda et al. 2009; Hashizume et al. 2010), India (Ruiz-Moreno et al. 2007;
Constantin de Magny et al. 2008), Ghana (Constantin de Magny et al. 2007) and South
Africa (Mendelsohn and Dawson 2008). The following subsections will review these
studies and the global and local climate parameters that were investigated in relation to
cholera incidence.

2.2.1 Global Climate Parameters and Cholera
Niño SST is an important global climate factor that has been associated with
cholera ecology in Bangladesh (Pascual et al. 2000). Pascual et al. (2000) was the first to
demonstrate this association in a study showing that peaks in monthly cholera rates in
the Bay of Bengal, Bangladesh fluctuated with the Niño 3.4 index from 1980 to 1998
during northern hemisphere (NH) spring. It was also observed that another peak
occurred in NH fall (unassociated with ENSO) suggesting that interannual and seasonal
climate variability impacted cholera transmission indirectly through its influence on local
27

coastal air-water temperatures and plankton blooms. Importantly the authors noted
that previous disease levels in the population were also likely to determine cholera
fluctuations overtime. In another study, Rodo et al. (2002) examined monthly cholera
data in relation to the SOI and Niño 3.4 index in order to estimate these relationships
6

across two periods in Dhaka, Bangladesh from: (a) 1893 to 1940 and (b) 1980 to 2001 .
The study found a stronger relationship in the latter period during which ENSO
explained 70.0% of the variance. Niño 3.4 was associated with peaks from 1991 to
7

1994; however, SOI was not. Cholera peaks were primarily associated with extreme
ENSO events, suggesting to the authors that the climate-cholera link was transient
overtime; and that perhaps there was a climate threshold during these periods.
To further investigate the relationship between ENSO and cholera in Bangladesh,
two studies assessed the comparative contributions of climate variability (referred to as
extrinsic) and herd immunity (referred to as intrinsic) on cholera incidence. In the first
study Koelle and Pascual (2004) examined monthly cholera mortality data in Dhaka,
Bangladesh from 1892 to 1940. The study investigated whether population immunity
(defined by birth rates and vaccination coverage data) could partly explain the temporal
variabilty observed by Rodo et al. (2002). These authors showed that immunity decayed
after 9 yrs, and may explain cholera variability incidence on 4 and 8 yr cycles. However,
there was no clear climate association with ENSO, SOI or local rainfall. In conclusion, the
6

Data were unavailable for the years in between the two periods.

7

This observation may reflect the ecological relationship between SST and cholera
ecology; the association between surface pressure and cholera has yet to be discerned.
28

study suggested that seasonality and partial population immunity contributed to rates
of cholera.
In a subsequent study, Koelle et al. (2005) demonstrated that ENSO’s
relationship was mediated by local climate impacts on cholera. In this study monthly
cholera cases in a rural region south of Dhaka, Bangladesh were examined in relation to
the Niño 3.4 index, Bay of Bengal (BOB) SSTA, rainfall and river discharge data for 1966
to 2002. Rainfall was included because it has been reported that cholera decreased
with the monsoon rains due to a dilution effect. The study found that cholera increased
after the monsoon rains because flooding contributed to population congregation
(density was also a factor), and collapse of sanitary conditions due to flooding. River
discharge was a proxy for water levels. Although the focus of the study was cholera
associated with the El Tor strain, cases of the classical strain were also included to
8

estimate herd immunity in the population. The study showed different associations
among the variables (e.g., Niño 3.4 and BOB SSTA and Niño 3.4 and cholera), which
depended on the temporal scale and the strain (e.g., the classical strain was associated
with low water levels). Herd immunity was shown to decay after 3yrs, but partial
immunity existed up to 10 yrs. The authors concluded that after large outbreaks, there
was a marked decrease in cholera incidence due to a decrease in a susceptible
population and increase in population immunity; the authors referred to these time

8

The study considered both strains because it has been reported that cross-immunity
among cholera serotypes can occur; cross-immunity can provide partial immunity
(Koelle et al. 2006).
29

intervals as ‘refractory periods’. It was hypothesized that even if favorable climate
conditions emerged, the likelihood of transmission would be low in times of refractory.
This was believed to explain the non-stationary link between climate and cholera in
Bangladesh.
Outside of Bangladesh, two studies in Africa examined cholera patterns in
relation to global climate parameters associated with the Indian Ocean. In a study by
Constantin de Magny et al. (2007), monthly cholera incidence was investigated in 5
countries (Cote d’Ivoire, Ghana, Togo, Benin, and Nigeria) in the Gulf of Guinea to
understand how the disease evolved overtime in relation to the Indian Oscillation Index
(IOI) and local rainfall from 1975 to 2002. The study found 2 to 5 yr cycles between
cholera and climate in countries except Cote d’Ivoire. Generally, the association
between IOI and rainfall and cholera incidence was particularly significant from 1989 to
1994. Rainfall was reported to be associated with flooding and water contamination.
The authors proposed that the lack of significant associations in Cote d’Ivoire could be
explained by refractory periods. In another study Constantin de Magny et al. (2007)
examined cholera incidence in Ghana from 1975 to 1995 in relation to IOI, SOI, rainfall,
and air temperature anomaly. The results were similar to those found in the Gulf of
Guinea study by Constantin de Magny et al. (2007); it was shown that cholera was
temporally associated with IOI and rainfall from 1989 to 1995.
2.2.2 Local Sea Surface Temperature
Lobitz et al. (2000) was one of the first to examine local climate and cholera in
coastal Bangladesh using Bay of Bengal SSTA and sea surface height anomaly (SSHA).
30

This study found a seasonal pattern between weekly SSTA and percent of confirmed
cholera cases from 1989 to 1995. There were significant associations in 1992, 1994, and
1995. It was also observed that SSHA preceded cholera outbreaks in 1993 and 1995.
Plankton counts were not measured, but it was suspected that rising coastal waters may
have transported plankton inland.
In another study, Constantin de Magny et al. (2008) examined SST, rainfall, and
phytoplankton biomass in relation to monthly cholera cases in two areas: Kolkata, India
and Matlab, N. Bay of Bengal from September 1997 to December 2006. It was assumed
that phytoplankton were environmental reservoirs for V. cholerae. SST was initially
included in the models, but was later removed because of collinearity; furthermore, it
was found that chlorophyll (a proxy for plankton productivity) and rainfall were better
explanatory variables. Importantly, the authors found that cholera hospital admissions
were associated with different climate pathways that varied by place and time (time lag
associations are discussed in section 2.2.4; also refer to Appendix 1). In Kolkata, which is
9

closer to the coast (relative to Matlab) , plankton blooms were associated with heavy
rains, nutrient run-off and blooms, while in Matlab, Bangladesh tidal intrusion
transported plankton inland to the low-lying coast. The authors suggested that cholera
transmission occurred when people used local rivers for cleaning and drinking water.
The importance of geographic differences in local climate and cholera dynamics
was further elucidated in a study by Emch et al. (2008), which investigated
environmental drivers of monthly cholera outbreaks in Matlab and Hue and Nihau Tran,
9

The distance between the two locations is approximately 210 km.
31

Vietnam from 1983 and 1985 to 2003. The authors showed that in Matlab, chlorophyll
was significant and positively associated with outbreaks; while river discharge was
negatively associated with outbreaks. SST, SSH, rainfall and air temperature were not
statistically significant. However, in Hue, elevated SST was significant and increased the
risk of a cholera outbreak; whereas, elevated SSH and river height decreased cholera
risk. SST was only significant when SSH was controlled in the model. In Nihau Tran, SST
and SSH were not significant, but an increase in river discharge and height were
positively associated with cholera outbreaks. Emch et al. (2008) suggested that river
discharge and height were related to flooding and possibly dual transmission pathways.
The differences in SST associations by place were not explained by the authors.
A possible explanation for the differences in SST associations in the Emch et al.
(2008) study could be local geography. Hue is separated from the ocean by several
kilometers of estuary, while Nihau Tran is a city on the central coast of Vietnam
suggesting that the physical environment and proximity to coast may have influenced
the climate mechanisms in each place. According to Mendelsohn and Dawson (2008),
topography may have explained why SST was a predictor of cholera and SSH was not in
KwaZulu-Natal, South Africa during an outbreak in 2000-01. In this study, the authors
explained that coastal intrusion did not occur via SSH because the low-lying coast in
KwaZulu-Natal becomes abruptly steep as you travel inland. Thus, the physical
geography may have prevented coastal instrusion and potential human-environment
interactions with V. cholerae. In a more recent study, Emch et al. (2010) showed that
results in Matlab were similar to those found by Emch et al. (2008). SST was not

32

significant and SSH and chlorophyll were associated with cholera incidence in the late
NH spring or pre-monsoon period. In this study, the authors suggested that the scale of
the study may have contributed to the difference in results (i.e., compared to other
studies that found an association).

2.2.3 Local Rainfall
As stated earlier, there is evidence that rainfall may increase and/or decrease
cholera risk (Kovats et al. 2003). For example, Ruiz-Moreno et al. (2007) demonstrated
that heavy rains increased the risk of cholera through exposure from floods, transport
and contamination of water supply in Madras, India. Similarly, Hashizume et al. (2008)
found that weekly cholera incidence in Dhaka, Bangladesh increased (14.0%) with a 10
mm increase in rain from 1996 to 2002. River levels partly explained this association but
overflow was not a likely transmission pathway. Instead, it was suggested that heavy
rains washed away predators of V. cholerae, and thereby, increased the bacteria’s
survival. In KwaZulu, South Africa, Mendelsohn and Dawson (2008) also found strong
associations between rainfall and cholera rates during an outbreak in 2000-01. These
authors hypothesized that heavy rains impacted the distribution of plankton, which
subsequently led to cholera transmission. Although they found statistically significant
results, they recommended longer time series analyses to support their findings.
Ruiz-Moreno et al. (2007) also showed that heavy rains decreased cholera risk by
diluting the concentration of bacteria in Madras, India, a pathway suggested earlier by
Koelle et al. (2005) in Bangladesh and Codeco (2001) in the Brazilian Amazon.
33

Furthermore, cholera transmission may also depend on the season and the rainfall
extreme (abundance or deficit). Akanda et al. (2009) demonstrated this aspect of the
rainfall-cholera relationship by examining Bay of Bengal SSTA and streamflow (i.e.,
associated with rainfall) from the Brahmatuputra and Ganges rivers in Bangladesh. This
study found that from 1980 to 2000 cholera outbreaks in the NH spring were associated
with rainfall deficient years; while in the NH fall, rainfall abundance years were
associated with outbreaks, most likely via coastal intrusion. It was also shown that SST
was significantly associated with cholera in the NH fall. These authors, like others
(Mendelsohn and Dawson 2008; Constantin de Magny et al. 2008), attributed rainfall’s
impact on cholera to conditions of the water supply in a place; therefore, it was
suggested that the combination of local rainfall and local infrastructure conditions may
affect the transmission cycle through influences on hygienic practices and other waterrelated activities that lead to exposure.

2.2.4 Temporal Lags
The literature review showed that climate and cholera associations are found at
global (e.g., Niño Region SST and IOI) and local (e.g., SST, rainfall, temperature and other
environmental variables) scales. It also suggested a need for climate-cholera studies
that are specific to ‘place’ to better understand how local climate impacts local cholera
ecology. Within the study of ‘place’, there is also a need to understand time based
associations. Climate was shown to influence cholera transmission at different time
scales (e.g., interannual and seasonal) in different geographies. These associations
34

could be characterized as periodic, quasi-periodic, or non-stationary depending on the
climate parameter. Furthermore, the literature indicated that there existed a wide
range of time lags (e.g., zero to 16 months) with apparent trends among the
parameters. Refer to Appendix 1 for a table of climate-cholera studies by area and the
time associations reported in several of the key studies discussed earlier.
Global parameters, such as Niño SST and the IOI, were associated with impacts
on cholera transmission with lags of 8 to 12 months. In contrast, river height and
rainfall, two local parameters, were associated with shorter time lags (e.g., Emch et al.
[2008] found zero-2 month lags). This difference between global and local parameters
may suggest that cholera impacts are also influenced by the distance from the climate
source (e.g., the further away from the climate source, the longer the time-delay). In
studies that examined local SST impacts on cholera, it was reported that time lags
ranged from zero to 9 months (e.g., Bouma and Pascual [2001] and Koelle et al. [2005]).
Temporal lags by ‘place’ (i.e., Piura, Peru) are important factors that will be explored in
this dissertation research because they will help to explain the time-space process
between climate teleconnections and impacts on cholera transmission.

2.3

Cholera Transmission and Human Ecology
Cholera is transmitted to humans through the ingestion of water and food

contaminated with V. cholerae. Once a person is infected the incubation period can be
1 to 5 days before the onset of symptoms (Glass and Black 1992: 141). Symptomatic
cholera infection may occur in several stages beginning with acute watery diarrhea and

35

vomiting followed by severe dehydration and death if supportive treatment via oral or
intravenous rehydration is not immediately initiated (Mahalanabis et al. 1992: 253). A
person must ingest a dosage of approximately 108 V. cholerae organisms in order to
become symptomatic (Glass and Black 1992: 140). According to the WHO (2009),
80.0% of persons that become symptomatic develop mild cholera symptoms, while
20.0% experience acute symptoms. Importantly, the WHO reports that 75.0% of
infected persons show no symptoms, but are still capable of transmitting the disease to
susceptible populations. This lack of complete understanding of underlying
population(s) at risk (e.g., susceptibles versus non-susceptibles) makes disease
surveillance and the planning of prevention and control efforts challenging.
Cholera is commonly transmitted through water contamination. Some common
sources of water contamination are municipal water supplies, surface water, lakes,
rivers and aquifers for drinking and bathing (Butler and Sack 1990) and open wells.
Municipal water supplies may become contaminated if there are leaks due to
infrastructure decline and/or are not adequately treated with chlorine that kill the V.
cholerae. It has also been reported that people who are in need of water illegally break
into water pipes, and in doing so contaminate the water supply. In places with
intermittent electricity (common in many rural areas in developing countries), the
municipal water supply can also become contaminated because the system may
interrupt chlorination (Tickner and Gouveia-Vigeant 2005). In communities where
sanitation services are unavailable or inadequate, it has been reported that diarrheal
disease is associated with wastewater and solid waste disposed in streets and landfills,

36

which can contaminate surface water, aquifers and wells, as well as rivers and lakes
through run-off (Govender et al. 2011). If wells and rivers are the only water available,
local people using these water sources can be exposed to V. cholerae as well as to other
water-borne pathogens. Another source of water contamination is tanker trucks that
may sell water obtained from the public water supply. Water trucks are found in many
places in developing countries where potable water infrastructure is unavailable in
homes or a clean water source is scarce. If the public water supply is infected, then the
water disbursed from tanker trucks can lead to widespread cholera transmission (WHO
2006).
Cholera is also transmitted through contaminated foods such as fruits and
vegetables washed with contaminated water or undercooked food. A person can
become infected or infect food if hygienic practices in the household are unsanitary
(e.g., do not wash hands or boil water). Furthermore, transmission can occur due to
improper food handling and storage. In homes of lower socioeconomic status,
unsanitary hygienic practices could be associated with lack of access to clean water
because they not have the financial resources to obtain this basic need. This
disadvantage can limit a person’s capacity to wash their hands and food. Washing fruits
and vegetables is particularly important in places where crops are irrigated with
contaminated water (Ticker and Gouveia-Vigeant 2005). Street vendors and local
markets are also sources of food contamination. They are common and cheap sources
of food and drink for many people, especially the poor living in developing countries. In
Peru and Guatemala vendors were found to spread cholera because they stored food in

37

places exposed to warmer temperatures, an environment conducive for bacteria growth
(Tauxe et al. 1995; Koo et al. 1996). They also used contaminated ice for beverages
(Ries et al. 1992; Tauxe et al. 1995).
In addition, a person may also become infected by the consumption of
undercooked food contaminated with V. cholerae. Tauxe et al. (1995) reported that
leftover rice and uncooked seafood (inadequately heated) were two routes of
transmission in Latin America. However, it should be noted that uncooked seafood is
commonly eaten in coastal communities in Latin America because of cultural traditions.
For example, ceviche, a raw seafood dish served during festivities, was associated with
the Latin American cholera epidemic. According to local practice, raw fish is marinated
in acidic lime- juice, which is believed to ‘cook’ the seafood and neutralize bacteria
(Tauxe et al. 1995).

2.4

Cholera and Population Vulnerability
Susceptibility to symptomatic cholera can vary by individual-level risk factors,

such as genetic predisposition, previous exposure to cholera, nutrition, age and gender.
Persons with blood type O appear to have a predisposition for severe cholera
(Gangarosa and Tauxe 1992: 355). The mechanism underlying this risk factor is unclear
but was observed during the 1991 cholera outbreak in Peru and Latin America (Huq et
al. 2001), and in studies in Bangladesh (Glass and Black 1992: 148). Another factor that
is important is herd immunity, presented earlier as an intrinsic variable in cholera
dynamics that can contribute to low transmission in a population (Koelle et al. 2005;

38

Koelle et al. 2006). Infants who are breastfed have been shown to receive protection
through their mother’s immunity (Glass and Black 1992: 137). Immunity can also be
obtained through a vaccination; however, it is advised that public authorities provide
vaccinations with caution because it only provides short-term protection (2 yrs). A
cholera vaccination is also limited in terms of its effectiveness for prevention, and is
reported to have adverse side effects (WHO 2009).

2.4.1 Cholera and Determinants of Health
Although V. cholerae is the causative agent of cholera infection, cholera
transmission is also dependent upon the socioeconomic and environmental conditions
in which people live and work. As described earlier, these factors are referred to as
determinants of health (Marmot 2005; Cockerham 2007; WHO 2008). These
determinants are associated with the basic needs of people (Gasper 1996). They may
include clean water and air, sufficient food, adequate shelter, basic education, safety
and security, and health care (WHO 2008). Whether these instruments of health are
met or deprived in a place can contribute to exposure, infection and ability or disability
to respond. People also need the economic means to obtain these resources. Having
these fundamental human necessities can contribute to the reduction of cholera
transmission and also help infected persons cope if they become ill. Furthermore,
public policies and underlying societal norms can also affect the determinants of
people’s health, and subsequent vulnerability to cholera undermining prevention and
control efforts.
39

According to Thisted (2003) infrastructure and income deprivation are potential
pre-conditions for cholera transmission. Cholera and other diarrheal diseases are
generally prevalent in places where water and sanitation infrastructure is inadequate,
declining or unavailable (WHO 2009). As discussed earlier, an important factor that
contributes to cholera risk is a lack of clean water. People may become exposed to
bacteria because they lack potable water connection in their homes or they are limited
to untreated municipal water supply, a marginal source of drinking water. They may
also lack the money to build adequate infrastructure in their homes or purchase filtered
water. The limited access of water in general, and clean water in particular, may affect
how often a person washes their hands or food.

10

Therefore, the disadvantage of not

having clean water can place some people at greater risk to cholera transmission, as
well as to other water and food-borne illnesses.
The availability and accessibility of health care services is also an important
determinant of how well people can cope and recover from infectious disease, such as
cholera (UNRISD 2007). Cholera rates are higher in places with few healthcare clinics
and services, such as those in rural areas and shantytowns. A study in Bangladesh found
that the case-fatality rates during a cholera epidemic were higher for people who lived
farther away from treatment centers (Ali et al. 2002). As people became infected and
developed symptoms they must immediately seek care and treatment before they
10

In a discussion with a nurse from the Ministry of Health in Sullana, Piura I learned that
many of the people in her district of lower socioeconomic status have limited access to
potable water, and therefore, they do not wash their hands fully; instead they only wash
their fingertips.
40

dehydrate, which can occur within a few hours. If oral rehydration therapy or
intravenous fluids are not provided during the incubation period, a person can die.
Furthermore, if people do not have access to health insurance or cannot afford to pay
for healthcare services, they may not seek treatment and fully recover.
Although not a direct determinant of health, public policy can play an influential
role in the public’s access to resources which help prevent disease or support coping
strategies (Wallace 1988; Briggs and Mantini-Briggs 2003). Two examples that are
relevant to the cholera epidemic in Peru and Latin America are structural adjustment
programs (SAPs) and public health educations messages.
SAPS are economic austerity measures in accordance with the International
Monetary Fund (IMF) and the World Bank designed to address economic problems and
stimulate economic growth through a process of loans, privatization, market
integration, and decrease in public spending. During the 1980s and 1990s, SAPS were
negatively associated with public suffering in developing countries because public
services cuts were disproportionately felt by groups of poorer socioeconomic status
(Abouharb and Cingranelli 2007; Jacobsen 2008: 254). When public services are cut the
allocation of monies into public health prevention of infectious diseases is decreased;
thereby, increasing the risk of transmission and illness in a population (Farmer 2001: 43;
Briggs and Mantini-Briggs 2003: 27).
Another example associated with public policy is the public health campaign
about seafood implemented during the cholera epidemic in Peru. These education
campaigns were promoted to alert the public about the association between cholera

41

and uncooked seafood. Specifically, it suggested avoiding the consumption of ceviche.
Local people on the coast, however, were apprehensive because it was Peru’s national
dish and an important part of their diet that has cultural meaning. Reportedly, the
warnings prevented many people, particularly the poor, from acquiring a cheap source
of protein. Consequently, the nutritional levels among the population declined
following these campaigns (MINSA 1994b; Cueto 2003).

2.4.2 Vulnerable Subpopulations
Cholera rates are highest among vulnerable populations, such as children, the
elderly and women in developing countries (WHO 2009). Although cholera is thought to
typically be associated with adults, a recent study across three countries (Indonesia,
India and Mozambique) demonstrated that children bear the greatest burden. The
authors showed that annual incidence rates ranged from 0.5 to 4.0 (per 1,000) in
children under 5 yrs of age (Deen et al. 2008). There are gender differences as well.
Women are at greater risk, relative to men, because they might participate more in
water-related activities, such as collecting and managing water resources. This has been
observed in rural households in many developing countries (United Nations
Development Fund for Women [UNIFEM] 2009).
Children, women and other groups may also be vulnerable to cholera because
they face unequal treatment in their households and societies. Moreover, these groups
may also be living in impoverished conditions. Undernourished children along with
persons who have compromised immune systems (e.g., those living with HIV) are the
42

most vulnerable to cholera mortality (WHO 2009). In rural communities, women could
be at greater risk of cholera than men because of unequal food distribution, education
and healthcare access. This was reported in Piura, Peru in the rural highlands (Reyes
2002).
Belonging to a certain social or ethnic groups may also be a risk factor for cholera
transmission. For example, it has been documented that persons of poorer
socioeconomic status or indigenous ethnicty face discrimination and unequal treatment
in Latin America. Often these groups may be associated with negative societal
stereotypes (e.g., ignorance and laziness). In the context of a cholera outbreak, such
images portray certain segments of populations at greater risk than others because of
their behaviors and culture (e.g., blame may be directed toward individuals rather than
the institutions that are responsible for providing basic needs) (Cueto 2003: 281; Briggs
and Mantini-Briggs 2003). For example, during cholera epidemics in Peru and
Venezuela, it was reported that public health campaigns were directed to ‘at risk’
groups, such as persons of poor socioeconomic status and indigenous groups
(Joralemon 1999: 53-55; Cueto 2003: 281; Briggs and Mantini-Briggs 2003). The
campaigns reinforced negative stereotypes, which influenced differential treatment. In
Peru, people from shantytowns (e.g., associated with poor socioeconomic status)
experienced discrimination in healthcare services. In some cases, people did not seek
healthcare because of the stigma of cholera (Cueto 2003: 283). In Venezuela, the
Warao people from the Delta Amacuro faced racial and ethnic profiling, which led to

43

limited access to treatment (Briggs and Mantini-Briggs 2003). Overall discrimination can
hinder the efforts of certain social and ethnic groups to respond to cholera illness.

44

3. CHAPTER 3: REEXAMINING EL NIÑO AND CHOLERA IN
PERU
In this Chapter, I reexamine El Niño’s link with the cholera epidemic in Peru. I
begin by recounting what is currently known about temperature-related impacts on
cholera incidence associated with El Niño. I then discuss several arguments, presented
in Chapter 1, which reflect characteristics of ENSO that have not been considered in
previous El Niño-cholera studies in Peru. These characteristics include: (a) the
importance of definition; (b) the La Niña factor; (c) geography of El Niño impacts; (d)
rainfall extremes; and (d) social dimensions. My aim is to set a precedent for the case
study of Piura, which is the focus of this dissertation research. I conclude the chapter by
stating the research hypotheses and objectives of this study.

3.1

Recounting El Niño Impacts on Cholera in Peru
Following the initial cholera outbreak in January 1991, evidence of El Niño-

related impacts on the cholera epidemic was suggested by research in Peru that
examined the effects of temperature on diarrheal disease (Salazar-Lindo et al. 1997;
Checkley et al. 2000; Lama et al. 2004) and V. cholerae (Franco et al. 1997; Lipp et al.
2003). According to Checkley et al. (2000), daily admissions of children with diarrhea at
the National Institute of Health (in downtown Lima) increased by 8.0% when mean air
temperature increased by 1.0°C from 1993 to 1997 in SH summer. These authors also
found that the effect of elevated temperature was greatest in SH winter (June to
August). Furthermore, peak in hospital admissions lagged by approximately one month

45

with the peak in air temperature. In another study, Lama et al. (2004) found a similar
association between monthly mean air temperature and diarrhea in adults at a hospital
in northern Lima from 1991 to 1998. Indirectly, these studies supported reports that
cholera incidence increased during the warmest months (e.g., January to March in Lima)
throughout the 1990s (MINSA 1994b; 1995a; Huanca 2004).
The association between El Niño and cholera incidence did not become apparent
until 1997 in Peru. The association was based on two coinciding events: the onset of
the strongest El Niño of the century and a resurgence of epidemic cholera in July 1997
(MINSA 1998d). According to the World Meteorological Organization (WMO), in April
1997, El Niño conditions were rapidly developing across the central and eastern
equatorial Pacific Ocean (1999: 29-38). By December El Niño was in a mature phase. It
was reported that sea surface temperatures exceeded 28.0-29.0°C along the coast, an
observation not seen since the extreme El Niño of 1982-83 (WMO 1999). El Niño’s
effects on local climate in Peru became evident as air temperatures in Lima grew
approximately 3.0-4.0°C above the mean (e.g., 21.0°C).

11

The development of an El Niño, particularly a strong one, during the winter
months meant that Peruvians experienced a warmer than average winter followed by a
warmer than average summer (Bell and Halpert 1998). The impacts for public health
and cholera risk were severe. For example, in North Lima at the Cayetano Heredia
Hospital, admissions for diarrhea and dehydration incidence rose by 35.0% during the
11

The mean was based on monthly air temperature in Lima from 1993 to 1997
(Checkley et al. 2000).
46

winter months (Salazar-Lindo 1997). Checkley et al. (2000) estimated an excess (due to
El Niño) of 6,225 daily admissions of children with diarrhea relative to an expected
number of cases based on a pre-El Niño pattern of cholera (for 1993 to 1996).
Speelmon et al. (2000) found that from November 1997 to July 1998 a rise in weekly
cholera cases was preceded by a rise in mean air temperature by 3 weeks. These
authors also found a temporal lag of an estimated 9 weeks between the presence of V.
cholerae in sewage waters and a rise in air temperature. This finding supported a
previous study by Franco et al. (1997), which showed that V. cholerae was present in
Lima water sources prior to cholera outbreaks from 1993 to 1995. Franco et al.
suggested that cholera transmission in Lima was amplified by the combined processes of
human fecal contamination and elevated air temperature, which led to a proliferation of
bacteria.
Other studies in Peru examined the effect of seawater temperature on V.
cholerae and cholera incidence. In one study, Lipp et al. (2003) detected V. cholerae in
samples collected monthly from seawater and plankton at 3 coastal sites (Trujillo –
North, Lima – Central, and Arequipa – South) from November 1998 to March 2000 and
October to December 2000. Measurements of air and sea surface temperature were
also taken at these locations. The authors observed that V. cholerae was significantly
correlated with and followed air temperature increases from January to March at each
site. Interestingly, seawater temperature was not significantly associated with V.
cholerae. Cholera cases were not measured in relation to these findings. In a second
study, Gil et al. (2004) revisited the same study areas as Lipp et al. (2003), but included

47

one additional site at Callao (considered part of the greater Lima). The study collected
similar variables as Lipp et al., but also measured them in relation to cholera incidence.
In this study the time period was extended to begin in October 1997 and end in June
2000. Gil et al. (2004) found a significant association between monthly cholera
incidence and elevated seawater temperature during the SH summer of 1998. The
detection of V. cholerae, particularly at the Lima site, coincided with these observations
suggesting to the authors that a link existed among seawater temperature, V. cholerae
and cholera incidence. The effect of elevated SST was particularly an important finding
because Lipp et al. (2003) had not observed this association; perhaps because the 199798 El Niño was not included in the study. It was also important because this finding
linked El Niño impacts on coastal water changes to cholera incidence on the coastal
mainland.

3.2

The Importance of Definition
The El Niño-cholera hypothesis rests on the assumption that El Niño impacted

the reproduction and transport of V. cholerae off the coast of Peru to initiate cholera
transmission from October 1990 to January 1991. Although plausible, and suggested
by studies in the previous section, the timing of El Niño with the emergence of cholera
in Peru remains debatable because researchers as well as Peruvian fisheries could not
agree on the number and timing of events during the first half of the 1990s (WMO 1999;
Glantz 2001: 21). This aspect raises an important question about a fundamental

48

component of the El Niño explanation: was an El Niño present before and during the
onset of the epidemic?
The first half of the 1990s was considered an “extraordinary” time for El Niños
(Glantz 2001: 21). According to Trenberth and Hoar (1996), 1990 to 1995 was the
longest El Niño of the twentieth century, estimated to occur once every 1500 to 3000
years. McPhaden (1993) and Kessler and McPhaden (1995) identified a prolonged El
Niño, but reported that it lasted from September 1991 to July 1993. Goddard and
Graham (1997) suggested that 3 events took place in: (1) March 1991 to June 1992; (2)
1993 (approximately February-March to October); and (3) 1994 (June to November).
There were also differences between regions. As Glantz (2001) notes, the Australians
saw it as a 5-year El Niño, while the Peruvians believed it was 3 consecutive events
within a 5-year period (99-100). Peruvian fishermen questioned the Australian view of a
five year El Niño because they noted that they were catching near-record-setting
anchovy landings in 1991, highly unlikely during an El Niño episode (Flores 1998). The
difference in views suggests that perhaps each country used an El Niño definition suited
to their own needs or region.
For example, Australia, like Peru, is a region well known for ENSO-related
teleconnections and impacts on its society. There, SST anomalies in the Niño 4 Region
are important to identify and characterize ENSO’s development in the equatorial Pacific
(Glantz 2000). Equally important for scientists is the use of the SOI in Australia; El Niños
are identified when values are consistently negative for several months (BOM 2005; The
Long Paddock 2010). From 1990 to 1995, the SOI values suggested to Australians and

49

NOAA—at the time—that El Niño-like conditions were present (reported by NOAA in
June 1993). Furthermore, there were reports of drought during this time in Austrialia, a
climate teleconnection often associated with El Niño (Glantz 2001: 99). In contrast,
Peruvian scientists may consider SST anomalies in the Niño 1+2 Region as better
estimates of El Niño in the equatorial Pacific (Glantz 2000; Lagos et al. 2008). By
observing this region, it appeared that several events may have developed, an
observation reported by Peruvian fishermen (Flores 1998).
If one compared the onset of the cholera epidemic (January 1991) with SST
anomalies in these two Niño regions, different associations may arise according to the
region and threshold chosen (Trenberth 1997). This is demonstrated in Figure 3.1,
which shows that using the Niño 4 region might satisfy the El Niño-cholera hypothesis
since SST anomalies in that area of the equatorial Pacific were substantially greater
relative to anomalies in the Niño 1+2 region. Whereas using the Niño 1+2 region might
suggest that El Niño began after the initial cholera outbreak in January 1991. That is,
assuming V. cholerae was present in coastal waters.

50

100000

3

90000

2.5
2

70000

1.5

60000
1
50000
0.5
40000
0

30000

-0.5

20000

-1

10000
0
1990

Monthly SSTA (degC)

Monthly Cholera cases in Peru

80000

-1.5
1991

1992
Cholera

1993

1994

Niño 1+2

1995

Niño 4

Figure 3.1 Monthly Cholera Cases in Peru and Niño 1+2 and Niño 4 Sea Surface Temperature Anomaly (SSTA) from 1990 to 1995.
Source: MINSA 2005b; NOAA 2009.
51

Evidence of El Niño
While defining El Niño in Peru is important, it is also necessary to identify El
Niño-related impacts on ecosystems and society in order to assess whether El Niño
contributed to the onset of the epidemic in 1990/91. However, identifying impacts in
the early 1990s is challenging since reports of El Niños in Peru during this time are not
well documented in scientific studies (Glantz 1998; personal communication August
2009) or even in popular media (Zapata- Velasco and Broad 2001: 191). Still, some
evidence in the literature suggests there were impacts on marine ecosystems and local
climate in Peru.
According to the Peruvian Marine Institute, 1991 was described as a year with
cold coastal waters and good for anchovy catch (Pizarro 1999). It implies that average
SST conditions in the eastern equatorial Pacific were present and that upwelling and
marine biological productivity was normal too. In other areas of the eastern Pacific
Ocean, ecological impacts (associated with El Niño) were reported off the coast of Costa
Rica (e.g., coral bleaching and mortality) in March to April 1992 (Jimenez and Cortes
2001), in the Galapagos (e.g., penguin populations declined associated with lower food
supply) from 1991 to 1993 (Hernan Vargas et al. 2006), and off the coast of Chile (e.g.,
changes in planktic fauna species) from November 1991 to March 1992 (Marchant et al.
1998).
There were also reports by NOAA’s ENSO Advisory (1992) that an El Niño was in
development during the SH summer of 1992; and that rainfall teleconnections were
reported in northern coastal Peru. Although speculative, this evidence along with

52

biological changes in marine ecosystems in 1991-92 suggests an El Niño connection well
after the cholera epidemic was underway.

3.3

The La Niña Factor
A dimension that has not been explored in relation to cholera in Peru is La Niña,

the cold phase of ENSO. La Niñas enhance cold SST conditions in the central and eastern
equatorial Pacific Ocean (Philander 1985). In relation to cholera, La Niña periods
appeared to decrease risk at a national level. Figure 3.2 shows how cholera cases
subsided significantly before and after the 1997-98 El Niño, coincidently with the
development of La Niñas in 1995/96 and 1998-2000 (NOAA 1996; Halpert and Bell 1997;
Trenberth 1997; Bell et al. 2000; Lawrimore et al. 2001). Concurrently, La Niñas may
also have increased cholera risk depending on the location in Peru as well as regional
and local conditions. For example, in February 1999 health ministries across Peru were
on alert because of rain-related disasters in the country’s south and central highland
and jungle regions. Ten Peruvian departments (Junin, Lima, Loreto, Cusco, Madre de
Dios, Apurimac, Ica, Huancavelica, Arequipa, and Ancash) were affected by floods and
were in need of assistance with food, water and sanitation (MINSA 1999b). The
implications for cholera transmission, however, were unclear because there appeared to
be a greater concern (in reports discussing La Niña) for respiratory disease, dermatitis
and latigazo (insect-borne disease that causes lashes on the skin) during that time
(MINSA 1999a). Nonetheless, segments of the population were experiencing climaterelated impacts on human health. La Niña conditions along with those in the tropical

53

Atlantic were blamed (MINSA 1999a). Examining these associations further is important
because La Niñas and their societal impacts in general are less studied in ENSO research
(Glantz 2002b: 8).

54

3

12000

2

10000
1
8000
0
6000
-1

4000
2000

-2

0

Monthly SST anomaly (degC)

4

14000
Monthly Cholera cases in Peru

16000

-3

1995

1996

1997

1998
Cholera

1999

2000

Niño 3.4

El Niños (red) and La Niñas (blue) events were defined according to the current NOAA definition (2010).
Figure 3.2 Cholera Cases in Peru and El Niños and La Niñas from 1995 to 2000.
Source: MINSA 2005b; NOAA 2009.
55

3.4

Geography of El Niño Impacts
Although a temperature link with cholera is plausible because of its potential

impact on the reproduction of the vibrio and marine hosts, whether this effect on
cholera was observed across Peru is uncertain. First, previous studies in Peru were
mainly based on observations in Lima, located on the central coast, where climate is
strongly modulated by maritime effects; it is also closer to the source of V. cholerae
(assuming it was present on the coast). Therefore, considering the country’s diverse
natural regions (e.g., coast, highlands, and jungle) and the geography of teleconnections
discussed earlier, it is uncertain that findings in Lima can be generalized to other areas
of Peru. Figures 3.3 to 3.5 compare plots of monthly cholera cases by year from 1993 to
1998 in three Peruvian departments (each associated with a natural region). Although
Lima (coast) and Cajamarca (mountain) appear to show a general trend (cholera peaks)
in the summer, there are some differences by years (e.g., 1993 and 1994); Loreto’s
(jungle) seasonal pattern was quite different in comparison to Lima and Cajamarca;
overall the cholera pattern in Loreto was lagged in time (peaks in the winter) and less
well-defined. Second, studies were also primarily focused on the time period after
1993. To date, the entire decade has never been examined including the time period of
the epidemic onset in 1990/91, which is the basis of the El Niño-cholera hypothesis
(Salazar-Lindo 2008). Therefore, investigating the time associations between El Niño
and its potential impacts on cholera throughout the 1990s will be necessary to fully
comprehend cholera patterns in Peru.

56

Cholera Cases in Lima

4500
4250
4000
3750
3500
3250
3000
2750
2500
2250
2000
1750
1500
1250
1000
750
500
250
0
J

F

M
1993

A

M
1994

J
1995

J

A
1996

Figure 3.3 Cholera Cases from 1993 to 1998 in Lima (coast).
Source: MINSA 2005b.
57

S
1997

O
1998

N

D

1200

Cholera Cases in Cajamarca

1000

800

600

400

200

0
J

F

M
1993

A

M
1994

J
1995

J

A
1996

Figure 3.4 Cholera Cases from 1993 to 1998 in Cajamarca (mountain).
Source: MINSA 2005b.

58

S
1997

O
1998

N

D

1200
1100
1000
Cholera Cases in Loreto

900
800
700
600
500
400
300
200
100
0
J

F

M
1993

A

M
1994

J
1995

J

A
1996

Figure 3.5 Cholera Cases from 1993 to 1998 in Loreto (jungle).
Source: MINSA 2005b.
59

S
1997

O
1998

N

D

3.5

Rainfall Extremes
Rainfall extremes is an important El Niño teleconnection in Peru (Lagos et al.

2008). In the late 1990s there were reports in Peru of diarrheal disease and cholera risk
attributed to heavy rains and climate-related disasters (MINSA 1997; 1998a; 1998d;
1999a). During the 1997-98 El Niño, rains in northern coastal Peru began in December
1997 and lasted until April 1998. Record-breaking river levels, flooding and landslides
were reported. Subsequently, these series of events contributed to damaged
infrastructure (e.g., homes and sewer systems) and human exposure not only to cholera
(MINSA 1998f), but other food and waterborne diseases, as well as skin infections and
conjunctivitis (PAHO 1998b; SENAMHI 2004). In other regions, such as the northern
highlands landslides destroyed homes and killed livestock (UNDRO 1998). During these
events public health responses were limited because of downed bridges and roads. In
addition there were reports of population displacement and refugees, which may have
potentially increased cholera exposure (PAHO 1998b). Although elevated temperatures
may have contributed to V. cholerae reproduction and subsequent cholera incidence in
Lima, contaminated water supplies via rains and floods may have been a more
important factor in other places in Peru.
Alternatively, rainfall deficit may have played a role in cholera transmission. For
example, in Loreto, which was discussed earlier, low river levels were associated with
cholera risk in places of poor water and sanitation infrastructure (MINSA 1995b; 1998b).
Under these circumstances, exposure to cholera may have occurred because persons

60

were limited to marginal water supplies (e.g., polluted river water) for drinking, cooking
and hygienic purposes.

