. < ‘ .. NE.
a é." {£33.
4 .. . 9. 33.:

.13. 3%

 

 

 

 

, S. ,
43.x?uu

 

. ...¥e&nr£.

.35.}. . yin. 2“.» V ‘ u.
fie. :: up

. I
3...... 1L... .. .5:
i2: 3. £9... .fihna
.337... .{fitgdlls raw.
1 71!: .x A .
1.7;. .
7: .
I. I. A:
11...}..39hv
r. r

z .
Uh;

Rug;
Fig.3...
3. .333.

it)... fiy‘t

:21”. ' 4:1... AV. Emma...
.=,.Ӥfl.....w gr

.1 1.5. . .Q
«31;... \

 

 

 

 

 

.. . , , . ; .«ncusflnugufiumw
. “Ev. ..

f 7745!"!

l LIBRARY
100 3 Michigan .State
Unnversnty

This is to certify that the
thesis entitled

EVALUATION OF AN EMERGENCY DEPARTMENT ASTHMA
SURVEILLANCE SYSTEM IN GRAND RAPIDS, MICHIGAN

presented by

Stacey Jane Elder

has been accepted towards fulfillment
of the requirements for the

Master of Science degree in Epidemiology

 

 

"Malta, ”5. flaw

Major Professor's Signature

/ r7] 2&96 .

 

Date

MSU Is an Afflnnative Action/Equal Opportunity Institution

on-1-.---I---u-o---v-I-o-o-IOu-o-o-n-I-I---n-o-o-----.--o-a-u-o-u-o-o-n-n-n-u-o--o---|_n--u-c-u-o-o-o-o-v-u-n-u-.-.-

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2/05 p:/CIRC/DateDue indd-p.1

 

EVALUATION OF AN EMERGENCY DEPARTMENT ASTHMA
SURVEILLANCE SYSTEM IN GRAND RAPIDS, MICHIGAN

By

Stacey Jane Elder

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

MASTERS OF SCIENCE
Department of Epidemiology

2006

ABSTRACT
EVALUATION OF AN EMERGENCY DEPARTMENT ASTHMA SURVEILLANCE
SYSTEM IN GRAND RAPIDS, MICHIGAN
By

Stacey Jane Elder

Asthma is a serious and potentially fatal problem affecting an estimated 20
million people in the United States. It is a disease that can usually be well
controlled with medication. Asthma accounts for approximately 1% of all health
care costs, 43% of which are due to emergency department (ED) visits. The
current study, performed in the greater Grand Rapids, Michigan area, was
designed to determine if ED physician billing records could serve as an accurate
and timely source of information for developing a surveillance system to monitor
trends in asthma-related ED visits. The available billing data included the
number of asthma-related ED visits per month by age, gender, number of repeat
visits, day of the week on which the event occurred, and the rate of hospital
admissions from the ED by age. A formal evaluation of the surveillance system
was performed, using criteria set forth by the Centers for Disease Control. This
included a special validation study to determine validity of the billing code data.
Results revealed that the ED asthma surveillance system reflected the seasonal
trends, as well as age and gender trends found in many previously published
results. Although the system was found to be highly accurate and reliable, it is
doubtful that broadening it statewide would be successful, due to a lack of

support from the Michigan Hospital and Healthcare Association.

ACKNOWLEDGEMENTS

In researching and writing this thesis, I have had the benefit of support and
encouragement from five wonderful groups of people: 1) the faculty and staff of
the department of epidemiology at Michigan State University;'2) the staff at
Medical Management Services; 3) the staff at Spectrum Health's Buttenrvorth
Hospital coding department; 4) the staff at Michigan Department of Community

Health; 5) my fellow classmates; and 6) my husband and parents.

Dr. Mathew Reeves of Michigan State University as my principal advisor for
developed my interest in this project and for provided guidance and funding.
Two other faculty members, Dr. Ellen Velie and Dr. P.K. Pathak of Michigan
State University and staff member Lora McAdams provided their support and
useful feedback while writing this thesis. The staff at both Medical Management
Services and Spectrum Health’s Butterworth Hospital provided me with the raw
data used for the analysis in this thesis and their time in answering my many
questions. Sarah Lyon-Callo lent me her invaluable knowledge of asthma. My
Classmates Emily Murray and Susan Bohm, as well as my supervisor, Dr. Ronald
Pisoni, offered me emotional support and invaluable advice when I felt as if I

would never finish this project. I thank you all.

Finally, I would like to thank my husband for his never-ending encouragement

and support during this endeavor. My parents were also full of encouragement.

TABLE OF CONTENTS

LIST OF TABLES ................................................................................................ vii
LIST OF FIGURES ............................................................................................. viii
CHAPTER 1 - JUSTIFICATION AND BACKGROUND ......................................... 1
A. Reasons for Studying Asthma ....................................................................... 1
AI Definition of Asthma ................................................................................ 1
A2 Prevalence and Mortality ......................................................................... 3
A3 Trends in Prevalence and Mortality ......................................................... 5
A4 Seasonal Variation in Asthma ................................................................. 8
A5 Disproportionate Effects on Minorities ................................................... 11
AG Risk Factors and Triggers of Asthma .................................................... 14
A] Preventability ......................................................................................... 17
B. Asthma Surveillance ................................................................................... 19
8.1 Definition of Public Health Surveillance ................................................. 19
3.2 Types of Surveillance Systems ............................................................. 20
8.2.1 Use of Death Certificates to Measure Asthma Burden .................... 20
8.2.2 Asthma-Related Hospitalizations .................................................... 22
3.23 National Data Sources for Measuring Incidence and Prevalence ...24
824 ED Asthma Surveillance ................................................................. 25
C. Validation of an ED Asthma Surveillance System ....................................... 26
D. Opportunities and Rationale for an ED Asthma Surveillance System ......... 27
D1 Justification and Advantages of an ED Asthma Surveillance System ...28
E. Limitations of ED Surveillance .................................................................... 29
F. OBJECTIVES ..................................... 31
F1 Objective 1 ............................................................................................. 31
F2 Objective 2 ............................................................................................. 31
F.3 Objective 3 ............................................................................................. 32
CHAPTER II - METHODS ................................................................................... 33
A. EMERGENCY DEPARTMENT ASTHMA SURVEILLANCE SYSTEM

(EDASS) .......................................................................................................... 33
A1 Description of Hospitals ......................................................................... 33
A2 Description of the Medical Record Coding Process - Hospital .............. 33

AB Description of the ED Physician Billing Company Data and Coding
Process ........................................................................................................ 35
A4 Data Collection ...................................................................................... 36
AS AnalySIs ....................................... 36
B. ED SURVEILLANCE SYSTEM EVALUATION ................ ' ........................... 3 7
8.1 Usefulness ............................................................................. -. ................ 37
B.2 Simplicity ............................................................................................... 38
B.3 Flexibility ................................................................................................ 39
B.4 Completeness ....................................................................................... 39

B.5 Acceptability .......................................................................................... 39

8.6 Sensitivity .............................................................................................. 40
B.7 Positive Predictive Value ....................................................................... 40
B.8 Representativeness ............................................................................... 40
B.9 Timeliness ............................................................................................. 41
8.10 Stability ................................................................................................ 41
C. VALIDATION STUDY ................................................................................ 41
O1 Sample Size .......................................................................................... 43
C2 Sampling Strategy - Cases ................................................................... 44
0.3 Sampling Strategy — Controls .................. g .............................................. 45
CHAPTER 3 — RESULTS ................................................................................... 47
A. DESCRIPTIVE ANALYSIS OF THE EMERGENCY DEPARTMENT
ASTHMA SURVEILLANCE SYSTEM ............................................................. 47
AI Number and Proportion of ED Visits and ED Asthma Visits per Month by
Site .............................................................................................................. 47
A2 Comparison of the Monthly Proportion of All ED Visits and ED Visits for
Asthma ......................................................................................................... 47
A21 Comparison of the Monthly Proportion of ED Asthma Visits by
Hospital .................................................................................................... 48
A22 Comparison of the Monthly Proportion of ED Asthma Visits by Age
................................................................................................................. 50
A23 Comparison of the Number of ED Asthma Visits by Age and Gender
................................................................................................................. 51
A24 Comparison of the Monthly Proportion of ED Asthma Visits by
Gender ..................................................................................................... 52
A3 Comparison of the Monthly Proportion of Repeat Asthma ED Visits ..... 53
A4 Comparison of the Monthly Proportion of Hospital Admissions from ED
Asthma Visits ............................................................................................... 53
A5 Comparison of ED Visits for Asthma Visits by Day of the Week and
Hospital ........................................................................................................ 55
B. VALIDATION STUDY ................................................................................. 56
8.1 Analysis ................................................................................................. 58
8.1.1 Sensitivity ........................................................................................ 58
8.1.2 Specificity ........................................................................................ 58
8.1.3 Positive Predictive Value ................................................................ 59
81.4 Measurement of agreement (Kappa) .............................................. 59
B15 False Negatives .............................................................................. 60
Possible Explanation for Inconsistency ........................................................ 61
B16 False Positives ............ ' .................................................................... 61
CHAPTER 4 - DISCUSSION ............................................................................. 64
A. DESCRIPTIVE ANALYSIS OF THE EMERGENCY DEPARTMENT
ASTHMA SURVEILLANCE SYSTEM ............................................................. 64
B. EMERGENCY DEPARTMENT SURVEILLANCE SYSTEM EVALUATION 68
B.1 Usefulness ............................................................................................. 68

8.2 Simplicity ............................................................................................... 69

8.3 Flexibility ................................................................................................ 70
8.4 Data Quality ........................................................................................... 70
8.5 Acceptability .......................................................................................... 71
8.6 Sensitivity ...................................................... _ ........................................ 71
8.7 Positive Predictive Value ....................................................................... 72
8.8 Representativeness ............................................................................... 72
8.9 Timeliness ............................................................................................. 73
8.10 Stability ................................................................................................ 73

C. RELIABILTY AND VALIDITY OF THE ED ASTHMA SURVEILLANCE
SYSTEM .......................................................................................................... 74
CHAPTER 5 - CONCLUSIONS ......................................................................... 79
REFERENCES ................................................................................................... 82

vi

LIST OF TABLES

TABLE 1
1999 annual rate of deaths with asthma as the underlying cause of death by age
group per 1,000,000 population. Age-adjusted to 2000 United States population.

From CDC Surveillance for Asthma—United States, 1980-1999 .......................... 4
TABLE 2

Breakdown by age of asthma prevalence and mortality rates for 1998 from
Surveillance for asthma, United States 1980-1999 ............................................... 5
TABLE 3

Annual prevalence, death, and ED visit rate comparisons between races. 1) per
1,000, 2) per 1 million , 3) per 10,000 ................................................................ 11
TABLE 4

Non-asthma diagnoses and their corresponding ICD-9 codes used for the control
population ........................................................................................................... 46
TABLE 5

A comparison of the numbers and frequencies of non-asthma diagnoses for the
total numbers and the validation sample ............................................................. 57
TABLE 6

A list of false negative results (cases with a diagnosis of asthma on medical
chart, but non-asthma on physician billing record) and their characteristics ....... 61

TABLE 7
A list of false positive results (controls with a diagnosis of not asthma on medical
chart, but asthma on physician billing record) and their characteristics .............. 63

vii

 

LIST OF FIGURES

FIGURE 1 . 7 —

Conceptual design to illustrate a model for public health surveillance. Mortality,
hospitalization, and ED/Urgent care are sentinel events. Hospitalizations,
ED/Urgent care visits, and office visits are measures of health care utilization.
Asthma prevalence and severity shows the underlying disease occurrence and

attack rates ......................................................................................................... 20
FIGURE 2

Venn diagram depicting the subsets compromising COPD and the relationships
between them. .................................................................................................... 30
FIGURE 3

Determining sensitivity, Specificity, and positive predictive value of the physician
billing record diagnosis, using the medical chart diagnosis as the ‘gold standard’

............................................................................................................................ 43
FIGURE 4

A comparison of the percentage of ED asthma visits and the percentage of visits

for all reasons by month, 2001 ............................................................................ 48
FIGURE 5

A hospital comparison of the percentage of ED asthma visits by month, 2001...49
FIGURE 6

Percent of ED asthma visits by month and age (child vs. adult), 2001 ............... 49
FIGURE 7

A comparison of the number of ED asthma visits by age and gender, 2001 ....... 50
FIGURE 8

Percent of ED asthma visits by month and gender, 2001 ................................... 51
FIGURE 9

Percent of repeat visits to the ED for asthma by month, 2001 ............................ 52
FIGURE 10

Hospital admission rates for asthma by month, 2001 ......................................... 54
FIGURE 11

Hospital admission rates for asthma by month and age, 2001 ........................... 55

viii

FIGURE 12

Hospital comparison of ED asthma visits by day of the week, 2001 ................... 56
FIGURE 13

Results of the comparison of ED medical records coding with ED physician billing
data ..................................................................................................................... 58

EVALUATION OF AN EMERGENCY DEPARTMENT ASTHMA
SURVEILLANCE SYSTEM IN GRAND RAPIDS, MICHIGAN

CHAPTER 1 - JUSTIFICATION AND BACKGROUND

A. Reasons for Studying Asthma

Asthma, a chronic inflammatory disorder of the airways, is a serious and
potentially fatal problem affecting an estimated 20.5 million people in the United
States.1 It is one of the most common chronic diseases in developing countries
and has been increasing in prevalence Since the late 1970s, despite better
understanding of the physiologic processes and improvements in treatments.
Asthma places a large burden on our health care system in terms of cost and
healthcare utilization, however it is controllable so there is potential to alleviate

much of this burden.

A.1 Definition of Asthma

Clinicians, physiologists, immunologists, and pathologists all have different
perspectives on asthma in terms of its definition, however the merging of these
many viewpoints into an acceptable working definition of asthma is important for
more specific and effective treatments of the disorder and for investigations into
its pathogenesis. Based on current knowledge, the following working definition of
asthma has been developed by the National Heart, Lung, and Blood Institute

(NHLBI):

Asthma is a chronic inflammatory disorder of the airways in which many cells and
cellular elements play a role, in particular, mast cells, eosinophils, T'Iymphocytes,
macrophages, neutrophils, and epithelial cells. In susceptible individuals, this
inflammation causes recurrent episodes of wheezing, breathlessness, chest
tightness, and coughing, particularly at night or in the early morning. These
episodes are usually associated with widespread but variable airflow obstruction
that is often reversible either spontaneously or with treatment. The inflammation
also causes an associated increase in the existing bronchial hyper-

responsiveness to a variety of stimuli. 2

In 1998, the Council of State and Territorial Epidemiologists (CSTE) developed
surveillance case definitions for asthma that can be applied to data collected
from either vital records, hospital discharge databases, population surveys, or
clinical and laboratory data.3 As an example, a hospital discharge is classified as
a probable case of asthma when the hospital discharge record lists one of the
following codes as a primary diagnosis: 1) ICD-9-CM 4930-4939 or 2) ICD-10-
CM J45.0-J45.9. For children <12 years of age, the lCD-9-CM codes 466 (acute
bronchitis or bronchiolitis), 491.20 and 491.21 (chronic bronchitis) may also be

considered as possible cases of asthma.

