' 3| '.'- I-Iv
1r ‘3 #1th

If.

  
   

 

.513 II 1I‘. ' ; 1&3.“ .i-:.IJ_I331

   
     
  
  

   
  
 
  

   
  
            
  

       
    
   

  

W.
33......
D”;.::; .
11m -
«I J
7.7.7::

“III 3111?... III II1I1;::,:,If. I3: II'III‘II

31" u . ’13" 3‘13
L .13!""‘3"3I=I~;1.3,.3_L":-‘1'“'731II
11"IIII'L1’MI 1,1. I. .:. III; I XIII -’
’ {1:35. . 13.33:» '3‘3'. 13:61.11. :' 33/. ‘
;'~v1 I‘II’I 931'.‘-’1I'I:II‘ 3 35131139221 1.3I1IIIIII.. I
.III‘t' . tifﬁ‘l" LII-S: 'Iﬁ~‘{‘z"1:'if; ‘25: tig‘.%,'11:.
‘7. I 353326333 HI;
I i 1,5331, [35131333 1'.‘ {II._:I“I
b.-‘J',3;‘1::U.II 3393-31-11,” '1 3 31511311211!“ '12!“
”33315-1311 "1'” 7W1, d“, 1 Ir 33“. I?“ E-
‘II «II IIIIII 1‘- 3333333351 I1; 4-32 III-11,111 1. III
13‘". ij H- uII ,, $14181 ;~ 1333141.. 3.3.;- at? 1" 3
III}, IJIIIIIIIJI! I "1331* 33331131“ 3I1II~:I.-.III II
I ,r: I" “III
33

 
   
         
       
      
  

       
  

.3. .11 . 3:15;; I“?! III

33 3"33'1‘13II‘I‘I‘IIII3' III' ”III, III?“ I 333333

I I I33» I n III, I .
- 141;“! 23'.-

gI‘Il‘IEwM-n}! 3331: ‘3 1:13: {"133 “”33 1,? 1333 .7 “[3133: 3123131531”
331731; I'M”; IIE’IIIIIII IIaIaeIsI‘

     
   
   
   

  

J:

W“.
£11???

a u a I

4....2‘33. '
zIFawt, “-3

2-."

  
 
 

31119:: 521

    

{II‘IIIII' 311;.3III'II
. I- “IE“III I'1'IvI‘I‘I3‘J‘
3:13 “”19"“ 13' III ‘

  

.~......

.4.-—.

 

         

,, :Wm'z'..." . .. ..
‘ ' ‘ ,0 .. .‘.
4 _ . . _:.
~“ ”.4" -
““ -<- _
., ~ _

   

J‘ ..

    
   
 
 

  
         
  

          
 
 

  

.4.

' 13 1 .
3-2. I.- {I 3355
,. 332311333333
11:15:81“ I; ‘
3'133‘1-‘l3h5_1317 1,3
‘ 1 3312531” -:a
.- 1.1;.) Ni 3’" VIII
4133 ’33 111:“

wow—«25::

L‘Zl

.. .aﬂ
£9
—" .Egu'.“

- ~¥--.:.:'. .
$3;

.7. _
.43....»
."'..>.

4,- ..
-;'.

    
 
     

       

’r-.4.

..:r":‘"
...-_"..a. -
_A~"-I

. ..
'""“..'..
.

.
..._._~
”Iii-Sf»: .
tficzr—t . 31-h”

31:2.

    
   

,3
IglL’iiI; ' 31$,

‘ '3' . 1 ‘1'. 1 I!
t {l I “w“‘z 3-;$|‘3:t':£1.112
#:1513531}: 'L‘I 3:31: ‘m‘htd 31:33"

I I II I ’
- ..,. III-III II-
1‘3" IIEI'EII‘} 33411:” 19
h' 5 I
1 .3313 '1335' 'I '
“31-1 3

‘ IIIl'E

  

.. 11‘.

  
 
   
     
 
 
  
  
  
 

:J

 

.— L
.p 1:.

 
 
   
 
 
 

 
 
 
  
 
 
  
  
  
   
 
 
  
 
 

'1’" :1}. 3" 13:3I 31331.43“ 3:31:31“: 11 _
.I ': 3': d‘ 3“ JIII' I {3‘31 I. 4. I 1
[3133- “nil; II IIxflIglg‘I :1.

  

 

    

     

       
   
 
   
   

' '. . 131111;."
ff; "41331333" I:
.1133:

33"“.
133,3

 

    
 
   
 
 

. 33:21.1},
M“ 1'1.”me I!
{3‘31”"3I‘L
I} L"? ‘3
i3. 13'313II'I1‘”
"-1. ‘II‘III’IIIIIIh ,

   
  
 

3‘83; 1
1 3'1. 1 .133 1IIIEEI'I t'n
3 'c I111':1‘||l;!:.
II ‘1 I, II _
1.133;”; I 31"

\V4 .
‘ ='.I. '. J ((((

         
   
 
 
  

.o—l.‘ ,.‘:I . ' ‘

    
 
   
    

   
  

1:35

   

 
 
 
   

 
   
   
  

'1 13,": ,‘Is. 'II". 13“" ' -
IIII 3'13“ 33‘1” 11,312.23; II 11.113313114333311 , f1 331
1 _.: I' I . II
3 3f 331133 I3.IIIIIiII III laigIquH I
‘133313‘3133333'3'1'gn‘. IIIII I I IIIIZII ,. ..
1373311? E] '13,," 1311:: 1'11." 133.3: .113’1-1 1'5“ {Iiz‘ = ‘
3‘! i {3‘13" 111,33‘111111333‘3‘13'13 "J L:
.-..I '3 :' II3IiI 3:". “JV-13335 311. If
w. I: = III; ,, .
I I“ " ’ " I-I 3'31‘3’3‘911fIILiIIIIia‘i
- I I?" III-«I 131131151. . 3121 {II
, Lit: 1IILI4IuI11-II3IIIIIIIL"3V f
‘9‘» I

 
 

. III» 3». I '.".'1;' I.»
:IIIII Iit’gl~1-~‘.I‘II' [ 3 3 135‘)"in I_' 333131
. I73.‘3,
III-IIIIIIII ‘ 1 I I
III- v3

. ”“1333;
3 ”1:" "01'“
33.3-11'1H1 11S” 3‘ I
8): r " '11'1|1‘1.3§[.1"| '1‘! It; .‘..
' 3:313 IIIIIIII1I33IIILIII3 3343 I131: I‘Iﬁ“ ‘ . 3 '3
9-1

33..
“f: I. I‘m
#:1333333“ 3:341, 3i! ‘1' i 3
1.313.113?!“

 
   
 
 
   

   

 
      
   
 
 

  

       

#7:... z- .. 3-
.me.- v:
.4
u. .-
_.,
{—
__...

  

    
    
 
 

        
  
 

 
  
    

 

'..‘.~.. ..
1-6”

 
 

_ ﬁxry‘éﬁ— ﬁw . -

   

    

   

THEES
Z.

NIVERSITY LIBRARI IES

IIIIIIIIIIIIIIIIIIIIIII I I I II I III

3 1293 01710

 

 

 

 

 

 

 

 

 

 

 

 

 

This is to certify that the

thesis entitled

SPORT CAUGHT GREAT LAKES FISH CONSUMPTION AND HUMAN
SERUM POLYCHLORINATED BIPHENYL LEVELS:CROSSfSECTIONAL
AND LONGITUDINAL RELATIONS IN MICHIGAN ANGLERS

presented by

Jianping He

has been accepted towards fulfillment
of the requirements for

Master's . Science
degree 1n

W@%

I

 

 

ﬁajor professor

Date LI/1‘5/0IC5

0-7639 MS U is an Afﬁrmative Action/Equal Opportunity Institution

 

 

  
      

 
 

LIBARY
Michigan State
Unlverslty

PLACE IN RETURN BOX
to remove this checkout from your record.
TO AVOID FINES return on or before date due.

 

DATE DUE DATE DUE DATE DUE

 

.1532? 200a
.9 01

I
L:

 

 

 

DWI? It 1

 

 

one

A
w

 

g 43 20m

__.

 

 

 

 

 

 

 

 

1/” mm."

 

SPORT-CAUGHT GREAT LAKES FISH CONSUMPTION AND HUMAN SERUM
POLYCHLORINATED BIPI-IENYL LEVELS: CROSS-SECTIONAL AND
LON GITUDINAL RELATIONS IN MICHIGAN AN GLERS

By

J ianping He

A THESIS

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

Master of SCIENCE

Department of Epidemiology

1998

ABSTRACT

SPORT-CAUGHT GREAT LAKES FISH CONSUMPTION AND HUMAN SERUM
POLYCHLORINATED BIPHENYL LEVELS: CROSS-SECTIONAL AND
LONGITUDINAL RELATIONS IN MICHIGAN AN GLERS

By

J ianping He

The present study combined 3 cross—sectional surveys conducted by Michigan
Department of Community Health in 1973-74, 1979-82, and 1989-91. The association
between ﬁsh consumption and serum contaminant levels and their joint change pattern
are examined longitudinally. Over the period of the 3 surveys, there was a decline in the
total annual pounds of ﬁsh consumed (median 40, 38, and 31, respectively). Men
reported consuming more ﬁsh than women at each survey. In addition to eating less ﬁsh,
ﬁsheaters reported modifying their ﬁsh preparation methods, and changing in the favorite
ﬁsh species eaten. Median PCB levels (ppb) were 2 to 3 times higher in ﬁsheaters than
nonﬁsheaters in each survey (14.0 versus 6.0; 19.4 versus 7.0; and 17.3 versus 5.8,
respectively). In repeated measure models analyses, ﬁsh consumption decreased
signiﬁcantly over 20-year study period (b=-21.l6 pound/decade, p<.001). After adjusting
for potential confounders, log PCB were associated with concurrent ﬁsh consumption;
log PCB levels signiﬁcantly increased from survey I to II (b=.31, p=.002) and remained
high in survey 111 (b=. 13, p=.19). It is concluded that sport caught Great Lakes ﬁsh
consumption has declined among Michigan anglers. Levels of PCB have not declined in

parallel with declining ﬁsh consumption probably reﬂecting the long PCB half-life.

ACKNOWLEDGMENTS

This endeavor would never have been possible without the help and support of
numerous individuals. First, I must thank my thesis advisor, Dr. Aryeh D. Stein, who
sacriﬁced countless hours providing guidance, reading drafts, and granting moral support
in the process of completing the work.

I am indebted to Professor Nigel Paneth, my academic advisor, for his
encouragement and numerous valuable suggestions. I would like thank Dr. Jeanne M.
Couval, the principal investigator of the project, for her guidance and for the considerable
time she spent reading and helping shape my thesis even during her late pregnancy. I
have been very fortunate to have the enthusiastic support and guidance of Dr. Dorothy R.
Pathak, Dr. Lawrence J. Fischer, and Dr. Joseph C. Gardiner.

This thesis relies heavily on data provided by Michigan Department of
Community Health, and thanks go to Dr. Harold E.B. Humphrey and Diane Getts for its
provision. I am especially grateful to Dr. Humphrey, upon whom I relied at various times
for information on the data.

I would like to thank the National Food Safety and Toxicology Center for their
ﬁnancial support. Special thanks extend to Michigan Department of Environmental

Quality and Agency for Toxic Substances and Disease Registry.

I would never have completed my journey through graduate school without the
support of family and friends. Particularly essential to this accomplishment is the
unending support and understanding of my husband, Xiao, and my daughter, Shiwei. I
am also thankful to all the faculty, staff, and a group of supportive classmates in
Department of Epidemiology at Michigan State University for their inspiration and

assistance.

TABLES OF CONTENTS

LIST OF TABLES ...................................................................................................... viii
LIST OF FIGURES ..................................................................................................... xii
LIST OF ABBREVIATIONS ..................................................................................... xiii
1 INTRODUCTION ............................................................................................. 1
1.1 The Great Lakes .................................................................................... 1
1.2 The nature of PCBs ............................................................................... 2
1.3 The toxicity of PCBs ............................................................................. 4
1.4 Findings from animal studies ................................................................ 6
1.5 Routes of human exposure to PCBs ...................................................... 8
1.6 Effects of PCBs on human health .......................................................... 9
1.7 The hypotheses and objectives of the present study ............................ 12
2 METHODOLOGY .......................................................................................... 14
2.1 Population and setting ......................................................................... 14
2.1.1 Study population ...................................................................... 14
2.1.2 Time windows ......................................................................... 15
2.1.2.1 1973-74 (survey 1) ....................................................... 15
2.1.2.2 1979-82 (survey II) ...................................................... 15
2.1.2.3 1989-91 (survey III) ..................................................... 16
2.1.3 Recruitment ............................................................................. 16
2.2 Data collection instruments ................................................................. 17
2.2.1 Fish consumption assessment .................................................. 17
2.2.2 General information module ................................................... 19
2.2.3 Measurement of serum PCBs and other
contaminants ........................................................................... 20
2.3 Data management and preparation ...................................................... 21
2.3.1 Deﬁnition of ﬁsh eating status ................................................ 21
2.3.2 Computation of ﬁsh consumption ........................................... 23
2.3.3 Choice of Aroclor .................................................................... 24
2.3.4 Integration of data within each survey ..................................... 24
2.3.5 Integration of data between surveys ........................................ 26
2.4 Statistical methods ............................................................................... 27
2.4.1 Univariate analysis .................................................................. 27

2.4.2 Multivariate analysis ............................................................... 29
3 FINDINGS ...................................................................................................... 31
3.1 Descriptive results and bivariate analysis ............................................ 31
3.1.1 Background characteristics of the study
population ................................................................................ 31
3.1.2 Characteristics of tobacco use and alcohol
consumption in the study population ....................................... 32
3.1.3 Fish consumption behaviors .................................................... 33
3.1.3.1 Survey I data ................................................................ 33
3.1.3.2 Survey H data ............................................................... 34
3.1.3.3 Survey III data ............................................................. 35
3.1.4 Laboratory measurements ........................................................ 36
3.1.4.1 Laboratory data at survey I .......................................... 36
3.1.4.2 Laboratory data at survey 11 ......................................... 36
3.1.4.3 Laboratory data at survey III ........................................ 37
3.1.5 Effect of general characteristics on the ﬁsh
consumption behaviors ............................................................ 38
3.1.6 Effect of general characteristics on contaminant
levels ........................................................................................ 39
3.1.7 Relationship between ﬁsh consumption and
contaminant levels ................................................................... 41
3.1.8 Pairwise comparisons of individuals surveyed
more than once ........................................................................ 41
3.1.8.1 Survey I versus survey II ............................................. 41
3.1.8.2 Survey 11 versus survey III ........................................... 43
3.1.8.3 Survey I versus survey III ............................................ 44
3.2 Multivariate analysis ........................................................................... 45
3.2.1 The structure of multivariate analysis ..................................... 46
3.2.1.1 Multiple linear regression analysis
within survey ............................................................... 46
3.2.1.2 Multiple linear regression analysis
between surveys ........................................................... 46
3.2.1.3 Repeated measure model analysis over
three surveys ................................................................ 47
3.2.2 The coding of multivariate analysis ........................................ 47
3.2.3 Results of multivariate analysis ............................................... 48
3.2.3.1 Fish consumption within surveys I, II,
and III .......................................................................... 48
3.2.3.2 Log levels of Aroclor 1254 in survey I ........................ 49
3.2.3.3 Log levels of Aroclor 1260 within
survey 1, II, and III ....................................................... 50
3.2.3.4 Log levels of DDT, and DDE within
survey 11 and III ........................................................... 51

vi

3.2.3.5 Log levels of PBB within survey II and

III ................................................................................. 51
3.2.3.6 Changes in ﬁsh consumption between

surveys-from survey I to II and from

survey 11 to III .............................................................. 52
3.2.3.7 Changes in levels of Aroclor 1260

between surveys-from survey I to II and

from survey II to III ..................................................... 53
3.2.3.8 Changes in levels of DDT from survey 11

to 1]] ............................................................................. 54
3.2.3.9 Changes in levels of DDE from survey H

to III ............................................................................. 54
3.2.3.10 Changes in levels of PBB from survey

11 to III .......................................................................... 55

3.2.3.11 Trends of changes in ﬁsh consumption
and log levels of Aroclor 1260 over the

three surveys ............................................................... 55
4 DISCUSSION .................................................................................................. 57
4.1 Strength of the present study ............................................................... 57
4.2 Time trends in ﬁsh consumption behaviors among
ﬁsheaters .............................................................................................. 57
4.3 Determinants of contaminant levels .................................................... 59
4.3.1 Fish consumption .................................................................... 59
4.3.2 Gender ..................................................................................... 60
4.4 Joint changes in ﬁsh consumption and PCB levels ............................. 62
5 LIMITATIONS ................................................................................................ 63
5.1 It is not a population-based sample. .................................................... 63
5.2 Information on ﬁsh consumption of the majority of
nonﬁsheaters was not available. .......................................................... 63
5.3 Information on PCB levels in Great Lakes ﬁsh was not
integrated into the analysis. ................................................................. 63
5.4 Few participants had three data points. ............................................... 64
6 FURTHER STUDY NEEDS ........................................................................... 65
APPENDICES ........................................................................................................... 129
BIBLIOGRAPHY ..................................................................................................... 13 1

vii

LIST OF TABLES

Table 1: Background characteristics of all study population, 1979-82 ..................... 66
Table 2: Background characteristics of male participants, 1979-82 ......................... 67
Table 3: Background characteristics of female participants, 1979-82 ...................... 68
Table 4: The characteristics of tobacco use and alcohol consumption

of male participants, 1989-91 ..................................................................... 69
Table 5: The characteristics of tobacco use and alcohol consumption

of female participants, 1989-91 .................................................................. 70
Table 6: The quantity of ﬁsh consumption among ﬁsheaters stratiﬁed

by gender, 1973-74 ..................................................................................... 71
Table 7: Fish consumption behaviors among ﬁsheaters stratiﬁed by

gender, 1973-74 .......................................................................................... 71
Table 8: The quantity of ﬁsh consumption among ﬁsheaters stratiﬁed

by gender, 1979-82 ..................................................................................... 72
Table 9: Fish consumption behaviors among ﬁsheaters stratiﬁed by

gender, 1979-82 .......................................................................................... 72
Table 10: The quantity of ﬁsh consumption among ﬁsheaters stratiﬁed

by gender, 1989-91 ..................................................................................... 73
Table 11: Fish consumption behaviors among ﬁsheaters stratiﬁed by

gender, 1989-91 .......................................................................................... 73
Table 12: Levels (ppb) of serum Aroclor 1254 and Aroclor 1260

stratiﬁed by gender and ﬁsh eating status, 197 3-74 .................................... 74
Table 13: Levels (ppb) of serum contaminants (Aroclor 1260, total

PCB, DDT, DDE, and PBB) stratiﬁed by gender and ﬁsh

eating status, 1979-82 ................................................................................. 75
Table 14: Levels (ppb) of serum contaminants (Aroclor 1260, total

PCB, DDT, DDE, and PBB) stratiﬁed by gender and ﬁsh

eating status, 1989-91 ................................................................................. 76
Table 15: Effect of the age on quantity of ﬁsh consumption stratiﬁed

by gender, 1979-82 ..................................................................................... 77
Table 16: Effect of education on quantity of ﬁsh consumption stratiﬁed

by gender, 1979-82 ..................................................................................... 78

viii

Table 17: Effect of region on quantity of ﬁsh consumption stratiﬁed by

gender, 1979-82 .......................................................................................... 79
Table 18-a: Effect of body mass index on quantity of ﬁsh consumption

in male ﬁsheaters, 1979-82 ......................................................................... 80
Table 18-b: Effect of body mass index on quantity of ﬁsh consumption

in female ﬁsheaters, 1979-82 ...................................................................... 80
Table 19: Effect of chemical exposure status on quantity of ﬁsh

consumption stratiﬁed by gender, 1979-82 ................................................. 81
Table 20: Effect of the age on contaminant levels stratiﬁed by ﬁsh

eating status and gender, 1979-82 ............................................................... 82
Table 21: Effect of education on contaminant levels stratiﬁed by ﬁsh

eating status and gender, 1979-82 ............................................................... 83
Table 22: Effect of region on contaminant levels stratiﬁed by ﬁsh

eating status and gender, 1979-82 ............................................................... 84
Table 23-a: Effect of body mass index on contaminant levels stratiﬁed

by ﬁsh eating status in male participants, 1979-82 ..................................... 85
Table 23-b: Effect of body mass index on contaminant levels stratiﬁed

by ﬁsh eating status in female participants, 1979-82 .................................. 85
Table 24: Effect of chemical exposure status on contaminant levels

stratiﬁed by ﬁsh eating status and gender, 1979-82 .................................... 86
Table 25-a: Effect of annual ﬁsh meals on contaminant levels in male

participants, 1979-82 .................................................................................. 87
Table 25-b: Effect of annual ﬁsh meals on contaminant levels in

female participants, 1979-82 ....................................................................... 88
Table 26-a: Effect of total weight of ﬁsh consumed on contaminant

levels in male participants, 1979-82 ........................................................... 89
Table 26—b: Effect of total weight of ﬁsh consumed on contaminant

levels in female participants, 1979-82 ........................................................ 90
Table 27: Changes in quantity of ﬁsh consumption among ﬁsheaters

stratiﬁed by gender from survey I to II ....................................................... 91
Table 28: Changes in ﬁsh consumption behaviors among ﬁsheaters

stratiﬁed by gender from survey I to II ....................................................... 91
Table 29: Changes in levels of serum Aroclor 1260 stratiﬁed by gender

and ﬁsh eating status from survey I to H ..................................................... 92
Table 30: Changes in quantity of ﬁsh consumption among ﬁsheaters

stratiﬁed by gender from survey 11 to 111 ..................................................... 93
Table 31: Changes in ﬁsh consumption behaviors among ﬁsheaters

stratiﬁed by gender from survey 11 to 111 ..................................................... 93

Table 32: Changes in levels of serum Aroclor 1260, DDT, DDE, and
PBB stratiﬁed by gender and ﬁsh eating status from survey

11 to III. ........................................................................................................ 94
Table 33: Changes in quantity of ﬁsh consumption among ﬁsheaters

stratiﬁed by gender from survey I to 111 ...................................................... 95
Table 34: Changes in ﬁsh consumption behaviors among ﬁsheaters

stratiﬁed by gender from survey I to II] ...................................................... 95

ix

Table 35: Changes in levels of serum Aroclor 1260 stratiﬁed by gender

and ﬁsh eating status survey I to survey III ................................................. 96
Table 36: Adjusted linear regression of annual ﬁsh meals on

demographic characteristics among ﬁsheaters stratiﬁed by

gender and survey ....................................................................................... 97
Table 37: Adjusted linear regression of total weight of ﬁsh consumed

on demographic characteristics among ﬁsheaters stratiﬁed

by gender and survey .................................................................................. 98
Table 38: Adjusted linear regression of levels of Aroclor 1254 (log

transformed) on demographic characteristics and ﬁsh

consumption stratiﬁed by gender, 1973-74 ................................................. 99
Table 39: Adjusted linear regression of levels of Aroclor 1260 (log

transformed) on demographic characteristics and ﬁsh

consumption stratiﬁed by gender and survey ............................................ 100
Table 40: Adjusted linear regression of levels of DDT (log

transformed) on demographic characteristics and ﬁsh

consumption stratiﬁed by gender and survey ............................................ 102
Table 41: Adjusted linear regression of levels of DDE (log

transformed) on demographic characteristics and ﬁsh

consumption stratiﬁed by gender and survey ............................................ 103
Table 42: Adjusted linear regression of levels of PBB (log

transformed) on demographic characteristics and ﬁsh

consumption stratiﬁed by gender and survey ............................................ 104
Table 43: Adjusted linear regression of change in annual ﬁsh meals of

predicted against demographic characteristics and baseline

ﬁsh consumption stratiﬁed by gender from survey I to II and

survey 11 to 111 ............................................................................................ 105
Table 44: Adjusted linear regression of change in Aroclor 1260 levels

of predicted against demographic characteristics, baseline

Aroclor 1260 levels, and change in ﬁsh consumption

stratiﬁed by gender from survey I to II and survey II to III ....................... 106
Table 45-a: Adjusted linear regression of change in DDT levels of

predicted against demographic characteristics, baseline DDT

levels, and change in annual ﬁsh meals stratiﬁed by gender

survey 11 to III ............................................................................................ 108
Table 45-b: Adjusted linear regression of change in DDT levels of

predicted against demographic characteristics, baseline DDT

levels, and change in quartile of change in total weight of

ﬁsh consumed stratiﬁed by gender survey II to I11 ................................... 109
Table 46-a: Adjusted linear regression of change in DDE levels of

predicted against demographic characteristics, baseline DDT

levels, and change in annual ﬁsh meals stratiﬁed by gender

survey 11 to 111 ............................................................................................ l 10
Table 46-b: Adjusted linear regression of change in DDE levels of

predicted against demographic characteristics, baseline DDT

levels, and change in quartile of change in total weight of

ﬁsh consumed stratiﬁed by gender survey II to 111 ................................... 111
Table 47: Adjusted linear regression of change in PBB levels of

predicted against demographic characteristics, baseline PBB

levels, and change in quartile of change in total weight of

ﬁsh consumed stratiﬁed by gender from survey 11 to 111 ........................... 112
Table 48: Adjusted linear regression of total weight of ﬁsh consumed

against general characteristics based on repeated measure

model ........................................................................................................ l 13
Table 49: Adjusted linear regression of Aroclor 1260 levels against
general characteristics based on repeated measure model ........................ 114

Table 50: Adjusted linear regression of Aroclor 1260 levels against

total weight of ﬁsh consumed and general characteristics

based on repeated measure model ............................................................. 115
Table 51: Adjusted linear regression of Aroclor 1260 levels of

predicted against ﬁsh consumption and general

characteristics stratiﬁed by gender based on repeated

measure model. ......................................................................................... 116

xi

LIST OF FIGURES

Figure 1: Structure of biphenyl .................................................................................. 117
Figure 2: Sample sizes of the three cross-sectional surveys ...................................... 118
Figure 3: Eleven western Michigan counties covered in the study ........................... 119
Figure 4: Fish consumption, by survey .................................................................... 120
Figure 5: Mentioning as one of the top three species of ﬁsh consumed,

by survey .................................................................................................... 121
Figure 6: Fish preparation methods, by survey ......................................................... 122
Figure 7: Serum levels of Aroclor 1260, by survey ................................................... 123
Figure 8: Multiple linear regression of log Aroclor 1260 levels against

quartile of total weight of ﬁsh consumed in males, by survey ................... 124

Figure 9: Multiple linear regression of log Aroclor 1260 levels against

quartile of total weight of ﬁsh consumed in females, by

survey ......................................................................................................... 125
Figure 10: Multiple linear regression of change in Aroclor 1260 levels

against quartile of change in total weight of ﬁsh consumed in

males and females ...................................................................................... 126
Figure 11: Trends in total weight of ﬁsh consumed over the 3 surveys .................... 127
Figure 12: Fish consumption and log serum Aroclor 1260 levels ............................. 128

xii

LIST OF ABBREVIATIONS

PCB .................................................................................. Polychlorinated biphenyl
DDT ..................................................................... Dichlorodiphenyl trichloroethane
DDE ...................................................................... The ethylene metabolite of DDT
PBB .................................................................................. Polybrominated biphenyl
survey I ........................................ Michigan angler PCB survey conducted in 1973-74
survey II ........................................ Michigan angler PCB survey conducted in 1979-82
survey III ........................................ Michigan angler PCB survey conducted in 1989-91

xiii

1 Introduction

1.1 The Great Lakes

The Great Lakes are the largest single collection of fresh water on the surface of
the earth, excluding the polar ice caps. About 25 percent of the Canadian population and
10 percent of the United States population live within the Great Lakes watershed(l). The
Great Lakes has been used a valuable resource for both the United States and Canada in
industry, agriculture, shipping, and recreation for more than 200 years.

Although the Great Lakes are open to the ocean through the St. Lawrence River,
the ecosystem of the Great Lakes is actually much like a closed system. The water
exchange rate is very low--it takes three years for Lake Erie to exchange its waters, and
173 years for Lake Superior. The long retention times make the Great Lakes act as the
reservoir for any introduction of change. By the early 1960’s, the environmental quality
of the Great Lakes had deteriorated signiﬁcantly as a result of eutrophication, overﬁshing,
and the widespread presence of toxic substances. Due to reports on detrimental effects on
ﬁsh and wildlife, and the potentially adverse effects on human health, an increasing
widespread concern has been shown since the 1970s.