3.6

Social Dimensions
Glantz (2005) in describing the factors that resulted in the outcomes of

Hurricane Katrina stated, “The combination of psychological, financial, and political
factors—together with a direct hurricane hit, the breakdown of the levees, and the
subsequent cascade of disasters underscored the vulnerabilities of the poor, the elderly,
children, and racial minorities.” This statement illuminates the notion that underlying
climate-related hazards are often, social, economic and political factors that contribute
to disasters, such as famines, population displacement, and epidemics (Sen 1981; Watts
and Bohle 1993; Glantz 2003: 253; Davis 2001: 279). Although the 1990s were
associated with several El Niños, it was also a time during which Peru was addressing
complex emergencies—unrelated to El Niño (UNDRO 1990a; 1990b), a multitude of
socioeconomic and infrastructure problems, as well as civil unrest (e.g., terrorism) (Nash
1991; Youngers 2000). This was the setting of the cholera epidemic in the early 1990s.
Preceding the epidemic, the Peruvian government was addressing two
humanitarian crises. The first was an earthquake in northeast Peru, which affected
70,000 and injured 1500 people in May 1990 (UNDRO 1990b). The second was an
agricultural state of emergency declared across 13 departments (out of 25) in June
1990. Reportedly over 2 million subsistence farmers in the highlands were gravely
affected by an ongoing drought and cold extremes experienced in 1989; and food and
61

water supplies were in decline (UNDRO 1990a). When cholera emerged in 1991,
approximately 45.0% of the country’s population did not have access to clean water and
59.0% were without sanitation services. In rural areas, conditions were much worse;
there, less than one-third of the population had access to clean water and other basic
services (PAHO 1991a). In addition, Peruvians were affected by economic reforms
implemented by the then elected President Alberto Fujimori (Nash 1991; Youngers
2000). As a result, public infrastructure and services were reduced including those in
the health sector (Cueto 2003). Ultimately, these events may have aggravated the preexisting living conditions of a population, who lacked immunity to cholera, increasing
their exposure and susceptibility to disease transmission.

3.7

Summary
In this chapter, I highlighted several factors of ENSO, which are important to

understand cholera transmission in Peru and warrant further investigation. These
potential avenues of research are: El Niño’s association with cholera may vary by the
definition applied; there is evidence that El Niño’s role might have been critical only
after the onset of the outbreak; La Niña may have contributed to a reduction in cholera
incidence; there is a need for studies in other areas of Peru given that El Niño-cholera
studies were limited to Lima; rainfall extremes may have been an important factor in
transmission; and social factors of vulnerability may have enhanced the severity of
impacts on cholera transmission. Each of these factors will be considered in this
dissertation study.

62

3.8

Research Questions, Hypotheses and Objectives
As stated earlier, the goal of my dissertation research is to reconstruct the

temporal and spatial associations among ENSO, social vulnerability and cholera
incidence in Piura, Peru from 1991 to 2001 in order to better understand El Niño’s
impact on the cholera epidemic in Peru. My overarching research questions are: (1)
What was the impact of ENSO on cholera incidence in Piura; and (2) How did social
vulnerability influence this relationship?

The research hypotheses are:


Hypothesis 1 (H1): There was a temporal association between ENSO, climate and
cholera cases in Piura in the 1990s. Furthermore, these associations were
stronger after 1992 compared to the onset of the epidemic in 1991.



Hypothesis 2 (H2): The spatial variability of the ENSO-climate-cholera
associations in Piura in 1997-98 will be explained by the spatial distribution of
social vulnerability. Moreover, the level of social vulnerability within districts in
Piura will either antagonize or buffer the effects of ENSO and climate on cholera
incidence.

The following objectives will address my research questions and hypotheses:
1) Develop a conceptual framework that characterizes the potential ecological
pathways and vulnerability conditions of cholera transmission in Piura;

63

2) Characterize the temporal associations between ENSO, climate and cholera
cases in Piura from 1991 to 2001;
3) Construct a social vulnerability index (SVI) that characterizes social
vulnerability to cholera in Piura in 1997-98;
4) Characterize the spatial and temporal associations between ENSO, climate
and cholera incidence by district in Piura for 1997-98 and estimate the degree to
which social vulnerability influenced this relationship;
5) Reflect on the findings from this study using a climate and development ethics
perspective in order to better formulate recommendations.

64

4. CHAPTER 4: RESEARCH DESIGN
In this chapter I present the research design of my dissertation project including
the research approach, theoretical concepts and conceptual framework that guide and
inform this study. I also provide a description of the study area and population. I
conclude with the data and methods I employed to complete each research objective .

4.1

Approach, Concepts and Framework
In this dissertation research, I used a climate affairs approach (Glantz 2003) to

develop a conceptual framework that integrates ENSO science and knowledge about
cholera impacts and interactions with theories of disease ecology and vulnerability from
the subfields of health and medical geography (Mayer 2000; Meade and Erickson 2005)
and human-environment geography (Cutter 2003; Turner et al. 2003; Zimmerer and
Bassett 2003) to explain cholera transmission. From an ethical viewpoint, this study is
guided by an ethics of climate and development, two emerging subfields in philosophy
(See section 6.5 for the ethical geographies of this study). The framework is further
informed by cholera research (Stock 1991; Cueto 2003; Nelson et al. 2009); climatecholera studies (refer to Chapter 2); and climate and society research (Glantz 2001;
Caviedes 2001; CCB 2010). Furthermore, my dissertation fieldwork conducted in the
summers of 2008 and 2009 in Lima and Piura, Peru was central to this project because it
grounded the conceptual framework and informed my research questions and
hypotheses.

65

4.1.1 Climate Affairs Approach
Climate affairs is a holistic approach to understanding the many facets of climate
(e.g., averages, extremes, variability and change) and how ecosystems and societies
interact with climate phenomena at multiple scales (Glantz 2003). The concept
emphasizes the importance and necessity of the contributions of physical, biological,
social sciences, and humanities to the understanding of the impacts of air and sea
interactions in the equatorial Pacific Ocean and worldwide (CCB 2010). It is principally
grounded in ENSO/climate science and knowledge, ecosystem-societal impacts and
vulnerability, and ethics and equity research and was therefore, an ideal lens by which
to reexamine ENSO’s link with the cholera epidemic in Peru and understand its potential
impact(s) on cholera transmission in Piura, Peru in the 1990s. The concept evolved from
the climate-society oriented research and collaborative activities of Dr. Michael H.
Glantz, which began in 1974 at the Environmental and Societal Impacts Group (ESIG), a
former program of the National Center for Atmospheric Research (NCAR), and which
later came to fruition as a research, training and education initiative in 2003. A large
segment of climate affairs-related activities and publications were centered on ENSO
including an El Niño Affairs program for Latin American countries
(http://ccb.colorado.edu/enos/). Currently, it continues to be implemented through the
Consortium for Capacity Building (CCB) at the University of Colorado, Boulder.
Figure 4.1 illustrates how a climate affairs approach, geographic concepts, ethics
and fieldwork were integrated into the research design of this study. A description of

66

the approach, theoretical concepts and conceptual framework developed to explain
cholera transmission in Piura will follow.

CLIMATE AFFAIRS APPROACH
↓
SCIENCE AND KNOWLEDGE
ENSO dynamics & society
Cholera transmission cycle
INTERACTIONS AND IMPACTS
ENSO-cholera
Local-climate-cholera
ENSO-local climate
Vulnerability-cholera
Vulnerability-ENSO-cholera
Vulnerability-local climate-cholera

↔

GEOGRAPHIC CONCEPTS
Disease ecology
(Medical/health tradition)
Vulnerability
(Human-environment tradition)

↔

ETHICS AND EQUITY
Differential impacts
↑
FIELDWORK

ETHICS CONCEPTS
Climate Ethics
Development Ethics
↑
THEORETICAL CONCEPTS

Figure 4.1 Climate Affairs as an Integrating Approach and Concept.

4.1.2 Theoretical Concepts
Disease Ecology
Disease ecology explains infectious disease transmission through the interactions
among biological agents (pathogens and vectors/reservoirs), population (human hosts)
and environment (Wilson 2003). Transmission occurs when an infectious agent enters
the human host through ingestion, respiration, absorption or sexual contact.
Environmental conditions and changes may contribute to the risk of exposure to
67

infection (e.g., via ecological changes in landscape and aquatic environments).
Infectious disease transmission can also occur through the migration of an infected
person or a biological agent (e.g., mosquito or tsetse fly). Who and how a population
becomes exposed to an agent and disease also varies by preexisting conditions:
population susceptibility, including genetics, previous exposure (immunity) to the agent,
and prior health status. From a medical geographic perspective, the concept of disease
ecology incorporates the notion of ‘place’ and considers transmission using a humanecological model within social, economic, cultural, built, and physical environments
(Meade and Erickson 2005). ‘Place’ is a key theme in medical and health geographies
(Kearns and Moon 2002). It is critical to the understanding of cholera ecology and
transmission in this dissertation because it emphasizes understanding the
characteristics and contextual setting of a ‘place’ in which human-cholera interactions
may have occurred and subsequent transmission may have taken place. Importantly,
‘place’ refers to the composition of a society (i.e., its people) and the context of the
society (Macintyre et al. 2002), e.g., social determinants as stated earlier (WHO 2008).
Vulnerability
Audy (1965) defined health as a process of adaptability or ability to defend or
recover from afflictions or hazards. Although he was referring to the susceptibility of a
population, his definition suggested that a ‘place’ is not only an environment where a
population becomes exposed to a biological hazard; it is also a setting where the
susceptibility of a population to transmission or a health hazard is influenced by a
population’s capacity to cope and respond when exposed. While culture is an important

68

factor that influences human behavior and response to disease (Roundy 1979), it is also
important to consider the underlying conditions and processes that make people, places
and societies susceptible to harm from an external hazard (Wisner et al. 2004: 167-200).
From this orientation, disease ecology and susceptibility of a population are linked to
the geographic theoretical concept of vulnerability. Like disease ecology, ‘place’ is an
important tenet of vulnerability research (Cutter 2003). Patterns of vulnerability vary by
the processes and hazards that exist in a place and region (Hewitt 1997). The
“geography of risk” is differential within and between societies (Hewitt 1997: 28), where
some subpopulations are more susceptible to harm and cope less than others (Cutter
and Finch 2008). To date multiple conceptions of vulnerability have emerged;
importantly, risk-hazards (Kates 1971; Burton et al. 1993), disasters (Hewitt 1997; Lewis
1999), and human ecology (Watts and Bohle 1993). The work of Glantz (1981; 2001)
and Caviedes (1973; 1982; 1984; 1985) have also contributed to this concept,
particularly with respect to El Niño impacts. More recently, vulnerability has been used
in climate-society research (Vogel 2001; Liverman et al. 2006; Adger 2006) and coupled
human-environment systems-thinking (Turner et al. 2003).
In the past, geographers have utilized the concepts of disease ecology and
vulnerability to understand global to local interactions that influence emerging
infectious diseases (e.g., diarrheal disease and HIV/AIDS in Africa). For example, in
Mozambique, Collins (1998) investigated diarrheal disease in coastal areas in relation to
population displacement and international development policies and found that
transmission was indirectly associated with global development programs influenced by

69

multi-lateral institutions. Structural changes led to impacts via land-use changes on the
local environment, which contributed to heightened disease incidence (Collins 1998: xiiixiv). Similarly, in sub-Saharan Africa, Mayer (2005) demonstrated that HIV/AIDS was
associated with SAPs, which led to reduced health services and treatment (Mayer 2005).
In Ghana, Oppong (1998) attributed the diffusion of HIV in subpopulations to population
vulnerability influenced by a national economic crisis that preceded an epidemic
(Oppong 1998). These studies in Africa illustrate that although HIV and diarrheal
disease (including cholera) are distinctly different infectious diseases, they share a
commonality: both are associated with a population’s vulnerability indirectly influenced
by global and local development.

4.1.3 Conceptual Framework
In this study, the particular conception of vulnerability that I integrated with
disease ecology was social vulnerability. Social vulnerability is characterized as social
factors and processes that contribute to a population’s susceptibility to harm in relation
to an environmental hazard (Cutter 2006). According to Wisner et al. (2004), these
factors are the root causes of vulnerability in society. Social vulnerability and the
concept of vulnerability are often associated with deprivation and disparities between
subpopulations (Cutter et al. 2003). Measuring social vulnerability is challenging, but
factors such as urban/rural, basic needs infrastructure (water, housing and sanitation),
and level of education can be utilized in geostatistical models to understand the
temporal-spatial relationships of such phenomenon (Cutter and Finch 2008). Using

70

social vulnerability, the human ecology of cholera transmission can be examined in
three dimensions: 1) exposure pathways to cholera transmission, 2) population
sensitivity or resistivity to cholera, and 3) capacities to cope with/respond to or recover
from cholera (i.e., resilience) (Turner et al. 2003). As such, cholera can be viewed as
both a hazard to society and an outcome from an external hazard/stressor (Wisner et al.
2003), e.g., associated with basic needs deprivation or extreme climate impacts. In this
study, the disease ecology of cholera and social vulnerability to cholera were situated
within a broader framework of vulnerability adapted from Turner et al. (2003).
Figure 4.2 is the conceptual framework adapted from Turner et al. (2003) to
situate my study within the climate affairs perspective. It illustrates the ecological
interactions and conditions that contributed to cholera transmission in the subregion of
Piura at global, national and local levels. The elements highlighted in white boxes
represent the factors that are empirically modeled in this study. Disease-ecological
vulnerability is composed of conditions and processes that are human-environment
coupled (Turner et al. 2003) and social-environmentally produced (Zimmerer and
Bassett 2003: 3). Furthermore, interactions and impacts depend on the spatial and
temporal scale(s) of analysis (Vincent 2007). According to the framework, cholera
incidence emerges from “place”-based interactions among population (e.g.,
demographics and health), local climate teleconnections, social environment (e.g., social
organization, culture, politics, microeconomics), infrastructure environment (roads,
bridges, health system infrastructure), and physical environment (land, weather, and
water). Pathogens and vectors are also included (e.g., V. cholerae and other disease

71

agents that persist in Piura). ENSO is an environmental factor that is cross-scalar with
global, national and local impacts. It influences local climate teleconnections, which has
associated impacts on human-pathogen-environment interactions in Piura.

72

GLOBAL

Ocean, Atmosphere, Land Influences
Outside Piura
PERU
PIURA

Social

Population

Local Climate

ENSO

Pathogens/
Vectors

Social
Vulnerability

Coping/
Responses

Coping/
Responses
Outside Piura

Cholera
Physical

Infrastructure

Social, Political, Economic, Development
Influences Outside Piura
Influences inside Piura
(included in models)
Influences inside Piura

Cross-scalar Influences
(included in models)
Cross-scalar Influences

Figure 4.2 Conceptual Framework adapted from Turner et al. (2003).

73

Influences outside Piura
Piura

Figure 4.3 shows ENSO impacts on the cholera transmission cycle in Piura. The
impact of ENSO can occur through direct and indirect impacts on V. cholerae and its
reservoirs in the ocean, coast and inland environments. ENSO can directly impact V.
cholerae and its reservoirs in the ocean through ocean-atmosphere interactions (e.g.,
changing marine conditions or coastal intrusion via storm surges). Through local climate
teleconnections, ENSO can indirectly influence the reproduction of V. cholerae in
aquatic inland environments (e.g., temperature changes). Local climate can also
influence the transport of V. cholerae into contact with humans via rainfall and flooding.
Even if V. cholerae are not present, humans can introduce the bacteria into the local
environment by migration of infected persons and through contaminated water from a
passing ship’s ballast. Potential exposure and transmission occurs when a susceptible
population comes into contact with V. cholerae by ingestion of contaminated water or
food. Preexisting conditions (e.g., genetic susceptibility, immunity, health and
nutritional status) influence whether a person becomes symptomatically or
asymptomatically infected. Cholera can potentially spread when an infected person
defecates and introduces V. cholerae into a local aquatic environment (i.e., municipal
water supply or river water). Cholera can also spread if the water or food becomes
contaminated within a household by an infected person. When local climate conditions
become favorable for the proliferation of V. cholerae, the amplification of transmission
by human-feacal spread becomes a potential pathway in which exposure and
transmission increase among a population.

74

ENSO Impacts

Local climate
(Persistence, reproduction and
transport of V. cholerae)

Local climate
(Persistence, reproduction and
transport of V. cholerae)
Marine reservoir and V. cholerae
Zooplankton, copepods, other
crustaceans
Phytoplankton, aquatic plants
Vibrios living marine environment
Environmental conditions
(Persistence, reproduction and
transport of V. cholerae)

Exposure and transmission routes
Contaminated water and food
Inland aquatic reservoir
V. cholerae living in inland aquatic
environments
Household spread
Human migration
(Amplification by fecal-oral spread)
Human migration
Susceptible human population
Pre-existing conditions
Symptomatic
Asymptomatic

OCEAN/COAST

COAST/INLAND

Figure 4.3 ENSO Impacts on the Cholera Transmission Cycle.
Adapted from Lipp et al. (2002) and Nelson et al. (2009).

75

Social vulnerability
Water and sanitation
infrastructure
Urban
Rural

However, ENSO and local climate impacts on cholera transmission are contingent
on the social vulnerability to cholera in Piura. Cholera transmission may either be
enhanced or buffered depending on the level of social vulnerability within Piura (e.g.,
access to water and sanitation infrastructure and living in urban versus rural areas).
Once a person is infected the ability to recover depends on the subsequent
coping/responses (e.g., local actions, access to healthcare, and public health responses)
in Piura and outside Piura (e.g., Peruvian civil defense, multi-lateral aid and assistance,
and NGOs). The framework also recognizes that cholera transmission and vulnerability
in Piura may have also been influenced by a broader context of national and global
pressures (Oppong 1998; Farmer 1999; Mayer 2000; Wisner et al. 2004: 184-185), such
as macroeconomic policies (e.g., associated with SAPs), terrorism (e.g., Shining Path
guerrillas), war (e.g., Ecuador border conflict), and an energy crisis (e.g., oil supply and
demand fluctuations related to the first Gulf War) (PAHO 1991a; Nash 1991; El Tiempo
1991b; Youngers 2000; Cueto 2003). As such, my conceptual framework acknowledges
political ecology (e.g., Zimmerer and Bassett [2003]), but also recognizes that examining
the links between subregion and district-level cholera incidence in Piura and global
development (and other large-scale social forces) goes beyond the scope of this study.

4.2

Study Area and Population
The study area is the health subregion of Piura in the Department of Piura,

located on the northern coast of Peru (approximately 900 km from Lima). The
2

department comprises of an area of 20,238 km with two well-defined land features,
76

including a low-lying savannah (3-meter elevation) on the coast and mountainous
forests (2,709- meter elevation) in the east. On the coast the climate is semi-arid and in
the mountains it is subtropical. The climate is dominated by the Humboldt Current,
conditions in the equatorial Pacific Ocean basin and the Andes Mountains, which lay in
the east. It has one important river, the River Piura, which is seasonal (Institute of
National Statistics and Information [INEI] 2000).
Figure 4.4 and 4.5 show monthly average temperatures and rainfall totals at the
Miraflores meteorological station in the capital city of Piura.

12

The average annual

temperature in Piura is 24.0°C with an average annual maximum and minimum of
31.0°C and 19.0°C. Average temperatures do not vary so much compared to rainfall
total which is highly seasonal (December to May). Piura typically receives less than 50
millimeters (mm) of rainfall annually, except during El Niño years when rainfall has been
recorded to be over 25 times the average total (Woodman 1998).

12

I used the capitol city average because it is widely used by officials in Piura and
represents the dataset which I am employing in this study.
77

Figure 4.4 Monthly Average Temperature (°C) at Miraflores station, Piura for 1971 to 2000.
Source: UDEP 2008
78

Figure 4.5 Monthly Average Rainfall (mm) at Miraflores station, Piura for 1971 to 2000.
Source: UDEP 2008.
79

Figure 4.6 is a district-level map of the subregion of Piura. It is one of two health
subregions. Piura consists of thirty-three of sixty-four districts in the Department of
Piura. In 1998, it had an estimated 847,257 habitants (56.0% of the Department’s total
population) representing 42 habitants per km2 (INEI 2000). Fifty-one percent of
residents lived in urban areas near the coast. The main sectors of employment were
agriculture, fisheries and mining (INEI 1996). Forty-eight percent of the population was
dependent (i.e., dependency ratio = 0.5) with most of that dependency attributed to
children under the age of 15 years (INEI 1993).

80

7
0

6

17

1

8

25 19
21 24 26
18

5

3
31 28
4
32 29
30

23

13

9

10

20
14

22

2

11

5°0'0"S

80°0'0"W

5°0'0"S

81°0'0"W

16 15

6°0'0"S

6°0'0"S

12

27

0

25

81°0'0"W

District
ID
PIURA
0
CASTILLA
1
CATACAOS
2
CURA MORI
3
EL TALLAN
4
LA ARENA
5
LA UNION
6
FRIAS
7
PACAIPAMPA
8
HUANCABAMBA 9
CANCHAQUE
10

50 Kilometers
80°0'0"W

District
EL CARMEN DE LA FRONTERA
HUARMACA
LALAQUIZ
SAN MIGUEL DE EL FAIQUE
SONDOR
SONDORILLO
CHULUCANAS
BUENOS AIRES
CHALACO
LA MATANZA
MORROPON

Figure 4.6 Map of the Subregion of Piura by District (ID).
81

ID
11
12
13
14
15
16
17
18
19
20
21

District
SALITRAL1
SAN JUAN DE BIGOTE
SANTA CATALINA DE MOSSA
SANTO DOMINGO
YAMANGO
SECHURA
BELLAVISTA DE LA UNION
BERNAL
CRISTO NOS VALGA
VICE
RINCONADA LLICUAR

ID
22
23
24
25
26
27
28
29
30
31
32

Development indicators in Piura suggest that basic needs are unmet for some
segments of the population. Illiteracy affected 25.0% of population (i.e., adults 15 years
+) in 1993. Importantly, water and sanitation infrastructure was not accessible to 27.0%
and 71.0% of the population (INEI 1993). The impact of basic needs unmet on
subpopulations was evident in the subregion’s health reports. There were high rates of
infant mortality (17.0 per 1000 live births) and maternal mortality (89.0 per 1000).
Moreover, 31.0% of diarrheal disease cases were associated with dehydration (INEI
1997).
Piura’s health capacity in 1997 was 57 laboratories and 199 health centers
including 3 major hospitals (INEI 2001). It is part of Peru’s health system that
encompassed (at the time) three organizations: Ministry of Health (87.0% of hospital
infrastructure and 87% of primary healthcare), Peruvian Institute of Social Security
(ESSALUD), and the private sector, which mainly provides ambulatory services. The
latter is a sector for wealthy Peruvians. Although the Ministry of Health was responsible
for most healthcare services to Peruvians, only a small percentage used those services
(e.g., 13.0% of those contributing to ESSALUD via employment and 9.8% of those with
private insurance). In many rural areas, communities did not have access to healthcare;
health care was mainly available in cities (Department for International Development
Health Systems [DFID] 1999). As part of a health reform in 1995, primary health care
was extended by forming ‘Local Committees for Health Administration’ (CLAS), which is
administered by local communities and funded by the treasury to expand healthcare to
underserved communities (DFID 1999). Despite this effort, which was part of a larger

82

national initiative called the Basic Health-for-All Program during the mid-1990s, it was
reported that the poorest (by poverty rates) sections of the Peru had the least resources
and human capital (PAHO 1998).

4.3

Data and Methods
I primarily used secondary data collected from various institutions in Peru in

Piura and Lima, online sources and various literatures (documents and sources are
described in Objective 1). A description of these data per objective will follow.

4.3.1 Objective 1: Develop a conceptual framework that characterizes the potential
ecological pathways and conditions of cholera vulnerability and transmission in Piura

Objective 1 Data and Methods
In the first objective, I developed a conceptual framework to explain cholera
vulnerability and transmission in Piura, Peru. The framework was presented in detail in
the previous sections. Developing the framework was a reflexive process guided by a
climate affairs approach and grounded in theoretical concepts of disease ecology and
vulnerability and fieldwork in the cities of Lima and Piura, Peru where I collected data,
information and documents in the Spanish language. In order to collect these data I
obtained an IRB (#X08-724 with exemption status) from Michigan State University in
order to consult with and document anecdotes from officials at a number of Peruvian
institutions to learn more about the cholera outbreaks in the 1990s, El Niño impacts,

83

and the type and quality of the data I collected. A list of the types of questions I asked
officials and a copy of a consent form are found in Appendix 2. I also took many field
notes where I documented my observations and also reflected on them. These
experiences and exchanges with the people in Piura provided me with ‘revealing points’
that helped me to connect important pieces of information and refine my framework
(Emerson et al. 1995: 144-167). These revelations were important because they helped
me ground my research questions and hypotheses. Thus, I was better informed about
Piura and how cholera transmission occurred and El Niño’s impact. I also collected
archival materials (e.g., newspapers, 1991 to 1993), newsletters, journal articles, and
health bulletins from various sources, including the library at University of Piura (UDEP),
University of the Pacific in Lima, INEI in Piura, the Ministry of Health in Piura,
International Institute of Nutrition in Lima, Institute of Peruvian Studies, Department of
Fisheries at the National University of Agriculture – La Molina in Lima, Center of
Research and Advocacy for Farmers in Piura (CIPCA), Applied Geography Research
Center at the Pontifical Catholic University of Peru, PAHO, and GTZ, a German nongovernmental organization in Piura. In addition I visited and used archives and
materials at the Consortium for Capacity Building (CCB) and the National Center for
Atmospheric Research (NCAR) in Boulder, Colorado. I consulted closely with Dr.
Michael H. Glantz about the El Niño materials I collected during fieldwork, and I gained
anecdotal knowledge from him about ENSO and society during the 1990s.

84

4.3.2 Objective 2: Characterize the temporal associations between ENSO, climate
and cholera cases in Piura from 1991 to 2001

Objective 2 Data
a) Cholera
I collected data on cholera cases for Piura during fieldwork in Peru from three
sources:
(1) The Department of Epidemiology at the Ministry of Health in Lima, Peru –
These data were provided in a spread sheet and consisted of weekly reports
for the time period (1991 to 2002).
(2) The Department of Epidemiology at the Ministry of Health and the
International Institute of Nutrition in Lima, Peru – Weekly Epidemiology
Bulletins (hard copies and disk) for the time period (1991 to 2004).
(3) The Ministry of Health for the Subregion Piura in Piura, Peru (MINSA Piura).
These data were provided in a spread sheet and consisted of weekly reports
for the time period (1991 to 2006).
These data included both laboratory confirmed cases, suspected (clinically
confirmed) cases and laboratory-negative cases. It was important to determine if the
data collected from these three sources were consistent so that I could select the most
‘reasonable’ dataset to use in my research. I found that the cholera data among these
sources were generally in agreement from 1991 to 1998. After 1998, cholera data
among these sources were not in agreement. In summary, I decided to use the cholera
data from source (3) for the time period 1991 to 2001. The total number of cases in the

85

Piura dataset that I used in this research was (n= 38,040). These weekly data were then
aggregated to cholera-case counts per month. Lastly, these time series data were
square-root transformed, a standard procedure used in wavelet analyses.
b) Sea Surface Temperature (SST)
Monthly SST index, mean and standardized anomaly data for 1971 to 2002 were
obtained for the following Niño Regions: Niño 3.4 (5°North-5°South) (120-170°West)
and Niño 1+2 (0-10°South) (90°West-80°West). These data were downloaded from the
Center for Climate Prediction (http://www.cpc.ncep.noaa.gov/data/indices/). Niño
index data are 3-month running averages of SST anomaly (e.g., December, January,
February; January, February, March, etc.). A Niño 1+2 index was not available via the
webpage so an index was calculated using the standardized anomaly data. Both indices
were used to define and identify ENSO events in the first part of this objective. Niño 3.4
was used for comparative purposes because it is a widely used global climate indicator
of ENSO conditions, and correlates strongly with rainfall in northern coastal Peru (Lagos
et al. 2008). I also obtained monthly SST mean and anomaly data for Paita for 1971 to
2001 from the University of Piura (UDEP). Paita is a coastal port located 3 meters above
sea-level and SST data from Paita is used to monitor SST off the coast of the Department
of Piura ([05°04’ 57”South] [81°06’ 57”West]). Sea surface temperature anomalies
(SSTA) data were based on the period 1971 to 2000. SSTA data were used to assess
patterns and associations between climate and cholera. In this study the Niño SSTA
parameters represent global SST conditions in the central and eastern equatorial Pacific
and the Paita parameter represents local SST conditions off the coast of Piura, Peru.

86

c) Temperature and Rainfall
Monthly temperature data (mean, maximum and minimum values) and rainfall
data (total mm) for the Miraflores station for 1971 to 2001 were obtained from UDEP.
Miraflores is a meteorological station located in the district of Castilla, which shares the
eastern border of the district Piura and is in the subregion of Piura ([05° 10'South]
[80°37'West]; altitude = 30 meters above sea level). A minimal number of values were
missing from the Miraflores datasets so I replaced those values with data from another
meteorological station called CORPAC, also located in Castilla ([05°12'South] [80°36'];
altitude = 49 meters above sea level). For each temperature parameter, standardized
anomalies were calculated by subtracting the monthly mean for the base period (1971
to 2000) from the mean from a particular month and then dividing by the monthly
standard deviation calculated for the base period. This procedure removed the annual
13

cycle. For rainfall, the data were normalized using square-root transformation. In this
study temperature maximum anomaly (TMAXA), temperature mean anomaly
(TMEANA), temperature minimum anomaly (TMINA), and rainfall parameters
represented the local climate conditions in the subregion of Piura.
Table 4.1 is a summary of statistics for cholera cases and the global and local
climate parameters (anomalies) described in this objective. I found that rainfall and
cholera case data were highly skewed even after performing square-root
transformation. In order to address this potential problem, I calculated standardized
13

This procedure is typically used to remove extremes in wavelet studies that examine
climate and disease relationships.
87

anomalies using base periods of 1971 to 2000 and 1991 to 2001. Using this procedure
there was an improvement in the frequency distribution of rainfall and cholera (Figures
4.7 and 4.8).

Table 4.1 Summary of Statistics for Cholera and Global and Local Climate Parameters for
the years 1971 to 2001
Variable

N

Minimum Maximum

Mean

Mean Std.
Error

Std.
Deviation

Cholera Cases*

132

0.0

10591.0

288.181

104.622

1202.0168

NIÑO 1+2 SSTA
(°C)

372

-2.1

4.5

0.011

0.0630

1.2150

NIÑO 3.4 SSTA
(°C)

372

-2.1

2.5

-0.004

0.0479

0.9233

Paita SSTA (°C)

372

-1.5

3.9

-0.001

0.0505

0.9735

TMAXA (°C)

372

-2.8

3.6

-0.032

0.0517

0.9978

TMEANA (°C)

372

-2.2

4.1

0.004

0.0505

0.9746

TMINA (°C)

372

-2.4

4.2

0.002

0.0504

0.9713

Rainfall (mm)

372

0.0

778.4

17.031

4.04415

78.0007

*Cholera data is for the time period 1991 to 2001

88

(a)

(b)

Figure 4.7 Histograms of (a) Rainfall (square-root transformed) (mm) by month; and (b) Rainfall (square-root transformed) Anomaly
(mm) by month.

89

(a)

(b)

Figure 4.8 Histograms of (a) Cholera Cases (square-root transformed) by month and (b) Cholera Cases (square-root transformed)
Anomaly by month.

90

Objective 2 Methods
a) Defining ENSO
In the first part of Objective 2, I adopted the current definition of ENSO used by
NOAA to estimate ENSO events (2010). The purpose of this procedure was to identify
ENSO events per region so that I can interpret and explain the ENSO background in the
results and discussion. El Niño (warm phase) and La Niña (cold phase) events are
defined as periods during which the values of the Niño 3.4 index exceeded a threshold
of +/-0.5 (°C) for at least 5 consecutive seasons. This definition was also applied to the
Niño 1+2 index in order to estimate ENSO events in that Niño region. Impacts on the
northern coast of Peru are of concern, and therefore, Niño 1+2 is important because of
its proximity to the Peruvian Coast and sensitivity to ocean-atmosphere changes in the
eastern and central Pacific Ocean.

14

Niño region indices were compared according to ENSO phase by identifying the
number of events, duration of events by months, percentage of months (number of
phase months/total number of months for the study time period), and timing of events
(beginning and end months). Neutral phase periods (non-El Niño or La Niña months)
were also identified. These estimates were studied in relation to cholera cases.
b) Wavelet Analyses
Wavelet analyses was used to characterize the temporal properties of each time
series (i.e., cholera and climate variables independently) and quantify associations in
14

Although Trenberth (1997) found that Niño 3.4 was closely associated with historical
studies of ENSO, he recommended the use of criteria that “suits” the region.
91

time frequency space between the following times series: (a) Niño 3.4 SSTA and cholera;
(b) Niño 1+2 SSTA and cholera; (c) Paita SSTA and cholera; (d) TMAXA and cholera; (e)
TMEANA and cholera; (f) TMINA and cholera; and (g) Rainfall and cholera. The time
period of analysis between cholera and climate associations is 1990 to 2001. In the
climate time series analysis (univariate), I considered the time period 1971 to 2000. The
longer time period gave me a better estimation of climate temporal patterns. However,
when I discuss cholera in reference to these associations I focused on 1990 to 2001,
which is relevant to the cholera epidemic in Peru.
The wavelet approach was chosen because it is commonly used in geophysical
studies to examine how phenomena (e.g., ENSO and Artic Oscillation) may vary at
localized time scales and at different time intervals (Torrence and Compo 1998; Grinsted
et al. 2004). More recently wavelet analyses have been used to examine non-stationary
associations (i.e., transient temporal properties) between climate and infectious
diseases (Cazelles et al. 2007) including cholera (Koelle et al. 2005; Constantin de Magny
et al. 2007), dengue (Cazelles et al. 2005; Johansson et al. 2009) and Leishmaniasis
(Chaves and Pascual 2006).
This study used three types of wavelet analyses: (a) wavelet transform
(univariate); (b) wavelet coherence (bivariate); and (c) cross-wavelet transform
(bivariate).
(a) Wavelet transform was used to examine the temporal properties of each
times series in order to identify time scales of power (i.e., frequency and also referred to
as periodicity) across different time intervals. Power increases from blue to red in the

92

wavelet power spectrum. Each scale is denoted by years and represents horizontal
slices of the variance across different time intervals (also referred to as bands) (Torrence
and Webster 1999). I identified those areas (i.e., time intervals and scales) with the
greatest power that are statistically significant (95.0% confidence) and outside the cone
of influence (COI), where edge effects at the beginning and end of the times series may
be interpreted but are not influential (Torrence and Compo 1998). The average power
and scales across the local wavelet spectra that are statistically significant (95.0%
confidence) were estimated using a measure called ‘global wavelet power spectrum’
(GWS).
(b) Wavelet coherence was used to estimate localized correlations between two
time series and to identify time scales and intervals where the time series co-varied.
Coherence increases from blue (low) to red (high) correlation. I was looking for areas
(i.e., time intervals and scales) with the greatest coherence that are statistically
significant (95% confidence) and outside the cone of influence (COI). In addition to
estimating correlations, coherence was also used estimate the direction of the
relationship (a range from ‘in phase’ as positive to ‘out of phase’ as negative) and
whether one time series leads or lags the other in this association. The phase of
coherence is indicated by arrows, as follows: north (climate lags by 90° or an estimated
6 months); south (climate leads by 90° or an estimated 6 months); east (climate and
cholera in phase at 0° or an estimated zero months); and west (climate and cholera out
of phase at 180° or an estimated zero months).

93

(c) Cross-wavelet transform was used to assess whether localized time scales
and intervals that co-varied also shared high power or frequency. The cross-wavelet
transform is important because it helps to assess whether the phase (denoted by
arrows) is consistent across scales. If arrows vary or conflict (e.g., in and out or lag and
lead) throughout a scale, it may suggest that the association is coincidental (Grinsted et
al. 2004). Therefore, the phase direction and the time lag or lead was interpreted
carefully and supported with additional methods.
(d) In this study, I also used cross-correlation analysis to support my
interpretations of the phase direction and time association between two times series as
outlined above.
The wavelet analyses were performed in Matlab R2009a (Matlab 2010) using
scripts by Torrence and Compo (1998) and Grinsted et al. (2004). Examples of these
scripts are provided in Appendix 3. Cross-correlation analysis was performed using SPSS
GRADPACK 17.0 (SPSS 2009).
4.3.3 Objective 3: Construct a social vulnerability index (SVI) that characterizes social
vulnerability to cholera in the subregion of Piura in 1997-98

Objective 3 Data
Demographic, socioeconomic, and infrastructure data at the district level (n = 33)
for the subregion of Piura were obtained from the census bureau (INEI) in Piura and
from the INEI webpage (http://www.inei.gob.pe/). These datasets included census and

94

basic needs information for 1993.
recent data were unavailable.

16

15

I chose this year because it is widely used and more

After reviewing the data, I discovered that the basic

needs data were incomplete; therefore I chose to use the census bureau data. This
dataset included 93 variables that were percentages (e.g., percent of population that
had access to potable water via water trucks, etc.). Thirteen variables were selected for
the social vulnerability index (SVI) based on theoretical and empirical associations with
cholera risk and transmission. Tables 4.2 and 4.3 show a summary of statistics for these
variables and correlations with cholera. Overall these variables included information on
education, housing for use as proxy indicators of broader social conditions, water and
sanitation infrastructure and urban because cholera in Peru was characterized as an
urban epidemic (PAHO 1991; Salazar-Lindo et al. 2008).

15

Basic need data in Peru is a composite measure using select census data to
characterize social and infrastructure poverty.
16
District level data in Peru was not collected periodically until 2005.
95

Table 4.2 Description of Variables used in Principal Components Analysis
Variable
ELEM_EDUC
ILLITERATE
IMPROV_HSING
KITCHEN_NONE
KITCHEN_SHARE
KITCHEN_YES
BATHRM_NO
URBAN
WATER_OTHER
WATER_PUMP
WATER_RIVER
WATER_TRUCKS
WATER_WELL

Description
Percentage of population that has only elementary education
Percentage of population that is adults (defined as 15 yrs+) that cannot read or write
Percentage of population that live in improvised housing
Percentage of population without kitchen in household
Percentage of population that shares a kitchen
Percentage of population with exclusive kitchen in household
Percentage of population without bathroom in household
Percentage of population that lives in urban area
Percentage of population whose source of water is listed as other
percentage of population whose source of water is a public water pump
percentage of population whose source of water comes is a local river
percentage of population whose source of water comes from water trucks
Percentage of population whose source of water is a public well

96

Table 4.3 Summary of Statistics for Variables and Pearson’s Correlations with Cholera
Variables
N
Cholera incidence for 199798 (per 1000)
33
ELEM_EDUC (%)
33
ILLITERATE (%)
33
IMPROV_HSING (%)
33
KITCHEN_NONE (%)
33
KITCHEN_SHARE (%)
33
KITCHEN_YES (%)
33
BATHRM_NO (%)
33
URBAN (%)
33
WATER_OTHER (%)
33
WATER_PUMP (%)
33
WATER_RIVER (%)
33
WATER_TRUCKS (%)
33
WATER_WELL (%)
33
N indicates the number of districts.