The diagnosis of asthma requires both a well-established clinical history and
complementary clinical examination findings. Unfortunately the diagnosis of
asthma is further complicated by:

1) the absence of any diagnostic ‘gold standard’,

2) the variable severity of the condition,
3) the variable seasonality of the condition, and
4) the reluctance on behalf of some clinicians to use the term asthma and

therefore make a formal clinical diagnosis.

A.2 Prevalence and Mortality

The prevalence of asthma varies from population to population, depending on
how the data are collected and what definition of asthma is used. Population-
based and epidemiologic surveys have not always used a consistent definition of
asthma. According to data collected from the Behavioral Risk Factor
Surveillance System (BRFSS) in 2000 from all 50 states, the point prevalence of
asthma in adults was 7.2 percent (95% confidence interval (CI) 7.0-7.4) and the
lifetime prevalence was 10.3 percent (95 % CI 8.9-11.7). 4 In Michigan, the
prevalence of asthma in 2000 according to the BRFS was 7.3 percent (CI 6.2-
8.5), while the lifetime prevalence was 10.3 percent (CI 8.9-11.7), both very Close
to the national estimates. The year 2000 was the first year that asthma data were
collected by the BRFSS in all 50 states. Before this the National Health Institute

Survey (NHIS) had been the only source for national prevalence data.

Data taken from a combination of national surveys, including the NHIS, the
National Ambulatory Medical Care Survey (NAMCS), the National Hospital
Ambulatory Medical Care Survey (NHAMCS), and the Mortality Component of

the National Vital Statistics System were used by the Centers for Disease

 

Control’s (CDC) National Center for Health Statistics (NCHS) in their surveillance
for asthma. A report released by CDC in March 2002 reported that 10,488,000
people in the United States had an asthma episode during a 12 month period. Of
these, over 3 million were children under the age of 15. 5 This makes asthma the
most common chronic childhood disease and the fourth leading cause of

disability in children. 6'7

In 2002, the mortality rate for asthma as an underlying cause of death was 1.5
deaths per 100,000 people for a total of 4,261 deaths in the United States. In
1998, among those less than 18 years of age the mortality rate was 0.3 deaths
per 100,000 people, compared to 1.9 deaths per 100,000 people in those greater
than or equal to 18 years of age. 8 Table 1 shows a breakdown of asthma

mortality by age for the United States.

 

Age Group 0-4 5-14 15-34 35-64 365 Total*

 

Death Rate 1.7 3.6 5.9 15.8 69.9 17.2

 

 

 

 

 

 

 

 

 

Table 1. 1999 annual rate of deaths with asthma as the underlying cause of
death by age group per 1,000,000 population. *Age-adjusted to 2000 United
States population. From CDC Surveillance for Asthma—United States, 1980-
19995

Age plays a major role in asthma prevalence and mortality as can be seen in
Table 2 taken from the NCHS data from 1980-1999. 8 Persons between the ages

of 5 and 14 have the highest prevalence of asthma, while those 65 years of age

and older have the lowest. Mortality due to asthma is very rare in children, but as

 

age increases, so does the mortality rate. Those aged 65 and older had the
lowest prevalence of asthma, but the highest mortality rate from asthma. There
may be an issue with validity of diagnosing asthma in older persons, as there are
many differential diagnoses for chronic respiratory conditions, such as

emphysema and COPD.

 

 

 

 

 

 

 

Age Asthma prevalence per Asthma mortality rate per
1,000 people in 1998 1 million people in 1998

0-4 46.4 2.1

5-14 57.8 3.8

15-34 37.5 6.4

35-64 35.7 17.8

365 28.7 86.9

Total 39.2 20.3

 

 

 

 

 

Table 2. Breakdown by age of asthma prevalence and mortality rates for 1998
from Surveillance for asthma, United States 1980-1999. 5

A.3 Trends in Prevalence and Mortality

When using data from the NHIS, it is not possible to compare data from before
1997 with that collected from 1997 and beyond as the survey questions
concerning asthma were Changed in 1997. Before 1997, anyone who was
reported to have had asthma in the last 12 months was counted as having
asthma. The redesigned NHlS now requires a medical diagnosis of asthma and
an episode or attack in the past 12 months in order for a person to be

categorized as having asthma. Since this definition requires the person to have

had an asthma episode in the last year, a person whose asthma was well
controlled through medication would not be counted as having asthma. AS a
result, prevalences of asthma since 1997 have been lower than those generated
before 1997. As an example, in 1996 the estimated prevalence of asthma was
54.6 cases per 1,000 people but was reduced to 39.2 cases per 1,000 people in

1998 using the new definition of asthma. 5

The prevalence and mortality rate of asthma have been increasing over the past
40 years for people of all ages. The increase in prevalence over the last several
decades has been 5% to 6% per year, according to some of the most well
conducted studies. 9 1° The United States and Canada experienced a 25%
increase in asthma mortality between 1960 and 1998. 11 According to the NHIS,
between 1980 and 1996 the self- or proxy-reported 12-month prevalence of
asthma in the United States alone increased 73.9% from nearly 7 million cases
or 31.4 cases per 1,000 people to 14.6 million persons or 54.6 cases per 1,000
people. The asthma prevalence in those who were 65 years of age and older
increased the least, from 31.9 cases per 1,000 people in 1980 to 39.4 cases per

1,000 people in 1995. 9

When the change in asthma prevalence was examined by age, the most rapid
rise between 1980 and 1995 occurred in children aged 0-4 years with an
increase from 22.2 cases per 1,000 people with self-reported asthma in 1980 to

60.5 cases per 1,000 people with self-reported asthma in 1995. 9

lntemationally, the prevalence of asthma is also increasing in countries such as
Australia, France, New Zealand, and the United Kingdom. 6 As an example, a
cross-sectional population prevalence survey questioned the parents of Children
7V2 to 8V2 years of age about wheezing in London, England in 1978 and 1991. In
1978 the prevalence of wheezing in children was 11.1%, but had increased by

16% to12.8% by 1991 in children of the same age in London. 12

The mortality rate from asthma increased between 1980 and 1996 from 14.4
deaths per million people (2,891 deaths) to 21.8 deaths per million people (5,667
deaths) in the United States. 5 Vital statistics data from 1960 and 1980 also Show
an increase in asthma mortality in Australia, Canada, Germany, Japan,
Singapore, Sweden, Switzerland, and the United Kingdom. However, more
recent data have shown a decrease in England and Wales. Countries that are
more developed in terms of urbanization, westemization, and increased
affluence, tend to have a higher prevalence of asthma, as well as greater
increases. 6 Reasons for this are not well understood, although greater amounts

of air pollution in these areas may play a part.

Studies are generally consistent in the observation that the prevalence of asthma
is higher in males up until adolescence whereupon the trend switches and
females have a higher prevalence thereafter. ‘3 14 15 16”Various explanations for

this crossover phenomenon have been proposed. For example, a study in New

 

Zealand on 13 year-old boys and girls suggested that one reason for boys having
a higher prevalence of asthma than giris may be due to them having a higher
sensitivity to allergens. ‘7 An explanation as to the higher prevalence in adult
women came from a study performed in the Ukraine. It was found that adult
women with mild to moderate asthma were at increased risk of an asthma
exacerbation due to psychological distress than are adult men with mild to

moderate asthma. ‘8

It has been noted that there is an association between asthma and menses. This
may explain the increased frequency of adult female ED visits and admissions to

the hospital. ‘9

A.4 Seasonal Variation in Asthma

Seasonal variation in asthma burden has been observed worldwide including
England, Finland, Italy, Japan, New Zealand, and the United States. In countries
where there are large differences in seasonal weather patterns, there tends to be
larger numbers of asthmatic episodes in mid fall and late spring. It has been
suggested that seasonal variation in grass pollen counts, temperature, humidity,
viral infections, fungal spore counts, and dust mite counts may all be contributing

factors to seasonal asthma variability. 20

As an example of seasonal variability, a prospective cohort study in Italy followed

two groups of asthmatic subjects; one had an allergy exclusively to dust mites

 

(n=165) and the other had allergies to dust mites and grass pollens (n=79).
Results showed that the group with only the dust mite allergy had a significant
increase in asthma attacks during the autumn months (September, October, and
November) as compared to spring (p<0.001) and winter (p<0.002). Those with
grass and mite allergies had a significantly higher number of asthma attacks
during the summer as opposed to the winter (p=0.017), followed by spring and
autumn. Results similar to these have been observed in other studies performed
in temperate climates where air humidity is variable throughout the year. Dust
mites prefer high humidity and autumn is usually the season when the humidity is

highest. 2°

An 8-year retrospective study performed in the United Kingdom used weekly
incidence data of asthma episodes reported in the Weekly Returns Service.-
(WRS) of the Royal College of General Practitioners to examine seasonal
variations in general practice office visits. Data were also obtained to document

hospital admissions and deaths related to asthma over the same time period. 21

Among those 0-14 years of age, visit rates and hospital admission rates had 6
total peaks throughout the year, the largest being in late September and lowest in
August. In the 15-44 year age group, there were two definite peaks in office
visits in June/July and September/October, but there was only one hospital
admissions peak in September/October. In the 45 and older group, general

practice visits and hospital admissions were highest in January and December.

with low hospital admission rates in March. Otherwise, visits and admissions
were fairly consistent during the rest of the year. The mortality rate for asthma
was highest in summer for those 15-44, but lowest for those 65 and older.
Deaths in those 65 and older were highest in the winter months. These results
show that variations in asthma burden are seasonal, but also vary with age.
They also indicate that hospital admissions, general practice visits, and deaths

due to asthma are not always congruent.

A 15-year retrospective study performed in Finland collected national-level data
on seasonal variations of asthma burden in children aged 5-9 years. 22 The
results showed that in May and October asthma episodes were highest at 35.6%
and 41.3%, respectively, above normal. January and July were the months with
the lowest asthma burden with trends 20.5% and 31.1%, respectively, below
normal. Asthma exacerbations were more frequent in girls than in boys in both
autumn and winter. These results follow the usual seasonal variations found in

temperate climates in the northern hemisphere.

In New Zealand an historical study used data from the New Zealand Health
lnforrnation service that was collected between 1978 and 1995 to determine
seasonal variations in hospitalizations and death rates due to asthma, based on
a total of 185,307 hospitalizations. ‘7 Seasonally, these results were very similar

to those seen in studies performed in the northern hemisphere.

10

A.5 Disproportionate Effects on Minorities

Many epidemiological studies performed in the United States concerning asthma
have focused on three racial groups: blacks, whites, and Hispanics. These
groups are compared ethnically rather than racially, as common phenotypes in
skin color rarely suggest common genotypes, but more often imply common
social structures, geographic origins, dietary preferences, and employment
patterns. Racial differences concerning asthma exist not only in the prevalence

of the disorder, but also in the mortality rate and health care utilization patterns.

 

 

 

 

 

 

 

 

White Black

Prevalence‘l
1980 31.4 33.1
1996 53.6 65.5

Death Rate2
1980 12.9 27.6
1999 14.2 38.7

ED Visits?
1992 43.7 143.2
1999 59.4 174.3

 

Table 3. Annual prevalence, death, and ED visit rate comparisons between
races. 1) per 1,000, 2) per 1 million , 3) per 10,000.6

The difference in asthma prevalence between blacks and whites has been
increasing in the last 2 decades (Table 3). In 1980, the estimated annual
prevalence of self-reported asthma for whites in the United States was 31.4 per
1,000 and was 33.1 per 1,000 for blacks. In 1996, the prevalence in whites rose

to 53.6 per 1,000 and in blacks it increased to 65.5 per 1,000. 6

11

Blacks in the United States have more than two times the annual rate of deaths
with asthma as the underlying cause of death diagnosis as whites (Table 3).
This difference has remained fairly consistent throughout the last 20 years. In
1980, the annual rate of death‘with asthma as an underlying cause in whites was
12.9 per 1 million and was 27.6 per 1 million in blacks. In 1999, the death rate

rose to 18.1 per 1 million in whites and to 48.0 per million in blacks. 6

Visits to the emergency room for asthma are also disproportionate in blacks
(Table 3). The rates are nearly three times higher among blacks than in whites. 6
In 1992, 43.7 per 10,000 whites compared with 143.2 per 10,000 blacks visited
the emergency room and were admitted rising in 1996 to 59.4 per 10,000 and

174.3 per 10,000, respectively.

Reasons for the racial differences in asthma prevalence have been studied. A
case control study performed in New York’s Mount Sinai Hospital examined all
asthma deaths and near deaths between 1986 and 1992 to determine risk
factors including ethnicity and poverty level. 23 The control group consisted of
randomly chosen visits for asthma in Mount Sinai Hospital occurring on the same
day as the fatal or near fatal event. Of the 13 asthma-related deaths and 20 near
deaths, all but one occurred in African-American and Hispanic patients compared
to the control group that was 70% white. Poverty level was measured by zip

code area and medical insurance status. Only 2.3% of the control group lacked

12

medical insurance as compared to 30.3% in the adverse outcomes group. There
was a total of 90.9% of adverse outcome asthma cases living in zip code areas

deemed economically depressed compared to only 8.8% of controls.

It was noted in 1986 that only 3% of the United States population lived in New
York City, but 6% of all asthma hospitalizations and 7% of all asthma deaths in
the United States occurred in New York City. 24 This may suggest that living in an
urban environment may be a risk factor for asthma. Asthma hospitalization and
mortality rates were 3 to 5.5 times higher for Hispanics and African Americans

than those for whites in New York City, suggesting that race is a major risk factor.