More than 1,000 chemicals have been detected in the waters, sediment, or biota of
the Great Lakes. Many of these toxic substances tend to absorb onto particles and
eventually settle to the bottom of the Great Lakes and become permanent or make their
impact until they have very slowly eliminated from the system (2-4). The International

Joint Commission has identiﬁed various persistent toxic substances in the Great Lakes,

including organo-chlorines (e.g. polychlorinated biphenyls), heavy metals (e. g.
methylmercury), and benzo[a]pyrene(a member of a class of substances known as PAHs).
Eleven of the most persistent substances were identiﬁed as ‘critical Great Lakes
pollutants’(2, 3). The critical Great Lakes pollutants identiﬁed by the International Joint
Commission are: Polychlorinated biphenyls (PCBs), dichlorodiphenyl trichloroethane
(DDT) and metabolites, benzo[a]pyrene, hexachlorobenzene, alkylated lead,
methylmercury, toxaphene, mirex, furans, dieldrin, and dioxins. Because of their
lipophilic nature and long half-lives, all 11 such toxic substances, including
polychlorinated biphenyls, tend to bioaccumulate in organisms and biomagnify in food
chain. PCBs were detected in signiﬁcant concentrations in ﬁsh and wildlife in the late
19605. Statistical sampling surveys of contaminants in the Great Lakes ﬁsh began in
1972. The Great Lakes have been monitored for contaminants in the water, the ﬂora, and
the fauna since the 19703. Several categories of pollutants have been investigated

extensively, including PCBs, DDT, and DDE(5).

1.2 The nature of PCBs

PCBs are chemical compounds having the empirical formula C1 ZHIMCln, with n
equals 1 to 10. The generalized molecular structure of chlorinated biphenyls and the
numbering of the carbon atoms in the rings are shown in Figure 1.

There are 209 theoretically possible discrete synthetic chemical compounds called
PCB congeners which can be divided into nine isometric groups and
decachlorobiphenyls, each of which has a systematic number(6). PCBs were ﬁrst

synthesized in 1881, but did not see industrial application in the United States until 1929.

Commercially, PCBs are manufactured by the batch chlorination of biphenyl, which
results in technical mixtures containing a given chlorine content, depending on the
duration of the chlorination process. PCB products had the trade name ‘Aroclor’ and
were manufactured by Monsanto in the United States until 1977(7). These Aroclor
compounds were characterized and named by their different degrees of chlorination. The
particular kind of Aroclor is identiﬁed by a four-digit number. The ﬁrst two digits refer
to the 12 carbon atoms, and the second two refer to the percent, by weight, of chlorine in
the mixture. Thus, Aroclor 1254 contains about 54% chlorine, and Aroclor 1260, about
60% chlorine.

Most PCBs are oily liquids or solids, clear to light brown in color, and have no
smell or taste. The physical-chemical properties of PCBs, which vary on a congener-by-
congener basis, are inﬂuenced by the different degrees of chlorination around the phenyl
rings. Because of their high dielectric constant, their chemical and thermal stability, their
non-ﬂammability, and their low cost, PCBs were once widely used(8). They mostly used
as insulating ﬂuids in electrical transformers and capacitors, heat transfer substances,
cutting oils, hydraulic ﬂuids, lubricating oils, and as plasticizers for making brittle plastic
pliable. Application was also found for these compounds in paints, in inks and dyes, as
ingredients in pesticides, in adhesives, in protective wood coatings, in carbonless-copy
paper. There are several properties of PCBs affecting their biodegradability, i.e., thermal
stability (low volatility), difﬁcult to oxidize and reduce, very low water solubility (i.e.
high lipophilicity, high dielectric constant), resistant to acid-base, hydrolysis, chemical
oxidation, photodegradation reactions, and existence of a large numbers of PCB

congeners, each of which have differing susceptibilities to biodegradation (9, 10). These

characteristics make PCBs persistent and accumulative in the environment. Some
components of the mixture are more easily degraded in the environment than others.
PCBs with fewer chlorine atoms, or with fewer chlorine atoms attached to the biphenyl
molecule, or with at least one paraposition on the molecule not occupied by a chlorine,
are more soluble, and thus more amenable to chemical and biological degradation. They
can be more easily metabolized and are therefore less persistent in the environment than
those PCBs with more chlorine atoms. The most toxic congeners however, are those that
hold a coplanar conformation with chlorine substitutes on the meta and para positions of
the phenyl rings (1 1). Since PCBs are poorly soluble in water and extremely soluble in
oils and fats, they tend to partition out of the aquatic ecosystem into biologic tissue.
Because of their persistence and the fact that they are poorly metabolized PCBs can
bioaccumulate in the food chain leading to an increase in concentration at each
succeeding trophic level (biomagniﬁcation)(12).

PCBs do not occur naturally and were never intended to be released into the
environment. PCBs have entered the environment partly because of accidental leaks and
ﬁres in electrical equipment, past disposal in dumps, accidents in transport, and leakage
from hazardous waste sites (6). They exist everywhere in the environment as a result of
their widespread usage and their long half lives(13). Although the production was ceased

in 1977, PCB residuals have been detected in animal and human tissue ever since.

1.3 The toxicity of PCBs

PCBs rarely cause acute toxicity. The dose required to kill 50% of a treated

sample of animals is quite high, which ranges anywhere from 0.5 g/Kg to 113ng body

weight(14). Most of the effects observed are the result of repetitive or chronic exposure.
People and most animals absorb PCBs through the skin, the lungs, and the
gastrointestinal tract. Once these chemicals are inside an individual they are transported
through the blood stream to various tissues. However, because they are highly lipophilic,
the PCBs largely tend to settle in adipose tissue(15).

The body of data reported to date indicates that the toxicity and half-lives of PCB
varies with the extent of chlorination of the biphenyl molecule. Our knowledge of their
effects is based on some studies of people exposed in the workplace or who ate
contaminated food, especially two tragic episodes in Japan and Taiwan, and on
experimental studies with animals. The toxic effects of PCBs on humans and animals
may vary according to the route of exposure, age, sex and the area of the body where the
PCBs are concentrated( 16).

Identiﬁcation of highly toxic polychlorinated dibenzofurans (PCDFs) in
commercial PCBs has contributed to the belief that they play a major role in the PCBs
toxicity(17). Some of the PCBs are stereochemically similar to the planar 2,3,7,8-
tetrachlorodibenzo-p-dioxin (T CDD). The toxicity of such PCBs can be expressed in
terms of TCDD-equivalency factors (TEFs), that is, expressed as a fraction of the toxicity
of TCDD, including body weight loss, thymic atrophy, reproductive toxicity, liver
enzyme induction, teratogenicity, immunotoxicity, alterations in lipid metabolism and
formation of reactive intermediate metabolites which may be responsible for cytotoxicity
and carcinogenicity(l8).

Planar PCBs and some of their mono-ortho substituted congeners exist in very

low concentrations in commercial PCB mixtures. Hence, their environmental occurrence

is relatively low compared to that of many other PCB congeners. Their mono-ortho, non-
coplanar derivatives were detected at much higher levels in commercial Aroclor mixtures
and this may be even more important in terms of toxic activity. In human adipose tissues,
the concentrations of non-coplanar PCBs may impose a greater threat than the PCDFs to

humans and probably to wild life also.

1.4 Findings from animal studies

Most laboratory-based animal studies conducted in the past 20 years used
mixtures of PCBs. In general, PCBs have been found to affect reproduction and the
immune response and to cause liver tumors in rodents(19). In animal studies, more
recently some of the isomers of the PCBs mixture have been found to be much more toxic
than others. The more toxic isomers constitute only a very small portion of the mixture,
particularly those with less chlorine by weight. Scientists were able to produce
hepatocellular carcinomas in rats with a German PCB mixture(20) or Aroclor 1260 (21,
22). When Aroclor 1254 was fed to rats, fewer liver tumors developed in exposed rats;
however, the incidence of gastric intestinal metaplasia and adenocardinoma of the
stomach increased (23). Aside from tumor formation, PCBs cause a variety of other
biological effects, such as the induction of enzyme(24). In some species they may cause
atrophy of the thymus, intrahepatic bile duct hyperplasia, hyperplasia of the epithelial
lining of the urinary bladder, atrophy of the sebaceous glands and hyperkeratosis of the
ducts (25).

Information on the effects on wild life of exposure to these compounds has come

from mice, mink, birds, rhesus monkeys, and seals(26-28). The Great Lakes bird

populations and mink and otter populations have demonstrated inverse relationship
between reproductive success and organo-chlorine levels(29). In laboratory-based
studies, reduced reproductive success has been observed in mink fed Great Lakes coho
salmon, carp, sucker, perch(30, 31) (32). The chemicals may interfere with fertility and
reproduction in bald eagles and river otters (26). Arnold (33) researched the prevalence
of endometriosis in rhesus monkeys ingesting Aroclor 1254 and the results suggested a
possible link between the PCB treatment and the occurrence of endometriosis but there
was no dose-response relationship. A study conducted by de Swart and co-workers (34)
demonstrated immunosuppression in seals, one of the top marine predators. The seals in
the experimental group were fed high levels of persistent lipophilic environmental
contaminants. Two and half years later, impaired natural killer and speciﬁc T cell
responses were found in the experimental group.

In general, PCB intoxication produces a wide range of symptoms including:
neurological dysfunction, altered neuromuscular reﬂexes, increased susceptibility to
infections, reproductive dysfunction, and lowered birth rates. Rhesus monkeys were
unique in having symptoms such as facial swelling, loss of hair, increased skin
pigmentation, and a skin condition known as chloracne(28). The difﬁculties in using
animal data to predict human health effects for PCBs and related compounds is that
animal species very greatly in their responses. Subhuman primates, mink, and guinea
pigs are particularly sensitive to the toxic effects of PCBs; other species, such as the rat,
the mouse, and the dog, can tolerate much higher doses. Further, many of the animal
studies use relatively high doses. Therefore, determining how such animal studies relate

to the human situation is difﬁcult.

1.5 Routes of human exposure to PCBs

In industry, absorption by the skin (dermal route) is the major route of exposure to
these contaminants(35). The primary route of general population exposure to PCBs is
ingestion of contaminated food(36, 37). Ingestion of untreated drinking water is a third
route of human exposure to organo-chlorines (38, 39). A fourth, less prominent,
exposure pathway is inhalation of polluted air. PCBs are also known to be passed from
the mother to the fetus through placental blood, and to the baby via breast milk(40-42).

In the Great Lakes region, consumption of contaminated ﬁsh has been identiﬁed
as an important exposure route (43) (44). PCBs enter the bodies of ﬁsh from water,
sediment and from eating prey that have PCBs in their bodies. PCBs build up in the
Great Lakes ﬁsh and can reach levels hundreds of thousands of times higher than the
levels in water(43). The state and federal inspection systems control exposure to PCBs
from commercial sources by seizure of contaminated ﬁsh before it reaches the market.
However, Michigan is the only state that is surrounded by four of the ﬁve Great Lakes
and it has the longest shoreline of any state. Many Michigan residents engage in sport
ﬁshing. Since many Great Lakes ﬁsh species are contaminated with a variety of organo-
chlorine compounds, particularly PCBs, sport ﬁshing represents an uncontrolled source of
PCB contaminated ﬁsh for distribution and human consumption. Due to the persistent
nature of these contaminants and their biomagniﬁcation and accumulation, Michigan
anglers who consume larger amount sport caught Great Lakes ﬁsh and wildlife than the
general population are likely to constitute a population at-risk of exposure to PCBs
compounds. It is estimated that perhaps four million of the 40 million people living in

the Great Lakes basin eat the ﬁsh and are therefore exposed to these chemicals (44).

Although we are not able to exactly quantify the amount of ﬁsh consumed by Michigan
anglers annually, it is estimated that sports ﬁshermen as a group consume in excess of the
national average of ﬁsh consumption(45). Sport ﬁsh anglers and their families thus

represent a population with potential elevated risk for exposure to PCB compounds.

1.6 Effects of PCBs on human health

The effects of acute human exposure to PCB mixtures containing other
contaminants are reasonably well documented, mainly as a result of two widely known
tragic episodes--l968 Japanese mass poisoning and the 1979 outbreak in Taiwan. Both
of these incidents were caused by the accidental introduction of PCBs and heat-degraded
.by-products into rice oil used for domestic cooking. These by-products included
polychlorinated dibenzofurans, terphenyls, and quarterphenyls. Since the disease was
caused by ingesting contaminated rice oil, it was called ‘Yusho’ in Japan and ‘Yu-cheng’
in Taiwan, which both meaning ‘rice-oil disease’(46, 47). Approximately 1,600
individuals in Japan suffered a classic set of disease symptoms. Chloracne was one of the
leading signs in those who became ill. A broad spectrum of effects was observed and
these are typiﬁed by the toxic symptoms observed by Kuratsune and coworkers (17).
These symptoms or signs included hyperpigmentation of the skin, distinctive hair
follicles, increased sweating at palms, acne-like skin eruptions and so on. Many of these
same symptoms were also observed in Yu-Cheng patients(48, 49). And it is clear from
numerous studies that both poisonings share a common etiology.

The effect of the acute poisoning of Yusho / Yu-Cheng victims has been

monitored over the succeeding decades. Offspring of exposed mothers have exhibited

10

dermatological, neurological, dental and immunological effects(50, 51) (52). Some
Japanese offspring, termed the ‘Yusho’ cohort, exhibited developmental abnormalities
including small birth weight, growth deﬁcits, hypotonicity, apathy and IQ impairments
(53). The Taiwanese children, victims of the ‘Yu-Cheng’ accident, have been the most
extensively investigated(54). These Children had higher than expected rates of low birth
weight, growth retardation, increased incidence of hyperpigmented skin, abnormal
dentition, otitis media (middle-ear infections) and lower scores on various tests of
cognitive development(55). However, interpretation of these ﬁndings is complicated by
the fact that there did not appear to be any relationship between available indices of
exposure and severity of effects, and by the fact that the PCBs to which the Taiwanese
were exposed contained unusually high concentrations of dibenzofurans, which are many
times more toxic than PCBs, and may have been responsible for some or all of the
observed effects (56, 57).

Occupational exposures have been reported since 1930, with the beginning of jobs
such as capacitor and transformer manufacture workers and machinists. Workplace
exposure to commercial PCBs can result in signiﬁcant uptake of these compounds and
this is reﬂected in the high serum or adipose tissue levels detected in the groups of
exposed workers(58). The effects of occupational exposure to PCBs are comparable to
some of the effects observed in laboratory animals. Chloracne and related skin problems
have been observed in several groups of workers(59). It was suggested that the air
concentrations of commercial PCBs > 0.2 mg/m3 were associated with this effect(60).
The effects of occupational exposure to PCB on the levels of several serum chemical and

hematological parameters have been reported by several studies. Mildly elevated SGOT

11

and glutamyl transpeptidase (GGTP) suggest some liver damage and induction of hepatic
monooxygenase enzyme (61). Warshaw and colleagues(60) reported a relatively high
incidence of pulmonary dysfunction in capacitor workers and the symptoms included
coughing, wheezing, tightness in the chest, and upper respiratory problems or eye
irritation. It has been speculated that prolonged exposure to potential carcinogens with a
long half-life could produce continuous target organ exposure to which should result in
tumor production. However, there is no strong evidence of elevated mortality rates in
human population (62-64). The epidemiological evidence is not of sufﬁcient strength to
link exposure to these compounds with the elevated cancer incidences(65, 66).

Serum levels of PCBs in the general population are usually less than 10-15 parts
per billion (ppb) and these concentrations are signiﬁcantly lower than observed in
occupationally exposed workers in whom serum PCB concentrations as high as 3,000 ppb
have been reported(67). Individuals who consume relatively large amounts of ﬁsh
represent a small sub-population exposed to higher levels of PCBs and this is reﬂected in
elevated serum PCB concentration (68). Researchers have suggested that PCBs carried
by lake trout and salmon can impair the children of mothers who eat a steady diet of the
ﬁsh(69-7 1). The developing fetus and neonate are considered to be at particular risk as
there is great potential for exposure to environmental contaminants in utero and through
breast milk(72). The developmental consequences of exposure to high concentrations of
organo-chlorinated contaminants include intrauterine growth retardation (IUGR),
shortened or prolonged gestational lengths, low birthweight, congenital malformations,

and spontaneous abortion (73-77).

12

Studies on male reproductive endpoints have shown reduced fertility in men
working in pesticide manufacturing plants (78). Carlsen and co-workers (79) reported a
decline in semen quality, compared with published information on semen quality
appearing in the literature over the preceding 50 years. Recent female reproductive
studies (80, 81) have shown consumption of contaminated sport ﬁsh was associated with
shorter menstrual cycles but there was no evidence of having a large detrimental effect on
time to pregnancy. Now new concerns are being raised that chemicals in the ﬁsh may
spread the onset of nervous system problems in the elderly. The body of research data
has demonstrated an age-related decline in cognitive and motor function in humans(82,
83). Because of this decline, the aged may be at increase risk for neurological

dysfunction following prolonged exposure to neurotoxicants, such as PCBs.

1.7 The hypotheses and objectives of the present study

Due to advantageous physical and chemical properties, PCBs were once used
extensively in a variety of applications. As a consequence of this use, accumulation and
storage of PCBs would theoretically make them available, under appropriate conditions,
for re-circulation within the body allowing continuous or repeated exposure of target
organs long after the original dose has ceased or when multiple low doses are received
and accumulated to a critical level. Therefore, human consumption of chemically
contaminated ﬁsh poses a potential health risk, the magnitude of which depends on the
amount of ﬁsh consumed and the degree of contamination. Evaluation of the potential
risks to populations that may be exposed to chemically contaminated ﬁsh requires

knowledge of the patterns and frequencies of ﬁsh consumption by these populations.

13

The objectives of the present study are: l) to examine ﬁsh consumption behaviors
measured by quantity of ﬁsh consumed, ﬁsh preparation, ﬁsh cooking methods, favorite
ﬁsh eaten, and changes in ﬁsh consumption behaviors among Michigan anglers over
time; 2) to describe variation in PCB levels in human serum and patterns of change in
study population; and 3) to integrate questionnaire information and laboratory ﬁndings on
human serum to evaluate the relationship between ﬁsh consumption and contaminant
levels in human serum and their joint change pattern over about 20 years. We
hypothesize that there has been a decrease in consumption of sport caught Great Lakes
ﬁsh among ﬁsheaters. Because the health advisory of ﬁsh consumption tends to target
women, we further hypothesize women reduced their total sport ﬁsh consumption more
than men. We hypothesize that elevation of PCB levels is largely due to consume of
sport caught Great Lakes ﬁsh. We expect to see PCB levels are higher among ﬁsheaters
than nonﬁsheaters and change in PCB levels is associated with concurrent sport caught
Great Lakes ﬁsh.

The present study will provide the opportunity to study human exposure to a
synthetic chemical removed from once common use and provide a signiﬁcantly historic

foundation for research testing health outcome hypotheses.

2 Methodology

2.1 Population and setting

2.1.1 Study population

The Michigan Department of Community Health (MDCH) has investigated
human exposure to waterborne contaminants, especially PCBs, for more than 25 years.
The data presented here were collected by MDCH from three cross-sectional surveys
between 1973 and 1991. The subjects of these surveys were not a random sample of the
general population residing along the whole shoreline of Lake Michigan. Instead, the
target population was licensed sport ﬁsh anglers. They were selected for study if their
recreational sport-caught Great Lakes ﬁsh consumption was signiﬁcantly high (24 pounds
or more sport ﬁsh consumed yearly) or if their sport caught Great Lakes ﬁsh consumption
was less than six pounds per year. The former comprised the ﬁsheater group; the latter,
the nonﬁsheater group. This strategy was deliberately designed to evaluate signiﬁcant
exposures which then would amplify the association between ﬁsh consumption and
human serum PCBs burden if the association was to exist. Study participation was
voluntary and without compensation. All participants signed an informed consent

statement approved by the MDCH.

l4

15

2.1.2 Time windows

2.1.2.1 1973-74 ( survey I)

The ﬁrst pilot survey in a series of MDCH sponsored studies, initiated in 1971,
focused on mercury exposure.

This was followed in 1973-74 by a study funded by the Food and Drug
Administration (FDA) to investigate human exposure to PCBs from consumption of sport
caught ﬁsh from Lake Michigan in 1973-74, which was the ﬁrst cross-sectional PCB
survey in Michigan anglers. The 1973-74 survey recruited 156 individuals with a median
age of 46 years, with a range of 37 (25 percentile) to 56 (75 percentile). Four major
communities along the western Michigan shoreline of Lake Michigan were involved in
the ﬁrst survey, South Haven, Ludington, Manistee, and Traverse City. These
communities represented the principal areas of Lake Michigan where PCB contaminated
ﬁsh had been identiﬁed. Algonac, a shoreline community on Lake St. Clair where PCB
levels in ﬁsh were not excessive, was also included, with a primary intention to make a

geographic comparison.

2.1.2.2 1979-82 ( survey H)

Additional funding from the Environmental Protection Agency (EPA) in 1979-81
was used to implement a large multi-site PCB survey in Michigan anglers residing in
Western Michigan along the shoreline of Lake Michigan. Eleven major communities
representing points of access to the lake were selected as regional study sites. These

communities spanned the entire shoreline of Lake Michigan. The sample size was

l6

enlarged to 1255 individuals. Participants in survey I were re-contacted and encouraged
to enroll in survey 11. 115 participants identiﬁed in survey I continued to participate in
survey 11. The remaining 1140 were new participants in the series of Michigan PCB

surveys.

2.1.2.3 1989-91 ( survey III)

Seven hundred and twenty eight persons participated in survey III. Each subject
from survey 11 was re-contacted and invited to participation. Nearly 57 percent
(717/1255) agreed to participate in the third survey, which accounted for 98 percent of
(717/728) participants in the third survey. Included among survey III participants were 64
persons who had participated in both surveys I and survey 11. Figure 2 illustrates the

sample sizes from these three cross-sectional surveys.

2.1.3 Recruitment

Participants were persons who have resided in selected communities for at least 1
year and were between the ages of 18 and 72. People who did not ﬁt the age range or
criteria of ﬁsh consumption, they did not want to be included, or did not wish to donate a
blood specimen for testing, were excluded from surveys.

All subjects (ﬁsheater or nonﬁsheater) in survey I and all ﬁsheaters in survey II
were directly selected by Michigan Department of Public Health ﬁeld workers. Field
workers went to boat launch sites, piers, bait shops, sporting goods shops, ﬁshing clubs to
ﬁnd participants. In order to speak to avid sport anglers, a ‘snowball approach’ was used
in recruiting potential participants. Interviewees were asked to give leads to other persons

who might also be interested in participation in the study, including his/her family

17

members. All potential participants were personally interviewed to determine their
eligibility (either as ﬁsheaters or as nonﬁsheaters). Those judged eligible, as determined
by the response to the brief ﬁsh consumption questions, were invited to participate in the
study.

Eleven Western Michigan communities involving in survey 11 were grouped into 3
regions by MDCH: the northern region went from Petoskey to Pentwater, the middle
region went from Montague to Grand Haven, and the southern region included
communities from Holland to Benton Harbor. A Michigan map displays the 11
communities in Figure 3. A random digit dialing method was performed in each of the
regions to recruit nonﬁsheaters for survey II. Over 1,000 households were contacted and
screened for eligible anglers. Four hundred and nineteen agreed to participate, creating a

regional comparison group for the ﬁsheater group in survey 11.

2.2 Data collection instruments

The questionnaires used in three surveys were mainly composed of two modules,

a ﬁsh consumption module and a general information module.

2.2.1 Fish consumption assessment

The ﬁsh consumption module contained detailed questions about sport caught
Great Lakes ﬁsh consumption behaviors. The main questions included that the number of
years eating sport caught Great Lakes ﬁsh, annual number of ﬁsh meals, usual portion
serving size, species of ﬁsh most frequently consumed (e. g. trout, salmon, perch, catﬁsh,
Whiteﬁsh, etc.), the most frequent cooking methods (e. g. pan-frying, deep-frying, broiling,

baking, etc.), ﬁsh preparation methods (trimming off dorsal or belly fat areas and

l8

removing skin prior to cook), and preservation (freezer or canned). MDCH ﬁeld
interviewers used a plastic ﬁsh model in a dinner plate in order to help individuals
determine more accurately the usual portion size of ﬁsh consumed during a meal. At
each time point of surveys, the interviewer began by asking the respondents about
participants’ ﬁsh consumption habits, speciﬁcally if they had eaten any sport caught Great
Lakes ﬁsh in the preceding year. Only if the respondents had consumed signiﬁcant
amounts of ﬁsh, i.e. greater than 24 pounds of Great Lakes sport ﬁsh, in the preceding 12
months were they asked to complete the ﬁsh consumption module. Participants in survey
11 who consumed little ﬁsh (less than 6 pounds), or did not eat ﬁsh at all in the last year,
completed the demographic module and skipped all the questions relevant to ﬁsh
consumption habits in ﬁsh consumption module.

The ﬁsh consumption module in the questionnaire of survey I did not include
comprehensive information on the usual portion serving size, ﬁsh preparation and
cooking methods. However, 66 percent (103/156) of participants of survey I provided
detailed ﬁsh consumption diaries, which contained type and quantity of ﬁsh consumed,
and how they were prepared and cooked, for every speciﬁc ﬁsh meal over a ﬁshing
season. An average portion serving size was calculated based on all ﬁsh meals for each
participant doing a ﬁsh diary. Preparation and cooking methods were not provided in the
ﬁsh consumption module of the questionnaire of survey II. Four hundred and thirty one
ﬁsheaters in survey II who reported more than six ﬁsh meals per year voluntarily
provided ﬁsh preparation charts to keep track of the type of ﬁsh consumed, preparation

methods, and the cooking methods. All ﬁsh consumption questions were built into the

l9

questionnaire of survey 111. Therefore, in the present study there existed differences with
regard to the sources of detailed ﬁsh consumption information.

Separate questionnaires were developed for ﬁsheaters and nonﬁsheaters in survey
11. The questionnaire for nonﬁsheaters removed the ﬁsh consumption module and made

it much shorter than that for ﬁsheaters.

2.2.2 General information module

The questionnaires included several important demographic and other control
variables that may inﬂuence the outcome of interest. Questions in the demographic
module included date of birth, gender, race, educational attainment, and region where
participants lived. Information on height and weight was reported in survey I and survey
II. Body mass index (BMI) was calculated as weight in kilograms divided by square of
height in centimeters. However, BMI was not available in survey 111, because survey 111
only provided the magnitude of yearly weight change. Participants were asked to report if
they were potentially exposed to PCB related industrial/agricultural chemicals, such as
hydraulic oils, capacitors, transformers, pesticides and so on. Information on self-
assessed health conditions was obtained in surveys I and II. In both surveys, participants
were asked if they were taking any medications for a health condition/illness, if they were
hospitalized in the past year, and if they were currently under the care of a physician for a
particular condition. Lifestyle information, for example, tobacco use and alcohol
consumption, included. Detailed information on smoking and drinking, including the age

of starting smoking and drinking included the age started smoking, the quantity of

20

cigarette smoked daily, the frequency and amount of drinking, etc., was obtained during
the interview of survey III.

Speciﬁc parameters of interest with regard to ﬁsh consumption behaviors and
other relevant variables in the questionnaire stated above have basically remained
unchanged across different surveys, which enhance the comparability of the data
generated across surveys and provide the basis to examine if the patterns of ﬁsh

consumption habits changed over about 20-year follow-up.

2.2.3 Measurement of serum PCB and other contaminants

Serum samples were drawn from the majority of participants, both ﬁsheaters and
non-ﬁsheaters, at all survey points to assess PCBs and nine other contaminants, detected
in the Great Lakes. Levels of serum contaminant were analyzed at the MDCH
Environmental Laboratories in Lansing. Serum PCB levels were evaluated by using the
modiﬁcations of the Webb-McCall packed-column gas chromatography methodology
(94). Quantitation of PCB from electron capture chromatograms was complicated during
1970s because the electron capture detector responds differently to each Aroclor. The
Webb-McCall packed-column method Utilizes several Aroclor standards against which
the quantitative composition of each electron capture chromatograms peaks can be
calculated. Aroclor 1260 is a standardized mixture that permits quantitation of certain
deﬁned sets of Webb-McCall peaks. The same was true for Aroclor 1254 and Aroclor
1016, but they refer to different sets of the Webb-McCall peaks. Most PCB contaminated
samples of ﬁsh, water, sediment, and human serum contain residues characteristic of

several Aroclors. The overlap of Aroclor 1016, Aroclor 1254, and Aroclor 1260 in these

21

three surveys is clariﬁed by a protocol that prescribed which peak and going to be
quantiﬁed by each of the PCB standards. In order to have comparable data on PCB using
a consistent laboratory technique, the Webb-McCall packed-column method was used in
all three surveys.

The detection limit for PCB was 3 parts per billion (ppb). Specimens with serum
PCB levels below the detection limit were assigned a value of 1.5 ppb in the present data
analysis. Only Aroclor 1254 and Aroclor 1260 were utilized as standards in 1973-74. In
laboratory data from surveys II and III, the samples were analyzed for Aroclor 1260 and
nine other contaminants, including DDT, DDE, PBBs, hexachlorobenzene (HCB),
oxychlordane, dieldrin, mirex, mercury, and lead. Levels of total serum lipids were also

tested in surveys 11 and III.

