8Social Capital and Physical HealthA Systematic Review of the Literature
DANIEL KIM, S.V. SUBRAMANIAN, AND ICHIRO KAWACHI
010203040506070809101112131415161718192021222324252627282930313233343536373839404142434445
Book_Kawachi, Subramanian & Kim_0387713107_proof3_180707
139
In this chapter, we describe the key findings from a systematic review of empiricalstudies linking social capital to physical health outcomes. As noted in the Intro-duction, as well as the chapters by van der Gaag and Webber (chapter 2), andLakon and colleagues (chapter 4), much of the public health literature has focusedon the health effects of social cohesion. That is, both ecological and multilevelstudies have sought to examine the health impacts of group cohesion measuredat different scales (e.g., neighborhoods, states, nations). In turn, a number ofindividual-level studies have sought to test the relationships between individualperceptions of social cohesion (e.g., trust of others) and health outcomes.Accordingly, our systematic review of the literature focuses on empirical studies ofsocial cohesion and physical health outcomes. There is a huge body of literaturedescribing the linkages between social integration, social networks, social support,and health (Berkman & Glass, 2000); however, the authors of these studies do nottypically classify their investigations under the heading of “social capital”, andindeed a substantial portion of this literature pre-dates the recent explosion ofinterest in social capital within the public health field.1 Similarly, there have beena number of empirical investigations in the health field using sociometric analysis.These studies have tended to focus on the “dark side” of social capital e.g., thecontagion of high risk behaviors within networks – such as the spread of suicidalideation (Bearman & Moody, 2004), injection drug use (Friedman & Aral, 2001),or alcohol and other drug use among adolescents (Valente, Gallaher, & Mouttapa,2004). The authors of chapter 4 would no doubt argue that these are studies ofsocial capital. However, since they did turn up in our search strategy for “socialcapital and health” (described further below), we shall not discuss them here(except to agree with the authors of chapter 4 that more studies of this type shouldbe encouraged).
1 Outside the public health field, scholars seem happy to mix them up. Thus in his chapteron social capital and health (chapter 20) in the book Bowling Alone (2000), Robert Putnamcites evidence from every type of study, including not only social cohesion, but also socialnetworks and social support.
8.1. Systematic Literature Review
We conducted a systematic literature review of all studies in English that haveexamined social capital in relation to measures of physical health, including all-cause mortality, self-rated health, and major chronic diseases or conditions (e.g.,cardiovascular disease, cancer, obesity, and diabetes), as well as acute infectiousdiseases. Citations were searched using the US National Library of Medicine’sPubMed database (which provides electronic citations from MEDLINE and otherlife science journals for biomedical articles) for the period between 1966 andNovember 1, 2006, corresponding to the keyword combinations of “social capital”with each of the following: “life expectancy”, “mortality”, “cardiovascular dis-ease”, “cancer”, “diabetes”, “obesity”, and “infectious diseases”. Articles werethen obtained and reviewed. Reference sections of retrieved articles were searchedto identify additional potential articles for inclusion. Tables 8.1 through 8.6 displaythe key characteristics and findings from these studies, stratified by the type ofstudy design (ecological, multilevel, individual-level) and the highest spatial levelof social capital (country, state/region, neighborhood/community), and are listedchronologically by year of publication within each grouping. From each study,we abstracted the study authors and year of publication, sample size and popula-tion/setting, age range for social capital and health outcome measures, type ofstudy design (cross-sectional versus prospective/longitudinal), measures of socialcapital and health/disease, factors included as covariates in statistical models (orstratified on), and individual-level and area-level effect estimates for social capital.For studies that only analyzed individual-level measures of social capital, our key-word search excluded a much more established body of literature that has focusedon social networks and social support (which we would argue conceptually belongto social capital). Nevertheless, our review identifies studies that have used indica-tors of social cohesion such as individual perceptions of trust and reciprocity, aswell as reports of civic engagement and social participation. For the outcome ofself-rated health, to facilitate comparison and discussion of the findings acrossstudies in which the outcome was dichotomous (fair/poor health versusexcellent/very good/good health), all odds ratios and 95% confidence intervalspresented in Table 8.2 for social trust and associational memberships correspondto associations between higher social capital and the relative odds of fair/poorself-rated health. These estimates were then plotted on the same graph for the sameindicators at each of the individual and contextual levels.
8.2. Social Capital, All-Cause Mortality, and Life Expectancy
Table 8.1 provides details of the 15 studies of social capital and life expectancyor all-cause mortality that met our inclusion criteria. Of these, only three stud-ies conducted multilevel analyses (two of which were prospective; Blakely etal., 2006; Mohan, Twigg, Barnard, & Jones, 2005), while the remaining studieswere ecological (only one of which was prospective; Milyo & Mellor, 2003).
010203040506070809101112131415161718192021222324252627282930313233343536373839404142434445
140 Kim et al.
Book_Kawachi, Subramanian & Kim_0387713107_proof3_180707
010203040506070809101112131415161718192021222324252627282930313233343536373839404142434445
Book_Kawachi, Subramanian & Kim_0387713107_proof3_180707
EC
OL
OG
ICA
L
STU
DIE
S:C
ount
ry le
vel
Lync
h et
al.,
200
116
cou
ntri
esSo
cial
cap
ital m
eas-
ures
: 18�
yH
ealth
out
com
em
easu
res:
All
ages
Soci
al m
istr
ust,
orga
niza
tiona
l m
embe
rshi
ps,
trad
e un
ion
mem
bers
hips
, vo
lunt
eeri
ng
Lif
e ex
pect
ancy
, all-
caus
e m
orta
lity
rate
s
GD
P pe
r ca
pita
;st
ratif
ied
by
gend
er
–1)
Soc
ial m
istr
ust:
Lif
e ex
pect
ancy
r �
�0.
14, p
�0.
65 (
men
)r
�0.
45, p
�0.
12 (
wom
en)
All-
caus
e m
orta
lity
rate
sr
��
0.06
, p �
0.84
(m
en)
r �
�0.
33, p
�0.
27 (
wom
en)
2) O
rgan
izat
iona
l mem
bers
hips
:L
ife
expe
ctan
cyr
��
0.07
, p �
0.82
(m
en)
r �
�0.
33, p
�0.
29 (
wom
en)
All-
caus
e m
orta
lity
rate
sr
�0.
17, p
�0.
59 (
men
)r
�0.
20, p
�0.
53 (
wom
en)
3) T
rade
uni
on m
embe
rshi
ps:
Lif
e ex
pect
ancy
r �
0.13
, p �
0.68
(m
en)
r �
�0.
31, p
�0.
30 (
wom
en)
All-
caus
e m
orta
lity
rate
sr
�0.
25, p
�0.
42 (
men
)r
�0.
36, p
�0.
23 (
wom
en)
4) V
olun
teer
ing:
Lif
e ex
pect
ancy
r �
0.28
, p �
0.40
(m
en)
r �
0.41
, p �
0.20
(w
omen
)
TAB
LE
8.1.
Soci
al c
apita
l, lif
e ex
pect
ancy
, and
all-
caus
e m
orta
lity.
Sam
ple
size
,A
utho
rs,
popu
latio
n/So
cial
cap
ital
Hea
lth o
utco
me
Indi
vidu
al-l
evel
Are
a-le
vel
year
setti
ngA
ge r
ange
mea
sure
mea
sure
Cov
aria
tes
effe
ct e
stim
ate
effe
ct e
stim
ate
(Con
tinu
ed)
010203040506070809101112131415161718192021222324252627282930313233343536373839404142434445
Book_Kawachi, Subramanian & Kim_0387713107_proof3_180707
Ken
nelly
et a
l.,20
0319
OE
CD
cou
ntri
esSo
cial
cap
ital m
eas-
ures
: 18�
yH
ealth
out
com
em
easu
res:
All
ages
Soci
al tr
ust,
asso
ciat
iona
lm
embe
rshi
ps,
volu
ntee
ring
Gen
der-
spec
ific
life
expe
ctan
cy, i
nfan
tm
orta
lity
rate
s,pe
rina
tal m
orta
l-ity
rat
es
GD
P pe
r ca
pita
,G
ini c
oeff
icie
nt,
phys
icia
ns p
erca
pita
, pro
port
ion
of p
ublic
exp
en-
ditu
re in
tota
lhe
alth
exp
endi
-tu
re, f
ruit
and
vege
tabl
e co
n-su
mpt
ion
per
capi
ta, t
obac
coco
nsum
ptio
n pe
rca
pita
, alc
ohol
cons
umpt
ion
per
capi
ta, c
ount
ry o
fJa
pan;
ana
lyse
sst
ratif
ied
by g
en-
der
and
acco
unt
for
surv
ey w
ave
–
All-
caus
e m
orta
lity
rate
sr �
�0.