A retrospective cohort study at a large hospital in Indiana sought to show that
patients who did not receive regular outpatient care from a primary care provider
for their asthma were more likely to be hospitalized for their asthma than those
who had regular care from a primary care provider. 25 Their results Showed that
African-American males and females had Significantly lower rates of outpatient
scheduled visits than white males and white females with age-adjusted rate
differences of 37 visits per 100 person-years (95% Cl: 26, 48) and 56 visits per
100 person-years (95% Cl: 46, 66), respectively. African American men had a
higher rate of emergency department visits than white males with an age-
adjusted rate difference of 4.2 visits per 100 person-years (95% CI: 1.7, 6.6).
African-American males also had a Significantly higher hospitalization rate for

asthma than white males with an age-adjusted rate difference of 7.6. Their

13

findings suggested that patients who lack regular care from primary care
physicians had higher rates of asthma-related hospitalizations and those who

lacked regular health care were more likely to be African-American.

A group of researchers from the United Kingdom explored the relationship
between socioeconomic deprivation and hospital admission rates for asthma
during 1989-1991. 26 Areas with high minority rates were observed to have
higher rates of socioeconomic deprivation. Overall results showed a fairly strong
association between the age-standardized admission ratios and socioeconomic
status (p=0.65; p=0.004). The association between the rate of admissions
through accident and emergency departments and the socioeconomic status was
very strong (p=0.83; p<0.001). These results Show a strong association between
those who live in socio-economically deprived areas (areas with high rates of
minorities) of the United Kingdom and hospital admissions for asthma, especially

for those admitted through accident and emergency departments.

A.6 Risk Factors and Triggers of Asthma

Risk factors for asthma include being of lower socioeconomic standing, living in
an urban area, being black, living in a developed country, being male before
adolescence, and being female after adolescence. 5 14 24 Triggers of asthma, on
the other hand, refer to those factors that may put an individual with asthma at
higher risk for having an asthma exacerbation. Triggers for asthma

exacerbations can include upper respiratory infections, flu, allergens, menses,

14

nighttime, and weather. Many environmental pollutants have long been thought
to act as triggers by increasing the risk for asthma exacerbations. These include
air pollutants such as ozone, nitrogen oxide, sulfur oxide, sulfate, carbon
monoxide, and fine particulate matter (PM). As asthma prevalences have been
observed to be highest in developed countries where problems with air pollution
and weather changes tend to be most severe, many studies have been
performed in developed countries in an attempt to identify the role of air pollution
and weather on asthma incidence, prevalence, and attack rates. These studies
often use ED visits as a proxy for incidence of asthma exacerbation. The

following are examples of some of these air pollution-based studies.

Researchers in France performed a time-series study in Paris to determine the
association between ambient air pollution and number of ED visits for asthma in
children (1-15) during 1988. 27 They used data from an existing air pollution-
monitoring network located throughout Paris to measure suspended particulates
of sulfur dioxide, nitrogen dioxide, and black smoke. Relative humidity and daily
average air temperature were also measured and pollen counts were recorded.
The study found that ozone was the only pollutant that caused a significant
increase in ED visits for asthma in children (relative risk=1.52, 95% Cl; 1.06,
2.19). This increased risk was observed 1 day after ozone levels increased by

100 micrograms per cubic meter.

15

An 8-year study of the effects of air pollution (ozone, sulfur dioxide, nitrogen
dioxide, sulfate, and total suspended particles) on ED visits for asthma was
performed in New Brunswick, Canada. 28 Predictors and confounding variables
controlled for included temperature, dew point, and relative humidity. The
frequency of visits to the ED for asthma increased by 33% (95% Cl; 10%, 56%)
when daily one hour maximum ozone concentrations exceeded 75 parts per
billion. The effect that ozone had on ED visits for weather or other air pollutants

did not Significantly change asthma.

A 9-month study in Seattle, Washington attempted to evaluate whether asthma
visits by children under 18 years of age to EDS in the inner city were associated
with outdoor air pollution (carbon monoxide, nitrogen dioxide, sulfur dioxide, and
PM) levels. 29 The study covered a 35 zip code area with 6 EDs. Zip code areas
were divided into 2 groups; high ED utilization (n=29) and low ED utilization
(n=6). A Significant association was found between PM of 10 micrometers (fine
PM) and ED visits in children in low utilization areas and overall areas when a
change of 11 micrograms per cubic meter in fine PM occurred (low utilization
area relative rate (RR)=1.14, 95% Cl; 1.02, 1.27 and all areas RR=1.14, 95% Cl;
1.05, 1.23). Carbon monoxide was significantly associated to ED visits for
asthma in children in low utilization and in all areas combined (low utilization
areas RR=1.15, 95% Cl; 1.05, 1.28 and all areas RR=1.10, 95% Cl; 1.02, 1.19).
When PM was measured by a light scattering method (51 micrometer in

aerodynamic diameter) it was also significantly associated with ED visits for

16

asthma in children in all areas (high utilization RR=1.13, 95% Cl; 1.02, 1.24; low
utilization RR=1.16, 95% Cl; 1.06, 1.27; and all areas RR=1.15, 95% Cl; 1.08,

1.23).

A.7 Preventability

Asthma is not a disorder that can be necessarily prevented, but with proper
management it can be controlled, meaning that the symptoms of asthma can be
prevented or greatly reduced, thus theoretically reducing ED visits. Beside the
fact that experiencing an asthma attack is frightening and uncomfortable, there
are a number of other important reasons for preventing them. Asthma can be a
life-threatening disease and furthermore, untreated asthma exacerbations may
cause long-term decline in lung function. Good asthma management can
prevent chronic and troublesome symptoms (e.g. coughing or shortness of
breath during the night, early morning or after exertion), maintain normal or near
normal pulmonary function, minimize the need for ED visits or hospitalizations, as
well as lead to less lost school or work days, thereby improving quality of life.

In 1999, the NHLBl published guidelines called ‘Guidelines to the Diagnosis and
Management of Asthma,’ designed to guide health professionals to achieve
control over asthma symptoms and standardize asthma diagnosis throughout the
United States. 30 These guidelines were updated in 2002. Persistent asthma is
best managed with daily long-term control medication that is most often an anti-
infiammatory therapy. Initially, continual monitoring is necessary to make certain

that the asthma is being well controlled. Follow-up visits to the doctor should be

17

made every month after diagnosis and then every 6 months as long as proper
control is sustained. Also, the patient or caretaker should be educated and
given an asthma management plan. Control of asthma is defined by the NHLBl
as having a peak expiratory flow (PEF) values of less than 10 to 20 percent
variability or PEF values consistently greater than 80 percent of the patient’s best
PEF value. Successful control of asthma also includes having minimal
symptoms, a minimal need for the use of short-acting Betaz-agonist, absence of

nighttime awakenings, as well as no limitations to activity.

In addition to these medically-based interventions, successful control of asthma
can potentially be achieved by reducing exposure to asthma triggers. Many
asthma exacerbations occur as the result of inhaled allergens (which act as
triggers) such as animal dander or fur, house-dust mites, cockroaches, indoor
fungi (molds), and outdoor allergens (e.g. pollen); occupational exposures;
irritants such as tobacco smoke or pollution. Other triggers include viral
respiratory illnesses, gastroesophageal reflux, use of beta-blockers,
rhinitis/sinusitis, and a sensitivity to aspirin or sulfites. Reducing exposure to
allergens can be achieved by removing animals and products made from
feathers, encasing mattresses and pillows in allergen impermeable covers,
washing the patient’s bed sheets and blankets in hot water, reducing humidity in
the home to less than 50 percent (i.e. no use of humidifiers), removing carpets in
the home, avoiding sleeping on upholstered furniture, minimizing the number of

stuffed toys, and by staying indoors with windows closed and air-conditioning on

18

during midday and afternoon when pollen and spore counts are highest. All
asthma patients should avoid being exposed to tobacco smoke as it is the
highest environmental indoor irritant to children and adult asthmatics. When
outdoor pollution is high, those with asthma Should be encouraged to restrain
from exertion and outdoor exercise. Since it has been shown that upper
respiratory inflammation can cause lower respiratory hyper-responsiveness,
those asthmatics acquiring rhinitis or Sinusitis should be medically treated to
encourage respiratory drainage. Patients with known sensitivities to aspirin,
sulfite, or beta-blockers should refrain from using products containing these
ingredients. Finally, for those with persistent asthma should be given an annual

flu vaccine. 3°

8. Asthma Surveillance

8.1 Definition of Public Health Surveillance

Public health surveillance has been defined by the CDC as the ongoing
systematic collection, analysis, and interpretation of health data essential to
planning, implementation, and evaluation of public health practice, closely
integrated with the timely dissemination of these data to those who need to know.
The final link in the surveillance chain is the application of these data to
prevention and control. A surveillance system includes a functional capacity for

data collection, analysis, and dissemination linked to public health programs. 31

19

8.2 Types of Surveillance Systems

While many potential data sources could be used to document the burden of
asthma such as hospitalizations, ED/Urgent care, and office visits (as depicted in
Figure 1), routine surveillance is currently limited to mortality and hospitalization
for asthma. The ED and urgent care facilities are very important yet

underdeveloped sources of asthma surveillance data.

Mortality

Hospitalizations

 

ED/Urgent Care Visits

 

Office Visit

 

A

Asthma Prevalence and Severitv

 

 

 

 

Figure 1. Conceptual design to illustrate a model for public health surveillance.
Mortality, hospitalization, and ED/Urgent care are sentinel events.
Hospitalizations, ED/Urgent care visits, and office visits are measures of health
care utilization. Asthma prevalence and severity Shows the underlying disease
occurrence and attack rates.

8.2.1 Use of Death Certificates to Measure Asthma Burden

The routine collection of asthma mortality data has major limitations because

deaths from asthma are rare. However, identifying individual deaths may be

20

useful in order to increase our understanding of how to prevent deaths,
particularly in children. It can also be used to understand mortality trends over
time, leading to changes in education and improvement in clinical management.
An asthma mortality surveillance system is based upon death certificates and the
coding of asthma as an underlying cause. Data are usually summarized by age,
gender, and race. A large problem that many studies have found in using death
certificates for collecting data is a lack of accuracy in terms of the correct
attribution of cause of death. 32 33 34 Several studies have shown that the number
of deaths due to asthma on death certificates is overestimated by between 13
and 26 percent. A large part of this overestimation is due to the misclassification
of patients who suffer from chronic obstructive pulmonary disease (COPD). 31 32
A study in New Zealand concurred with a Similar study in the United Kingdom
when the certified cause of death and its subsequent nosological coding was
compared with the opinion of a panel of respiratory physicians who had detailed
information on the medical history and the circumstances surrounding the
deaths. Asthma mortality in persons 15-64 years of age was found to have been
overestimated the most (by 26 percent), whereas mortality occurring in the 0-14
year age group was the least overestimated (only by 5.6 percent), 35 a finding
which was also found in 2 other studies. 33 34 The failure of certifying doctors and
coroners to follow appropriate procedures for the identification of the primary

condition leading to death, or the misdiagnosis of lung disease other than asthma

was said to account for the most inaccuracies in certification.

21

8.2.2 Asthma-Related Hospitalizations

The collection of asthma hospitalization data provides a useful measure of health
care utilization and may also serve as a useful indirect measure of disease
severity and disease control. In the US, data on hospitalizations are reported
by the National Hospital Discharge Survey (NHDS), which collects data on the
inpatient records of 270,000 patients at 500 hospitals on an annual basis. 36
Data collected for the NHDS includes age, sex, race, ethnicity, marital status,
expected sources of payment, admission and discharge dates, and lCD-9

diagnoses and procedure codes.

Hospitalization surveillance can be used to target new initiatives such as
providing better asthma education. In southeast Michigan, an area where
hospitalization rates for asthma are high, asthma has been made a health priority
in a seven-county area. One county health department has organized an
advisory committee to develop strategies for prevention. 37 It can be used to
identify research needs. Researchers in Wisconsin have demonstrated that
asthma surveillance data can be used to investigate correlations between
environmental events and asthma morbidity. 38 It can also be used to evaluate
trends by race, sex, and county. They found that between 1990-1992 and 1996-
1998 overall asthma hospitalization rates declined by nearly 10% from 13.4 to
12.1 per 10,000 patients. The American Heart and Lung Association found that
the hospital discharge rate of asthma increased dramatically from 1970 to 1986,

it then plateaued into the mid 19905, declined in 1996-97, rebounded in 1998 and

22

has remained stable since then. 39 Finally, hospital discharge data can serve as
a key indicator of the impact that hospitalization rates have on the health care

system.

Surveillance systems based on hospitalization data usually collect data from pre-
existing hospital administrative data. Studies of these systems deem them to
have good accuracy in their inter-observer variability as was seen from a medical
record review performed at the Mayo Clinic in Rochester, Minnesota. 4° The
degree of association when 2 nurses reviewed medical records was between
Kappa=0.48 and Kappa=0.58 when making a definite or probable diagnosis of
asthma. Agreement in the classification of asthma severity, however, was
marginal with concordance between nurses occurring between 10 and 48

percent of the time.

Another study performed by the Department of Veteran Affairs attempted to find
the degree of association between administrative files and clinical records for
principal diagnoses for inpatient discharges. Results indicated that the
agreement was good for respiratory-related illnesses (Kappa=0.664 and Kappa:

0.724). 4‘

The problems with this type of surveillance system are that it only provides
information on the most severe cases of asthma and it is unable to provide any
type of incidence or prevalence data. Also the determinants of hospitalization

are not just limited to the disease event itself. For example, studies have

23

identified risk factors for asthma-related hospitalization, which include food
allergen sensitivity in inner city children, 42 second hand smoke exposure, 43 and

being of black race or female, 44 among others.

8.2.3 National Data Sources for Measuring Incidence and Prevalence

AS documented in section A.2, national prevalence and incidence data
concerning asthma is based on existing surveys such as the National Health

Institute Survey (NHIS) and the Behavioral Risk Factor Surveillance Survey.

The NHIS is a cross-sectional annual telephone interview survey that collects
information on the general health of a random sample of Americans. It consists
of an adult section and a child section containing questions concerning
information on health insurance, access to health care, health care utilization,
and medical conditions are included in the survey results. The NHIS provides a
self-reported estimate of asthma prevalence by asking, “Has a doctor or other

health care professional ever told you that you had asthma?”