2.3 Data management and preparation

2.3.1 Deﬁnition of ﬁsh eating status

One of the objectives of this study is to compare the ﬁsh consumption and
contaminant levels longitudinally, and we therefore needed to develop a criterion to
deﬁne if a participant was a ﬁsheater or non-ﬁsheater in the study analysis. And this
criterion had to be consistent across the three surveys. The new deﬁnition used in present
study was that in survey I a ﬁsheater was a person who reported eating an average of 24
pounds or more sport caught Great Lakes ﬁsh in the preceding 12 months. Forty-four
persons who annually ate moderate amount of sport caught Great Lakes ﬁsh (less than 24
pounds but greater than 6 pounds) represented as a mild range group in Humphrey’s 1976

report of survey I to the FDA. Applying the deﬁnition, these participants will be deﬁned

22

as a non-ﬁsheaters. They will be treated in the same way as those annually ate less than 6
pounds, and would therefore remain in the analysis as nonﬁsheaters.

For participants in survey II, ﬁsheaters were deﬁned by their ﬁsh eating status,
which based upon whether or not their annual weight of ﬁsh consumed exceeded 24
pounds. Generally, ﬁsh consumption behaviors among nonﬁsheaters in survey 11 were
not available, since the questionnaires were administered separately based on the ﬁsh
eating status. However, information on ﬁsh consumption habits for a very small
proportion of new participants (1.6%) in survey II, who claimed they annually ate more
than 24 pounds but actually ate less than that amount of sport caught Great Lakes ﬁsh,
were also available. Fish eating status was determined at the beginning of the interview
based on participant’s self-reported amount of ﬁsh consumed (24 pound/year or more:
ﬁsheater; 6 pound/year or less: nonﬁsheater). Once participant was classiﬁed as a
nonﬁsheater in survey 11, a nonﬁsheater questionnaire, which did not include ﬁsh
consumption module, would be used. We were not able to know the detailed ﬁsh
consumption behaviors in nonﬁsheaters, including total weight of ﬁsh consumed.
Probably some participants, who classiﬁed as nonﬁsheaters, actually ate quite large
amount of ﬁsh. However, we believe that number should be very small, because in the
process of recruitment of the regional control group, based on the response to the
screening ﬁsh consumption, those self-reported consuming 24 pounds or more were
recruited into the ﬁsheater group instead of the nonﬁsheater group.

Participants in survey 111 were basically a subset of survey 11, as 98 percent
(717/7 28) of participant’s ﬁsh eating status had been decided at the time of either survey I

or II and kept unchanged. The ﬁsh eating status of 11 (728 - 717) new participants in

23

survey 111 were also determined by total weight of ﬁsh consumed the year prior to survey
111 was conducted.

In general, the ﬁsh eating status was based on the participant’s total weight of
sport caught Great Lakes ﬁsh consumption at the ﬁrst time they were enrolled in survey.
Once the participant’s ﬁsh eating status was decided, it would remain the same no matter
whether or not his/her annual total pounds of ﬁsh consumed changed during follow-up

time.

2.3.2 Computation of ﬁsh consumption

Consumption of sport caught Great Lakes ﬁsh was considered as the primary
exposure variable in the present study. In each time point of surveys, the categorical
designation as a ﬁsheater or non-ﬁsheater was used to represent cumulative exposure
status. The current exposure was expressed as two continuous variables: 1) number of
annual sport caught Great Lakes ﬁsh meals-calculated by summing the reported number
of ﬁsh meals consumed in the preceding 12 months; 2) the total weight (pounds) of ﬁsh
consumed--measured by multiplying annual meals of sport caught Great Lakes ﬁsh by
usual portion serving size, assuming the portion serving size was consistent across the
preceding 12 months. The portion serving size itself was ascertained across three-time
point of surveys and provided the basis of looking into the change pattern too. For the
majority of nonﬁsheaters, with no information on ﬁsh consumption, their annual ﬁsh
meals and total weight of ﬁsh consumed were set to zero in the multiple linear regression

analyses to estimate the prediction equations of contaminant levels.

24

2.3.3 Choice of Aroclor

Aroclor 1254 was used as a standard in survey I but not in survey II or survey 1]].
Aroclor 1016 was used as a standard in survey II but not in survey I or survey III. Since
only Aroclor 1260 used as a quantitative standard in the Webb-McCall packed gas-
columns laboratory technique consistently over the three surveys, it was used to examine
the pattern of changing PCB levels across the surveys.

Actually, data from serum PCB concentration analysis do not reﬂect the analysis
of a speciﬁc Aroclor mixture, for example, Aroclor 1260. Rather, the serum contains a
mixture of PCB congeners that is different from that in chemist to produce a standard
mixture of PCBs that could reﬂects the ever changing components of the serum mixture
of PCBs. The laboratory selects a particular Aroclor mixture (or a set of Aroclors) to
serve as a standard so that the chromatographic peaks can be translated into micrograms
of total PCBs. The numbers that are generated from this procedure are rather arbitrary
but since the same procedure was used for all samples (i.e., Aroclor 1260 was used as a
standard) the numbers are comparable. So it can be used with some conﬁdence to

estimate the variation of total PCBs in serum.

2.3.4 Integration of data within each survey

Within each survey, information from the questionnaire was merged with
laboratory data by matching on each participant’s unique identiﬁcation number.
Questionnaire information in survey I was obtained from its pilot study, and from the
main study phases of survey I during 1973 to 1974. Questionnaires used in the pilot study

and the main study were basically the same. There were 241 records in survey I with 156

25

unique identiﬁcation numbers. Seventy-one participants had one record, which provided
questionnaire information through either the pilot or the main study at survey time. All
these 71 records were included in this analysis. The other eighty-ﬁve participants had
two records collected in both pilot and main study. There were at least three choices of
which data to use from the participants who have both main and pilot data --1) simply
choose main study records; 2) simply choose pilot study records; 3) combine information
from two studies, and average all the key variables. For instance, calculate the mean of
annual total ﬁsh meals consumed based on pilot and main questionnaire responses.
Using the average value could increase the precision of measurements for an individual
by reducing the variance. From an individual point of view, it may better represent the
‘true’ value. However, these 85 records would be not comparable to 71 records from
participants who only provided one record because of the effect of regression toward the
mean. Therefore, the third choice was ruled out.

In order to make a rational choice, a reproducibility analysis was performed for
the key variables. These variables were general characteristics such as sex, height,
weight, education attainment, as well as main ﬁsh consumption variables including
portion serving size, and most common species of ﬁsh consumed within this short period
of time. Information was grouped into categorical variables and continuous variables.
Kappa values were calculated for categorical variables to assess the agreement between
pilot and main studies. Pearson correlation analyses were performed for continuous
variables. The results of the reproducibility analysis are displayed in Appendix 1.
Excellent agreement was observed in general characteristics such as gender, height and

weight. Kappa values for categorical key ﬁsh consumption variables (e.g. the 3 most

26

common species of ﬁsh consumed) showed good agreement. All Kappa values were
greater than 0.4. The differentiation was observed within one-degree category in a cross-
tabulation. In sum, we considered there was good agreement between two measurements
collected from pilot and main studies. In the present data analysis, information from
main study of these 85 participants was selected. In combination with the rest of the 71
records, a total of 156 records were used in further analysis as questionnaire data of
survey 1.

Many of participants in survey I provided multiple samples of blood. Some of
them gave as many as 9 blood specimens during 1973 and 1974. The original objective
in study design was to test a seasonal variation of the PCB levels. Humphrey (1976)
examined the survey data and concluded that there are no statistically signiﬁcant changes
in PCB blood levels during the survey period. Because there were no apparent short-term
change was observed, the ﬁrst measure value of each participant was used in the present
data analysis to represent the participant’s PCB body burden at the time survey I
conducted.

After merging within surveys, the number of participants with questionnaire and

laboratory data in surveys I, II, and 111 year were 153, 1093, and 413, respectively.

2.3.5 Integration of data between surveys

Due to losses to follow-up and changes in sampling strategy over time, varying
numbers of individuals bad data for multiple time points. Data between surveys were
merged by matching the unique identiﬁcation number of the participant. One hundred

fourteen participants had questionnaire and laboratory data for both surveys I and II; 43

27

participants had complete data were available for surveys I and III; 391 participants had
complete data for survey II and III. A total of 462 participants had complete data from at

least two surveys; forty-three participants had full information across all the three survey

points.

2.4 Statistical methods

2.4.1 Univariate analysis

Univariate analyses stratiﬁed by gender were performed to provide a summary of
descriptive statistics for the study population. With 73 percent (115/156) of participants
of survey I re-enrolled in survey H, the overwhelming proportion of participants in survey
111 having participated in survey H, and survey II having the largest sample size, survey 11
data could be treated as a representative of the study population. In the present study, the
demographic description of the study population was therefore only performed for survey
11 data. Age was calculated from the participant’s birth date given during survey 11
interview. Only survey III data provide information on tobacco use and alcohol
consumption, therefore the information on these variables was based on survey III data
only.

Within each survey, univariate analysis of ﬁsh consumption variables was
restricted to the exposure group (ﬁsheaters) as no information on ﬁsh consumption was
available in the majority of non-ﬁsheaters. Mean, standard deviation, median, and range
(25, 75 percentiles) were calculated for continuous ﬁsh consumption variables, such as
annual ﬁsh meals, usual portion size in ounces, total pounds of ﬁsh consumed in the past

year. Frequencies were used to describe the categorical ﬁsh consumption variables, such

28

as, the most common species of ﬁsh consumed, common ﬁsh preparation and cooking
methods. Mean, Standard deviation, median and range (25, 75 percentiles) were
calculated for laboratory measurements stratiﬁed by gender and by ﬁsh eating status at
each survey point. Laboratory values of all contaminant levels reported as less than the
detectable limit would be coded as the middle value between zero and the detect limit.
For instance, where the detect limit for PCB was 3 parts per billion (ppb), all the values
reported under the detectable limit were set to 1.5 ppb. .

All tests for statistical signiﬁcance were set at the two-sided 0.05 level.
Differences in means were examined by using Student T test for two group comparisons,
one-way AN OVA was used to compare for 3 or more groups. The linear associations
were also examined. Distribution frequencies of occurrences were tested using the Chi-
square test. Estimates of the mean and median values for contaminant levels were both
available from the analysis; when the data appeared log-normally distributed, the median
provided a better sense of the center of the contaminant concentrations. In addition,
median estimates are more resistant to the effect of outlying observations than are means.
Tests on contaminant levels were performed both on natural log scale and on an
arithmetic scale to evaluate the sensitivity of the test when different scales were used.

For the cross-sectional analysis at each survey point, the data were analyzed as
follows: 1) ﬁsh consumption and contaminant levels were categorized as none, low,
medium and high. The Chi-square test was employed to evaluate relationship between
ﬁsh consumption factors and general characteristics such as age, education, region, body
mass index (if applicable), and self-reported chemical exposure; 2) Pearson or Spearman

correlation coefﬁcients were generated to assess the direction and strength of possible

29

associations between contaminant levels and continuous ﬁsh consumption variables.
Subsequently, simple linear regression was use to examine the relationship between

laboratory measurements and continuous ﬁsh consumption variables.

2.4.2 Multivariate analysis

Multiple linear regression analyses were performed to identify signiﬁcant
predictors of serum contaminant levels after controlling for potential confounders by
including these factors in models. The primary exposure variable in models was sport
caught Great Lakes ﬁsh consumption. As described above, cumulative exposure was
expressed as a categorical variable, i.e. ﬁsh eating status (ﬁsheater vs. nonﬁsheater).
Current exposure was described as two variables, i.e. annual number of ﬁsh meals and the
total weight of ﬁsh consumed in the preceding year. The collinearity of these variables
needed to be considered

Model construction was initiated with a basic regression equation that included
age, gender, education, and region as predictor variables. After selection of the most
appropriate ﬁsh consumption variable between annual ﬁsh meals and total weight of ﬁsh
consumed, additional secondary exposure variables, such as body mass index and self-
reported chemical exposure, were examined. The selection of variables to be evaluated in
the regression models was based upon biological relevance to the outcome of interest-
contaminant levels and demonstrated association with the outcome of interest in
preliminary bivariate analyses. Instead of using annual ﬁsh meals as a continuous
variable, quartiles of annual ﬁsh meals were used in prediction equations of contaminant

levels within each survey. The reason for this choice is that annual ﬁsh meals is not a

30

constant continuous variable; the same number of annual ﬁsh meals could represent
different quantities of ﬁsh among different ﬁsheaters. The same was true for total weight
of ﬁsh consumed. Once the model was determined, standard diagnostic techniques were
used to examine the goodness of ﬁt of the ﬁnal models and the validity of the underlying
assumptions of linear regression.

Pairwise comparisons (survey I versus 11; survey 11 versus 111; survey I versus III)
were performed for participants with two-time point survey data. Paired student t-tests
were used for continuous variables to assess whether a statistically signiﬁcant difference
between two surveys had occurred with regard to quantity of ﬁsh consumption, ﬁsh
consumption behaviors, and contaminant levels. The McNemar test or Wilcoxon rank
sum test was generated for categorical variables, such as the commonly eaten ﬁsh species,
ﬁsh preparation practices, and common cooking methods. The difference of ﬁsh
consumption and contaminant levels between two time point measurements was
calculated to express the magnitude of change. Furthermore, regression analyses were
used to assess the change in ﬁsh consumption over time adjusted for age, gender,
education, region, baseline body mass index, and baseline contaminant levels. The same
approach would be used to assess the change in the population contaminant levels over
time adjusted for ﬁsh consumption as well as potential confounders stated above.

For participants with data from at least two-time points, time trends in ﬁsh
consumption, in Aroclor 1260 levels, and their joint change pattern were examined. This
analysis is complicated by the dependence among repeated observations made on the
same participant, data were analyzed by Proc Mixed with the statement ‘repeated’ in

SAS.

3 Findings
3.1 Descriptive results and bivariate analysis

3.1.1 Background characteristics of study population (Tables 1- 3)

Nearly 49 percent (610/1255) of participants were ﬁsheaters, of whom male
ﬁsheaters were twice more than female ﬁsheaters (66.3 percent versus 33.7 percent,
p<.001) (Table 1). This difference may reﬂect that annually more ﬁshing licenses issued
to male than to females. The mean age of ﬁsheaters was three years older than that of
nonﬁsheaters (46 versus 43, p<.001), reﬂecting more nonﬁsheaters aged 15-24 and more
ﬁsheaters above age 45. Fisheaters education levels were lower than those of
nonﬁsheaters. Overall, 17 percent of ﬁsheaters reported having less than a high school
level education, compared with less than 15 percent of nonﬁsheaters. Fisheaters were
taller and heavier than nonﬁsheaters and they were more likely to report having been
exposed to chemicals at the work place or at home. About the same proportion of
ﬁsheaters and nonﬁsheaters lived in the three regions of Michigan, i.e. Northern, Middle,
and Southern established by the sampling scheme.

Once stratiﬁed by gender background characteristics of the study population,
some difference between ﬁsheaters and nonﬁsheaters became non-statistically signiﬁcant.
There were no signiﬁcant differences observed with regard to educational attainment,
region, height, and self-reported exposure to any chemicals between male ﬁsheaters and
nonﬁsheaters (Table 2). Male ﬁsheaters were a little older and somewhat heavier than

nonﬁsheaters. Nonﬁsheaters tended to be more highly educated but it was not

31

32

signiﬁcant (p=0.07). In females, ﬁsheaters’ age is older than that of nonﬁsheaters
(p=0.04). Overall, 52 percent of ﬁsheaters were 45 or more years of age, compared with
44 percent of nonﬁsheaters (T able 3). The association between ﬁsh eating status and
education in ﬁsheaters was signiﬁcant (p=0.04), reﬂecting more nonﬁsheaters having a
high school education or above. No differences were observed between ﬁsheaters and

nonﬁsheaters in terms of region, height, weight, and self-reported chemicals exposure.

3.1.2 Characteristics of tobacco use and alcohol consumption in study population

(Tables 4, 5)

The proportion of never smoking was signiﬁcantly higher in male nonﬁsheaters
than that in ﬁsheaters, 36 percent and 22 percent, respectively (Table 4). Among male
participants who reported current or prior smoking, the average age of starting smoking
was 18 years. Prior smokers in both the ﬁsheater group and nonﬁsheater group reported
that the mean age of they last smoked regularly was about 38 years of age. Current
smokers, whether ﬁsheaters or nonﬁsheaters, reported smoking about 25 cigarettes a day.
Seventy seven percent of participants were current alcohol consumers, similar in both
ﬁsheaters and nonﬁsheaters. However, abstainers were more frequent among male
nonﬁsheaters than among ﬁsheaters. Fisheaters were more likely to be beer drinkers, and
they usually drank more bottles/cans beer at a sitting than did their nonﬁsheater
counterparts. Reported drinking of wine was more likely in nonﬁsheaters, regardless of
whether judged by drinking frequency or by the quantity consumed at a sitting. No
difference was observed in terms of liquor drinking habits. Fish consumption status did

not appear to be a signiﬁcant determinant of tobacco use or overall alcohol consumption

33

among women, but was associated with beer drinking (Table 5). Nearly 37 percent of
female ﬁsheaters reported they drank beer more than once a week, compared with 22

percent of female nonﬁsheaters.

3.1.3 Fish consumption behaviors

3.1.3.1 The ﬁrst survey (Tables 6, 7):

One hundred and ﬁfty six persons participated in 1973-74 survey, of whom 74
(47.4%) reported they annually ate 24 or more pounds of sport caught Great Lakes ﬁsh.
These 74 constituted the ﬁsheater group. The median number of annual ﬁsh meals
reported consumed by male ﬁsheaters is 56, which less than that for female ﬁsheaters, 72
(Table 6). However, male ﬁsheaters tended to eat a larger quantity of ﬁsh at each meal
than did female ﬁsheaters. The median usual portion serving size was 12.5 ounces for
male and 9.0 ounces for female ﬁsheaters. As a consequence, the total annual weight of
ﬁsh consumed by female ﬁsheaters was very similar to that consumed by male ﬁsheaters,
with the median total weight of ﬁsh consumed in the previous year being about 40 pounds
for both males and females.

Fish consumption behaviors regarding the most common ﬁsh species eaten, as
well as cooking and preparation methods are displayed in Table 7. Almost all ﬁsheaters
cited trout as one of their three favorite species of ﬁsh consumed. Nearly 65 percent
referred to salmon and over 30 percent to perch. Less than a quarter of participants
mentioned Whiteﬁsh or catﬁsh. Similar patterns were observed in both male and female
ﬁsheaters. Since every participant could report up to three most favorite ﬁsh species, the

total proportion does not sum to 100. 50.8 percent of ﬁsheaters preferred pan-frying to

34

other cooking methods. Less than one-third of ﬁsheaters cited broiling, or baking as one
of their three favorite cooking methods. Nearly eight percent of male ﬁsheaters, but no
female ﬁsheaters referred to deep-frying. The proportions of ﬁsheaters reporting always
trimming off fatty areas and removing skin in ﬁsh preparation prior to cooking were
uniformly small for both males and females. No ﬁsheaters reported they always removed
ﬁsh skin prior to cooking. 9.5 percent of female ﬁsheaters mentioned that they trimmed
dorsal or belly fat areas prior to ﬁsh preparation, but 5.3 male ﬁsheaters reported

trimming off belly fat area and none reported trimming of dorsal fat area.

3.1.3.2 The second survey (Tables 8, 9):

Six hundred and ten out of 1255 participants (48.6 %) were ﬁsheaters. As shown
in Table 8, the proportion of ﬁsheaters was higher in men than in women, 66 % versus
34%. Differences between men and women were observed with respect to total ﬁsh
meals in a year and usual portion serving size. Female ﬁsheaters reported a higher
median number of annual ﬁsh meals consumed than did male ﬁsheaters (58 versus 53),
but the median usual portion serving size among women was one-third smaller than
among their male counterparts (8 ounces versus 12 ounces). Based on the total weight of
ﬁsh consumed in the preceding year, annual ﬁsh consumption for female ﬁsheaters was
lower than that of male ﬁsheaters, 33 pounds compare to 40 pounds.

Trout and salmon were most frequently cited by ﬁsheaters as one of the 3 most
common species of ﬁsh consumed in survey II (Table 9). Forty two percent of ﬁsheaters
named perch, only 2 percent mentioned catﬁsh, and none referred to Whiteﬁsh. 65 percent

of male ﬁsheaters reported that pan-frying was one of their three favorite cooking

35

methods, followed by broiling and deep-frying. None mentioned baking. Over half of
female ﬁsheaters cited broiling and pan-frying as one of their 3 most frequent cooking
methods. About one-third percent mentioned deep-frying, only 3.3 percent named
baking. With regard to ﬁsh preparation, 86.1 percent of ﬁsheaters reported they always
trimmed away the dorsal and belly fat area, and nearly half of ﬁsheaters reported they

always removed ﬁsh skin prior to cooking.

3.1.3.3 The third survey (Tables 10, 11):

Seven hundred and twenty eight persons participated in survey III (1989-91), of
whom two-thirds were men; 49.3 percent (359/728) of the sample were classiﬁed as
ﬁsheaters. The median number of male ﬁsheaters’ annual ﬁsh meals, 34, was
substantially higher than for females, 27. And the median usual serving size for males
was larger than for females, 8 ounces versus 6 ounces. The total weight of annual ﬁsh
consumption in male ﬁsheaters was higher than in female ﬁsheaters, 18 pounds compare
to 11 pounds (T ablelO).

Fish consumption behaviors in ﬁsheaters on survey III displays in Table 11.
Around 50 percent of ﬁsheaters picked trout, salmon, or perch as one of the three most
frequently eaten ﬁsh, but everyone mentioned Whiteﬁsh. Concerning cooking methods,
pan-frying was most commonly mentioned as one of top 3 frequently used cooking
methods. Approximately 30 percent of ﬁsheaters picked baking, broiling, and deep-
frying. With reference to ﬁsh preparation prior to cooking, about 90 percent of ﬁsheaters
reported they always trimmed away dorsal and belly fatty area, while 82 percent

mentioned removing ﬁsh skin prior to cooking.

36

Figure 4 illustrates quantity of ﬁsh consumption among ﬁsheaters over the three
surveys. Figure 5 displays proportions mentioning as one of the top three species of ﬁsh

consumed over the three surveys. Figure 6 shows ﬁsh preparation methods by survey.

3.1.4 Laboratory measurements

3.1.4.1 Laboratory data at survey I (Table 12):

One hundred sixty three persons donated blood samples in survey 1, of whom
male participants accounted for 47.9 (78/ 163) percent. 47.2 (77/163) percent were
ﬁsheaters. One participant’s gender was unknown. Male participants had higher median
levels of Aroclor 1254 in serum than did females, 35 ppb and 23 ppb, respectively (Table
12). The same pattern appeared for Aroclor 1260. The median levels of Aroclor 1260
were 12.5 ppb for men, 6.5 ppb for women. When PCB levels were stratiﬁed by ﬁsh
eating status, Aroclors 1254 and 1260 serum levels in ﬁsheaters were considerably higher

than in nonﬁsheaters of both genders.

3.1.4.2 Laboratory data at survey H (Table 13):

In survey II, serum samples were tested for the presence of levels of Aroclor 1260,
DDT, DDE, PBB, Aroclor 1016, hexachlorobenzene (HCB), beta-isomer of hexachloro-
cyclohexane (Beta BHC), oxycholrdane, trans-nonachlor, and dieldrin. The contaminant
levels in men were consistently higher than females with regard to Aroclor 1260, DDT,
DDE, and PBB. The median levels of Aroclor 1260 in men and women were 16.3 ppb
and 8.3 ppb, respectively. Median levels of PCB total were higher in men 17.6 ppb than

in women 9.0 ppb . Levels of 22.5 ppb of DDT, 18.9 ppb of DDE, 1.9 ppb of PBB were

37

found in men, compared with 14.2 ppb of DDT, 12.2 ppb of DDE, and 1.2 ppb of PBB in
women. Aroclor 1260, PCB total, DDE, and DDT values were all signiﬁcantly higher in
ﬁsheaters than in nonﬁsheaters for both sexes. After stratifying by ﬁsh eating status,
male ﬁsheaters had the highest contaminant levels of Aroclor 1260, DDT and DDE;
female nonﬁsheaters had the lowest levels of these contaminants, while contaminant
levels in female ﬁsheaters were higher than male nonﬁsheaters but lower than those in
male ﬁsheaters. However, PBB levels did not show this pattern. Since very few serum
samples had detectable levels of other contaminants, such as Aroclor 1016, HCB, beta
BHC, oxychlordane, trans-nonachlor, and dieldrin, the results on these contaminants were

not presented here and would not be used in any further analysis.

3.1.4.3 Laboratory data at survey 111 (Table 14):

The median levels of contaminant concentrations and ranges of contaminants
detected in human serum samples at survey III are displayed in Table 14. The
contaminant levels in men were consistently higher than in women in terms of all the
contaminant levels tested in the survey, namely Aroclor 1260, DDT, DDE, and PBB. The
median levels of contaminants in men were 13.1 ppb of Aroclor 1260, 12.6 ppb of DDT,
11.4 ppb of DDE, and 2.8 ppb of PBB, compared with 6.8 ppb Aroclor 1260, 7.8 ppb
DDT, 7.0 ppb DDE, and 1.8 ppb PBB in women, respectively. Values of total PCB in
survey 111, as measured by Aroclors 1260 and 1016 as standards, were almost the same as
Aroclor 1260, because Aroclor 1016 was hardly found in participants’ serum specimens
at that time. Again, all contaminant levels were higher in ﬁsheaters than in nonﬁsheaters,

the highest levels were always found in male ﬁsheaters.

38

Figure 7 illustrates serum levels of Aroclor 1260 stratiﬁed by gender and ﬁsh

eating status over the three surveys.

3.1.5 Eﬁ‘ect of general characteristics on the main ﬁsh consumption behaviors

(Tables 15 - 19)

A borderline signiﬁcant linear trend between increasing age and increasing mean
number of annual ﬁsh meals was demonstrated in male ﬁsheaters (linear trend F=4.03,
p=.05), but not in females. There was no linear trend between increasing age and
increasing ﬁsh consumption with respect to the total weight of ﬁsh consumed in males or
females (Table 15).

Education was signiﬁcantly associated with ﬁsh consumption as measured by
number of annual ﬁsh meals or total amount (pounds) of ﬁsh consumed. Fisheaters who
reported having high school education or above ate less ﬁsh than those did not attain high
school education level. This was especially true for female ﬁsheaters (Tables 16).

Male ﬁsheaters in the southern region reported eating less annual ﬁsh meals than
those in the other two regions, whereas for women there was more of a north to south
consumption in ﬁsh meals yearly (Linear trend F=3.44, p=.06 for men; Linear trend
F=5.32, P=.02 for women). Male ﬁsheaters living in the middle shoreline of Lake
Michigan reported eating larger amount (pounds) of ﬁsh than those in either the south or
north. Female ﬁsheaters, however, consumed more pounds of ﬁsh among in the northern
region (Table 17).

Body mass index in male ﬁsheaters was signiﬁcantly associated with ﬁsh

consumption in terms of annual ﬁsh meals (chi-square=44.14, p<.001) as well as total

39

weight of ﬁsh consumed (chi-square=42.90, p<.001). In male ﬁsheaters, but not in
females, a signiﬁcant linear trend was also shown between body mass index and the total
weight of ﬁsh consumed although the upward trend was irregular (Linear trend F=5.50,
p=.02) (Tables l8-a, l8-b).

Self-reported chemical exposure status was not associated with the quantity of

ﬁsh consumption (Table 19).

3.1.6 Effect of general characteristics on contaminant levels (Tables 20-24)

After stratifying by gender and ﬁsh eating status, the means of Aroclor 1260,
DDT, and DDE levels in ﬁsheaters were two to four times higher than in nonﬁsheaters at
each age group. The magnitude of differences in these contaminant levels across
different age groups was greater in males than in females. Linear trends were established
between age and these contaminant levels both in ﬁsheaters and nonﬁsheaters. No linear
association between ﬁsheater status and PCB levels or between age and PBB levels was
found (F=2.38, p=. 12) (Table 20).

A weak inverse association between education levels and mean Aroclor 1260
levels was demonstrated in ﬁsheaters (F=5.0, p=.03 for male ﬁsheaters; F=l3.0, p<.001
for female ﬁsheaters) as well as male nonﬁsheaters (F=4.66, p=.03). This effect
pronounced most when comparing college education and non-college educated men.
Education level appeared to be a signiﬁcant determinant of mean DDT and DDE levels
both in female ﬁsheaters (DDT: linear trend F=18.91, p<.001; DDE: linear trend
F=2l.38, p<.001) and female nonﬁsheaters (DDT: linear trend F=l3.6l, p<.001; DDE:

linear trend F=l4.82, p<.001), but not in male participants. Female nonﬁsheaters with

40

higher levels of education were more likely to have slightly lower PBB levels than less
educated female nonﬁsheaters (linear trend F=5.33, p=.02). Such a linear trend between
education and levels of PBB was found neither in ﬁsheaters of both sexes nor in male
nonﬁsheaters (Table 21).