53, p
�0.
09 (m
en)
r ��
0.59
, p �
0.06
(wom
en)
1) S
ocia
l tru
st:
Lif
e ex
pect
ancy
ß >
0, p
�0.
47 (
men
)ß
> 0
, p �
0.25
(w
omen
)In
fant
mor
talit
y ra
tes
ß <
0, p
�0.
31Pe
rina
tal m
orta
lity
rate
sß
< 0
, p �
0.14
2) A
ssoc
iatio
nal m
embe
rshi
ps:
Lif
e ex
pect
ancy
ß >
0, p
�0.
14 (
men
)ß
> 0
, p �
0.32
(w
omen
) In
fant
mor
talit
y ra
tes
ß <
0, p
�0.
65Pe
rina
tal m
orta
lity
rate
sß
> 0
, p �
0.22
3) V
olun
teer
ing:
Lif
e ex
pect
ancy
ß >
0, p
�0.
76 (
men
)ß
> 0
, p �
0.48
(w
omen
)In
fant
mor
talit
y ra
tes
ß >
0, p
�0.
46Pe
rina
tal m
orta
lity
rate
sß
> 0
, p �
0.15
TAB
LE
8.1.
(Con
tinu
ed).
Sam
ple
size
,A
utho
rs,
popu
latio
n/So
cial
cap
ital
Hea
lth o
utco
me
Indi
vidu
al-l
evel
Are
a-le
vel
year
setti
ngA
ge r
ange
mea
sure
mea
sure
Cov
aria
tes
effe
ct e
stim
ate
effe
ct e
stim
ate
010203040506070809101112131415161718192021222324252627282930313233343536373839404142434445
Book_Kawachi, Subramanian & Kim_0387713107_proof3_180707
Stat
e or
reg
iona
lle
vel
Kaw
achi
et a
l.,19
97
Wilk
inso
n et
al.,
1998
Siah
push
&Si
ngh,
199
9
Mily
o &
Mel
lor,
2003
Vee
nstr
a, 2
002
Ken
nedy
et a
l., 1
998
39 U
S st
ates
39 U
S st
ates
7 st
ates
/te
rrito
ries
in
Aus
tral
ia in
eac
hof
sev
en y
ears
(n
�49
)2
sam
ples
: 48
US
stat
es; 3
9 U
Sst
ates
29 h
ealth
dis
tric
ts in
the
prov
ince
of
Sask
atch
ewan
,C
anad
a
40 r
egio
ns in
Rus
sia
Soci
al c
apita
l mea
s-ur
es: 1
8�y
Hea
lth o
utco
me
mea
sure
: all
ages
Soci
al c
apita
l m
easu
res:
18�
yH
ealth
out
com
em
easu
re: a
ll ag
es
Soci
al c
apita
l m
easu
res:
15�
yH
ealth
out
com
em
easu
re: a
ll ag
es
Soci
al c
apita
l m
easu
res:
18�
yH
ealth
out
com
em
easu
res:
All
ages
Soci
al c
apita
l m
easu
re: 1
8�y
Hea
lth o
utco
me
mea
sure
s: A
llag
es
Soci
al c
apita
l m
easu
re: 1
6�y
Hea
lth o
utco
me
mea
sure
s: A
llag
es
Soci
al m
istr
ust,
lack
of h
elpf
ulne
ss,
volu
ntar
y gr
oup
mem
bers
hips
Soci
al m
istr
ust
Perc
enta
ge o
f la
bor
forc
e w
ith u
nion
mem
bers
hips
Putn
am s
ocia
l cap
i-ta
l ind
ex (
deri
ved
from
14
indi
ca-
tors
), s
ocia
l m
istr
ust
Soci
al c
apita
l ind
ex(a
ssoc
iatio
nal
mem
bers
hips
,so
cial
invo
lve-
men
t, el
ecto
ral
part
icip
atio
n)
Mis
trus
t in
loca
l and
in r
egio
nal g
ov-
ernm
ent,
lack
of
soci
al c
ohes
ion
at
Age
-sta
ndar
dize
dal
l-ca
use
mor
talit
y ra
tes
Age
-sta
ndar
dize
dal
l-ca
use
mor
talit
y ra
tes
Age
-sta
ndar
dize
dal
l-ca
use
mor
talit
y ra
tes
Age
-sta
ndar
dize
dal
l-ca
use
mor
talit
y ra
tes
Age
-sta
ndar
dize
dal
l-ca
use
mor
talit
y ra
tes
Lif
e ex
pect
ancy
,ag
e-st
anda
rdiz
ed
all-
caus
e m
orta
lity
rate
s
Stat
e pr
eval
ence
of
pove
rty –
Cal
enda
r ye
ar
Prop
ortio
n of
pop
u-la
tion
in p
over
ty
Inco
me
ineq
ualit
y,ge
nder
com
posi
-tio
n, to
tal c
rim
e
Per
capi
ta in
com
e,pr
opor
tion
inpo
vert
y, p
er-
ceiv
ed e
cono
mic
hard
ship
in
– – – – – –
1) S
ocia
l mis
trus
t:ß
> 0
, p <
0.0
12)
Lac
k of
hel
pful
ness
:ß
> 0
, p <
0.0
13)
Vol
unta
ry g
roup
mem
bers
hips
:ß
< 0
, p <
0.0
1
r �
0.76
, p <
0.0
5
ß >
0, p
< 0
.05
1) P
utna
m s
ocia
l cap
ital i
ndex
:ß
< 0
, p <
0.0
12)
Soc
ial m
istr
ust:
ß >
0, p
< 0
.01
ß >
0, p
�0.
81
1) M
istr
ust i
n lo
cal g
over
nmen
t:
Lif
e ex
pect
ancy
ß <
0, p
�0.
02 (
men
)ß
< 0
, p �
0.05
3 (w
omen
)A
ll-ca
use
mor
talit
y ra
tes
(Con
tinu
ed)
010203040506070809101112131415161718192021222324252627282930313233343536373839404142434445
Book_Kawachi, Subramanian & Kim_0387713107_proof3_180707
Skra
bski
et a
l.20
0320
cou
ntie
s in
H
unga
rySo
cial
cap
ital m
eas-
ure:
16�
y H
ealth
out
com
em
easu
res:
45–
64 y
wor
k, la
ck o
fin
tere
st in
pol
itics
Soci
al m
istr
ust,
rec-
ipro
city
, rec
eive
dhe
lp f
rom
civ
icor
gani
zatio
ns
Age
-spe
cifi
c (a
ges
45–6
4)
and
gend
er-
spec
ific
al
l-ca
use
mor
talit
y ra
tes
regi
on, p
er c
apita
crim
e ra
te; a
naly
-se
s st
ratif
ied
byge
nder
GD
P pe
r ca
pita
,in
com
e, e
duca
-tio
n, p
reva
lenc
e of
smok
ing,
ave
rage
alco
hol c
onsu
mp-
tion,
une
mpl
oy-
men
t rat
e;
–
ß >
0, p
�0.
01 (
men
)ß
> 0
, p �
0.06
(w
omen
)2)
Mis
trus
t in
regi
onal
gove
rnm
ent:
Lif
e ex
pect
ancy
ß <
0, p
�0.
15 (
men
)ß
< 0
, p �
0.24
(w
omen
)A
ll-ca
use
mor
talit
y ra
tes
ß >
0, p
�0.