The BRFSS is used to monitor state-level prevalence of the major behavioral
risks among adults associated with premature morbidity and mortality. 45 The
basic philosophy is to collect data on actual behaviors, rather than on attitudes or
knowledge, that would be especially useful for planning, initiating, supporting,

and evaluating health promotion and disease prevention programs. Asthma data

24

were only included for the first time in 2000. Two questions will now be included,
“Have you ever been told by a doctor that you have asthma?” and “Do you still

have asthma?” Although these surveys are an excellent way to estimate national
disease prevalence from different angles, they fail to provide data on a local level

or information on incidence.

8.2.4 ED Asthma Surveillance

The emergency department is an important setting for asthma surveillance. In
the United States there are nearly 2 million ED visits for asthma each year, 46
many of which should be preventable. 47 The collection of data on ED asthma
visits is an important direct indicator of both the amount of disease burden and
the healthcare utilization due to asthma. Although an ED asthma surveillance
system is not capable of measuring disease prevalence, it can be an indirect
indicator of disease severity and asthma control (i.e. management and treatment)
at the local level. An example of ED surveillance using a survey as the means
for data collection is the National Hospital Ambulatory Care Survey (NHAMCS).
48 The annual survey is designed to collect data on the utilization of ambulatory
care services in hospital emergency and outpatient departments from non-
institutional general and short-stay hospitals in the United States. Physicians are
assigned to a randomly chosen 1-week period in which they are to report data
from a systematic random sample of office visits. Data are obtained on patient’s

symptoms, physician’s diagnosis, medications provided or ordered, patient’s

25

demographics, diagnostic procedures, patient management, and plans for future
treatment. Examples of survey results include the number of ED visits, visit rate,

percent of all visits and those specific to asthma.

C. Validation of an ED Asthma Surveillance System

Validation is the process of establishing that a method or system is accurate.
Measurement validity is an expression of the degree to which a measurement
measures what it purports to measure. In an ED asthma surveillance system the
medical record is considered the ‘truth’ or the ‘gold standard.’ When comparing
ED medical records to ED billing data on diagnosis, we are trying to find the
extent to which the ED billing data agrees with the ‘gold standard.’ Results are

expressed in terms of sensitivity, specificity, and positive predictive value.

. ED data based on billing system data generated by either hospital or physician
groups can measure the number and characteristics of asthma visits, but
questions as to the accuracy of billing data have emerged. There have been few
studies performed to measure the accuracy of using ED billing data for case
identification. Using medical records as a ‘gold standard,’ a validation study was
performed by the New York City Department of Health using 11 public hospitals
in New York City. 49 It was found that only 75.5 percent of those who had a
primary diagnosis of asthma in their medical records had a diagnosis of asthma

in their billing data (this therefore represents the sensitivity of billing data).

26

Conversely, 96.4 percent of those who had an asthma diagnosis in their billing
data also had a primary asthma diagnosis in their medical records (this therefore

represents the positive predictive value of billing data).

Any system should be validated to document how accurate it is. If there is poor
comparability between the system and what it is intended to be measure, then
little confidence can be placed in the data. Poor accuracy of an ED asthma
surveillance system would mean that the number of asthma visits were being
either over or underestimated. A decision would then have to be made as to
whether the system used to measure ED asthma visits could be improved or

whether a new system to measure ED asthma visits Should be found.

D. Opportunities and Rationale for an ED Asthma Surveillance System

At a conference entitled ‘ED Data Surveillance,’ sponsored be the National
Association of Health Data Organizations (NAHDO) in Washington, DC. in April
2002, it was concluded that the development of statewide ED data systems to
measure all types of health events should be given high priority. Existing data
sources such as the NHAMCS and NHIS survey have limitations. For example,
due to the clustered nature of their samples, variance estimates may be large for
characteristics not evenly distributed across hospitals. Also, for rare events, data

collected over many years are necessary before any conclusions could be made.

27

D.1 Justification and Advantages of an ED Asthma Surveillance System

Asthma-related ED visits represent a large portion of asthma morbidity for which
data are not currently available. Surveillance of ED asthma-related visits
generated from already existing data sources, such as billing data, would be
useful to Characterize ED asthma patients in terms of their demographic profile
(such as age, gender, and race); the date and time of presentation; their
geographic location (through zip code of residence); hospital admission rates; ED
repeat visit rates; presence of co-morbidities; payer source (insurance type); and
direct medical costs.

The reasons for developing the system include monitoring trends in ED asthma
visits, developing ideas for interventions to improve asthma care, and having the
ability to evaluate these interventions. Since many ED visits for asthma are
thought to be preventable, these data can potentially be used as a measure of
asthma control (i.e. management and treatment) at a community-level, and as an
outcome measure for population- or system-level interventions. Visits to the ED
for asthma represent an easily measured outcome that is related both to the
severity of the illness and the quality of the previous clinical management

received by the patient. 50

Due to this desire to develop a statewide ED surveillance system, we wanted to
design and implement a pilot ED asthma surveillance system based on billing
system data to identify all visits due to asthma at 3 EDS in the Grand Rapids,-

Michigan area, each serving different populations; i.e. inner-city, suburban, and

28

rural. A pilot study was used in order to determine characteristics of the system
including its feasibility, usefulness, accuracy, acceptability, and costs. AS
opposed to using an active approach to the collection of data, by placing an
emphasis on case finding, we chose to use a passive approach to surveillance,

involving the use of pre-existing billing data. 51

E. Limitations of ED Surveillance

The diagnosis of asthma is a complex problem, in part, because there is no ‘gold
standard’ for the diagnosis of asthma and because there are several differential
diagnoses that can be potentially confused with asthma in both children and
adults. In children upper airway diseases (e.g. pneumonia, both viral and
bacterial), and obstructions involving large or small airways can all be confused
with asthma. Diagnosis of asthma in very young children is difficult in part
because of the inability to perform pulmonary function tests. 52 In adults, COPD
(includes emphysema and chronic bronchitis (Figure 2)), congestive heart failure,
pulmonary embolism, laryngeal dysfunction, pulmonary infiltration with
eosonophilia, obstruction due to benign or malignant tumors, and vocal cord
dysfunction can all be confused with asthma. COPD and congestive heart failure
are common and increase dramatically with age. An adult patient who smokes or
has smoked in the past is more likely to be categorized as suffering from COPD
than asthma although the optimal combination of factors and cut-points in order.
to distinguish asthma from COPD is not known and is currently under

investigation. 2 53 54 However, it is likely that they will never be satisfactorily

29

separated. Some believe that adult asthma is under-diagnosed as the diagnosis

of COPD is almost always given to smokers. 53

Chronic bronchitis Emphysema
/
\

 

COPD- '<——— Airflow Obstruction

 

 

 

 

Figure 2. Venn diagram depicting the subsets compromising COPD and the
relationships between them. 5"

Quandaries over asthma diagnosis are further exaggerated by the knowledge,
attitudes, and practices of ED physicians. Many physicians are either unaware
that guidelines for the diagnosis and management of asthma exist or have not
yet read them. Even when ED physicians have read the guidelines there is
disparity among them about what questions to ask when taking a medical history
and of patients who have asthma. 55 56 Unfortunately the diagnosis of asthma is
further complicated by the reluctance on behalf of some clinicians to use the term
asthma and therefore make a formal diagnosis, plus ED physicians are reluctant
to make a new diagnosis of asthma since they rarely have access to the patient’s
complete medical history. However, most ED asthma visits are in people with a

pre-existing asthma diagnosis. However, even when there is a diagnosis for

30

asthma, it has been noted that physicians tend to underestimate the severity of
asthma which often leads to under-treatment 57 and many healthcare providers
overestimate patients’ asthma control while underestimating the need for patient

education and intervention. 58

F. OBJECTIVES

A pilot ED asthma surveillance system based on physician billing data was set up
at 3 hospitals in the Grand Rapids area beginning in September of 2000. The

following specific objectives will be undertaken.

F.1 Objective 1

The first objective of this study is to analyze and describe the data collected from
a pilot ED asthma surveillance system over a one-year period, January 1, 2001-
December 31, 2001. The characteristics of asthma visits in terms of gender,
age, day of the week, month, hospital admissions, and repeat visits will be

presented.

F.2 Objective 2

The second objective is to perform a formal evaluation of a pilot emergency
department asthma surveillance system based on the criteria set forth by the

Centers for Disease Control, 59 which include the system’s usefulness, Simplicity,

31

flexibility, data quality, acceptability, sensitivity, positive predictive value,

representativeness, timeliness, and stability.

F.3 Objective 3

The final objective is to perform a specific validation study measuring the
accuracy of the ED asthma surveillance system by comparing medical record
codes generated from the hospital with those from the physician billing company.
This objective addresses the characteristics of data quality, sensitivity, and
positive predictive value from objective 2. A determination of the reliability of the

data will also be performed using the Kappa statistic.

32

CHAPTER II - METHODS

A. EMERGENCY DEPARTMENT ASTHMA SURVEILLANCE SYSTEM
(EDASS)

A.1 Description of Hospitals

An ED asthma surveillance system (EDASS) was set Up at three hospitals in the
Grand Rapids, Michigan region. Grand Rapids is located in Kent County in
southwest Michigan and is the second largest city in Michigan, having a
population of about 575,000. Two of the 4 major hospitals in Grand Rapids are
Butterworth Hospital and Blodgett Hospital, owned by Spectrum Health System.
Buttenlvorth Hospital is located in downtown Grand Rapids, an urban
environment, whereas Blodgett Hospital is located in eastern Grand Rapids in a
more suburban area. The third hospital, Gerber Memorial, is located in Newaygo
County, about 35 miles north of Grand Rapids in the small rural town of Fremont,
Michigan where it is the only hospital. Newaygo County has a population of

about 50,000.

A.2 Description of the Medical Record Coding Process - Hospital

For each ED visit, medical coders in the medical records department at each
hospital assign a final diagnosis code (lCD-9 codes), as well as secondary
diagnosis codes when necessary, after reviewing the complete ED medical

record. This department is separate from the hospital-billing department. The

33

final primary and secondary diagnoses are coded after the medical coders have
assessed the documentation provided in the physician’s dictation notes, progress
notes, nursing notes, and test results (i.e. laboratory, radiology, etc.). The
physician’s notes should contain a complete overview of the visit, including a
description of the chief complaint, the history of the presenting illness (HPI), the
patient’s past medical history, the physical examination findings, and the final or

provisional diagnoses.

The chief complaint is first assessed to see whether, in the case of asthma, it is
consistent with an acute onset respiratory condition. Examples of chief
complaints consistent with asthma may include difficulty breathing, asthma

exacerbation, wheeze, and shortness of breath.

The HPI includes signs or indicators of the current illness including, the context of
the present illness, the severity, signs and symptoms, modifying factors, and the

duration.

The patient’s past medical history includes information relevant to the patient’s
chief complaint (e.g. known asthmatic), family history relevant to the patient’s
chief complaint, social history (i.e. smoking status, drug use), and a review of the

patient’s body systems.

Other information, such as the findings from the physician’s examination, is also
assessed to ensure that the presenting illness, history, physical findings, and ED

course of action are consistent with a diagnosis of asthma.

34

The ED course of action and medical decision-making contained in the doctor’s
dictation, documents the diagnostic procedures and treatments undertaken
during the ED visit, as well as the patient’s discharge instructions. AS such,

these are very important in confirming the diagnosis to the coder.

The diagnosis recorded in the medical record can be one of two types, final or
provisional. A final diagnosis in the physician’s notes is always coded as the final
diagnosis by the coder. A provisional diagnosis provided in the medical record
becomes the final diagnosis when a final diagnosis is not provided or it becomes
a secondary diagnosis in the case where a final diagnosis is provided by the
physician. On occasion a hospital coder can make a change to the final or
provisional diagnosis if test results that were unavailable at the conclusion of the
ED visit indicate a better alternative diagnosis.

A.3 Description of the ED Physician Billing Company Data and Coding
Process

AS with hospital medical coding, the diagnosis is coded based on documentation
in the registration form, physician’s notes, progress notes, nursing notes, order
Sheet, and test results. Unlike the hospital coders, physician billing company
coders are given leeway to change a provisional diagnosis, but not a final
diagnosis, if the coder disagrees with the provisional diagnosis after having
reviewed the rest of the medical Chart. However, a physician coder cannot make

a change to a final diagnosis

35

A.4 Data Collection

Because ED physician services are not billed through the hospital, a separate
system operated by a private company, Medical Management Services (MMS),

in Grand Rapids, Michigan is responsible for processing ED physician billing
claims at the 3 hospitals. Starting in October of 2000, all ED bills with a primary
diagnosis of asthma were extracted from MMS on a monthly basis. The number
of monthly ED visits were reported in aggregate form and were categorized by
age group (<2, 2-5, 6-11, 12-17, 18-34, 35-44, 45-54, 55-64, 65-74, and >74),
gender, gender and age, day of the week of the visit, the number of hospital
admissions that resulted from these ED visits was also reported by age (<2, 2-17,
>17), as was the number of repeat visits for asthma during a given month.
Repeat visits were defined as two or more Visits to the ED with a primary
diagnosis of asthma occurring within the same calendar month. These data were
provided by MMS on an Excel Spreadsheetand sent via e-mail on a quarterly

basis to MSU.

A.5 Analysis

Two major outcomes were defined: 1) the absolute number of visits per month
and, 2) the proportion of visits per month across the full calendar year for 2001.
Comparisons between the 3 hospitals using these 2 outcomes were then
generated using gender and age as the major explanatory variables. The

absolute number and monthly proportions for day of the week, hospital

36

admissions, and repeat visits were also generated and compared between
hospitals as a mixture of secondary outcomes and secondary variables. Rates
(i.e. the number of visits per population time) could not be calculated since the

true underlying population denominators are unavailable.
8. ED SURVEILLANCE SYSTEM EVALUATION

In 2001, the CDC released an Updated publication entitled, ‘Guidelines for the
Evaluation of Public Health Surveillance Systems.’ 59 These guidelines identified
the following 10 characteristics to be considered in an evaluation of a public
health surveillance system: usefulness, simplicity, flexibility, data quality,
acceptability, sensitivity, positive predictive value, representativeness, timeliness,
and stability. These guidelines will be used to evaluate our pilot EDASS. The

characteristics are defined below.

8.1 Usefulness

Usefulness refers to whether or not the surveillance system contributes to the
prevention and control of asthma, including an improved understanding of the
public health implications of asthma. The system can also be deemed useful if
its data contributes to performance measures, including health indicators that are
used in needs assessments and accountability systems. The system may be
considered useful if it satisfactorily addresses at least one of the following

questions.