Region was signiﬁcantly associated with DDT levels both in male ﬁsheaters
(linear trend F=20.27, p<.001) and male nonﬁsheaters (linear trend F=4.68, p=.03) with
the highest levels in northern Michigan. Mean DDT levels were twice as high in
Northern than southern residents. The linear trends with respect to DDE and Aroclor
1260 levels were pronounced only in male ﬁsheaters. Trends were similar, but not
statistically signiﬁcant in females. Mean PBB levels were strongly associated with
region, regardless of gender and ﬁsh eating status. Participants who lived in middle part
shoreline of Lake Michigan had the highest levels of PBB, compared with those lived in
northern and southern region (Table 22). The results were consistent with greatest
amount of affected PBB farms in the middle region.

Body mass index was signiﬁcantly linearly related to DDT and DDE levels in
both sexes, regardless of ﬁsh eating status. The same pattern was observed in 1260
levels for men but not for women. An inverse linear trend between body mass index and
levels of PBB was shown in female ﬁsheaters (F=7.32, p=.01) (Tables 23-a, 23-b).

No association between self-reported chemical exposure status and these four

contaminant levels were found (Table 24).

41

3.1.7 Relationship between ﬁsh consumption and contaminant levels (Tables 25, 26)

Signiﬁcant linear trends were demonstrated between ﬁsh consumption both in
terms of annual ﬁsh meals (T ables 25-a, 25-b) and total weight of ﬁsh consumed (Table
26-a, 26-b) and Aroclor 1260, DDT, DDE levels. Similar patterns were observed in men

and women. There was no such trend in PBB levels.

3.1.8 Pairwise comparisons of individuals surveyed more than once

3.1.8.1 Survey 1 versus survey 11 (Tables 27-29)

Fifty-two men and 63 women participated in survey I and survey II. Paired
comparisons stratiﬁed by gender are displayed in Table 27. Information on quantity of
ﬁsh consumption was available in 55 ﬁsheaters-34 males and 21 females. In male
ﬁsheaters, a small non-signiﬁcant decrease in ﬁsh consumption was observed with'regard
to total ﬁsh meals and also a small decreasing in usual portion size. Total weight of ﬁsh
consumed in a year did decline signiﬁcantly (paired t=2.0, p=0.05). In female ﬁsheaters,
time change was strongly associated with ﬁsh consumption behaviors. The means annual
ﬁsh meals were 94 in survey I and 61 in survey II (paired t=3.88, p=0.01). A
considerable decrease was presented in mean portion serving size, which changed from
10.3 ounces in survey I to 7.9 ounces in survey II (paired t=7.9, p=0.01). The mean total
weight of ﬁsh consumed in the preceding year decreased nearly 50 percent-from 60.8
pounds to 33.1 pounds in females (paired t=3.43, p=0.01).

Data on surveys I and 11 indicated an overall similar proﬁle in male and female
ﬁsheaters with respect to ﬁsh preparation and cooking methods, and favorite ﬁsh species

consumed. The proportions of ﬁsheaters reporting trimming away dorsal and belly fatty

42

area were low in survey I (under 10 percent), but substantially higher in survey II (over 50
percent) (Table 28). None of ﬁsheaters reported removing ﬁsh skin in survey 1, but over
20 percent reported doing so in survey 11. With regard to cooking methods, pan-frying
was the commonest procedure cited by males in both surveys and by females in survey 1.
Women also picked baking as one of the three most common cooking methods, but no
signiﬁcant difference was observed between surveys I and II data. A trend towards less
baking and more pan-frying was seen in both genders. Trout and salmon were
consistently picked by ﬁsheaters as their preferred ﬁsh species in both surveys. A slight
shift from trout to salmon was seen in both men and women--the proportion citing trout
being one of top 3 went down (McNemar test p=0.03 for men, p=0.06 for women), and
the proportion citing salmon went up but there is no statistical signiﬁcance. Although
approximately 20 percent of subjects listed Whiteﬁsh as preferred in survey 1, none of
ﬁsheaters named Whiteﬁsh as one of top 3 ﬁsh species in survey 11, a reduction was
signiﬁcant for men (P=.004).

An upward shift of serum Aroclor 1260 was observed between the two surveys in
the 58 men and 65 women for whom serum data were available. Results presented in
Table 29. Over this period, levels of Aroclor 1260 signiﬁcantly increased from 21.3 to
34.0 ppb for men, 8.3 to 15.2 ppb for women (Paired t test: t=4.26, p<0.001 for men;

=6.61,.p<0.001 for women). These signiﬁcantly increased changes in Aroclor 1260 were
pronounced in both sexes, regardless of ﬁsh eating status. The magnitude of increase in
Aroclor 1260 was higher in ﬁsheaters than in nonﬁsheaters, and was higher in men than

in women.

43

3.1.8.2 Survey II versus survey 111 (Tables 30-32)

Substantial reductions in the overall mean number of annual ﬁsh meals and in
portion serving size were apparent in both male and female ﬁsheaters, and as a result, a
considerable decrease in total weight of ﬁsh consumed in a year was demonstrated (Table
30). Fisheaters reported annually consuming 52 pounds for men, 40 pounds for women
in survey 11, but their consumption was reduced to 25 and 16 pounds in survey 111,
respectively.

Data pertaining to trimming away dorsal and belly fat and removing ﬁsh skin
prior to cooking showed that these practices had substantially increased in men and
women. Approximately 72 percent in men and 62 percent in women ﬁsheaters reported
trimming away fat areas in survey II, whereas these proportions went up to over 90
percent in men and 88 percent in women in survey III (Table 31). As many as 80 percent
of ﬁsheaters reported removing ﬁsh skin before cooking in survey III, compared with
only one-third of ﬁsheaters who did so in survey 11. The proportions of ﬁsheaters that
named pan-frying and deep-frying as one of the three commonest cooking methods
remained basically unchanged. About a 30 percent increase was observed in the
proportion of ﬁsheaters who mentioned baking as one of the top three cooking methods in
survey 111, compared with very few ﬁsheaters who did so in the previous two surveys.
Fewer ﬁsheaters referred to broiling as one of the top three cooking methods in survey III,
but a statistically signiﬁcant reduction in this cooking method was observed in men only.
A noteworthy change appeared in the common species of ﬁsh consumed. With respect
to one of three most common species of ﬁsh consumed, less than 47 percent of ﬁsheaters

cited trout and less than 60 percent cited salmon in survey III, as compared with over 90

percent for both species in survey II. The magnitude of differences was larger in terms of
trout than of salmon. The change in referring to Whiteﬁsh as one of 3 top favorite ﬁsh
species was very striking-from none in survey II to all in survey III.

A considerable downward shift was found both in ﬁsheaters and nonﬁsheaters
with respect to DDT, and DDE levels in serum (Table 32). The mean serum Aroclor
1260 levels decreased signiﬁcantly from 26.3 to 21.7 ppb in men (t=6.86, p<.001). A
reduction of marginal statistical signiﬁcance also appeared in females (11.1 ppb in survey
II to 10.0 ppb in survey III; t=1.92, p=.06). After stratifying by ﬁsh eating status,
decreasing changes of Aroclor 1260 levels in all subgroups were statistically signiﬁcant
with exception of female nonﬁsheaters. Over the same time period, there was slight
increase seen in the levels of serum PBB distributions in females (2.5 ppb in survey 11 and

2.6 ppb in survey III, respectively) but not in males.

3.1.8.3 Survey I versus survey III (Tables 33-35)

Twenty eight ﬁsheaters, 15 of whom were male, 13 female, provided information
on ﬁsh consumption in both surveys I and III. Forty three participants had Aroclor 1260
values during these two-time points. Although decreasing tends in annual ﬁsh meals and
serving size in male ﬁsheaters were not statistically signiﬁcant over this time period, the
total weight of ﬁsh consumed in a year decreased considerably (44 pounds and 25.8
pounds between survey I and 111, respectively) (t=2.25, p=.04) (Table 33). A substantial
decrease in number of annual ﬁsh meals (t=3.30, p=0.01) and total weight of ﬁsh

consumed was also seen in female ﬁsheaters (t=3.02, p=.02).

45

The proportions of ﬁsheaters reported trimming away fat areas and removing ﬁsh
skin increased in a great deal from survey I to survey III . An increasing proportion of
ﬁsheaters cited baking and pan-frying as one of the top three cooking methods these
change were not statistically signiﬁcant (Table 34). More ﬁsheaters referred to Whiteﬁsh
as one of the three favorite species of ﬁsh consumed in survey 111. Signiﬁcant reductions
in the proportions of ﬁsheaters choosing trout and salmon occurred, especially in males.

Changes in Aroclor 1260 levels based on 15 male and 28 female paired blood
samples from survey I to III are presented in Table 35. Aroclor 1260 levels in males
tended to be higher in survey III than in survey 1, but no statistical signiﬁcance was found
for either ﬁsheaters or nonﬁsheaters. Levels of Aroclor 1260 in females were

signiﬁcantly higher in survey 111.

3.2 Multivariate analysis

The bivariate analyses reported above document substantial variability in ﬁsh
consumption behavior and serum contaminant levels. This section presents the results of
multivariate statistical analyses of the survey data undertaken in order to address the
following question--when the various factors considered are taken into account
simultaneously (i.e., controlled statistically), which factors emerged as the key
dimensions underlying the observed variations in ﬁsh consumption behaviors and/or
serum contaminant levels? The answer to this question is important to decision-makers,
both for the purpose of targeting research efforts at particular population subgroups, as
well as for use in reﬁning health education strategies so as to more effectively ‘reach’

different population sub-groups.

46

3.2.1 The structure of multivariate analysis

3.2.1.1 Multiple linear regression analysis within each survey

Within each survey, quantity of ﬁsh consumed and log transformed contaminant
levels were selected and analyzed as dependent variables. Participants’ age, education
level, region, and body mass index were used as potential predictors in regression models
of ﬁsh consumption. In models for predicting contaminant levels, current sport ﬁsh
exposure (annual ﬁsh meals) and cumulative sport ﬁsh exposure (ﬁsh eating status) were
treated as main exposure variables and included in the models. The models were
controlled for potential confounding factors by including these factors into the models.

Analyses were performed for males and females separately.

3.2.1.2 Multiple linear regression analysis between surveys

For participants with data from two time points, differences for two primary ﬁsh
consumption variables (annual ﬁsh meals and total weight of ﬁsh consumed) and changes
in contaminant levels (Aroclor 1260, and DDT, DDE, PBB, if applicable) were
calculated. The absolute changes in these two measures of ﬁsh consumption and in
contaminant levels between two-survey time were chosen and analyzed as outcomes or
dependent variables in multiple linear regression. Changes in annual ﬁsh meals and
changes in total weight of ﬁsh consumed were included in multiple linear regression
models separately. Apart from general characteristics considered in bivariate analyses-
age, education, region, baseline body mass index, baseline annual ﬁsh meals were also
included in models, where baseline mean data obtained in the earlier survey of the two

compared surveys. Change in annual ﬁsh meals served as the main predictor variable of

IT

47

interest, and was included in regression models of predicting differences in contaminant
levels, after adjustment for age, education, region, baseline body mass index, ﬁsh eating

status, and baseline contaminant levels.

3.2.1.3 Repeated measures model analysis over the three surveys

Over the three cross-sectional surveys, participants were involved in either one, or
two, or all three surveys. Two to three repeated measures were obtained from 462
participants over the three survey times. Since the analysis is complicated by the
dependence among repeated observations made on the same participant, and data on some
participants may be unbalanced or partially incomplete, such scenarios can be handled by
Procedure Mixed with statement ‘repeated’ available through Statistical Analysis
Software (SAS). The repeated measure models were used to examine time trends in
quantity of ﬁsh consumption and joint changes in levels of serum Aroclor 1260 over

about 20 years period of the three surveys.

3.2.2 The coding of multivariate analysis

Linear regression results are presented in the form of regression coefﬁcients and
respective standard errors. Regression coefﬁcients are interpretable as the association of
the average change in outcome for a unit change in a given predictor variable, adjusted
for potential confounders. The regression coefﬁcients provide a measure of adjusted, or
net, quantitative association. Levels of statistical signiﬁcance are indicated next to each
regression coefﬁcient. The operational deﬁnitions of the outcome variables and the

predictor variables considered in the analysis are indicated in the box below.

48

 

 

 

 

 

Variable Regression coding
Gender Female (reference)
Male
Age Entered as continuous variable (year)
Education 4 dummy variables:
Elementary school (reference)
Some high school
High school graduation
Some college
College graduation and above
Region 2 dummy variables:

thody mass index

North (reference)
Middle
South

Entered as continuous variable

 

Fish eating status

Nonﬁsheaters (reference)
Fisheaters

 

Annual ﬁsh meals

Entered as continuous variable

 

Total weight of ﬁsh consumed

Entered as continuous variable (lbs)

 

Quartile of total weight of ﬁsh consumed

3 dummy variables:
1 (reference)

3
4

 

Change in annual ﬁsh meals

Entered as continuous variable (meal)

 

Change in total weight of ﬁsh consumed

Entered as continuous variable (lbs)

 

Aroclor 1260, DDT, DDE, PBB levels

Entered as continuous variable in
scale either arithmetic (ppb) or log
transformed

 

Change in Aroclor 1260, DDT, DDE, PBB
levels

Entered as continuous variable (ppb)

 

 

Survey time

2 dummy variables.
survey I (1973-74) (reference)
survey II (1979-81)
survey 111 (1989-91)

 

3.2.3 Results of multivariate analysis

3.2.3.1

The two outcomes chosen for further analysis for ﬁsh consumption at each survey

Fish consumption within surveys I, II, and III (T ables 36, 37)

time were 1) annual ﬁsh meals, and 2) total weight of ﬁsh consumed. Except for region,

 

49

general characteristics of male ﬁsheaters, namely age, education, and body mass index
were not signiﬁcantly linear associated with either annual ﬁsh meals or total weight of
ﬁsh consumed within each survey time point. Survey II data shows that males who lived
in the south reported a considerably larger amount of ﬁsh consumed as measured by total
pounds of ﬁsh consumed yearly, compared with those living in north of Michigan
(B=10.60, p=.01) (Table 37). Females living in the middle area of Michigan reported a
considerably higher quantity of ﬁsh consumption than those living in north Michigan as
measured either by annual ﬁsh meals (B=36. 17, p=.002) (Table 36) or total weight of ﬁsh
consumed (B=28.25, p=.001) (Table 37) in survey II. The ﬁrst and second surveys
showed that women having some high school, had graduated from high school, or had
some college, had fewer annual ﬁsh meals. However, this relationship was reversed in
survey III data, making it difﬁcult to be interpreted. It is noteworthy to point out that
factors considered in models to predict the quantity of ﬁsh consumption were not strong
predictors, especially in surveys 11 and III. In combination, these four predictors only

explained 3 to 19 percent of the variances of models, as indicated by R square values.

3.2.3.2 Log levels of Aroclor 1254 in survey I (Table 38)

After controlling for demographic characteristics and body mass index, quartiles
of total weight of ﬁsh consumed (pounds) appeared to be predictors of Aroclor 1254
levels (log transformed) in both sexes. The signiﬁcant linear association between age and
log levels of Aroclor 1254 was demonstrated in female ﬁsheaters (b=0.02, p<.001). This
association was identiﬁed in male ﬁsheaters, but it was not statistically signiﬁcant

(b=0.01, p=.07). The older the participant’s age, the higher the levels of Aroclor 1254.

50

In the models, adjusting for general characteristics, either eating status or quartile of
annual ﬁsh meals is signiﬁcantly associated with log levels of Aroclor 1254 while these
two variables included in the models separately. The association between eating status
and Aroclor 1254 levels became non-statistically signiﬁcant once including quartile of
annual ﬁsh meals into the models. The regression coefﬁcients of quartile of total weight
of ﬁsh consumed did not change much (less than 5 percent) with or without eating status

in the models (Data not presented here).

3.2.3.3 Log levels of Aroclor 1260 within surveys I, II, and III (Table 39)

Within each survey in both sexes, the results showed that the larger the amount of
ﬁsh consumed, the higher the Aroclor 1260 levels. Participants’ ﬁsh eating status were
also signiﬁcantly associated with log levels of Aroclor 1260 in both surveys II and III in
men and survey III in women. An association was seen between increasing age and
elevated log Aroclor 1260 levels in both surveys 11 and III in men and all three surveys in
women. Participants, who were male, older, a ﬁsheater, or reported having large
quantity of ﬁsh consumed annually, were more likely to have a high level of Aroclor
1260. The same patterns were demonstrated in women. In addition, female’s education
level was signiﬁcantly related to her log levels of Aroclor 1260 in surveys II and III.
Women having high school education or above presented considerably lower log levels of
Aroclor 1260, compared with those educated at only elementary level. After adjusting for
general characteristics, regression coefﬁcients for the separate model with eating status or
with quartile of total weight of ﬁsh consumed were statistically signiﬁcant. In models

with both sport ﬁsh exposure variables, eating status contributed to the models is getting

51

more signiﬁcant in survey III, while quartile of total weight of ﬁsh consumed contributed
more in the earlier surveys. Figures 8 and 9 display multiple linear regression of log
Aroclor 1260 levels against quartile of total weight of ﬁsh consumed over the three

surveys for males and females, respectively.

3.2.3.4 Log levels of DDT and DDE within surveys 11 and III (Tables 40, 41)

Log DDT and DDE levels were highly associated (Pearson r=0.98, p<.001). After
adjusting for potential confounding factors, quartile of total weight of ﬁsh consumed was
signiﬁcant associated with log levels of these two contaminants in survey 11 in both sexes
and in survey III in women, whereas survey 111 data in men shows a signiﬁcant
association between quartile of annual ﬁsh meals and log levels of DDT or DDE which
was diminished once ﬁsh eating status was included in the models. A signiﬁcant linear
association between age and log levels of DDT or DDE was consistently shown in men
and women. Education level is inversely related to log DDT or DDE levels in both
surveys for females but not for males. Women who living in middle or south Michigan

had higher DDT and DDE levels than those in north Michigan.

3.2.3.5 Log levels of PBB within surveys II and 111 (Table 42)

Region appears to be a strong predictor of log PBB levels in both sexes at surveys
11 and III. Participants living in middle or south Michigan had higher PBB levels than
those in north Michigan. No statistically signiﬁcant association between log PBB levels
and ﬁsh consumption as measured by ﬁsh eating status or by quartile of total weight of

ﬁsh consumed was observed in either sex. But in survey III, women with the third or

52

forth quartiles of total weight of ﬁsh consumed had higher log PBB levels compared with

those with lowest quartile.

3.2.3.6 Changes in ﬁsh consumption between surveys--- from survey I to II and from

survey 11 to III (Table 43)

Results of regression analyses showed the same pattern based on the two
dependent variables of ﬁsh consumption with respect to quartile of absolute change in
annual ﬁsh meals and absolute change in total weight of ﬁsh consumed. Therefore,
Table 43 only presented results of regression analysis on change in total weight of ﬁsh
consumed based on two time point data from survey I to II or survey II to III. There were
114 participants with data on demographic characteristics and Aroclor 1260 levels during
survey I and 11. Age, education, region, and baseline body mass index were included as
independent variables in regression models to explain the variability of change in total
weight of ﬁsh consumed. All results are stratiﬁed by sex. Baseline total weight of ﬁsh
consumed, in addition to age, gender, region, and baseline body mass index, was a
covariate in the ﬁnal model, with p values less than 0.001 in both sexes. The regression
coefﬁcients (standard error) for baseline total weight of ﬁsh consumed were 0.73 (0.14)
for male ﬁsheaters and 0.59 (0.07) for female ﬁsheaters, respectively. The R square was
0.47 and 0.68 for the model in male ﬁsheaters and female ﬁsheaters, respectively. The
decline in ﬁsh consumption was greater in ﬁsheaters who reported having higher baseline
ﬁsh consumption than that in ﬁsheaters having lower baseline ﬁsh consumption.

There were 391 participants with information on ﬁsh consumption from surveys 11

and III. Two hundred and three of 391 (51.9 percent) participants were regular ﬁsheaters,

53

of whom 139 were male. Separate data for both sexes showing the linear relation of
general characteristics to change in total weight of ﬁsh consumed from survey II to III are
presented in Table 43. Only baseline ﬁsh consumption entered the model at the
signiﬁcance level of p less than 0.05; other factors considered in models were not strong

predictors of changes in total weight of ﬁsh consumed.

3.2.3.7 Changes in levels of Aroclor 1260 between surveys----from survey I to II and

survey II to III (Table 44)

In women, the largest quartile of change in total weight of ﬁsh consumed from
survey II to 111 is strong predictor of changes in Aroclor 1260, as compared to the lowest
quartile of change in total weight of ﬁsh consumed (b=3.61, p=.03); cumulative sport
ﬁsh exposure status and baseline Aroclor 1260 levels also signiﬁcantly contributed to the
model of change in levels of Aroclor 1260 from survey II to II] (b=-3.56, p=.02 for ﬁsh
eating status; b=0.26, p<.001 for baseline Aroclor 1260 levels, respectively), this
associations was not found in data from survey I to 11. Change in Aroclor 1260 levels in
men has different pattern from women. Quartile of change in total weight of ﬁsh
consumed did not appear to be a strong predictor of changes in Aroclor 1260 levels in
men based on data of survey II to III, while male participants with larger baseline body
mass index had smaller changes in Aroclor 1260 levels (b=-61.7l p=.02); no signiﬁcant
association between change in PCB levels and other covariates was found in both survey
intervals. Figures 10 shows multiple linear regression of change in Aroclor 1260 levels
against quartile of change in total weight of ﬁsh consumed in males and females over the

three surveys.

54

3.2.3.8 Changes in levels of DDT from survey 11 to III (T ables 45-a, 45-b)

Based on the two-time points data of survey 11 and III, regression of change in
DDT against quartile of changes in total weight of ﬁsh consumed after controlling for
age, education, region, baseline BMI, and baseline DDT levels are presented, stratiﬁed by
gender, in Tables 45-a and 45-b. Fish consumption was a signiﬁcant predictor of
changes in DDT levels; the change in contaminant levels were strong predicted by
changes in annual ﬁsh meals in men (b=0.09, p=.03), and by cumulative sport ﬁsh
exposure status in women (b=-3.43, p=.03). Baseline DDT levels were positively
associated with changes in DDT levels in both sexes. Men with larger body mass indexes
had slower reductions of serum DDT levels (b=-123.00, p=.003). Older women
experienced considerably slower reductions in DDT levels (b=-0.15, p=.004) (Table 45-
a). However, when quartile of change in total weight of ﬁsh consumed was included in
the model instead of absolute change in annual ﬁsh meals, no association between change
in DDT levels and quartile of change in total weight of ﬁsh consumed was found in both

sexes (Table 45-b).

3.2.3.9 Changes in levels of DDE from survey 11 to survey III (Table 46-a, 46-b)

Change in annual ﬁsh meals signiﬁcantly contributed to change in DDE levels.
After stratiﬁed by gender, this association was only found in men (b=0.08, p=.02) (Table
46-a). when quartile of change in total weight of ﬁsh consumed was included in the
model instead of absolute change in annual ﬁsh meals, the highest quartile of change in
ﬁsh consumption was marginal associated with change in DDE levels, as compared with

the lowest quartile of ﬁsh consumption in men (b=8.24, p=.05) (Table 46-b). No

55

signiﬁcant association between change in DDT levels and ﬁsh consumption was found in
women. Consistently, baseline DDE levels were also signiﬁcantly associated with DDE
change from survey II to III in both sexes. Baseline body mass index in men was

inversely related to change in DDE levels (b=-117.55, p=.003) (Table 46-b).

3.2.3.10 Changes in Levels of PBB from survey II to III (Table 47)

After adjusting for potential confounders, change in PBB levels in both sexes only
depended on the baseline PBB levels and was not associated with either ﬁsh eating status

or change in ﬁsh consumption.

3.2.3.11 Trends of changes in ﬁsh consumption, in log Aroclor 1260 levels, and joint

changes in ﬁsh consumption and log Aroclor 1260 over the three surveys (Tables 48-51)

Repeated measures analyses were performed based on 462 participants with two-
or three-time point data in order to evaluate the changes in ﬁsh consumption, in log
Aroclor 1260 levels, and joint changes in ﬁsh consumption and log Aroclor 1260 levels
over three surveys. Signiﬁcantly decreasing trends were observed in ﬁsh consumption
among ﬁsheaters as measured by the annual ﬁsh meals (b=-21.93 meal/decade, p<.001)
or total weight of ﬁsh consumed (b=-21. 16 pound/decade, p<.001) after controlled for
potential confounding factors-gender, age, education, and region. Table 48 displays total
weight of ﬁsh consumed against 3 times of survey after controlling for gender, age,
education, and region. Figure 11 illustrate time trends in total weight of ﬁsh consumed in
the period of the three surveys. Log levels of Aroclor 1260 elevated from survey I to II
(b=0. 10 ppb/decade, p=0.37) and signiﬁcantly decreased in survey III compared with

survey I (b=-0.42 ppb/decade, p<.001), after controlling for sex, age, education, region,

56

but not for ﬁsh consumption (Table 49). However, when including variable-total weight
of ﬁsh consumed in the model stated above, log levels increased signiﬁcantly from survey
Ito II (b=.31, p=.002) and remained high in survey III (b=.13, p=.19) (Table 50).
Signiﬁcant linear relationship between log Aroclor 1260 levels and group of total weight
of ﬁsh consumed was seen in both sexes after adjusting for age, gender, education, and

region (Table 51 and Figure 12).

4 Discussion

4.1 Strength of the present study

Most previous studies investigating the relation between sport caught Great Lakes
ﬁsh consumption and human serum levels of contaminants were cross-sectional (43, 84).
But Hovinga’s work(85, 86) examined change in PCB and DDT levels from two
Michigan anglers survey conducted in 1982 and 1989. In the present study, the unique
strength is that this secondary data analyses combined three cross-sectional surveys
conducted in 1973-74, 1979-82, and 1989-91. The association of contaminant levels and
ﬁsh consumption behaviors could not only be examined within each survey separately,
but between surveys. Pairwise comparisons could be made among participants with two-
time survey data; furthermore, nearly one-third of participants had two- or three- time
points data which provided an basis for evaluation of patterns of changes in ﬁsh
consumption and in contaminant levels through long term, low dose environmental
exposure to contaminated ﬁsh over about a 20-year time window.
4.2 Time trends in ﬁsh consumption behaviors among ﬁsheaters

It is clear from three- survey data that considerable decline has been observed in
annual ﬁsh meals and total weight of ﬁsh consumed over 20 years among ﬁsheaters. The
combination of eating less sport caught Great Lakes ﬁsh, of modifying ﬁsh-cleaning or
preparation methods, and changing the choice of favorite ﬁsh species sought and eaten,
are the three most common changes. This may come as no great surprise, as awareness of

the potential toxicity of sport caught Great Lakes ﬁsh motivated health departments in

57

58

several states around the Great Lakes to issue ﬁsh consumption advisories. Survey 1]]
data shows that the majority of ﬁsheaters now trim away fatty areas and remove ﬁsh skin
prior to cooking. The shift of favorite ﬁsh species began during the period of survey I
and II, but a larger shift occurred between survey 11 to IH- with every ﬁsheater in the study
citing Whiteﬁsh, a less contaminated ﬁsh species, as one of the most frequent ﬁsh
consumed and the proportion of ﬁsheaters citing trout and salmon dropping signiﬁcantly.
Our results are comparable to the survey conducted by Connelly and Knuth (87). They
studied 8,000 licensed sport anglers from all the US Great Lakes States and found that 36
percent of the respondents had made changes in their ﬁsh consumption behaviors with
respect to consuming less ﬁsh and modifying their ﬁsh preparation methods. In response
to increasing study ﬁndings from ﬁsh, wild life, and animal studies, ﬁsheaters become
more aware of their general health and reduced their sport ﬁsh consumption accordingly.
It is also possible that the number of ﬁshermen who depend on ﬁsh as a food supplement
has decreased because of economic improvement. The purpose of ﬁshing is probably
more for recreation now than for seeking food searching. Changes in ﬁsh-cleaning and
preparation ways and favorite ﬁsh species shift may be attributable to state health ﬁsh
advisories. It may be possible that the shift in the most common ﬁsh species consumed to
Whiteﬁsh may happen to be more Whiteﬁsh are available than salmon and trout because
of ecosystem changes. This needs to be conﬁrmed. No clear pattern of cooking method
or change in cooking methods could be identiﬁed in ﬁsheaters. Fish advisories encourage
ﬁsheaters to broil or bake ﬁsh and to avoid frying. Yet ﬁsheaters have their personal

preferences about produces the best taste, despite acknowledgment of other advice.