0497
(m
en)
3) L
ack
of s
ocia
l coh
esio
n at
wor
k:L
ife
expe
ctan
cyß
< 0
, p �
0.01
(m
en)
ß <
0, p
�0.
04 (
wom
en)
All-
caus
e m
orta
lity
rate
sß
> 0
, p �
0.02
(m
en)
4) L
ack
of in
tere
st in
pol
itics
:L
ife
expe
ctan
cyß
< 0
, p �
0.02
(m
en)
ß <
0, p
�0.
06 (
wom
en)
All-
caus
e m
orta
lity
rate
sß
> 0
, p �
0.10
(m
en)
ß >
0, p
< 0
.10
(wom
en)
1) S
ocia
l mis
trus
t:ß
> 0
, p <
0.0
1 (m
en)
ß >
0, p
< 0
.01
(wom
en)
2) R
ecip
roci
ty:
ß <
0, p
< 0
.01
(men
)ß
< 0
, p <
0.0
1 (w
omen
)
TAB
LE
8.1.
(Con
tinu
ed).
Sam
ple
size
,A
utho
rs,
popu
latio
n/So
cial
cap
ital
Hea
lth o
utco
me
Indi
vidu
al-l
evel
Are
a-le
vel
year
setti
ngA
ge r
ange
mea
sure
mea
sure
Cov
aria
tes
effe
ct e
stim
ate
effe
ct e
stim
ate
010203040506070809101112131415161718192021222324252627282930313233343536373839404142434445
Book_Kawachi, Subramanian & Kim_0387713107_proof3_180707
(Con
tinu
ed)
Skra
bski
et a
l., 2
004
Tur
rell
et a
l., 2
006
Nei
ghbo
rhoo
d le
vel
Loc
hner
et a
l., 2
003
150
subr
egio
ns in
Hun
gary
Pers
ons
aged
25–
74ye
ars
in 4
1 st
atis
-tic
al lo
cal a
reas
inth
e st
ate
of T
as-
man
ia, A
ustr
alia
342
neig
hbor
hood
sin
Chi
cago
in th
eU
S
Soci
al c
apita
l mea
s-ur
es: 1
8�y
Hea
lth o
utco
me
mea
sure
s: 4
5–64
y
Soci
al c
apita
l mea
s-ur
es: 1
8�y
Hea
lth o
utco
me
mea
sure
s: 2
5–74
y
Soci
al c
apita
l mea
s-ur
es: 1
8�y
Hea
lth o
utco
me
mea
sure
s: 4
5–64
y
Soci
al m
istr
ust,
rec-
ipro
city
, mem
ber-
ship
in c
ivic
orga
niza
t-io
ns,
relig
ious
gro
upin
volv
emen
t
Soci
al tr
ust,
soci
alco
hesi
on, p
oliti
-ca
l par
ticip
atio
n
Tru
st, r
ecip
roci
ty,
asso
ciat
iona
lm
embe
rshi
ps
Age
-spe
cifi
c (a
ges
45–6
4) a
nd g
ende
r-sp
ecif
ic a
ll-ca
use
mor
talit
y ra
tes
All-
caus
e ag
e-st
anda
rdiz
ed
mor
talit
y ra
tes
All-
caus
e m
orta
lity
rate
s (a
ges
45–6
4)
anal
yses
str
atif
ied
by g
ende
r
Inco
me
per
capi
ta,
mea
n ye
ars
ofed
ucat
ion,
pre
va-
lenc
e of
sm
okin
g,av
erag
e al
coho
lco
nsum
ptio
n,co
llect
ive
effi
cacy
;an
alys
es s
trat
ifie
dby
gen
der
Are
a: age,
gen
der,
soci
oeco
nom
icdi
sadv
anta
ge,
geog
raph
icre
mot
enes
s,ne
ighb
orho
odsa
fety
Soci
oeco
nom
ic d
ep-
riva
tion;
ana
lyse
sst
ratif
ied
by
– – –
3) R
ecei
ved
help
from
civ
icor
gani
zatio
ns:
ß >
0, p
< 0
.01
(men
)
1) S
ocia
l mis
trus
t:ß
> 0
, p <
0.0
1 (m
en)
ß >
0, p
< 0
.01
(wom
en)
2) R
ecip
roci
ty:
ß <
0, p
< 0
.01
(men
)ß
< 0
, p <
0.0
1 (w
omen
)3)
Mem
bers
hip
in c
ivic
orga
niza
tion
s:ß
< 0
, p <
0.0
1 (m
en)
ß <
0, p
< 0
.01
(wom
en)
4) R
elig
ious
gro
up in
volv
emen
t:ß
> 0
, p <
0.0
1 (m
en)
ß <
0, p
< 0
.01
(wom
en)
1) S
ocia
l tru
st:
ß >
0, p
> 0
.05
2) P
oliti
cal p
artic
ipat
ion:
ß >
0, p
> 0
.05
3) T
rust
in p
ublic
and
pri
vate
inst
itutio
ns:
ß >
0, p
> 0
.05
4) N
eigh
borh
ood
inte
grat
ion:
ß <
0, p
> 0
.05
5) N
eigh
borh
ood
isol
atio
n:ß
> 0
, p >
0.0
5
1) T
rust
:W
hite
wom
enß
< 0
, p <
0.0
1W
hite
men
010203040506070809101112131415161718192021222324252627282930313233343536373839404142434445
Book_Kawachi, Subramanian & Kim_0387713107_proof3_180707
MU
LTIL
EV
EL
STU
DIE
S:N
eigh
borh
ood-
or
regi
onal
-lev
el s
ocia
lca
pita
l
race
/eth
nici
ty a
ndge
nder
ß <
0, p
< 0
.01
Bla
ck w
omen
ß <
0, p
< 0
.05
Bla
ck m
enß
< 0
, p >
0.0
52)
Rec
ipro
city
: W
hite
wom
enß
< 0
, p <
0.0
1W
hite
men
ß <
0, p
< 0
.05
Bla
ck w
omen
ß <
0, p
> 0
.05
Bla
ck m
enß
< 0
, p <
0.0
53)
Ass
ocia
tiona
l mem
bers
hips
:W
hite
wom
enß
< 0
, p <
0.0
1W
hite
men
ß <
0, p
< 0
.01
Bla
ck w
omen
ß <
0, p
> 0
.05
Bla
ck m
enß
< 0
, p <
0.0
1
TAB
LE
8.1.
(Con
tinu
ed).
Sam
ple
size
,A
utho
rs,
popu
latio
n/So
cial
cap
ital
Hea
lth o
utco
me
Indi
vidu
al-l
evel
Are
a-le
vel
year
setti
ngA
ge r
ange
mea
sure
mea
sure
Cov
aria
tes
effe
ct e
stim
ate
effe
ct e
stim
ate
010203040506070809101112131415161718192021222324252627282930313233343536373839404142434445
Book_Kawachi, Subramanian & Kim_0387713107_proof3_180707
Wen
et a
l., 2
005
Moh
an e
t al.,
200
5
12,6
72 a
dults
dia
g-no
sed
and
hosp
i-ta
lized
with
one
of 1
3 se
riou
s ill
-ne
sses
in 5
1 zi
pco
des
in C
hica
go
7,57
8 ad
ults
in 3
96el
ecto
ral w
ards
inE
ngla
nd
Soci
al c
apita
l mea
s-ur
es: 1
8�y
Hea
lth o
utco
me
mea
sure
s:67
�ye
ars
Soci
al c
apita
l and
heal
th o
utco
me
mea
sure
s: 1
8–94
y
Soci
al s
uppo
rt,
soci
al n
etw
ork
dens
ity, p
artic
ipa-
tion
in lo
cal
orga
niza
tions
,vo
lunt
ary
asso
ciat
ions
Indi
vidu
al le
vel:
Bel
ongi
ng to
com
-m
unity
, rel
iabl
efr
iend
s, fr
eque
ncy
of fe
elin
g lo
nely
War
d le
vel:
Vol
unte
erin
g, p
ar-
ticip
atio
n in
soci
al a
ctiv
ities
,al
trui
stic
act
ivi-
ties,
pol
itica
lac
tiviti
es, e
lec-
tora
l par
ticip
at-
ion,
impo
rtan
ceof
loca
l fri
ends
,at
titud
es to
war
dsbe
long
ing
tone
ighb
orho
od,
will
ingn
ess
tow
ork
to im
prov
ene
ighb
orho
od,
talk
ing
to n
eigh
-bo
rs, f
requ
ency
of m
eetin
g lo
cal
peop
le, p
erce
ived
All-
caus
e m
orta
lity
(dic
hoto
mou
s)
All-
caus
e m
orta
lity
Indi
vidu
al le
vel:
age,
gen
der,
race
/et
hnic
ity, M
edic
aid
reci
pien
t, co
-mor
bidi
tyZ
ip c
ode
leve
l:so
cioe
cono
mic
stat
us
Age
, gen
der,
soci
alcl
ass,
hou
seho
ldte
nure
, sm
okin
g,al
coho
l con
sum
p-tio
n, e
xerc
ise,
die
t
–
1) P
erce
ived
belo
ngin
g to
com
mun
ity:
OR
�1.