Mean Std. Dev Correlation with Cholera
1.77
69.12
24.64
1.52
15.85
3.00
81.24
71.55
51.03
5.33
7.33
41.64
4.24
12.73

1.03
10.40
12.02
3.19
7.12
2.24
7.28
21.94
37.20
13.30
7.19
32.95
11.63
10.74

1.00
-0.47
-0.63
0.36
0.19
0.59
-0.38
-0.18
0.71
0.44
0.12
-0.65
0.28
0.06

Objective 3 Methods
a) Principal Components Analysis
The SVI was constructed following a method by Cutter et al. (2003). Variables
were explored for multicolinearity. A correlation matrix of these variables is displayed
in Table 4.4. In order to address multicolinearity, the SVI input variables were reduced
into orthogonal dimensions using principal components analysis (PCA) in SPSS
GRADPACK 17.0 (SPSS 2009). The PCA settings were rotated using varimax and sorted.
Dimensions that accounted for at least 5.0% of the total variance were extracted.
Factors within the dimensions were selected based on a loading value (0.5 or above),
theory, and literature review. For example, although a factor may not have obtained a

97

loading value of 0.5, inclusion of the factor in the SVI was contingent on findings from
theoretical and empirical studies about cholera transmission (discussed in the literature
review). Factor scores for each observation (district) derived from the PCA analysis
were saved. Scores represent how influential a variable is in a pattern by the weight
assigned in the PCA (Rummel 1967). Each dimension was given equal weight for
simplification. Scores for each dimension were then summed and divided by the
number of dimensions. Scores were mapped using ArcGIS version 9.2 (ESRI 2010) to
illustrate the spatial patterns of vulnerability for the SVI and individual dimensions (SVI
factors, hereafter) by districts. District scores for SVI and SVI factors were also mapped
and ranked in ascending order (low vulnerability to high vulnerability) for comparison
purposes.

98

Table 4.4 Correlation Matrix for Variables used in Principal Components Analysis
1

2

3

4

5

6

7

8

9

pw3
pw3
pw4
pw5
pw6
pw7
ed3
k1
k2
k3
urban
how5
bth5
illit_a
1

pw4

pw5

pw6

pw7

ed3

k1

k2

k3

1.000
0.136
0.076
-0.503
-0.129
-0.306
-0.196
0.258
0.115
0.456
0.328
-0.354
0.004

0.136
1.000
0.003
-0.335
-0.022
0.278
-0.231
0.359
0.140
0.127
-0.082
-0.036
0.206

0.076
0.003
1.000
-0.239
-0.042
0.020
-0.457
0.274
0.377
0.322
0.194
0.095
-0.306

-0.503
-0.335
-0.239
1.000
-0.312
0.542
0.459
-0.721
-0.237
-0.928
-0.477
0.414
0.556

-0.129
-0.022
-0.042
-0.312
1.000
-0.157
-0.124
0.220
0.053
0.347
-0.052
0.149
-0.298

-0.306
0.278
0.020
0.542
-0.157
1.000
0.106
-0.340
0.012
-0.578
-0.629
0.758
0.688

-0.196
-0.231
-0.457
0.459
-0.124
0.106
1.000
-0.177
-0.952
-0.511
-0.348
-0.033
0.231

0.258
0.359
0.274
-0.721
0.220
-0.340
-0.177
1.000
-0.130
0.662
0.394
-0.352
-0.411

10

0.115
0.140
0.377
-0.237
0.053
0.012
-0.952
-0.130
1.000
0.304
0.228
0.146
-0.090

2

3

4

5

urban

0.456
0.127
0.322
-0.928
0.347
-0.578
-0.511
0.662
0.304
1.000
0.523
-0.379
-0.668
6

11

how5

0.328
-0.082
0.194
-0.477
-0.052
-0.629
-0.348
0.394
0.228
0.523
1.000
-0.426
-0.494
7

12

bth5

illit_a

-0.354
-0.036
0.095
0.414
0.149
0.758
-0.033
-0.352
0.146
-0.379
-0.426
1.000
0.396

WATER_PUMP, WATER_WELL, WATER_TRUCKS, WATER_RIVER, WATER_OTHER, ELEM_EDUC, KITCHEN_YES,
8
9
10
11
12
13
KITCHEN_SHARE, KITCHEN_NO, URBAN, IMPROV_HSING, BATHRM_NO, and ILLITERATE

99

13

0.004
0.206
-0.306
0.556
-0.298
0.688
0.231
-0.411
-0.090
-0.668
-0.494
0.396
1.000

4.3.4 Objective 4: Characterize the spatial and temporal associations between ENSO,
climate and cholera incidence by district in Piura for 1997-98 and estimate the degree
to which social vulnerability influenced this relationship

Objective 4 Data
a) Cholera
Cholera data by district (n=33) per week for January 1997 to December 1998 in
Piura were obtained from the Department of Epidemiology at the Ministry of Health in
Lima, Peru. I aggregated these data to a monthly scale using cholera case-counts in
1997 to 1998 and monthly cholera case-counts for 1998.

17

In addition population

estimates by district for 1998 were obtained from the INEI webpage. I then calculated
cholera incidence rates (herein referred to as, cholera incidence) at two scales (total
cholera incidence for 1997 to 1998 and monthly cholera incidence in 1998) by dividing
the total number of cholera cases by the estimated population in 1998. The total
cholera case-counts for 1997 to 98 and monthly cholera case-counts for 1998 were
square-root transformed for analytical analyses. The reasoning for the two scales of
data is explained in the subsequent methods section.
b) SVI
The SVI and SVI factors by district (n = 33) estimated in the previous section were
also utilized in this objective.

17

Because 1997 was a year with less than 50 cholera cases in the subregion, I decided to
sum 1997 and 1998 cases by district.
100

c) Climate
Monthly values of global and local sea surface temperature anomalies (Niño 3.4,
1+2 and Paita); local temperature anomalies (TMAXA, TMEANA and TMINA) and rainfall
for 1997 to 1998 in Piura (obtained from datasets described in section 4.3.2.1) were also
utilized in this objective. From these data, a new set of time series were created with 1
to 7 month lag associations. The temporal lag implied that climate may have led cholera
by 1 to 7 months. In total there were 8 time series (0 to 7 month lags) for each of the
climate parameters. Each time series had 0 to 7 month lags with a total of 12 months.
Objective 4 Methods
After examining the data that I collected while I was in Peru, I realized that my
data had issues related to sample size and scale mismatch. First, the sample size (n=33
districts) was too small to fully estimate the modifying (indirect) effect(s) of social
vulnerability on the cholera and ENSO, climate relationships, as outlined in my Objective
4 in my proposal. Second, I only had climate data from one meteorological station and
therefore, I did not have informaton on the climate variablility across the district.
Instead I decided to use the monitoring data to study the climate and cholera
relationship in separate models for each district.

18

Third, I wanted to estimate the

effect of SST on cholera at a district-level. However, since SST is based on parameters in

18

After consulting with Prof. Norma Ordinola, my collaborator at the University of Piura,
I decided to use the data at the Miraflores station located in the city of Piura. Although
reliable, the data were only representative of climate in districts along the low lying
coast and may not capture climatic conditions further inland or at higher elevations
(personal communication June 2009).
101

the equatorial Pacific Ocean, obtaining a measure of SST variability at a district-level is
not possible. I, therefore, examined the effect(s) of each hazard (e.g., social
vulnerability and climate) on cholera incidence by district in separate models and then
descriptively compared the mapped-spatial patterns in coefficients from those models.
In the first part of this objective, I examined the spatial relationships between
cholera incidence (1997-1998) and SVI and SVI factors at the district level (n=33). In the
second part, I examined the temporal relationships between cholera incidence and
global and local climate variables by district (n=33) and month (n=12) in 1998. This
estimated the potential temporal impact of climate (monthly) on cholera across districts
in Piura.
a) Objective 4 Methods Part 1
Unadjusted global (ordinary least squares regression [OLS]) and local
(geographically weighted regression [GWR]) models were estimated to assess the effect
of SVI or SVI factors on cholera incidence (dependent variable) in the subregion of Piura
at the district level. The OLS regression analysis estimated the strength of association
(r2) and level of significance (p-values) in these relationship(s) across the study area.
Subsequently, local GWR models were estimated to visualize and explore the spatial
variation in model coefficients (i.e., intercepts and slopes) and residual-error terms.
The results from the global and local models were compared in order to assess how
localized cholera incidence was in each region (in constrast to being evenly distributed
across the subregion). GWR is a spatial regression technique that accounts for spatial
autocorrelation in the data; therefore in this objective it was used to characterize non-

102

stationary spatial associations, where the relationships between variables vary by
location (Fotheringham et al. 2002). All models were estimated using OLS and GWR
functions in the Spatial Statistics tool in ArcGIS (v.9.3.1) (ESRI 2010).
b) Objective 4 Methods Part 2
OLS regression analyses estimated the association(s) between cholera incidence
and global and local climate parameters for each district in Piura. Bivariate analyses
were conducted where the monthly cholera incidence was the dependent variable and
global sea surface temperature (Niño 3.4 SSTA and Niño 1+2 SSTA), local sea surface
temperature (Paita SSTA), local temperature (TMAXA, TMEANA and TMINA), and local
rainfall were explanatory variables. Temporal lag associations from zero to 7 months
were explored. I identified the strength of climate association (r-squared) and
significance (p-value) by district, variable and time lag. All OLS analyses were performed
in Matlab R2009a (Matlab 2010). The districts with the strongest association(s) in
cholera incidence and climate parameters and temporal lags (strongest teleconnections)
were then joined into one table and mapped in ArcGIS version 9.2 (ESRI 2010) to
explore the spatial variation of the climate-cholera relationships.

4.3.5 Objective 5: Reflect on the findings from this study using a climate and
development ethics perspective in order to better formulate recommendations

Objective 5 Approach
In this objective, I wrote a reflective statement that draws on the ethics of
climate and development literatures to address the moral questions that underlie my
103

research questions and research process. In this coda, my goal was to use the
methodological contributions of the study presented as a starting point for suggestions
about approaches to ethical issues in interdisciplinary research and policy around
climate and health, and broader societal issues pertaining to global climate change and
global development. I refer to these ethical challenges as ‘ethical geographies’. I
addressed challenges pertinent not only to cholera and development, but climate and
development as well. The notions of “climate ethics” and “climate justice” have been
developing since the 1990s due to growing concerns about extreme weather and a
changing climate. A particular concern for public health practitioners is that impacts will
be disproportionately felt in developing countries, where currently, large segments of
their populations are deprived of basic human needs. This statement links these ethical
concerns with those in global development. In addition questions were considered that
arose during the research process, such as trade-offs and “useability” of the research
(Glantz 2001a: 7). Trade-off refers to the opportunity costs incurred by the
collaborators and others that have assisted the researcher in this project. “Useability”
refers to the potential benefits of the research to society that includes its application
(Glantz 2001a: 7) and credibility and communicability of the conclusions to society
(Brown and Doberneck 2009). Specifically I referred to literatures on climate ethics and
justice (Jamieson and Glantz 2000; Glantz 2003; Jamieson 2010), ethics of global
development (Gasper 2004; Crocker 2008), and the capability approach (Nussbaum
2000; Sen 1999; 2009; Esquith and Gifford 2010) to address the ethical geographies in

104

my coda. This statement was used to inform the recommendations that I present in the
concluding chapter of this dissertation research.

105

5. CHAPTER 5: RESULTS
In this chapter, I present key results for Objectives 2, 3, and 4. First I compare
ENSO events by Niño region. I then discuss the wavelet analysis, which estimated the
temporal associations between climate parameters and cholera cases in the subregion
of Piura from 1991 to 2001. Following the temporal analysis, I present the SVI by district
in Piura and the analysis estimating the spatial associations between (a) SVI and cholera
incidence (1997-98) and (b) climate parameters and cholera incidence in 1998.

5.1

ENSO Events by Niño Index (Objective 2)
Figures 5.1 and 5.2 are time plots of the Niño 3.4 and 1+2 region indices. An

ENSO event was defined as a period during which the values of the indices (3-month
running means for SSTA) in a given Niño region exceeded a threshold of +/-0.5 (°C) for at
least 5 consecutive months.

NINO 3.4 Index (degC)

3
2
1
0
-1
-2
-3

1975

1980

1985

1990

1995

2000

Figure 5.1 Niño 3.4 Index from 1971 to 2001 at ±0.5 (°C) threshold (gray dotted lines).

106

5
NINO 1+2 Index (degC)

4
3
2
1
0
-1
-2
-3

1975

1980

1985

1990

1995

2000

Figure 5.2 Niño 1+2 Index from 1971 to 2001 at ±0.5 (°C) threshold (gray dotted lines).

Tables 5.1 and 5.2 compare El Niño (red) and La Niña (blue) events according to
the Niño 3.4 and 1+2 Indices for 1990 to 2001. The 2 regions illustrate different Niño
signature patterns. For the entire period, 39 months were assigned to El Niño and 36
months were assigned to La Niña in the Niño 3.4 region. This suggests that ENSO events
were in mode 52.0% (75 of 144 months) of the time in the central equatorial Pacific
Ocean. This comprises of 3 El Niños and 3 La Niñas. On the eastern end of the
equatorial Pacific Ocean, the Niño 1+2 region shows that La Niñas appear to be the
dominant mode. Fifty-nine months were assigned to La Niña, while only 27 months
were El Niño. In contrast to the Niño 3.4 region, that consisted of 2 El Niños and 7 La
Niñas. Here, an El Niño or La Niña event was underway 60.0% (86 of 144 months) of the
time.

107

Table 5.1 Comparison of El Niño (red) and La Niña (blue) events according to the Niño
3.4 Index from 1990 to 2001
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001

J
0.1
0.4
1.8
0.3
0.2
1.2
-0.7
-0.4
2.3
-1.4
-1.6
-0.6

F
0.2
0.3
1.6
0.4
0.2
0.9
-0.7
-0.3
1.9
-1.2
-1.4
-0.5

M
0.2
0.3
1.5
0.6
0.3
0.7
-0.5
0.0
1.5
-0.9
-1.0
-0.4

A
0.2
0.4
1.4
0.7
0.4
0.4
-0.3
0.4
1.0
-0.8
-0.8
-0.2

M
0.2
0.6
1.2
0.8
0.5
0.3
-0.1
0.8
0.5
-0.8
-0.6
-0.1

J
J
A
S
0.2 0.3 0.3 0.3
0.8 1.0 0.9 0.9
0.8 0.5 0.2 0.0
0.7
0.4 0.4 0.4
0.5 0.6 0.6 0.7
0.2 0.0 -0.2 -0.5
-0.1 0.0 -0.1 -0.1
1.3 1.7 2.0 2.2
0.0 -0.5 -0.8 -1.0
-0.8 -0.9 -0.9 -1.0
-0.5 -0.4 -0.4 -0.4
0.1 0.2 0.2 0.1

O
0.3
1.0
-0.1
0.4
0.9
-0.6
-0.2
2.4
-1.1
-1.1
-0.5
0.0

N
0.3
1.4
0.0
0.3
1.2
-0.7
-0.3
2.5
-1.3
-1.3
-0.6
-0.1

D
0.4
1.6
0.2
0.2
1.3
-0.7
-0.4
2.5
-1.4
-1.6
-0.7
-0.1

Table 5.2 Comparison of El Niño (red) and La Niña (blue) events according to the Niño
1+2 Index from 1990 to 2001
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001

J
-0.4
-0.3
0.6
0.1
-0.1
0.7
-0.5
-0.7
3.6
-0.4
-0.7
-0.5

F
-0.2
-0.1
0.8
0.4
-0.4
0.4
-0.4
-0.1
3.2
-0.3
-0.4
0.1

M
-0.1
-0.1
1.5
0.8
-0.7
-0.2
-0.7
0.6
3.0
-0.4
-0.1
0.7

A
-0.2
0.0
1.9
0.9
-1.0
-0.8
-1.0
1.5
3.2
-0.5
-0.1
0.5

M
-0.2
0.0
1.8
0.9
-0.9
-0.8
-1.4
2.4
3.0
-1.0
-0.2
-0.2

J
-0.4
0.2
1.0
0.7
-0.7
-0.7
-1.5
3.2
2.4
-1.1
-0.7
-0.9

J
-0.5
0.2
0.3
0.4
-0.7
-0.6
-1.5
3.7
1.6
-1.2
-0.9
-1.0

A
-0.6
0.3
-0.2
0.3
-0.7
-0.5
-1.4
3.9
1.0
-1.2
-0.8
-1.0

S
-0.5
0.3
-0.2
0.3
-0.2
-0.6
-1.1
3.9
0.6
-1.0
-0.6
-1.1

O
-0.6
0.4
-0.2
0.2
0.4
-0.5
-1.2
3.9
0.2
-1.1
-0.7
-1.2

N
-0.6
0.5
-0.2
0.0
0.7
-0.7
-1.2
3.9
-0.1
-1.0
-0.7
-1.1

D
-0.6
0.5
-0.2
-0.1
0.8
-0.6
-1.1
4.0
-0.4
-1.0
-0.8
-1.0

Tables 5.3 and 5.4 display the timing of ENSO events with the duration of
months per ENSO event from 1990 to 2001. From 1990 to 1995 there is a marked
contrast between the two Niño Indices. The difference in the number of El Niño/ La
Niña months and their timing is notable. There were a total of 26 El Niño months and 4
108

months assigned to La Niña in the Niño 3.4 region. The first El Niño began in May 1991
and lasted until July 1992. Another event took place from May 1994 to March 1995.
One La Niña event was qualified from September 1995 into the summer season of 1996.
In contrast, the Niño 1+2 region was assigned 8 months as El Niño and 21 months as La
Niña. The only El Niño event in this region began in November 1991 and ended June
1992. A La Niña preceded this warm event beginning in July 1990 and lasted 6 months.
The next La Niña events qualified in March 1994 for 6 months and April 1995 for 9
month.

Table 5.3 Niño 3.4 Index: Timing of El Niño and La Niña events from 1990 to 2001
El Niños
Year Begin Year End Duration (months)
1991
May 1992 Jul
15
1994
May 1995 Mar
11
1997
May 1998 May
13
La Niñas
Year Begin Year End Duration (months)
1995
Sep 1996 Mar
7
1998
Jul
2000 Jun
24
2000
Oct 2001 Feb
5

109

Table 5.4 Niño 1+2 Index: Timing of El Niño and La Niña events from 1990 to 2001
El Niños
Year Begin Year End Duration (months)
1991
Nov 1992 Jun
8
1997
Mar 1998 Sep
19
La Niñas
Year Begin Year End Duration (months)
1990
Jul
1990 Dec
6
1994
Mar 1994 Aug
6
1995
Apr 1996 Jan
10
1996
Mar 1997 Jan
11
1999
Apr 2000 Jan
10
2000
Jun 2001 Jan
8
2001
Jun 2002 Jan
8

Compared to the first half of the time period, the two Niño indices show fewer
differences in El Niño/La Niña patterns from 1996 to 2001. Except for 1996 during
which the Niño 3.4 region contrasts the Niño 1+2 region by the number of La Niña
months (e.g., 3 versus 11), the subsequent years show both regions follow similar
dominant ENSO modes including the extreme El Niño in 1997-98 followed by several La
Niña months until 2001. In the Niño 3.4 region an El Niño began in May 1997 lasting 13
months, while the Niño 1+2 region qualifies El Niño earlier (March) and lasts longer (19
months). In June 1998 there is an abrupt switch of modes from warm to cold in the
Niño 3.4 region. In the following month a La Niña qualified and persisted for 2 years
(ending June 2000). Another La Niña qualified in October 2000 and ended February
2001. In the Niño 1+2 region there is a neutral period of 6 months before a 10-month
La Niña begins in April 1999. This episode was followed by two La Niñas in June 2000
110

and June 2001. These tables demonstrate the need to clarify the definition of ENSO in
climate and health studies.

5.2

Cholera Cases by Niño Index and ENSO Event (Objective 2)
Tables 5.5 and 5.6 show a comparison of ENSO months and cholera cases.

When comparing ENSO events/periods by index with the occurrence of cholera in Piura
by month, an interesting observation is revealed. According to both Niño indices,
cholera was more likely to occur in a neutral month than an ENSO event month. In fact,
although Niño 3.4 and 1+2 indices had different ENSO patterns, particularly in the first
half of the decade, it can be discerned that cholera emerged during a neutral period
preceded by neutral months in the Niño 3.4 region and La Niña months in the Niño 1+2
region. Overall, in the Niño 3.4 region, cholera was reported during 43 neutral months
compared to 35 El Niño months and 32 La Niña months. In the Niño 1+2 region, cholera
was found in 51 neutral months, while cholera was only reported in 22 El Niño months
and 35 La Niña months.

111

Table 5.5 Comparison of ENSO months and cholera cases using the Niño 3.4 Index from
1990 to 2001
J

F

M

A

M

J

J

A

S

O

N

D

1990
1991 81 6804 10591 4360 1181 308 243 40 13 23 0
0
1992 96 230 1582 2838 2202 684 89 41 53 55 62 58
1993 118 65
168
130
99
46 10 33 61 81 162 181
1994 353 143
98
98
43
3
0
4 1 1
3
2
1995 17
20
26
50
61
7
12 7 2 8
3
0
1996 3
10
1
4
0
1
0
2 3 2
1
0
1997 3
4
3
0
0
1
0
2 1 2
5
4
1998 156 1472 1534 559 252 59 44 30 7 9
9
16
1999 1
19
12
12
4
2
2
5 1 1
0
3
2000 1
3
3
7
2
0
0
1 0 2
1
0
2001 1
0
1
7
0
0
0
0 0 0
0
0
Table 5.6 Comparison of ENSO months and cholera cases using the Niño 1+2 Index from
1990 to 2001
J
F
M
A
M
J
J
A S O
N
D
1990
1991 81 6804 10591 4360 1181 308 243 40 13 23
0
0
1992 96 230 1582 2838 2202 684 89 41 53 55 62 58
1993 118
65
168 130
99 46 10 33 61 81 162 181
1994 353 143
98
98
43
3
0 4 1 1
3
2
1995 17
20
26
50
61
7 12 7 2 8
3
0
1996
3
10
1
4
0
1
0 2 3 2
1
0
1997
3
4
3
0
0
1
0 2 1 2
5
4
1998 156 1472 1534 559 252 59 44 30 7 9
9 16
1999
1
19
12
12
4
2
2 5 1 1
0
3
2000
1
3
3
7
2
0
0 1 0 2
1
0
2001
1
0
1
7
0
0
0 0 0 0
0
0

112

5.3

Wavelet Transform Analyses (Objective 2)
The aim of this section is to examine the temporal patterns of cholera cases and

global and local climate parameters from 1991 to 2001. Figure 5.3 is cholera case
anomaly in Piura from 1991 to 2001. The time plot shows that cholera case anomaly
peaked in 1991, 1992, 1994, and throughout 1998. Figures 5.4 to 5.10 show time plots
of global and local climate parameter anomalies from 1975 to 2001. The most notable
peaks across most climate series were in 1982-83 and 1997-98 during the strongest El
Niños of the twentieth century. The latter El Niño event coincided with a rise of cholera
in Piura and throughout Peru. Prior to the 1997-98 El Niño, there were several peaks
observed in the first half of the 1990s for all climate parameters. Understanding these
temporal patterns using wavelet analysis is the focus of the next sections.

113

Cholera Cases (sqrt) Anomaly

3
2
1
0
-1
-2

1992 1993 1994 1995 1996 1997 1998 1999 2000 2001

Figure 5.3 Cholera Cases (square-root transformed) Anomaly from 1991 to 2001.
The gray dotted line indicates the mean.

114

3

NINO 3.4 SSTA (degC)

2
1
0
-1
-2
-3

1975

1980

1985

1990

Figure 5.4 Niño 3.4. Sea Surface Temperature Anomaly (SSTA) (°C) from 1971 to 2001.
The gray dotted line indicates the mean.

115

1995

2000

5

NINO 1+2 SSTA (degC)

4
3
2
1
0
-1
-2
-3

1975

1980

1985

1990

Figure 5.5 Niño 1+2 Sea Surface Temperature Anomaly (SSTA) (°C) from 1971 to-2001.
The gray dotted line indicates the mean.

116

1995

2000

4

Paita SSTA (degC)

3
2
1
0
-1
-2

1975

1980

1985

1990

Figure 5.6 Paita Sea Surface Temperature Anomaly (SSTA) (°C) from 1971 to 2001.
The gray dotted line indicates the mean.

117

1995

2000

Temperature MAXA (degC)

4
3
2
1
0
-1
-2
-3

1975

1980

1985

1990

Figure 5.7 Temperature Maximum Anomaly (TMAXA) (°C) in Piura from 1971 to 2001.
The gray dotted line indicates the mean.

118

1995

2000

Temperature MEANA (degC)

5
4
3
2
1
0
-1
-2
-3

1975

1980

1985

1990

Figure 5.8 Temperature Mean Anomaly (TMEANA) (°C) in Piura from 1971 to 2001.
The gray dotted line indicates the mean.

119

1995

2000

Temperature MINA (degC)

5
4
3
2
1
0
-1
-2
-3

1975

1980

1985

1990

Figure 5.9 Temperature Minimum Anomaly (TMINA) (°C) in Piura from 1971 to 2001.
The gray dotted line indicates the mean.

120

1995

2000

Rainfall (sqrt) Anomaly (mm)

6
4
2
0
-2

1975

1980

1985

1990

Figure 5.10 Rainfall (square-root transformed) Anomaly (mm) in Piura from 1971 to 2001.
The gray dotted line indicates the mean.

121

1995

2000

Cholera Cases Anomaly

Figure 5.11 shows the wavelet transform of cholera cases anomaly in Piura.
Cholera cases in Piura show high significant power at the 1 – 1.5 yr scale from 1991 until
mid-1994. Periodicity at this scale was reliable only after mid-1992 (because data in
1991 were inside the COI). Still it illustrates there was high cholera activity during the
initial outbreak. After 1994 until 1998, there was moderate power at a 2 yr scale.
Although periodicity was not statistically significant at this time interval, the shape of
the spectra is marked during the resurgence of cholera in 1997 to 1998 at low power
and at scales less than 1 year. Why cholera activity from 1997 to 1998 does not exhibit
high power or is not statistically significant is surprising and interesting given it was the
third largest outbreak in Piura. On average the GWS indicated that periodicity of
cholera cases during the 1990s was not statistically significant.

122

Period (Year)

0.25
0.5
1
2
4
0

1992 1993 1994 1995 1996 1997 1998 1999 2000 2001
-4

-3.5

-3

-2.5

-2

-1.5

-1

-0.5

0

0.5

2

4

6

8

Power
1

Figure 5.11 Wavelet Transform Analysis of Cholera Cases Anomaly in Piura by month from 1991 to 2001.
The left panel represents the wavelet transform by period (scale by year) and across time. Power, indicating periodicity or
frequency, increases from blue to red in the wavelet transform, and statistical significance (95.0% confidence level) is indicated by
areas with black outlines outside the cone of influence (COI), where edge effects are not influential. The right panel represents the
global wavelet power spectrum, an estimate of the average scales and powers; statistical significance (95.0% confidence level) is
denoted by the black dashed line across the local wavelet spectra.

123

Sea Surface Temperature Anomaly
The wavelet transform analyses for global and local SSTA are shown in Figures
5.12 to 5.14. Before 1991, Niño 3.4 SSTA showed high significant power from 1981 to
1990 at a 4-5 yr scale (Figure 5.12). After 1991, although there appears to be high
power during the initial cholera outbreak (1991 to 1995), periodicity for Niño 3.4 SSTA
was not statistically significant. However, from 1995 to 1998, which includes the
resurgence of cholera, there was high significant power for Niño 3.4 SSTA at a 3-4 yr
scale. In the Niño 1+2 region, SSTA showed high significant power throughout the 1980s
and 1990s at a 4 yr scale (Figure 5.13). Periodicity was most notable in 1982-83 and
during the resurgence of cholera in 1997-98 at a 2 to 4 yr scale. In Paita, local SSTA
showed high significant power before 1986 and from 1995 to 1998 at a 2 to 4 yr scale
(Figure 5.14). On average the GWS indicated that significant periodicity for Niño 3.4,
Niño 1+2, and Paita SSTA was found at scales of 3 to 5 yrs.

124

0.25

Period (Year)

0.5
1
2
4
8
1975
-4

-3.5

1980
-3

-2.5

1985
-2

-1.5

1990
-1

1995
-0.5

0

2000
0.5

0
1

10
Power

20

Figure 5.12 Wavelet Transform Analysis of Niño 3.4 Sea Surface Temperature Anomaly (SSTA) (°C) by month from 1971 to 2001.
The left panel represents the wavelet transform by period (scale by year) and across time. Power, indicating periodicity or
frequency, increases from blue to red in the wavelet transform, and statistical significance (95.0% confidence level) is indicated by
areas with black outlines outside the cone of influence (COI), where edge effects are not influential. The right panel represents the
global wavelet power spectrum, an estimate of the average scales and powers; statistical significance (95.0% confidence level) is
denoted by the black dashed line across the local wavelet spectra.

125

0.25

Period (Year)

0.5
1
2
4
8
1975
-4

-3.5

1980
-3

-2.5

1985
-2

-1.5

1990
-1

1995
-0.5

0

2000
0.5

0

20
Power

40

1

Figure 5.13 Wavelet Transform Analysis of Niño 1+2 Sea Surface Temperature Anomaly (SSTA) (°C) by month from 1971 to 2001.
The left panel represents the wavelet transform by period (scale by year) and across time. Power, indicating periodicity or
frequency, increases from blue to red in the wavelet transform, and statistical significance (95.0% confidence level) is indicated by
areas with black outlines outside the cone of influence (COI), where edge effects are not influential. The right panel represents the
global wavelet power spectrum, an estimate of the average scales and powers; statistical significance (95.0% confidence level) is
denoted by the black dashed line across the local wavelet spectra.

126

0.25

Period (Year)

0.5
1
2
4
8
1975
-4

-3.5

1980
-3

-2.5

1985
-2

-1.5

1990
-1

1995
-0.5

0

2000
0.5

0
1

5
10
Power

Figure 5.14 Wavelet Transform Analyses of Paita Sea Surface Temperature Anomaly (SSTA) (°C) by month from 1971 to 2001.
The left panel represents the wavelet transform by period (scale by year) and across time. Power, indicating periodicity or
frequency, increases from blue to red in the wavelet transform, and statistical significance (95.0% confidence level) is indicated by
areas with black outlines outside the cone of influence (COI), where edge effects are not influential. The right panel represents the
global wavelet power spectrum, an estimate of the average scales and powers; statistical significance (95.0% confidence level) is
denoted by the black dashed line across the local wavelet spectra.

127

Temperature Anomaly
The wavelet transform analyses of local temperature anomalies are displayed in
Figures 5.15 (TMAXA), 5.16 (TMEANA) and 5.17 (TMINA). Moderate to high significant
power was found before the initial cholera outbreak in 1983 at a 1 yr scale for TMAXA
and from 1982 to 1987 at a 4 yr scale for TMEANA and TMINA. Significant periodicity
for local temperature parameters was not observed during the first half of the 1990s.
From 1995 to 1999, however, significant periodicity was found for TMAXA, TMEANA,
and TMINA ranging from moderate to high power at scales of 2 to 4 yrs. Significant
power was evident among all three parameters during the resurgence of cholera in
1997-98. On average the GWS indicated that significant periodicity for TMAXA was
found at a 2 yr scale and at a 4 yr scale for TMEANA and TMINA.

128

0.25

Period (Year)

0.5
1
2
4
8
1975
-4

-3.5

1980
-3

-2.5

1985
-2

-1.5

1990
-1

1995
-0.5

0

2000
0.5

0

10
Power

20

1

Figure 5.15 Wavelet Transform Analyses of Temperature Maximum Anomaly (TMAXA) (°C) by month in Piura from 1971 to 2001.
The left panel represents the wavelet transform by period (scale by year) and across time. Power, indicating periodicity or
frequency, increases from blue to red in the wavelet transform, and statistical significance (95.0% confidence level) is indicated by
areas with black outlines outside the cone of influence (COI), where edge effects are not influential. The right panel represents the
global wavelet power spectrum, an estimate of the average scales and powers; statistical significance (95.0% confidence level) is
denoted by the black dashed line across the local wavelet spectra.

129

0.25

Period (Year)

0.5
1
2
4
8
1975
-4

-3.5

1980
-3

-2.5

1985
-2

-1.5

1990
-1

1995
-0.5

0

2000
0.5

0

5
10
Power

1

Figure 5.16 Wavelet Transform Analyses of Temperature Mean Anomaly (TMEANA) (°C) by month in Piura from 1971 to 2001.
The left panel represents the wavelet transform by period (scale by year) and across time. Power, indicating periodicity or
frequency, increases from blue to red in the wavelet transform, and statistical significance (95.0% confidence level) is indicated by
areas with black outlines outside the cone of influence (COI), where edge effects are not influential. The right panel represents the
global wavelet power spectrum, an estimate of the average scales and powers; statistical significance (95.0% confidence level) is
denoted by the black dashed line across the local wavelet spectra.

130

0.25

Period (Year)

0.5
1
2
4
8
1975
-4

-3.5

1980
-3

-2.5

1985
-2

-1.5

1990
-1

1995
-0.5

0

2000
0.5

0

5
10
Power

1

Figure 5.17 Wavelet Transform Analyses of Temperature Minimum Anomaly (TMINA) (°C) by month in Piura from 1971 to 2001.
The left panel represents the wavelet transform by period (scale by year) and across time. Power, indicating periodicity or
frequency, increases from blue to red in the wavelet transform, and statistical significance (95.0% confidence level) is indicated by
areas with black outlines outside the cone of influence (COI), where edge effects are not influential. The right panel represents the
global wavelet power spectrum, an estimate of the average scales and powers; statistical significance (95.0% confidence level) is
denoted by the black dashed line across the local wavelet spectra.

131

Rainfall Anomaly
Figure 5.18 is the wavelet transform of local rainfall anomaly. High significant
power was found before the initial cholera outbreak in 1982 to 1984 and during the
resurgence of cholera in 1997-98 at multiple scales ranging from .8 to 2 yr scale. There
was also high to moderate significant power at several scales less than 1 yr throughout
most of the 1980s and in 1996. On average the GWS indicated that significant
periodicity for rainfall anomaly was significant at less than 1 yr scale.

132

0.25

Period (Year)

0.5
1
2
4
8
1975
-4

-3.5

1980
-3

-2.5

1985
-2

-1.5

1990
-1

1995
-0.5

0

0

2000
0.5

2

4

Power
1

Figure 5.18 Wavelet Transform Analyses of Rainfall Anomaly (mm) by month in Piura from 1971 to 2001.
The left panel represents the wavelet transform by period (scale by year) and across time. Power, indicating periodicity or
frequency, increases from blue to red in the wavelet transform, and statistical significance (95.0% confidence level) is indicated by
areas with black outlines outside the cone of influence (COI), where edge effects are not influential. The right panel represents the
global wavelet power spectrum, an estimate of the average scales and powers; statistical significance (95.0% confidence level) is
denoted by the black dashed line across the local wavelet spectra.

133

5.4

Wavelet Coherence Analyses (Objective 2)

5.4.1 Climate-Cholera Associations
Sea Surface Temperature Anomaly and Cholera Associations
The wavelet coherence between cholera cases anomaly and SSTA were
examined first. The analysis shows that moderate to high significant coherence existed
between cholera and Niño 3.4 SSTA in mid-1993 and during the resurgence of cholera
from 1997 to 1999 at a 1.5 yr scale (Figure 5.19). The phase angle suggests that at these
time intervals cholera followed Niño 3.4 by angles greater than 90° or approximately
more than 6 months.
Figure 5.20 shows the wavelet coherence between cholera cases anomaly and
Niño 1+2 SSTA. Moderate significant coherence was found between cholera and Niño
1+2 in 1993-94 at a 1.5 yr scale where SSTA led cholera by 90°. During the resurgence of
cholera, Niño 1+2 SSTA led cholera by 2-3 months from 1997 to 1999 at a 2 yr scale.
With respect to local sea surface temperature conditions in Piura, moderate significant
coherence was found between cholera cases anomaly and Paita SSTA (Figure 5.21).
Paita SSTA, like Niño 3.4 SSTA, led cholera by more than 6 months in 1993 and 1998 at a
scale of 1.5 yr.

19

19

The time interval scales were approximately 1993-94 and 1998-99.
134

Figure 5.19 Wavelet Coherence of Niño 3.4 Sea Surface Temperature Anomaly (SSTA) and Cholera Cases in Piura from 1991 to 2001.
The panel represents the wavelet coherence by period (scale by year). Coherence was estimated from low (blue) to high (red)
correlation. The direction (phase) of each coherence and temporal lead/lag (i.e., the phase angle or difference) was indicated by
arrows, as such: north (climate lags); south (climate leads); east (climate and cholera in phase); and west (climate and cholera out of
phase). Statistical significance (95.0% confidence level) is indicated by areas with black outlines outside the cone of influence (COI),
where edge effects are not influential.

135

Figure 5.20 Wavelet Coherence of Niño 1+2 Sea Surface Temperature Anomaly (SSTA) and Cholera Cases in Piura from 1991 to 2001.
The panel represents the wavelet coherence by period (scale by year). Coherence was estimated from low (blue) to high (red)
correlation. The direction (phase) of each coherence and temporal lead/lag (i.e., the phase angle or difference) was indicated by
arrows, as such: north (climate lags); south (climate leads); east (climate and cholera in phase); and west (climate and cholera out of
phase). Statistical significance (95.0% confidence level) is indicated by areas with black outlines outside the cone of influence (COI),
where edge effects are not influential.

136

Figure 5.21 Wavelet Coherence of Paita Sea Surface Temperature Anomaly (SSTA) and Cholera Cases in Piura from 1991 to 2001.
The panel represents the wavelet coherence by period (scale by year). Coherence was estimated from low (blue) to high (red)
correlation. The direction (phase) of each coherence and temporal lead/lag (i.e., the phase angle or difference) was indicated by
arrows, as such: north (climate lags); south (climate leads); east (climate and cholera in phase); and west (climate and cholera out of
phase). Statistical significance (95.0% confidence level) is indicated by areas with black outlines outside the cone of influence (COI),
where edge effects are not influential.

137

Temperature Anomaly and Cholera Associations
Wavelet coherence analyses between cholera cases anomaly and local
temperature anomalies are shown in Figures 5.22 to 5.24. High significant coherence
was found between cholera and TMAXA throughout the 1990s at the 2-3 yr scale (Figure
5.22). For TMEANA, moderate significant coherence was found with cholera at two
time intervals: 1993 to the middle of 1994 at a 1.5 yr scale and from 1995 to 1999 at a
2-3 yr scale (Figure 5.23). Similarly, there was moderate significant coherence between
cholera and TMINA from 1993 to 1995 at 1.5 yr scale and from 1996 until 1999 at scales
ranging from 1.5 to 2.5 yr scale (Figure 5.24). However, these relationships should be
considered with caution because the phase relationships and time lags were unclear in
the graphs (Refer to cross-wavelet transform results in section 5.4.2).

138

Figure 5.22 Wavelet Coherence of Temperature Maximum Anomaly (TMAXA) and Cholera Cases in Piura from 1991 to 2001.
The panel represents the wavelet coherence by period (scale by year). Coherence was estimated from low (blue) to high (red)
correlation. The direction (phase) of each coherence and temporal lead/lag (i.e., the phase angle or difference) was indicated by
arrows, as such: north (climate lags); south (climate leads); east (climate and cholera in phase); and west (climate and cholera out of
phase). Statistical significance (95.0% confidence level) is indicated by areas with black outlines outside the cone of influence (COI),
where edge effects are not influential.

139

Figure 5.23 Wavelet Coherence of Temperature Mean Anomaly (TMEANA) and Cholera Cases in Piura from 1991 to 2001.
The panel represents the wavelet coherence by period (scale by year). Coherence was estimated from low (blue) to high (red)
correlation. The direction (phase) of each coherence and temporal lead/lag (i.e., the phase angle or difference) was indicated by
arrows, as such: north (climate lags); south (climate leads); east (climate and cholera in phase); and west (climate and cholera out of
phase). Statistical significance (95.0% confidence level) is indicated by areas with black outlines outside the cone of influence (COI),
where edge effects are not influential.