59

4.3 Determinants of contaminant levels

4.3.1 Fish consumption

Although each of the general factors considered was observed to be associated
with serum contaminant levels and their changes at one point or another in the analysis,
two factors emerge from the surveys as being especially strong determinants of levels of
Aroclor 1260, DDT, DDE. The ﬁrst of these is ﬁsh consumption. Serum contaminant
levels of ﬁsheaters are strikingly higher than those of participants from the same area who
report eating less sport Great Lakes ﬁsh in a year, even after adjusting for potential
confounding factors. A dose-response relationship was observed-the higher the ﬁsh
consumption, the higher the levels of contaminants. Statistical coherence in the form of a
linear dose-response relationship strongly supports the causal relationship, but any other
statistical relationship needs to be explored further.

In the multiple linear models, adjusting for general characteristics, either eating
status or quartile of annual ﬁsh meals is signiﬁcantly associated with log levels of
contaminants. Once both ﬁsh eating status and annual ﬁsh meals were included in the
models, one or the other is statistically signiﬁcant. In the earlier survey data, total weight
of ﬁsh consumed contributed more to the models than did eating status, while ﬁsh eating
status contributed more to the models in the later survey data. Almost every participants’
ﬁsh eating status was deﬁned at survey I or survey 11, based on quantity of ﬁsh consumed
and remained unchanged over the survey time. Hence, quantity of ﬁsh consumption, such
as total weight of ﬁsh consumed or number of annual ﬁsh meals, may can more
information than does the binary variable-ﬁsh eating status. In survey 111, ﬁsh eating

status became more important than quantity of ﬁsh consumption to the models, which

60

reﬂects cumulative sport ﬁsh exposure being more important than current exposure in the
late survey time. The overall environment has been improved for more than 20 years.
The levels of contaminants in ﬁsh have decreased greatly. Among ﬁsheaters, the drop in
quantity of ﬁsh consumed and the shift in ﬁsh consumption behaviors to recommended
ﬁsh preparation practices have been demonstrated. As a result, effect of current sport ﬁsh
exposure on levels of contaminants has been diminished, but cumulative sport exposure
indicating historical exposure still remains a strong predictor of current contaminant
levels.
4.3.2 Gender

The second factor is gender. The results indicate that contaminants found in
human serum especially DDT, DDE and Aroclor 1260 were consistently and signiﬁcantly
higher in males than females. This could be explained by men having more contaminated
ﬁsh exposure than women. Since the lipophilic nature of these contaminants, levels of
total lipids in human blood need to be taken into account when examining the
relationship between gender and levels of contaminants. Storage and excretion of these
contaminants in human tissue may be inﬂuenced by the lipid levels. The descriptive
statistics of total lipids levels stratiﬁed by gender and ﬁsh eating status based on survey 11
data is presented in Appendix II, and results of Pearson correlation analyses are displayed
in Appendix III. Lipid levels are signiﬁcantly higher than in males than females. Lipid
levels are positively associated with levels of contaminants (Pearson r=0.36 for Aroclor
1260; r=0.34 for DDT; r=0.34 for DDE; all p values <.001; Pearson r=0.09 for PBB,
p=.002 ). Lipid levels alone contributed to 13% of the variance of log Aroclor 1260

levels, 12% of the variance of log DDT levels, and 11% of the variance of log DDE

61

levels; all p values are less than .001. Therefore, regression analyses have been rerun
after controlling for lipid levels by including lipid levels in the models. The results show
an association between the quantity of ﬁsh consumption and levels of Aroclor 1260, or
DDT, or DDE, after adjusted potential confounding factors included in previous models
as well as lipid levels. The lipid levels did not explain gender differences in log PCB
body burden based on our data.

Hovinga(85) found that between 1982 and 1989, mean serum DDT levels
decreased substantially in both 115 ﬁsheaters and 95 comparisons, while no correlation
was found between changes in DDT body burden and ﬁsh consumption. Hovinga’s work
failed to detect the effect of ﬁsh consumption on changes in DDT, which is probably
attributable to the relativély small sample size. In the present study, reduction in levels of
DDT, and DDE contaminants from survey II to survey 111 has been demonstrated and
conﬁrms that ﬁnding in Hovinga’s work. Data from the present study further
demonstrate that changes in DDT and DDE levels are determined by not only the baseline
DDT or DDE levels, but by the number of annual ﬁsh meals as well. While stratiﬁed by
sex, the effect of eating less ﬁsh on reduction in DDT or DDE is pronounced for men
only. Since the use of DDT was banned in the United States in 1972, DDT and DDE
levels in western populations have declined since the 197OS(88, 89). The measurable
reduction in DDT or DDE body burdens found in this study may be due in part to the
overall reduction in environmental DDT or DDE contamination in the environment and
in ﬁsh tissue. But also it appears that high-risk populations are heeding ﬁsh consumption
advisories and therefore reducing their quantity of ﬁsh consumption, which also leads to

reduce in DDT/DDE body burden.

62

4.4 Joint change in ﬁsh consumption and contaminant levels

Although studies of Great Lakes ﬁsh and wildlife conﬁrm that the levels of PCB
have declined since the mid-1970’s(90-93), and the results from this study documented
downward trends in ﬁsh consumption, over the time period of the surveys, the levels of
Aroclor 1260 have not yet decreased. The lack of concurrent decline in PCB
concentrations in human serum probably reﬂects the persistent nature of PCBs. An
apparently declining trend has been demonstrated in DDT concentrations in at-risk sub-
population (ﬁsheaters) from 19805 to 19903, but considering PCB was banned later than
DDT. A decrease in body burdens of PCBs in human population may be expected to
follow declining DDT levels. And because of the long half-live of PCBs, this decline
may be expected to lag considerably behind the PCB decline in the environment. There
may have other reasons for the static PCB levels in the target population studied.
Although PCB production was ceased in the United States in 1977, very large quantities
of PCB remain in industrial use. Thus, waste site leakage or other sources of PCB
contamination such as atmospheric deposition may be major sources of exposure rather
than ﬁsh consumption. Restrictions on PCB production alone may reﬂect not ensure
apparently decreasing levels of PCBs exposure in human populations.

PBB is a unique contaminant for Michigan residents. It is not a ﬁsh-bom
contaminant and its levels were not associated with the quantity of ﬁsh consumption.
This has been conﬁrmed in this study, which indirectly illustrates that our study data are

reliable.

5 Limitations
5.1 It was not a population-based sample.

Because the recruitment in the three surveys was based on volunteers, selection
bias may exist. Participants who remained in the study from the earlier studies may be
more aware of the safety of ﬁsh consumption. This may affect their ﬁsh consumption
behaviors.

5.2 Information on ﬁsh consumption in the majority of nonﬁsheaters was not
available.

Fish consumption information on the majority of nonﬁsheater was not available.
Probably some of nonﬁsheaters ate a small amount of ﬁsh, but it is impossible to quantify
their ﬁsh consumption in this study. The values of these participants’ quantity of ﬁsh
consumption were set to zero in the analysis. This may slightly dilute the association
between ﬁsh consumption and levels of contaminants, but would not alter our overall
conclusions.

5 .3 Information on PCB levels in the Great Lakes ﬁsh was not integrated into the
analysis.

We are not able to integrate PCB levels of ﬁsh tissue in the analysis due to lack of
detailed information on PCB levels in the Great Lakes ﬁsh. Therefore, it is impossible to I
ascertain exposure more precisely. Since PCB levels in ﬁsh are associated with ﬁsh’s

characteristics, for example, ﬁsh species, size, age, etc.

63

5.4 Few participants had three data points.
Participants remained in the surveys were more likely to be ﬁsheaters. However,
in terms of gender, age, and body mass index, there is no difference between participants

with one survey data and with 2- or 3- survey data (data not present here).

6 Future study needs

Although earlier studies showed higher body burdens of PCBs in populations who
consumed a lot of ﬁsh from contaminated waters, the health effects in humans are
controversial. PCB body burdens have been measured in the majority of the
epidemiological studies. However, PCB may simply reﬂect exposure to other ﬁsh-bome
contaminants.

No information is available in the United States on the levels of polychlorinated
dibenzodioxins PCDDs and PCDFs in anglers who consume a great deal of ﬁsh presumed
to be contaminated by these chemicals. In light of the above statement, future
epidemiological studies should focus on highly exposed, susceptible populations such as
occupational exposed workers, reproductive aged women, children, and the aged. Results
from these studies would provide important information on the risk of perinatal or adult
exposure to PCBs, although generalization of results obtained in these populations to the
general population may be fraught with difﬁculties.

A downward trend of human serum PCB, DDT and DDE has been shown since
19803. And a continuous decline of PCB, DDT and DDE human body burden could be
expected due to documented decreases in the overall environment. Fish advisories need

to be modiﬁed based on updated information on contaminant levels.

65

TABLES

66

Table 1. Background characteristics of all study population, 1979-82

 

 

 

Fisheaters Non-Fisheaters x2] t P-value
Number of participants 610 645
(%) Gender: 64.62 <.001
Male 66.3 42.9
Female 33.7 57.1
(Mean) Age in years 46.0 43.1 3.54 <.001
(%) Age: 28.68 <.001
15 - 24 3.6 10.2
25 - 34 20.0 24.7
35 - 44 26.3 21.4
45 - 54 21.2 18.3
55 - 64 17.9 15.0
65+ 11.0 10.4
(%) Education: 12.08 .02
Elementary school 4.6 5.2
Some high school 12.4 9.2
High school graduate 42.0 35.8
Some college 23.3 26.6
College graduate 17.7 23.2
(%) Area of residence 1.71 .43
North 28.9 27.7
Middle 31.6 35.2
South 39.5 37.1
(Mean) Height in inches 68.5 67.1 6.17 <.001
(Mean) Weight in pounds 175.6 158.2 9.08 <.001
(%) Exposed to any kind of 7.55 .01
chemicals: Yes 32.5 25.4
No 67.5 74.6

 

 

 

 

 

 

67

Table 2. Bagground characteristics of male participants, 1979-82

 

 

 

Fisheaters Non-Fisheaters le t P-value
Number of participants 403 241
(%) Age: 12.74 .03
15 - 24 4.0 7.5
25 - 34 18.4 24.9
35 - 44 28.5 21.2
45 - 54 19.9 18.7
55 - 64 18.1 13.7
65+ 11.2 14.1
(%) Education: 8.83 .07
Elementary school 4.7 7.2
Some high school 12.7 9.7
High school graduate 39.2 32.5
Some college 24.7 24.1
College graduate 18.7 26.6
(%) Area of residence 1.46 .48
North 29.3 25.3
Middle 31 .3 34.9
South 39.5 39.8
(Mean) Height in inches 70.6 70.4 0.74 .45
(Mean) Weight in pounds 191.5 178.3 5.88 <.001
(%) Exposed to any kind of 0.00 0.99
chemicals: Yes 41 .4 41 .5
No 58.6 58.5

 

 

 

 

 

 

68

Table 3. Baclgpound characteristics of female participants, 1979-82

 

 

 

Fisheaters Non-Fisheaters 78/ t P-value
Number of participants 205 321
(%) Age: 11.69 .04
15 - 24 2.5 9.0
25 - 34 23.5 24.6
35 - 44 21.6 22.1
45 - 54 24.0 19.9
55 - 64 17.6 17.4
65+ 10.8 6.9
(%) Education: 9.93 .04
Elementary school 4.5 2.6
Some high school 11.9 8.7
High school graduate 47.5 38.8
Some college 20.8 27.5
College graduate 15.3 22.3
(%) Area of residence 1.08 .58
North 28.3 29.3
Middle 32.2 35.5
South 39.5 35.2
(Mean) Height in inches 64.5 64.6 0.42 .67
(Mean) Weight in pounds 144.4 142.8 0.72 .47
(%) Exposed to any kind of 0.01 .92
chemicals: Yes 14.6 14.3
No 85.4 85.7

 

 

 

 

 

 

69

Table 4. The characteristics of tobacco use and alcohol consumption of male participants, 1989-91

 

 

 

Fisheaters Non-Fisheaters le t P-value
Number of participants 238 152
(%) Smoking 9.89 .01
Never 22.3 36.2
Ex-smoker 48.9 44.3
Current smoker 28.8 19.5
Among ex_smoker and cur_smoker
(Mean) Age started smoking 18.0 18.0 0.11 .91
Among ex_smoker
(Mean) Age last smoked regularly 38.9 38.3 0.27 .79
Among cur_smoker
(Mean) number of cigarettes a day 25.5 25.2 0.11 .91
(%) Drinking 10.64 .004
Abstainer 7.6 15.9
Ex_user 15.1 7.3
Current user 77.3 76.8
Among current drinkers:
(Mean) bottles/cans beer at a sitting 1.1 0.8 2.10 .04
(%) Beer drinking 7.86 .05
None 7.1 13.8
< 1 /wk 32.1 40.5
1-3 lwk 30.4 24.1
4+ lwk 30.4 21.6
(Mean) Glasses wine at a sitting 0.3 0.5 3.10 .002
(%) Wine drinking 8.02 .02
None 64.7 50.9
< 1 [wk 30.4 37.1
1+ lwk 4.9 12.1
(Mean) Drinks of liquor at a sitting 0.7 0.6 1.67 .10
(%) Liquor drinking 1.56 .67
None 45.1 49.1
< 1 /wk 33.7 30.2
1-3 lwk 8.7 11.2
4+ lwk 12.5 9.5

 

 

 

 

 

 

70

Table 5. The characteristics of tobacco use and alcohol consumption of female participants, 1989-91

 

 

 

Fisheaters Non-Fisheaters 78/ t P-value
Number of participants 120 214
(%) Smoking 0.23 .89
Never 52.6 54.4
Ex-smoker 27.2 24.8
Current smoker 20.2 20.9
Among ex_smoker and cur_smoker
(Mean) Age started smoking 18.7 20.5 1.81 .07
Among ex_smoker
(Mean) Age last smoked regularly 39.3 41.5 0.68 .50
Among cur_smoker
(Mean) number of cigarettes a day 20.7 19.3 0.58 .56
(%) Drinking 0.12 .94
Abstainer 21.0 19.8
Ex_user 10.1 9.4
Current user 68.9 70.8
Among current drinkers:
(Mean) bottles/cans beer at a sitting 0.4 0.4 0.14 .89
(%) Beer drinking 8.18 .04
None 28.0 39.3
< 1 /wk 35.4 38.7
1-3 lwk 18.3 14.7
4+ lwk 18.3 7.3
(Mean) Glasses wine at a sitting 0.5 0.5 0.01 .99
(%) Wine drinking 0.44 .80
None 47.6 45.3
< 1 /wk 41.5 40.7
1+ lwk 11.0 14.0
(Mean) Drinks of liquor at a sitting 0.7 0.6 1.50 .14
(%) Liquor drinking 3.61 .31
None 48.8 50.7
< 1 /wk 28.0 34.7
1-3 lwk 9.8 8.0
4+ lwk 13.4 6.7

 

 

 

 

 

 

Table 6. The quantity of ﬁsh consumption among ﬁsheaters stratiﬁed by gender, 1973-74

71

 

 

 

 

 

 

 

 

 

 

 

Male Female Total
Number of ﬁsheaters 44 29 74*
Annual ﬁsh meals
MeaniSD 66.2:t43.3 90.9:t48.1 76.41463
Median 56.0 72.0 66.0
Range(Q25-Q75) 40.0-69.0 61 .0-104.0 49.0-90.5
Portion serving size (02)
MeaniSD 13.0:44 9.4i3 .4 1 1.5:1:4.3
Median 12.5 9.0 10.0
Range(Q25-Q75) 9.3-16.0 6.5-1 1.0 8.0-16.0
Total weight of ﬁsh consumed (1b)
MeaniSD 50.9:I:31.6 52.2i35.9 51.7i33.0
Median 39.8 40.0 40.0
Range(Q25-Q75) 28.0-58.5 29.3-54.9 28.4-56.3
*: One participant's gender was unknown.
Table 7. Fish consumption behaviors among ﬁsheaters stratiﬁed by gender, 1973-74
Male Female Total
Number of ﬁsheaters 44 29 74“
One of top 3 ﬁsh species consumed (%)
Trout 100.0 93.1 97.3
Salmon 65.9 62. 1 64.9
Perch 31.8 41.4 36.5
Catﬁsh 4.5 3.4 4.1
Whiteﬁsh 25.0 17.2 21.6
Number of ﬁsheaters“ 38 21 61 ***
One of top 3 cooking methods (%)
Broiled 31.6 28.6 29.5
Baked 18.4 52.4 29.5
Pan fried 55.3 47.6 50.8
Deep fried 7.9 0.0 4.9
Fish preparation prior to cooking (%)
Always trim away dorsal fat area 0.0 9.5 3.3
Always trim away belly fat area 5.3 9.5 6.6
Always remove skin 0.0 0.0 0.0

 

 

 

 

 

 

Note: Total proportion in every sub-category would not sum to 100, because every participants could report
up to three most favorite ﬁsh species or cooking methods.

"' Information was based on ﬁsh diaries.

** One participant’s gender was unknown.

*"Two participants’ genders were unknown.

Table 8. The quantity of ﬁsh consumption among ﬁsheaters stratiﬁed bygender, 1979-82

72

 

 

 

 

 

 

 

 

 

 

 

Male Female Total
Number of ﬁsheaters 403 205 610*
Annual ﬁsh meals
MeaniSD 61.01288 71.0i49.5 64.4:t37.3
Median 53.0 58.0 54.0
Range(Q25-Q75) 44.0-71 .5 51.0-82.3 47.0-76.0
Portion serving size (02)
MeaniSD 13.9i6.5 10.1i4.6 12.6:i:6.2
Median 12.0 8.0 12.0
Range(Q25-Q75) 9.0- 16.0 8.0-12.0 8.0-16.0
Total weight of ﬁsh consumed (lb)
MeaniSD 51.7:I:38.0 43.41386 48.93383
Median 40.0 33.0 38.0
Range(Q25-Q75) 28.0-60.0 26.0-46.8 27.0—57.0
"' Two participants’ genders were unknown.
Table 9. Fish consumption behaviors among ﬁsheaters stratiﬁed bygender, 1979-82
Male Female Total
Number of ﬁsheaters 403 205 610”
One of top 3 ﬁsh species consumed (%)
Trout 94.3 91.2 93.3
Salmon 89.6 93.2 90.8
Perch 43.9 39.0 42.3
Catﬁsh l .5 2.0 1.6
Whiteﬁsh 0.0 0.0 0.0
Number of ﬁsheaters“ 280 121 402”
One of top 3 cooking methods (%)
Broiled 51.8 55.4 52.7
Baked 0.0 3.3 1.0
Pan fried 65.0 54.5 61.9
Deep fried 33.6 32.2 33.3
Fish preparation prior to cooking (%)
Always trim away dorsal fat area 84.6 89.3 86.1
Always trim away belly fat area 84.6 89.3 86.1
Always remove skin 48.6 52.1 49.5

 

 

 

 

 

 

Note: Total proportion in every sub-category would not sum to 100, because every participants could report
up to three most favorite ﬁsh species or cooking methods.

* Information was based on ﬁsh chart.

"Two participants’ genders were unknown.

Table 10. The quantity of ﬁsh consumption amorg ﬁsheaters stratiﬁed by gender, 1989-91

73

 

 

 

 

 

 

 

 

 

 

Male Female Total
Number of ﬁsheaters 238 120 359*
Annual ﬁsh meals
MeaniSD 41 .6:I:32.6 35.21311 39.62324
Median 34.0 27.0 31.0
Range(Q25-Q75) 18.0-52.0 12.0-52.0 15.0-52.0
Portion serving size (02)
MeaniSD 9413.7 7 2:35 87:38
Median 8.0 6.0 8.0
Range(Q25-Q75) 8.0-12.0 4.0-8.0 2.0, 16.0
Total weight of ﬁsh consumed (1b)
MeaniSD 24.71229 15.91167 21.93214
Median 18.0 10.8 15.5
Range(Q25-Q75) 9.4-31 .4 4.5-19.5 6.4-30.0
* One participants’ gender was unknown.
Table 11. Fish consumption behaviors amongﬁsheaters stratiﬁed by mder, 1989-91
Male Female Total
Number of ﬁsheaters 238 120 359*
One of top 3 ﬁsh species consumed (%)
Trout 47. 1 45.8 46.5
Salmon 58.4 60.8 59.3
Perch 52.1 45 .0 49.9
Catfish 0.4 0.0 0.3
Whiteﬁsh 100.0 100.0 100.0
One of top 3 cooking methods (%)
Broiled 29.0 33.3 30.4
Baked 27.7 39.2 31.8
Pan fried 46.6 40.0 44.3
Deep fried 29.8 26.7 29.0
Fish preparation prior to cooking (%)
Always trim away dorsal fat area 92.0 88.3 90.8
Always trim away belly fat area 92.4 88.3 91.1
Always remove skin 83.2 78.3 81.6

 

 

 

 

 

 

Note: Total proportion in every sub-category would not sum to 100, because every participants could report
up to three most favorite ﬁsh species or cooking methods.
* One participants’ gender was unknown.

Table 12. Levels (ppb) of serum Aroclor 1254 and Aroclorl260 stratiﬁed by gender and eating status,

 

 

1973-74
N MeaniSD Median Q25-Q75
Aroclor 1254
Male 78 52.1i47.7 35.0 25.0-66.5
Fisheater 47 65.5:t55.4 40.0 27.0-91.0
Nonﬁsheater 3 1 31 .6i20.3 29.0 15.0-42.0
Female 84 27.9i17 .1 23.0 15.3-35.0
Fisheater 29 39.21205 35.0 23.0-50.5
Nonﬁsheater 55 21 .9:i:1 1.4 20.0 15.0-27.0
Aroclor 1260
Male 78 21 .5:I:21 .4 12.5 90.-23.25
Fisheater 47 26.0:I:25.3 17.0 10.0-32.0
Nonﬁsheater 31 12. 11:86 10.0 6.0-15.0
Female 84 8.2i6.2 6.5 4.0-10.0
Fisheater 29 12.4i7.5 1 1.0 7.5-15.5
Nonﬁsheater 55 5.9-3.9 5.0 4.0-8.0

 

75

Table 13. Levels (ppb) of serum contaminants (Aroclor 1260, PCB total, DDT, DDE, and PBB) stratiﬁed
by gender and eating status, 1979-82

 

 

N MeaniSD Median Q25-Q75
Aroclor 1260
Male 615 23.5i23.5 16.3 8.5-29.5
Fisheater 401 30.7:258 22.9 14.5-38.6
Nonﬁsheater 214 10.1:t7.9 8.1 5.6-11.5
Female 502 12.91146 8.3 4.9-15.5
Fisheater 205 19.61: 1 9.0 14.5 8.2-22.8
Nonﬁsheater 297 8.3-7.7 6.0 3.9-10. l
PCB total
Male 615 25.2:t25.4 17.6 9.0-32.3
Fisheater 401 3293278 24.7 15 .5-4 1 .5
Nonﬁsheater 214 10.7i8.9 8.2 5.6—12.1
Female 502 14.1:t16.2 9.0 5.0-17.0
Fisheater 205 21 .4:t21 .2 15. 1 8.9-25 .4
Nonﬁsheater 297 9.0i8.5 6.3 4.0-10.7
DDT
Male 615 42.5:t63.4 22.5 1 1.6-43.8
Fisheater 401 56.4i74.1 30.8 17.0-53.8
Nonﬁsheater 214 16.31152 12.4 6.3-21.2
Female 502 22.7:t33.2 14.2 8.0-25.6
Fisheater 205 32.5:iz45.5 19.9 1 1.0-34.3
Nonﬁsheater 297 15.912182 1 1.1 6.6-19.5
DDE
Male 615 36621554 18.9 10.1-37.2
Fisheater 401 48.6:t64.9 26.7 14.9-44.7
Nonﬁsheater 214 14.211 3.1 11.0 5.5-18.4
Female 502 l9.1:1:28.1 12.2 6.9-21.4
Fisheater 205 27.6:1:39.8 16.8 9.7-28.8
Nonﬁsheater 297 13.21: 1 2.6 9.5 5.9-16.8
PBB
Male 615 3.4:t4.4 1.9 0.5-4.2
Fisheater 401 36:49 1 .9 0.5-4.2
N onﬁsheater 214 3233.4 2.0 0.5-4.3
Female 502 2. 13:25 1.2 0.5-2.5
Fisheater 205 1.9:l:2. 1 l .1 0.5-2.5

Nonﬁsheatcr 297 22:27 1 .3 0.5-2.6

 

76

Table 14. Levels (ppb) of serum contaminants (Aroclor 1260, PCB total, DDT, DDE, and PBB) stratiﬁed
by gender and eating status, 1989-91

 

N MeaniSD Median Q25-Q75
Aroclor 1260
Male 226 20.9i27.0 13.1 6.3-25.3
Fisheater 140 29.01314 21.0 12.0-33.6
Nonﬁsheater 86 7.8i6.2 6.0 3.7-9.6
Female 189 9.9i9.0 6.8 4.3-13.1
Fisheater 65 14.8:t10.4 12.8 7.3-18.2
Nonﬁsheater 124 7.3i6.6 5 .6 4.0-9.1
PCB total
Male 226 20.9:1:27.0 l3. 1 625-253
Fisheater 140 29.013] .4 21 .0 12.0-33.6
Nonﬁsheater 86 7.8i6.2 6.0 3.7-9.6
Female 189 1 00:19.02 6.9 4.4-13.4
Fisheater 65 14.8i10.4 12.8 7.3-18.2
Nonﬁsheatcr 124 7 .5i7.0 5.7 4.0-9.1
DDT
Male 226 19.1i25.5 12.6 5.8-20.9
Fisheater 140 22.6i26.7 14.8 9.0-26.4
Nonﬁsheater 86 8.31:7. 1 6.2 3.7-12.4
Female 189 1 1.7:I:12.5 7.8 4.5-14.7
Fisheater 65 14.1:t15.3 10.2 5.3-16.2
Nonﬁsheater 124 8.6i7.5 - 5.9 3.4-10.8
DDE
Male 226 17.1:I:22.6 11.4 5.6-18.8
Fisheater 140 22.6:l:26.7 14.8 9.0-26.4
Nonﬁsheater 86 8.31:7. 1 6.2 3.7-12.4
Female 189 10.5:tl 1.1 7.0 4.0-13.2
Fisheater 65 14.1:t15.3 10.2 5.3-16.2
Nonﬁsheater 124 86:75 5.9 3.4-10.8
PBB
Male 226 4.5i4. 4 2.8 1.6-6.0
Fisheater 140 4.71:4.8 2.7 1.6-6.0
Nonﬁsheater 86 4.2:l:3.8 2.8 3.7-9.6
Female 189 2.7:t2.8 1.8 0.5-3.5
Fisheater 65 2.51224 1.8 0.5-3.6

Nonﬁsheater 124 28:29 1 .8 0.5-3.5

 

77

Table 15. Effect of the ag on gurantity of ﬁsh consumption stratiﬁed by gender, 1979-82

 

 

 

Agelyear)
<35 35-44 45-54 55+ leF valuc‘MI p value
Male
No. of Cases 90 115 80 1 18
Mean. No. of ﬁsh meals" 54.81 60.73 62.05 63.25 4.03 .05
Pct. No. of ﬁsh meals 16.65 .05
None 47.6 30.7 38.4 36.8
Low 22.6 22.9 16.8 18.9
Middle 16.1 25.9 22.4 25.9
High 13.7 20.5 22.4 18.4
Mean. pounds of ﬁsh consumed“ 53.09 52.25 49.68 49.18 0.72 .40
Pet. pounds of ﬁsh consumed 16.48 .06
None 47.6 30.7 38.4 36.8
Low 17.9 24.7 20.8 27.6
Middle 17.9 18.7 19.2 20.0
High 16.7 25.9 21.6 15.7
Female
No. of Cases 53 44 49 58
Mean. No. of ﬁsh meals“ 68.13 58.13 73.45 77.51 2.28 .13
Pet. No. of ﬁsh meals 9.40 .40
None 67.1 61.7 60.2 58.1
Low 10.6 20.0 15.9 15.4
Middle 9.3 11.3 9.7 11.8
High 13.0 7.0 14.2 14.7
Mean. pounds of ﬁsh consumed“ 46.49 35.38 40.53 43.98 0.01 .94
Pet. pounds of ﬁsh consumed 10.23 .33
None 67.1 61.7 60.2 58.1
Low 11.2 14.8 8.0 16.9
Middle 10.6 12.2 18.6 11.0
High 11.2 11.3 13.3 14.0

 

*: Analyses were restricted to ﬁsheaters.
I""': F values were testing for linearity

Table 16. Effect of education levels on quantity of ﬁsh consumption stratiﬁed by gender, 1979-82

 

 

 

 

Education
< High school >= High school x2 /t value p value
Male
No. of Cases 70 331
Mean. No. of ﬁsh meals"I 68.71 58.73 2.64 .01
Pct. No. of ﬁsh meals 5.73 .13
None 37.3 37.9
Low 14.5 22.0
Middle 22.7 22.9
High 25.5 17.2
Mean. pounds of ﬁsh consumed“ 56.88 49.82 1.41 .16
Pet. pounds of ﬁsh consumed 4.30 .23
None 37.3 37.9
Low 18.2 24.2
Middle 18.2 19.3
High 26.4 18.6
Female
No. of Cases 33 169
Mean. No. of ﬁsh meals“ 94.15 65.05 1.75 .09
Pet. No. of ﬁsh meals 16.88 <.001
None 51.5 62.3
Low 10.3 16.5
Middle 10.3 10.8
High 27.9 10.4
Mean. pounds of ﬁsh consumed"I 60.61 38.32 2.45 .02
Pet. pounds of ﬁsh consumed 20.65 <.001
None 51.5 62.3
Low 4.4 14.7
Middle 16.2 12.6
High 27.9 10.4

 

I": Analyses were restricted to ﬁsheaters.