11, 9
5%
CI
�0.
93�
1.32
2) R
elia
ble
frie
nds:
OR
�1.
05, 9
5%
CI
�0.
63�
1.78
3) F
requ
ency
of
feel
ing
lone
ly:
OR
�1.
30, 9
5%
CI
�0.
98�
1.72
1) S
ocia
l sup
port
:H
R �
0.99
6, p
> 0
.05
2) S
ocia
l net
wor
k de
nsity
:H
R �
1.02
, p >
0.0
53)
Loc
al o
rgan
izat
ions
:H
R �
0.99
4, p
> 0
.05
4) V
olun
tary
ass
ocia
tions
:H
R �
1.00
5, p
> 0
.05
Low
est l
evel
s of
:1)
Any
vol
unte
erin
g: O
R �
1.35
,95
% C
I �
1.06
�1.
712)
Vol
unte
erin
g (1
1+ d
ays
over
past
yea
r): O
R �
1.31
, 95%
C
I �
1.03
�1.
673)
Par
ticip
atio
n in
soc
ial
orga
niza
tions
: OR
�1.
36, 9
5%
CI
�1.
07�
1.73
4) P
artic
ipat
ion
in a
ltrui
stic
or
gani
zatio
ns:O
R �
1.27
, 95%
CI
�1.
00�
1.57
5) P
oliti
cal a
ctiv
ities
: OR
�1.
27,
95%
CI
�1.
01�
1.60
6) E
lect
oral
par
ticip
atio
n:
OR
�1.
03, 9
5% C
I �0.
81�
1.29
7) I
mpo
rtan
ce o
f lo
cal f
rien
ds:
OR
�1.
20, 9
5%
CI
�0.
96�
1.51
8) A
ttitu
des
tow
ards
bel
ongi
ng to
neig
hbor
hood
: OR
�0.
93, 9
5%C
I �
0.73
�1.
189)
Will
ingn
ess
to w
ork
to im
prov
ene
ighb
orho
od: O
R �
1.09
, 95%
CI
�0.
86�
1.38
10)
Talk
ing
to n
eigh
bors
: OR
�1.
04, 9
5% C
I �
0.83
�1.
30
(Con
tinu
ed)
010203040506070809101112131415161718192021222324252627282930313233343536373839404142434445
Book_Kawachi, Subramanian & Kim_0387713107_proof3_180707
Bla
kely
et a
l., 2
006
All
25–7
4 ye
ar-o
lds
in 1
,683
Cen
sus
area
uni
ts in
New
Zea
land
Soci
al c
apita
l mea
s-ur
e: 1
5�y
Hea
lth o
utco
me
mea
sure
s: 2
5–74
y
frie
ndlin
ess
ofar
ea, b
lood
dona
tion
Cen
sus
area
un
it le
vel:
volu
ntee
ring
All-
caus
e m
orta
lity,
stra
tifie
d by
ge
nder
Indi
vidu
al le
vel:
age,
rac
e/ e
thni
c-ity
, mar
ital s
tatu
s,in
com
e, e
duca
-tio
n, c
ar a
cces
s,em
ploy
men
t sta
-tu
s, u
rban
re
side
nce
Nei
ghbo
rhoo
d le
vel:
soci
oeco
nom
icde
priv
atio
n
–
11)
Freq
uenc
y of
mee
ting
loca
lpe
ople
: OR
�0.
80, 9
5%
CI
�0.
63�
1.02
12)
Perc
eive
d fr
iend
lines
s of
are
a: O
R �
0.84
, 95%
C
I �
0.67
�1.
0613
) B
lood
don
atio
n: O
R �
1.05
,95
% C
I �
0.83
�1.
32
Low
vol
unte
eris
m:
RR
�0.
95,
95%
CI
�0.
89�
1.02
(m
en)
RR
�0.
96, 9
5%
CI
�0.
88�
1.04
(w
omen
)
TAB
LE
8.1.
(Con
tinu
ed).
Sam
ple
size
,A
utho
rs,
popu
latio
n/So
cial
cap
ital
Hea
lth o
utco
me
Indi
vidu
al-l
evel
Are
a-le
vel
year
setti
ngA
ge r
ange
mea
sure
mea
sure
Cov
aria
tes
effe
ct e
stim
ate
effe
ct e
stim
ate
010203040506070809101112131415161718192021222324252627282930313233343536373839404142434445
8. Social Capital and Physical Health 149
Book_Kawachi, Subramanian & Kim_0387713107_proof3_180707
Among ecological studies, the unit of analysis for social cohesion variedwidely, from the country level down to the neighborhood level, whereas multi-level studies assessed social capital at the regional or neighborhood, but notcountry levels. In the country-level ecological studies, nations that wereincluded consisted primarily of OECD nations, and excluded developingnations. Within-country ecological studies analyzed population samples in theUS, Canada, Australia, as well as Russia and Hungary, while the multilevelanalyses employed samples in the US, England, and New Zealand.
The vast majority of studies focused on a single indicator of social capital, suchas social trust, associational memberships, and reciprocity, and were derived byaggregating survey responses among adults to the area level, while one study(Milyo & Mellor, 2003) applied the Putnam social capital index (based on 14 state-level social capital indicators), and another study (Siahpush & Singh, 1999) inves-tigated the association for the percentage of the labor force with unionmemberships. Most ecological studies examined all-cause mortality rates as thehealth outcome across all age groups, including children and adolescents (appro-priately summarized through age-standardization), but without stratification byage. A small subset of studies confined the examination of mortality to those ofmiddle age (45–64 years) (Lochner, Kawachi, Brennan, & Buka, 2003; Skrabski,Kopp, & Kawachi, 2003, 2004). Two of the three multilevel analyses analyzed therisk of all-cause mortality among adults in most age groups, while the other analy-sis (Wen, Cagney, & Christakis, 2005) was restricted to an elderly population(67� years), and estimated the relative hazards of dying among those diagnosedand hospitalized with serious illnesses.
Adjustment for potential confounders in ecological studies was variable, withsome studies limiting control to gender and area-level deprivation (e.g., Lynchet al., 2001), and other studies controlling for ecological factors expectedly corre-lated with health behaviors, that could plausibly mediate the effects of social capi-tal (Kennelly, O’Shea, & Garvey, 2003; Skrabski et al., 2003, 2004). In multilevelstudies, suitable control was made for several individual-level factors includingdemographic characteristics (e.g., age, gender, and race/ethnicity) and socioeco-nomic status (e.g., income or education), through adjustment in statistical modelsor stratification. Nonetheless, control at the area level was confined to area-levelsocioeconomic deprivation (Blakely et al., 2006; Wen et al., 2005), or was absentaltogether (Mohan et al., 2005), so that residual confounding bias due to effects ofother area-level factors such as racial/ethnic heterogeneity cannot be excluded.