37

1) Does the system provide estimates of the magnitude of morbidity and
mortality related to asthma, including the identification of factors associated
with asthma?

2) Does the system detect trends that signal changes in the occurrence of
asthma, including the detection of epidemics?

3) Does the system permit assessment of the effect of prevention and control
programs?

4) Does the system lead to improved clinical, behavioral, social, policy, or
environmental practices?

5) Does the system stimulate research intended to lead to prevention or

control?

8.2 Simplicity

Simplicity of a surveillance system refers to its structure and ease of operation.
Surveillance systems should be as simple as possible while still meeting their
objectives. Characteristics to be considered when evaluating a surveillance
system include:

1) the number of organizations involved in receiving case reports;

2) the level of integration with other systems;

3) the method of collecting the data, including the number and types of

reporting sources and the time spent collecting data;

4) the method of managing the data;

5) the methods for analyzing and disseminating the data;

38

6) the staff training requirements; and

7) the time spent on maintaining the system.

3.3 Flexibility

Flexibility refers to the adaptability of the surveillance system to changing
information needs with little additional time, personnel, or allocated funds. A
flexible system Should be able to accommodate new health-related events,

changes in case definitions, and variations in funding or reporting sources.

8.4 Completeness

The completeness and validity of the data recorded in the surveillance system is
a measure of the system’s data quality. A Special study has been performed to
evaluate the validity, including the sensitivity, specificity, and positive predictive
value, of the lCD-9 billing codes assigned by the billing company and is included

in this thesis.

8.5 Acceptability

Acceptability reflects the willingness of persons and organizations to participate
in the surveillance program. Factors influencing the acceptability of a
surveillance system include:

1) the public health importance of the related event;

2) the dissemination of aggregate data to reporting sources and interested

39

parties;
3) the ease and cost of data reporting; and

4) the ability of the system to protect privacy and confidentiality.

8.6 Sensitivity

The sensitivity of a surveillance system can refer to either the proportion of cases
of a disease detected and therefore is a measure of completeness, or to the
ability of the system to monitor Changes in the number of cases over time. The

former definition of sensitivity is addressed in our validation study.

8.7 Positive Predictive Value

The positive predictive value of a surveillance system is the proportion of
reported cases that actually have the disease under surveillance. This aspect of

the evaluation is also addressed in the Special validation study.

8.8 Representativeness

A representative surveillance system accurately describes the occurrence of a
disease over time and its distribution in the population by place and person.
Representativeness may also be used to refer to its generalizability. The
representativeness of a public health surveillance system is assessed by

comparing the characteristics of reported events to all such actual events.

40

8.9 Timeliness

Timeliness reflects the time required between steps of the surveillance system.
For the ED asthma surveillance system these steps would include the time
between the ED visit, hospital coding, physician billing, data entry by the
physician billing company, the transfer of this infomation to MSU, and analysis

and dissemination of the data.

3.10 Stability

The stability of a public health surveillance system reflects the ability to collect,
manage, and provide data properly without failure and the ability to be
operational when it is needed. These features are also referred to as reliability
and availability, respectively. Methods of measuring a system’s stability can
include finding the desired and actual amount of time required for the system to
collect or receive data, manage the data (transfer, entry, editing, storage), and
release the data. 59

Each of these 10 attributes was considered during the evaluation of the pilot

EDASS in Grand Rapids, Michigan. Our findings are reviewed in Chapter 4.

C. VALIDATION STUDY

To assess the validity of the coding used in the ED physician billing data, a

separate study was performed using records from Butterworth Hospital. This

41

hospital was selected because it is the largest of the three hospitals. The study
was conducted in children only, defined as 2 through 17 years of age. We
measured the validity or accuracy of the codes by using the primary diagnosis
code contained in the ED medical records (obtained from the hospital) as the
gold standard. We then compared these codes to the corresponding primary
diagnosis code in the ED billing data. A primary diagnosis of asthma was
defined as any lCD-9 493.0-493.9 code. A sample of 150 asthma and non-
asthma diagnoses was randomly selected from the hospital medical records
department. Non-asthma diagnoses were selected from a list of other acute
respiratory diseases and conditions (see section C3). The medical records and
the ED billing data were matched using the medical record number and the date
and time of the visit. From this comparison, the sensitivity, specificity, and

positive predictive values (PPV) were calculated, as shown in Figure 3.

Medical Record Diagnosis Code

 

 

 

Asthma* No asthma
Billing
Data Asthma* a b
Diagnosis
Code No Asthma c d

 

 

 

 

 

Sensitivity = a/(a+c)
Specificity = d/(b+d)

PPV = a/(a+b)

42

Kappa = (a+d)/(1 - ((a+c)/100 - (b+d)/100))
*Asthma defined as lCD-9 493 code
Figure 3. A 2x2 table for determining sensitivity, specificity, and positive

predictive value of the billing data diagnosis code, using the medical record
diagnosis code as the ‘gold standard’

Sensitivity in this study was defined as the probability, given the final diagnosis in

the medical record was asthma, that the diagnosis in the billing data was asthma.

Specificity in this study was defined as the probability, given the final diagnosis in
the medical record was not asthma, that the diagnosis in the billing data was also

not asthma.

The positive predictive value in this study was defined as the probability, given
that the diagnosis in the billing data was asthma, that the final diagnosis in the

medical record was also asthma.

Kappa in this study was defined as the measure of the level of agreement
between the medical record code and the billing code beyond chance.
C.1 Sample Size

Sample size estimates for the study were based on the desire to measure
sensitivity and specificity with sufficient precision. The sample size was

determined using the following formula. 60

43

N = (2,12)2 V(81)/L2
where:
N is the required sample size
Zen is the type I error (two-tailed hypothesis), set at 0.05 (Z=1.96)
V(01) = 01 (1-91) the variance of the sensitivity where 01 is the hypothesized
sensitivity

L is the desired width of 1/2 of the CI. The desired width of the Cl was +/- 0.07.

From a previous validation study performed in New York City of ED billing data
for asthma, it was noted that the sensitivity (61) was 75%. Using this figure,

setting a=0.05, and L=0.07, resulted in a sample size of 147.

N = (1 .962(0.75)(1-0.75))/(0.07)2= 147
This number was rounded up to 150 in case of problems in obtaining data were
encountered. The sample size was then doubled to account for the case group
(i.e. sensitivity) and the control group (i.e. specificity) for a total sample size of

300.

C2 Sampling Strategy — Cases

A case for the validation study was considered to have been any patient between
the ages of 2 and 17 years who presented to the Butterworth Hospital ED
between August 1, 2000 and July 31, 2001 with a chief complaint of asthma or

acute respiratory symptoms (i.e. cough, wheeze, shortness of breath, and/or

44

chest tightness) that had a final diagnosis of asthma (coded to lCD-9 493) by the
hospital upon discharge from the ED. There were 514 visits that met this
definition. An employee in the medical records department of Butterworth
Hospital randomly selected 150 patients from this list. The hospital provided the
following data on these cases: the patient’s medical record number, the date and
time of the visit, the chief complaint with its ICD-9 code, and the attending doctor.
Patients who had more than one asthma visit throughout the 1-year sampling

period, were only included in the sample once.

C.3 Sampling Strategy - Controls

A control was considered any patient between the ages of 2 and 17 years who
presented to the Butten~orth Hospital ED between August 1, 2000 and July 31,
2001 having a chief complaint of asthma or any combination of acute respiratory
symptoms (i.e. cough, wheeze, shortness of breath, and/or chest tightness) that
had a final diagnosis of a respiratory-related condition that was not asthma.
From this group of 838 patients, a random sample of 150 was obtained. The

final diagnoses of this group with their lCD-9 codes are shown in Table 4.

 

 

 

Differential Diagnosis lCD-9
Otitis Media, unspecified 382.9
CrOUP 464.4

 

Acute Upper Respiratory Infection of 465.9
multiple or unspecified sites

 

 

 

 

45

 

Acute Bronchitis or Bronchiolitis 466

 

 

 

 

Acute Bronchiolitis due to other 466.19
infectious organisms

Pneumonia, unspecified 486
Bronchitis, not specified as acute or 490
chronic

Other Diseases of the Trachea and 519.1

Bronchus, NOS

Disease of the Respiratory System, not 519.8
elsewhere Specified

Unspecified Disease of the Respiratory 519.9

 

 

 

 

 

 

 

System

Hypewentilation 786.01
Shortness of Breath 786.05
Wheezing 786.07
Other Dyspnea and Respiratory 786.09
Abnormalities

Cough 786.2

 

 

 

 

Table 4. Non-asthma diagnoses and their corresponding lCD-9 codes used for
the control population

46

CHAPTER 3 — RESULTS

A. DESCRIPTIVE ANALYSIS OF THE EMERGENCY DEPARTMENT ASTHMA

SURVEILLANCE SYSTEM

A.1 Number and Proportion of ED Visits and ED Asthma Visits per Month
by Site

During the one-year period, January 1, 2001 through December 31, 2001, there
were 146,508 total ED visits to the three hospitals, of which 2,095 (1.4%) were
for asthma. Among all asthma visits 1,402 (67%) were from Butterworth
Hospital, 487 (23%) were from Blodgett Hospital, and 204 (10%) were from
Gerber Memorial Hospital. The main outcome measures derived from the billing
data were the number of visits, the number of repeat visits and, the number of

hospital admissions

A.2 Comparison of the Monthly Proportion of All ED Visits and ED Visits for
Asthma

A comparison of the monthly proportion of ED visits for all reasons and all
asthma ED visits is shown in Figure 4. The proportion of all ED visits varied little,
peaking in January with 10 percent of all visits and was lowest in March with 7.6
percent of all visits. In contrast, ED asthma visits showed dramatic variability
across the year. The proportion ED asthma Visits was highest in the month of

March with nearly 12 percent of all ED asthma visits and lowest in July with 5

47

percent of ED asthma visits. There were 2 main peaks for ED asthma visits that

occurred in March and September-October with a smaller peak in May.

14

 

 

 

 

E-l—All ED
Visits

 

 

Percent of Visit:

 

+Asthma I
ED Visits I

 

 

 

 

 

 

JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC
Month

 

 

Figure 4. A comparison of the proportion of ED asthma visits and the proportion
of visits for all reasons by month, 2001.

A.2.1 Comparison of the Monthly Proportion of ED Asthma Visits by

Hospital

When comparing the percent of ED asthma visits by month between the 3
hospitals, it can be seen in Figure 5 that they followed a similar pattern, although

Blodgett Hospital, the suburban hospital, had a higher- peak in the fall than in

spnng.

48

,2 A
‘° :73 /\ /
.\/ \i/Ngiatf

 

 

 

 

oo

 

 

 

_1

Percent of Vlsits

 

 

 

 

 

JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC
Month

Figure 5. A hospital comparison of the percentage of ED asthma visits by month,
2001.

 

.3
O)

...s
A

 

 

/\
fl
6 v//h\\\’ (+Adultl

o I ' T 7 I 1 T TI T r I
JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC

Month

 

..s
O

L
>'/

 

 

 

Percent of Visits
co

 

 

 

 

 

Figure 6. Percent of ED asthma visits by month and age (child vs. adult), 2001.

49

A.2.2 Comparison of the Monthly Proportion of ED Asthma Visits by Age

There were 896 ED asthma visits made by children (less than 18) and 1195
made by adults (those 18 and older). Figure 6 shows that the percent of adult
visits for asthma in March, September, and October were more than twice that of
the number in July when the percentage of visits was lowest. Among children,
the highest proportion of visits was seen in March and the lowest was in August.
The most notable difference between adults and children was seen in the months

of March, September and October.

400

 

 

350

 

 

300

 

 

 

 

 

 

g 250

I)

5 _.___..L_
a. +Female,
S 200 l
o +Male

n

E

3

z

150

 

 

100 4—

 

50-

 

 

 

0 -
<2 2-5 6-11 12-17 18-34 35-44 45-54 55-64 65-74

 

Age

Figure 7. A comparison of the number of ED asthma visits by age and gender,
2001.

50

A.2.3 Comparison of the Number of ED Asthma Visits by Age and Gender

Because of the strong interaction between age and gender we evaluated both
risk factors together (Figure 7). As expected, boys had a much greater number
of asthma visits than girls, but the gap narrowed as adolescence approached.
After adolescence the number of female visits compared to male visits rose
dramatically and stayed higher throughout the adult years. The greatest
difference in ED asthma visits occurred in the 35-44 year old age group where

the number of female visits was nearly 2 times higher than that of male visits.

 

14

 

 

 

 

 

+ Female ;
—I—— Male i

 

 

 

 

Percent of Visits

 

 

 

 

 

 

JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC
Month

Figure 8. Proportion of ED asthma visits by month and gender, 2001.

51

A.2.4 Comparison of the Monthly Proportion of ED Asthma Visits by
Gender

There were 889 male ED asthma visits and 1162 female visits. Figure 8 shows
that the monthly proportion of ED asthma visits by gender is similar to Figure 4
showing two peaks in the winter and early spring and in the fall in females.
Females had a higher number of Visits than males and had a much more
pronounced peak in the fall compared to males. The resemblance between
Figures 6 and 8 shows that gender and age (defined as child vs. adult) interact,
meaning that the majority of child ED visits for asthma are made by males and

the majority of ED visits for asthma adults are made by females.

14

, /\ .
.../ \ /\
V \

 

 

 

Percent of Repeat Visits
0) oo

 

 

 

O . . . l 1 I . T
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

Month

 

Figure 9. Monthly proportion of repeat ED asthma visits, 2001.

52

A.3 Comparison of the Monthly Proportion of Repeat Asthma ED Visits

The monthly proportion of repeat visits is Shown in Figure 9. A repeat asthma
visit was defined as a second asthma visit within the same calendar month.
March had the highest percentage of repeat visits with over 13% while July had
the lowest with less than 3%. The trend mimicked that of Figure 4, the
percentages of all ED asthma visits, in that there were peaks in both early spring

and fall.

A.4 Comparison of the Monthly Proportion of Hospital Admissions from ED
Asthma Visits

There were 277 hospital admissions from the ED for asthma in 2001. ED asthma
hospital admission remained fairly consistent throughout the year (Figure 10).
This trend was noticed in both Children and adults (Figure 11). The adult
admission rate was at its highest in August when the child admission rate was
lowest. Annual admission rates were 13 percent overall, 11 percent for children,

and 15 percent for adults.