140

Figure 5.24 Wavelet Coherence of Temperature Minimum Anomaly (TMINA) and Cholera Cases in Piura from 1991 to 2001.
The panel represents the wavelet coherence by period (scale by year). Coherence was estimated from low (blue) to high (red)
correlation. The direction (phase) of each coherence and temporal lead/lag (i.e., the phase angle or difference) was indicated by
arrows, as such: north (climate lags); south (climate leads); east (climate and cholera in phase); and west (climate and cholera out of
phase). Statistical significance (95.0% confidence level) is indicated by areas with black outlines outside the cone of influence (COI),
where edge effects are not influential.

141

Rainfall and Cholera Associations
Figure 5.25 is the wavelet coherence between rainfall anomaly and cholera cases
anomaly. Moderate to high significant coherence was found during the resurgence of
cholera in 1997 until 1999 at multiple scales ranging from 7 months to 2.25 yr. The
phase relationship between cholera and rainfall indicated that the phase direction
approached in-phase as the scale decreased by month (i.e., lag approached zero as time
delay decreased). Rainfall led cholera by approximately zero to one month lag.

142

Figure 5.25 Wavelet Coherence of Rainfall Anomaly and Cholera Cases in Piura from 1991 to 2001.
The panel represents the wavelet coherence by period (scale by year). Coherence was estimated from low (blue) to high (red)
correlation. The direction (phase) of each coherence and temporal lead/lag (i.e., the phase angle or difference) was indicated by
arrows, as such: north (climate lags); south (climate leads); east (climate and cholera in phase); and west (climate and cholera out of
phase). Statistical significance (95.0% confidence level) is indicated by areas with black outlines outside the cone of influence (COI),
where edge effects are not influential.

143

5.4.2 Cross-Wavelet Transform: Climate-Cholera
Figures 5.26 to 5.32 show the cross-wavelet transforms of cholera with the
global and local climate parameters. In general the cross-wavelet transform analyses
indicated that areas with significant coherence at scales greater than 1 yr shared high
common power, whereas scales lower than 1 yr shared low common power. It was also
found that a strong association could exist between different levels of power (Torrence
and Webster 1999). For example, during the resurgence of cholera there was significant
high coherence between cholera and rainfall from 1997 to 1999, yet in the crosswavelet analysis, there was moderate common power. It suggests that correlation in
time does not always lead to frequency of occurrence between cholera and climate.
Furthermore, the cross-wavelet results revealed that the phase and time relationships
between cholera and local temperature (TMAXA, TMEANA and TMINA) were likely
coincidental given that the phase arrows in the graphs were inconsistent across the
frequency scales where significant coherence was noted (Refer to Grinsted et al. (2004)
and associated webpage in References for more on the interpretation of the crosswavelet).

144

Figure 5.26 Cross-Wavelet of Niño 3.4 Sea Surface Temperature Anomaly (SSTA) and Cholera Cases in Piura from 1991 to 2001.
The cross-wavelet transform indicates where the two time series share common power by period (scale by year). Common power
increases from blue to red. The direction (phase) and temporal lead/lag (i.e., the phase angle or difference) are indicated by arrows,
as such: north (climate lags); south (climate leads); east (climate and cholera in phase); and west (climate and cholera out of phase).
Statistical significance (95.0% confidence level) is indicated by areas with black outlines outside the cone of influence (COI), where
edge effects are not influential.

145

Figure 5.27 Cross-Wavelet of Niño 1+2 Sea Surface Temperature Anomaly (SSTA) and Cholera Cases in Piura from 1991 to 2001.
The cross-wavelet transform indicates where the two time series share common power by period (scale by year). Common power
increases from blue to red. The direction (phase) and temporal lead/lag (i.e., the phase angle or difference) are indicated by arrows,
as such: north (climate lags); south (climate leads); east (climate and cholera in phase); and west (climate and cholera out of phase).
Statistical significance (95.0% confidence level) is indicated by areas with black outlines outside the cone of influence (COI), where
edge effects are not influential.

146

Figure 5.28 Cross-Wavelet of Paita Sea Surface Temperature Anomaly (SSTA) and Cholera Cases in Piura from 1991 to 2001.
The cross-wavelet transform indicates where the two time series share common power by period (scale by year). Common power
increases from blue to red. The direction (phase) and temporal lead/lag (i.e., the phase angle or difference) are indicated by arrows,
as such: north (climate lags); south (climate leads); east (climate and cholera in phase); and west (climate and cholera out of phase).
Statistical significance (95.0% confidence level) is indicated by areas with black outlines outside the cone of influence (COI), where
edge effects are not influential.

147

Figure 5.29 Cross-Wavelet of Temperature Maximum Anomaly (TMAXA) and Cholera Cases in Piura from 1991 to 2001.
The cross-wavelet transform indicates where the two time series share common power by period (scale by year). Common power
increases from blue to red. The direction (phase) and temporal lead/lag (i.e., the phase angle or difference) are indicated by arrows,
as such: north (climate lags); south (climate leads); east (climate and cholera in phase); and west (climate and cholera out of phase).
Statistical significance (95.0% confidence level) is indicated by areas with black outlines outside the cone of influence (COI), where
edge effects are not influential.

148

Figure 5.30 Cross-Wavelet of Temperature Mean Anomaly (TMEANA) and Cholera Cases in Piura from 1991 to 2001.
The cross-wavelet transform indicates where the two time series share common power by period (scale by year). Common power
increases from blue to red. The direction (phase) and temporal lead/lag (i.e., the phase angle or difference) are indicated by arrows,
as such: north (climate lags); south (climate leads); east (climate and cholera in phase); and west (climate and cholera out of phase).
Statistical significance (95.0% confidence level) is indicated by areas with black outlines outside the cone of influence (COI), where
edge effects are not influential.

149

Figure 5.31 Cross-Wavelet of Temperature Minimum Anomaly (TMINA) and Cholera Cases in Piura.
The cross-wavelet transform indicates where the two time series share common power by period (scale by year). Common power
increases from blue to red. The direction (phase) and temporal lead/lag (i.e., the phase angle or difference) are indicated by arrows,
as such: north (climate lags); south (climate leads); east (climate and cholera in phase); and west (climate and cholera out of phase).
Statistical significance (95.0% confidence level) is indicated by areas with black outlines outside the cone of influence (COI), where
edge effects are not influential.

150

Figure 5.32 Cross-Wavelet of Rainfall Anomaly and Cholera Cases in Piura from 1991 to 2001.
The cross-wavelet transform indicates where the two time series share common power by period (scale by year). Common power
increases from blue to red. The direction (phase) and temporal lead/lag (i.e., the phase angle or difference) are indicated by arrows,
as such: north (climate lags); south (climate leads); east (climate and cholera in phase); and west (climate and cholera out of phase).
Statistical significance (95.0% confidence level) is indicated by areas with black outlines outside the cone of influence (COI), where
edge effects are not influential.

151

5.4.3 Cross-Correlation Analysis: Climate-Cholera
The results from cross-correlation analyses between the cholera and global and
local climate time series are displayed in Table 5.7. Cholera had a strong association
with SSTA parameters at 6 and 7 month lags. There were also strong associations at 5
month lag (Niño 1+2 SSTA) and 8 month lag (Paita). With rainfall, cholera was
correlated at 1-2 month lags and 6-7 month lags. It was also found that the direction of
associations between cholera and global and local climate parameters was positive and
that climate led cholera. The temporal lags associated with SSTA and local temperature
anomaly parameters were typically longer compared to rainfall anomaly. In sum, the
cross-correlation analyses supported the wavelet coherence results (See Table 5.8 for a
summary of wavelet coherence by lag, direction and phase). In addition, it was shown
that the strongest associations were with Niño 3.4 SSTA.

152

Table 5.7 Cross-Correlations between Cholera Case Anomaly and Global and Local
Climate Time Series
Lag by
NIÑO 3.4
NIÑO 1+2
PAITA
Rainfall
Std.
month
SSTA
SSTA
SSTA
Anomaly
Error
-12
0.23
0.04
0.01
-0.09
0.091
-11
0.25
0.02
0.01
-0.10
0.091
-10
0.25
0.03
0.01
-0.08
0.091
-9
0.22
0.01
-0.01
-0.05
0.090
-8
0.20
-0.02
-0.04
-0.06
0.090
-7
0.17
-0.04
-0.09
-0.10
0.089
-6
0.14
-0.03
-0.11
-0.12
0.089
-5
0.13
-0.02
-0.13
-0.13
0.089
-4
0.13
0.01
-0.13
-0.13
0.088
-3
0.14
0.04
-0.09
-0.11
0.088
-2
0.15
0.09
-0.05
-0.01
0.088
-1
0.19
0.15
0.00
0.02
0.087
0
0.25
0.20
0.06
0.06
0.087
1
0.32
0.25
0.12
0.14
0.087
2
0.37
0.28
0.18
0.16
0.088
3
0.42
0.30
0.22
0.12
0.088
4
0.47
0.33
0.26
0.09
0.088
5
0.50
0.35
0.31
0.12
0.089
6
0.53
0.37
0.34
0.14
0.089
7
0.52
0.37
0.35
0.14
0.089
8
0.50
0.34
0.33
0.09
0.090
9
0.45
0.29
0.29
0.07
0.090
10
0.39
0.23
0.23
0.03
0.091
11
0.33
0.17
0.15
-0.05
0.091
12
0.26
0.12
0.08
-0.09
0.091
Gray areas indicate the strongest associations for each climate parameter.

Table 5.8 Summary of Wavelet Coherence Analyses by Lag, Direction, and Phase
Cholera Cases Anomaly
Climate Variable
Lag
Direction of Relationship Phase
NIÑO 3.4 SSTA
>6months
Positive
Lead
NIÑO 1+2 SSTA
6 months*
Positive
Lead
PAITA SSTA
>6 months
Positive
Lead
Rainfall A
Zero
Positive
In
* led by 6 months at 1.5 yr scale and 2-3 months at 2 yr scale.

153

5.5

The Social Vulnerability Index (Objective 3)
The aim of this section is to examine the spatial associations between cholera

incidence and social vulnerability by district in Piura in 1997-98. In 1997 and 1998
health authorities in the subregion of Piura reported a total of 4,172 confirmed and
suspected cholera cases across 33 districts. The spatial distribution of district-level
cholera by total number of cases and incidence rate (per 1000 persons at risk) from
1997-98 is displayed in Figure 5.33 and 5.34. Please refer back to Figure 4.6 reference
map for a complete description of the names of each district in the subregion of Piura.
The greatest number of cases were reported in the northern part of the subregion (e.g.,
Piura [n = 1160], Castilla [n = 749] and Chulucanas [n = 525]); while the highest incidence
rates were found in the southern and central parts of the subregion (e.g., Sechura
[11.1/1000] and Rinconada [16.0/1000], and San Juan de Bigote [13.8/1000]).

154

80°0'0"W

6°0'0"S

6°0'0"S

5°0'0"S

5°0'0"S

81°0'0"W

Cholera Cases
1 - 20
21 - 62
63 - 128
129 - 434
435 - 749
0
81°0'0"W

25

750 - 1160

50 Kilometers
80°0'0"W

Figure 5.33 Total cholera cases in Piura for 1997-98, based on natural breaks
classification.

155

80°0'0"W

6°0'0"S

6°0'0"S

5°0'0"S

5°0'0"S

81°0'0"W

Cholera Rate (per 1000)

0
81°0'0"W

25

0.04 - 0.47 (n = 7)
0.48 - 1.19 (n = 4)
1.20 - 2.29 (n = 7)
2.30 - 5.42 (n = 6)
5.43 - 9.51 (n = 5)
9.52 - 15.94 (n = 4)

50 Kilometers
80°0'0"W

Figure 5.34 Cholera incidence rate (per 1000) for 1997-98 in Piura, based on natural
breaks classification.

In order to better understand the spatial distribution of cholera incidence in
Piura, I examined the effect of social vulnerability on cholera. Using PCA analysis, I
constructed a Social Vulnerability Index (SVI). The SVI estimated social vulnerability to
cholera transmission based on factors mainly related to water and sanitation
accessibility. Tables 5.9 and 5.10 show the results from the PCA analysis including the
156

total variance explained and the dominant loadings (>0.50) for each dimension (factor).
Four factors were extracted that explained 78.0 % of the variance in the data. A
description of each factor, their dominant variables and how each variable contributes
to the SVI (e.g.., positive/negative) will follow.

Table 5.9 SVI - Total Variance Explained in the Principal Components Analysis (PCA)
Initial Eigenvalues
% of
Component Total
Variance
1
4.867
37.441
2
2.286
17.582
3
1.529
11.760
4
1.403
10.793
5
.937
7.204
6
.717
5.514
7
.488
3.751
8
.304
2.340
9
.239
1.838
10
.111
.857
11
.085
.654
12
.034
.259
13
.001
.007

Cumulative
%
37.441
55.023
66.784
77.576
84.780
90.294
94.046
96.386
98.224
99.081
99.734
99.993
100.0

157

Rotation Sums of Squared Loadings
% of
Cumulative
Total
Variance
%
3.744
28.802
28.802
2.596
19.968
48.770
2.208
16.982
65.751
1.537
11.825
77.576

Table 5.10 Rotated Component Matrix in the Principal Components Analysis (PCA)
1
2
3
4
ELEM_EDUC
0.945 0.034 0.021 -0.088
BATHRM_NO
0.774 0.226 -0.205 0.256
IMPROV_HSING -0.739 0.297 0.067 -0.119
ILLITERATE
0.730 -0.193 0.038 -0.493
URBAN
-0.638 0.382 0.495 0.320
KITCHEN_NO
0.034 0.947 -0.007 -0.085
KITCHEN_YES
0.105 -0.938 -0.200 -0.014
WATER_TRUCKS -0.086 0.622 0.053 0.119
WATER_WELL
0.314 0.085 0.818 -0.118
KITCHEN_SHARE -0.408 -0.007 0.711 0.300
WATER_RIVER
0.569 -0.288 -0.676 -0.244
WATER_PUMP
-0.396 0.121 0.473 -0.453
WATER_OTHER -0.021 0.044 0.130 0.845
Rotation Method: Kaiser Normalization; bold indicates dominant loadings.

SVI Factor 1 (SVIF1) – Rural (River Water)
The first factor included 6 variables which explained 28.8 % of the variance.
Elementary education (ELEM_EDUC), illiteracy among adults (ILLITERATE), without
bathroom in household (BATHRM_NO), and river water (WATER_RIVER) were positive
variables; and urban (URBAN) and living in improvised housing (IMPROV_HSING) were
negative variables. These loadings may represent districts where rural people may have
limited education and bathroom infrastructure to wash and dispose of waste and thus,
use river water to meet their basic needs. Living without bathroom facilities and using
river water for basic needs may be an important transmission cycle in rural areas.
SVI Factor 2 (SVIF2) – Urban (Water Truck)
The second factor included 4 variables which explained 20.0 % of the variance.
Persons without a kitchen (KITCHEN_NO) and water trucks (WATER_TRUCKS) were
positive; and exclusive kitchen (KITCHEN_YES) was negative. Although urban (URBAN)
158

was not a dominant loading, I included it because it was positive (0.38) and may suggest
that this factor characterizes non-rural areas. Overall these loadings are indicative of
people living in an urban area, without a kitchen, and in need of water trucks to meet
their basic needs. Furthermore, these characteristics are suggestive of people living in
urban shantytowns, without access to proper cooking facilities or access to potable
water, who rely on water trucks. Cholera transmission may have occurred through
contaminated water via water trucks.
SVI Factor 3 (SVIF3) – Urban (Public Well Water)
The third factor included 3 variables which explained 17.0 % of the variance.
Shared kitchen (KITCHEN_SHARE) and public well water (WATER_WELL) were positive;
and river water (WATER_RIVER) was negative. This factor is suggestive of people living
in urban areas in houses with a shared a kitchen, using public well water. Cholera may
have been transmitted in these communities through the common sharing of a kitchen
(i.e., exposure through an infected person in the household) and contaminated well
water.
SVI Factor 4 (SVIF4) – Other Water Sources
The fourth factor included two variables which explained 11.8 % of the variance.
‘Other water sources’ (WATER_OTHER) and urban (URBAN) were positive. This factor is
suggestive of people living in urban areas with access to other water source(s), which
may refer to street vendors. Reportedly, consuming water (as well as food) from street
vendors was an important risk factor for cholera transmission in Piura and Peru (Ries et

159

al 1992). Therefore, particular attention will be given to this factor in the interpretation
of my findings.
In sum, the SVI and its four factors (SVI 1-4) describe the different vulnerability
pathways by which cholera may have been transmitted in the population of Piura. The
next section presents the spatial distribution of the SVI and the four factors (hereafter
referred to as sub-indices) in Piura.

5.5.1 Mapping the SVI by District
Figures 5.35 to 5.39 are maps of the SVI and the sub-indices. The overall SVI
ranged from 1.30 to -1.26. Both the highest and lowest vulnerability was concentrated
on the west coast of Piura (Figure 5.35). The most vulnerable district was Rinconada
Llicuar, which also had the highest estimated cholera incidence in 1997-98 (16.0/1000
persons at risk). Refer to Appendix 4 for Table of the SVI and sub-indices from high to
low social vulnerability compared with cholera incidence rates. The least vulnerable
district was Piura, where cholera incidence was 5.4/1000 persons. The low SVI score
and cholera incidence in Piura might be expected because it is the capital of the
Department of Peru, where a concentration of public resources may have been
available.

160

80°0'0"W

6°0'0"S

6°0'0"S

5°0'0"S

5°0'0"S

81°0'0"W

SVI
< -2.5 Std. Dev.
-2.5 - -1.5 Std. Dev.
-1.5 - -0.50 Std. Dev.
-0.50 - 0.50 Std. Dev.
0.50 - 1.5 Std. Dev.
1.5 - 2.5 Std. Dev.
0
81°0'0"W

25

> 2.5 Std. Dev.

50 Kilometers
80°0'0"W

Figure 5.35 Social Vulnerability Index (SVI), based on standard deviation classification.
Figure 5.36 shows the distribution of rural and river water vulnerability (SVIF1).
The SVIF1 ranged from 1.54 to -3.39. The highest to lowest vulnerability appears to be
contrasted between districts in the eastern part (further inland) of Piura and those in
the west coast of Piura. The most vulnerable districts are: Pacaipampa (SVIF1 = 1.54),
Sondorillo (SVIF1 = 1.44), and Huarmaca (SVIF1 = 1.43). Interestingly, all of these
districts reported low cholera incidence (< 2.0/1000 persons) in 1997-98. Some possible
161

reasons for these observations are discussed further in section 6.3. The least vulnerable
districts were Piura (SVIF1 = -3.39) and its neighbor Castilla (SVIF1 = -2.55), which had an
estimated cholera incidence of 7.4/1000 persons.

80°0'0"W

6°0'0"S

6°0'0"S

5°0'0"S

5°0'0"S

81°0'0"W

SVI Factor 1
< -2.5 Std. Dev.
-2.5 - -1.5 Std. Dev.
-1.5 - -0.50 Std. Dev.
-0.50 - 0.50 Std. Dev.
0.50 - 1.5 Std. Dev.
0
81°0'0"W

25

1.5 - 1.6 Std. Dev.

50 Kilometers
80°0'0"W

Figure 5.36 Rural and River Water (SVIF1), based on standard deviation classification.
Figure 5.37 shows the distribution of urban and water truck vulnerability (SVIF2).
The SVIF2 ranged from 2.59 to – 1.69. The highest vulnerability was concentrated on
162

the west coast, while the lowest vulnerability was in the west coast to central part of
Piura. The most vulnerable districts were: Vice (SVIF2 = 2.59) and Sechura (SVIF2 =
2.12). Cholera incidence in Sechura (11.1/1000 persons) was notable compared to Vice
(3.7/1000 persons). The least vulnerable districts were: El Tallan (SVIF2 = -1.69) in the
west and Santo Domingo (SVIF2 = - 1.47) in the central.

80°0'0"W

6°0'0"S

6°0'0"S

5°0'0"S

5°0'0"S

81°0'0"W

SVI Factor 2
< -1.5 Std. Dev.
-1.5 - -0.50 Std. Dev.
-0.50 - 0.50 Std. Dev.
0.50 - 1.5 Std. Dev.
1.5 - 2.5 Std. Dev.
0
81°0'0"W

25

> 2.5 Std. Dev.

50 Kilometers
80°0'0"W

Figure 5.37 Urban and Water Truck (SVIF2), based on standard deviation classification.
163

Figure 5.38 shows the distribution of urban and well water vulnerability (SVIF3).
The SVIF3 ranged from 2.66 to -1.56. The highest vulnerability was found in the west
coast, while the lowest vulnerability was in the coast and central Piura. The most
vulnerable districts were La Matanza (SVIF3 = 2.66) and El Tallan (SVIF3 = 2.45). Cholera
incidence in these two districts was <2.5/1000 persons. The least vulnerable districts
were La Laquiz (SVIF3 = -1.56) and El Carmen de la Frontera (SVIF3 = -1.41).
80°0'0"W

6°0'0"S

6°0'0"S

5°0'0"S

5°0'0"S

81°0'0"W

SVI Factor 3
< -1.5 Std. Dev.
-1.5 - -0.50 Std. Dev.
-0.50 - 0.50 Std. Dev.
0.50 - 1.5 Std. Dev.
0
81°0'0"W

25

> 1.5 Std. Dev.

50 Kilometers
80°0'0"W

Figure 5.38 Urban and Well Water (SVIF3), based on standard deviation classification.
164

Figure 5.39 shows the distribution of ‘other water sources’ vulnerability. The
SVIF4 ranged from 4.60 to -1.34. The highest vulnerability was found on the west coast.
The lowest vulnerability was distributed throughout the west coast, central and eastern
parts of Piura. The most vulnerable district was Rinconada Llicuar, which was also the
most vulnerable according to the overall SVI. The least vulnerable district was La Arena
(SVIF4 = -1.34). La Arena had an estimated cholera incidence of 3.9/1000 persons.
80°0'0"W

6°0'0"S

6°0'0"S

5°0'0"S

5°0'0"S

81°0'0"W

SVI Factor 4
< -0.50 Std. Dev.
-0.50 - 0.50 Std. Dev.
0.50 - 1.5 Std. Dev.
1.5 - 2.5 Std. Dev.
0
81°0'0"W

25

> 2.5 Std. Dev.

50 Kilometers
80°0'0"W

Figure 5.39 Other Water Sources (SVIF4), based on standard deviation classification.
165

5.6

Measuring Associations between the SVI and Cholera Incidence (Objective 4)
The results of the global regression (OLS) analyses estimating the effect of SVI

and SVIF1-4 on cholera incidence are presented in Table 5.11. These analyses show
that the SVI is significantly associated with cholera incidence (B= 0.65, p-value = 0.07) at
the 0.01 level. Of the SVI sub-indices, using ‘water from other sources’ (SVIF4) had a
significant and positive association with cholera incidence (B = -0.46, p-value = 0.008).
In contrast, living in rural areas and using water from the river (SVIF1) had a significant
and negative association with cholera incidence (B= -0.46, p-value = 0.008).

Table 5.11 Summary Statistics for Ordinary Least Squares Regression (OLS)
Coefficient Standardized Error t value p value
Intercept
1.7712
0.1735
10.2073 0.0000
SVI
0.6500
0.3524
1.8443 0.0747
Intercept
SVIF1

1.7712
-0.4636

0.1634
0.1659

10.8417 0.0000
-2.7947 0.0088

Intercept
SVIF2

1.7712
0.2856

0.1757
0.1784

10.0821 0.0000
1.6009 0.1195

Intercept
SVIF3

1.7712
0.3087

0.1745
0.1772

10.1529 0.0000
1.7423 0.0914

Intercept
1.7712
0.1580
11.2074 0.0000
SVIF4
0.5194
0.1605
3.2362 0.0029
Bold indicates variables that are significant at 95.0% confidence level.
2

Table 5.12 compares the global model and local model estimates (GWR] by r
and residuals. Generally the GWR results supported the OLS findings except for the

association with overall SVI (i.e., was not significant). SVIF1 and SVIF4 had significant
166

associations with cholera incidence. Compared to the other indices the variance
explained by SVIF1 (R2 = 0.45) and SVIF4 (R2 = 0.46) was substantially greater.
Furthermore, their residuals were also relatively smaller, suggesting a closer fit between
the predicted values and the observed data.

Table 5.12 Comparison Statistics for Geographically Weighted Regression (GWR) and
Ordinary Least Squares Regression (OLS)
AICc
R-Squared
Adj. R-Squared Residuals
Variable GWR OLS
GWR
OLS
GWR
OLS
GWR
SVI
89.68 95.38 0.4089 0.0989 0.3230 0.0698
20.20
SVIF1
86.62 91.40 0.4515 0.2012 0.3789 0.1755
18.75
SVIF2
89.93 96.19 0.4271 0.0764 0.3299 0.0466
19.58
SVIF3
92.20 95.73 0.3548 0.0892 0.2634 0.0598
22.06
SVIF4
85.08 89.21 0.4617 0.2525 0.3995 0.2284
18.40
Bold indicates variables that are significant at the 95.0% confidence level
5.6.1 Mapping Associations between SVI and Cholera Incidence
Figures 5.40 to 5.45 are maps of the predicted values and standardized residuals
20

for SVI , SVIF1 and SVIF4 from the GWR analyses. The predicted values represent the
local effect of social vulnerability (specified by the defined mechanisms) on cholera
incidence by districts. The standardized residuals represent risks factors other than
social vulnerability –i.e., unexplained risk factors for cholera incidence across districts.
For SVI and the sub-indices, there was a consistent spatial pattern in predicted values;
such that, estimations are distinctly higher to lower from the west coast to the eastern
part of Piura. This pattern was also observed in the standardized residuals maps;
20

I included the SVI for comparative purposes since it represents overall social
vulnerability.
167

however, values from west to east were generally moderate (Std. Dev. < 1.5). This
finding suggests that there may be an association between areas where local cholera
rate estimations were high to moderate and areas where much remains unexplained by
the GWR models. Thus, these districts could be important places of inquiry to
investigate climate associations with cholera in 1997-98.

168

80°0'0"W

6°0'0"S

6°0'0"S

5°0'0"S

5°0'0"S

81°0'0"W

Predicted
1.17 - 1.28 (n = 4)
1.29 - 1.39 (n = 8)
1.40 - 1.47 (n = 5)
1.48 - 1.86 (n = 4)
0

25

1.87 - 2.42 (n = 8)

50 Kilometers

2.43 - 3.01 (n = 4)
81°0'0"W

80°0'0"W

Figure 5.40 Cholera Incidence Predictions by Social Vulnerability Index (SVI). Predicted
values, based on natural breaks classification, from geographically weighted regression
(GWR).

169

80°0'0"W

Std. Residuals
< -2.5 Std. Dev.
-2.5 - -1.5 Std. Dev.
-1.5 - -0.5 Std. Dev.
-0.5 - 0.5 Std. Dev.
0.5 - 1.5 Std. Dev.
0

25

1.5 - 2.5 Std. Dev.

50 Kilometers

> 2.5 Std. Dev.
81°0'0"W

80°0'0"W

Figure 5.41 Cholera Incidence Predictions by Social Vulnerability Index (SVI).
Standardized residuals, based on the standard deviation classification, from
geographically weighted regression (GWR).

170

6°0'0"S

6°0'0"S

5°0'0"S

5°0'0"S

81°0'0"W

80°0'0"W

6°0'0"S

6°0'0"S

5°0'0"S

5°0'0"S

81°0'0"W

Predicted
1.17 - 1.28 (n = 4)
1.29 - 1.39 (n = 8)
1.40 - 1.47 (n = 5)
1.48 - 1.86 (n = 4)
0

25

1.87 - 2.42 (n = 8)

50 Kilometers

2.43 - 3.01 (n = 4)
81°0'0"W

80°0'0"W

Figure 5.42 Cholera Incidence Predictions by Social Vulnerability Index Factor 1 (SVIF1).
Predicted values, based on natural breaks classification, from geographically weighted
regression (GWR).

171

80°0'0"W

Std. Residuals

6°0'0"S

6°0'0"S

5°0'0"S

5°0'0"S

81°0'0"W

< -2.5 Std. Dev.
-2.5 - -1.5 Std. Dev.
-1.5 - -0.5 Std. Dev.
-0.5 - 0.5 Std. Dev.
0.5 - 1.5 Std. Dev.
0

25

1.5 - 2.5 Std. Dev.

50 Kilometers

> 2.5 Std. Dev.
81°0'0"W

80°0'0"W

Figure 5.43 Cholera Incidence Predictions by Social Vulnerability Index Factor 1 (SVIF1).
Standardized residuals, based on the standard deviation classification, from
geographically weighted regression (GWR).

172

80°0'0"W

6°0'0"S

6°0'0"S

5°0'0"S

5°0'0"S

81°0'0"W

Predicted
0.91 - 1.04 (n = 3)
1.05 - 1.32 (n = 6)
1.33 - 1.69 (n = 11)
1.70 - 2.15 (n = 8)
0

25

2.16 - 2.68 (n = 4)

50 Kilometers

2.69 - 3.90 ( n = 1)
81°0'0"W

80°0'0"W

Figure 5.44 Cholera Incidence Predictions by Social Vulnerability Index Factor 4 (SVIF4).
Predicted values, based on natural breaks classification, from geographically weighted
regression (GWR).

173

80°0'0"W

Std. Residuals

6°0'0"S

6°0'0"S

5°0'0"S

5°0'0"S

81°0'0"W

< -2.5 Std. Dev.
-2.5 - -1.5 Std. Dev.
-1.5 - -0.5 Std. Dev.
-0.5 - 0.5 Std. Dev.
0.5 - 1.5 Std. Dev.
0

25

1.5 - 2.5 Std. Dev.

50 Kilometers

> 2.5 Std. Dev.
81°0'0"W

80°0'0"W

Figure 5.45 Cholera Incidence Predictions by Social Vulnerability Index Factor 4 (SVIF4).
Standardized residuals, based on the standard deviation classification, from
geographically weighted regression (GWR).

Districts where high (red) to moderate (orange –light orange) predicted values
and high standardized residuals overlapped are highlighted in Table 5.13. These
districts included (by order of ID): 1) Piura (0); Castilla (1); Catacaos (2); La Arena (5); La
174

Union (6); Huarmaca (12); Chulucanas (17); Chalaco (19); Salitral1 (22); San Juan de
Bigote (23); Sechura (27); Bernal (29); Cristo Nos Valga (30); Vice (31); and Rinconada
Llicuar (32). Twelve of these areas were identified in 3 of 4 indices. Although
predicted values for cholera in San Juan de Bigote were moderate, it was included
because residuals for this district were consistently high (Std. Dev. > 2.5) across indices.
It also reported the second highest cholera rate (13.8 per 1000) in 1997-98. In the next
section, cholera incidence in these selected districts will be examined for associations
with global and local climate parameters.

Table 5.13 Selected Districts from Geographically Weighted Regression (GWR) based on
Predicted Values and Standardized Residuals
ID
0
1
2
5
6
12
17
19
22
23
27
29
30
31

SVI
PIURA
CASTILLA
CATACAOS
LA ARENA
LA UNION

SVIF1

CHULUCANAS
SALITRAL
SAN JUAN DE
BIGOTE
SECHURA
BERNAL
CRISTO NOS
VALGA

RINCONADA
32 LLICUAR

CHALACO
SALITRAL1
SAN JUAN DE
BIGOTE
SECHURA
BERNAL
CRISTO NOS
VALGA
VICE
RINCONADA
LLICUAR

SVIF4
PIURA
CASTILLA
CATACAOS

LA UNION

SVIF3
PIURA
CASTILLA
CATACAOS
LA ARENA
LA UNION

CHULUCANAS

CASTILLA
CATACAOS
LA ARENA
LA UNION
HUARMACA

SVIF2

CHULUCANAS

CHULUCANAS

SALITRAL1
SAN JUAN DE
BIGOTE
SECHURA
BERNAL
CRISTO NOS
VALGA

SALITRAL1
SAN JUAN DE
BIGOTE
SECHURA
BERNAL
CRISTO NOS
VALGA

SALITRAL1
SAN JUAN DE
BIGOTE
SECHURA

RINCONADA
LLICUAR

RINCONADA
LLICUAR

RINCONADA
LLICUAR

CATACAOS

175

LA UNION

CRISTO NOS
VALGA

5.7

Measuring Climate Associations with Cholera by District (Objective 4)
In this section I examine the associations between cholera incidence and global

and climate parameters by districts in 1998. Figure 5.46 is monthly cholera incidence
(per 100,000 persons) for Piura in 1997-98. The graph illustrates that cholera incidence
increased sharply in December of 1997 after a decline that year in the number of
reported cases (<40). Cholera incidence peaked in February and March of 1998 and
then gradually declined throughout the remainder of the year. In 1998, most districts in
the study area observed this pattern of cholera incidence, i.e., sharp rise in SH summer,
fall in SH winter (See Figures 5.47 to 5.52 for monthly cholera incidence per 1000
persons by districts).

176

Cholera Rate (per 100000)

15

10

5

J

F M A M J

J A S O N D J F M A M J
1997-98

J A S O N D

Figure 5.46 Cholera Incidence Rate per 100,000 (square-root transformed) for Piura in 1997-98 by month.

177

2.0
1.8

Cholera Rate (per 1000)

1.6
1.4
1.2
1.0
0.8
0.6
0.4
0.2
0.0
J

F

M

A

M

PIURA

CASTILLA

J
CATACAOS

J

A
CURA MORI

S

O

N

EL TALLAN

Figure 5.47 Cholera Incidence Rate per 1000 (square-root transformed) by district and month in 1998.

178

D

2.0
1.8

Cholera Rate (per 1000)

1.6
1.4
1.2
1.0
0.8
0.6
0.4
0.2
0.0
J

F

M
LA ARENA

A

M

LA UNION

J
FRIAS

J

A

PACAIPAMPA

S

O

N

HUANCABAMBA

Figure 5.48 Cholera Incidence Rate per 1000 (square-root transformed) by district and month in 1998.

179

D

0.9
0.8

Cholera Rate (per 1000)

0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
J
CANCHAQUE

F

M

A

M

EL CARMEN DE LA FRONTERA

J
HUARMACA

J

A
LALAQUIZ

S

O

N

SAN MIGUEL DE EL FAIQUE

Figure 5.49 Cholera Incidence Rate per 1000 (square-root transformed) by district and month in 1998.

180

D

1.6
1.4

Cholera Rate (per 1000)

1.2
1.0
0.8
0.6
0.4
0.2
0.0
J
SONDOR

F

SONDORILLO

M

A

CHULUCANAS

M

J

BUENOS AIRES

J

A

CHALACO

S

O

LA MATANZA

N

MORROPON

Figure 5.50 Cholera Incidence Rate per 1000 (square-root transformed) by district and month in 1998.

181

D

Figure 5.51 Cholera Incidence Rate per 1000 (square-root transformed) by district and month in 1998.

182

3.0

Cholera Rate (per 1000)

2.5
2.0
1.5
1.0
0.5
0.0
J

F

M

BELLAVISTA DE LA UNION

A
BERNAL

M

J

J

A

CRISTO NOS VALGA

VICE

S

O

N

RINCONADA LLICUAR

Figure 5.52 Cholera Incidence Rate per 1000 (square-root transformed) by district and month in 1998.

183

D

In 1997-98, the rise in cholera incidence coincided with the strongest El Niño of
the century, which had a strong influence on regional and local climate in Piura. In
Figure 5.53, a rise in SSTA (Niño 1+2, Niño 3.4 and Paita) was observed in the SH
summer of 1997. SSTA anomalies remained above 1°C until the next summer in 1998.
Local mean and minimum temperature anomalies in Piura also show a pattern similar to
SSTA in 1997-98 (Figure 5.54). Although a rise in anomalies was observed for local
maximum temperature in Piura too, it was followed by a decline from September 1997
to January 1998 (Figure 5.54). Figure 5.55 is a graph of monthly rainfall anomaly (mm)
in Piura in 1997-98. Rainfall’s pattern during this time appeared similar to cholera
incidence; rainfall total rose sharply in the SH summerand subsequently declined as the
SH winter approached.

184

5
NINO 1+2
NINO 3.4
PAITA

4

SSTA (degC)

3
2
1
0
-1
-2
J

F M A M J

J A S O N D J F M A M J
1997-98

J A S O N D

Figure 5.53 Sea Surface Temperature Anomaly (SSTA) by month for Niño 3.4, Niño 1+2, and Paita in 1997-98.

185

5
TMAXA
TMEANA
TMINA

TEMPERATURE (degC)

4
3
2
1
0
-1
-2
J

F M A M J

J A S O N D J F M A M J
1997-98

J A S O N D

Figure 5.54 Temperature Maximum (TMAXA), Mean (TMEANA), and Minimum (TMINA) Anomalies by month in 1997-98.

186

30

RAINFALL (mm)

25
20
15
10
5
0
J

F M A M J

J A S O N D J F M A M J
1997-1998

Figure 5.55 Rainfall (square-root transformed) by month in 1997-98.

187

J A S O N D

The spatial associations between climate parameters and monthly cholera
incidence by districts (n = 13) were examined using OLS regression. Thirteen districts
were selected for a monthly analysis of 1998 based on (a) the GWR analysis; (b) the
number of reported cases (>40); and (c) the number of months of reported cholera
(=>5).

21

Figure 5.56 is a map of the selected districts for the climate-cholera analysis. A

total of 689 regressions were performed. Approximately 47.0% (324) of associations
were statistically significant at the 0.05 level. All the associations were positive and the
2

strength of these associations (measured by r ) varied by climate variable, temporal lag
2

and district. See Appendix 5 for a table that shows the r and p values (<0.05) by
district. The climate parameters with the greatest number of associations were
TMEANA (n = 63) and TMINA (n = 61). The climate parameters with the lowest number
of associations were rainfall (n = 33) and TMAXA (n = 34). The districts with the most
associations were La Union (n = 41), La Arena (n = 38), Sechura (n = 38), and Salitral1 (n
= 37). The least number of associations were found in Huancabamba (n = 9), Rinconada
Llicuar (n = 6), and Chalaco (n = 3). In San Juan de Bigote no statistically significant
associations were observed. The most common time lags were 0-2 months; the least
common was 7 months.

21

I initially began with 15 districts from the GWR analysis and then eliminated 3 districts
based on the criteria discussed above. I included the district of Huancabamba even
though it was not selected in the GWR analysis because this district reported cholera
cases 9 out of 12 months.
188

0

5°0'0"S

80°0'0"W

5°0'0"S

81°0'0"W

19
17

1

9
6

23

5
32

22

2

6°0'0"S

6°0'0"S

12

27

0

25

50 Kilometers

81°0'0"W

80°0'0"W

District
ID District
PIURA
0 CHULUCANAS
CASTILLA
1 CHALACO
CATACAOS
2 SALITRAL1
LA ARENA
5 SAN JUAN DE BIGOTE
LA UNION
6 SECHURA
HUANCABAMBA 9 RINCONADA LLICUAR
HUARMACA
12

ID
17
19
22
23
27
32

Figure 5.56 Map of Selected Districts for Climate-Cholera Analysis.

189

5.7.1 Mapping Climate Associations with Cholera by District
The climate parameters and temporal lags (>0) which exhibited the strongest
2

associations (r > 0.70) were Niño 3.4 SSTA (1 month lag), Paita SSTA (1 month lag),
Rainfall (1 month lag), Rainfall (2 month lag), TMAXA (6 month lag), and TMEANA (6
month lag). The climate parameter with the weakest associations was Niño 1+2 SSTA.
Associations with lags greater than zero were highlighted because of their potential
usefulness to policymakers for early warning systems. Figures 5.57 to 5.60 are maps of
cholera incidence associations with Niño 3.4 SSTA (1 month lag) and Paita SSTA (1
month lag). The strongest associations for these climate parameters were in west coast
districts, such as Piura, La Arena, La Union and Sechura. These same districts are also
places where cholera was strongly associated with rainfall (1 month lag) (Figures 5.61
and 5.62). A cholera-rainfall (1 month lag) link is also strong in two additional west
coast districts, Castilla and Catacaos. At a 2 month lag (Figures 5.63 and 5.64), cholera’s
association with rainfall has greater spatial variability; and strong associations shift
eastward. Here, Chulucanas and Salitral1 demonstrate strong associations, whereas La
Union remains the only strong association on the west coast. The strongest associations
between cholera and TMAXA and TMEANA were found at a 6 month lag (Figures 5.65 to
5.68). Both parameters had the best associations with cholera incidence on the west
coast in the districts of La Union, La Arena, and Sechura 22. There were also strong

22

2

Sechura (r = 0.63).
190

associations between cholera incidence and TMEANA in Huarmaca (eastern Piura) and
Chulucanas (central Piura).