Social cohesion was fairly consistently associated in a protective direction withmortality outcomes at the state, regional, and/or neighborhood levels in the US,Russia, and Hungary, whereas the relationships were statistically non-significantin other countries including Canada, Australia, and New Zealand as well as incross-national studies. Among the three multilevel studies, findings were moremixed, with only one study (Mohan et al., 2005) observing significant associa-tions for selected social capital measures (volunteering, organizational participa-tion, and non-electoral political participation, but not informal socializingdomains) after adjustment for individual-level social capital indicators.
010203040506070809101112131415161718192021222324252627282930313233343536373839404142434445
150 Kim et al.
Book_Kawachi, Subramanian & Kim_0387713107_proof3_180707
8.3. Social Capital and Self-Rated Health
Altogether 32 studies met our inclusion criteria for social capital and self-ratedhealth (Table 8.2). Only one of these studies was ecological, while 24 were multi-level (with higher-level units ranging from the country level to the state andneighborhood or community level), and seven were conducted at the individuallevel. Only two studies (both multilevel; Mellor & Milyo, 2005; Zimmerman &Bell, 2006) were prospective, while all other studies were cross-sectional.
As with studies involving mortality, studies of self-rated health have predomi-nantly analyzed single indicators of social cohesion such as trust, associationalmembership, and reciprocity. Studies that incorporated a large number of indica-tors combined indicators either through factor analysis or by taking the mean ofstandardized values for multiple indicators, with one such study measuring bothcommunity- and individual-level bonding and bridging social capital (Kim, Sub-ramanian, & Kawachi, 2006a). In nine of the 25 multilevel studies, individual andcollective social capital were simultaneously examined, with individual-levelsocial capital being measured via the same survey items (without aggregation) asat the area level.
Most studies dichotomized the outcome of self-rated health into fair/poor versusexcellent/very good/good health, though some studies analyzed the outcome as acontinuous or ordinal variable.
The sole ecological study (Lynch et al., 2001) was conducted with countries asthe unit of analysis, and adjusted for gross domestic product (GDP) per capita.The majority of multilevel studies adjusted for key individual-level covariatesincluding age, gender, race/ethnicity, and income or education. Meanwhile,adjustment for potential confounders at the area level ranged widely, with somestudies making no adjustment at all, and other studies controlling for multiplepotential confounders (see for e.g., Browning & Cagney, 2003).
In multilevel studies, measures of social capital at the individual level were forthe most part significantly associated with better self-rated health. By contrast,the association between area social cohesion and self-rated health was moremixed, especially after adjustment for individual-level covariates (Table 8.2).These contrasts between the individual and area level are apparent in Figures 8.1through 8.4, which plot the odds ratios and 95% confidence intervals for theassociations between higher social trust and associational memberships andfair/poor self-rated health (Figures 8.1 and 8.3 at the individual level, and Figures8.2A and 8.4A at the area level after adjustment for individual-level social capital,respectively).
There was also evidence of attenuation of the odds ratios with the addition ofindividual-level social capital indicators, in some instances to statistical non-significance: Figures 8.2B and 8.4B show the odds ratio estimates for area-levelsocial trust and associational memberships in the multilevel analyses withoutadjustment for individual-level social capital. All of these studies were cross-sectional in design. Here, a general pattern emerges of stronger inverse and statis-tically significant odds ratios prior to multivariate adjustment, compared to the
010203040506070809101112131415161718192021222324252627282930313233343536373839404142434445
Book_Kawachi, Subramanian & Kim_0387713107_proof3_180707
EC
OL
OG
ICA
L
STU
DIE
S:C
ount
ry-l
evel
so
cial
cap
ital
Lync
h et
al.,
200
1
MU
LTIL
EV
EL
STU
DIE
S:C
ount
ry-l
evel
so
cial
cap
ital
Hel
liwel
l &
Putn
am, 2
004
16 c
ount
ries
83,5
20 a
dults
in 4
9co
untr
ies
Soci
al c
apita
l mea
s-ur
e &
hea
lth
outc
ome
mea
sure
: 18+
y
Soci
al c
apita
l m
easu
re &
heal
th o
utco
me
mea
sure
: 18�
y
Soci
al m
istr
ust,
orga
niza
tiona
lm
embe
rshi
ps,
trad
e un
ion
mem
bers
hips
,vo
lunt
eeri
ng
Indi
vidu
al le
vel:
soci
al tr
ust (
gen-
eral
, in
polic
e)as
soci
atio
nal
mem
bers
hips
, N
atio
nal l
evel
:so
cial
trus
t,as
soci
atio
nal
mem
bers
hips
Prop
ortio
n re
port
-in
g fa
ir/
poor
hea
lth
Con
tinuo
us (
high
er�
bette
r he
alth
)
GD
P pe
r ca
pita
Indi
vidu
al le
vel:
age,
gen
der,
mar
ital s
tatu
s,em
ploy
men
t sta
-tu
s, im
port
ance
of
God
/rel
igio
n, f
re-
quen
cy o
f at
tend
-in
g re
ligio
usse
rvic
eN
atio
nal l
evel
:m
edia
n in
com
e,im
port
ance
of
God
/rel
igio
n,
–
1) G
ener
al s
ocia
ltr
ust:
ß >
0, p
< 0
.01
2) T
rust
in p
olic
e:ß
> 0
, p <
0.0
13)
Ass
ocia
tiona
lm
embe
rshi
ps:
ß >
0, p
< 0
.05
1) S
ocia
l mis
trus
t:r
�0.
47, p
�0.
112)
Org
aniz
atio
nal
mem
bers
hips
:r
��
0.36
, p �
0.25
3)
Tra
de u
nion
mem
bers
hips
:r
��
0.17
, p �
0.58
4) V
olun
teer
ing:
r �
�0.
80, p
�0.
003
1) G
ener
al s
ocia
l tru
st:
ß >
0, p
< 0
.01
2) A
ssoc
iatio
nal m
embe
rshi
ps:
ß >
0, p
< 0
.01
TAB
LE
8.2.
Soci
al c
apita
l and
sel
f-ra
ted
phys
ical
and
gen
eral
hea
lth.
Sam
ple
size
,A
utho
rs,
popu
latio
n/So
cial
cap
ital
Form
of
self
-rat
ed
Indi
vidu
al-l
evel
Are
a-le
vel
year
setti
ngA
ge r
ange
mea
sure
heal
th m
easu
reC
ovar
iate
sef
fect
est
imat
eef
fect
est
imat
e
(Con
tinu
ed)
010203040506070809101112131415161718192021222324252627282930313233343536373839404142434445
Book_Kawachi, Subramanian & Kim_0387713107_proof3_180707
Poor
tinga
, 200
6a
Stat
e-le
vel s
ocia
lca
pita
lK
awac
hi e
t al.,
199
9
Subr
aman
ian
et a
l.,20
01
42,3
58 a
dults
in
22 E
urop
ean
coun
trie
s
167,
259
adul
ts in
39
US
stat
es
144,
692
adul
ts in
39
US
stat
es
Soci
al c
apita
l m
easu
re &
hea
lthou
tcom
e m
easu
re: 1
5�y
Soci
al c
apita
l m
easu
re &
hea
lthou
tcom
e m
easu
re: 1
8�y
Soci
al c
apita
l m
easu
re &
hea
lthou
tcom
e m
easu
re: 1
8–98
y
Indi
vidu
al le
vel:
soci
al tr
ust,
asso
ciat
iona
lm
embe
rshi
psN
atio
nal l
evel
:so
cial
trus
t,as
soci
atio
nal
mem
bers
hips
Stat
e le
vel:
soci
al tr
ust,
reci
-pr
ocity
, gro
upm
embe
rshi
ps
Stat
e le
vel:
soci
al m
istr
ust
(con
tinuo
us %
)
Dic
hoto
mou
s
Dic
hoto
mou
s
Dic
hoto
mou
s
gove
rnan
ce
qual
ity
Indi
vidu
al le
vel:
age,
gen
der,
edu-
catio
n, in
com
e
Indi
vidu
al le
vel:
age,
gen
der,
race
/et
hnic
ity, i
ncom
e,m
arita
l sta
tus,
smok
ing,
obe
sity
,he
alth
insu
ranc
eco
vera
ge, h
ealth
chec
kup
in la
sttw
o ye
ars
Indi
vidu
al le
vel:
age,
gen
der,
race
/eth
nici
ty,
mar
ital s
tatu
s,in
com
e, s
mok
ing,
heal
th in
sura
nce
cove
rage
, hea
lth
1) H
igh
soci
al tr
ust:
OR
�0.