79

Table 17. Effect of region on quantity of ﬁsh consumption stratiﬁed by gender, 1979-82

 

 

 

 

Region
North Middle South {I F value" p value
Male
No. of Cases 1 18 126 159
Mean. No. of ﬁsh meals" 63.11 62.38 56.96 3.44 .06
Pet. No. of ﬁsh meals 2.78 .84
None 34.6 41.9 38.0
Low 21.2 20.0 20.4
Middle 22.9 21.4 23.5
High 21.2 16.7 18.0
Mean. pounds of ﬁsh consumed“ 49.15 58.55 46.79 0.54 .46
Pct. pounds of ﬁsh consumed 17.47 .01
None 34.6 41.9 38.0
Low 29.6 14.8 25.1
Middle 15.6 19.0 21.2
High 20.1 24.3 15.7
Female
No. of Cases 58 66 81
Mean. No. of ﬁsh meals“ 80.29 67.92 62.00 5.32 .02
Pet. No. of ﬁsh meals 11.16 .08
None 63.2 64.4 58.8
Low 9.2 15.0 20.1
Middle 10.5 10.0 10.8
High 17.1 10.6 10.3
Mean. pounds of ﬁsh consumed" 48.00 39.77 38.12 2.52 .11
Pct. pounds of ﬁsh consumed 3.99 .68
None 63.2 64.4 58.8
Low 13.8 11.7 13.4
Middle 9.2 12.8 15.5
High 13.8 11.1 12.4

 

*: Analyses were restricted to ﬁsheaters.

**: F values were testing for linearity

80

Table 18-a. Effect of body mass index on quantity of ﬁsh consumption in males, 1979-82
Body Mass Index

 

 

 

.10—.23 24-25 .26—.28 29.45 12/ Fvalue“ p value
No. of Cases 102 101 99 101
Mean. No. of ﬁsh meals" 56.74 59.08 61.88 62.46 2.17 .14
Pet. No. of ﬁsh meals 44.14 <.001
None 55.6 41.2 31.6 24.4
Low 16.9 17.6 19.6 28.1
Middle 16.3 24.8 27.8 21.9
High 1 1.3 16.4 20.9 25.6
Mean. pounds of ﬁsh consumed* 44.91 44.40 57.95 53.93 5.50 .02
Pet. pounds of ﬁsh consumed 42.90 <.001
None 55.6 41.2 31.6 24.4
Low 16.9 23.0 22.8 29.4
Middle 14.4 20.0 23.4 18.1
High 13.1 15.8 22.2 28.1

 

Table 18-b. Effect of body mass index on quantity of ﬁsh consumption in females, 1979-82
Body Mass Index
.18-.20 .2l-.22 .23-.26 27.39 {I F value“ p value

 

No. ofCases 51 51 51

51
Mean. No. of ﬁsh meals“ 60.49 72.75 72.61 70.08 0.81 .36
Pct. No. of ﬁsh meals 13.07 .16
None 68.5 62.1 52.3 64.8
Low 14.6 18.2 17.4 10.9
Middle 9.2 8.3 12.1 11.7
High 7.7 11.4 18.2 12.5
Mean. pounds of ﬁsh consumed" 42.89 38.45 41.61 44.57 0.13 .72
Pet. pounds of ﬁsh consumed 15.19 .09
None 68.5 62.1 52.3 64.8
Low 10.8 15.9 13.6 10.9
Middle 13.8 12.1 15.9 9.4
High 6.9 9.8 18.2 14.8

 

*: Analyses were restricted to ﬁsheaters.

“z F values were testing for linearity

81

Table 19. Effect of the chemicals exposure status on quantity of ﬁsh consumption stratiﬁed by gender,
1979-82
Chemical exposure

 

 

 

Yes No le t value p value
Male
No. of Cases 167 236
Mean. No. of ﬁsh meals“ 58.57 61.70 1.12 .26
Pet. No. of ﬁsh meals 1.25 .74
None 39.0 37.9
Low 21.0 20.2
Middle 23.6 22.0
High 16.5 19.9
Mean. pounds of ﬁsh consumed“ 52.02 50.33 0.44 .66
Pet. pounds of ﬁsh consumed 3.50 .32
None 39.0 37.9
Low 20.2 24.9
Middle 21.7 17.0
High 19.1 20.2
Female
No. of Cases 30 175
Mean. No. of ﬁsh meals“ 72.81 68.86 0.42 .67
Pet. No. of ﬁsh meals 2.03 .57
None 60.5 62.2
Low 13.2 15.6
Middle 9.2 10.7
High 17.1 11.6
Mean. pounds of ﬁsh consumed“ 41.53 41.7 0.02 .98
Pet. pounds of ﬁsh consumed 1.78 .62
None 60.5 62.2
Low 15.8 12.4
Middle 9.2 13.3
High 14.5 12.0

 

*: Analyses were restricted to ﬁsheaters.

82

Table 20. Effect of the aggon contaminants levels stratiﬁed by eating status and gender, 1979-82

 

 

 

Ass (Lear)
<35 35-44 45-54 55+ F value" p value

Male
Mean. Aroclor 1260 (ppb)

Non-ﬁsheater 6.45 8.75 1 1.68 12.60 34.16 <.001

Fisheater 18.35 26.79 35.56 40.46 78.84 <.001
Mean DDT (ppb)

Non-ﬁsheater 7.97 13.17 22.17 22.50 39.91 <.001

Fisheater 24.75 41.79 69.66 85.64 86.21 <.001
Mean DDE (ppb)

Non-ﬁsheater 6.93 1 1.50 19.27 19.46 43.00 <.001

Fisheater 21.13 35.80 60.68 73.77 84.18 <.001
Mean PBB (ppb)

Non-ﬁsheater 3.10 3. 10 4.26 2.65 <0.01 .98

Fisheater 3.91 3.51 3.84 3.28 (0.01 1.00
Female
Mean. Aroclor 1260 (ppb)

Non-ﬁsheater 4.50 8.58 9.97 9.91 50.44 <.001

Fisheater 10.87 12.78 22.16 30.68 69.72 <.001
Mean DDT (PPb)

Non-ﬁsheater 8.07 18.47 15.33 21.63 44.97 <.001

Fisheater 16.24 18.86 35 .71 55.13 65.73 <.001
Mean DDE (ppb)

Non-ﬁsheater 6.95 13.26 13.98 18.43 52.1 1 <.001

Fisheater 13.60 15.87 30.70 46.97 62.51 <.001
Mean PBB (PPb)

Non-ﬁsheater 1 .96 l .89 1 .84 2.95 3.41 .07

Fisheater 2.06 1.72 1.97 1.93 0.05 .82

 

*: F values were testing for linearity and were calculated after contaminant levels had been log transformed.

83

Table 21. Effect of education on contaminants levels stratiﬁed by eating status and gender, 1979-82

 

 

 

 

Education
Elemen. Some High Some college F value“ p value
school high school college grad.
school grad.

Male
Mean. Aroclor 1260 (PPb)

Non-ﬁsheater 12.47 9.28 1 1.59 7.79 8.28 4.66 .03

Fisheater 33.76 32.00 33.51 29.13 24.28 5.00 .03
Mean DDT (ppb)

Non-ﬁsheater 24.25 14.13 16.66 13.17 15.84 1.24 .27

Fisheater 49.34 57.51 68.07 48.16 41.95 3.29 .07
Mean DDE (ppb)

Non-ﬁsheater 20.90 13.1 1 14.45 1 1.45 13.52 2.92 .09

Fisheater 42.08 50.09 58.76 41.32 36.02 3.49 .06
Mean PBB (ppb)

Non-ﬁsheater 3 .22 5. 17 3.42 2.46 2.84 1.65 .20

Fisheater 2.58 4.43 3.44 3.42 3.83 0.09 .77
Female
Mean. Aroclor 1260 (ppb)

Non-ﬁsheater 10.14 8.19 8.19 7.04 7.54 2.60 .11

Fisheater 49.10 27.05 17.64 15.69 15.93 13.00 <.001
Mean DDT (ppb)

Non-ﬁsheater 26.83 17.41 15.44 15.80 1 1.22 13.61 <.001

Fisheater 95.77 51.16 28.66 24.66 20.66 18.91 <.001
Mean DDE (ppb)

Non-ﬁsheater 23.13 14.94 13.59 1 1.54 9.60 14.82 <.001

Fisheater 82.17 44.35 24.39 20.77 17.45 21.38 <.001
Mean PBB (ppb)

Non-ﬁsheater 3.38 2.40 2.28 2.13 1.97 5.33 .02

Fisheater 1.82 1.95 2.05 1.55 1.83 2.05 .15

 

*: F values were testing for linearity and were calculated after contaminant levels had been log transformed.

Ml

MI

84

Table 22. Effect of region on contaminant levels stratiﬁed by eating status and gender, 1979-82

 

 

 

 

Livingarea
North Middle South F value“ p value

Male
Mean. Aroclor 1260 (ppb)

Non-ﬁsheater 1 1.96 9.03 9.09 0.80 .37

Fisheater 36.17 31.40 26.01 12.43 <.001
Mean DDT (ppb)

Non-ﬁsheater 20.79 15. 19 14.27 4.68 .03

Fisheater 74.33 62.22 38.56 20.27 <.001
Mean DDE (ppb)

Non-ﬁsheater 17.91 13.17 12.49 3.18 .08

Fisheater 64.40 53.91 32.78 20.71 <.001
Mean PBB (PPb)

Non-ﬁsheater 2.88 4.83 2.20 8.90 .003

Fisheater 4.33 5.60 1.49 62.60 <.001
Female
Mean. Aroclor 1260 (ppb)

Non-ﬁsheater 9.29 7.06 7.63 0.49 .48

Fisheater 22.05 19.03 18.39 0.07 .79
Mean DDT (ppb)

Non-ﬁsheater 17.23 15.94 13.20 1.25 .26

Fisheater 35. 18 33.92 29.42 0.60 .44
Mean DDE (ppb)

Non-ﬁsheater 14.45 12.40 1 1.75 0.33 .56

Fisheater 29.99 29. 10 24.77 0.94 .33
Mean PBB (ppb)

Non-ﬁsheater 1.83 3.54 1.10 10.47 .001

Fisheater l .76 2.86 1.27 4.25 .04

 

*: F values were testing for linearity and were calculated after contaminant levels had been log transformed.

85

Table 23-a. Effect of body mass index on contaminant levels stratiﬁed by eating status in males, 1979-82

 

Body Mass Index
.10-.23 .24-.25 .26-.28 .29-.45 F value"I p value

Mean. Aroclor 1260 (ppb)

Non-fisheater 7.52 9.22 10.01 12.43 1 1.00 .001

Fisheater 28.20 29.25 34.19 31.12 3.83 .05
Mean DDT (PPb)

Non-ﬁsheater 8.59 15.85 20.58 20.10 20.75 <.001

Fisheater 50.33 50.50 59.24 65.62 13.47 <.001
Mean DDE (ppb)

Non-fisheater 7.53 13.67 17.72 17.59 24.91 <.001

Fisheater 43.02 43.65 51.1 1 56.75 13.57 <.001
Mean PBB (PPb)

Non-ﬁsheater 3.08 3.10 3.51 3.17 0.03 .86

Fisheater 4.51 3.18 3.56 3.14 1.50 .22

 

Table 23-b. Effect of body mass index on contaminant levels stratified by eating status in females, 1979-82
Body Mass Index
.18-.20 .21-.22 .23—.26 .27-.39 F value" p value

 

Mean. Aroclor 1260 (ppb)

Non-ﬁsheater 6.51 9.09 7.82 8.38 1.51 .22

Fisheater 15 .94 22.28 18.26 22.31 0.33 .57
Mean DDT (PPb)

Non-ﬁsheater 1 1.54 16.56 13.79 19.71 3.88 .05

Fisheater 19.48 36.49 32.10 42.28 5.86 .02
Mean DDE (ppb)

Non-ﬁsheater 10.12 14.22 12.36 14.62 3.78 .05

Fisheater 16.74 30.99 27.70 35.70 5.83 .02
Mean PBB (ppb)

Non-ﬁsheater 2.21 2.37 1.72 2.53 0.08 .78

Fisheater 2.41 1.95 1.83 1.53 7.32 .01

 

*: F values were testing for linearity and were calculated after contaminant levels had been log transformed.

86

Table 24. Effect of chemicals exposure status on contaminant levels stratiﬁed by eating status and gender,
1979-82

 

Chemical exposure

 

 

 

Yes No t value" p value

Male
Mean. Aroclor 1260 (ppb)

Non-ﬁsheater 9.78 9.85 0.88 .38

Fisheater 28.55 32.15 1.09 .28
Mean DDT (ppb)

Non-ﬁsheater 15.82 16.58 0.71 .48

Fisheater 53.10 58.71 1.24 .22
Mean DDE (ppb)

Non-ﬁsheater 13.81 14.34 0.68 .50

Fisheater 45 .75 50.62 1.21 .23
Mean PBB (ppb)

Non-ﬁsheater 3.26 3.18 0.37 .71

Fisheater 3.65 3.56 0.10 .92
Female
Mean. Aroclor 1260 (ppb)

Non-ﬁsheater 6.75 8. 15 1.05 .30

Fisheater 17.42 20.01 0.80 .43
Mean DDT (ppb)

Non-ﬁsheater 13.83 15.69 0.45 .65

Fisheater 26.5 1 33.53 0.40 .69
Mean DDE (PPb)

Non-ﬁsheater 12.1 1 12.94 0.24 .81

Fisheater 22.65 28.50 0.31 .76
Mean PBB (PPb)

Non-ﬁsheater 2.31 2.16 1.38 .17

Fisheater 1.74 1.95 0.54 .59

 

*: t values were calculated after contaminant levels had been log transformed.

87

Table 25-a. Effect of annual ﬁsh meals on contaminants levels in males, 1979-82
group of annual ﬁsh meals

 

 

None 12—48 49-63 64-180 XZIF value“ p value
Mean. Aroclor 1260 (ppb) 10.37 26.21 30.36 35.73 261.14 <.001
Pct. Aroclor 1260 212.91 <.001
Low 70.4 17.4 14.4 10.1
Middle 22.8 42.4 39.7 33.6
High 6.8 40.2 45.9 56.3
Mean DDT (ppb) 17.52 43.71 58.45 67.17 140.48 <.001
Pct. DDT 124.30 <.001
Low 60.2 23.5 19.2 15.1
Middle 29.1 37.9 37.0 31.1
High 10.7 38.6 43.8 53.8
Mean DDE (ppb) 16.65 48.33 69.66 73.25 140.10 <.001
Pct. DDE 120.90 <.001
Low 59.2 22.7 20.5 16.0
Middle 30. 1 40.2 34.9 29.4
High 107 37.1 44.5 54.6
Mean PBB (ppb) 15.19 37.26 50.49 58.24 0.56 .45
Pct. PBB 2.26 .89
Low 31.6 32.6 34.9 36.1
Middle 32.5 31.8 34.2 34.5
High 35.9 35.6 30.8 29.4

 

Aroclor 1260 (ppb): Low=(1.5, 10.5); Middle=(10.6, 23.9); High=(24.0, 202.7)

DDT (ppb): Low=(0.5, 15.1); Middle=(15.2, 34.6); High=(34.7, 512.6)
DDE (ppb): Low=(0.5,12.8); Middle=(12.9,28.8); High=(28.9, 425.3)
PBB (ppb): Low=(0.5, 1.0); Middle=( 1.1, 3.2); High=( 3.3, 33.1)

*: F values were testing for linearity and were calculated after contaminant levels had been log transformed.

Table 25-b. Effect of annual ﬁsh meals on contaminants levels in females, 1979-82

 

goup of annual ﬁsh meals

 

 

 

None 1452 53-72 73-600 xszvalue" pvalue
Mean. Aroclor 1260 (ppb) 8.17 17.14 15.75 25.87 124.77 <.001
Pct. Aroclor1260 118.63 <.001
Low 47.9 11.3 16.4 9.2
Middle 36.5 35.0 27.3 23.1
High 15.6 53.8 56.4 67.7
Mean DDT (ppb) 15.73 24.27 36.69 38.96 57.55 <.001
Pct. DDT 45.27 <.001
Low 42.7 20.0 23.6 15.4
Middle 34.7 32.5 29.1 32.3
High 22.6 47.5 47.3 52.3
Mean DDE (ppb) 13.08 20.67 31.18 33.20 52.78 <.001
Pct. DDE 44.17 <.001
Low 42.7 23.8 23.6 13.8
Middle 34.7 27.5 32.7 33.8
High 22.6 48.8 43.6 52.3
Mean PBB (ppb) 2.21 1.93 1.58 2.09 0.35 .55
Pet. PBB 6.76 .34
Low 39.6 48.8 50.9 36.9
Middle 27.4 18.8 27.3 27.7
High 33.0 32.5 21.8 35.4
Aroclor1260(ppb):Low=(1.5,5.7); Middle=(5.8, 12.5); High=(12.6, 146.9)
DDT (ppb): Low=(05, 9.2); Middle=(93, 20.3); High=(20.4, 403.9)
DDE(ppb): Low=(0.5, 8.3); Middle=(8.4,17.1); High=(l7.2, 38.8)
PBB(ppb): Low=(0.5,0.5); Middle=(0.6, 2.0); High=( 2.1, 16.2)

I“: F values were testing for linearity and were calculated after contaminant levels had been log transformed.

M
Pc

Ml

89

Table 26-a. Effect of the total weight of ﬁsh consumed on contaminants levels in males, 1979-82
group of total weight of ﬁsh consumed

 

 

 

None 7-32 33-53 54-413 xz/F value"I p value
Mean. Aroclor 1260 (ppb) 10.37 29.57 30.06 32.29 211.83 <.001
Pct. Aroclor 1260 206.27 <.001
Low 70.4 15.5 12.3 14.2
Middle 22.8 39.9 37.7 38.6
High 6.8 44.6 50.0 47.2
Mean DDT (ppb) 17.52 51.80 59.31 58.22 110.89 <.001
Pct. DDT 119.23 <.001
Low 60.2 19.6 18.9 19.7
Middle 29.1 37.8 32.0 36.2
High 10.7 42.6 49.2 44.1
Mean DDE (ppb) 15.19 44.43 51.02 50.55 106.97 <.001
Pct. DDE 113.62 <.001
Low 59.2 20.3 18.9 20.5
Middle 30.1 37.2 32.0 35.4
High 10.7 42.6 49.2 44.1
Mean PBB (ppb) 3.32 3.41 3.30 3.93 0.20 .65
Pct. PBB 4.28 .64
Low 31.6 38.5 32.0 32.3
Middle 32.5 30.4 38.5 32.3
High 35.9 31.1 29.5 35 .4
Aroclor 1260 (ppb): Low=(1.5, 10.5); Middle=(10.6, 23.9); High=(24.0, 202.7)
DDT (ppb): Low=(0.5, 15.1); Middle=(15.2, 34.6); High=(34.7, 512.6)
DDE (ppb): Low=(0.5, 12.8); Middle=(12.9, 28.8); High=(28.9, 425.3)
PBB (ppb): Low=(0.5, 1.0); Middle=( 1.1, 3.2); High=( 3.3, 33.1)

‘: F values were testing for linearity and were calculated after contaminant levels had been log transformed.

M
Pc

Ml
PC

Mt
PC‘

90

Table 26—b. Effect of total weight of ﬁsh consumed on contaminants levels in females, 1979-82
group of total weight of ﬁsh consumed

 

 

 

None 727 28-41 42413 leF value" p value
Mean. Aroclor 1260 (PPb) 8.17 17.54 16.41 25.02 122.08 <.001
Pct. Aroclor 1260 117.75 <.001
Low 47.9 8.8 14.9 12.3
Middle 36.5 36.8 26.9 23.1
High 15.6 54.4 58.2 64.6
Mean DDT (PPb) 15.73 31.92 27.27 38.39 51.64 <.001
Pct. DDT 49.05 <.001
Low 42.7 16.2 26.9 15.4
Middle 34.7 33.8 32.8 27.7
High 22.6 50.0 40.3 56.9
Mean DDE (PPb) 13.08 27.06 23.34 32.66 47.98 <.001
Pct. DDE 45.25 <.001
Low 42.7 19.1 25.4 16.9
Middle 34.7 30.9 34.3 27.7
High 22.6 50.0 40.3 55.4
Mean PBB (ppb) 2.21 1.95 1.55 2.16 0.26 .61
Pct. PBB 5.34 .50
Low 39.6 51.5 46.3 38.5
Middle 27.4 22. 1 26.9 23.1
High 33.0 26.5 26.9 38.5
Aroclor 1260 (ppb): Low=(1.5, 5.7); Middle=(5.8, 12.5); High=(12.6, 146.9)
DDT (ppb): Low=(0.5,9.2); Middle=(9.3, 20.3); High=(20.4, 403.9)
DDE (ppb): Low=(0.5, 8.3); Middle=(8.4, 17.1); High=(17.2, 38.8)
PBB (ppb): Low=(0.5,0.5); Middle=(0.6, 2.0); High=( 2.1, 16.2)

*: F values were testing for linearity and were calculated after contaminant levels had been log transformed.

1 .

91

Table 27. Changes in quantity of ﬁsh consumption among ﬁsheaters stratiﬁed by gender from survey I to

 

 

 

 

survey 11
Mean
# pairs I II t-value“ p—value

Annual ﬁsh meals

male 34 68.0 57.4 1.21 .24

female 21 93.6 61.3 3.88 .001
Serving size (oz)

male 34 13.2 12.0 1.45 .16

female 21 10.3 7.9 3.18 .005
Total weight of ﬁsh consumed (lb)

male 34 54.8 41.2 2.00 .05

female 21 60.8 33. 1 3.43 .003
* paired t-test

Table 28. Changgs in ﬁsh consumption behaviors among ﬁsheaters stratiﬁed bxgender from survey I to II

 

 

 

Male Female
#pairs 1 II p-value* ifpairs I II p-value“
Always trim away prior to cooking
Dorsal fat area 34 0.0 50.0 <.001 21 9.5 52.4 .004
Belly fat area 34 5.9 50.0 <.001 21 9.5 52.4 .004
Remove skin 34 0.0 20.6 .02 21 0.0 23.8 .06
One of top 3 cooking methods
Broiled 34 23.5 32.4 .58 21 28.6 23.8 1.00
Baked 34 11.8 0.0 . 13 21 38.1 14.3 .06
Pan Fried 34 47.1 44.1 1.00 21 38.1 19.0 .29
Deep Fried 34 8.8 23.5 .18 21 0.0 14.3 .25
One of top 3 ﬁsh species consumed
Trout 34 100 82.4 .03 21 95.2 71.4 .06
Salmon 34 61.8 64.7 1.00 21 61.9 76.2 .51
Perch 34 32.4 50.0 .15 21 38.1 28.6 .75
Catﬁsh 34 5.9 0.0 .50 21 4.8 0.0 1.00
White ﬁsh 34 26.5 0.0 .004 21 14.3 0.0 .25

 

 

* McNemar test

92

Table 29. Changes in levels serum Aroclor 1260 stratiﬁed by gender and eating status from survey I to II

 

 

 

Mean
Aroclor 1260 (ppb) # pairs I II t-value"I p-value
Male 58 21.3 34.0 -6.70 <.001
Fisheater 26.3 43.2 -5.19 <.001
Nonﬁsheater 12.4 17.7 4.37 <.001
Female 65 8.3 15.2 -6.66 <.001
Fisheater 12.7 22.9 -5. 13 <.001
Nonﬁsheater 6.2 l 1.6 -4.84 <.001

 

* paired t-test was performed after log transformation.

93

Table 30. Changes in quantity of ﬁsh consumption among ﬁsheaters stratiﬁed by gender from survey II to
survey III

 

 

 

Mean
# pairs 11 III t-value"I p-value

Annual ﬁsh meals

male 238 61.3 41.9 7.87 <.001

female 120 66.0 35.9 7.54 <.001
Serving size (02.)

male 238 13.7 9.29 l 1.08 <.001

female 120 10.1 7.1 8.02 <.001
Total weight of ﬁsh consumed (1b.)

male 238 51.7 24.7 9.77 <.001

female 120 40.2 15.8 9.91 <.001

 

* paired t-test

Table 31. Changes in ﬁsh consumption behaviors among ﬁsheaters stratiﬁed by gender from survey 11 to
survey 111

 

 

 

Male Female
#pairs 11 III p-value* #pairs II III p-value*
Always trim away prior to cooking
Dorsal fat area 238 71.8 91.2 <.001 120 61.7 88.3 <.001
Belly fat area 238 71.8 92.4 <.001 120 61.7 88.3 <.001
Remove skin 238 39.9 83.2 <.001 120 37.5 78.3 <.001
One of top 3 cooking methods
Broiled 238 44.5 29.0 <.001 120 42.5 33.3 .15
Baked 238 0.0 27.7 <.001 120 2.5 39.2 <.001
Pan Fried 238 55.9 46.6 .05 120 37.5 40.0 .78
Deep Fried 238 27.7 29.8 .65 120 25.0 26.7 .86
One of top 3 ﬁsh species consumed
Trout 238 95.0 47.1 <.001 120 90.8 45.8 <.001
Salmon 238 89.9 58.4 <.001 120 93.3 60.8 <.001
Perch 238 41.2 52.1 .007 120 31.2 45.0 .04
Catﬁsh 238 1.7 0.4 .38 120 3.3 0.0 .13
White ﬁsh 238 0.0 100 <.001 120 0.0 100 <.001

 

 

* McNemar test

94

Table 32. Changes in levels of serum Aroclor 1260, DDT, DDE, and PBB stratiﬁed by gender and ﬁsh
eating status from survey 11 to survey 111

 

 

 

Mean
# pairs 11 III t-value* p-value
Aroclor 1260 (PPb)

Male 212 26.3 21.7 8.68 <.001
Fisheater 139 34.9 29.2 7.53 <.001
Nonﬁsheater 73 10.5 8.0 2.88 .01

Female 179 1 1.1 10.0 1.92 .06

Fisheater 64 17.0 14.8 2.16 .04
Nonﬁsheater 1 15 7.8 7.2 0.95 .35
DDT (ppb)

Male 212 48.6 19.9 15.54 <.001
Fisheater 139 65.4 25.4 13.12 <.001
Nonﬁsheater 73 17.8 9.8 8.62 <.001

Female 179 20.8 11.8 11.10 <.001

Fisheater 64 28.2 15.8 8.36 <.001

Nonﬁsheater 1 15 16.5 9.6 7.82 <.001

DDE (ppb)

Male 212 42.0 17.8 15.1 <.001
Fisheater 139 56.4 22.7 12.79 <.001
Nonﬁsheater 73 15.5 8.7 8.34 <.001

Female 179 16.9 10.5 10.22 <.001

Fisheater 64 24.0 14. 1 7.63 <.001
Nonﬁsheater 1 15 12.9 8.5 7.22 <.001
PBB (ppb)

Male 212 4.8 4.3 0.71 .48
Fisheater 139 5.2 4.6 0.68 .50
Nonﬁsheater 73 4.0 3.7 0.28 .78

Female 179 2.6 2.6 -2.81 .01

Fisheater 64 2.3 2.5 -2.54 .01

Nonﬁsheater 1 15 2.7 2.7 -1.64 . 10

 

"' paired t-test was performed after contaminant levels had been log transformation.

95

Table 33. Changes in quantity of ﬁsh consumption among ﬁsheaters stratiﬁed by gender from survey I to
survey III.