191

80°0'0"W

6°0'0"S

6°0'0"S

5°0'0"S

5°0'0"S

81°0'0"W

NINO 3.4 R2

0
81°0'0"W

25

0.1241 - 0.1821
0.1822 - 0.2757
0.2758 - 0.4561
0.4562 - 0.6000
0.6001 - 0.7401

50 Kilometers
80°0'0"W

Figure 5.57 Niño 3.4 Sea Surface Temperature Anomaly (SSTA) (1 month lag)
Associations with Cholera Incidence Rates (per 1000), R-squared values based on natural
breaks classification.

192

80°0'0"W

6°0'0"S

6°0'0"S

5°0'0"S

5°0'0"S

81°0'0"W

NINO 3.4 P-Value
0.0003 - 0.0031
0.0032 - 0.0257
0.0258 - 0.0383
0.0384 - 0.0500
0
81°0'0"W

25

0.0501 - 0.2614

50 Kilometers
80°0'0"W

Figure 5.58 Niño 3.4 Sea Surface Temperature Anomaly (SSTA) (1 month lag)
Associations with Cholera Incidence Rates (per 1000), P values based on natural breaks
classification.

193

80°0'0"W

6°0'0"S

6°0'0"S

5°0'0"S

5°0'0"S

81°0'0"W

PAITA R2
0.1360 - 0.1378
0.1379 - 0.2622
0.2623 - 0.5001
0.5002 - 0.6000
0
81°0'0"W

25

0.6001 - 0.7320

50 Kilometers
80°0'0"W

Figure 5.59 Paita Sea Surface Temperature Anomaly (SSTA) (1 month lag) Associations
with Cholera Incidence Rates (per 1000), R-squared values based on natural breaks
classification.

194

80°0'0"W

6°0'0"S

6°0'0"S

5°0'0"S

5°0'0"S

81°0'0"W

PAITA P-Value
0.0004 - 0.0054
0.0055 - 0.0120
0.0121 - 0.0232
0.0233 - 0.0500
0
81°0'0"W

25

0.0501 - 0.2382

50 Kilometers
80°0'0"W

Figure 5.60 Paita Sea Surface Temperature Anomaly (SSTA) (1 month lag) Associations
with Cholera Incidence Rates (per 1000), P values based on natural breaks classification.

195

80°0'0"W

6°0'0"S

6°0'0"S

5°0'0"S

5°0'0"S

81°0'0"W

RAINFALL R2
0.1745 - 0.1991
0.1992 - 0.4503
0.4504 - 0.5000
0.5001 - 0.7000
0
81°0'0"W

25

0.7001 - 0.8863

50 Kilometers
80°0'0"W

Figure 5.61 Rainfall Total (1 month lag) Associations with Cholera Incidence Rates (per
1000), R-squared values based on natural breaks classification.

196

80°0'0"W

6°0'0"S

6°0'0"S

5°0'0"S

5°0'0"S

81°0'0"W

RAINFALL P-Value
0.0000 - 0.0001
0.0002 - 0.0037
0.0038 - 0.0179
0.0180 - 0.0500
0
81°0'0"W

25

0.0501 - 0.1767

50 Kilometers
80°0'0"W

Figure 5.62 Rainfall Total (1 month lag) Associations with Cholera Incidence Rates (per
1000), P values based on natural breaks classification.

197

80°0'0"W

6°0'0"S

6°0'0"S

5°0'0"S

5°0'0"S

81°0'0"W

RAINFALL R2
0.3308
0.3309 - 0.4471
0.4472 - 0.5533
0.5534 - 0.7000
0
81°0'0"W

25

0.7001 - 0.8613

50 Kilometers
80°0'0"W

Figure 5.63 Rainfall Total (2 month lag) Associations with Cholera Incidence Rates (per
1000), R-squared values based on natural breaks classification.

198

80°0'0"W

6°0'0"S

6°0'0"S

5°0'0"S

5°0'0"S

81°0'0"W

RAINFALL P-Value
0.0000 - 0.0002
0.0003 - 0.0019
0.0020 - 0.0100
0.0101 - 0.0500
0
81°0'0"W

25

0.0501 - 0.0504

50 Kilometers
80°0'0"W

Figure 5.64 Rainfall Total (2 month lag) Associations with Cholera Incidence Rates (per
1000), P values based on natural breaks classification.

199

80°0'0"W

6°0'0"S

6°0'0"S

5°0'0"S

5°0'0"S

81°0'0"W

TMAXA R2
0.2062 - 0.2768
0.2769 - 0.4265
0.4266 - 0.5000
0.5001 - 0.6272
0
81°0'0"W

25

0.6273 - 0.7917

50 Kilometers
80°0'0"W

Figure 5.65 Temperature Maximum Anomaly (TMAXA) (6 month lag) Associations with
Cholera Incidence Rates (per 1000), R-squared values based on natural breaks
classification.

200

80°0'0"W

6°0'0"S

6°0'0"S

5°0'0"S

5°0'0"S

81°0'0"W

TMAXA P-Value
0.0001 - 0.0021
0.0022 - 0.0091
0.0092 - 0.0151
0.0152 - 0.0500
0
81°0'0"W

25

0.0501 - 0.1381

50 Kilometers
80°0'0"W

Figure 5.66 Temperature Maximum Anomaly (TMAXA) (6 month lag) Associations with
Cholera Incidence Rates (per 1000), P values based on natural breaks classification.

201

80°0'0"W

6°0'0"S

6°0'0"S

5°0'0"S

5°0'0"S

81°0'0"W

TMEANA R2
0.2626
0.2627 - 0.3791
0.3792 - 0.5000
0.5001 - 0.6000
0
81°0'0"W

25

0.6001 - 0.7488

50 Kilometers
80°0'0"W

Figure 5.67 Temperature Mean Anomaly (TMEANA) (6 month lag) Associations with
Cholera Incidence Rates (per 1000), R-squared values based on natural breaks
classification.

202

80°0'0"W

6°0'0"S

6°0'0"S

5°0'0"S

5°0'0"S

81°0'0"W

TMEANA P-Value
0.0003 - 0.0021
0.0022 - 0.0053
0.0054 - 0.0105
0.0106 - 0.0500
0
81°0'0"W

25

0.0501 - 0.0885

50 Kilometers
80°0'0"W

Figure 5.68 Temperature Mean Anomaly (TMEANA) (6 month lag) Associations with
Cholera Incidence Rates (per 1000), P values based on natural breaks classification.

In sum, associations between cholera and climate parameters were strongest on
the west coast. The time lag associated with local temperature anomaly link to cholera
was greater (6 months) than those associated with SSTA and rainfall (1 month). In terms
of parameters, rainfall had the strongest association (p > 0.85) with cholera incidence
among all climate parameters for several districts.
203

6. CHAPTER 6: DISCUSSION
In my dissertation research I investigated the temporal and spatial associations
between ENSO and cholera incidence in Piura, Peru from 1991 to 2001. I also examined
whether the temporal associations were stronger after 1992 and whether social
vulnerability could explain the spatial variability of the ENSO-cholera associations during
the 1997-98 El Niño. In this chapter, I first discuss the key findings of this dissertation in
order to answer my research questions and hypotheses. I then present a reflective
statement about the ethical geographies of my dissertation results and the overall
research project.

6.1

The ENSO Context
The comparative examination of ENSO events by Niño region revealed that there

were important differences and similarities between the Niño 3.4 and Niño 1+2 indices
throughout the 1990s. From 1990 to 1995, El Niño conditions were more prevalent in
the Niño 3.4 region compared to the Niño 1+2 region, where La Niña conditions were
the dominant mode. There were also differences in the timing of the onset of El Niño
conditions in each of the regions. For instance, during the first year of the cholera
epidemic the Niño 3.4 index indicated that an El Niño began in May of 1991 and lasted
until July of 1992. However, according to the Niño 1+2 index, it was shown that an El
Niño began in November of 1991 and lasted until June of 1992. Of import here is that
each region’s ENSO characteristics, while different, commonly suggested that El Niño
developed well after the onset of the cholera epidemic in October of 1990 (based on
204

Seas et al. [2001]). Similarly, in 1997, the two Niño regions have different onset dates
(i.e., March versus May of 1997), but also show and support the notion that El Niño
conditions were present during the onset of the cholera resurgence. In the subsequent
sections, I interpret my results given this ENSO context.

6.2

Temporal Associations between Cholera and ENSO in Piura, Peru: 1991-2001
In Objective 2, I demonstrated the existence of temporal associations between

cholera cases in Piura and global and local sea surface temperature anomalies (i.e., Niño
3.4 SSTA, Niño 1+2 SSTA and Paita SSTA) and local rainfall anomaly. These associations
measured as the wavelet coherence between cholera and climate can be summarized as
non-stationary positive relationships, where cholera and climate parameters were
correlated and co-varied in localized time frequency space. Significant coherence
between cholera and sea surface temperature anomalies were detected at two time
intervals at a scale band of 1.5-2 yr as follows: (a) cholera and Niño 3.4 SSTA – 1993 and
1997 to 1999; (b) cholera and Niño 1+2 SSTA – 1993-94 and 1997 to 1999; and (c)
cholera and Paita SSTA – 1993-94 and 1998-99. Significant coherence between cholera
and local rainfall anomaly was detected from 1997 to 1999 at several scales ranging
from 7 months to 2.25 yr. Although an association between cholera and local
temperature anomalies (i.e., maximum, mean and minimum) was found, their phase
relationships (i.e., direction of association) were inconsistent across scales, suggesting
that these associations were dubious and therefore, unreliable (Grinsted et al. 2004).
It was also shown that the temporal associations between cholera and global and local

205

SSTA, and local rainfall anomalies, were positive and lagged in time such that cholera
followed each climate series as follows by time interval: (i) in 1993-94 – Niño 3.4 SSTA
and Paita SSTA by more than 6 months; and Niño 1+2 SSTA by 6 months; (ii) in 1997-99 Niño 3.4 SSTA and Paita SSTA by more than 6 months; Niño 1+2 by 2-3 months; and local
rainfall anomaly by zero to one month.

6.2.1 Cholera and Global Climate
On a global scale, the exposure pathways associated with the Niño 3.4 region
suggested that sea surface temperature anomalies contributed to cholera incidence in
Piura through a six-month process (i.e., the time lag) via ecological impacts on: (a)
distant marine conditions which in turn affected vibrios or aquatic reservoirs off the
coast, supporting previous studies in Dhaka, Bangladesh, which found that cholera
fluctuated with Niño 3.4 SST (Pascual et al. 2000; Rodo et al. 2002; Koelle et al. 2005);
and/or (b) local climate inland (i.e., via teleconnections) which in turn may have
influenced vibrio reproduction (e.g., temperature – Pascual et al. 2000), distribution
(e.g., rainfall – Ruiz-Moreno et al. 2007), or transport (e.g., rainfall – Ruiz-Moreno et al.
2007).
Two key factors associated with the pathways described above are the time lag
and the distance from the ocean-climate source of impact to the vibrio source of
infection. In a study by Pascual et al. (2000), it was suggested that cholera incidence
followed Niño 3.4 SST by 8 to11 months because of the distance (of the teleconnection)
between Dhaka, Bangladesh and the central equatorial Pacific Ocean. Therefore, the
206

time relationship found between cholera incidence and Niño 3.4 SSTA in this
dissertation research may be explained by Piura’s geographic proximity to air-sea
interactions in the central equatorial Pacific Ocean. Thus, the shorter time delay (i.e., 6
months +, but less than a year) observed in Piura may be explained by the shorter
distance to the Niño 3.4 region.
While distance may explain part of the temporal link between cholera and Niño
3.4, it cannot fully explain cholera’s link with the Niño 1+2 region. Niño 1+2 represents
sea surface temperature conditions off the coast of northern Peru and southern coast of
Ecuador. Given the proximity of this Niño region, one could assume based on the Niño
3.4 finding that the shorter distance to the climate source of impact would also suggest
a shorter time delay. Although this study did find a shorter time lag (i.e., 2-3 months)
between cholera and Niño 1+2 SSTA from 1997 to 1999, it also found that Niño 1+2 lead
cholera by 6 months in 1993-94, just as the Niño 3.4 estimates indicated. This finding
may suggest that understanding the temporal link requires a closer look at the
geography of the region.
First, the characteristics of each region in terms of sea surface temperature
conditions at each time interval could reveal additional information to help understand
the temporal pathway associated with Niño 1+2 SSTA (see Table 5.1). For example,
interestingly, in 1993 it was estimated that both regions were in neutral conditions.
During that year cholera cases followed the seasonal pattern observed nationally by
public health authorities in Peru (i.e., up in summer, down in winter, up as summer
approaches – see Table 5.3). It was not until the spring of 1994 that the SST conditions

207

in the two regions began to diverge; an El Niño developed in the Niño 3.4 region, while a
La Niña developed in the Niño 1+2 region. Actually, it appeared that cholera cases
declined with the onset of ENSO events. In contrast, in 1997 to 1999, most monthly sea
surface temperature conditions were either El Niño or La Niña in each region.
Furthermore, these conditions were associated with strong El Niño and La Niña events
and the cholera resurgence. This observation may suggest that the difference in the
magnitude of anomalies from 1997 to 1999 may have had a stronger influence on the
ecology of cholera in Piura, and therefore, should also be studied in order to understand
temporal associations between cholera and SST.
Second, in addition to sea surface temperature conditions in the region, it may
also be important to consider the region’s proximity to the South American continent
and Piuran population. This is not to suggest that distance is a factor again; rather it is
to propose that coastal land features coupled with population factors (e.g.,
sociodemographic) on the immediate coast may explain the differences between the
Niño 3.4 and Niño 1+2 regions and their associations with cholera. Unlike the Niño 3.4
region, the Niño 1+2 region is adjacent to the South American continent. Importantly it
borders a low lying desert environment inhabited by coastal communities of Piura
where livelihoods consist of fishing and farming. This may support Epstein’s initial
theory about climate and cholera which suggests that run-off from agricultural activities,
as one example, may have contributed to the plankton blooms on the coast of Peru,
which subsequently lead to transmission inland (1993). It highlights that the complexity
of local topography, climate and human-environment interactions on Piura’s coast

208

contributed to different temporal associations between cholera and the eastern end
and central equatorial Pacific Ocean.

6.2.2 Cholera and Local Climate
On a local scale, it was shown that cholera was temporally associated with sea
surface temperature anomalies at Paita in 1993-94 and 1997 to 1999. As discussed
earlier, Paita is a port in Piura. Although it is located on the immediate coast, the SST
impact on cholera was delayed by more than 6 months. This observation reiterates the
importance of locality in the temporal process of cholera transmission as stated above
because it provides evidence that local SST associations previously found in Lima (e.g.,
time lags were 0-1 month – Gil et al. [2004]) are different by location. The association
at Paita may in part be explained by the influence of the Niño 3.4 region on Paita SSTA.
Paita is highly sensitive to changes in ocean-atmosphere interactions in the equatorial
Pacific Ocean, namely El Niño (Rodriguez 2005). A wavelet coherence analysis (data not
shown) of Paita and Niño 3.4 indicated that there existed strong teleconnections
between this global and local climate parameter throughout most of the 1990s at a
range of time lags of zero to 1 month. This association was most evident in 1997-98.
In addition to SST impacts, this study also found a temporal association between
cholera and rainfall anomaly in Piura during the resurgence of cholera in 1997-98.
Rainfall was approximately in phase with cholera at multiple scales suggesting that
rainfall may have affected cholera transmission at several time periods through 199798. This is an important finding because it provides quantitative evidence that

209

substantiates reports by PAHO during that time period of collapsed infrastructure from
flooding and subsequent contamination of the water supply in Piura (see Chapter 3).
Although it can be assumed that heavy rainfall was the exposure pathway, it should also
be noted that rainfall can play a dual role in cholera transmission (Kovats et al. 2003;
Ruiz-Moreno et al. 2007). Therefore, rainfall deficit may also have impacted cholera
transmission through its potential effects on water supply (Akanda et al. 2008). Lastly, it
should be noted that there was a strong interannual relationship (e.g., 3 to 8 years)
between rainfall in Piura and Niño 3.4 SSTA, Niño 1+2 SSTA, and Paita SSTA throughout
the 1990s (based on wavelet coherence and cross-wavelet data not shown). I draw
attention to this association because it may potentially explain the interrelated links
among global climate, local climate, and local cholera incidence in Piura, Peru. Studies
by Pascual et al. (2000) and Bouma and Pascual (2001) suggest that this is a potential
mechanism that mediates the climate and cholera relationships in Bangladesh and India.
Therefore, it warrants further investigation.

6.2.3 Was there a temporal association between cholera incidence in Piura and
ENSO in the 1990s? Was this association stronger after 1992?
In this dissertation research, I demonstrated that global and local sea surface
temperature anomalies and local rainfall anomaly positively influenced cholera
incidence in Piura in the 1990s. The scales of these relationships suggested that there
was an interannual component to cholera’s temporal variability following the second
wave of the cholera epidemic in 1992 (i.e., 1993-94) and during the resurgence of

210

cholera in 1997-98. Furthermore, there was an intrannual component of cholera
associated with local rainfall anomaly from 1997 to 1999. The wavelet results along
with the comparison of ENSO events by Niño 3.4 and Niño 1+2 regions strongly suggest
that the 1997-98 El Niño and a subsequent La Niña (i.e., estimated in both regions) were
temporally associated with cholera incidence from 1997 to 1999. These findings
support a previous study in Peru that found an association between coastal seawater
and cholera (Gil et al. 2004); however, it also highlights a geographic difference (i.e., in
terms of time lag) between those results in Lima and Piura. Although the wavelet
analysis also showed a temporal association at the time period of 1993-94, the results
from the comparison could not support an association with ENSO because neutral sea
surface temperature conditions estimated in 1993 were followed by diverging sea
surface temperature conditions (i.e., El Niño in Niño 3.4 and La Niña in Niño 1+2) in each
region. Therefore, it could be argued that the association between cholera and ENSO
was stronger after 1992. Lastly, it was discerned that the temporal associations
between cholera and sea surface temperatures in the Niño regions were potentially
mediated through local sea surface temperatures at Paita and local rainfall in Piura.

6.3

ENSO and the Social Vulnerability of Cholera Incidence in Piura, Peru: 1997-98
In Objectives 3 and 4, I examined the spatial distribution of social vulnerability

and its association with cholera incidence by district (n = 33) in the subregion of Piura in
1997-98. In a subsequent analysis based on the findings of social vulnerability, I also
examined the association between global and local climate and cholera incidence by

211

district in 1998. In the social vulnerability analysis I constructed an overall social
vulnerability index and four sub-indices of vulnerability: (SVIF1) rural and river water;
(SVIF2) urban and water truck; (SVIF3) urban and public water well; and (SVIF4) ‘other’
water sources. All of these indices were based on potential risk factors associated with
cholera transmission, such as factors related to contaminated water exposure from poor
sanitation accessibility.
Mapping the overall social vulnerability and sub-indices revealed that the highest
overall social vulnerability, urban and water truck vulnerability, urban and water well
vulnerability, and ‘other water sources’ vulnerability was generally found on the west
coast of Piura. Furthermore, it suggested that those living on the coast were at highest
risk. This could be explained by populous urban areas found on the immediate coast of
Piura. River and rural water vulnerability (SVIF1) was found in the eastern part of Piura,
which reflects a population that may be living in rural areas where infrastructure and
public services are poor and people use river water. By comparing cholera incidence of
the most vulnerable districts, I demonstrated that the district with the highest
vulnerability also had the highest cholera incidence in Piura (i.e., Rinconada Llicuar
located on the west coast based on overall SVI and SVIF4). I also demonstrated that the
highest vulnerability by district did not always reflect the highest cholera incidence. For
example, several districts that were estimated as having the highest vulnerability based
on SVIF1 were places where cholera incidence was lowest (<2.0/1000 persons) in 199798. Another example is the district of Piura, which was found to be the least vulnerable
according to overall social vulnerability. These findings suggested that although some

212

places exhibited characteristics of cholera risk, there were other factors in these districts
(not captured by the SVI) that may have protected the population from cholera
transmission, such as population density, geographic isolation or perhaps the
environments in these districts were not suitable for bacteria proliferation. In the case
of the district of Piura, it may have been the availability of public resources and services,
which enhanced the resilience of the population since it is where the capital of the
Department is located.

6.3.1 Cholera and Social Vulnerability by District: 1997-98
Using global and local regression, I demonstrated that rural and river water
vulnerability and ‘other water sources’ vulnerability were significant factors that
explained the spatial distribution of cholera incidence in Piura. However, I also found
that each vulnerability index had a different impact. The effect of rural and river water
vulnerability on cholera transmission was negative; while the effect of ‘other water
sources’ vulnerability on cholera transmission was positive. While the former finding
may appear counterintuitive at first, it may also suggest that other factors not captured
by the sub-index prevented cholera transmission, such as population size or geographic
isolation. For example, cholera incidence in areas with rural and river water
vulnerability may also have been places with seasonal migrants or persons traveling
between urban and rural areas. Cholera cases in rural areas may have been isolated and
did not pose a threat to cholera transmission in those areas. Another possible
explanation is that the optimal conditions for V. cholerae reproduction may not have
213

been present in these districts. Therefore, low population along with poor ecology may
have prevented cholera transmission among these communities.
As suggested previously (Chapter 5), ‘other water sources’ vulnerability was an
interesting finding because it may refer to street vendors or a source not accounted for
in the vulnerability analysis. According to Ries et al (1992), ice from street vendors was
one of the primary vehicles of cholera transmission in the city of Piura in the first few
months of the cholera epidemic in 1991. Although I was unable to obtain
documentation for 1997-98 to support this exposure pathway, it is well known that
street vendors are a popular source of cheap food and drink throughout many Latin
American countries particularly during the summer months when temperatures
increase. Therefore, it may have been an important risk factor for cholera during the
1997-98 El Niño.
Furthermore, my results from the social vulnerability analysis revealed that there
was a pattern of vulnerability and unexplained risk of high to low values from west to
east associated with rural and river water vulnerability and ‘other water sources’
vulnerability. Of notable interest were those districts that exhibited high to moderate
vulnerability and high to moderate unexplained risk. I proposed that climate in part
might account for the unexplained risk and therefore, I explored these districts further
to understand the spatial variation of cholera vulnerability in Piura.

214

6.3.2 Cholera and Global and Local Climate by District: 1998
Following the spatial analysis of cholera and social vulnerability, I investigated
the associations between cholera incidence and global and local sea surface
temperatures and local temperatures and rainfall by district (n = 13) and month (n = 12)
in Piura in 1998. Using global regression, I demonstrated that climate had a positive
effect on cholera incidence but that the strength of these associations (i.e., determined
2

by the r ) varied by parameter, temporal lag and place (district and location). The
strongest associations were found in districts located on the west coast of Piura. One
possible reason is that the climate data employed in this study were measurements
from the district of Piura on the west coast. Therefore, it is likely the data is
representative of conditions in this part of Piura. Alternatively, the stronger climatecholera connection on the coast could be related to the influence of El Niño’s
teleconnections, which are stronger in Peru’s north coast. It supports Glantz’s
‘geographic’ notion that teleconnections are strongest the closer a place is to the
central equatorial Pacific Ocean; equally, the farther a teleconnection is, the weaker
that association will be (2001).
The strongest associations with cholera incidence were: Niño 3.4 SSTA (1 month
lag), Paita SSTA (1 month lag), rainfall (1 month lag), and rainfall (2 month lag). One
interesting observation is that the time lag for Niño 3.4 was notably different from the
wavelet results (e.g., 6 months); the time lag for Paita SSTA appears to agree with the
wavelet which found a shorter time delay in 1997-98 (e.g., 2-3 months). The sea surface
temperature associations support previous studies in Peru (Gil et al. 2004) and South
215

Africa (Mendelsohn and Dawson 2008), which documented short time delays between
SST and cholera. Niño 1+2 SSTA was weakly associated with cholera.
In contrast to the wavelet results, local temperature maximum (TMAXA) and
temperature mean (TMEANA) were strongly associated with cholera incidence. The
time lags associated with TMAXA and TMEANA were 6 months, similar to results in
Bangladesh (Pascual et al. 2000) and Ghana (de Magny et al. 2008). These findings also
support evidence in Peru that temperature increases may have impacted the
reproduction of V. cholerae in local water sources inland (Franco et al. 1994; Speelmon
et al. 2000). In addition, these findings may suggest that exposure through a
temperature pathway may have been important for cholera transmission in central and
eastern Piura (e.g., districts of Huarmaca and Chulucanas).
One of the most important findings was the association between cholera and
2

rainfall (1 month lag). It was the strongest climate association (r = 0.85) in this
segment of the analysis. It reinforced empirical evidence found in the wavelet analysis
that rainfall was an important transmission factor for cholera transmission during the
1997-98. It was particularly stronger in the districts of the west coast; however, as the
time lag increased to 2 months, the strongest associations moved eastward, which
indicates that during the 1997-98, rainfall impacts may have shifted to the highlands. It
may also reflect a time lag between El Niño and climate conditions in the highlands since
they are further away from the ocean.

216

6.3.3 How did social vulnerability influence the climate-cholera relationships in
Piura? Does the spatial distribution of social vulnerability within districts in Piura
explain the spatial variability of the ENSO-cholera associations in Piura in 1997-98?

In this dissertation project, I also demonstrated that rural and river water
vulnerability and ‘other water sources’ vulnerability were significant factors that
explained the spatial distribution of cholera incidence across districts in Piura during the
1997-98 El Niño. Moreover, I showed that global and local climate parameters were
significantly associated with cholera incidence in 1998. In order to assess whether social
vulnerability could explain the climate-cholera associations in Piura, I selected the 10
districts with the strongest associations in the climate-cholera analysis for a
discussion.

23

Figure 6.1 is a map showing these districts. They mainly represent districts

on the west coast since the strongest associations were located there; however there
are a few in the central and eastern part of Piura too. I also focused on the climate
variables, which had the strongest associations with cholera incidence.

23

Chalaco, Rinconada Llicuar, and San Juan de Bigote were not included in this
assessment because teleconnections in these districts were weakly associated with
2

cholera (r = < 0.60) or not statistically significant.
217

0

5°0'0"S

80°0'0"W

5°0'0"S

81°0'0"W

17

1

9
6

5
22

2

6°0'0"S

6°0'0"S

12

27

0

25

50 Kilometers

81°0'0"W

District
PIURA
CASTILLA
CATACAOS
LA ARENA
LA UNION

80°0'0"W

ID
0
1
2
5
6

District
HUANCABAMBA
HUARMACA
CHULUCANAS
SALITRAL1
SECHURA

ID
9
12
17
22
27

Figure 6.1 Map of Selected Districts for Comparative Analysis.

218

2

Table 6.1 compares social vulnerability and climate associations (based on r

values) with cholera by district. The distribution of values by district suggests that ‘other
water sources’ vulnerability partly explained the distribution of TMEANA (6 months lag)
which was concentrated on the west coast. It may also explain how cholera
transmission was enhanced in most districts where rainfall was strongly associated with
incidence. Rainfall’s impact was found in 8 out of 10 districts. At a 1 month lag,
rainfall’s impacts are distinctly associated with coastal districts (e.g., Piura, Castilla,
Catacaos, La Arena, La Union and Sechura). The combination of flooding and
overwhelmed water and sanitation infrastructure (due to flooding sewers) may have
contaminated the munipal water supply; subsequently, exposure to V. cholerae may
have occurred through drinks from street vendors or other water sources (not identified
in the census data).

219

2

Table 6.1 Comparison of Social Vulnerability and Climate Associations (r ) with Cholera
ID
District
SVIF1 SVIF4 NIÑO 3.4 SSTA1 PAITA SSTA1 Rainfall1 Rainfall2 TMAXA6 TMEANA6
0 PIURA
0.05 0.26
0.68
0.61
0.85
0.40
0.57
0.57
1 CASTILLA
0.08 0.24
0.60
0.55
0.85
0.45
0.56
0.56
2 CATACAOS
0.10 0.27
0.46
0.48
0.84
0.64
0.51
0.50
5 LA ARENA
0.03 0.30
0.74
0.71
0.81
0.55
0.76
0.63
6 LA UNION
0.02 0.32
0.69
0.73
0.89
0.82
0.79
0.75
9 HUANCABAMBA 0.26 0.13
0.18
0.26
0.37
0.67
0.41
0.38
12 HUARMACA
0.25 0.17
0.36
0.42
0.45
0.64
0.46
0.63
17 CHULUCANAS
0.16 0.20
0.41
0.49
0.59
0.86
0.42
0.68
22 SALITRAL1
0.22 0.18
0.44
0.50
0.36
0.75
0.43
0.70
27 SECHURA
0.04 0.34
0.67
0.63
0.84
0.54
0.63
0.63
The colors indicate in which districts the SVI indices may influence climate associations

220

At a 2 month lag, rainfall impacts 1 coastal district (La Union), 1 central district
(Chulucanas) and 1 eastern district (Salitral1). The latter two places are interesting. For
example, both districts share similar time lag associations with rainfall (possibly
influenced by the distance from the equatorial Pacific Ocean); however, they are
influenced by vulnerability in different ways. For example, while cholera risk was
enhanced by ‘other water sources’ vulnerability in Chulucanas; rural and river water
vulnerability reduced risk for cholera transmission in Salitral1, where it appears that
mean temperature also influenced incidence. This suggests that while climate processes
have the potential to impact cholera, there are other factors that prevent human
transmission. As discussed previously (in section 6.3.1), factors such as the population
size or geographic isolation (from infected migrants from the coast) may have prevented
cholera transmission. Furthermore, although river water was the main source of water,
it may not have been contaminated. These factors should be explored further in future
studies.
The comparative analysis also suggests that in the coastal districts of La Arena
and La Union, the impact of several climate variables on cholera incidence was
enhanced by ‘other water sources’ vulnerability. Of import here is that several
transmission pathways may have been present independently or interactively in these
districts. In contrast, in Huancabamba and Huarmaca located in the east, like the district
of Salitral1, rural and river water vulnerability may have lessened or prevented the
impacts of climate on cholera transmission; climate impacts in these districts were less
influential (associations were weak).
221

6.4

Study Limitations
This study has several caveats and provides lessons for future cholera research.

One limitation is the possibility that cholera cases were under-reported and that cases
were clinically confirmed and not laboratory confirmed. Relatedly, cholera case data
were unclear after 1998; several datasets I obtained had varying counts from 1999 to
2001. I also observed this discrepancy in the department and national datasets; the
number of cholera cases reported by MINSA conflicted with PAHO. There was also a
lack of population data over time; therefore I was unable to calculate incidence rates in
the temporal analysis. Another limitation is that I was also unable to fully examine the
initial cholera outbreak in 1991-92 during the wavelet analysis in Objective 2. Much of
my data fell within the cone of influence. I suspect that the short length of the cholera
time series (11 yrs) limited the number of scales that I could analyze (Compo, personal
communication January 2011). There was also the lack of spatial cholera data before
1997 and after 1998 in the subregion of Piura. Therefore, my results from the temporalspatial analysis in Objective 4 may not be applied to the entire decade. Yet, another
limitation is related to the vulnerability index, which was static rather than dynamic. I
utilized census data in 1993 to represent social conditions in 1998. Furthermore, the SVI
indices cannot capture the full complexity of human vulnerability to cholera; this is a
typical critique of vulnerability indicators and indices (Hahn et al. 2009). One way I
attempted to circumvent these limitations was to utilize literature and anecdotal
evidence that I collected in Piura and Lima during my fieldwork. I also consulted with my

222

collaborators in Piura and Lima. In the future, I will also consider using interviews
(public health authorities) to fill in gaps of knowledge.
In addition, my climate data was not representative of the entire subregion. I
used one station which was biased towards the west coast of Piura. However, it could
be argued that I measured cholera associations with this station using a ‘teleconnected
approach’ to impacts; to some degree neighboring stations in the subregion of Piura are
likely correlated with one another, and therefore, the information obtained in this
project was still useful to gain some understanding of the association between climate
and cholera in Piura. Furthermore, my assessment of the influence of social
vulnerability on the ENSO-cholera association during the 1997-98 El Niño was limited by
scalar differences. It highlights the challenges that face researchers in the quest to
understand climate and social impacts on disease and health, particularly in developing
countries. Lastly, from this research I did not learn about individual exposures to
cholera or how cholera was transmitted among individuals in a population. I also
recognize that there may be gender and age differences in how people are exposed to
cholera and these differences may vary by socioeconomic status, occupation and
migration.

6.5

Ethical Geographies of the Cholera Epidemic in Piura
In conclusion of this chapter, I present a coda that reflects my ethical perspective

on the findings in this dissertation as well as the research process as it relates to my
fieldwork in Peru. My aim is to highlight several ethical challenges/issues raised by the
223

study of the cholera epidemic in Piura. I will focus on two ethical lines of inquiry. The
first ethical consideration addresses the empirical findings in this dissertation. The
second ethical consideration addresses issues that arose while conducting this research,
e.g., context of the fieldwork and the research setting.
I begin this statement by defining my ethical lens which considers both an ethics
of climate and an ethics of development. These perspectives are needed to evaluate
the ethical considerations described above because cholera transmission is indirectly
influenced by climate variability and by the human development conditions in which
people live. Furthermore, during my fieldwork I encountered several issues associated
with inequity in scholarship and epistemological concerns in regards to local
understandings of the cholera epidemic and El Niño. I go on to present several ethical
challenges/issues which I herein refer to as ‘ethical geographies’ that emerged from this
dissertation.

6.5.1 Ethics of Climate and Development
Climate ethics is a growing field that has emerged in philosophy to address the
moral implications of anthropogenic climate change for society (Gardiner 2010: 3). In
this reflective statement, I employ the lens of an ethics of climate based on the climate
affairs approach used in this study. Climate ethics from this perspective is defined as
moral issues or questions associated with harms or challenges to society that arise from
climate-society interactions; it also considers the principles that ought to guide a society
in its decisions about climate and climate-related impacts (Glantz 2003: 165). According
224

to its mantra, a climate affairs ethics views climate phenomena and climate-related
issues more broadly then the current climate ethics literature.

24

Therefore, I suggest

that a climate affairs-based ethics is more suitable to address moral questions about the
cholera epidemic in Piura and its relationship with ENSO, which is natural climate
variability.

25

This approach to climate ethics considers topics, such as inter- vs.

intragenerational equity conflict, precautionary principle, North-South divide (e.g., the
socioeconomic development gap between developed and developing countries), and
environmental justice (Glantz 2003: 65-74). It is also concerned with questions about
the ethics of impacts and vulnerability, a topic which has received less attention. From
this conception, a climate affairs-based ethics shares concerns with an ethics of global
development (Gasper 2005; Crocker 2008).
Development ethics is a subfield in philosophy which is concerned with ethical
issues and questions that arise from global development policies and practice (Crocker
2008: 1), as well as the interpretation of the concept (Gasper 2005: 1). Namely,
development ethics addresses the social and economic inequities that are experienced
by societies by uneven global development (de Blij 2009); furthermore, it emphasizes
the greatest concern for the poorest and socially disadvantaged subpopulations. One
particular conception of development ethics that is useful to examine ENSO and the
24

Generally, the climate ethics literature has been concerned with arguments that
support the reduction of greenhouse gas emissions to address human-induced climate
change. See Gardiner et al. (2010) for a collection of essays on the subject.
25

I make this distinction here because studies suggest that ENSO’s behavior has not
been impacted thus far by anthropogenic climate change.
225

cholera epidemic in Piura is the capability approach to human development (Sen 1999;
2010; Nussbaum 2003; 2011). The notion of human capability suggests that human
development be viewed through a broader context than just incomes, resources, or
goods. It provides the conceptual space to evaluate human well-being through
‘capabilities’, which are elements that enable humans to function and flourish in society
(Sen 1999). Important to this concept is ‘freedoms’ which represent people’s ‘real’
opportunities that are instrumental and intrinsically valuable to their flourishing
(Aristotle’s notion as cited in Nussbaum 2000). Well-being is judged by the extent to
which freedoms can be achieved and the actual achievements, which freedoms and
capabilities facilitate for people. The Human Development Index (HDI) by the United
Nations is based on human capabilities (UNDP 2009). Combining climate ethics with a
capability approach provides this reflective statement a great insight to evaluate ethical
issues that arise in health, medical and human-environment geographic research, and in
this case, ENSO and the cholera epidemic in Piura.
6.5.2 Ethical Geographies
In this section I discuss several ethical geographies that emerged from this
research including my experience during fieldwork in Peru in 2008 and 2009 and
subsequent collaborations with my institutional partners in Piura and Lima, Peru.
6.5.2.1 Ethical Geography 1
ENSO impacts may have exacerbated cholera transmission in vulnerable places
within the subregion of Piura, Peru. Furthermore ENSO impacts on cholera may
have affected some places more than others depending on the characteristics of
social vulnerability in those places.
226

According to the Rocky Ethics Institute (http://rockblogs.psu.edu/climate/), the
magnitude of impact from a changing climate will be catastrophic for the most
vulnerable populations, and thus it is an ethical reason for society to respond. Although
the context of this statement is a changing climate and what we should do about
greenhouse gas emissions, it can also be applied to ENSO and its impacts on the cholera
epidemic in Piura. ENSO alters climate patterns around the world and those changes in
climate have been associated with hazards that affect the health of vulnerable
populations (Kovats et al. 2003; Glantz 2001; 2003).
Evidence in this dissertation suggested that indeed ENSO may have increased the
burden of cholera transmission on the west coast of the subregion of Piura, Peru,
namely through flooding, for example, because of heavy rains and the subsequent
collapse of sanitation infrastructure and/or contamination of the public water supply.
Moreover, cholera impacts were experienced differentially based on the social
vulnerability of water and sanitation infrastructure within that place. Many people may
have been exposed because they purchased water or food from street vendors
influenced by socioeconomic constraints or unavailability of potable water in their
homes. From a climate ethics and capability approach, it suggests that cholera
transmission increased in Piura in places characterized by deprivation of capabilities
associated with limited access to food and water and public services, which protect
people from waterborne diseases.

227

This underlying context of the cholera epidemic in Piura, specifically during the
1997-98 El Niño, brings to light the human development and the social environment in
Piura. The setting suggests that conditions in Piura were not so good to begin with and
that some people in Piura were already living in socially marginal conditions that created
the setting for potential infectious disease transmission. When El Niño conditions
developed in April-May of 1997 followed by El Niño-related teleconnections in the
December of 1997, extreme weather and climate contributed to ecologically favorable
conditions for bacteria and pathogens that intersected with the development context in
Piura which led to subsequent exposure to cholera and social harm during that austral
summer.
During the 1997-98 El Niño, there were also reports of climate-related impacts
on other infectious diseases, which are typically reported during extreme climate events
in Piura, and coincidently are associated with poverty in the region. For example, from
January to March of 1998, when El Niño conditions were peaking, there were 42,000
cases of malaria (i.e., based on two types: P. falciparum and P. vivax) and 3,200 cases of
conjunctivitis because of ecological changes in local aquatic environments and impacts
26

on disease vectors.