66, 9
5%
CI
�0.
62–0
.70
2) H
igh
asso
cia-
tiona
l mem
ber-
ship
s:O
R �
0.76
, 95%
C
I �
0.70
–0.8
2
– –
1) H
igh
soci
al tr
ust:
OR
�0.
91, 9
5%
CI
�0.
73–1
.14
2) H
igh
asso
ciat
iona
l m
embe
rshi
ps: O
R �
0.91
,95
% C
I �
0.71
–1.1
7
1) H
igh
soci
al tr
ust:
OR
�0.
71, 9
5%
CI
�0.
67–0
.75
2) H
igh
reci
proc
ity:
OR
�0.
68, 9
5%
CI
�0.
64–0
.71
3) H
igh
grou
p m
embe
rshi
ps:
OR
�0.
82, 9
5%C
I �
0.76
–0.8
8
Hig
her
soci
al tr
ust:
OR
�0.
99;
95%
CI
�0.
98–0
.996
TA
BL
E8.
2. (
Con
tinu
ed)
Sam
ple
size
,A
utho
rs,
popu
latio
n/So
cial
cap
ital
Form
of
self
-rat
edIn
divi
dual
-lev
elA
rea-
leve
l ye
arse
tting
Age
ran
gem
easu
rehe
alth
mea
sure
Cov
aria
tes
effe
ct e
stim
ate
effe
ct e
stim
ate
010203040506070809101112131415161718192021222324252627282930313233343536373839404142434445
Book_Kawachi, Subramanian & Kim_0387713107_proof3_180707
Mel
lor
& M
ilyo,
2005
Nei
ghbo
rhoo
d- o
rco
mm
unit
y-le
vel
soci
al c
apit
alSu
bram
ania
n et
al.,
2002
2 sa
mpl
es:
~68,
000
adul
ts in
39 U
S st
ates
;~7
6,00
0 ad
ults
in48
US
stat
es
21,4
56 a
dults
in 4
0U
S co
mm
uniti
es
Soci
al c
apita
l m
easu
re: 1
8�y
Hea
lth o
utco
me
mea
sure
: 16�
y
Soci
al c
apita
l mea
s-ur
e &
hea
lth o
ut-
com
e m
easu
re:
18–8
9 y
Stat
e le
vel:
soci
alm
istr
ust,
grou
pm
embe
rshi
ps,
Putn
am s
ocia
lca
pita
l ind
ex(d
eriv
ed f
rom
14
indi
cato
rs)
Com
mun
ity
leve
l:so
cial
trus
t (g
ener
al, t
rust
inne
ighb
ors,
cow
orke
rs, f
ello
wco
ngre
gant
s, s
tore
empl
oyee
s, lo
cal
polic
e)
Ord
inal
(fi
ve c
ate-
gori
es; h
ighe
r �
bette
r he
alth
)
Dic
hoto
mou
s
chec
kup
in la
stye
arSt
ate
leve
l:m
edia
nho
useh
old
inco
me,
inco
me
ineq
ualit
ySt
ate-
indi
vidu
alin
tera
ctio
n:st
ate
inco
me
ineq
ualit
y x
indi
-vi
dual
inco
me
inte
ract
ions
Indi
vidu
al le
vel:
age,
gen
der,
race
/eth
nici
ty,
mar
ital s
tatu
s,in
com
e, e
duca
-tio
n, h
ealth
in
sura
nce
cove
r-ag
e, c
entr
al c
ity/
MSA
res
iden
ce,
Stat
e le
vel:
med
ian
hous
ehol
d in
com
e
Indi
vidu
al le
vel:
age,
gen
der,
race
/et
hnic
ity, m
arita
lst
atus
, inc
ome,
educ
atio
n
–
Hig
h so
cial
trus
t:O
R �
0.55
, 95%
C
I �
0.50
-0.6
1
1) S
ocia
l mis
trus
t:ß
< 0
, p
< 0
.05
2) G
roup
mem
bers
hips
:ß
> 0
, p
> 0
.05
3) P
utna
m s
ocia
l cap
ital i
ndex
: ß
> 0
, p
< 0
.05
Hig
h so
cial
trus
t: O
R �
0.87
, 95%
CI �
0.62
-1.2
1In
tera
ctio
n m
odel
s:si
gnif
ican
t pos
itive
inte
ract
ion
betw
een
high
com
mun
ityso
cial
trus
t and
hig
h in
divi
d-ua
l soc
ial t
rust
(Con
tinu
ed)
010203040506070809101112131415161718192021222324252627282930313233343536373839404142434445
Book_Kawachi, Subramanian & Kim_0387713107_proof3_180707
Bro
wni
ng &
Cag
ney,
200
3
Wen
et a
l., 2
003
2,21
8 ad
ults
in 3
33ne
ighb
orho
ods
inth
e ci
ty o
fC
hica
go, U
S
3,45
9 ad
ults
in 2
75ne
ighb
orho
odcl
uste
rs in
Chi
cago
Soci
al c
apita
l mea
s-ur
e &
hea
lth o
ut-
com
e m
easu
re:
18�
y
Soci
al c
apita
l mea
s-ur
e &
hea
lth o
ut-
com
e m
easu
re:
18�
y
Nei
ghbo
rhoo
d le
vel:
frie
ndsh
ip s
ocia
lsu
ppor
t and
ne
twor
ks
Nei
ghbo
rhoo
d le
vel:
soci
al re
sour
ces
(rec
ipro
city
, den
-si
ty o
f soc
ial n
et-
wor
ks, s
ocia
lco
hesi
on, i
nfor
mal
soci
al c
ontr
ol)
Dic
hoto
mou
s
Ord
inal
(fo
ur c
ate-
gori
es; h
ighe
r �
bette
r he
alth
)
Indi
vidu
al le
vel:
age,
gen
der,
race
/et
hnic
ity, m
arita
lst
atus
, inc
ome,
educ
atio
n, h
ouse
-ho
ld te
nure
, yea
rsin
nei
ghbo
rhoo
d,fo
reig
n-bo
rn
stat
us, i
nter
view
year
Nei
ghbo
rhoo
d le
vel:
tota
l pop
ulat
ion,
resi
dent
ial s
tabi
l-ity
, im
mig
rant
conc
entr
atio
n,pr
ior
neig
hbor
-ho
od h
ealth
, dis
-or
der,
anom
ie,
tole
ranc
e of
ris
kbe
havi
or, c
olle
c-tiv
e ef
fica
cy
Indi
vidu
al le
vel:
age
,ge
nder
, rac
e/et
hnic
ity, m
arita
lst
atus
, inc
ome,
educ
atio
n, s
mok
-in
g, h
yper
tens
ion,
inte
rvie
w y
ear
– –
Hig
h so
cial
sup
port
and
ne
twor
ks:
OR
�0.
89, 9
5%
CI
�0.
78-1
.02
Hig
her
soci
al r
esou
rces
:O
R �
1.19
p
< 0
.05
TA
BL
E8.