 

 

Mean
# pairs I III t-value"I p-value

Annual ﬁsh meals

male 15 67.4 43.5 1.44 .12

female 13 123.0 31.0 3.30 .01
Serving size (02.)

male 15 12.5 9.6 1.74 .10

female 13 9.1 7.4 1.60 .15
Total weight of ﬁsh consumed (1b.)

male 15 44.0 25.8 2.25 .04

female 1 3 68. 1 13.4 3.02 .02

 

* paired t-test

Table 34. Changes in ﬁsh consumption behaviors among ﬁsheaters stratiﬁed by gender from survey Ito III

 

 

 

Male Female
#pairs I III p-value* #pairs I III p-value“
Always trim away prior to cooking
Dorsal fat area 15 0.0 80.0 .001 13 15.4 69.2 .04
Belly fat area 15 6.7 80.0 .001 13 15.4 69.2 .04
Remove skin 15 0.0 53.3 .008 13 0.0 46.2 03
One of top 3 cooking methods
Broiled 15 13.3 6.7 1.00 13 30.8 30.8 1.00
Baked 15 13.3 40.0 .22 13 30.8 61.5 .29
Pan Fried 15 46.7 60.0 .73 13 30.8 46.2 .69
Deep Fried 15 13.3 13.3 1.00 13 0.0 7.7 1.00
One of top 3 ﬁsh species consumed
Trout 15 100 33.3 .002 13 92.3 61.5 .13
Salmon 15 46.7 46.7 1.00 13 61.5 30.8 .13
Perch 15 40.0 60.0 .38 13 38.5 23.1 .69
Catﬁsh 15 33.3 100 .002 13 - - -
White ﬁsh 15 0.0 80.0 <.001 13 23.8 100 .002

 

 

* McNemar test

96

Table 35. Changes in levels (ppb) of serum Aroclor 1260 stratiﬁed by gender and eating status from survey

 

 

 

Ito survey III
Mean
Aroclor 1260 (ppb) # pairs I III t-value* p-value
Male 15 20.1 32.8 -0.98 .35
Fisheater 6 38.8 70.5 -0.52 .63
Nonﬁsheater 9 10.8 14.0 -0.78 .45
Female 28 7.5 12.7 -3.76 .001
Fisheater 9 13.6 19.2 -2.59 .03
Nonﬁsheater 19 4.7 9.6 -2.95 .01

 

* paired t-test was performed after contaminant levels had been log transformation.

97

Table 36. Adjusted linear regression of annual ﬁsh meals on demographic characteristics among ﬁsheaters
stratiﬁed by gender and survey

 

 

 

 

 

Men Women
BiSE p thSE p
Survey I
R square . l 3 .38
Age .40:1:.80 .62 -.7 1196 47
Education
Elementary - - - -
Some high school -26.26i35.89 .47 -41.87:1:57.09 .48
High school graduate -5.85:1:29.86 .85 -93.75:1:-43.96 .05
Some college -31.38i36.93 .40 -70.81:1:41.92 .12
College and above -30.99:1:32.90 .35 -42.35i44.90 .36
Region
North - -
Middle -10.4 11:22.00 .64 -80.78:t43.46 .09
South I t t *
Body mass index 120.372t233.01 .61 546.20:|:429.32 .23
Survey 11
R square .04 .09
Age .18zl:.11 .11 3921:.27 .14
Education
Elementary - - - ‘-
Some high school 5.20:t7.66 .50 -25.10:1:17.73 .16
High school graduate -1 9816.92 .77 -30.72i15.78 .05
Some college .26:t7.35 .97 -32.78:1:16.74 .05
College and above -8.31i7.46 .27 -28.61i17.32 .10
Region
North - - - -
Middle 6.96:1:4.04 .09 36.17:1:1 1.24 .002
South 3.52:1:3.33 .29 4.40:t7.76 .57
Body mass index 53.51:1:41.98 .20 120018793 .89
Survey III
R square .04 .19
Age .3321:.24 .19 -.02i.24 .94
Education
Elementary - - - -
Some high school 6.90:1:19.55 .72 16.76zl:21.81 .45
High school graduate 16.99i17.66 .34 33.73i17.68 .06
Some college 16.81:1:17.77 .35 41.08:1:18.24 .03
College and above 10.57i18.45 .58 32.07i18.43 .09
Region
North - - - -
Middle 6.25:1:9.48 .5 1 25.6921:10.77 .02
South -5.88i7.7 8 .45 -4.54:1:-6.87 .51
Body mass index "‘ * * *

 

 

 

B :1: SE: Regression coefﬁcient :1: Standard error

- : Reference group
*: No data

98

Table 37. Adjusted linear regression of total weight of ﬁsh consumed on demographic characteristics
amongﬁsheaters stratiﬁed by gender and survey

 

 

 

 

 

Men Women
BiSE p thSE p
Survey I
R square .35 .32
Age .61i.50 .24 -.43.t.76 .58
Education
Elementary - - - -
Some high school 11.3 11122.58 .62 9.91:1:45.l7 .83
High school graduate 23.0321:18.79 .23 -10.79i34.78 .76
Some college 76323.23 .97 -4.63:t33.17 .89
College and above 1.89:1:20.70 .93 50.68:l:35.53 .18
Region
North - - - -
Middle -23.92:1: 1 3.84 .10 -24.2 1134.39 .49
South * * * *
Body mass index 239.632t146.59 .1 l 327.18:1:339.74 .35
Survey 11
R square .05 .09
Age -. 181:. 15 .22 -.002i.21 .99
Education
Elementary - - - -
Some high school 1.04:1:9.94 .92 ~10.81:1:13.71 .43
High school graduate -3.19i8.99 .72 -l6.55:1:12.20 .18
Some college -4.2l:1:9.54 .66 -27.43i12.94 .04
College and above -17.60i9.68 .07 -23.043cl3.38 .09
Region
North - - - -
Middle 5.27525 .32 28.25zt8.69 .001
South 106014.32 .01 .17:1:6.00 .98
Body mass index 87.53:t54.48 .11 12.25i67.95 .86
Survey III
R square .03 .07
Age 08:. 18 .66 -.09i. 13 .52
Education
Elementary - - - -
Some high school 4.25il4.16 .76 4.61:1:12.32 .71
High school graduate 7.23:1:12.80 .57 12.3321:9.99 .22
Some college 8.52:1:12.87 .50 12.82i10.31 .22
College and above .47:1:l3.37 .97 8.11:1:10.42 .44
Region
North - - - -
Middle .77:1:6.87 .91 5.83zt6.09 .34
South -6.14i5.63 .28 .7 1:388 .86
Body mass index "' * * *

 

 

 

B :1: SE: Regression coefﬁcient i Standard error

- : reference group
*: no data

Table 38. Adjusted linear regression of levels of Aroclor 1254 (log transformed) on demographic

99

characteristics and quartiles of total weight of ﬁsh consumed stratiﬁed byggnder, 1973-74

 

 

 

Men Women
B1SE p B1SE p
R square .53

Age .011.01 .07 .021.01 <.001
Education

Elementary - - - -

Some high school -.271.35 .45 221.32 .49

High school graduate -.051.26 .84 -.111.18 .54

Some college .101.32 .75 -. 121.20 .57

College and above -.401.26 .12 091.20 .67
Region

North - - - -

Middle .201.18 .28 . 191.14 . 18

South * "‘ .401.48 .41
Body mass index .0212.32 .99 -3.2111.56 .04
Eating status

Non-Fisheater - - - -

Fisheater . 121.30 .69 -.021.26 .93
Quartile of total weight of ﬁsh consumed

1 (male: 24-28; female: 24-28 ) - - - -

2 (male: 29-39; female: 29-39) .711.26 .01 .471.19 .02

3 (male: 40-53; female: 40-54) .711.39 .07 ,79122 .001

4 (male: 54-413; female: 55-163) 1.191.38 .003 .871.33 .01

 

 

 

Note: Aroclor 1254 was tested in 1973-74 survey only.

B 1 SE: Regression coefﬁcient 1 Standard error

- : Reference group
*: No data

100

Table 39. Adjusted linear regression of levels of Aroclor (log transformed) on demographic characteristics

and quartile of total weight of ﬁsh consumed stratiﬁed by gender and survey

 

 

 

 

Men Women
B18E P 815E p
Survey I
R square .53 .63
Age .011.01 .09 .021.01 .002
Education
Elementary - - - -
Some high school -.291.37 .45 .17136 .63
High school graduate -.061.28 .82 .121.21 .55
Some college -.01134 .98 -.131.23 .56
College and above -.511.27 .07 .171.22 .45
Region
North - — - -
Middle .111.19 .58 .411.16 .01
South "' "' .311.54 .57
Body mass index 29012.46 .25 -6.2211.75 .001
Eating status
Non-Fisheater - - - -
Fisheater . 171.32 .59 -. 121.30 .68
Quartile of total weight of ﬁsh consumed
1 (male: 24-28; female: 24-28 ) - - - -
2 (male: 29-39; female: 29-39) .741.28 .01 631.22 .01
3 (male: 40-53; female: 40-54) .731.41 .08 1.021.24 <.001
4 (male: 54-413; female: 55-163) 1.231.40 .004 1.391.37 <.001
Survey II
R square .48 .42
Age 021.00 <.001 .021.00 <.001
Education
Elementary - - - -
Some high school -.101.13 .43 -.261.16 .11
High school graduate -.001. 12 .99 -.281.14 .05
Some college -.081.12 .54 -.361. 15 .02
College and above -.101. 12 .40 -.371. 15 .02
Region
North - - - -
Middle 361.08 <.001 . 181. 10 .04
South . 121.06 .06 .031.07 .63
Body mass index 1.071.78 .17 -1.041.80 .19
Eating status
Non-Fisheater - - - -
Fisheater .381. 1 8 .04 .061. 15 .71
Quartile of annual ﬁsh meals
1 (male: 7-28; female: 7-28 ) - - - -
2 (male: 29-40; female: 29-33) .571.18 .002 - -
3 (male: 41-60: female: 34-46) .761.19 <.001 .761.15 <.001
4 (male: 61-413; female: 47413) .741.19 <.001 .871.16 <.001

 

 

(To be Continued)

101

Table 39. Adjusted linear regression of levels of Aroclor 1260 (log transformed) on demographic
characteristics and quartile of total weiLht of ﬁsh consumed stratiﬁed by gender and survey (continueg

 

 

 

 

 

Men Women
B1SE P B1SE P
Survey III
R square .48 .49
Age 021.00 <.001 .031.00 <.001
Education
Elementary - - - -
Some high school -. 131.28 .63 -1.011.25 <.001
High school graduate -.071.23 .78 -.6l1.19 .002
Some college -.091.24 .70 -.581.20 .005
College and above -.341.24 .16 -.641.21 .002
Region
North - - - -
Middle .461. 16 .005 .3 11. 14 .04
South .301.13 .02 .231.1 1 .03
Body mass index * * * *
Eating status
Non-Fisheater - - - -
Fisheater 1.041. 13 <.001 531.10 <.001
Quartile of total weight of ﬁsh consumed
1 (male: 0.4-9; female: 0.3-4 ) - - -
2 (male: 10-18; female: 5-10) .101.15 .49 .471.14 .001
3 (male: 19-31; female: 11-19) .071.17 .69 .351.12 .004
4 (male: 32-159; female: 20%) 271.18 .14 .491.13 <.001

 

B 1 SE: Regression coefﬁcient 1 Standard error

Reference group
No data

Table 40. Adjusted linear regression of levels of DDT (log transfonned) on demographic characteristics and

102

quartile of total weight of ﬁsh consumed stratiﬁed by gender and survgg

 

 

 

 

 

 

Men Women
B1SE p 8185 p
Survey 11
R square .40 .32
Age 031.00 <.001 031.00 <.001
Education
Elementary - - - -
Some high school 011.18 .98 -. 171.18 .35
High school graduate .171.16 .28 -.211. 16 .18
Some college .101.17 .54 -.291. 17 .08
College and above .121.17 .46 -.521. 17 .003
Region
North - - - -
Middle .591. 10 <.001 091.1 1 .43
South . 161.08 .05 .05108 .53
Body mass index 4.041105 <.001 1.111.90 .22
Eating status
Non-Fisheater - - - -
Fisheater .391.24 1 1 081.17 .63
Quartile of total weight of ﬁsh consumed
1 (male: 7-28; female: 7-28) - - - -
2 (male: 29-40; female: 29-33) .451.24 07 - -
3 (male: 41-60; female: 34-46) .691.26 .01 .531.17 .002
4 (male: 61-413; female: 47413) 661.25 .01 531.18 <005
Survey III
R square .36 .27
Age 031.00 <.001 .031005 <.001
Education
Elementary - - - -
Some high school -. 181.33 .58 -.871.35 .01
High school graduate -. 101.28 .71 361.27 .19
Some college -.151.29 .60 -.301.29 .29
College and above -.431.13 .15 -.56:1;29 .06
Region
North - - - -
Middle 631.19 .001 .391.20 .06
South .421. 15 .001 .331. 15 .03
Body mass index * * * *
Eating status
Non-Fisheater - - - -
Fisheater .781. 16 <.001 .261. 15 .08
Quartile of total weight of ﬁsh consumed
1 (male: 0.4—9; female: 0.3-4 ) - - - -
2 (male: 10-18; female: 5-10) -061.18 .76 341.20 .08
3 (male: 19-31; female: 11-19) -.081.20 .70 .361.17 .04
4 (male: 32-159; female: 20-93) 051.22 83 281.18 .13

 

Note: DDT was tested in 1979-82 and 1989-91 survey only.
B 1 SE: Regression coefﬁcient 1 Standard error; - : Reference group; *: No data

103

Table 41. Adjusted linear regression of levels of DDE (log transformed) on demographic characteristics and
quartile of total weight of ﬁsh consumed stratiﬁed by gender and survey

 

 

 

 

Men Women
B1SE p B1SE p
Survey II
R square .39 .32
Age 031.00 <.001 .03100 <.001
Education
Elementary - - - -
Some high school .101.17 .55 -. 141.18 .45
High school graduate .211.15 .17 -. 181.15 .23
Some college .121. 16 .46 -.281.16 .09
College and above .141.16 .39 -.5 11.17 .002
Region
North - - - -
Middle .571. 10 <.001 071.1 1 .48
South . 141.08 .07 .05108 .48
Body mass index 4.191103 <.001 .721.87 .41
Eating status
Non-Fisheater - - - -
Fisheater 371.24 . 1 1 081.17 .65
Quartile of total weight of ﬁsh consumed
1 (male: 7-28; female: 7-28 ) - - - -
2 (male: 29-40; female: 29-33) 401.24 .10 - -
3 (male: 41-60; female: 34-46) .63124 .01 .491.17 .003
4 (male: 61-413: female: 47-413) 031.25 .01 .481.18 .001
Survey [11
R square .38 .27
Age 031.00 <.001 .03101 <.001
Education
Elementary - - - -
Some high school -.181.32 .58 -.831.35 .02
High school graduate -111.27 .69 -.351.27 .21
Some college -.101.28 .73 .28129 .33
College and above -.411.28 .14 -.531.29 .07
Region
North - - - -
Middle .601.18 .001 .391.20 .06
South .431. 15 .003 .331. 15 .03
Body mass index * * * *
Eating status
Non-Fisheater - - - -
Fisheater .771. 16 <.001 .261. 14 .08
Quartile of total weight of ﬁsh consumed
1 (male: 0.4-9; female: 0.3-4 ) - - - -
2 (male: 10-18; female: 5-10) -041.17 .80 351.19 .07
3 (male: 19-31; female: 11-19) -.081. 19 .68 .361.17 .03
4 (male: 32-159; female: 20-93) .1 11.21 .62 .291.18 .11

 

 

*Notez DDE was tested in 1979-82 and 1989-91 survey only.

B 1 SE: Regression coefﬁcient 1 Standard error

- : Reference group
*: No data

 

104

Table 42. Adjusted linear regression of levels of PBB (log transformed) on demographic characteristics and
quartile of total weight of ﬁsh consumed stratiﬁed by gender and survey

 

 

 

 

 

 

Men Women
B18E p B1SE p
Survey I]
R square .28 .22
Age -003100 .25 0031.00 .28
Education
Elementary - - - -
Some high school 241.19 .21 -. 141.20 .48
High school graduate 041.17 .83 -.221.17 .20
Some college -. 101.18 .55 -.291.18 .10
College and above -081.18 .65 -.411.18 .02
Region
North - - - -
Middle 1 071.1 1 <.001 .641.12 <.001
South 1.19109 <.001 .91109 <.001
Body mass index -3061l.14 .01 -2.231.97 .02
Eating status
Non-Fisheater - - - -
Fisheater 341.26 . 19 -061. 19 .76
Quartile of total weight of ﬁsh consumed
1 (male: 7-28; female: 7-28 ) - - - -
2 (male: 29-40; female: 29-33) -.351.26 .18 - -
3 (male: 41-60; female: 34-46) -.321.28 .22 -.011.19 .94
4 (male: 61-413; female: 47-413) -.411.28 .14 041.20 .84
Survey III
R square .17 .17
Age .00101 .50 .01101 . 16
Education
Elementary - - - -
Some high school 391.33 .24 071.38 .86
High school graduate -.071.28 .80 -021.29 .95
Some college -.251.29 .38 -. 161.31 .61
College and above -.241.29 .40 -.301.31 .34
Region
North - - - -
Middle .631. 19 .001 441.22 .04
South .901. 15 <.001 .701. 16 <.001
Body mass index * * * *
Eating status
Non-Fisheater - - - -
Fisheater 031.16 .85 -.281. 16 .07
Quartile of total weight of ﬁsh consumed
1 (male: 0.4-9; female: 0.3-4 ) - - - -
2 (male: 10-18; female: 5-10) -.041.18 .82 .101.21 .65
3 (male: 19-31; female: 1 1-19) -.041.20 .86 .371.18 .04
4 (male: 32-159; female: 20-93) .151.22 .49 591.19 .002

 

Note: PBB was tested in 1979-82 and 1989-91 survey only.

B 1 SE: Regression coefﬁcient 1 Standard error

- : Reference group
*: No data

105

Table 43. Adjusted linear regression of change in total weight of ﬁsh consumed of predicted against
demographic characteristics and baseline ﬁsh consumption stratiﬁed by gender, from survey I to II and from

 

 

 

 

survey H to 111
Men Women
8181?. p 81813 p
Survey I to II
R square .47 .68
Age . 171.38 .65 -. 141. 17 .42
Education
Elementary - - - -
Some high school -28.99117.21 .10 -39.89110.30 <.001
High school graduate -26.7211306 .05 -7.9815.84 .18
Some college -26.30115.9l .11 -14.8816.25 .02
College and above -8.97112.58 .48 -150716.74 .03
Region
North - - - -
Middle 0319.45 .99 -7. 1014.66 . 1 3
South * * -6.68116.54 .69
Baseline body mass index -139.341123.88 .27 43.78150.49 .39
Baseline total weight of ﬁsh consumed .731. 14 <.001 .59107 <.001
Survey II to 111
R square .72 .78
Age -.1 11.13 .38 021.06 .77
Education
Elementary - - - -
Some high school -3.8518.78 .66 -1.1013.97 .78
High school graduate 80217.70 .30 2621309 .40
Some college -7.7417.97 .33 -1.3013.27 .69
College and above -2061803 .80 -.3013.29 .93
Region
North - - - -
Middle -3.3814.71 .47 -2.7212.26 .23
South 24313.79 .52 -0511.67 .78
Baseline body mass index -10.6114900 .83 120811803 .50
Baseline total weight of ﬁsh consumed .82104 <.001 .84104 <.001

 

 

 

B 1 SE: Regression coefﬁcient 1 Standard error

- : Reference group
*: No data

106

Table 44. Adjusted linear regression of change in Aroclor 1260 levels of predicted against demographic
characteristics, baseline Aroclor 1260 levels, and change in ﬁsh consumption stratiﬁed by gender, from
survey 1 to II and from survey II to 111

 

 

 

Men Women
B1SE p BiSE p
Survey I to II
R square .26 .27
Age 261.32 .42 -. 121.10 .24
Education
Elementary - - - -
Some high school -13.32114.75 .37 -4.3615.94 .47
High school graduate -11.27110.37 .28 9113.35 .79
Some college -20.29113.82 .15 2.181332 .51
College and above -5.90110.35 .57 8413.60 .82
Region
North - - - -
Middle -7.9217.80 .32 -2.1812.58 .40
South * * -6.8518.86 .44
Baseline body mass index -36.02197.49 .71 -7. 16128.00 .80
Eating status
Non-Fisheater - - - -
Fisheater -. 1 0318.84 .91 -7.2213.62 .05
Baseline Aroclor 1260 -.231.19 .23 -.271.24 .24
Quartile of change in total weight
of ﬁsh consumed
1 (malez-95 to ~7; femalez-Sl to 0) - - - -
2 (male: -6 to 5; female: 1 to 6) -l.16111.53 .92 -3.9213.69 .29
3 (male: 6 to 26; female: 7 to 17) 4.811946 .61 -.441304 .89
4 (male: 27 to 105;female: 18 to 95) -9.50110.54 .37 64514.41 .15

 

 

(T o be continued)

107

Table 44. Adjusted linear regression of change in Aroclor 1260 levels of predicted against demographic
characteristics, baseline Aroclor 1260 levels, and change in ﬁsh consumption stratiﬁed by gender, from

 

 

 

 

 

survey I to II and from survey [1 to III (continued)
Men Women
B1SE p B1SE p
Survey II to III
R square .09 .26
Age .02107 .82 -03103 .27
Education
Elementary - - - -
Some high school 2231492 .65 12612.13 .55
High school graduate 2.781433 .52 -.5411.64 .74
Some college -.2014.44 .96 1.1411.7l .50
College and above 8714.45 .85 -.7211.73 .68
Region
North - - - -
Middle -30612.63 .25 -.6711. 18 .57
South 36912.08 .08 1,441.88 . 10
Baseline body mass index -61.71127.30 .02 -10.l319.28 .28
Eating status
Non-Fisheater - - - -
Fisheater 1 .4012.93 .63 -3.5611 .54 .02
Baseline Aroclor 1260 levels .04104 .21 .26105 <.001
Quartile of change in total weight of
ﬁsh consumed
1 (male:-l 13 to —5; femalez-46 to -2) - - - -
2 (male: -4 to 6; female: -1 to 0) 5312.52 .83 -.111.94 .91
3 (male: 7 to 23;female: 1 to 14) 39412.85 .17 25211.75 .15
4 (male: 34 to 395;fema1e: 15 to 89) 3.291298 .27 3.611100 .03

 

B 1 SE: Regression coefﬁcient 1 Standard error

- : Reference group
*: No data

 

108

Table 45-a. Adjusted linear regression of change in DDT levels of predicted against demographic
characteristics and baseline DDT levels, and change in annual ﬁsh meals stratiﬁed by gender, from survey

11 to survey III

 

 

 

Men Women
B1SE p B1SE p
Survey II to II]
R square .88 .86

Age -. 151.12 .23 -. 151.05 .004
Education

Elementary - -

Some high school -10.2717.79 .19 47413.42 .17

High school graduate 2.931696 .67 30112.64 .25

Some college -l.9317.13 .79 55812.81 .05

College and above -.4117. 15 .95 20212.82 .35
Region

North - - - -

Middle 1.241424 .77 -1.2211.98 .54

South -2.661-3.36 .43 - 1 . 1211 .44 .44
Baseline body mass index -132.00143.10 .003 142311584 .37
Eating status

Non-Fisheater - - - -

Fisheater 51013.28 .12 -3.4311.65 .04
Baseline DDT levels .77102 <.001 .68102 <.001
Change in annual ﬁsh meals 091.04 .03 .02102 .36

 

 

 

B 1 SE: Regression coefﬁcient 1 Standard error

- : Reference group
*: No data

109

Table 45-b. Adjusted linear regression of change in DDT levels of predicted against demographic
characteristics and baseline DDT levels, and quartile of change in total weight of ﬁsh consumed stratiﬁed
by gender, from survey II to III

 

 

 

 

 

Men Women
B1SE p B1SE p
Survey 11 to survey III
R square .88 .86
Age -. 151.12 .22 -. 151.05 .003
Education
Elementary - -
Some high school 9.971796 .21 46913.48 .18
High school graduate 3.4717.ll .63 28612.70 .29
Some college -1.4017.23 .85 5.291285 .06
College and above -.211726 .98 2.471287 .39
Region
North - - - -
Middle 1.651483 .70 -1.1311.98 .57
South -3.171-3.39 .35 -1.1 111.46 .45
Baseline body mass index 4360014421 .002 140911587 .38
Eating status
Non—Fisheater - - - -
Fisheater -6.3014.47 . 16 -3.5112.54 .17
Baseline DDT levels .77102 <.001 .68102 <.001
Quartile of Change in total weight of
ﬁsh consumed
1 (male:-l 13 to -5; femalez-46 to -2) - - - -
2 (male: -4 to 6; female: —1 to 0) 3.621409 .38 22711.54 .14
3 (male: 7 to 23;female: 1 to 14) 59314.55 .19 2.531295 .39
4 (male: 34 to 395;fema1e: 15 to 89) 80814.77 .07 30412.69 .26

 

B 1 SE: Regression coefﬁcient 1 Standard error
- : Reference group
*2 No data

110

Table 46-a. Adjusted linear regression of changes in DDE levels of predicted against demographic
characteristics and baseline DDE levels, and change in annual ﬁsh meals stratiﬁed by gender, from survey II

 

 

 

to 111
Men Women
B1SE p B1SE p
Survey 11 to III
R square .87 .81
Age -. 161.11 .13 -.06104 .10
Education
Elementary - - - -
Some high school 93216.89 .18 28912.64 .28
High school graduate 2.1416.16 .73 2.681204 .19
Some college 2.701682 .67 30712.17 .16
College and above -8816.33 .89 10112.18 .46
Region
North - - - -
Middle 10913.76 .77 -.3311.54 .83
South 2.521298 .40 -.9911.1 1 .38
Baseline body mass index -11325138.17 .003 -199711204 .10
Eating status
Non-Fisheater - - - -
Fisheater -5 .091290 .08 -1.2411 .29 .34
Baseline DDE levels .77102 <.001 .58102 <.001
Change in annual ﬁsh meals 081.04 .02 011.02 .65

 

 

 

B 1 SE: Regression coefﬁcient 1 Standard error

- : Reference group
‘: No data

Table 46—b. Adjusted linear regression of change in DDE levels of predicted against demographic

lll

characteristics and baseline DDE levels, and quartile of change in total weight of ﬁsh consumed stratiﬁed

by gender, from survey 11 to III

 

 

 

Men Women
B18E p B1SE p
Survey II to III
R square .81
Age -.171.11 .13 -.07104 .09
Education
Elementary - - - -
Some high school 90317.05 .20 2.531209 .35
High school graduate 2.741629 .66 22912.09 .27
Some college -2. 1416.40 .74 20412.20 .23
College and above -.5916.42 .93 1.241222 .58
Region
North - - - -
Middle 1.551383 .69 -.2011.53 .89
South 2.971300 .32 -8511 . 13 .45
Baseline body mass index -117.55139.14 .003 -19.9911204 .10
Eating status
Non-Fisheater - - — -
Fisheater 0.091395 . 12 -2.2111.96 .26
Baseline DDE levels .77102 <.001 .58102 <.001
Quartile of Change in total weight of
ﬁsh consumed
1 (malez-113 to -5; femalez-46 to -2) - — - -
2 (male: -4 to 6; female: -1 to 0) 3.441302 .34 1.181120 .32
3 (male: 7 to 23;female: 1 to 14) 5.181403 .20 3.081228 .18
4 (male: 34 to 395;fema1e: 15 to 89) 8.241424 .05 2.451208 .24

 

 

 

B 1 SE: Regression coefﬁcient 1 Standard error

- : Reference group
*: No data

Table 47. Adjusted linear regression of change in PBB levels of predicted against demographic

112

characteristics and baseline PBB levels, and quartile of change in total weight of ﬁsh consumed stratiﬁed by

_gender, from survey 11 to survey III

 

 

 

Men Women
B1SE p B1SE p
Survey II to III
R square .40 .23
Age -02101 .10 -011.01 .26
Education
Elementary - - -
Some high school -.201.71 .79 -.231.66 .73
High school graduate -27102 .67 -.511.51 .32
Some college -.431.63 .50 -. 181.54 .74
College and above -08104 .90 -201.55 .72
Region
North - - -
Middle -041.38 .92 -.241.38 .53
South -.321-.31 .31 -011.29 .97
Baseline body mass index 1.431395 .72 -3021298 .31
Eating status
Non-Fisheater - - -
Fisheater 201.39 .61 -.381.48 .43
Baseline PBB levels 251.02 <.001 251.04 <.001
Quartile of change in total weight of
ﬁsh consumed
1 (male:-113 to -5; femalez-46 to -2) - - - -
2 (male: -4 to 6; female: -1 to 0) -081.36 .83 .441.30 .14
3 (male: 7 to 23;female: 1 to 14) -.481.41 .23 011.56 .28
4 (male: 34 to 395 ;female: 15 to 89) 031.42 .94 .751.51 .15

 

 

 

B 1 SE: Regression coefﬁcient 1 Standard error

- : Reference group
*: No data

113

Table 48. Adjusted linear regression of total weight of ﬁsh consumed against general characteristics based
on repeated measure model

 

 

B1SE p
Gender

Female - -
Male 13.381303 <.001
Ase -061.11 .59
Education

Elementary - -

Some high school 142619.30 .13

High school graduate 98618.43 .24

Some college 11.431803 .19

College and above 3.371882 .70
Region

North - -

Middle -4.5313.01 . 13

South -9.5915. 1 3 .06
Survey time

survey I (1973-74) - -

survey 11 (1979-82) 10.121502 -04

survey 111 (1989-91) -35.8515. 13 <.001

 

B 1 SE: Regression coefﬁcient 1 Standard error
- : Reference group

114

Table 49. Adjusted linear regression of Aroclor 1260 levels against general characteristics based on

repeated measure model

 

 

B1SE p
Gender

Female - -
Male .69106 <.001
As9 021.002 <.001
Education

Elementary - -

Some high school .22121 .27

High school graduate .331. 19 .08

Some college .351. 19 .07

College and above 041.19 .85
Region

North - -

Middle 241.06 <.001

South -201. 12 .10
Survey time

survey I (1973-74) - -

survey II (1979-82) .101.11 .37

survey III (1989-91) -.421.11 <.001

 

B 1 SE: Regression coefﬁcient 1 Standard error

- : Reference group

115

Table 50. Adjusted linear regression of Aroclor 1260 levels against total weight of ﬁsh consumed and
_general characteristics based on repeated measure model

 

 

B1SE p
Gender
Female - -
Male .51106 <.001
Age 021.002 <.001
Education
Elementary - -
Some high school 081.19 .67
High school graduate 211.17 .22
Some college 231.17 . 19
College and above 061.18 .75
Region
North - -
Middle .1 11.06 .05
South -. 141.1 1 .20
Total weight of ﬁsh consumed 011.00 <.001
Survey time
survey I (1973-74) - -
survey II (1979-82) 811.10 <.002
survey 111 (1989-91) -.l31.10 .19

 

B 1 SE: Regression coefﬁcient 1 Standard error
- : Reference group

116

Table 51. Adjusted linear regression of Aroclor 1260 levels of predicted against ﬁsh consumption and

_general characteristics stratiﬁed by gender based on repeated measure model

 

 

 

Men Women
B1SE p B1SE p
Age .02100 <.001 021.00 <.001
Education
Elementary - - - -
Some high school .28122 .20 -.641.36 .08
High school graduate 261.19 .17 -.431.34 .21
Some college .40120 .05 -.501.35 .15
College and above .101.20 .60 -.451.35 .20
Region
North - - - -
Middle .13108 .10 .14107 .06
South -.261.14 .06 -.211.15 .17
Survey time
survey I (1973-74) - - - -
survey 11 (1979-82) 241.14 .08 .551.12 <.001
survey 111 (1989-91) -.051.15 .73 231.13 ' .07
Group of total weight of ﬁsh consumed
1(None) - - - -
2 (male: 124; female: 10) .741.11 <.001 .571.11 <.001
3 (male: 25-48: female: 7-24) 1.10110 <.001 731.10 <.001
4 (male: 48+: female: 25+ ) 1.341.10 <.001 891.09 <.001

 

 

 

FIGURES

Onto onto wave

AO/ Alv-

onto 05.8 See

 

 

 

 

3:235 10 83083 05. A oSwE

118

@8135 m :6 E BSEQEHQV

 

 

 

 
 

 

 

 

ae
wmm 14
wmenz mmﬁnz 1 eﬂuz
K . :e f m:
11 411
game: $33: $38:
E 335.01 ﬂ thee/Sm Hemotsm

 

@833 230303-380 m 2: 80¢ moﬁm 038mm .N 8:me

 

1.1

           

36.—82>» .m

8.2.3.. d
8352..”
55 8.3.3.5.."
5.8»...—

42 .318

119

nausea: 3.!