In addition there were economic impacts on infrastructure in

multiple sectors. Estimated losses ($ U.S. in millions) were reported in agriculture
(41.5), sanitation (13.0), and health (0.5) (Sandoval 1999). Although the health sector

26

According to Norma Ordinola at UDEP, during the summers and particularly during El
Niños, there are plagues of insects. Apparently, there are so many that people have the
tendency to scratch their more often which can lead to infection if there hands are dirty.
228

appeared to be the least affected, the damages incurred by extreme weather on health
infrastructure may have impeded society’s capacity to respond in Piura during that time.
The broader implication of this story is that populations within the subregion of
Piura may have experienced a confluence of climate extremes, infrastructure
deprivation, and an environment conducive to infectious disease transmission. Within
such a context, the study of the cholera epidemic in Piura does not only raise questions
about distributive justice (e.g., how are impacts distributed); it also raises questions
about the type of hazards and impacts that populations experienced (e.g., co-exposures
to climate and social hazards and co-infectious disease morbidities) and the relationship
of these experiences to human capabilities and development. One important question
to consider is to understand how human capabilities change as the hazards and impacts
change across time and space.
6.5.2.2 Ethical Geography 2
Making the research process and collaborations with partners in Piura equitable
and participatory.
My dissertation fieldwork in Piura and Lima exposed me to several interrelated
ethical challenges associated with what I term ‘inequity in scholarship’ which could also
be linked to the North-South divide. Before I begin describing these challenges, I would
like to first reiterate the purpose of my field work in Peru to set the context for these
challenges.
In the summers of 2008 and 2009 I visited Piura and Lima to collect data
including documents in order to develop my conceptual and empirical models in my
229

dissertation. I also collected anecdotal information from many officials at institutions in
Lima and Piura. This was important because I needed to understand how the data were
collected and interpreted. I was also looking for social and political dimensions of the
cholera epidemic to inform my analysis. Many of the documents I obtained were in
Spanish so I needed to communicate frequently with my collaborators in Piura during
and after my visits. I also needed a guide to help me navigate through the institutions
as well as keep an eye on me since I was a foreigner (i.e., for safety reasons). My main
contact in Piura was Prof. Norma Ordinola of the University of Piura (UDEP) via a referral
by Dr. Michael H. Glantz at CCB. Therefore, my collaborators were essential to this
dissertation.
As a researcher and graduate student of an American University, I entered the
collaborative process under the naïve guise that this research experience would be
reciprocal and mutually beneficial. By this I am referring to one of the principles of
community engagement as defined by Fitzgerald et al. (2005). Reciprocal and mutually
beneficial refers to a bi-directional relationship of information and benefits derived from
the scholarship. It suggests that knowledge is generated with partners and that this
process is transformational for society. However, as I would learn thereafter, achieving
equity in scholarship as an American conducting research in a developed country setting
was not so easy.
To begin, I entered the research process with a certain set of goals that would
hopefully advance my career and capabilities. It was apparent to me what I could gain

230

from the experience. For my partners in Piura, however, it may not have been so clear.
I am not sure that they were even concerned about gaining. In my project, my partners
were clearly interested in helping me the graduate student and fellow South American.
I am an American but also one with a background and ethnicity that may have given me
some advantages for collaboration. I also came into the process with advantages in
terms of socioeconomic status. As I recall, Prof. Norma Ordinola pointed this fact out,
and stated that even as a graduate student my social conditions were better than a
Professor’s in Piura. Therefore, I should be careful when I communicate with officials
and other academics in Piura. Overall my experience was positive when I worked with
institutions in Piura.
My collaborators were invaluable to my dissertation research. They trusted me
and invested a great deal of time and effort to assist me. Furthermore, they collected
and handed me what appeared to be their entire dataset. One ethical geography that
arised from this experience was whether my partners were trading off important time
and resources to communicate and work with me. Dr. Stephen Esquith of the
Philosophy Department at MSU alerted me to this type of opportunity cost which
collaborators/ partners in developing countries may face. This question leads to a
subsequent consideration of the usability of my research and my relationship to Piura.
The former is more difficult to answer and likely requires a forum for participation and
feedback. It could be true that my approach to understanding the cholera epidemic

231

may only be useful to academic circles. That is, since cholera is no longer present and
current health challenges may seem of greater import (e.g., dengue).

27

Another ethical geography associated with fieldwork was communicating with
my partners in Piura. I found that infrastructure and language were barriers. I realize
that my partners in Piura do not always have access to good internet service. They are
also likely to be busy with their own projects. I also find that since I am no longer a
native speaker of Spanish, I am less engaged because it takes much longer time to
develop a message to my partners. It is much easier to communicate in person in
Spanish. Although I was able to obtain enough information about my project, I felt that
it was not the most effective or equitable way to work on a research project. My work
could have been better informed with more communication and more direct
communication (e.g., phone instead of internet). Moreover, it was difficult to sustain (in
my opinion) a working relationship this way.

6.5.2.3 Ethical Geography 3
Local understandings of El Niño in Peru are marginalized.
The last ethical geography refers to the marginalization of Peru’s scientific
understandings of El Niño and impacts. Currently, the definition of El Niño-Southern
Oscillation is based upon NOAA’s interpretation which relies on the Niño 3.4 region in

27

The number of classic dengue fever cases rose from 495 in 1999 to 9,739 in 2004
(MINSA 2005a).
232

the equatorial Pacific Ocean. The importance of sea surface temperatures in this region
which straddles the Niño 4 and Niño 3 regions is it has been shown to allow for a better
predictability of ENSO’s global teleconnections. While I acknowledge that the NOAA
definition is important, I also recognize that Peruvian scientists, as well as the South
American scientific community, have their own interpretation of ENSO. In fact, for
Peruvians, it could be argued that the El Niño phenomenon, as they refer to it, is
important to their society because they are sensitive to its associated environmental
changes since generally conditions off the coast of Peru are typically in a neutral state of
ENSO. Therefore, while Peruvians consider the NOAA definition, they also rely upon
their own criteria using monitoring stations spanning the coast of Peru (see Chapter 2,
footnote 4). This definition known as the SCOR definition was proposed by South
American scientists in 1983 and was subsequently rejected by the international scientific
community (Trenberth 1997; Lagos et al. 2008). It is my assumption that this
community was largely made up of Americans and Europeans who influenced this
decision.
I became aware of the importance of this issue while reviewing the literature
and conducting fieldwork in Piura. As I began to have conversations with Prof. Norma
Ordinola, my contacts at various institutions, and even local people (e.g., cab drivers)
about El Niño in the early 1990s, I realized that an El Niño in 1991 was not remembered
by anyone that I spoke with. It suggested that Piurans did not experience an El Niño
during that time. What many remembered, however, was the experience of heavy rains
in Piura in the austral summer of 1992, which local data confirms. It is this piece of
233

information that helped inform my hypotheses. It also supported my doubts that an El
Niño had occurred in the initial stages of the epidemic in 1991. Therefore, Colwell’s
hypothesis (1996) may have been based on the global understanding of ENSO. Well,
why do I consider this an ethical issue? These insights from my fieldwork raise
questions about knowledge inequalities between the North and South scientific
communities. It could be argued that knowledge generated about El Niño in Peru is not
widely known, and therefore, not as widely accepted (e.g., as NOAA), because of
resources disparities and perhaps even language barriers. This speaks to the fact that in
order for scientific knowledge to become widely accepted, one must publish in peerreviewed journals, most likely, in English. In order to do so, funding and fluency in
English is necessary.
6.5.3 Conclusions
Using the combined perspectives of climate and development ethics, this coda
highlighted several ethical geographies that arose from the findings in this dissertation
and the research process itself. The first ethical geography associated the findings with
issues of distributive justice and suggested that the cholera outbreak in Piura was
actually part of a broader social and environmental crisis with implications for the
geography of human capabilities and development. The second and third ethical
geographies raised questions about inequity in climate impacts scholarship. These
issues illuminated disparities that existed in the study area, and suggested that future
scholarship be mindful of this context and work towards minimizing this type of
injustice.
234

7. CHAPTER 7: CONCLUSIONS
Using a climate affairs approach, I reconstructed the temporal and spatial
associations among ENSO, social vulnerability and cholera incidence in Piura, Peru from
1991 to 2001 in order to better understand El Niño’s impact on the cholera epidemic in
Peru. Moreover, I explored what more can be learned about cholera emergence and
transmission if we considered the broader aspects of the ENSO cycle including social
dimensions. My findings suggested that cholera’s temporal association with ENSO was
transient throughout the 1990s; the strongest associations were found during the 199798 El Niño, supporting previous studies in Peru. I also found that an important temporal
pathway in cholera transmission occurred through the interactions of global and local
sea surface temperatures with rainfall. Furthermore, I showed that the spatial
distribution of social vulnerability by district can in part explain the associations
between global and local climate and cholera during the 1997-98 El Niño; the findings
also suggested that social vulnerability modifies the climate-cholera relationship based
on the type of water-infrastructure vulnerability. Moreover, at a district-level analysis
(i.e., distinguishing the wavelet analysis from the regression analysis), it was revealed
that the associations between climate and cholera varied by time lag, district and
climate variable. In terms of geographic patterns, it appears that districts on the west
coast of the subregion of Piura were the most vulnerable to climate-cholera impacts. In
addition, the spatial analysis provided further support for the role of rainfall on cholera
transmission within Piura. Lastly, important to the understanding of these findings is

235

that interpretation of ENSO and its association with cholera will depend highly on the
Niño region chosen for analysis, as Trenberth (1997) had demonstrated previously.
Dissemination of findings
I plan to disseminate and share my findings with scholars, public health
practicioners, policymakers, and civil society by presenting at conferences, such as the
Association of American Geographers (AAG) Annual Meeting, the International Medical
Geography Symposium (IMGS) (at MSU in 2013), and the American Meteorological
Society (AMS), and at workshops, such as NOAA’s Annual Climate Prediction Applications
Science Workshop (CPASW). I will also prepare several manuscripts for publication

including: a wavelet anaylsis of climate and cholera in Piura; an assessment of the
impacts of global and local climate anomalies on cholera incidence in Piura, Peru during
the 1997-98 El Niño; a geographically weighted regression analysis of social vulnerability
and cholera incidence in Piura; and the importance of definition and region in
understanding the relationship between sea surface temperature and cholera in Peru. I
will also pursue a research grant so that I may return to Peru to share my findings with
MINSA and my institutional collaborators in Piura and Lima, Peru. This activity will also
contribute to future collaborations and capacity in Piura. In addition, I will integrate my
study knowledge including my fieldwork experiences into undergraduate and graduate
coursework materials and lectures, as well as outreach and engagement activities that
foster capacity building.

236

Future areas of scholarship: implications for research, policy and ethics
Understanding the climate dimension of cholera transmission in Peru remains an
important topic today because, although, the last confirmed cases of cholera (related to
the El Tor strain) were reported in 2002, the first case of the epidemic strain V. cholerae
0139, previously reported in Bangladesh (WHO 2009), was documented in South Lima in
2004 (Instituto Nacional de Salud 2006). Furthermore, a cholera epidemic in Haiti that
erupted in October 2010 revealed that cholera remains a potential threat not only to
Peru but to other parts of the region; and maybe more so in the face of a changing
warming climate (Shah 2011).
Below are several implications that can be drawn from this study to build
capacity and contribute to the future prevention of cholera and other climate-sensitive
diseases in Piura and the Latin American region.
a) Building capacity through impacts and time-lag information
One important finding which warrants further investigation is the associations
among Niño 3.4 sea surface temperatures, local sea surface temperatures in Paita, and
rainfall in Piura. We should learn more about the pathways in which each variable was
associated with cholera incidence and how each may have interacted with one another
to impact disease transmission. For example, how are the time-lag associations among
the climate variables related? Furthermore, what are the characteristics of rainfall
impacts (e.g., abundant versus deficient) by place (e.g., coast versus mountain or areas
away from coast)? Understanding these important pieces of information could inform

237

future public health responses in Piura to ENSO and other climate-related extremes.
Moreover, such information could improve current early warning systems in Piura for
climate-sensitive diseases, such as diarrheal (non-cholera), mosquito-borne and
respiratory diseases. This endeavor would enhance Piura’s El Niño contingency plan
which is organized through several institutions including the Ministry of Health and
UDEP (Sandoval 1999; MINSA 2005a; Rodriguez, personal communication 2009).
b) Understanding the first few years of the outbreak
My study was unable to fully examine the initial cholera outbreak in Piura from
1991 to 1992, and therefore it is crucial to gain a better understanding of this time
period because it may help to understand the origin of the epidemic. I highlight two
points of investigation. One is to understand how cholera incidence diffused within and
across Piura. Another is to understand how climate impacted the emergence of cholera.
One research suggestion is to conduct a qualitative analysis that retraces the first
suspected cases of cholera by place through newspapers (e.g., the local newspaper El
Tiempo) and other documents that may be available at local health outposts and census
offices in Piura. Using these sources, it may also be possible to identify if there were any
extreme climate-related events that coincide with incidence. Important to this
endeavor will be to sustain collaborations with the University of Piura (UDEP), the
Ministries of Health in Piura and Lima, and the local census offices.

238

c) ENSO and attribution
The study of the cholera epidemic in Piura offers lessons about ENSO and disease
attribution. In order to fully understand an association between ENSO and infectious
diseases, researchers and public health practitioners should take into account the
importance of understanding the ENSO definition and Niño region in climate-health
impact studies. Identifying El Niño periods is not a simple task given that there are
many definitions that exist and can be applied; the association with infectious disease
can vary by the criteria of the definition (e.g., anomaly thresholds) and the region
chosen to represent climate anomalies (e.g., Niño 3.4 or or Niño 1+2). As such, the
researcher should carefully interpret and assess which region data is appropriate for the
area of study. Furthermore, ENSO has many characteristics, which should be fully
explored if possible, to assess an association. In my study, two important characteristics
were rainfall teleconnections and social dimensions.
d) Unexplained social vulnerability
As this dissertation has demonstrated, there are geographic differences of social
vulnerability to cholera that require better understanding. For example, what were the
contributing factors in districts where river and rural water vulnerability lessened the
likelihood of transmission? In addition, can we identify the transmission pathways in
districts where ‘other water’ sources vulnerability increased the risk for transmission?
In order to further understand these findings and their application in future prevention
of cholera, it may prove useful to examine the contribution of population immunity on

239

cholera incidence. This would require obtaining birth rates and vaccination coverage
information, two variables used by Koelle and Pascual (2004) to assess herd immunity in
Bangladesh. Furthermore, this new information on immunity should take into account
that 75.0% of infected persons are asymptomatic (WHO 2009). Another factor which
may be important is migration (urban to rural or mountain to coast). In my analysis of
MINSA epidemiology reports, I noticed that alerts were posted about the potential risk
of cholera transmission during the summer season (e.g., agricultural migrants),
coincidently when cholera incidence increased at a national level.
Yet another factor to understand is water infrastructure and humanenvironment interactions which led to transmission in Piura. Specifially, in regards to
‘other water’ sources vulnerability, I suspected that water from street vendors was the
source of transmission based on previous evidence in Peru (Tauxe et al. 1995) and Piura
(Ries et al. 1992). The latter study was based on the city of Piura and, therefore, it may
not have represented the source of transmission in other areas within the subregion of
Piura. Thus, additional knowledge about water sources and transmission in Piura should
be pursued. These data could be obtained by interviewing officials at MINSA. As I
discovered during fieldwork, there are still many public health staff at MINSA in Piura,
which were present during the years of cholera epidemics in Piura. Learning about
these factors would enable us to better estimate incidence, understand societal factors,
and enhance prevention efforts.

240

e) Resilience actions
One important aspect of vulnerability which was not addressed in this study is
resilience. During the cholera epidemics in 1991 and 1998, there were also reports of
societal responses to mitigate the impacts of cholera. In addition to education
programs to prevent diarrheal disease and cholera, MINSA set up sanitary and oral
rehydration posts (MINSA 1994b). Furthermore, given Peru’s past history with El Niño,
several actions from different public sectors were taken to respond to emerging El Niño
threats in 1997 (Sandoval 1999; MINSA Piura 2005a), which may have contributed to
cholera prevention too. Therefore, I recommend that practitioners and policymakers in
Piura that are responsible for public health and development initiatives consider
understanding actions that enhance both health and development by not only
considering the multiple ways that societies can become exposed to infectious diseases,
but also actions that promote capacities to cope.
f) Cholera as part of broader health crisis
Future health and development policies should also consider that the Ministry of
Health in Piura may be facing several concurrent health and social crises at multiple
scales. From this perspective, policy, practice, and research should consider a more
holistic approach to society, health and disease, such as the ‘syndemic’ perspective
which is currently being promoted at the CDC (http://www.cdc.gov/syndemics/).
Syndemic environments are places where social, economic, political, and environmental
factors interact and contribute to an excess burden of disease in vulnerable populations.
The concept of “syndemics” was first introduced by Singer (1994) to capture and
241

understand how poverty, poor nutrition, and socioeconomic stressors increased the risk
for exposure to HIV transmission and related opportunistic infections in Harlem, NYC.
Utilizing this orientation of analysis to understand cholera within a medical geographic
tradition would greatly compliment my findings on social vulnerability. Moreover, if a
political ecology of health approach were integrated too, this would greatly enhance our
understanding of scalar interactions including those related to global development
(Mayer 2000).
g) Addressing inequity in scholarship
One ethical issue that arose during this dissertation research was inequity in
scholarship. In order to address this problem, I acknowledged and credited my partners
in Piura in presentations and papers. I also intend to publish articles with them, as well
as work with these partners more equitably (e.g., from the beginning of the research
process) in the future. I have already published one editorial with Prof. Norma Ordinola,
my collaborator in Piura (http://ccb.colorado.edu/enos/editorials/sept09.html). Also,
through my work at CCB with Dr. Michael H. Glantz, I developed a bibliography of El
Niño-related works that include the materials I collected in Peru and links to webpages
from various Latin American institutions that address El Niño and climate impacts (This
information is available online at: http://ccb.colorado.edu/enos/comunidad.html). The
collection and sharing of El Niño-related materials from Latin America addresses the
problem of marginalized ENSO-society science in Latin America. This is an important
future area of research which I intend to pursue.

242

h) Climate, cholera and marine reservoirs
Lastly, in order to further link my findings to the broader knowledge of
climate/environmental factors of cholera transmission, it is necessary to examine the
association between potential marine reservoirs off the coast of Piura and cholera
incidence. Through UDEP partnerships, it may be possible to collect marine
environmental data as well as remote sensing data that measures biological productivity
(e.g., chlorophyll, which serves as a proxy for plankton) in the equatorial Pacific Ocean.
This would link my findings to previous work in Lima (Gil et al. 2004). Furthermore,
collaborations with researchers that examine climate and cholera incidence in
Bangladesh (e.g., Dr. Mercedes Pascual [University of Michigan] and Dr. Michael Emch
[University of North Carolina]) should be fostered in order to compare results in Peru.
In closing, I identified important areas of future scholarship and their relevance to
research, policy and ethics. They highlight the importance of retrospective analyses in
climate-health impacts research and the multidisciplinary nature of this work. It is
hoped that this research will contribute to future climate-informed initiatives that
enhance societal capacities while focusing on population health and the monitoring of
populations during future climate events in Piura, Peru and the Latin American region.

243

8. APPENDICES

244

APPENDIX 1
Table A-1 Time Lag Associations in Climate Cholera Studies

Cholera Variable
% cholera
Cholera rate
Cholera rate
% cholera
% cholera
% cholera
% cholera
% cholera death

Climate
Variables
SSH
High rainfall
Low rainfall
High rainfall
Low rainfall
Niño 3.4 SST
Temperature
Niño 3.4 SST

Lag
1 month
1-5 weeks
1-16 weeks
29-31 weeks
10 weeks
11 months
4-6 months
10 months

S
P
P
P
P
P
P
P
P

1997-2000

Cholera

Lake water cond

zero month

P

1997-2000

Cholera

Rainfall

8 week

N

1997-2000

Cholera

Water temp

4-8 weeks

P

1997-2000

Cholera

Water temp

6 weeks

P

2008 Bangladesh, Matlab

1997-2006

Cholera

Chlorophyll

1 month

P

2008 Bangladesh, Matlab
2008 Bangladesh, Matlab

1983-2003
1983-2003

Cholera outbreak Chlorophyll
Cholera outbreak Chlorophyll

2 months
zero month

P
P

Authors
Lobitz et al.
Hashizume et al.
Hashizume et al.
Hashizume et al.
Hashizume et al.
Pascual et al.
Pascual et al.
Rodo et al.

Year
2000
2008
2008
2010
2010
2000
2000
2002

Huq et al.

2005

Huq et al.

2005

Huq et al.

2005

Huq et al.

2005

Constantin de
Magny et al.
Emch et al.
Emch et al.

Study Area
Bangladesh
Bangladesh, Dhaka
Bangladesh, Dhaka
Bangladesh, Dhaka
Bangladesh, Dhaka
Bangladesh, Dhaka
Bangladesh, Dhaka
Bangladesh, Dhaka
Bangladesh, L.
Bakerhganj
Bangladesh, L.
Bakerhganj
Bangladesh, L.
Bakerhganj
Bangladesh, L.
Bakerhganj

Time Period
1992-1995
1996-2002
1996-2002
1983-2008
1983-2008
1980-1998
1980-1998
1893-1940

245

Table A-1 (cont’d)
Authors
Constantin de Magny et al.
Constantin de Magny et al.
Constantin de Magny et al.
Constantin de Magny et al.
Ruiz Moreno et al.
Ruiz Moreno et al.
Ruiz Moreno et al.
Emch et al.
Emch et al.
Emch et al.
Emch et al.
Emch et al.
Emch et al.
Emch et al.
Koelle et al.
Koelle et al.
Koelle et al.
Koelle et al.
Koelle et al.
Bouma & Pascual
Bouma & Pascual

Year
2007
2007
2007
2007
2007
2007
2007
2008
2008
2008
2008
2008
2008
2008
2005
2005
2005
2005
2005
2001
2001

Study Area
Time Period
Ghana
1975-1995
Ghana
1975-1995
Ghana
1975-1995
Ghana
1975-1995
India - NE, Madras
1901-1940
India - S, Madras
1901-1940
India - S, Madras
1901-1940
Vietnam, Hue
1985-2003
Vietnam, Hue
1985-2003
Vietnam, Nha Trang 1985-1995
Vietnam, Nha Trang 1985-1995
Vietnam, Nha Trang 1985-1995
Vietnam, Nha Trang 1985-1995
Vietnam, Nha Trang 1985-1995
Bangladesh, Matlab 1966-2002
Bangladesh, Matlab 1966-2002
Bangladesh, Matlab 1966-2002
Bangladesh, Matlab 1966-2002
Bangladesh, Matlab 1966-2002
Bengal
1891-1940
Bengal
1891-1940

246

Cholera Variable
Cholera
Cholera
Cholera
Cholera
Cholera deaths
Cholera deaths
Cholera deaths
Cholera outbreak
Cholera outbreak
Cholera outbreak
Cholera outbreak
Cholera outbreak
Cholera outbreak
Cholera outbreak
Cholera
Niño 3.4 SSTA
Cholera
Cholera
Cholera
Cholera deaths
Cholera deaths

Climate Variables
Indian Ocean I
Temperature
Rainfall
Rainfall
Rainfall
Rainfall
Rainfall
SSH
SSH
Cai River disc
Cai River height
Dinh River height
Dinh River height
Rainfall
BOB SST
BOB SST
NE India rainfall
Niño 3.4 SSTA
River disch. A
BOB SST
Niño Year

Lag
16 months
12 months
.75 month
16 months
zero month
3-7 months
zero month
2 months
zero month
zero month
zero month
2 months
zero month
zero month
0-9 months
2-3 months
14 months
8-10 months
7 months
zero month
12 months

S
P
N
N
N
P
P
N
N
N
P
P
P
P
P
P
P
N
P
N
P
P

APPENDIX 2

IRB#X08-724 – Questions and Consent Form
List of Questions that I will ask individuals at Peruvian institutions:
Please note that these questions will be adapted to each type of institution I will
approach for data.

What kinds of data are available at your institution?
If data are available, are they accessible?
If data are available, how are they organized?
If data are available, in what format do the data exist?
If data are available, at what scales and time periods are the data available for?
If data are available, may I review the data?
If data are available, how may I obtain the data?
If data are not available, can you refer me to an institution, which have these types of
data available?

247

Consent Form

I, _____________________, at the institution ____________________, on this date
______________________, agree to participate in a discussion about data sources
relevant to the pertinent research study that Ivan Ramirez, a PhD student in the Dept. of
Geography at Michigan State University, is conducting for his dissertation while in Peru.
Participation is voluntary, you may choose not to participate at all, or you may refuse to
answer certain questions or discontinue your participation at any time without
consequence. I acknowledge that the information generated from this discussion may
be used for research purposes and will be properly cited and acknowledged. If I wish to
remain as an anonymous source, then I will inform Mr. Ramirez.

If you have concerns or questions about this study, such as scientific issues, how to do
any part of it, or to report an injury, please contact the researchers:

Ivan J. Ramirez, M.A.
Department of Geography, Michigan State University
118 Geography Building
East Lansing, MI 48824
517.355.4649, ijr2105@msu.edu

Dr. Sue Grady
Department of Geography, Michigan State University
130 Geography Building
East Lansing, MI 48824
517.342.9998, gradys@msu.edu

248

APPENDIX 3

Wavelet Transform Script
%WAVETEST Example Matlab script for WAVELET, using Cholera (Square root
transformed) Anomaly dataset
cholera = cholera_sqrt_a;
Script developed by Torrence and Compo (1998) found at:
http://paos.colorado.edu/research/wavelets/
%------------------------------------------------------ Computation
% normalize by standard deviation (not necessary, but makes it easier% to compare
with plot on Interactive Wavelet page, at
% http://paos.colorado.edu/research/wavelets/plot/ (this was excluded from the script
because the climate datasets I used were already standardized anomalies.
variance = std(cholera)^2;
n = length(cholera); %length of dataseries
dt = 0.083; %amount of time between each Y value, i.e. the sampling time. this says 1
month
time = [0:length(cholera)-1]*dt + 1991.0 ; % construct time array
xlim = [1991,2001]; % plotting range
pad = 1; % pad the time series with zeroes (recommended)
dj = 0.25; % this will do 4 sub-octaves per octave
s0 = 2*dt; % the smallest scale of the wavelet. this says start at a scale of 2 months
j1 = 5/dj; % the # of scales minus one. this says do 5 powers-of-two with dj suboctaves each
lag1 = 0.72; % lag-1 autocorrelation for red noise background
mother = 'Morlet'; %the mother wavelet function.
% Wavelet transform:
[wave,period,scale,coi] = wavelet(cholera,dt,pad,dj,s0,j1,mother);
power = (abs(wave)).^2 ;
% compute wavelet power spectrum
% Significance levels: (variance=1 for the normalized cholera)
[signif,fft_theor] = wave_signif(1.0,dt,scale,0,lag1,-1,-1,mother);
sig95 = (signif')*(ones(1,n)); % expand signif --> (J+1)x(N) array
sig95 = power ./ sig95;
% where ratio > 1, power is significant
% Global wavelet spectrum & significance levels:
global_ws = variance*(sum(power')/n); % time-average over all times
dof = n - scale; % the -scale corrects for padding at edges
249

Wavelet Transform Script (cont’d)
global_signif = wave_signif(variance,dt,scale,1,lag1,-1,dof,mother);
whos
%------------------------------------------------------ Plotting
%--- Contour plot wavelet power spectrum
subplot('position',[0.1 0.37 0.65 0.28])
levels = [0.0625,0.125,0.25,0.5,1,2,4,8,16,32,64] ;
Yticks = 2.^(fix(log2(min(period))):fix(log2(max(period))));
contourf(time,log2(period),log2(power),log2(levels)); %*** or use 'contourfill'
%imagesc(time,log2(period),log2(power)); %*** uncomment for 'image' plot
ylabel('Period (Year)')
set(gca,'XLim',xlim(:))
set(gca,'YLim',log2([min(period),max(period)]), ...
'YDir','reverse', ...
'YTick',log2(Yticks(:)), ...
'YTickLabel',Yticks)
% 95% significance contour, levels at -99 (fake) and 1 (95% signif)
hold on
contour(time,log2(period),sig95,[-99,1],'k');
hold on
% cone-of-influence, anything "below" is dubious
plot(time,log2(coi),'k')
hold off
%--- Plot global wavelet spectrum
subplot('position',[0.77 0.37 0.2 0.28])
plot(global_ws,log2(period))
hold on
plot(global_signif,log2(period),'--')
hold off
set(gca,'YLim',log2([min(period),max(period)]), ...
'YDir','reverse', ...
'YTick',log2(Yticks(:)), ...
'YTickLabel','')
set(gca,'XLim',[0,1.25*max(global_ws)])
250

Wavelet Coherence Script
%% Wavelet coherence Example
%
% USAGE: [Rsq,period,scale,coi,sig95]=wtc(x,y,[,settings])
%
% Settings: Pad: pad the time series with zeros?
%.
Dj: Octaves per scale (default: '1/12')
%.
S0: Minimum scale
%.
J1: Total number of scales
%.
Mother: Mother wavelet (default 'morlet')
%.
MaxScale: An easier way of specifying J1
%.
MakeFigure: Make a figure or simply return the output.
%.
BlackandWhite: Create black and white figures
%.
AR1: the ar1 coefficients of the series
%.
(default='auto' using a naive ar1 estimator. See ar1nv.m)
%.
MonteCarloCount: Number of surrogate data sets in the significance
calculation. (default=300)
%.
ArrowDensity (default: [30 30])
%.
ArrowSize (default: 1)
%.
ArrowHeadSize (default: 1)
%
% Settings can also be specified using abbreviations. e.g. ms=MaxScale.
% For detailed help on some parameters type help wavelet.
%
% Example:
% t=1:200;
% wtc(sin(t),sin(t.*cos(t*.01)),'ms',16)
%
% Phase arrows indicate the relative phase relationship between the series
% (pointing right: in-phase; left: anti-phase; down: series1 leading
% series2 by 90°)
%
% Please acknowledge the use of this software in any publications:
% "Crosswavelet and wavelet coherence software were provided by
% A. Grinsted."
%
% (C) Aslak Grinsted 2002-2004
%
% http://www.pol.ac.uk/home/research/waveletcoherence/

251

Wavelet Coherence Script (cont’d)
% ------------------------------------------------------------------------% Copyright (C) 2002-2004, Aslak Grinsted
% This software may be used, copied, or redistributed as long as it is not
% sold and this copyright notice is reproduced on each copy made. This
% routine is provided as is without any express or implied warranties
% whatsoever.
% ------validate and reformat timeseries.
[x,dt]=formatts(x);
[y,dty]=formatts(y);
if dt~=dty
error('timestep must be equal between time series')
end
t=(max(x(1,1),y(1,1)):dt:min(x(end,1),y(end,1)))'; %common time period
if length(t)<4
error('The two time series must overlap.')
end
n=length(t);
%----------default arguments for the wavelet transform----------Args=struct('Pad',1,...
% pad the time series with zeroes (recommended)
'Dj',1/12, ... % this will do 12 sub-octaves per octave
'S0',2*dt,... % this says start at a scale of 2 months
'J1',[],...
'Mother','Morlet', ...
'MaxScale',[],... %a more simple way to specify J1
'MakeFigure',(nargout==0),...
'MonteCarloCount',300,...
'BlackandWhite',0,...
'AR1','auto',...
'ArrowDensity',[30 30],...
'ArrowSize',1,...
'ArrowHeadSize',1);
Args=parseArgs(varargin,Args,{'BlackandWhite'});
if isempty(Args.J1)
if isempty(Args.MaxScale)
Args.MaxScale=(n*.17)*2*dt; %auto maxscale
end
Args.J1=round(log2(Args.MaxScale/Args.S0)/Args.Dj);
end

252

Wavelet Coherence Script (cont’d)
ad=mean(Args.ArrowDensity);
Args.ArrowSize=Args.ArrowSize*30*.03/ad;
%Args.ArrowHeadSize=Args.ArrowHeadSize*Args.ArrowSize*220;
Args.ArrowHeadSize=Args.ArrowHeadSize*120/ad;
if ~strcmpi(Args.Mother,'morlet')
warning('WTC:InappropriateSmoothingOperator','Smoothing operator is designed for
morlet wavelet.')
end
if strcmpi(Args.AR1,'auto')
Args.AR1=[ar1nv(x(:,2)) ar1nv(y(:,2))];
if any(isnan(Args.AR1))
error('Automatic AR1 estimation failed. Specify it manually (use arcov or arburg).')
end
end
nx=size(x,1);
%sigmax=std(x(:,2));
ny=size(y,1);
%sigmay=std(y(:,2));
%-----------:::::::::::::--------- ANALYZE ----------::::::::::::-----------[X,period,scale,coix] =
wavelet(x(:,2),dt,Args.Pad,Args.Dj,Args.S0,Args.J1,Args.Mother);%#ok
[Y,period,scale,coiy] = wavelet(y(:,2),dt,Args.Pad,Args.Dj,Args.S0,Args.J1,Args.Mother);
%Smooth X and Y before truncating! (minimize coi)
sinv=1./(scale');
sX=smoothwavelet(sinv(:,ones(1,nx)).*(abs(X).^2),dt,period,Args.Dj,scale);
sY=smoothwavelet(sinv(:,ones(1,ny)).*(abs(Y).^2),dt,period,Args.Dj,scale);
% truncate X,Y to common time interval (this is first done here so that the coi is
minimized)
dte=dt*.01; %to cricumvent round off errors with fractional timesteps
idx=find((x(:,1)>=(t(1)-dte))&(x(:,1)<=(t(end)+dte)));
X=X(:,idx);
sX=sX(:,idx);
coix=coix(idx);

253

Wavelet Coherence Script (cont’d)
idx=find((y(:,1)>=(t(1))-dte)&(y(:,1)<=(t(end)+dte)));
Y=Y(:,idx);
sY=sY(:,idx);
coiy=coiy(idx);
coi=min(coix,coiy);
% -------- Cross wavelet ------Wxy=X.*conj(Y);
% ----------------------- Wavelet coherence --------------------------------sWxy=smoothwavelet(sinv(:,ones(1,n)).*Wxy,dt,period,Args.Dj,scale);
Rsq=abs(sWxy).^2./(sX.*sY);
if (nargout>0)||(Args.MakeFigure)
wtcsig=wtcsignif(Args.MonteCarloCount,Args.AR1,dt,length(t)*2,Args.Pad,Args.Dj,Args.S
0,Args.J1,Args.Mother,.6);
wtcsig=(wtcsig(:,2))*(ones(1,n));
wtcsig=Rsq./wtcsig;
end
if Args.MakeFigure
Yticks = 2.^(fix(log2(min(period))):fix(log2(max(period))));
if Args.BlackandWhite
levels = [0 0.5 0.7 0.8 0.9 1];
[cout,H]=safecontourf(t,log2(period),Rsq,levels);
colorbarf(cout,H)
cmap=[0 1;.5 .9;.8 .8;.9 .6;1 .5];
cmap=interp1(cmap(:,1),cmap(:,2),(0:.1:1)');
cmap=cmap(:,[1 1 1]);
colormap(cmap)
set(gca,'YLim',log2([min(period),max(period)]), ...
'YDir','reverse', 'layer','top', ...
'YTick',log2(Yticks(:)), ...
'YTickLabel',num2str(Yticks'), ...
'layer','top')
ylabel('Period')
hold on
254

Wavelet Coherence Script (cont’d)
%phase plot
aWxy=angle(Wxy);
aaa=aWxy;
aaa(Rsq<.5)=NaN;
%[xx,yy]=meshgrid(t(1:5:end),log2(period));
phs_dt=round(length(t)/Args.ArrowDensity(1));
tidx=max(floor(phs_dt/2),1):phs_dt:length(t);
phs_dp=round(length(period)/Args.ArrowDensity(2));
pidx=max(floor(phs_dp/2),1):phs_dp:length(period);
phaseplot(t(tidx),log2(period(pidx)),aaa(pidx,tidx),Args.ArrowSize,Args.ArrowHeadSize);
if ~all(isnan(wtcsig))
[c,h] = contour(t,log2(period),wtcsig,[1 1],'k');%#ok
set(h,'linewidth',2)
end
%suptitle([sTitle ' coherence']);
plot(t,log2(coi),'k','linewidth',3)
hold off
else
H=imagesc(t,log2(period),Rsq);%#ok
%[c,H]=safecontourf(t,log2(period),Rsq,[0:.05:1]);
%set(H,'linestyle','none')
set(gca,'clim',[0 1])
HCB=safecolorbar;%#ok
set(gca,'YLim',log2([min(period),max(period)]), ...
'YDir','reverse', 'layer','top', ...
'YTick',log2(Yticks(:)), ...
'YTickLabel',num2str(Yticks'), ...
'layer','top')
ylabel('Period')
hold on
%phase plot
aWxy=angle(Wxy);
aaa=aWxy;
aaa(Rsq<.5)=NaN; %remove phase indication where Rsq is low
%[xx,yy]=meshgrid(t(1:5:end),log2(period));
255

Wavelet Coherence Script (cont’d)
phs_dt=round(length(t)/Args.ArrowDensity(1));
tidx=max(floor(phs_dt/2),1):phs_dt:length(t);
phs_dp=round(length(period)/Args.ArrowDensity(2));
pidx=max(floor(phs_dp/2),1):phs_dp:length(period);
phaseplot(t(tidx),log2(period(pidx)),aaa(pidx,tidx),Args.ArrowSize,Args.ArrowHeadSize);
if ~all(isnan(wtcsig))
[c,h] = contour(t,log2(period),wtcsig,[1 1],'k');%#ok
set(h,'linewidth',2)
end
%suptitle([sTitle ' coherence']);
tt=[t([1 1])-dt*.5;t;t([end end])+dt*.5];
hcoi=fill(tt,log2([period([end 1]) coi period([1 end])]),'w');
set(hcoi,'alphadatamapping','direct','facealpha',.5)
hold off
end
end
varargout={Rsq,period,scale,coi,wtcsig};
varargout=varargout(1:nargout);
function [cout,H]=safecontourf(varargin)
vv=sscanf(version,'%i.');
if (version('-release')<14)|(vv(1)<7)
[cout,H]=contourf(varargin{:});
else
[cout,H]=contourf('v6',varargin{:});
end
function hcb=safecolorbar(varargin)
vv=sscanf(version,'%i.');
if (version('-release')<14)|(vv(1)<7)
hcb=colorbar(varargin{:});
else
hcb=colorbar('v6',varargin{:});
end
_______________________________________________________________________

256

APPENDIX 4

Table A-2 Comparison of Social Vulnerability Index and Sub-Indices from High Social
Vulnerability (+) to Low Social Vulnerability (-) and Cholera Incidence by District ID
ID SVI
C
ID SVIF1 C
ID SVIF2 C
32 1.30 15.9 8
1.54 0.0 31 2.59 3.7
31 0.87 3.7 16 1.44 0.6 27 2.12 11.1
30 0.59 12.1 12 1.43 1.2 29 2.01 5.8
27 0.53 11.1 15 0.95 0.1 2
1.10 7.2
20 0.48 1.4 11 0.89 0.4 17 0.84 6.6
12 0.42 1.2 7
0.69 0.4 5
0.64 3.9
29 0.37 5.8 20 0.58 1.4 12 0.49 1.2
6 0.29 9.5 13 0.56 0.8 11 0.45 0.4
16 0.28 0.6 26 0.56 0.3 14 0.40 1.5
28 0.26 3.3 31 0.56 3.7 15 0.40 0.1
3 0.19 2.3 14 0.47 1.5 6
0.36 9.5
4 0.18 2.2 9
0.42 1.7 3
0.17 2.3
5 0.14 3.9 19 0.32 3.9 13 0.10 0.8
17 0.12 6.6 32 0.25 15.9 26 0.09 0.3
22 0.12 6.5 4
0.20 2.2 22 0.07 6.5
15 0.09 0.1 30 0.17 12.1 32 0.01 15.9
8 0.06 0.0 3
0.11 2.3 23 -0.06 13.8
2 0.04 7.2 25 0.00 1.4 8 -0.09 0.0
11 -0.06 0.4 6 -0.04 9.5 16 -0.13 0.6
14 -0.13 1.5 22 -0.06 6.5 1 -0.16 7.4
23 -0.15 13.8 5 -0.08 3.9 10 -0.21 1.5
13 -0.19 0.8 10 -0.08 1.5 21 -0.29 0.7
9 -0.24 1.7 23 -0.15 13.8 30 -0.37 12.1
26 -0.26 0.3 28 -0.24 3.3 0 -0.42 5.4
7 -0.29 0.4 24 -0.35 4.0 9 -0.44 1.7
10 -0.32 1.5 17 -0.37 6.6 28 -0.56 3.3
21 -0.44 0.7 29 -0.53 5.8 20 -0.70 1.4
18 -0.49 0.5 18 -0.71 0.5 18 -1.17 0.5
19 -0.49 3.9 27 -0.79 11.1 24 -1.31 4.0
24 -0.64 4.0 2 -0.88 7.2 7 -1.35 0.4
25 -0.64 1.4 21 -0.92 0.7 19 -1.43 3.9
1 -0.74 7.4 1 -2.55 7.4 25 -1.47 1.4
0 -1.26 5.4 0 -3.39 5.4 4 -1.69 2.2
C represents the total cholera incidence rate for 1997 and 1998 for comparison