2. (
Con
tinu
ed)
Sam
ple
size
,A
utho
rs,
popu
latio
n/So
cial
cap
ital
Form
of
self
-rat
edIn
divi
dual
-lev
elA
rea-
leve
l ye
arse
tting
Age
ran
gem
easu
rehe
alth
mea
sure
Cov
aria
tes
effe
ct e
stim
ate
effe
ct e
stim
ate
010203040506070809101112131415161718192021222324252627282930313233343536373839404142434445
Book_Kawachi, Subramanian & Kim_0387713107_proof3_180707
Dru
kker
et a
l., 2
003
Lin
dstr
öm e
t al.,
2004
Hel
liwel
l &
Putn
am, 2
004
3,40
1 ad
oles
cent
s in
36 n
eigh
bor-
hood
s in
M
aast
rich
t,N
ethe
rlan
ds
3,60
2 ad
ults
in 7
5ne
ighb
orho
ods
inM
alm
ö, S
wed
en
28,7
66 a
dults
in 4
0U
S co
mm
uniti
es
Soci
al c
apita
l m
easu
re: 2
0–65
yH
ealth
out
com
em
easu
re: ~
10–1
2 y
Soci
al c
apita
l mea
s-ur
e &
hea
lth o
ut-
com
e m
easu
re:
20–8
0 y
Soci
al c
apita
l mea
s-ur
e &
hea
lth o
ut-
com
e m
easu
re:
18–9
9 y
Nei
ghbo
rhoo
d le
vel:
soci
al c
ohes
ion
and
trus
t
Indi
vidu
al le
vel:
soci
al
part
icip
atio
n
Indi
vidu
al le
vel:
asso
ciat
iona
lm
embe
rshi
ps,
soci
al tr
ust (
gen-
eral
, in
neig
hbor
s,po
lice)
Com
mun
ity
leve
l:as
soci
atio
nal
Con
tinuo
us
Dic
hoto
mou
s
Con
tinuo
us (
high
er�
bette
r he
alth
)
Nei
ghbo
rhoo
d le
vel:
pove
rty,
aff
lu-
ence
, inc
ome
ineq
ualit
y, e
duca
-tio
n, h
ealth
-en
hanc
ing
serv
ices
, cri
me
expo
sure
, pri
orhe
alth
Indi
vidu
al le
vel:
gend
er, g
rade
rete
ntio
nH
ouse
hold
leve
l:oc
cupa
tiona
l sta
-tu
s, e
duca
tion,
fam
ily w
elfa
rest
atus
, sin
gle
pare
ntN
eigh
borh
ood
leve
l:so
cioe
cono
mic
depr
ivat
ion,
res
i-de
ntia
l ins
tabi
lity
Indi
vidu
al le
vel:
age,
gen
der,
coun
try
of o
rigi
n,ed
ucat
ion
Indi
vidu
al le
vel:
age,
gen
der,
mar
i-ta
l sta
tus,
empl
oym
ent s
ta-
tus,
impo
rtan
ce o
fG
od/r
elig
ion,
fre
-qu
ency
of
atte
nd-
ing
relig
ious
–
Hig
h so
cial
pa
rtic
ipat
ion:
OR
�0.
34, 9
5%
CI
�0.
27–0
.43
1) A
ssoc
iatio
nal
mem
bers
hips
:ß
> 0
, p <
0.0
12)
Gen
eral
trus
t: ß
> 0
, p <
0.0
13)
Tru
st in
nei
gh-
bors
: ß
> 0
, p <
0.0
1
ß <
0, p
> 0
.05
–
1) A
ssoc
iatio
nal m
embe
rshi
ps:
ß >
0, p
> 0
.05
2) S
ocia
l tru
st: ß
> 0
, p <
0.0
1
(Con
tinu
ed)
010203040506070809101112131415161718192021222324252627282930313233343536373839404142434445
Book_Kawachi, Subramanian & Kim_0387713107_proof3_180707
Vee
nstr
a, 2
005a
Zie
rsch
et a
l., 2
005
1,18
4 ad
ults
in 2
5co
mm
uniti
es in
the
prov
ince
of
Bri
tish
Col
umbi
a,C
anad
a
2,40
0 ad
ults
in s
ub-
urba
n ne
ighb
or-
hood
s in
Ade
laid
e,
Aus
tral
ia
Soci
al c
apita
l mea
s-ur
es &
hea
lth o
ut-
com
e m
easu
re:
18�
y
Soci
al c
apita
l mea
s-ur
e &
hea
lth o
ut-
com
e m
easu
re:
18�
y
mem
bers
hips
,so
cial
trus
t
Indi
vidu
al le
vel:
soci
al tr
ust,
polit
-ic
al tr
ust,
soci
alpa
rtic
ipat
ion
Nei
ghbo
rhoo
d le
vel:
soci
al tr
ust,
soci
al c
onne
c-tio
ns/c
ohes
ion
Dic
hoto
mou
s
Self
-rep
orte
d ph
ysi-
cal h
ealth
(c
ontin
uous
)
serv
ice,
com
mut
etim
e to
wor
kC
omm
unit
y le
vel:
med
ian
inco
me,
impo
rtan
ce o
fG
od/r
elig
ion
Indi
vidu
al le
vel:
age,
gen
der,
fore
ign-
born
,in
com
e,
educ
atio
n
Indi
vidu
al le
vel:
age
,ge
nder
, inc
ome,
educ
atio
n, h
ouse
-ho
ld te
nure
, yea
rsat
add
ress
Nei
ghbo
rhoo
d le
vel:
pollu
tion,
saf
ety
4) T
rust
in p
olic
e:ß
> 0
, p <
0.0
1
Soci
al tr
ust:
OR
�0.
73, 9
5%
CI
�0.
56–0
.97
p �
0.03
Poli
tica
l tru
st:
OR
�0.
57, 9
5%C
I �
0.44
–0.7
5p
< 0
.01
Part
icip
atio
n in
vo
lunt
ary
asso
ciat
ions
:O
R �
0.96
, 95%
C
I �
0.84
–1.1
0p
�0.
58 –
–
1) N
eigh
borh
ood
trus
t:ß
> 0
, p >
0.0
52)
Nei
ghbo
rhoo
d co
nnec
tion
s/co
hesi
on:
ß >
0, p
> 0
.05
TA
BL
E8.
2. (
Con
tinu
ed)
Sam
ple
size
,A
utho
rs,
popu
latio
n/So
cial
cap
ital
Form
of
self
-rat
edIn
divi
dual
-lev
elA
rea-
leve
l ye
arse
tting
Age
ran
gem
easu
rehe
alth
mea
sure
Cov
aria
tes
effe
ct e
stim
ate
effe
ct e
stim
ate
010203040506070809101112131415161718192021222324252627282930313233343536373839404142434445
Book_Kawachi, Subramanian & Kim_0387713107_proof3_180707
Step
toe
& F
eldm
an,
2001
Kav
anag
h et
al.,
2006
b
654
adul
ts in
37
neig
hbor
hood
s in
Lon
don,
Eng
land
15,1
12 a
dults
in 4
1st
atis
tical
loca
lar
eas
in th
e st
ate
of T
asm
ania
,A
ustr
alia
Soci
al c
apita
l mea
s-ur
e &
hea
lth o
ut-
com
e m
easu
re:
18–9
4 y
Soci
al c
apita
l mea
s-ur
e &
hea
lth o
ut-
com
e m
easu
re:
18–9
7 y
Nei
ghbo
rhoo
d le
vel:
soci
al c
ohes
ion
Are
a le
vel:
soci
altr
ust,
soci
al c
ohe-
sion
, pol
itica
lpa
rtic
ipat
ion
Self
-rep
orte
d ph
ysi-
cal f
unct
ion
(dic
hoto
mou
s)
Dic
hoto
mou
s
Indi
vidu
al le
vel:
age,
sex
, so
cioe
cono
mic
depr
ivat
ion
Nei
ghbo
rhoo
dle
vel:
soci
oeco
-no
mic
dep
riva
tion
Indi
vidu
al le
vel:
age,
mar
ital s
ta-
tus,
inco
me,
edu
-ca
tion,
indi
geno
usst
atus
, sm
okin
g A
rea
leve
l: s
ocio
eco-
nom
ic d
isad
van-
tage
, geo
grap
hic
rem
oten
ess,
neig
hbor
hood
safe
ty
– –
Low
soc
ial c
ohes
ion:
OR
�2.
31,
95%
CI
�1.
16-6
.63
Men
:1)
Soc
ial t
rust
:ß
< 0
, p �
0.01
2) P
oliti
cal p
artic
ipat
ion:
ß
> 0
, p �
0.88
3) T
rust
in p
ublic
and
pri
vate
inst
itutio
ns:
ß >
0, p
�0.