53

Admission Rate

15 7’V\/\ /

 

 

 

 

 

 

 

 

 

10 V L+ Admit Rate“:i
5
O ‘ ‘ fl ' T I > T E
Q- A V‘- V O 3 A 0
3?? (39 Q? ‘3? Q? )0 30 Y9 (96 00 ‘50 0Q/
Month

Figure 10. Hospital admission rate for asthma by month, 2001.

54

 

25

2, It

W V V W |
: +Child Admit;
. Rate '
l
d 3 /\/“ I—l—AdultAdmit
I Rate i
l J
5 V

0 r I . T a I I 1 m
$5“ «‘3’ $9 (3% 3‘5 39‘; 30» e30 9‘8 06‘ e04 6°C)

Month

 

_|
01

 

 

..s
O

 

Percent Admissions

 

 

 

 

 

Figure 11. Hospital admission rate for asthma by month and age, 2001.

A.5 Comparison of ED Visits for Asthma Visits by Day of the Week and
Hospital

In observing the percent of asthma visits by day of the week and hospital (Figure
12) it can be seen that Monday and Sunday had the highest percent of visits,
whereas Friday and Saturday had the least. This trend was seen among the 3
hospitals studied. The only variation from the overall day of the week comparison

came from Butterworth Hospital where the percentage of visits for Monday was

55

less than that of the percentage on Sunday. The other 2 hospitals had opposite

results for this attribute.

 

25

20
. + Butterw orth I
i + Blodgett ‘
I + Gerber

 

 

 

 

1O

 

Percent of Vlslts

 

 

 

 

Mon Tue Wed Thur Fri Sat Sun
Day

Figure 12. Hospital comparison of percent of ED asthma visits by day of the

week, 2001.

B. VALIDATION STUDY

During the 12-month period, August 1, 2000 and July 31, 2001 at Butterworth
Hospital, there were 514 total ED visits by children aged 2-17 years with asthma
(lCD-9 493) as a primary diagnosis from which 150 ED visits were randomly
chosen. There were 838 visits by patients 2-17 years of age who visited
Butterworth hospital between August 1, 2000. and July 31,2001 with a chief
complaint of acute asthma-like respiratory symptoms. Of these visits, 150were

randomly chosen. Table 5 shows frequencies of non-asthma diagnoses

56

occurrence and frequencies of those sampled. The frequencies of the total ‘and

the sampled diagnoses are Similar.

 

lCD-9 Non-asthma Diagnosis Frequency (%) Frequency (%)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Total Sampled
332.9 Otitis Media, unspecified 6.4 3.3
464.4 Croup 15.8 20.7
465.9 Acute Upper Respiratory 34 30.0
Infection of multiple or
unspecified sites
466 Acute Bronchitis or 5.5 8.0
Bronchiolitis
466.19 Acute Bronchiolitis due to 1.2 2.7
other infectious organisms
486 Pneumonia, unspecified 1 1.9 13.3
490 Bronchitis, not specified 2.4 3.3
as acute or chronic
519.1 Other Diseases of the 2.7 1.3
Trachea and Bronchus,
NOS
519.8 Disease of the Respiratory 4.3 2.0
System, not elsewhere
Specified
519.9 Unspecified Disease of 1.0 0
the Respiratory System
786.01 Hyperventilation 0.8 1.3
786.05 Shortness of Breath 1.2 1.3
786.07 Wheezing 3.3 2.0
786.09 Other Dyspnea and 3.1 6.0
Respiratory Abnormalities
786.2 Cough 5.8 4.7
Total 100.0 100.0

 

 

 

 

 

 

Table 5. A comparison of the numbers and frequencies of non-asthma diagnoses
for the total numbers and the validation sample.

57

8.1 Analysis

8.1.1 Sensitivity

The results of the comparison of ED medical records coding with ED physician
billing data shown in Figure 15 found that of the 150 ED visits for asthma, 133
had a physician ED diagnosis that corresponded ED to this diagnosis, leaving 17

discordant cases (false negative results).

Medical Record Diagnosis Code

 

 

 

 

Asthma No asthma Total
Physician Asthma 133 13 146
Billing
Data No Asthma 17 137 154
Diagnosis
Code Total 1 50 1 50 300

 

 

 

 

 

 

These data result in a sensitivity of 0.89.
Sensitivity = 133/150 = 0.89
95% confidence interval (CI) = (0.85 - 0.93)

Figure 13. Results of the comparison of ED medical records coding with ED
physician billing data

8.1.2 Specificity

The results of the comparison of ED medical reCordS with ED physician billing

data (Figure 14) found that of the 150 ED visits for non-asthma, 137 had an

58

physician ED diagnosis that corresponded to this diagnosis. The data Showed
that 91% of the non-asthma controls had ED bills with a diagnosis other than

asthma.

Specificity = 137/150 = 0.91

95% CI = (0.88 - 0.94)
8.1.3 Positive Predictive Value
The probability that ED physician billing data had a primary diagnosis of asthma
given the corresponding medical record had a primary diagnosis of asthma was

91%.

PPV =133/146 = 0.91

95% CI = (0.88 — 0.94)

8.1.4 Measurement of agreement (Kappa)

The probability beyond chance that the medical record codes and the billing

codes agreed was 0.80.

Kappa = ((137+133)/100 - 0.5)/(1-0.50) = 0.80

59

8.1.5 False Negatives

In an attempt to explain the false negative results, we performed an independent
evaluation of the medical records for these 17 subjects. We examined date, age,
history of asthma, and the billing diagnosis code. Based on these attributes, we
assigned the diagnosis of asthma as probable, possible, or not asthma. If an
alternative respiratory disease was clearly present in the chart (e.g. upper
respiratory infection, pneumonia) which should have been the primary diagnosis
instead of asthma, then these were labeled ‘not asthma’ in Table 6. A label of
‘possible’ was assigned if the patient may have had asthma, but the medical
records did not indicate that the case met the 1998 CSTE guidelines 3 for definite
asthma (i.e. the cases had episodic symptoms of airflow obstruction that could be
attributed to another diagnosis). Possible was assigned to 1 of the cases,
indicating that this case could have been indicative of either the physician or
medical record diagnosis. A case was labeled ‘probable’ if the medical record
gave no indication that the case was not asthma (i.e. the case had serious
asthma-like symptoms and no secondary diagnosis of a respiratory infection). If
a patient had a history of asthma and no secondary diagnosis of a respiratory
infection, a case was more likely to be a probable or possible case of asthma. A
probable case was assigned to 9 of the cases, meaning that after careful perusal
of the medical record, these cases agreed with the medical record diagnosis of
asthma. An lCD-9 code of 786 was given to 15 of the 17 cases with a diagnosis
of asthma on the medical record, but non-asthma on the billing record. By these

standards, of these 17 cases, 9 were definite asthma that the billing data missed,

60

 

but in 7 of the 17 cases, the diagnosis was not asthma, so the billing data was

 

 

correct.
Possible Explanation for Asthma
Inconsistency History Diagnosis
Age Billing Between Billing and Medical of Consistent with
Code Records Asthma Medical Record
Hospitalized - no follow-up
2 786.09 information Yes Probable
3 786.07 Possible pneumonia Yes Not Asthma
5 786.09 lNone Yes Probable
10 519.1 Acute bronchospasm Yes Probable
3 786.09 RAD secondary Dx - viral illness No Not Asthma
11 786.09 Acute RAD Yes Probable
INew asthma, secondary Dx -
7 786.09 dyspnea Yes Probable
13 786.09 Asthma exac Yes Probable
2 786.09 Acute 493 Yes Probable
14 786.09 RAD No Probable
13 786.09 Secondary Dx - viral URI Yes Not Asthma
7 486 Secondary Dx — pneumonia Yes Not Asthma
3 786.09 Dx=diff breathing, hospitalized Yes Not Asthma
3 786.07 RAD Yes Probable
Diff breath, secondary Dx - ’
3 786.09 tonsilitis Yes Not Asthma
Acute asthma, later changed to
4 786.09 pneumonia Yes Not Asthma
9 786.07 Possibly not severe enough Yes Possible

 

 

 

 

 

 

 

Table 6. A list of false negative results (cases with a diagnosis of asthma on
medical chart, but non-asthma on physician billing record) and their
characteristics

8.1.6 False Positives

Of the 150 medical records where the primary diagnosis was something other
than asthma, 137 of the ED bills were consistent with a diagnosis other than

asthma, leaving 13 discordant controls (false positive results). Table 7 shows the

61

results of an independent assessment performed in an attempt to explain
whether or not these cases were consistent with the medical record. One of the
cases was first ruled asthma, but a radiology report received after the visit
overruled this diagnosis, which was then Changed to pneumonia. This case was

dropped as a false positive result.

In 9 of the 12 cases remaining cases, where the medical record diagnosis was
not asthma, the primary diagnosis was a type of infection. In these cases, the
billing record was wrong, since the infection should always be coded as the
primary diagnosis. In the other 3 cases where the primary diagnosis was not an
infection, the medical record diagnosis was RAD. These cases could have

qualified as primary diagnoses of asthma, making the billing record correct.

62

 

 

 

 

Secondary Non-Asthma
Possible History Diagnosis Diagnosis
Age Final Diagnosis Explanation of of Consistent with
in Medical Record Asthma Asthma lMedical Record
ACUTE URI Symptoms
4 46” NOS like 493 No Yes Yes
ACUTE URI No exposure
3 465'9 NOS URI No Yes Yes
ACUTE URI Slight
8 465'9 NOS wheeze Yes Yes Yes
ACUTE URI Peak flow
11 ”659 NOS 125-300 Yes Yes Yes
ACUTE URI Peak flow
15 465'9 NOS improved Yes Yes Yes
ACUTE URI Peak flow
6 465'9 NOS 75-150 No Yes Yes
More serious
2 78“” RAD than RAD No No No
Slight
7 519'8 RAD wheeze No No No
ACUTE URI Slight
3 465'9 NOS wheeze Yes Yes Yes
Original
8 486 gEELAAIGEDST/IIA’ diagnosis of
asthma Yes N/A Yes
Admitted 3x
2 465.9 QgLéTE UR' in pastfor
493 Yes Yes Yes
Possible
3 786'07 RAD pneumonia Yes No No
ACUTE Slight
3 ”660 BRONCHITIS wheeze No No Yes

 

 

 

 

 

 

Table 7. A list of false positive results (controls with a diagnosis of not asthma on
medical chart, but asthma on physician billing record) and their characteristics.

63

 

 

CHAPTER 4 - DISCUSSION

A. DESCRIPTIVE ANALYSIS OF THE EMERGENCY DEPARTMENT ASTHMA

SURVEILLANCE SYSTEM

The central finding in the descriptive analysis of the ED asthma surveillance
system concerned the monthly variation in overall ED asthma visits. High peaks
in asthma ED visits are usually seen in spring and mid fall, which is broadly
consistent with our findings. However, the highest proportion of visits typically

occurs in early to mid fall (late September-October), ‘0 11 12

whereas the highest
proportion of visits in EDASS occurred in March. An explanation for this event
may come from the fact that the fall months of 2001 in Michigan were very mild
compared to previous years. The suburban hospital, Blodgett, actually mimicked
the typical pattern in asthma ED visits, but Since the urban hospital, Butterworth,
had a significantly greater number of visits than the 2 other hospitals, it was the

driving force behind the overall trends. The rural hospital, Gerber Memorial had

peaks in Spring and Fall that were very similar to each other.

The comparison of the monthly variation in ED asthma visits and all ED visits in
2001 Showed very little difference in the month-to-month percentage of all Visits.
January was the only month that was slightly inconsistent with this trend with

overall ED visits being higher than the rest of the year. The monthly comparison
10 11 12

of ED asthma visits was fairly consistent with the expected peaks in Spring

64

and fall and the valleys in summer and winter as was seen by many researchers
including, Dales RE et al. in Ottawa, Canada in 1993-1997. 61 The percentage
of ED asthma visits was much higher than the percentage of overall visits for
March and September-October, while ED asthma visits were much lower than

the overall percentage during June-August.

Our EDASS findings involving age (child versus adult) were somewhat unusual
as compared with other asthma surveillance data. Children tended to have
higher percentages of asthma events than adults in spring and fall with the
highest peak in the fall, whereas adults have been found to have similar
percentages of events in spring and fall. EDASS showed child and adult asthma
visit curves to be fairly similar; the main difference between them was seen in the
fall. In contrast to many researchers including, Fleming et al., 21 adults in our
study had the characteristically high asthma peak during Spring that is usually
more notable in children, whereas the children’s fall peak was much smaller
comparatively to the adults’ and to their spring peak. From the age, gender, and
age/gender comparisons, it was clear that the relationship of asthma related ED
visits with gender was confounded by age. Females have a much higher
percentage of asthma visits than males after adolescence, so our adult data
closely resembled the trend for females. An explanation for the difference in the
number of ED visits for asthma in males and females over the age of 65 comes

I. 62

from researchers Dodge eta who observed that men over the age of 65 are

more often diagnosed with emphysema as compared to asthma given identical

65

clinical presentation consistent with asthma in females over 65. This could
account for the higher number of ED asthma visits that was noted for females
over 65 in EDASS or it could truly represent a lower incidence of asthma in

males than in females.

Logically, the relative proportion of monthly repeat ED asthma visits would be the
Similar to the proportion of monthly ED asthma visits. Gerber Memorial Hospital
showed an opposite pattern in that repeat visits were highest in August when
overall visits were at their lowest, whereas ED visits for asthma were high in
September when overall ED asthma visits to Gerber were at their lowest (not
shown). Once again, very little can be concluded from this variability in repeat
visits as the number of repeat ED asthma visits was very small. At Blodgett
Hospital there were also months where repeat visits and all asthma visits to that
hospital did not concur. This was seen in February where the percentage of ED
asthma visits was low, but the percentage of repeat visits was high. Butterworth
Hospital’s repeat visits varied in a similar way to the percentage of ED asthma
visits. It follows that with a greater percentage of ED asthma visits to a hospital,

there was less variation between ED asthma visits and repeat ED asthma visits.

Numbers of ED asthma visits and hospital admissions for asthma from EDASS
followed extraordinarily similar patterns throughout 2001. This result concurs
with the result that Fleming et al. found during a year-long study of hospital

admissions in UK. Even when broken down by hospital, the number of ED

66

asthma visits shown by EDASS was proportional to that of the number of
admissions for asthma except a Slight variation during the month of July at
Gerber Memorial Hospital when the number of admissions rose slightly while the
number of ED visits was decreasing. It is difficult to conclude a great deal from

Gerber Memorial Hospital due to its very small numbers of ED asthma visits.