257

Table A-2 (cont’d)
ID SVIF3 C
ID SVIF4 C
20 2.66 1.4 32 4.60 15.9
4
2.45 2.2 30 1.38 12.1
5
1.34 3.9 28 1.33 3.3
6
1.30 9.5 31 0.51 3.7
30 1.19 12.1 27 0.41 11.1
16 0.64 0.6 18 0.24 0.5
3
0.54 2.3 13 0.15 0.8
22 0.52 6.5 25 0.11 1.4
28 0.50 3.3 24 0.10 4.0
21 0.39 0.7 2
0.09 7.2
27 0.39 11.1 7
0.09 0.4
32 0.35 15.9 19 0.06 3.9
12 0.31 1.2 9
0.03 1.7
17 0.18 6.6 29 0.01 5.8
29 -0.03 5.8 3 -0.05 2.3
2 -0.15 7.2 22 -0.07 6.5
31 -0.16 3.7 1 -0.08 7.4
1 -0.17 7.4 11 -0.16 0.4
23 -0.22 13.8 23 -0.16 13.8
18 -0.30 0.5 17 -0.18 6.6
8 -0.50 0.0 4 -0.25 2.2
0 -0.51 5.4 15 -0.33 0.1
7 -0.58 0.4 10 -0.34 1.5
14 -0.64 1.5 6 -0.46 9.5
15 -0.64 0.1 12 -0.56 1.2
10 -0.67 1.5 26 -0.56 0.3
19 -0.93 3.9 20 -0.62 1.4
9 -0.97 1.7 8 -0.70 0.0
24 -1.00 4.0 0 -0.71 5.4
26 -1.12 0.3 14 -0.76 1.5
25 -1.20 1.4 16 -0.84 0.6
11 -1.41 0.4 21 -0.94 0.7
13 -1.56 0.8 5 -1.34 3.9
C represents the total cholera incidence rate for 1997 and 1998 for comparison

258

APPENDIX 5

Table A-3 Climate-cholera Associations by R-Squared and P-Values from Ordinary Least
Squares Regression (OLS) by District
ID
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0

District
PIURA
PIURA
PIURA
PIURA
PIURA
PIURA
PIURA
PIURA
PIURA
PIURA
PIURA
PIURA
PIURA
PIURA
PIURA
PIURA
PIURA
PIURA
PIURA
PIURA
PIURA
PIURA
PIURA
PIURA
PIURA
PIURA
PIURA
PIURA
PIURA
PIURA
PIURA
PIURA
PIURA
PIURA
PIURA

2

Variable
Lag r
p value
Rainfall
1 0.85 0.0000
PAITA SSTA
0 0.71 0.0006
Rainfall
0 0.71 0.0006
NIÑO 3.4 SSTA
1 0.68 0.0010
TMEANA
5 0.67 0.0012
TMAXA
5 0.65 0.0014
NIÑO 3.4 SSTA
0 0.65 0.0016
NIÑO 3.4 SSTA
2 0.63 0.0021
PAITA SSTA
1 0.61 0.0026
TMINA
5 0.58 0.0038
TMAXA
4 0.58 0.0039
TMEANA
3 0.58 0.0043
TMAXA
6 0.57 0.0043
TMEANA
6 0.57 0.0045
TMINA
0 0.56 0.0050
TMINA
6 0.56 0.0054
TMEANA
0 0.54 0.0065
NIÑO 3.4 SSTA
3 0.54 0.0068
TMINA
3 0.52 0.0080
TMEANA
4 0.52 0.0085
PAITA SSTA
2 0.51 0.0086
TMINA
1 0.50 0.0101
TMEANA
2 0.50 0.0102
TMEANA
1 0.49 0.0112
NIÑO 1+2 SSTA 2 0.48 0.0128
TMINA
2 0.46 0.0161
NIÑO 1+2 SSTA 3 0.45 0.0163
NIÑO 3.4 SSTA
4 0.44 0.0189
PAITA SSTA
3 0.44 0.0189
TMAXA
7 0.44 0.0189
TMAXA
3 0.42 0.0222
Rainfall
2 0.40 0.0264
NIÑO 1+2 SSTA 1 0.40 0.0271
NIÑO 1+2 SSTA 0 0.39 0.0295
TMINA
4 0.39 0.0299

259

Table A-3 (cont’d)
ID
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1

District
CASTILLA
CASTILLA
CASTILLA
CASTILLA
CASTILLA
CASTILLA
CASTILLA
CASTILLA
CASTILLA
CASTILLA
CASTILLA
CASTILLA
CASTILLA
CASTILLA
CASTILLA
CASTILLA
CASTILLA
CASTILLA
CASTILLA
CASTILLA
CASTILLA
CASTILLA
CASTILLA
CASTILLA
CASTILLA
CASTILLA
CASTILLA
CASTILLA
CASTILLA
CASTILLA
CASTILLA
CASTILLA

2

Variable
Lag r
Rainfall
1 0.85
PAITA SSTA
0 0.65
Rainfall
0 0.65
TMAXA
5 0.64
TMEANA
5 0.60
NIÑO 3.4 SSTA
1 0.60
NIÑO 3.4 SSTA
0 0.57
TMAXA
6 0.56
TMAXA
4 0.56
TMINA
6 0.56
TMEANA
6 0.56
NIÑO 3.4 SSTA
2 0.56
PAITA SSTA
1 0.55
TMINA
0 0.51
TMINA
5 0.51
TMEANA
3 0.49
TMEANA
0 0.48
NIÑO 3.4 SSTA
3 0.47
PAITA SSTA
2 0.45
Rainfall
2 0.45
TMINA
3 0.44
TMEANA
4 0.44
TMINA
1 0.44
TMEANA
1 0.41
TMEANA
2 0.40
TMAXA
7 0.40
NIÑO 1+2 SSTA 3 0.39
NIÑO 1+2 SSTA 2 0.39
NIÑO 3.4 SSTA
4 0.38
PAITA SSTA
3 0.38
TMAXA
3 0.37
TMINA
2 0.36

260

p value
0.0000
0.0015
0.0016
0.0017
0.0031
0.0031
0.0046
0.0050
0.0053
0.0053
0.0053
0.0054
0.0054
0.0088
0.0089
0.0114
0.0123
0.0143
0.0172
0.0174
0.0185
0.0190
0.0191
0.0245
0.0272
0.0277
0.0290
0.0296
0.0320
0.0334
0.0364
0.0378

Table A-3 (cont’d)
ID
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2

District
CATACAOS
CATACAOS
CATACAOS
CATACAOS
CATACAOS
CATACAOS
CATACAOS
CATACAOS
CATACAOS
CATACAOS
CATACAOS
CATACAOS
CATACAOS
CATACAOS
CATACAOS
CATACAOS
CATACAOS
CATACAOS
CATACAOS
CATACAOS
CATACAOS
CATACAOS
CATACAOS
CATACAOS
CATACAOS
CATACAOS
CATACAOS
CATACAOS

2

Variable
Lag r
Rainfall
1 0.84
Rainfall
2 0.64
TMEANA
5 0.54
TMAXA
5 0.53
PAITA SSTA
0 0.53
TMAXA
6 0.51
TMEANA
6 0.50
PAITA SSTA
1 0.48
NIÑO 3.4 SSTA
2 0.48
TMAXA
7 0.47
TMINA
0 0.47
NIÑO 3.4 SSTA
3 0.46
TMINA
5 0.46
NIÑO 3.4 SSTA
1 0.46
TMEANA
0 0.44
PAITA SSTA
2 0.44
TMINA
6 0.43
TMEANA
7 0.43
TMINA
7 0.41
NIÑO 3.4 SSTA
0 0.41
PAITA SSTA
3 0.41
NIÑO 3.4 SSTA
4 0.40
NIÑO 1+2 SSTA 3 0.39
TMEANA
3 0.36
TMINA
3 0.36
NIÑO 1+2 SSTA 0 0.35
TMINA
2 0.34
Rainfall
0 0.34

261

p value
0.0000
0.0018
0.0063
0.0071
0.0076
0.0091
0.0105
0.0120
0.0120
0.0133
0.0133
0.0148
0.0148
0.0159
0.0186
0.0191
0.0211
0.0211
0.0243
0.0252
0.0260
0.0262
0.0308
0.0383
0.0392
0.0444
0.0452
0.0471

Table A-3 (cont’d)
ID
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5

District
LA ARENA
LA ARENA
LA ARENA
LA ARENA
LA ARENA
LA ARENA
LA ARENA
LA ARENA
LA ARENA
LA ARENA
LA ARENA
LA ARENA
LA ARENA
LA ARENA
LA ARENA
LA ARENA
LA ARENA
LA ARENA
LA ARENA
LA ARENA
LA ARENA
LA ARENA
LA ARENA
LA ARENA
LA ARENA
LA ARENA
LA ARENA
LA ARENA
LA ARENA
LA ARENA
LA ARENA
LA ARENA
LA ARENA
LA ARENA
LA ARENA
LA ARENA
LA ARENA
LA ARENA

2

Variable
Lag r
PAITA SSTA
0 0.87
TMAXA
5 0.81
Rainfall
1 0.81
NIÑO 3.4 SSTA
0 0.78
TMAXA
6 0.76
NIÑO 3.4 SSTA
1 0.74
PAITA SSTA
1 0.71
TMEANA
5 0.70
TMINA
1 0.70
TMEANA
1 0.70
NIÑO 3.4 SSTA
2 0.69
TMINA
0 0.66
PAITA SSTA
2 0.66
Rainfall
0 0.66
TMEANA
6 0.63
TMEANA
0 0.63
NIÑO 3.4 SSTA
3 0.62
NIÑO 1+2 SSTA 0 0.61
TMEANA
3 0.58
TMINA
5 0.58
TMAXA
7 0.58
TMINA
3 0.57
TMINA
6 0.56
Rainfall
2 0.55
TMEANA
4 0.54
NIÑO 3.4 SSTA
4 0.52
TMEANA
2 0.52
TMAXA
4 0.51
TMINA
2 0.49
NIÑO 1+2 SSTA 1 0.49
PAITA SSTA
3 0.48
TMINA
4 0.42
NIÑO 1+2 SSTA 4 0.42
NIÑO 1+2 SSTA 2 0.41
NIÑO 1+2 SSTA 3 0.40
NIÑO 1+2 SSTA 5 0.39
NIÑO 1+2 SSTA 6 0.35
NIÑO 3.4 SSTA
5 0.34

262

p value
0.0000
0.0001
0.0001
0.0001
0.0002
0.0003
0.0006
0.0006
0.0007
0.0007
0.0008
0.0013
0.0014
0.0014
0.0020
0.0022
0.0023
0.0026
0.0038
0.0041
0.0043
0.0047
0.0054
0.0055
0.0063
0.0080
0.0081
0.0092
0.0108
0.0113
0.0127
0.0233
0.0233
0.0242
0.0281
0.0300
0.0409
0.0479

Table A-3 (cont’d)
ID
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6

District
LA UNION
LA UNION
LA UNION
LA UNION
LA UNION
LA UNION
LA UNION
LA UNION
LA UNION
LA UNION
LA UNION
LA UNION
LA UNION
LA UNION
LA UNION
LA UNION
LA UNION
LA UNION
LA UNION
LA UNION
LA UNION
LA UNION
LA UNION
LA UNION
LA UNION
LA UNION
LA UNION
LA UNION
LA UNION
LA UNION
LA UNION
LA UNION
LA UNION
LA UNION
LA UNION
LA UNION
LA UNION
LA UNION
LA UNION
LA UNION
LA UNION

2

Variable
Lag r
Rainfall
1 0.89
Rainfall
2 0.82
TMAXA
6 0.79
PAITA SSTA
0 0.77
TMEANA
6 0.75
PAITA SSTA
1 0.73
TMAXA
5 0.71
NIÑO 3.4 SSTA
1 0.69
TMAXA
7 0.69
TMINA
6 0.68
NIÑO 3.4 SSTA
2 0.68
TMEANA
5 0.68
PAITA SSTA
2 0.66
NIÑO 3.4 SSTA
0 0.66
TMINA
0 0.65
TMINA
1 0.65
NIÑO 3.4 SSTA
3 0.64
TMEANA
0 0.63
TMEANA
1 0.59
NIÑO 3.4 SSTA
4 0.57
NIÑO 1+2 SSTA 0 0.57
PAITA SSTA
3 0.56
TMINA
5 0.52
TMEANA
7 0.52
TMINA
3 0.50
TMEANA
3 0.50
Rainfall
0 0.50
TMEANA
4 0.48
TMINA
2 0.48
TMEANA
2 0.47
NIÑO 1+2 SSTA 1 0.45
NIÑO 1+2 SSTA 4 0.43
NIÑO 1+2 SSTA 3 0.43
NIÑO 3.4 SSTA
5 0.43
NIÑO 1+2 SSTA 2 0.40
NIÑO 1+2 SSTA 6 0.40
NIÑO 1+2 SSTA 5 0.39
TMINA
7 0.38
TMINA
4 0.37
TMAXA
4 0.37
PAITA SSTA
4 0.33
263

p value
0.0000
0.0001
0.0001
0.0002
0.0003
0.0004
0.0006
0.0009
0.0009
0.0009
0.0009
0.0010
0.0014
0.0014
0.0016
0.0017
0.0018
0.0022
0.0036
0.0043
0.0046
0.0050
0.0080
0.0085
0.0101
0.0101
0.0103
0.0120
0.0126
0.0141
0.0178
0.0206
0.0208
0.0209
0.0265
0.0285
0.0290
0.0338
0.0347
0.0355
0.0491

Table A-3 (cont’d)
ID
9
9
9
9
9
9
9
9
9

District
HUANCABAMBA
HUANCABAMBA
HUANCABAMBA
HUANCABAMBA
HUANCABAMBA
HUANCABAMBA
HUANCABAMBA
HUANCABAMBA
HUANCABAMBA

ID
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12

District
HUARMACA
HUARMACA
HUARMACA
HUARMACA
HUARMACA
HUARMACA
HUARMACA
HUARMACA
HUARMACA
HUARMACA
HUARMACA
HUARMACA
HUARMACA
HUARMACA
HUARMACA
HUARMACA
HUARMACA
HUARMACA
HUARMACA
HUARMACA
HUARMACA
HUARMACA
HUARMACA

2

Variable Lag r
Rainfall
2 0.67
TMINA
7 0.44
TMAXA
7 0.44
TMAXA
6 0.41
Rainfall
3 0.40
TMEANA 6 0.38
Rainfall
1 0.37
TMEANA 7 0.37
TMINA
6 0.34
2

Variable
Lag r
Rainfall
2 0.64
TMEANA
6 0.63
TMINA
6 0.62
PAITA SSTA
0 0.50
TMINA
1 0.46
TMAXA
6 0.46
Rainfall
1 0.45
PAITA SSTA
1 0.42
TMAXA
5 0.41
NIÑO 3.4 SSTA
2 0.40
NIÑO 3.4 SSTA
3 0.39
TMEANA
1 0.38
NIÑO 1+2 SSTA 4 0.37
NIÑO 3.4 SSTA
0 0.37
NIÑO 3.4 SSTA
1 0.36
TMINA
0 0.36
PAITA SSTA
2 0.36
NIÑO 1+2 SSTA 0 0.36
TMEANA
4 0.36
NIÑO 3.4 SSTA
4 0.34
Rainfall
0 0.34
TMEANA
0 0.34
NIÑO 1+2 SSTA 3 0.33

264

p value
0.0012
0.0180
0.0191
0.0255
0.0273
0.0330
0.0349
0.0355
0.0471

p value
0.0019
0.0021
0.0024
0.0098
0.0148
0.0151
0.0169
0.0232
0.0252
0.0269
0.0289
0.0315
0.0346
0.0357
0.0383
0.0391
0.0393
0.0396
0.0408
0.0451
0.0456
0.0460
0.0489

Table A-3 (cont’d)
ID
17
17
17
17
17
17
17
17
17
17
17
17
17
17
17
17
17
17
17
17
17
17
17
17
17
17
17
17
17
17
17
17
17
17

District
CHULUCANAS
CHULUCANAS
CHULUCANAS
CHULUCANAS
CHULUCANAS
CHULUCANAS
CHULUCANAS
CHULUCANAS
CHULUCANAS
CHULUCANAS
CHULUCANAS
CHULUCANAS
CHULUCANAS
CHULUCANAS
CHULUCANAS
CHULUCANAS
CHULUCANAS
CHULUCANAS
CHULUCANAS
CHULUCANAS
CHULUCANAS
CHULUCANAS
CHULUCANAS
CHULUCANAS
CHULUCANAS
CHULUCANAS
CHULUCANAS
CHULUCANAS
CHULUCANAS
CHULUCANAS
CHULUCANAS
CHULUCANAS
CHULUCANAS
CHULUCANAS

ID
19
19
19

District
CHALACO
CHALACO
CHALACO

2

Variable
Lag r
Rainfall
2 0.86
TMEANA
7 0.71
TMINA
7 0.71
TMEANA
6 0.68
TMINA
6 0.66
PAITA SSTA
3 0.59
TMAXA
7 0.59
Rainfall
1 0.59
NIÑO 3.4 SSTA
4 0.56
NIÑO 3.4 SSTA
3 0.55
NIÑO 3.4 SSTA
5 0.54
PAITA SSTA
2 0.51
NIÑO 3.4 SSTA
2 0.51
NIÑO 1+2 SSTA 7 0.51
NIÑO 1+2 SSTA 3 0.51
Rainfall
3 0.49
PAITA SSTA
1 0.49
TMEANA
5 0.47
TMINA
0 0.45
NIÑO 3.4 SSTA
6 0.44
TMEANA
0 0.44
PAITA SSTA
0 0.43
TMAXA
6 0.42
NIÑO 1+2 SSTA 4 0.42
NIÑO 1+2 SSTA 0 0.42
TMINA
1 0.42
TMINA
3 0.41
PAITA SSTA
4 0.41
NIÑO 3.4 SSTA
1 0.41
NIÑO 1+2 SSTA 2 0.38
TMINA
5 0.37
TMEANA
3 0.36
TMEANA
1 0.36
TMINA
2 0.35
2

Variable Lag r
Rainfall
2 0.51
TMINA
6 0.43
TMEANA 6 0.36

p value
0.0093
0.0199
0.0401
265

p value
0.0000
0.0006
0.0006
0.0009
0.0013
0.0035
0.0036
0.0037
0.0051
0.0058
0.0065
0.0087
0.0090
0.0091
0.0094
0.0111
0.0118
0.0133
0.0174
0.0181
0.0190
0.0204
0.0217
0.0221
0.0226
0.0235
0.0240
0.0254
0.0257
0.0327
0.0348
0.0376
0.0396
0.0439

Table A-3 (cont’d)
ID
22
22
22
22
22
22
22
22
22
22
22
22
22
22
22
22
22
22
22
22
22
22
22
22
22
22
22
22
22
22
22
22
22
22
22
22
22

District
SALITRAL1
SALITRAL1
SALITRAL1
SALITRAL1
SALITRAL1
SALITRAL1
SALITRAL1
SALITRAL1
SALITRAL1
SALITRAL1
SALITRAL1
SALITRAL1
SALITRAL1
SALITRAL1
SALITRAL1
SALITRAL1
SALITRAL1
SALITRAL1
SALITRAL1
SALITRAL1
SALITRAL1
SALITRAL1
SALITRAL1
SALITRAL1
SALITRAL1
SALITRAL1
SALITRAL1
SALITRAL1
SALITRAL1
SALITRAL1
SALITRAL1
SALITRAL1
SALITRAL1
SALITRAL1
SALITRAL1
SALITRAL1
SALITRAL1

2

Variable
Lag r
Rainfall
2 0.75
PAITA SSTA
3 0.71
TMEANA
6 0.70
TMINA
6 0.66
TMAXA
7 0.60
NIÑO 3.4 SSTA
4 0.60
NIÑO 1+2 SSTA 7 0.59
NIÑO 3.4 SSTA
3 0.59
NIÑO 3.4 SSTA
5 0.58
TMEANA
7 0.56
TMINA
1 0.55
PAITA SSTA
2 0.53
NIÑO 3.4 SSTA
2 0.52
NIÑO 1+2 SSTA 0 0.51
Rainfall
3 0.51
TMINA
3 0.51
PAITA SSTA
4 0.50
PAITA SSTA
1 0.50
TMEANA
1 0.49
NIÑO 1+2 SSTA 3 0.48
NIÑO 3.4 SSTA
6 0.47
NIÑO 1+2 SSTA 1 0.45
NIÑO 3.4 SSTA
1 0.44
NIÑO 1+2 SSTA 2 0.43
TMAXA
6 0.43
TMINA
7 0.43
PAITA SSTA
0 0.42
TMINA
2 0.39
TMEANA
3 0.39
TMEANA
2 0.39
NIÑO 1+2 SSTA 4 0.38
TMEANA
5 0.37
TMEANA
4 0.36
Rainfall
1 0.36
TMINA
4 0.34
NIÑO 3.4 SSTA
0 0.34
TMEANA
0 0.34

266

p value
0.0002
0.0006
0.0007
0.0012
0.0029
0.0031
0.0035
0.0036
0.0040
0.0054
0.0059
0.0071
0.0082
0.0087
0.0088
0.0094
0.0098
0.0101
0.0118
0.0130
0.0137
0.0173
0.0193
0.0206
0.0213
0.0213
0.0222
0.0288
0.0292
0.0309
0.0319
0.0368
0.0388
0.0406
0.0461
0.0465
0.0467

Table A-3 (cont’d)
ID
27
27
27
27
27
27
27
27
27
27
27
27
27
27
27
27
27
27
27
27
27
27
27
27
27
27
27
27
27
27
27
27
27
27
27
27
27
27

District
SECHURA
SECHURA
SECHURA
SECHURA
SECHURA
SECHURA
SECHURA
SECHURA
SECHURA
SECHURA
SECHURA
SECHURA
SECHURA
SECHURA
SECHURA
SECHURA
SECHURA
SECHURA
SECHURA
SECHURA
SECHURA
SECHURA
SECHURA
SECHURA
SECHURA
SECHURA
SECHURA
SECHURA
SECHURA
SECHURA
SECHURA
SECHURA
SECHURA
SECHURA
SECHURA
SECHURA
SECHURA
SECHURA

2

Variable
Lag r
Rainfall
1 0.84
PAITA SSTA
0 0.71
TMEANA
5 0.71
NIÑO 3.4 SSTA
1 0.67
NIÑO 3.4 SSTA
2 0.64
PAITA SSTA
1 0.63
TMEANA
6 0.63
TMINA
5 0.63
NIÑO 3.4 SSTA
0 0.63
TMAXA
6 0.63
TMINA
0 0.62
TMINA
6 0.60
TMAXA
5 0.60
PAITA SSTA
2 0.59
NIÑO 3.4 SSTA
3 0.58
TMEANA
3 0.57
TMINA
3 0.57
TMEANA
0 0.57
TMAXA
7 0.55
Rainfall
2 0.54
TMINA
1 0.54
Rainfall
0 0.53
TMEANA
1 0.53
PAITA SSTA
3 0.52
NIÑO 3.4 SSTA
4 0.52
TMEANA
2 0.49
NIÑO 1+2 SSTA 2 0.48
NIÑO 1+2 SSTA 0 0.47
TMINA
2 0.47
NIÑO 1+2 SSTA 1 0.46
TMEANA
4 0.46
NIÑO 1+2 SSTA 3 0.43
TMAXA
4 0.38
NIÑO 3.4 SSTA
5 0.38
NIÑO 1+2 SSTA 6 0.37
NIÑO 1+2 SSTA 5 0.34
NIÑO 1+2 SSTA 4 0.34
TMINA
4 0.33

267

p value
0.0000
0.0006
0.0006
0.0011
0.0017
0.0019
0.0020
0.0020
0.0021
0.0021
0.0025
0.0030
0.0030
0.0034
0.0041
0.0043
0.0044
0.0047
0.0056
0.0065
0.0067
0.0072
0.0074
0.0078
0.0078
0.0114
0.0124
0.0134
0.0143
0.0149
0.0159
0.0197
0.0324
0.0340
0.0355
0.0449
0.0457
0.0488

Table A-3 (cont’d)
ID
32
32
32
32
32
32

District
RINCONADA LLICUAR
RINCONADA LLICUAR
RINCONADA LLICUAR
RINCONADA LLICUAR
RINCONADA LLICUAR
RINCONADA LLICUAR

2

Variable
Lag r
TMINA
6 0.46
Rainfall
1 0.44
Rainfall
2 0.39
PAITA SSTA 0 0.36
TMEANA
6 0.35
TMAXA
5 0.34

268

p value
0.0150
0.0179
0.0302
0.0394
0.0417
0.0460

REFERENCES

269

REFERENCES

Abouharb, M.R. and D. Cingranelli, 2007: Human Rights and Structural Adjustment.
Cambridge University Press.
Adeel, Z. and M.H. Glantz., 2001: El Niño of the century: once burnt, twice shy? Global
Environmental Change, 11, 171-174.
Adger, N.W., 2006: Vulnerability. Global Environmental Change, 16, 268-281.
Akanda, A.S., S.J. Antarpreet, and S. Islam, 2009: Dual peak cholera transmission in
Bengal Delta: A hydroclimatological explanation. Geophysical Research Letters, 36
[Available online at
http://www.agu.org.proxy2.cl.msu.edu/journals/gl/gl0919/2009GL039312/2009GL0393
12.pdf]
Alam, M. and Coauthors, 2007: Viable but nonculturable Vibrio cholera 01 in biofilms in
the aquatic environment and their role in cholera transmission. Proceedings of the
National Academy of Sciences, 104, 17801-17806.
Ali, M., M. Emch, J.P. Donnay, M. Yunus, and R.B. Sack, 2002: The spatial epidemiology
of cholera in an endemic area of Bangladesh. Social Science and Medicine, 55, 10151024.
Amat y Leon, C., B. Seminario, M. Paz Cigaran, S. Bambaren, L. Macera, M.T. Cigaran,
and D. Vasquez, 2008: El Cambio Climático No Tiene Fronteras: Impacto Del Cambio
Climatico En La Comunidad Andina (Climate Change has no Frontiers: Impacts on the
Andean Community). Comunidad Andina. (in Spanish)
Audy, J.R., 1971: Measurement and diagnosis of health. Geographical Analysis, 5.
Barber, R.T. and F.P. Chavez, 1983: Biological consequences of El Niño. Science, 22,
1203-1210.
Barnett, T. and P. Blaikie, 1992: AIDS in Africa. Its present and future impact. John Wiley.
Mahalanabis, D., A.M. Molla, and D. Sack, 1992: Clinical management of cholera. In
Cholera, Barua, D., W.B. and Greenough III, Eds. Plenum Medical.
Bell, G.D. and M.S. Halpert, 1998: Climate assessment for 1997. Bull. Amer. Meteor. Soc.,
79, S1-50.
Bell, G.D. and Coauthors, 2000: Climate assessment for 1999. Bull. Amer. Meteor. Soc.,
81, S1-50.
270

Besser, R.E., and Coauthors, 1995: Prevención de la transmissión del cólera: evaluación
rápida de la calidad del agua municipal en Trujillo, Peru (Prevention of cholera
transmisión: rapid evaluation of water quality of municipal water supply in Trujillo,
Peru). Boletin Oficina Sanit Panam, 119, 189-193. (in Spanish)
Bjerknes, J., 1969: Atmospheric teleconnections from the equatorial Pacific. Mon. Wea.
Rev., 97, 163-172.
Bouma, M.J. and M. Pascual, 2001: Seasonal and interannual cycles of endemic cholera
in Bengal 1891–1940 in relation to climate and geography. Hydrobiologia, 460, 147–156.
Bouma, M.J., R.S. Kovats, S.A. Goubet, J.St.H. Cox and A. Haines, 1997: Global
assessment of El Niño’s disaster burden. The Lancet, 350, 1435-1438.
Briggs, C.L. and C. Mantini-Briggs, 2003: Stories in the Time of Cholera: Racial Profiling
during a Medical Nightmare. University of California Press.
Brooke, J., 1991: Feeding on 19th century conditions, cholera spreads in Latin America.
NY Times, 21 April, [Available online at
http://www.nytimes.com/1991/04/21/weekinreview/the-world-feeding-on-19thcentury-conditions-cholera-spreads-in-latin-america.html]
Bureau of Meteorology in Australia (BOM), 2005: El Niño, La Niña and Australia’s
climate. [Available online at http://www.bom.gov.au/info/leaflets/Niño-Niña.pdf]
Bureau of Meteorology in Australia (BOM), 2010: About ENSO monitoring graphs.
[Available online at http://www.bom.gov.au/climate/enso/indices/about.shtml]
Burton, I., R.W. Kates, and G.F. White, 1993: The Environment as Hazard. 2nd ed.
Guildford Press.
Cane, M.A., 2004: The evolution of El Niño, past and future. Earth Planet. Sci. Lett., 164,
1-10.
Carr, M. and K. Broad, 2000: Satellites, society, and the Peruvian fisheries during the
1997-1998 El Niño. In Satellites, Océanography and Society, D. Halpern, Ed., Elsevier
Science B.V., pp. 171-191.
Carillo, C., 1892: Disertación sobre las Corrientes Oceanicas y Estudios de la Corriente
Peruana de Humboldt (Examination of Ocean Currents and the study of the Humboldt
Current). Boletines de Sociedad Geografico Lima, 11, 84. (in Spanish)
Caviedes, C.N., 1973: Secas and El Niño: two simultaneous climatical hazards in South
America. Procession of the Association of American Geographers, 5, 44-49.
--------1982: Natural hazards in South America: in search of a method and a theory.
GeoJournal, 6, 101-109.

271

--------1984: El Niño 1982-83. Geographical Review, 74, 267-290.
--------2001: El Niño in History: Storming through the Ages. University Press of Florida.
Caviedes, C.N. and T.J. Fik, 1992: The Peru-Chile eastern fisheries and climatic
oscillation. In Climate Variability, Climate Change and Fisheries, M.H. Glantz, Ed.
Cambridge University Press, pp. 355-376.
Cazelles, B., M. Chavez, A.J. McMichael, and S. Hales, 2005: Nonstationary influences of
El Niño on the synchronous dengue epidemics in Thailand. Public Library of Science, 2,
e106. [Available online at http://www.plos.org/]
Cazelles, B., M. Chavez, G. Constantin de Magny, J. Guegan and S. Hales, 2007: Timedependent spectral analysis of epidemiological time-series with wavelets. The Journal of
the Royal Society Interface, 4, 625-636.
Center for Disease Control and Prevention (CDC), 1991: International notes cholera –
Peru, 1991. Morbidity and Mortality Weekly Report, 40, 108-109. [Available online at
http://www.cdc.gov/mmwr/preview/mmwrhtml/00001912.htm]
--------2009: Syndemics definition. Syndemics Prevention Networks. [Available online at
http://www.cdc.gov/syndemics/definition.htm]
Centro de Operaciones de Emergencia (CTAR). 1998. Memoria 1997-1998 (Memory of
1997-1998), CTAR. (in Spanish)
Chaves, L.F. and M. Pascual, 2006: Climate Cycles and Forecasts of Cutaneous
Leishmaniasis, a Nonstationary Vector-Borne Disease. Public Library of Science, 3, e295.
[Available online at
http://www.plosmedicine.org/article/info%3Adoi%2F10.1371%2Fjournal.pmed.003029
5]
Chavez, F.P., A. Bertrand, R. Guevarro-Carrasco, P. Soler and J. Csirke, 2008: The
northern Humboldt Current system: ocean dynamics, ecosystem processes, and
fisheries. Progress in Océanography, 79, 1-15. [Available online at
http://www.imarpe.gob.pe/paita/documentos/Editorial_Conf_Humboldt.pdf]
Checkley, W., L.D. Epstein, R.H. Gilman, D. Figueroa, R.I. Cama, J.A. Patz, and R.E. Black,
2000: Effect of El Niño and ambient temperature on hospital admissions for diarrhoeal
diseases in Peruvian children. The Lancet, 355, 442–450.
Cisneros, F. and N. Mujica, 1999: Impacto del cambia climatico en la agricultura: efectos
del fenómeno El Niño en los cultivos de la costa central (Impacts from climate change on
agricultura: the effect of El Niño on crops in the central coast). In Peru: Vulnerabilidad
Frente Al Cambio Climático: Aproximaciones a la experiencia con el fenómeno El Niño
(Peru: Vulnerability to Climate Change: The experience of the El Niño Phenomenon as an
Analogue), B. Marticorena, Ed. Consejo Nacional de Ambiente, pp. 115-136. (in Spanish)
272

Cockerham, W.C., 2007: Social Causes of Health and Disease. Polity Press.
Codeco, C.T., 2001: Endemic and epidemic dynamics of cholera: the role of the aquatic
reservoir. BMC Infectious Diseases, 1. [Available online at
http://www.biomedcentral.com/1471-2334/1/1]
Collins, A.E., 1998: Environment, Health and Population Displacement: Development and
Change in Mozambique’s Diarrhoeal Disease Ecology. Ashgate.
Consejo Nacional del Ambiente (National Environmental Agency) (CONAM), 2002:
Estrategia Nacional de Cambio Climatico, Peru (National Climate Change Strategy).
Version 8, 13 December. [Available online at
http://sinia.minam.gob.pe/index.php?idElementoInformacion=1] (in Spanish)
Consortium for Capacity Building (CCB), 2010: Climate Affairs: What do we mean and
why? [Available online at http://ccb.colorado.edu/climate.php]
Colwell, R.R., 1996: Global climate and infectious disease: The cholera paradigm.
Science, 274, 2025-2031.
Colwell, R.R., 2004: Infectious disease and environment: cholera as a paradigm for
waterborne disease. International Microbiology, 7, 285-289.
Colwell, R.R. and W.M. Spira, 1992: The ecology of Vibrio cholerae. In Cholera, D. Barua
and W.B. Greenough III, Eds., Plenum Medical, pp. 107-127.
Consejo Nacional de Ambiente (CONAM), 1999: Peru: Vulnerabilidad Frente Al Cambio
Climático: Aproximaciones a la experiencia con el fenómeno El Niño. CONAM.
Constantin de Magny, G. and R.R. Colwell, 2009. Cholera and climate: a demonstrated
relationship. Transactions of the American Clinical and Climatological Association, 120,
119-128.
Constantin de Magny, G., B. Cazelles, and J.F. Guegan, 2007: Cholera threat to humans
in Ghana is influenced by both global and regional climate variability. EcoHealth 3, 223231.
Constantin de Magny, G., J.F. Guegan, M. Petit, and B. Cazelles, 2007: Regional-scale
climate-variability synchrony of cholera epidemics in West Africa. BMC Infectious
Diseases, 7. [Available online at http://www.biomedcentral.com/1471-2334/7/20]
Constantin de Magny, G., and Coauthors, 2008: Environmental signatures associated
with cholera epidemics. Proceedings of the National Academy of Sciences of the United
States of America, 105, 17676-17681.

273

Constantin de Magny, G., and Coauthors, 2011: Role of Zooplankton Diversity in Vibrio
cholerae Population Dynamics and in the Incidence of Cholera in the Bangladesh
Sundarbans. Applied and Environmental Microbiology, 77, 6125-6132.
Cordova Aguilar, H., 1996: Piura. Geospacios 9. Universidad de la Serena, Peru. (in
Spanish)
Crocker, D., 2008: The Ethics of Global Development. Cambridge University Press.
Cueto, M., 2001: The Return of Epidemics: Health and Society in Peru during the
Twentieth Century. Ashgate, pp. 107-137.
Cutter, S.L., 2003: The vulnerability of science and the science of vulnerability. Annals of
the Association of American Geographers, 93, 1-12.
Cutter, S.L. and C. Finch, 2008: Temporal and spatial changes in social vulnerability to
natural hazards. Proceedings of the National Academy of Sciences of the United States of
America, 105, 2301-2306.
Cutter, S.L., B.J. Boruff, and W.L. Shirley, 2003: Social vulnerability to environmental
hazards. Social Science Quarterly, 84, 242-261.
Davis, M., 2001: Late Victorian Holocausts: El Niño Famines and the Making of the Third
World. Verso.
de Blij, H., 2009: The Power of Place: Geography, Destiny, and Globalization's Rough
Landscape. Oxford University Press.
del Aguila, R., B. Benavides, E. Jacoby, and J. Novara, 1992: Reconocimiento de cólera
por personas sintomaticas después del brote epidémico en la Udes Lima y la Subregion
Luciano Castilla-Region Grau (Identification of persons with symptomatic cholera after
outbreaks in Lima and the Subregion Luciano Castilla-Region Grau). Revista Peruana de
Epidemiologia, 5, 5-9. (in Spanish)
Deen, J.L. and Coauthors, 2008: The high burden of cholera in children: comparison of
incidence from endemic areas in Asia and Africa. Public Library of Science Neglected
Tropical Diseases, 2, e173. [Available online at
http://www.plosntds.org/article/info:doi/10.1371/journal.pntd.0000173]
Department for International Development Health Systems Resource Center, UK (DFID),
1999: Peru: country health briefing paper.
Diaz, H.F., M.P. Hoerling, and J.K. Eishield, 2001: ENSO variability, teleconnections and
climate change. Int. J. Climatol., 21, 1845-1862.
Eakin, H. and A. Lynd Luers, 2006: Assessing the vulnerability of social-environmental
systems. Annual Review of Environmental Resources, 31, 365-394.