ERIE 3

\

 

 

 

beam 05 E @8960 39:50 58:82 8883 555 .m 8:me

120

 

 

 

(wmwpmmm

 

353:8 so: 3 220B .80...

VNODO'NO

('10) 0213 Buuuee unlpew

an.» 9:28 .8an 28:. so: .353

>033 3 .:o:mE:m:oo 5E .26. 8:me

 

8

3

sum 1m; lenuuu unlpon

121

0.2.5". 2:!

 

 

 

c
on
%
3“
m
3.1..
w
8m
3
8
on

 

5 3.5mm. w
.1255.-
_ ease. 5:223
a
a x
8 m
u
8 B
8—
coE_am 52...
zeta 3

60:83:00 nmm m0 360% 085 no“ mo 0:: we egouﬁdz

 

Buluolluew 95

.m 8:me

122

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

a
. . W
, 1 . 3 m
. 8 m
m.
i 8 B
=. 3:5 a
.1255 I - 8p
>02:
_ m I Ex» :0: no>o 89. «>2»:
r . . F M11“. Q
m. . I _
an M , 4 an
n m
o m l 3
m... m
m u - 8
m. w
a .r .1 I 8
1 8w

 

 

 

 

3:. >38 .33: «ES «>25?

35 Ea 2.2. 25.: «>25?

>038 3 £888: 5:98:89 65 .0 8:me

 

Baluonuovl 91.

123

838... 832
00.08.0202 3.00:0. 3.08.2202 3.00:0".

 

 

5 >023
:85 I
>023 l

 

 

033 3 602 820094 we £0>0~ 880m

 

 

b 8:wE

(qdd10921 801110 $19491 Halpaw

124

>263 3-32 8m was; wage new 602830 .09“ .8.“ 363.53.
333 $-33 v5 >255 333.2 8% maﬁa wanna can 435 £228.60 ems 8m 33:25. *

 

 

 

 

 

 

$78 3.9 2.2 3...
3.
_ u o
* * q . m... M
1 . m
. 3 m
. a
. 3
>25. 3.82
2.75 8.; 3.9a. am.» 973 3.3 a; cwéu
. no. 3.
b: o 0 a
+ 1 a... m . 2. m
* + I . m I . m
. 2 m .2 m
l N 1 u
n.“ - m.“
5:3 3.2.2 >25. 2.42.2

>95; 3 “WES: E 883:8 :mm .3 EwEB 38 mo 0:325
Gamma £32 cog 8302 mg mo =23“;on Son: 39:52 .w Emmi

125

Eta 3.3% Ba £58m wcuao cum £03825 .0? 8m 683.23.
>033 $-32 Ea >033 v5-2.3 .8.“ maﬁa wage 5m .95 £03830 own .5“. 38:914. ...

 

 

 

 

 

 

8.8 2.: 2.... I"...
. no.
u c
+ + + m... M
. m
. 3 m
P a
- m."
>253 3.82
9?: g 3.3 84. 87mm 3.3 mg“ 34%
4 ma. . no.
u u c n o
* 4 m... M * n... .m
+ + . m * . m
+ mé m . rep M
L a . a
nu . 3
5:3 3.2... F 2.2:» «was.

323 E .moﬁﬁﬂ E 383:8 :mm mo Ewﬁ? 180:0 25.8%
HmEmwa 20>“: cog H2002 woﬁ mo 8&8on 53:: 032:2 .o 8:me

126

35.3 $500 3... .230. com. .2092. 05.00.... .95 05.003 623.. 62.00.60 .0w0 00.. 0030:}. *

 

 

 

 

 

 

 

 

 

8.9 3; o 8 w. a. 331 33 up up 0:. o o: 03 pm.
. ov- O . ov- O
. a. . a.
p _r 0 O m . h * 0 a m
* w a w * H . m
L 2 . 2 ..
L on m 1 00 m
g 8 ( .. On (
00.050. 0. 5 >023 8 .. >020» Ea... 09.2.0 00.02.0— :. = >020» 8 _ >02...» Ea... 09.0.5
002» on... 02? 0-20:. 8.2.8 0080 020. 5.800-
w 0*- 3 1 0V. 0
on. L on.
L at W. 1 ON. W
1 3. m _ 1 0.- a
_ + 0 o .- n o I
w w . or m ﬂ — 4 or M
. on m . 8 m
l 8 ( I On (
00.05 c. 5 >023 8.10200 ES. 09.0.5 00.0... c. = >023 2 _ >023 Eat 09.0.5

0083.80 3% mo 3&0? .88 E 0wcmno .0 05.8.5 33%.“
3.0.6. com. 2200?. 5 $030 .3 00303.02 000:: 03232 d. 8%.”.

127

£808 2:988 Banana co comma 3&8 £038.60 .0? he 353:5. *

 

 

 

5-2:: 3.23 3-33
_ _ ﬁ q o m.
m
- S m.
5
W
1 ON :0...
4 cm w.
0
m
1 av m
m
i on w
m
29:2-.- 1 cm W
«_mEl-I - on 0

 

 

@833 m 85 388:8 nmm mo £303 :88 E $5? .: oSwE

128

 

£0608 8388 @0338 no @343 0:5 533 .8me 623260 6%“ 8m @8393 *

 

 

 

 

§.§§§3§1§3§ Egitggsgo
+m~ vuﬁ o; 2.02 +3 in «a; 2.02
o W J . . . a 1
m m
p M w M
a M
m. L «
\\ 1 N M \\ M
n m
m.. . n
1 N m m
1 v
e 1.:

22:0“—

ﬂo>2 82 8302 888 mg was soumﬁsmaoo 5E .S oSwE

APPENDICES

129

 

 

 

 

 

Appendix I:
Reproducibilitj analysis of I972 and 1974 survey
Variable Statistic P-value
Gender Kappa=l .0 <.001
Height Pearson r=0.95 <.001
Weight Pearson r=0.96 <.001
Portion serving size Spearman r=0.73 <.001
One of 3 top fish species:
Trout Kappa=0.6 <.001
Salmon Kappa=0.6 <.001
Catﬁsh Kappa=l .0 <.001
Perch Kappa=0.4 <.001
Whiteﬁsh Kappa=0.9 <.001
Appendix II
Total lipid levels Lng/dlLstratiﬁed by gender and fish eating status, 1979-82
N MeaniSD Median Q25-Q75
Male 615 584.6il4l.l 563.0 4815-6605
Fisheater 401 602311399 590.0 4960-6810
Nonﬁsheater 214 551.0:l:l37.5 525.5 456.5-609.8
Female 502 543.8il33.1 523.0 458.0-612.0
Fisheater 205 555 .51: 1 39.1 534.0 4665-6300
Nonﬁsheater 297 535.7i128.3 5 l 8.5 4550-6000

 

130

Appendix III
Correlations between total lipids levels (mg/dl) and log levels of Aroclor 1260, DDT, DDE, and PBB,
1979-82

 

 

Correlation Coefﬁcient p-value
Aroclor 1260 0.36 <.001
DDT 0.34 <.001
DDE 0.34 <.001
PBB 0.09 .002

 

Appendix IV
Unadjusted regression coefﬁcient of log levels of Aroclor 1260, DDT, DDE, and PBB of predicted against
total lipid levels, 1979-82

 

 

Reggssion coefﬁcient R mare p-value
Aroclor 1260 .002 .13 <.001
DDT .003 .12 <.001
DDE .002 .l l <.001

PBB <.001 .01 .002

 

BIBLIOGRAPHY

10.

11.

12.

13.

14.

BIBLIOGRAPHY
Hicks H. The Great Lakes: a historical overview. Toxicol Ind Health
1996;12:301-13.
IJC. Great Lakes water quality agreement of 1978. 1978.
UC. Great Lakes water quality agreement of 1983. 1983.

Barber MC, Suarez LA, Lassiter RR, Delﬁno JJ. Toxic substances in the Great
Lakes. Govt Reports Announcements & Index 1991;16: 1462-1468.

Anderson H, Falk C, Fiore B, et al. Consortium for the Health Assessment of
Great Lakes Sport Fish Consumption. Toxicol Ind Health 1996;12:369-73.

Hutzinger 0, Safe S, Zitko V. The chemistry of PCBs. CRC Press, Cleveland,
Ohio. 1974.

EPA US. Polychlorinated biphenyls (PCBs): manufacturing, processing,
distribution in commerce and use prohibitions: use in electrical equipment. Fed.
Reg. 1982.

Brinkman VA, DeKok A. Production, properties and usage. New York: Elsevier,
1980.

Lech JJ, Peterson FE. Biotransformation and persistence of polychlorinated
biphenyls (PCBs) in ﬁsh. Boston: Butterworth, 1983.

Evans MS, Noguchi GE, Rice CP. The biomagniﬁcation of polychlorinated
biphenyls, toxaphene, and DDT compounds in a Lake Michigan offshore food
web. Arch Environ Contam Toxicol 1991 ;20:87-93.

Crine JP. Hazards, Decontamination, and replacement of PCB, environmental
science reseach. Plenum Press, New York. 1986;37:1-12.

Goodman JI. Concepts in toxicology and carcinogenesis: A basis for discussing
the potential environmental hazard posed by PCBs. Boston: Butterworth, 1983.

Hoet P, Lauwreys R. Industrial chemical exposure-Z’nd Edition, Lewis Publishing
Inc. 1993.

Sullivan J, Krieger G. Hazardous materials toxicology, Williams and Wilkins
Publishing Corp., Baltimore. 1992:748-751.

131

15.

16.

17.

l8.

19.

20.

21.

22.

23.

24.

25.

26.

27.

132

Buehler F, P. S, Schlatter C. Kinetics of PCB elimination in man. Chemosphere
1988;17:1717.

Safe S. PCBs and human health. Berlin, Heidelberg: Springer-Verlag, 1987.

Kuratsune M. Yusho, with reference to Yu-Cheng, in halogenated biphenyls,
terphenyls, N aphthalenes, Dibenzodioxins and related products. Elsevier,
Amsterdam, 1989.

EPA US. Workshop report on toxicity equivalency factors for polychlorinated
biphenyl congeners, US. Environmental Protection Agency, EPA/625/3-91/020.
1991.

Kimbrough RD. Laboratory and human studies on polychlorinated biphenyls
(PCBs) and related compounds. Environ Health Perspect 1985;59:99-106.

Schaeffer E, Greim H, Goessner W. Pathology of chronic polychlorinated
biphenyl (PCB) feeding in rats. Toxicol Appl Pharmacol 1984;75:278-288.

Norback DH, Weltrnan RH. Polychlorinated biphenyl induction of hepatocellular
carcinoma in the Sprague-Dawley rat. Environ Health Perspect 1985;60:97-105.

Kimbrough RD, Squire RA, Linder RE, Strandberg JD, Montali RJ, Burse VW.
Induction of liver tumors in Sherman Strain rats by polychlorinated biphenyl
Aroclor 1260. J Natl Cancer Inst 1975;55:1453-59.

Kerkvliet NI, Kimeldorf DJ. Antitumor activity of a polychlorinated biphenyl

mixture, Aroclor 1254, in rats inoculated with Walker 256 carcinosarcoma cells. J
Natl Cancer Inst 1977;59:951-955.

Safe S, Bandiera S, Sawyer T, Robertson L, Safe L. PCB: Structure-function
relationships and mechanism of action. Environ Health Perspect 1985;60:47-56.

Smialowicz RJ, Andrews JE, Riddle MM, Rogers RR, Luebke RW, Copeland
CB. Evaluation of the immunotoxicity of low-level PCB exposure in the rat.
Toxicology 1989;56: 197.

Colbom T. Epidemiology of Great Lakes bald eagles. J Toxicol Environ Health
1991;33:395-453.

Gilbertson M, Kubiak T, Ludwig J, Fox G. Great Lakes embryo mortality, edema,
and deformities syndrome (GLEMEDS) in colonial ﬁsh-eating birds: similarity to
chick-edema disease. J Toxicol Environ Health 1991;33:455-520.

28.

29.

30.

31.

32.

33.

34.

35.

36.

37.

38.

39.

133

Thomas PT, Hindsill RD. Effect of polychlorinated biphenyls on the immune
responses of rhesus monkeys and mice. Toxicol Appl Pharmacol 1978;44:41.

Peterson RE, Theobald HM, Kimmel GL. Developmental and reproductive
toxicity of dioxins and related compounds cross species comparisons. Crit Rev
Toxicol 1993;23:283-335.

Aulerich RJ, Ringer RK. Current status of PCB toxicity to mink, and effect on
their reproduction. Arch. Environ Contam Toxicol 1977;6:279-292.

Homshaw TC, Aulerich RJ, Johnson HE. Feeding Great Lakes ﬁsh to mink:
effects on mink and accumulation and elimination of PCBs by mink. J Toxicol
Environ Health l983;1 1:933-946.

Kamrin MA, Carney EW, Chou K, et al. Female reproductives and developmental
toxicology: overview and current approaches. Toxicol lett 1994;74:99-119.

Arnold DL, Nera EA, Stapley R, et al. Prevalence of endometriosis in rhesus
(Macaca mulatta) monkeys ingesting PCB (Aroclor 1254): review and evaluation.
Fundam Appl Toxicol 1996;31:42-55.

de Swart RL, Ross PS, Vos JG, Osterhaus AD. Impaired immunity in harbour
seals (Phoca vitulina) exposed to bioaccumulated environmental contamiants:

review of a long-term feeding study. Environ Health Perspect 1996; 104 (suppl
4):823-828.

Schmid P, Buhler F, Schlatter C. Denna] absorption of PCB in man.
Chemosphere 1992;24: 1283.

Swain WR. An overview of concern with po;ychlorinated biphenyls in the Great
Lakes. Ann Arbor, MI: Ann Arbor Sciences Press, 1983.

Fiore B], Anderson HA, Hanrahan LP, Olson LJ, Sonzogni WC. Sport ﬁsh
consumption and body burden levels of chlorinated hydrocarbons: a study of
Wisconsin anglers. Arch Environ Health 1989;44:82-8.

EPA US. Great Lakes basin risk characterization study. USEPA Great Lakes
National Program Ofﬁce. Washington, DC: United STates Environmental
Protection Agency; 111: 4-6. 1992.

Bro Rasmussen F. Contamination by persistent chemicals in food chain and
human health. Sci Total Environ 1996;188:845-60.

40.

41.

42.

43.

45.

46.

47.

48.

49.

50.

51.

52.

134

Jacobson JL, Fein GG, Jacobson SW, Schwartz PM, Dowler J K. The transfer of
polychlorinated biphenyls (PCBs) and polybrominated biphenyls (PBBs) across
the human placenta and into maternal milk. Am J Public Health 1984;74:378-9.

Fitzgerald EF, Hwang SA, Brix KA, Bush B, Quinn J, Cook K. Chemical
Contaminants in the milk of Mohawk women from Akwesasne. New York State
Department of Health, Albany, NY 1992;43.

Furst P, Furst C, Wilmers K. Human milk as a bioindicator for body burden of
PCDDs, PCDFs, organochlorine pesticides, and PCBs. Environ Health Perspect
1994; l : 187-93.

Humphrey HEB. Population Studies Of Pcbs In Michigan Residents, 1983.

Schwartz PM, Jacobson SW, Fein G, Jacobson JL, Price HA. Lake Michigan ﬁsh
consumption as a source of polychlorinated biphenyls in human cord serum,
maternal serum, and milk. Am J Public Health 1983;73:293-6.

Humphrey HEB. Chemical contaminants in the Great Lakes: The human health
aspect. New York: John Wiley and Sons, 1988.

Kunita N, Kashimoto T, Miyata H, Fukushima S, Hori S, Obana H. Causal agents
of Yusho. New York: Alan R. Liss, INC., 1979.

Hsu ST, Ma CI, Hsu SKI-I, Wu SS, Hsu N, Yeh CC. Discovery and epidemiology
of PCB Poisoning in Taiwan. New York: Alan R. Liss, INC., 1979.

Li WM, Chen CJ, Wong CK. PCB poisoning of 27 cases in three generations of a
large family. Clinical Med 1981;7z23.

Masuda Y. Health status of Japanese and Taiwanese after exposure to
contaminated rice oil. Environ Health Perspect 1985;60:321-5.

Chen PH, Wong CK, Rappe C, Nygren M. Polychlorinated biphenyls,
dibenzofurans and quaterphenyls in toxic rice-bran oil and in the blood and tissues
of patients with PCB poisoning (Yu-Cheng) in Taiwan. Environ Health Perspect
1985;59:59-65.

Hsu ST, Ma CI, Hsu SK, et al. Discovery and epidemiology of PCB poisoning in
Taiwan: a four-year followup. Environ Health Perspect 1985;59:5-10.

Kunita N, Hori S, Obana H, et al. Biological effect of PCBs, PCQs and PCDFs
present in the oil causing yusho and yu-cheng. Environ Health Perspect
1985;59:79-84.

53.

54.

55.

56.

57.

58.

59.

61.

62.

63.

135

Yamashita F, Hayashi M. Fetal PCB syndrome: clinical features, intrauterine
growth retardation and possible alteration in calcium metabolism. Environ Health
Perspect 1985;59:41-5.

Masuda Y, Kuroki H, Haraguchi K, Nagayama J. PCB and PCDF congeners in
the blood and tissues of yusho and yu-cheng patients. Environ Health Perspect
1985;59:53-8.

Huisman M, Eerenstein SE, Koopman Esseboom C, et al. Perinatal exposure to
polychlorinated biphenyls and dioxins through dietary intake. Chemosphere
1995;31:4273-87.

Kashimoto T, Miyata H, Fukushima S, Kunita N, Ohi G, Tung TC. PCBs, PCQs
and PCDFs in blood of yusho and yu-cheng patients. Environ Health Perspect
1985;59:73—8.

Miyata H, Fukushima S, Kashimoto T, Kunita N. PCBs, PCQs and PCDFs in
tissues of yusho and yu-cheng patients. Environ Health Perspect 1985;59:67-72.

Wolff MS, Thornton J, Fischbein A, Lilis R, Selikoff U. Disposition of
polychlorinated biphenyls in occupationally exposed persons. Toxicol. Appl.
Pharmacol. 1982;62:294-306.

Chase KH, Wong 0, D. T, Bemey BW, Simon RK. Clinical and metabolic
abnormalities associated with occupational exposure to polychlorinated biphenyls
(PCBs). J Occup Med 1982;24:109.

Warshaw R, Fischbein A, Thornton J, Miller A, Selikoff U. Decrease in vital
capacity in PCB-exposed workers in a capacitor manufacturing facility. Ann. NY.
Acad. Sci. 1979;320:273.

Safe S. Polychlorinated biphenyls (PCBs) and polybrominated biphenyls (PBBs):
biochemistry, toxicology and mechanism of action. CRC Crit Rev Toxicol
1984;13:319.

Brown DP, Jones M. Mortality and industrial hygiene study of workers exposed to
polychlorinated biphenyls. Arch. Environ. Health 1981;36:120.

Brown DP. Mortality of workers exposed to polychlorinated biphenyls-an update.
Arch. Environ Health 1987;42:333.

Sinks T, Steele G, Smith AB, Watkins K, Shults RA. Mortality among workers
exposed to polychlorinated biphenyls. Am J Epid 1992;136:389.

65.

66.

67.

68.

69.

70.

71.

72.

73.

74.

75.

76.

136

Gustavsson P, Hogstedt C, Rappe C. Shortterm mortality and cancer incidence in
capacitor manufacturing workers exposed to polychlorinated biphenyls. Am J Ind
Med 1986;10:341.

Bertazzi PA, Riboldi L, Pesatori A, Radice L, Zocchetti C. Cancer mortality of
capacitor manufacturing workers. Am J Ind Med 1987;1 l: 165.

Smith AB, Schloemer J, Lowry LK, et al. Metabolic and health consequences of
occupational exposure to polychlorinated biphenyls (PCBs). Brit J Ind Med
1982;38:55.

Fitzgerald EF, Brix KA, Deres DA, et al. Polychlorinated biphenyl (PCB) and
dichlorodiphenyl dichloroethylene (DDE) exposure among Native American men
from contaminated Great Lakes ﬁsh and wildlife. Toxicol Ind Health
1996;12:361-8.

Jacobson JL, Humphrey HE, Jacobson SW, Schantz SL, Mullin MD, Welch R.
Determinants of polychlorinated biphenyls (PCBs), polybrominated biphenyls
(PBBs), and dichlorodiphenyl trichloroethane (DDT) levels in the sera of young
children. Am J Public Health 1989;79: 1401-4.

Jacobson JL, Jacobson SW, Humphrey HE. Effects of in utero exposure to
polychlorinated biphenyls and related contaminants on cognitive functioning in
young children. J Pediatr 1990;116:38-45.

Jacobson JL, Jacobson SW. Evidence for PCBs as neurodevelopmental toxicants
in humans. Neurotoxicology 1997; 18:4 15-24.

Jacobson JL, Jacobson SW. Teratogen update: polychlorinated biphenyls.
Teratology 1997;55:338-47.

Foster WG. The reproductive toxicology of Great Lakes contaminants. Environ
Health Perspect 1995;9263-9.

J ordan-Simpson D, Walsh P, Sherman G. Reproductive outcomes-a background
paper for the state of the Lakes ecosystem conference. Laboratory Center for
Disease Control, Health Canada. 1994.

Buck GM. Epidemiologic perspective of the developmental neurotoxicity of PCBs
in humans. Neurotoxicol Teratol 1996;18:239-41; discussion 271-6.

Jacobson J L, Jacobson SW, Humphrey HE. Effects of exposure to PCBs and
related compounds on growth and activity in children. Neurotoxicol T eratol
1990;12:319-26.

77.

78.

79.

80.

81.

82.

83.

84.

85.

86.

87.

88.

137

Mendola P, Buck GM, Vena JE, Zielezny M, Sever LE. Consumption of PCB-
contaminated sport ﬁsh and risk of spontaneous fetal death. Environ Health
Perspect 1995;103:498-502.

Whorton D, Krauss RM, Marshall S, Milby TH. Infertility in male pesticide
workers. Lancet l977;2: 1259-1261.

Carlsen E, Giwereman A, Keiding N, Skakkebaek NE. Evidence for decreasing
quality of semen during the past 50 years. Brit Med J 1992;305:609-613.

Mendola P, Buck GM, Sever LE, Zielezny M, Vena J E. Consumption of PCB-
contaminated freshwater ﬁsh and shortened menstrual cycle length. Am J
Epidemiol 1997;146:955-60.

Buck GM, Sever LE, Mendola P, Zielezny M, Vena J E. Consumption of
contaminated sport ﬁsh from Lake Ontario and time-to-pregnancy. Am J
Epidemiol 1997;146:949—54.

Jacobson SW, Fein GG, Jacobson JL, Schwartz PM, Dowler J K. The effect of
intrauterine PCB exposure on visual recognition memory. Child Dev
1985;56:853-60.

Schantz SL, Sweeney AM, Gardiner J C, et al. Neuropsychological assessment of
an aging population of Great Lakes ﬁsheaters. Toxicol Ind Health 1996;12:403-
17.

Humphrey HE, Budd ML. Michigan’s ﬁsheater cohorts: a prospective history
of exposure. Toxicol Ind Health 1996;12:3-4.

Hovinga ME, Sowers M, Humphrey HE. Historical changes in serum PCB and
DDT levels in an environmentally-exposed cohort. Arch Environ Contam Toxicol
1992;22:362-6.

Hovinga ME, Sowers M, Humphrey HE. Environmental exposure and lifestyle
predictors of lead, cadmium, PCB, and DDT levels in Great Lakes ﬁsh eaters.
Arch Environ Health 1993;48:98-104.

Connelly NA, Knuth B. Great Lakes ﬁsh consumption health advisories: angler
response to advisories and evaluation of communication techniques. Great Lakes
Protection Fund Final Report. 1993.

Skaare J U, Tuveng J M, H.A. S. Organochlorine pesticides and polychlorinated
biphenyls in maternal adipose tissue, blood, milk, and cord blood from mothers
and their infants living in Norway. Arch Environ Contam Toxicol 1989;17:55-63.

89.

90.

91.

92.

93.

94.

138

Kutz FW, Wood PH, Bottimore DP. Organochlorine pesticides and

polychlorinated biphenyls in human adipose tissue. Reviews Environ Contarn
Toxicol 1991;120:1-85.

Borgmann U, Whittle DM. Contaminant concentration trends in Lake Ontario
Lake trout: 1977 to 1988. J Great Lakes Res 1991;17:368-381.

Miller MA, Madenjian CP, Masnado RG. Patterns of organochlorine
contamination in lake trout from Wisconsin waters of the Great Lakes. J Great
Lakes Res 1992;18:742-754.

Stow CA, Carpenter SR, Eby LA, Amrhein RJ, Hesselberg RJ. Evidence that
PCBs are approaching stable concentrations in lake Michigan ﬁshes. Ecological
Appl. 1995;5z248-258.

De Vault DS, Hesselberg R, Rodgers P, Feist TJ. Contaminant trends in lake trout
and walleye from the Laurentian Great Lakes. J. Great Lakes Res 1996;22:884-
895.

Webb RG, McCall AC, Quantitative PCB standards for electron capture gas
chromatography. J. Chromatographic Science, 11:366-373. 1973

”111111111111111111111“