Data looking at the day of the week on which visits took place are a relatively
new frontier. EDASS detected very little variation in the comparison of the day of
the week of an ED asthma visit. The slight variation that was seen, though,
demonstrated that Sunday and Monday had the highest number of ED asthma
visits and Friday and Saturday had the least. All of the hospitals Shared this
trend. A possible explanation for the differences seen between the numbers of
Friday/Saturday visits and Sunday/Monday visits may be that Since Friday and
Saturday are normally considered days at the end of the work or school week,
and may involve less work or school stress. Sunday and Monday are usually
considered days at the beginning of the work or school week when work or
school stress is higher. This stress may cause anxiety, leading to an asthma
event. 1” lnforrnation that could be useful and easily provided by the ED physician
billing company is a breakdown of the day of the week of each visit by age and
gender. This information could help us to understand the implications behind the
day of the week on which the asthma event occurred such as if stress is a major
factor in severe asthma exacerbations and if it is gender or age related. Another

variable could provide some very useful information is the type of health

67

insurance a patient has (i.e. private, Medicaire, Medicaid, HMO, or no insurance).
As with zip code, the type of insurance held by the patient could provide further

insight into the role of socioeconomic status with visits to the ED for asthma.
B. EMERGENCY DEPARTMENT SURVEILLANCE SYSTEM EVALUATION

The evaluation of the ED asthma surveillance system was performed using the
CDC’s ‘Guidelines for the Evaluation of Public Health Surveillance Systems.“ 59
Because only one year of complete data were available, the experience was

limited. A discussion of the evaluation of the 10 characteristics suggested for

use by the CDC is presented below.

8.1 Usefulness

The ED asthma surveillance system at its current stage is unable to estimate the
magnitude of morbidity and mortality related to asthma due to its not being
population-based. If the system were established statewide, the data collected
may be able to estimate the magnitude of morbidity in relation to ED use, but
would still be unable to address the magnitude of mortality, since this information
is not available from the physician billing records. To date, the data obtained
from the system have only been able to develop baseline results, being in place
for less than 2 years. With more time, the system is expected tothave the ability
to detect trends signaling changes in the occurrence of asthma. The system has

not yet been tested to investigate its ability to permit the assessment of the effect

68

of prevention and control programs, but it is expected that it will be sufficiently
responsive to be able to perform this task. With this capability, the system will
also have the ability to stimulate research intended to lead to the prevention and
control of asthma events, through the improvement of clinical, behavioral, policy,
and environmental practices. Considering what the surveillance system is able
to do thus far and what it has the capacity to achieve in the future, it has the

potential to be a very useful system.

8.2 Simplicity

The EDASS is simple, easy to operate, and provides the data necessary to
establish that an asthma episode has occurred. A case is easily ascertained as
the case definition is straightforward: anyone who was declared by the ED _‘
physician billing company to have asthma. Currently, Michigan State University’s
(MSU) department of epidemiology is the only organization receiving these case
reports. The surveillance system is integrated with the ED physician billing
company’s system, which is the only data-reporting source. The data are stored
on a computer hard drive and on a floppy disk, making them easy to manage.
The data are analyzed using Microsoft Excel, to determine monthly absolute
numbers of visits and visit frequencies for hospital comparisons. The data have
yet to be disseminated. Staff training on how to operate and maintain the
system, manage the data and perform data analysis are yet to be determined.
The time Spent maintaining the surveillance system is minimal. There are no .

multiple levels of reporting, Special or follow-up laboratory tests, and asthma

69

 

 

cases are not investigated. Because data for the EDASS are generated from
billing records, it is always available. In consideration of what attributes involving
simplicity the EDASS has, it meets the CDC’s standards for a public health

surveillance system.

3.3 Flexibility

The EDASS has shown the ability to adapt to changing information needs with
little additional time and no additional personnel or funds. Changes in age group
distributions were easily made and new information such as zip codes and type
of insurance can easily be added. The system can also accommodate other
health-related events as was witnessed during the validation study when we
required information on ED Visits for events other than asthma, but may have
been confused with asthma. In this sense, according to CDC’S standards, the

EDASS can be deemed flexible.

8.4 Data Quality

In reviewing the completeness and validity of the data gathered from the EDASS,
the surveillance system has been shown to have data of reasonably high quality,
with high sensitivity, specificity, and positive predictive values, as was seen in the

results of the Special validity study that was performed.

70

 

 

8.5 Acceptability

The Michigan Department of Community Health, the Michigan Asthma Task
Force Committee, and Spectrum Health have accepted the EDASS. Through
these channels, the dissemination of the aggregate data collected from the
EDASS will become part of a website concerning asthma in Michigan. It is
unknown what the acceptability of the system would be elsewhere, either at other l-'
hospital medical records departments, hospital billing departments, or outside

billing companies like MMS.

 

The ease and cost of reporting these data at a statewide level have yet to be
determined. Because the system has the ability to protect privacy since data are
collected in aggregate form and are not linked to any confidential information,

EDASS Should have a greater chance for acceptability.

8.6 Sensitivity

Sensitivity of the EDASS can be referred to on two levels. The first involves the
quality of case diagnosis in the ED and case reporting by the billing company,
while the other involves the ability to monitor changes in the number of cases
over time.

The EDASS has reasonably high probability for identifying an asthma visit from
the ED physician-billing data when the final diagnosis in the medical record was

asthma (sensitivity = 0.89), Showing a reasonably high quality of case reporting.

71

As was seen in the results section, the EDASS is very capable of monitoring

changes in the number of visits for asthma over time.

8.7 Positive Predictive Value

Having the ability to have accurate reporting of asthma cases that actually have
asthma is of high importance. A low positive predictive value could report trends
in ED asthma Visits that may not exist, being only an artifact of the surveillance
system. These false reports could then lead to unnecessary research to find an
Intervention for an asthma problem that is nonexistent. The EDASS was able to
collect a high proportion of asthma cases that actually have asthma as was

demonstrated in the results of validation study (PPV = .91 ).

8.8 Representativeness

The ability to accurately describe the occurrence of asthma events over time and
their distribution in the population by place and person is a feature that is of great
significance in ED asthma surveillance. Because the EDASS only monitored 3
hospitals as opposed to all hospitals in the area, it can only describe the
occurrence of asthma over time (i.e. the monthly trends), but was not able to
show the distribution of asthma in the population by place and person. Therefore
it is not generalizable Since we do not have complete knowledge concerning the
proportion of the Grand Rapids area population that visited these three EDS.

Once a statewide system is in place, the system will be more representative.

72

 

 

8.9 Timeliness

The ED asthma surveillance system’s speed varies by step. A 6 to 7 day time
period was generally needed from the time of the ED Visit to the time when the
hospital’s medical records department completed lCD-9 coding. It takes 2-3
days to send the hospital-coded record to the ED physician billing, with 55 days
required between the receipt of the hospital-coded medical record and
completion of its physician billing coding. The data tables can then be quickly

and easily populated. Sending the data via e-mail to MSU requires a few minutes

 

at the most. The total time from an ED visit for asthma to the time the data from

that visit are ready to be sent to MSU could be between 8 and 44 days. 5
However, for routine analysis, we receive the data at 3-month intervals. The
speed between steps is not a key factor for the EDASS Since asthma is not a
communicable disease and our reasons for having the system do not require a
great deal of rapidity. The reporting of data at 3-month intervals is adequate for

our needs therefore the timeliness of the EDASS is acceptable.
8.10 Stability

In reviewing the stability of the ED asthma surveillance system, it is best to view
it as two components: reliability and availability. The system is highly reliable in
that it is able to collect, manage, and provide data properly without failure.

Occasionally, the number of visits recorded in the age and gender tables do not

 

match the number of Visits documented in the day of the week tables; a simple

73

data entry error. When this occurs a thorough search of the data is necessary to

locate any possible errors.

A stable performance of any surveillance system is crucial to its viability,
regardless of the adverse event. A lack of dedicated resources such as too few
personnel or a lack of funds can be disastrous to a system’s reliability and
availability. The EDASS system has always been operational when it is needed,
but has yet to be fully tested. For our purposes and objectives, the EDASS has

been stable.

The guidelines produced by the CDC on evaluating public health surveillance
systems, provided a thorough and insightful look into the manner in which the
operation of our surveillance system excels and fails to perform as it should. For
many of the characteristics covered by the guidelines, our system has yet to be
tested, but overall, the ED asthma surveillance system has worked well for our

needs.

C. RELIABILTY AND VALIDITY OF THE ED ASTHMA SURVEILLANCE
SYSTEM

In the past, the use of billing records as a source of data for surveillance systems
has been shown to be a source of invalid data, significantly underestimating the
number of asthma visits in some EDS. 57 A validation study of an ED asthma

surveillance system in New York used 12 hospitals in New York City. 49 Their

74

 

 

results varied widely between hospitals, from 41 to 96 percent sensitivity.
However, we decided to use this form of data collection through MMS, an ED
physician billing company, for an ED asthma surveillance system. We measured
the system’s accuracy and reliability through a special study from data that were
also provided by MMS as well as Buttenlvorth Hospital medical records
department. The data gathered during the validation study from 300 hospital
medical records and their respecting ED physician bills suggested that the use of
ED physician billing as a source of data in the EDASS was a valid source for the
ascertainment of asthma cases. The sensitivity of the system was very high at
89 percent, as were the specificity and PPV (both 91 percent). The reliability of
the system with a Kappa statistic of 0.80, was fairly good, although we would like
to have an even higher level of agreement between the hospital medical record
and the physician billing codes. In performing the validation study retrospectively,
we were able to obtain data without the knowledge of those who code the .
diagnosis for ED physician billing company, thereby avoiding the introduction of

bias.

Because symptoms of asthma vary widely, clinical judgment is required in the
diagnosis of asthma. With the introduction of clinical judgment comes the
potential for bias. ED doctors and ED physician billing coders do not always
agree in the diagnosis of asthma or in the diagnosis of something that resembles
asthma. The agreement beyond chance or reliability in the NewYork City study

was between 0.55 and 0.61, showing agreement not much higher than thatdue

75

to chance alone. 49 The degree of agreement of the medical record code and the
billing code beyond chance in our study was 0.80, showing that the reliability of

our data was very good.

In our study we used only the primary diagnosis given by the ED physician as our
‘gold standard’ definition of an asthma case. Through communication with
researchers at the New York City Department of Health 49 and another group in

Washington State it was Ieamed that they both used this same definition.

A limitation to this study is that the validation of the ED asthma surveillance
system involved only one of the 3 hospitals. Butterworth Hospital was used
because it is the largest of the 3 hospitals and generates the most data. A
limitation to this study was that the assumption was made that validation results
from Gerber and Blodgett Hospitals would be Similar to that of Butterworth
Hospital partly because they employ the same company to perform their ED
physician billing. Also, Blodgett Hospital is operated by Spectrum Health, the
same company that manages Butterworth Hospital. The assumption that results
from all 3 hospitals would be similar may be unsubstantiated for a few reasons.
The staff employed at each of these hospitals may come from a variety of
backgrounds, Since the 3 hospitals are in 3 different areas: urban, suburban, and
rural. The differences in emploYees’ working environments may influence their
abilities to perform their jobs with consistency. For example, a larger hospital

such as Butterwoth may require a greater amount of work per employee than a

76

 

 

smaller hospital such as Gerber, possibly causing these employees to have a
higher probability of making errors. Different hospitals may have different 5'
methods or approaches to coding, eg. using the lCD-9 versus the lCD-10 or they
may require different levels of education or experience for their coding 1
employees. Therefore, although Butterworth Hospital provided over 50 percent
of the data used in the EDASS, they were not the sole source of data. If doctors
in other communities use different criteria for the diagnosis of asthma, it can be
expected that the sensitivity, specificity, and PPV may differ from that seen in this
pilot study. Gerber Memorial and Blodgett Hospitals should have their billing
records validated in the future since these results and those from New York have
shown that sensitivity can be variable between areas (hospitals, cities, states).
However, the occurrence of differences in sensitivity, Specificity, and PPV across
communities may lead to additional efforts for a more uniform approach to .

diagnosis.

Other major limitations of the system are its inability to calculate event rates, the
magnitude of mortality and morbidity related to asthma, or Show the distribution
of asthma in the population by place and person, as it is not population-based.
The system is not generalizable as complete knowledge concerning the
proportion of the Grand Rapids area population that visited the 3 EDS. In
addition, there is not widespread support for the creating a mandatory statewide
reporting system and a voluntary reporting system is not supported by the

Michigan Hospital and Healthcare Association. Because the State of Michigan

77

does not require medical facilities to record information on race, we were unable
to obtain data on this variable. AS has been previously noted, race has been
shown to have disproportionate affects on minorities. 5 If this data were
available, it would be interesting to track trends to see if we could replicate this
disproportionality, thereby proposing studies to possibly aid in better healthcare

utilization for minorities.

78

CHAPTER 5 - CONCLUSIONS

The pilot ED asthma surveillance system in the Grand Rapids, Michigan area
was designed to measure the ability of an ED physician billing data system to
track asthma visits in emergency rooms and to demonstrate that a system of this

type could produce much needed data elsewhere.

Through a review of the descriptive analysis of the data from January 1, 2001
through December 31, 2001, an evaluation of the EDASS by CDC’S standards,
and a special validation study, it has been demonstrated that the EDASS is a

viable source of ED asthma data.

Since the analysis of this year of data was performed, a zip code variable has
been added in order to identify the area of residency of the visiting patient. This
new variable may help provide some insight into the socioeconomic status of the
patients who are visiting the ED with asthma and may also lead to the potential of
calculating Visit rates by zip code. This information could aid in the
Characterization of the asthma population who use EDS and, in turn, could help
identify ‘at-risk’ populations for Specific interventions. It is believed that those
asthma patients of lower socioeconomic status have unmet needs, under-using

necessary services and failing to achieve optimal control over their disease. 63 64

79

In reviewing the characteristics of the ED asthma surveillance system using the
CDC’S ‘Guidelines for the Evaluation of Public Health Surveillance Systems’, it
was found that the system is useful, simple, flexible, complete, and has very
good sensitivity and positive predictive value. It is timely for our needs, although
a billing-based system such as this would be too slow for a system involving the
surveillance of infectious diseases. In an attempt to create a statewide ED
surveillance system based on Similar billing data through MMS, there has been a
hindrance due to the unwillingness of the Michigan Hospital and Healthcare
Association to support the system, but otherwise there is willingness in terms of
this billing company, the Michigan Department of Community Health, the
Michigan Asthma Task Force Committee, and Spectrum Health to participate in

the surveillance program.