274

Eguiguren, D.V., 1894: Las lluvias en Piura (Rains in Piura). Boletín de Sociedad
Geografico de Lima, 4, 241-258. (in Spanish)
El Tiempo, 1991a: Paralización total hubo en la Region por falta de luz (Without
electricity the region [Grau – Piura and Tumbes] was paralyzed). 1 February, Piura. (in
Spanish)
El Tiempo, 1991b: Se agudiza deficiencia energética y de agua (Energy and water
deficiencies worsen). 28 January, Piura. (in Spanish)
Emch, M., M. Yunus, V. Escamilla, C. Feldacker, and M. Ali, 2010: Local population and
regional environmental drivers of cholera in Bangladesh. Epidemiology, 9. [Available
online at http://www.ehjournal.net/content/9/1/2]
Emch, M., C. Feldacker, M. Yunus, P.K. Streafield, V.D. Thiem, D.G. Canh, and M. Ali,
2008: Local environmental predictors of cholera in Bangladesh and Vietnam. American
Journal of Tropical Medicine and Hygiene, 78, 823-832.
Emerson, R.F., R.I. Fretz, and L.L. Shaw, 1995: Writing Ethnographic Fieldnotes.
University of Chicago Press.
Epstein, P.R., 1992: Cholera and the environment. The Lancet, 339, 1167-1168.
--------1993: Algal blooms in the spread and persistence of cholera. Biosystems, 31, 209221.
Epstein, P. R., T.E. Ford, and R.R. Colwell, 1993: Health and climate change: Marine
ecosystems. The Lancet, 342, 1216-19.
Escribano, R. and Coauthors, 2004: Biological and chemical consequences of the 19971998 El Niño in the Chilean coastal upwelling system: a synthesis. Deep-Sea Research,
11, 2389-2411.
Esquith, S. and F. Gifford, (Eds.), 2010: Capabilities, Power, and Institutions: Toward a
More Critical Development Ethics. The Pennsylvania State University Press.
ESRI, 2010: ArcGIS. October 2010. [Available at http://www.esri.com/]
Farmer, P., 2001: Infections and Inequalities: The Modern Plagues. University of
California.
Fitzgerald, H.E., P. Smith, P. Book, K. Rodin, and the CIC Committee on Engagement,
2005: Draft CIC Report: Resource guide and recommendations for defining and
benchmarking engagement. CIC. (Available at
http://apa.wisc.edu/communityengagement/CIC_EngagementReportREV2-22-05.pdf)
Flohn, H. and H. Fleer, 1975: Climate teleconnections with the equatorial Pacific and the
role of ocean/atmosphere coupling. Atmosphere, 13, 96-109.
275

Flores, M., 1998: Fisheries. In Assessment of the use of remote sensing and other
information related to ENSO: the use of ENSO information in Peru, M.H. Glantz, Ed.,
NASA/NCAR/Peru Project Final Report. Boulder, CO. Environmental and Societal Impacts
Group, National Center for Atmospheric Research, pp. 37-42.
Fotheringham, A.S., C. Brunsdon, and M. Charlton, 2002: Geographically Weighted
Regression: The Analysis of Spatially Varying Relationships. Wiley and Sons.
Franco, A.A., and Coauthors, 1997: Cholera in Lima, Peru, correlates with prior isolation
of Vibrio Cholerae from the Environment. American Journal of Epidemiology, 146, 10671075.
Galvão, L.A.C., S. Edwards, C. Corvalan, C., K. Fortune, and M. Akerman, 2009: Climate
change and social determinants of health: two interlinked agendas. Global Health
Promotion, Supplement (1), 81–84.
Gangarosa, E.J. and R.V. Tauxe, 1992: Epilogue: the Latin American cholera epidemic. In
Cholera, D. Barua and W.B. Greenough III, Eds., Plenum Medical, pp. 351-357.
Gardiner, S., S. Caney, D. Jamieson, and H. Shue, Eds., 2010: Climate Ethics: Essential
Readings. Oxford University Press.
Garreaud, R.D., M. Vuille, R. Compagnucci, and J. Marengo, 2008: Present-day South
American climate. Palaeogeography, Paleoclimatology, Paleoecology, 281, 180-195.
Gasper, D., 1996: Needs and basic needs – a clarification of foundational concepts for
development ethics and policy. In Questioning Development, G. Koehler et al., Eds.,
Metropolis, pp. 71-101.
--------2005: The Ethics of Development. Edinburgh University Press.
Gil, A.I., July 2009: Personal communication.
Gil, A.I. and Coauthors, 2004: Occurrence and distribution of Vibrio cholerae in the
coastal environment of Peru. Environmental Microbiology, 6, 699-706.
Glantz, M.H., 1981: Considerations of the societal value of an El Niño forecast and the
1972-73 El Niño. In Resource Management and Environmental Uncertainty: Lessons from
Coastal Upwelling Fisheries, M.H. Glantz and J.D. Thompson, Eds., John Wiley, pp. 449476.
--------1998: Assessment of the use of remote sensing and other information related to
ENSO: the use of ENSO information in Peru, NASA/NCAR/Peru Project Final Report.
Boulder, CO. Environmental and Societal Impacts Group, National Center for
Atmospheric Research, pp. 37-42.
--------2000: Why care about El Niño and La Niña? In Satellites, Océanography and
Society, D. Halpern, Ed., Elsevier Science B.V., pp.149-169.
276

--------2001b: Once Burned, Twice Shy: Lessons Learned from the 1997-98 El Niño. United
Nations University Press, pp. 1-10.
--------2001a: Currents of Change: Impacts of El Niño and La Niña on Climate and Society.
Cambridge University Press.
--------2002b: La Niña: an overview of the process. In La Niña and Its Impacts: Facts and
Speculation. M.H. Glantz, Ed. United Nations University Press, pp. 3-24.
--------2002a: Preface. In La Niña and Its Impacts: Facts and Speculation. M.H. Glantz, Ed.
United Nations University Press, pp. xiii-xv.
--------2003: Climate Affairs: A Primer. Island Press.
--------2005: Hurricane Katrina as a ‘teachable moment’. Advances in Geosciences, 14,
287-294. [Available online at http://www.adv-geosci.net/14/287/2008/adgeo-14-2872008.html]
--------July 2008: Personal Communication.
--------July 2009: Personal Communication.
--------August 2009: Personal Communication.
Glantz, M.H. and Z. Adeel, 2000: Climate Affairs as a next-generation environmental
science. Global Environmental Change, 10, 81-85.
Glass, R.I. and R.E. Black, 1992: The epidemiology of cholera. In Cholera, Barua, D., W.B.
and Greenough III, Eds. Plenum Medical.
Glass, R.I., M. Claeson, P.A. Blake, R.J. Waldman, and N.F. Pierce, 1991: Cholera in Africa:
Lessons on transmission and control for Latin America. The Lancet, 338, 791-795.
Goddard, L. and N.E. Graham, 1997: El Niño in the 1990s. J. Geophys. Res., 102, 1042310436.
Goddard, L., S.J. Mason, S.E. Zebiak, C.F. Ropelewski, R. Basher, and M. Cane, 2001:
Current approaches to seasonal-to-annual climate predictions. International Journal of
Climatology, 21, 1111-1152.
Gommes, R., J. du Guerney, M.H. Glantz, and L. Hsu, 2004: Climate and HIV/AIDS : a
hotspots analysis for early warning rapid response systems. A joint publication of the
United Nations Development Programme, Food and Agricultural Organization of the
United Nations and the National Center for Atmospheric Research. April 2005 [Available
at http://www.hivdevelopment.org/Publications_english/Climate%20and%20HIV.htm].
Govender, T., J.M. Barnes, and C.H. Pieper, 2011: Contribution of water pollution from
inadequate sanitation and housing quality to diarrheal disease in low-cost housing
277

settlements of Cape Town, South Africa. American Journal of Public Health, 101, e4-e9
[Available online at
http://ajph.aphapublications.org/doi/full/10.2105/AJPH.2010.300107]
Graham, N.E. and W.B. White, 1988: The El Niño cycle: a natural oscillator of the Pacific
Ocean-Atmosphere system. Science, 240, 1293-1302.
Grimm, A.M. and R.G. Tedeschi, 2009: ENSO and Extreme Rainfall Events in South
America. J. Climate, 22, 1589-1609.
Grinsted, A., J.C. Moore, and S. Jevrejeva, 2004: Application of the cross wavelet
transform and wavelet coherence to geophysical time series. Nonlinear Processes in
Geophysics, 11, 561-566.
--------2008. Matlab package for performing cross-wavelet and wavelet coherence.
[Available online at http://www.pol.ac.uk/home/research/waveletcoherence/]
Gueri, A., 1984: Lessons learned: health effects of El Niño in Peru. Disasters:
Preparedness and Mitigation, 19. PAHO. [Available online at
http://helid.digicollection.org/en/d/Jdi019e/2.html]
Halpert, M.S. and G.D. Bell, 1997: Climate assessment for 1996. Bull. Amer. Meteor. Soc.,
78, S1-49.
Hahn, M., A.M. Riederer, and S.O. Foster, 2009: The Livelihood Vulnerability Index: a
pragmatic approach to assessing risk of climate variability and change—a case study in
Mozombique. Global Environmental Change, 19, 74-88.
Hashizume M., B. Armstrong, S. Hajat, Y. Wagatsuma, A.S. Faruque, T. Hayashi, and D.
Sack, 2008: The effect of rainfall on the incidence of cholera in Bangladesh.
Epidemiology, 19, 103–110.
Hashizume, M, A.S. Faruque, Y. Wagatsuma, T. Hayashi, and B. Armstrong, 2010:
Cholera in Bangladesh: Climatic Components of Seasonal Variation. Epidemiology, 21,
706-710.
Hernan Vargas, F., S. Harrison, S. Rea, and D.W. Macdonald, 2006: Biological effects of El
Niño on the Galapagos penguins. Biological Conservation, 127, 107-114.
Hewitt, K.,1997: Regions of Risk: A Geographical Introduction to Disasters. Longman.
Horel, J.D. and A.G. Cornejo-Garrido, 1986: Convection along the Coast of Northern Peru
during 1983: Spatial and Temporal Variation of Clouds and Rainfall. Mon. Wea. Rev.,
114, 2091-2105
Huanca, R.I., 2004: El fenómeno El Niño y su influencia en el desarollo de algunos brotes
epidémicos: Cono norte-Lima (The El Niño phenomenon and its influence on the
emergence of epidemics: North Lima). Revista de Viernes Medico, 29, 38-43. (in Spanish)
278

Huq, A. and Coauthors, 2005: Critical factors influencing the occurrence of vibrio
cholerae in the environment of Bangladesh. Applied and Environmental Microbiology,
71, 4645-4654.
Huq A., R.R. Colwell, R. Rahman, A. Ali, M. Chowdhury, S. Parveen, D.A. Sack, and E.
Russek-Cohen, 1990: Detection of V. cholerae 01 in the aquatic environment by
fluorescent monoclonal antibody and culture method. Applied and Environmental
Microbiology, 56, 2370-2373.
Huq, A., R.B. Sack, and R.R. Colwell, 2001: Cholera and global ecosystems. In Ecosystem
Change and Public Health: A Global Perspective, J.L. Aron and J. Patz, Eds. Johns Hopkins
University, pp. 327-352.
Instituto Nacional de Salud (National Institute of Health), 2006: Fortalecimiento de la
vigilancia de Laboratorio: Vibrio cholerae, Fenómeno El Niño Peru 2006-2007
(Strengthening laboratory surveillance: Vibrio cholerae, the El Niño phenomenon Peru).
Boletín del Instituto Nacional de Salud, 50. (in Spanish)
Instituto Nacional de Estadistica e Informatica (Institute of National Statistics and
Information) (INEI), Peru, 1994: Piura: Compendio Estadistico 1993-1994 (Piura:
Summary of Statistics 1993-1994), INEI. (In Spanish
--------1996: Encuesta Demográfica y de Salud Familiar 1996 (Study of Family
Demography and Health 1996). [Available online at http://desa.inei.gob.pe/endes/] (in
Spanish)
--------2000: Encuesta Demográfica y de Salud Familiar 2000 (Study of Family
Demography and Health 2000). [Available online at http://desa.inei.gob.pe/endes/] (in
Spanish)
--------2001: Encuesta Demográfica y de Salud Familiar 2000 (Study of Family
Demography and Health 2000).
--------2006: Necesidades Basicas Insatisfechas (Basic Needs Unmet). [Available online at
http://www.inei.gob.pe/biblioineipub/bancopub/Est/Lib0068/n00.htm] (in Spanish)
International Research Institute for Climate and Society (IRI), 2007: The ENSO system.
[Available online at http://iri.columbia.edu/climate/ENSO/enso.html]
Jacobsen, K., 2008: Introduction to Public Health. Jones and Bartlett.
Jamieson, D., 2010: Ethics, Public Policy, and Global Warming. In Climate Ethics:
Essential Readings, S. Gardiner, S. Caney, D. Jamieson, and H. Shue, Eds., Oxford
University Press, pp. 77-86.
Jamieson, D. and M.H. Glantz, 2000: Societal response to Hurricane Mitch and intraversus intergenerational equity issues: whose norms should apply? Risk Analysis, 6, 869882.
279

Jimenez, C. and J. Cortes, 2001: Effects of the 1991-92 El Niño on scleractinian corals of
the Costa Rican central Pacific coast. Revista de Biologia Tropical, 49, 239-250.
Johansson M.A., D. Cummings, and G.E. Glass, 2009: Multiyear climate variability and
dengue—El Niño Southern Oscillation, weather, and dengue incidence in Puerto Rico,
Mexico, and Thailand: a longitudinal data analysis. Public Library of Science, 6,
e1000168. [Available online at
http://www.plosmedicine.org/article/info%3Adoi%2F10.1371%2Fjournal.pmed.100016
8]
Joralemon, D., 1999: Exploring Medical Anthropology. Pearson.
Kates, R. W., 1971: Natural hazard in human ecological perspective: hypotheses and
models. Economic Geography, 47, 438-451.
Kearns, R. and G. Moon, 2002: From medical to health geography: novelty, place and
theory after a decade of change. Progress in Human Geography, 26, 605-625.
Kessler, W.S. and M.J. McPhaden, 1995: Oceanic equatorial waves and the 1991-93 El
Niño. J. Climate, 8, 1757-1774.
Kiladis, G.N. and H.F. Diaz, 1989: Global climate anomalies associated with extremes in
the Southern Oscillation. J. of Climate, 2, 1069-90.
Klauer, A., 2000: El Niño-La Niña: El fenómeno océano-atmosférico del Pacifico Sur, un
reto para la ciencia y la historia (El Niño-La Niña: the ocean-atmosphere phenomenon in
the South Pacific, a challenge for science and the history). [Available online at
http://www.wilmercarbajal.com/documentos/El_Niño.pdf] (in Spanish)
Koelle K. and M. Pascual, 2004: Disentangling extrinsic from intrinsic factors in disease
dynamics: a nonlinear time series approach with an application to cholera. American
Naturalist, 163, 901–913.
Koelle K., X. Rodo, M. Pascual, M. Yunus, and G. Mostafa, 2005: Refractory periods and
climate forcing in cholera dynamics. Nature, 436, 696–700.
Koelle, K., M. Pascual, and M. Yunus, 2006: Serotype cycles in cholera dynamics.
Proceedings of the Royal Society, Series B, Biological Sciences, 273, 2879-2886.
Koo, D. and Coauthors, 1996: Epidemic cholera in Guatemala, 1993: transmission of a
newly introduced epidemic strain by street vendors. Epidemiology and Infection, 116,
121-126.
Kousky, V.E., M.T. Kagano and I.F.A. Cavalcanti, 1984: A review of the Southern
Oscillation: oceanic-atmospheric circulation changes and related rainfall anomalies.
Tellus, 36a, 490-504.

280

Kovats, R.S., M.J. Bouma, S. Hajat, E. Worrall, and A. Haines, 2003: El Niño and health.
The Lancet, 362, 1481 – 1489.
Lagos, P., 1998: Use of the information and prediction of El Niño in the Peruvian
agrarian sector. In Assessment of the use of remote sensing and other information
related to ENSO: the use of ENSO information in Peru, M.H. Glantz, Ed.,
NASA/NCAR/Peru Project Final Report. Boulder, CO. Environmental and Societal Impacts
Group, National Center for Atmospheric Research, pp. 29-35.
Lagos, P. and J. Buizer, 1992: El Niño and Peru: a nation’s response to interannual
climate variability. In Natural and Technological Disasters: Causes, Effects and
Preventive Measures, Majumdar et al., Eds., The Pennsylvania Academy of Science, pp.
223-238.
Lagos, P., Y. Silva, E. Nickl, and K. Mosquera, 2008: El Niño-related precipitation
variability in Peru. Advances in Geosciences, 14, 231-237. [Available online at
http://www.adv-geosci.net/14/231/2008/]
Lama, J.R., C.R. Seas, R. Leon-Barua, E. Gotuzzo, and R.B. Sack, 2004: Environmental
temperature, cholera, and acute diarrhea in adults in Lima, Peru. Journal of Health and
Population Nutrition, 22, 399-403.
Lawrimore, J.H. and Coauthors, 2001: Climate assessment for 2000. Bull. Amer. Meteor.
Soc., 82, S1-55.
Lewis, J., 1999: Development in Disaster-prone Places: Studies of Vulnerability.
Intermediate Technology Publications.
Lipp, E.K., A. Huq, and R.R. Colwell, 2002: Effects of global climate on infectious disease:
the cholera model. Clinical Microbiology Reviews, 15, 757 – 770.
Lipp, E.K., and Coauthors, 2003: Direct detection of Vibrio cholerae and ctxA in Peruvian
coastal water and plankton by PCR. Applied and Environmental Microbiology, 69, 36763680.
Liverman, D.M. and R. Merideth, 2002: Climate and society in the US Southwest: The
context for a regional assessment. Climate Research, 21, 199-218.
Lobitz, B., L. Beck, A. Huq, B. Wood, G. Fuchs, A.S.G. Faruque, and R.R. Colwell, 2000:
Climate and infectious disease: use of remote sensing for detection of V. cholerae by
indirect measurement. Proceedings of the National Academy of Sciences of the United
States of America, 97, 1438-1443.
Macintyre, S., A. Ellaway and S. Cummins, 2002: Place effects on health: how can we
conceptualise, operationalise and measure them? Social Science & Medicine, 55, 125139.

281

Marchant, M., D. Hebbeln, and G. Wefer, 1998: Seasonal flux patterns of planktic
foraminifera in the Peru-Chile Current. Deep-Sea Research I, 45, 1161-1185.
Marengo, J.A. and Coauthors, 2008: The drought of Amazonia in 2005. J. Climate, 21,
495-516.
Marengo, J.A., 1999: Estudios sobre tendencias y variabilidad climatica en el tropico
sudamericano: Peru y Brasil (A study of climate influences and variability in the South
American tropics: Peru and Brazil). In Peru: Vulnerabilidad Frente Al Cambio Climático:
Aproximaciones a la experiencia con el fenómeno El Niño (Peru: Vulnerability to Climate
Change: The experience of the El Niño Phenomenon as an Analogue), B. Marticorena,
Ed., Consejo Nacional de Ambiente, pp. 185-204. (in Spanish)
Marmot, M., 2005: Social determinants of health inequalities, The Lancet 365, 10991104.
Mathworks (Matlab), 2010: Matlab Student Version R2009a. [Available online at
http://www.mathworks.com/]
Mayer, J.D., 2000: Geography, ecology, and emerging infectious diseases. Social Science
& Medicine, 50, 937-952.
--------2005: The geographical understanding of HIV/AIDS in Sub-Saharan Africa.
Norwegian Journal of Geography, 59, 6-13.
McCarthy S.A., and F.M. Khambaty, 1994: International dissemination of epidemic Vibrio
cholerae by cargo ship ballast and other nonpotable waters. Applied and Environmental
Microbiology, 60, 2597-2601.
McPhaden, M., 1994: TOGA-TAO and the 1991-93 El Niño-Southern Oscillation Event.
National Oceanic and Atmospheric Administration. [Available online at
http://www.pmel.noaa.gov/pubs/docs/mcph1401.html]
McPhaden, M.J., S.E. Zebiak, and M.H. Glantz, 2006: ENSO as an integrating concept in
earth science. Science, 314, 1740-1745.
Meade, M.S. and R.J. Earickson, 2005: Medical Geography, 2nd ed., Guilford Press, pp.
218.
Mendelsohn, J. and T. Dawson, 2008: Climate and cholera in KwaZulu-Natal, South
Africa: the role of environmental factors and implications for epidemic preparedness.
International Journal of Hygiene and Environmental Health, 211, 156-162.
Ministerio de Salud (Ministry of Health), Piura (MINSA Piura), 2005: Enfermedades
emergentes-reemergentes y modificaciones en el perfil epidemiologico del noroeste
Peruano post Fenómeno El Niño (Emergent and reemergent diseases and interventions
in northwest Peru in a post El Niño year), 2005. Powerpoint Presentation. (in Spanish)
282

--------2007: Weekly cholera cases in the Subregion of Piura, Peru for 1991 to 2006.
Ministerio de Salud (Ministry of Health)(MINSA), Lima, Peru, 1994b: Recopilación y
sistematizacion de las condiciones ambientales y actividades desarrolladas en el control
de la epidemia del cólera (A summary of environmental conditions and activities to
control the cholera epidemic). MINSA. (in Spanish)
--------1994a: Reporte Epidemiológico (Epidemiology Report), Week 26. (in Spanish)
--------1995b: Reporte Epidemiológico(Epidemiology Report), Week 39. (in Spanish)
--------1995a: Reporte Epidemiológico (Epidemiology Report), Week 52 . (in Spanish)
--------1997: Reporte Epidemiológico (Epidemiology Report), Week 38. (in Spanish)
--------1998f: Reporte Epidemiológico (Epidemiology Report), Week 8. (in Spanish)
--------1998e: Reporte Epidemiológico (Epidemiology Report), Week 17. (in Spanish)
--------1998d: Reporte Epidemiológico (Epidemiology Report), Week 21. (in Spanish)
--------1998c: Reporte Epidemiológico (Epidemiology Report), Week 26. (in Spanish)
--------1998b: Reporte Epidemiológico (Epidemiology Report), Week 32. (in Spanish)
--------1998a: Reporte Epidemiológico (Epidemiology Report), Week 52. (in Spanish)
--------1999b: Reporte Epidemiológico (Epidemiology Report), Week 5. (in Spanish)
--------1999a: Las últimas consecuencias del fenómeno “La Niña” en el Peru (The
phenomenon “La Niña” and recent impacts in Peru). Reporte Epidemiológico
(Epidemiology Report), Week 7. (in Spanish)
--------2000: Reporte Epidemiológico (Epidemiology Report), Week 10. (in Spanish)
--------2001: Los desastres en el Peru (Disasters in Peru). Reporte Epidemiológico
(Epidemiology Report), Week 9, p. 1. (in Spanish)
--------2005b: Weekly cholera cases dataset for 1991 to 2002 in Peru. The Department of
Epidemiology, MINSA in Lima.
--------2005a: Reporte Epidemiológico (Epidemiology Report), Week 1. (in Spanish)
Mourino-Perez, R.R, 1998: Océanography and the seventh cholera pandemic.
Epidemiology, 9, 355-357.
Nash, N.C., 1991: Fujimori in the time of cholera. NY Times, 24 February. [Available
online at http://www.nytimes.com/1991/02/24/weekinreview/the-world-fujimori-inthe-time-of-cholera.html]

283

National Oceanic and Atmospheric Administration (NOAA), 1992: ENSO Advisory, May.
Climate Analysis Center.
--------1993: ENSO Advisory, June. Climate Analysis Center.
--------1996: Climate Conditions during the 1994-95 and 1995-96 Winter Season. Special
Climate Summary-96/1, March 1996. Climate Prediction Center. [Available online at
http://www.cpc.noaa.gov/products/special_summaries/96_1/]
--------2000: The 1998–2000 Pacific Cold Episode (La Niña). Climate Prediction Center.
[Available online at
http://www.cpc.noaa.gov/products/assessments/assess_99/enso.html]
--------2005: Frequently asked questions about El Niño and La Niña. Climate Prediction
Center. [Available online at
http://www.cpc.noaa.gov/products/analysis_monitoring/ensostuff/ensofaq.shtml]
--------2009: Monthly atmospheric and SST indices. [Available online at
http://www.cpc.ncep.noaa.gov/data/indices/]
--------2010: Cold and warm episodes by season. [Available online at
http://www.cpc.ncep.noaa.gov/products/analysis_monitoring/ensostuff/ensoyears.sht
ml]
Nelson, E.J., J.B. Harris, J.G. Morris Jr., S.B. Calderwood, and A. Camilli, 2009: Cholera
transmission: the host, pathogen and bacteriophage dynamic. Nature, 7, 693-702.
Ndayimirije, N. and Coauthors, 1993: Epidemic cholera – Burundi and Zimbabwe, 19921993. Morbidity Mortality Weekly Report, June 4, 42.
Nussbaum, M.C., 2000: Women and Human Development: The Capabilities Approach.
Cambridge University Press.
--------2011: Creating Capabilities: The Human Development Approach. Harvard
University Press.
Oppong, J.R. 1998. A vulnerability of the geography of HIV/AIDS in Ghana, 1986-1995.
The Professional Geographer, 50, 437-448.
Ordinola, N., 2002: The consequences of cold events for Peru. In La Niña and Its
Impacts: Facts and Speculation, M.H. Glantz, Ed., United Nations University Press, pp.
146-150.
--------June 2009: Personal Communication.
Orlove, B.S., J.C.H. Chiang and M.A. Cane, 2000: Forecasting Andean rainfall and crop
yield from the influence of El Niño on Pleiades visibility. Nature, 403, 68-71.

284

Pan American Health Organization (PAHO), 1991b: Cholera situation in the Americas.
Epidemiological Bulletin, 12, 1-24.
--------1991a. Cholera situation in the Americas: An update. 1991. Epidemiological
Bulletin, 2, 1-16.
--------1998b: Peru: Fenómeno “El Niño”. Informe estratégico #2, OPS-Peru, Week
10.[Available online at [Available online at
http://www.cepis.org.pe/eswww/elNiño/inf02.html] (in Spanish)
--------1998a: El Niño and its impact on health. 122 nd Meeting, Washington, D.C. June
1998, Provisional agenda item 4.4.
[Available online at http://www.bvsde.paho.org/bvsacd/cd68/ElNiño.pdf]
--------2000: Peru. In Crónicas de Desastres: Fenómeno El Niño, 1997-1998 (Disaster
Chronicles: The 1997-1998 El Niño Phenomenon). Chapter 11. [Available online at
http://www.paho.org/spanish/ped/ElNiño.htm] (in Spanish)
--------2008: The number of cholera cases in the Americas, 1990-2008. [Available online
at http://www.paho.org/English/AD/DPC/CD/cholera-1990-2008.pdf]
Pascual, M., X. Rodo, S.P. Ellner, R.R. Colwell, and M.J. Bouma, 2000: Cholera Dynamics
and El Niño-Southern Oscillation. Science, 289, 1766-1769.
Pascual M, M.J. Bouma, and A.P. Dobson, 2002: Cholera and climate: revisiting the
quantitative evidence. Microbes and Infection, 4, 237–245.
Philander, S.G.H., 1985: El Niño and La Niña. J. Atmos. Sci., 42, 2652-2662.
Philander, S.G., 2006: Our Affair with El Niño: How We Transformed an Enchanting
Peruvian Current into a Global Climate Hazard. Princeton University Press.
Pizarro, L., 1999: El Niño phenomenon and natural disasters in Peru. Internacional
Workshop on Natural Disaster Management, 10-12 June, 1999, Beijing, China. [Available
online at www.crid.or.cr/digitalizacion/pdf/eng/doc15155/doc15155-contenido.pdf]
Pyle, G.F., 1979: Applied Medical Geography. V.H. Winston & Sons.
Ramírez, N.C., I.J. Ramírez, and N. Ordinola, 2009: El Niño, capacidades existentes en
Piura, Peru (El Niño capacities in Piura, Peru). El Niño Affairs webpage, Consortium for
Capacity Building, Boulder, CO. [Available online at
http://ccb.colorado.edu/enos/editorials/sept09.html] (in Spanish)
Reid, P. and C. Vogel, 2006: Living and responding to multiple stressors in South Africa-Glimpses from KwaZulu-Natal. Global Environmental Change, 16, 195-206.

285

Reyes, R.R., 2002: Gendering responses to El Niño in rural Peru. Gender and
Development, 10, 60-69.
Ries, A.A., and Coauthors, 1992: Cholera in Piura, Peru: A modern urban epidemic. The
Journal of Infectious Diseases, 166, 1429-1433.
Rodó, X., M. Pascual, G. Fuchs, and S. Faruque, 2002: ENSO and cholera: a nonstationary
link related to climate change? Proceedings of the National Academy of Sciences, 99,
12901-12906.
Rodriguez, R., A. Macabres, B. Luckman, M. Evans, M. Masiokas, and T.M. Ektvedt, 2005:
“El Niño” events recorded in dry-forest species of the lowlands of northwest Peru.
Dendrochronologia, 22, 181-186.
Ropelewski, C. and M. Halpert, 1987: Global and regional scale precipitation patterns
associated with the El Niño/Southern Oscillation. Mon. Wea. Rev., 115, 1606-1626.
Ropelewski, C. and M. Halpert, 1992: Surface temperature patterns associated with the
Southern Oscillation. J. Climate, 5, 577-593.
Roundy, R., 1979: Human behaviour and disease hazards: spatial perspectives on rural
health. GeoJournal, 3, 579-586.
Ruiz-Moreno, D., M. Pascual, M. Bouma, A. Dobson, and B. Cash, 2007: Cholera
seasonality in Madras (1901-1940): dual role for rainfall in endemic and epidemic
regions. EcoHealth, 4, 52-62.
Rummel, J., 1967: Understanding factor analysis. Conflict Resolution, 11, 444-480.
[Available online at http://intersci.ss.uci.edu/wiki/pw/Rummel1967FactorAnal.pdf]
Salazar-Lindo, E., C. Seas, and D. Gutiérrez, 2008: ENSO and cholera in South America:
what can we learn about it from the 1991 cholera outbreak? International Journal of
Environment and Health, 2, 30-36.
Salazar-Lindo, E. P. Pinell-Salles, A. Maruy, and E. Chea-Woo, 1997: El Niño and
diarrheoa and dehydration in Lima, Peru. The Lancet, 350, 1597-1598.
Salazar-Lindo, E. M. Alegre-chang, M. Rodriguez, P. Carrion, and N. Razzeto, 1993: The
Peruvian cholera epidemic and the role of chlorination in its control and prevention. In
Safety of Water Disinfection: Balancing Chemical and Microbial Risks, G.F. Craun, Ed. ILSI
Press.
Sandoval, P.S., 1999: Evaluación de Daños y Acciones del Fenómeno El Niño (Evaluation
of damages caused by the El Niño phenomenon and actions taken). Oficina de
Planificación, Dirección de Salud Regional (Planning Office, Regional Health Post), Piura,
Peru. (in Spanish)

286

Santos, J.L., 2008: The impact of El Niño-Southern Oscillation events on South America.
Advances in Geosciences, 6, 221-225.
Seas, C., J. Miranda, A.I. Gil, R. Leon-Barua, J. Patz, A. Huq, R.R. Colwell, and R.B. Sack,
2000: New insights on the emergence of cholera in Latin America during 1991: the
Peruvian experience. The American Journal of Tropical Medicine and Hygiene, 62, 513 –
517.
Sen, A., 1982: Poverty and Famines: An Essay on Entitlements and Deprivation. Oxford
University Press.
--------1999: Development as Freedom. Oxford University Press.
--------2010: The Idea of Justice. Belknap Press.
Shah, S., 2011: Climate’s Strong Fingerprint in Global Cholera Outbreaks. Environment
360, Yale University. [Available online at
http://e360.yale.edu/feature/climates_strong_fingerprint_in_global_cholera_outbreaks
/2371/]
Singleton, F.L., R.W. Attwell, M.S. Jangi, and R.R. Colwell, 1982: Influence of salinity and
organic nutrient concentration on survival and growth of Vibrio cholerae in aquatic
microcosms. Applied and Environmental Microbiology, 43, 1080-1085.
Singer, M., 1994: AIDS and the health crisis of the U.S. urban poor; the perspective of
the critical anthropology. Social Science and Medicine, 39, 931-948.
Servicio Nacional de Meteorologia e Hidrologia (SENAMHI), Lambayeque, 2004: El
evento El Niño Oscilacion Sur 1997-1998: su impacto en el departamento de
Lambayeque (The 1997/98 ENSO event: impacts in the Department of Lambayeque).
Chiclayo, Peru. [Available online at http://www.senamhi.gob.pe/?p=0702] (in Spanish)
--------2008: Impactos Hidrologicos de los eventos El Niño-La Niña: cuencas de los rios
Topara, San Juan, Pisco, Ica y Grande (Impacts from El Niño-La Niña on hydrology:
Topara, San Juan, Pisco, Ica, and Grande River Basins). Lima, Peru. (in Spanish)
--------2009: Escanarios de cambio climatico en la cuenca del Rio Urubamba para el ano
2100: resumen técnico (Climate change scenarios for the Urubamba River Basin in 2100:
a technical summary). Lima, Peru. [Available online at
www.senamhi.gob.pe/main_down.php?ub=cmn&id...URUBAMBA] (in Spanish)
Speelmon, E.C., W. Checkley, R.H. Gilman, J. Patz, M. Calderon, and S. Manga, 2000:
Cholera incidence and El Niño-related higher ambient temperature. Journal of the
American Medical Association, 283, 3072-3074.
Stock, R.F, 1976: Cholera in Africa. International African Institute.

287

Swerdlow, D.L., and Coauthors, 1992: Waterborne transmission of epidemic cholera in
Trujillo, Peru: lessons for a continent at risk. The Lancet, 340, 28-33.
Tamplin, M.L., A.L. Gauzens, A. Huq, D.A. Sack, and R.R. Colwell, 1990: Attachment of
Vibrio cholera serogroup 01 to zooplankton and phytoplankton of Bangladesh waters.
Applied and Environmental Microbiology, 56, 1977-1980.
Tarazona, J. and S. Valle, 1999: Impactos potenciales del cambio climatico global sobre
el ecosistema marino Peruano (Potencial impacts from global climate change on the
Peruvian marine ecosystem). In Peru: Vulnerabilidad Frente Al Cambio Climático:
Aproximaciones a la experiencia con el fenómeno El Niño (Peru: Vulnerability to Climate
Change: The experience of the El Niño Phenomenon as an Analogue), B. Marticorena,
Ed., Consejo Nacional de Ambiente, pp. 95-114. (in Spanish)
Tauxe, R., E.D. Mintz, and R.E. Quick, 1995: Epidemic cholera in the New World:
translating field epidemiology into new strategies. Emerging Infectious Diseases, 1, 4.
[Available online at http://www.cdc.gov/ncidod/EID/vol1no4/tauxe.htm]
The Long Paddock, 2010: The Long Paddock website, Queensland Government.
[Available online at http://www.longpaddock.qld.gov.au/]
Thisted, R.A., 2003: Are there social determinants of health and disease? Perspectives in
Biology and Medicine, 46, S65-73.
Tickner, J. and T. Gouveia-Vigeant, 2005: The 1991 cholera epidemic in Peru: Not a case
of precaution gone awry. Risk Analysis, 25, 495-502.
Torrence, C. and G.P., Compo, 1998: A practical guide to wavelet analysis. Bulletin of the
American Meteorological Society, 79, 61-78.
Torrence, C. and P.J. Webster, 1999: Interdecadal changes in the ENSO-Monsoon
System. Journal of Climate, 12, 2679-2690.
Trenberth, K.E., 1997: The definition of El Niño. Bull. Amer. Meteor. Soc, 78, 2771-2777.
[Available online at http://www.cgd.ucar.edu/cas/papers/clivar97/en.dfn.html]
Trenberth, K.E. and T.J. Hoar, 1996: The 1990-1995 El Niño-Southern Oscillation event:
longest on record. Geophysical Res. Lett., 23, 57-60.
Turner, B. L., II, R. E. Kasperson, P. A. Matson, J. J. McCarthy, R. W. Corell, L. Christensen,
N. Eckley, J. X. Kasperson, A. Luers, M. L. Martello, C. Polsky, A. Pulsipher, and A. Schiller,
2003: Science and Technology for Sustainable Development Special Feature: A
framework for vulnerability analysis in sustainability science. The Proceedings of the
National Academy of Sciences of the United States of America, 100, 8074-8079.
United Nations Development Programme (UNDP), 2009: The Human Development
Concept. [Available online at http://hdr.undp.org/en/humandev/]
288

United Nations Office for the Coordination of Humanitarian Affairs (UNDRO), 1990b:
Peru Earthquake May 1990 UNDRO Situation Reports, 1-7. [Available online at
http://www.reliefweb.int/rw/rwb.nsf/db900SID/OCHA-64C3R8?OpenDocument]
--------1990a: Peru Drought June 1990 UNDRO Situation Reports, 1-5. [Available online at
http://www.reliefweb.int/rw/rwb.nsf/db900sid/OCHA-64BQQC?OpenDocument]
--------1998: Peru El Niño floods OCHA Situation Report, 11. [Available online at
http://www.reliefweb.int/rw/rwb.nsf/db900SID/ACOS-64D2FY?OpenDocument]
United Nations Research Institute for Social Development (UNRISD), 2007: Equitable
access to health care and infectious disease control. Report of an International
Symposium, 13-15 February, Rio de Janeiro, Brazil. [Available online at
http://apps.who.int/tdr/publications/tdr-research-publications/equitable-access-healthcare/pdf/conference_news.pdf]
Universidad de Piura (University of Piura [UDEP]), 2008: Datos Climaticos (Climate
Datasets). Piura, Peru.
Valverde, A., 1998: Influence of El Niño in the systems of hydroelectric generation. In
Assessment of the use of remote sensing and other information related to ENSO: the use
of ENSO information in Peru, M.H. Glantz, Ed., NASA/NCAR/Peru Project Final Report.
Boulder, CO. Environmental and Societal Impacts Group, National Center for
Atmospheric Research, pp. 45-48.
Vincent, K., 2007: Uncertainty in adaptive capacity. Global Environment Change, 17, 1224.
Wallace, R., 1988: A synergism of plagues: “Planned shrinkage,” contagious housing
destruction, and AIDS in the Bronx. Environmental Research, 47, 1-33.
Walker, G.T., 1924: Correlation in seasonal variations of weather, IX: a further study of
world weather. Memoirs of the Indian Meteorological Department XIV (part IX), pp.
275-332.
Watts, M. and H.G. Bohle, 1993: The space of vulnerability: the causal structure of
hunger and famine. Progress in Human Geography, 17, 43-67.
Wilson M.L., 2001: Ecology and infectious disease. In Ecosystem Change and Public
Health: A Global Perspective, J.L. Aron and J. Patz, Eds. Johns Hopkins University, pp.
283-324.
Wisner, B., P. Blaikie, T. Cannon, and I. Davis, 2004: At Risk: Natural hazards, People’s
Vulnerability and Disasters. Routledge.
Woodman, R., 1998: El Fenómeno El Niño y el Clima en el Peru (The El Niño
phenemonon and climate in Peru). In El Peru en los Albores del Siglo XXI (Peru at the
289

Turn of the 21st Century), Ediciones del Congreso del Peru, Lima-Peru, pp. 201-242. (in
Spanish)
World Health Organization (WHO), 2006: Vended water. Water and Sanitation Health.
[Available online at
http://www.who.int/water_sanitation_health/gdwqrevision/vendedwater/en/]
-------------2008: Protecting health from climate change: Report for World Health Day
2008. [Available online at
http://www.who.int/globalchange/publications/CC_Report_WHD2008/en/index.html]
-------------2009: Cholera Factsheet. [Available online at
http://www.who.int/mediacentre/factsheets/fs107/en/index.html]
World Meteorological Organization (WMO), 1999: The 1997-1998 El Niño Event: A
Scientific and Technical Retrospective. WMO-No. 905. World Meteorological
Organization.
Wolter, K., 2010: Multivariate ENSO Index. Earth Systems Research Laboratory.
[Available online at http://www.esrl.noaa.gov/psd/enso/enso.mei_index.html]
Wyrtki, K., 1975: Teleconnections in the equatorial Pacific Ocean. Science, 180, 66-68.
Youngers, C.A., 2000: Deconstructing democracy: Peru under President Alberto Fujimori.
Washington Office on Latin America.
Zapata-Velasco, A. and K. Broad, 2001: Peru country case study: Impacts and responses
to the 1997-98 El Niño event. In Once Burned, Twice Shy? Lessons Learned from the
1997-98 El Niño, M. H. Glantz, Ed., United Nations University Press, pp. 186-199.
Zimmerer, K.S. and T.J. Bassett, 2003: Political Ecology: An Integrative Approach to
Geography and Environment-Development Studies. The Guilford Press.
Zuckerman, J.N., L. Rombo, and A. Fisch, 2007: The true burden and risk of cholera:
Implications for prevention and control. Lancet Infectious Diseases, 7, 521-530.

290