The system has not been tested thoroughly enough as to be deemed reliable and
available or stable. The one major characteristic this ED asthma surveillance
system lacks at this time is representativeness, as it is not generalizable to a

wider population.

The data produced by the validity study showed a high level of accuracy and
reliability. The sensitivity of the ED asthma surveillance system at Butterworth
Hospital was 89 percent, the specificity and PPV were 91 percent, and the

agreement of the 2 coding systems beyond chance was 0.80.

80

ED asthma surveillance is a tool that can help identify trends in ED asthma Visits
as well as aid in the planning and assessment of asthma interventions. This tool
can also provide insights into the cause and pathogenesis of asthma and
augment research in that it can disseminate population-based data in a timely
manner, it is ongoing, and it could be linked to those involved in prevention and

control activities such as the Michigan Asthma Task Force Committee.

Many states have developed or are in the act of developing statewide ED
surveillance systems as is encouraged by the CDC. We believe that the EDASS
received a fairly good evaluation rating for how it has been used thus far,
supporting strong consideration for a statewide system. Although, implementing
a surveillance system such as this statewide, may be difficult as many ED

physician billing companies may consider data reporting to be a burden.

81

REFERENCES

1 Centers for Disease Control, National Center for Health Statistics, National Health Interview
Survey, 2004, http://www.cdc.gov/asthma/nhis/04/pdfs/table3-1.pdf.

2 National Institute of Health, National Heart, Blood and Lung Institute. Guidelines for the
Diagnosis and Management of Asthma, July 1997; 97- 4051.

3 Council of State and Territorial Epidemiologists. Environmental and Chronic Disease
Committees. Council of State and Territorial Epidemiologists Position Statement # EC/CD 1.
Council of State and Territorial Epidemiologists Annual Meeting, 1998.

‘ Reese S et al. Self-reported Asthma Prevalence Among Adults—United States, 2000. 17 Aug
2001, MMWR; 50(32): 682-686.

5 Mannino DM, Homa DM, Akinbami LJ, Moorman JE, Gwynn C, Redd SC. Surveillance for
Asthma - United States, 1980-1999. MMWR, March 29, 2002;51(SS01):1-13.

6 Camargo CA, Richardson LD. Chapter 5, “Epidemiology of Asthma,’ In: Brenner 8E, ed.
Emergency Asthma. New York, NY: Marcel Dekker, Inc. 1999:59-80.

7 Centers for Disease Control. Asthma Mortality and Hospitalization Among Children and Young
Adults—United States, 1980-1993. 1996 MMWR; 45: 350-353.

8 National Center for Health Statistics, Health-E Stats, New Asthma Estimates: Tracking
Prevalence, Health Care, and Mortality.
http:l/www.cdc.qov/nchs/products/puMubd/hestatS/asthmafisthmaEhtm.

9 Mannino DM, Homa DM, Pertowski CA, et al Surveillance for Asthma- United States, 1960-
1995. MMWR, 1998; 47.1 -27.

1° Ertle AR, London MR. Insights into Asthma Prevalence in Oregon. Journal of Asthma, 1998;
35:281-289.

1‘ Hartert TV, Peebles RS. Epidemiology of Asthma: The Year in Review. Current Opinion in
Pulmonary Medicine, 2000;6z4-9.

’2 Anderson HR, Butland 8K, Strachan DP. Trends in prevalence and severity of Childhood
asthma. June 1994, Brithish Medical Journal; 308: 1600-1604.

13 Miller JE. Predictors of asthma in young children: does reporting source affect our
conclusions? American Journal of Epidemiology! 2001 Aug 1;154(3):245-50.

1‘ Sears MR, Burrows B, Flannery EM, Herbison GP, Holdaway MD. Atopy in childhood. I.
Gender and allergen related risks for development of hay fever and asthma. Clinical Exp Allergy
1993 Nov ,23(1 1): 941-8.

‘5 Montefort S, Lenicker HM, Caruna S, Agius Muscat H. Asthma, rhinitis and eczema in Maltese
13-15 year-old schoolchildren - prevalence, severity and associated factors [ISAAC].
International Study of Asthma and Allergies in Childhood. Clinical Exp Allergy 1998
Sep;28(9):1089-99.

82

‘6 Sunyer J, Anto JM, Kogevinas M, Barcelo MA, Soriano J'B, Tobias A, Muniozguren N,
Martinez-Moratalla J, Payo F, Maldonado JA. Risk factors for asthma in young adults. Spanish
Group of the European Community Respiratory Health Survey. European Respiratory Journal,
1997 Nov;10(11):2490-4.

‘7 KimbeII-Dunn M, Pearce N, Beasley R. Seasonal variation in asthma hospitalizations and
death rates in New Zealand. Respirology, 2000; 5: 241 -246.

‘8 Tovt-Korshynska MI, Dew MA, Chopey IV, Spivak MY, Lemko IS. Gender differences in
psychological distress in adults with asthma. Journal of Psychosom Res, 2001 Nov;51(5):629-37.

19Brenner BE. Where Might We Go: Trends in Emergency Asthma Research. Chapter 2,
‘Epidemiology of Asthma’, In: Brenner BE, ed. Emergency Asthma. New York, NY: Marcel
Dekker, Inc.; 1999:33-36.

2° Riccioni G, Di Stefano F, De Benedictis M, Verna N, Paolini F, Di Sciascio MB, Della Vecchia
R, et al. Seasonal Variability of Non-Specific Bronchial Responsiveness in Asthmatic Patients
with Allergy to House Dust Mites. Allergy and Asthma, Jan 2001; 22(1): 5-9.

21 Fleming DM, Cross KW, Sunderland R, Ross AM. Comparison of the seasonal patterns of
asthma identified in general practitioner episodes, hospital admissions, and deaths. Thorax,
2000; 55: 662-665.

22 Harju T, Keistinen T, Kivela SL. Seasonal variation in childhood asthma hospitalizations in
Finland, 1972-1992. European Journal of Pediatrics, 1997; 156: 436-439.

23 Corn B, Hamrung G, Ellis A, Kalb T, Sperber K. Patterns of Asthma Death and Near-Death in
an Inner-City Tertiary Care Teaching Hospital. Journal of Asthma, 1995; 32(6):405-12.

2‘ Carr W, Zeitel L, Weiss K. Variation in Asthma Hospitalizations and Death in New York City.
American Journal of Public Health, 1992; 82:59-65.

25 Murray MD, Stang P, Tierney WM. Health Care Use by Inner-city Patients with Asthma. Journal
of Clinical Epidemiology, 1997; 50(2): 167-174. '

26 Watson JP, Cowen P, Lewis RA. The relationship between asthma admission rates, routes of
admission, and socioeconomic deprivation. European Respiratory Journal, 1996; 9: 2087-2093.

27 Fauroux B, Sampil M, Quénel P, Lemoullec Y. Ozone: A trigger for hospital pediatric asthma
emergency room visits. Pediatric Pulmono/ogy, 2000; 30: 41-46.

28 Stieb DM, Burnett RT, Beveridge RC, Brook JR. Association between ozone and asthma
emergency department visits in Saint John, New Brunswick, Canada. Environmental Health
Perspectives, Dec 1996; 104(12): 1354-1360.

29 Norris G, YoungPong SN, Koenig JQ, Larson TV, Sheppard L, Stout JW. An association
between fine particles and asthma emergency department visits for children in Seattle.
Environmental Health Perspectives, June 1999; 107: 489-493.

3° National Heart, Lung, and Blood Institute. Guidelines for the Diagnosis and Management of
Asthma, Expert Panel Report 2: 1-148.

3‘ Centers for Disease Control. Comprehensive Plan for Epidemiologic Surveillance. Centers for
Disease Control: Atlanta, Georgia, 1986.

83

32 Model D. Preventable factors and death certification in deaths due to asthma. Respiratory
Medicine, 1995; 89: 21-25.

33 Sears MR, Rea HH, de Boer G, Beaglehole R, Gillies AJD, Holst PE, O'Donnell TV, Rothwell
RPG. Accuracy of certification of deaths due to asthma: a national study. American Journal of
Epidemiology. 1986; 124(6): 1004-101 1 .

3‘ Hills EA, Somner AR, Stewart CJ, Adelstein AM, Clarke C, Clark 'TJH, Dunhill MS, Johnson AJ,
Nunn AJ, Stableforth DE, Whitfield AGW. Accuracy of death certificates in bronchial asthma.
Thorax, 1984; 39: 505-509.

35 Model D. Preventable factors and death certification in death due to asthma. Respiratory
Medicine, 1 995;89:21-25.

36 National Center for Health Statistics. National Hospital Discharge Survey,
http://www.cdc.gov/nchs/aboutlmajor/hdasd/nhdses.htm.

37 K. Wilcox, Michigan Department of Health, personal communication, 1997.

 

3‘8 Fiore BJ, Olson JA, Hanrahan LP, Anderson HA. Asthma hospitalizations in Wisconsin: A
missed opportunity for prevention. Wisconsin Medical Journal, 2000; 99(8):52-6.

39 American Lung Association's Epidemiology & Statistics Unit Research Program Services.
Trends in Asthma Morbidity and Mortality, 2005. http:l/www.kintera.org/atf/cf/%7B7A8D42C2-
FCCA-4604-8ADE-7FSD5E762256%7D/ASTHMA1 .PDF.

‘0 Beard CM, Yunginger JW, Reed CE, O’Connell EJ, Silverstein MD. Interobserver variability in
medical record review: an epidemiological study of asthma. Journal of Clinical Asthma, 1992;
45(9): 1013-1020.

‘1 Kashnear TM. Agreement Between Administrative Files and Written Medical Records: a case
of the Department of Veterans Affairs. 1998, Medical Care; 36(9): 1324-1336.

42 Wang J, Visness CM, Sampson HA. Food allergen sensitization in inner-city children with
asthma. Journalof Allergy and Clinical lmmunoogyl. 2005;1 15(5):1076—80.

43 Eisner MD, Klein J, Hammond SK, Koren G, Lactao G, lribarren, C. Directly measured second
hand smoke exposure and asthma health outcomes. Thorax, 2005; 60:814-821.

‘4 Gunderson EKE, Garland CF, Gorham ED. Health Surveillance for asthma in the US Navy:
Experience of 9,185,484 person-years. AEP, 2005;15(4):310-315.

‘5 National Center for Chronic Disease Prevention and Health Promotion, Behavioral Risk Factor
Surveillance System. http://www.ch.qov/brfss/#about BRFSS.

‘6 Centers for Disease Control and Prevention. (2002a). Asthma prevalence, health care use, and
mortality 2000-2001. Hyattsville, MD:Centers for Disease Control and Prevention.

‘7 Rance KS, Trent CA. Profile of a primary care practice asthma program: Improved patient
outcomes in a high-risk population. Journal of Pediatric Health Care, 2005;19(1):25-32.

‘8 Centers for Disease Control. NHAMCS Description. National Center for Health Statistics
Hyattsville, MD, 2005.

84

‘9 From a presentation of the New York City Department of Health.

5° Richardson LD. Access, the Emergency Room, and Asthma. Chapter 37, ‘Epidemiology of
Asthma', In: Brenner BE, ed. Emergency Asthma. New York, NY: Marcel Dekker, Inc.; 1999:33-
36.

51 Aldrich T, Griffith J. Surveillance activities in disease and exposure activities. Chapter 5,
“Developing a surveillance program,’ In: Aldrich T, Griffith J, ed. Environmental Epidemiology and
Risk Assessment. New York, NY: Van Nostrand Reinhold; 1993:85-86.

52 Larsen GL. Differences between adult and childhood asthma. Journal of Allergy and Clinical
Immunology. Sept 2000; 106(3): 8153-8157.

53 Jablonski RA. Discovering asthma in the older adult, The Nurse Practitioner, March 2000;
25(3):14-39.

54 Chen JC, Mannino DM. Worldwide epidemiology of chronic obstructive pulmonary disease.
Current Opinion in Pulmonary Medicine, 1999;5z93-99.

55 Wolle JM, Cwi J. Physicians' prevention-related practice behaviors in treating adult patients
with asthma: results of a national survey. Journal of Asthma, 1995;32(2):131-140.

56 Grunfeld A, Beveridge RC, Berkoitz J, Fitzgerald JM. Management of acute asthma in
Canada: an assessment of emergency physician behavior. Journal of Emergency Medicine,
1997;15(4):547-56.

5’ WoIfden LL, Diette GB, Krishnan JA, Skinner EA, Steinwachs DM, Wu AW. Lower physician
estimate of underlying asthma severity leads to undertreatment. Archives of Internal Medicine,
2003; 163:231-236.

58 Davis D, Thomppson O’Brien MA, Freemantle N, Wolf F, Mazmanian P, Taylor-Vaisey A. Do
conferences, workshops, rounds, and other continuing education activities change physician
behavior or health care outcomes? Journal of American Medicine, 1999; 282:867-874.

59 Centers for Disease Control, US Department of Health and Human Services. Updated
Guidelines for Evaluating Public Health Surveillance Systems. MMWR, 27 July 2001; 50(RR-
13)I1-36.

6° Obuchowski NA. Sample size calculations in studies of test accuracy. Statistical Methods in
Medical Research, 1 998:7:371 -92

61 Dales RL, Cakmak S, Burnett RT, Judek S, Coates F, Brook JR. Influence of ambient fungal
spores on emergency department for asthma to a regional children’s hospital. American Journal
of Respiratory and Critical Care Medicine, 2000; 162(6):2087-2090.

62 Dodge R, Cline MG, Burrows B. Comparison of asthma, emphysema, and chronic bronchitis in
a general population sample. American Review of Respiratory Disease, 1986; 133:981-6.

‘53 Fossett J, Perloff J. The “New” Health Reform and Access to Health Care: The problem of the
inner city, Washington, DC: The Kaiser Commission on the Future of Medicaid, 1995.

64 Austin R. Asthma and the unmet needs of minorities. In Bernstein E and Bernstein J, eds.,

Case Studies in Emergency Medicine and the Health of the Public, Sundbury, MA: Jones and
Bartlett Publishers, 1996, 152-157.

85

 

.1

l"WillIf“INTITll'almll

302 8450524

   

 

. A34“;—

. nvfi."‘ ' ""