924)
Nei
ghbo
rhoo
d in
tegr
atio
n:ß
< 0
, p �
0.53
5) N
eigh
borh
ood
alie
natio
n:ß
< 0
, p �
0.02
Wom
en:
1) S
ocia
l tru
st:
ß <
0, p
�0.
012)
Pol
itica
l par
ticip
atio
n:ß
> 0
, p �
0.91
3) T
rust
in p
ublic
and
pri
vate
inst
itutio
ns:
ß >
0, p
�0.
964)
Nei
ghbo
rhoo
d in
tegr
atio
n:ß
< 0
, p �
0.30
5) N
eigh
borh
ood
alie
natio
n:ß
< 0
, p �
0.31
(Con
tinu
ed)
010203040506070809101112131415161718192021222324252627282930313233343536373839404142434445
Book_Kawachi, Subramanian & Kim_0387713107_proof3_180707
Zim
mer
man
&B
ell,
2006
4,
817
adul
ts in
855
US
coun
ties
and
45 s
tate
s
Soci
al c
apita
l m
easu
re: 1
8�y
Hea
lth o
utco
me
mea
sure
: 40–
45 y
Stat
e le
vel:
soci
alca
pita
l ind
ex(d
eriv
ed f
rom
nine
indi
cato
rs o
fso
cial
trus
t, ci
vic
enga
gem
ent,
and
anom
ie)
Dic
hoto
mou
sIn
divi
dual
leve
l:ge
nder
, rac
e/
ethn
icity
, mar
ital
stat
us, u
rban
res
i-de
nce,
reg
ion,
inco
me,
edu
ca-
tion,
pov
erty
sta
-tu
s, e
mpl
oym
ent
stat
us, h
ealth
insu
ranc
e st
atus
Cou
nty
leve
l:pr
o-po
rtio
n w
ealth
y,un
skill
ed w
ages
,ho
usin
g af
ford
-ab
ility
, cri
me
rate
,pr
opor
tion
unem
-pl
oyed
, pro
port
ion
Bla
ck, p
ropo
rtio
nH
ispa
nic,
mea
nin
com
e, m
ean
year
s of
edu
ca-
tion,
inde
x of
avai
labi
lity
ofps
ychi
atri
c se
rv-
ices
, ind
ex o
fav
aila
bilit
y of
heal
th s
ervi
ces
Stat
e le
vel:
gene
ros-
ity
of s
tate
sp
endi
ng
–O
R �
1.09
, 95%
CI
�0.
56–2
.12
TA
BL
E8.
2. (
Con
tinu
ed)
Sam
ple
size
,A
utho
rs,
popu
latio
n/So
cial
cap
ital
Form
of
self
-rat
edIn
divi
dual
-lev
elA
rea-
leve
l ye
arse
tting
Age
ran
gem
easu
rehe
alth
mea
sure
Cov
aria
tes
effe
ct e
stim
ate
effe
ct e
stim
ate
010203040506070809101112131415161718192021222324252627282930313233343536373839404142434445
Book_Kawachi, Subramanian & Kim_0387713107_proof3_180707
Dru
kker
et a
l., 2
005
Poor
tinga
, 200
6d
Kim
et a
l., 2
006a
801
adol
esce
nts
in34
3 ne
ighb
or-
hood
s in
Chi
cago
;53
3 ad
oles
cent
sin
36
neig
hbor
-ho
ods
in
Maa
stri
cht,
Net
herl
ands
7,39
4 ad
ults
in4,
332
hous
ehol
dsin
720
pos
tal
code
sec
tors
inth
e U
K
24,8
35 a
dults
in 4
0U
S co
mm
uniti
es
Soci
al c
apita
l mea
s-ur
e: 1
8�y
Hea
lth o
utco
me
mea
sure
: 12
y (C
hica
go);
10–1
3 y
(Maa
stri
cht)
Soci
al c
apita
l mea
s-ur
e &
hea
lth o
ut-
com
e m
easu
re:
16�
y
Soci
al c
apita
l mea
s-ur
e &
hea
lth o
ut-
com
e m
easu
re:
18–9
9 y
Nei
ghbo
rhoo
d le
vel:
soci
al c
ohes
ion
and
trus
t
Indi
vidu
al le
vel:
Soci
al s
uppo
rt,
soci
al tr
ust,
civi
cpa
rtic
ipat
ion
Com
mun
ity
leve
l:re
cipr
ocity
Indi
vidu
al le
vel:
form
al b
ondi
ngso
cial
cap
ital,
trus
t in
mem
bers
of o
ne’s
rac
e/
ethn
icity
, for
mal
brid
ging
, inf
or-
mal
bri
dgin
g,so
cial
trus
t
Con
tinuo
us (
high
er�
bette
r he
alth
)
Dic
hoto
mou
s
Dic
hoto
mou
s
Indi
vidu
al le
vel:
gend
er, a
ge/g
rade
rete
ntio
n,ra
ce/e
thni
city
Hou
seho
ld le
vel:
occu
patio
nal s
ta-
tus,
edu
catio
n,fa
mily
wel
fare
stat
us, s
ingl
e pa
rent
Nei
ghbo
rhoo
d le
vel:
soci
oeco
nom
icde
priv
atio
n
Indi
vidu
al le
vel:
age,
gen
der,
phys
-ic
al a
ctiv
ity
Hou
seho
ld le
vel:
soci
al c
lass
,ho
useh
old
tenu
re
Indi
vidu
al le
vel:
age,
gen
der,
race
/eth
nici
ty,
mar
ital s
tatu
s,in
com
e, e
duca
tion
Com
mun
ity
leve
l:m
ean
age,
per
cent
low
inco
me,
sta
teco
mm
unity
–
1) S
ever
e la
ck o
fso
cial
sup
port
: O
R �
2.17
, 95%
C
I �
1.72
-2.7
32)
Hig
h so
cial
trus
t:O
R �
0.69
, 95%
CI
�0.
58-0
.82
3) H
igh
civi
c pa
rtic
-ip
atio
n:
OR
�0.
62, 9
5%C
I �
0.51
-0.7
6
1) H
igh
form
albo
ndin
g so
cial
cap
ital:
OR
�0.
77, 9
5%
CI
�0.
66-0
.88
2) H
igh
trus
t in
mem
bers
of
one’
sra
ce/e
thni
city
: O
R �
0.88
, 95%
C
I �
0.79
-0.9
8
Maa
stri
cht:
ß <
0, p
> 0
.05
Chi
cago
, non
-His
pani
cs:
ß >
0, p
> 0
.05
Chi
cago
, His
pani
cs:
ß >
0, p
> 0
.05
Rec
ipro
city
: O
R �
0.52
, 95%
CI
�0.
33-0
.83
1-SD
* hi
gher
bon
ding
soc
ial
capi
tal:
OR
�0.
86, 9
5%
CI
�0.
80-0
.92
1-SD
* hi
gher
bri
dgin
g so
cial
capi
tal:
OR
�0.
95, 9
5%
CI
�0.
88-1
.02
Inte
ract
ion
mod
els:
sig
nifi
cant
lyw
eake
r bo
ndin
g so
cial
cap
ital
asso
ciat
ions
am
ong
Bla
cks
and
thos
e in
the
“Oth
er”
(Con
tinu
ed)
010203040506070809101112131415161718192021222324252627282930313233343536373839404142434445
Book_Kawachi, Subramanian & Kim_0387713107_proof3_180707
Kim
et a
l., 2
006b
24,8
35 a
dults
in
40
US
com
mun
ities
Soci
al c
apita
l mea
s-ur
e &
hea
lth o
ut-
com
e m
easu
re:
18–9
9 y
Com
mun
ity
leve
l:bo
ndin
g so
cial
cap
i-ta
l, br
idgi
ngso
cial
cap
ital
Indi
vidu
al le
vel:
soci
al tr
ust,
info
r-m
al s
ocia
l int
er-
actio
ns, d
iver
sity
of f
rien
dshi
p ne
t-w
orks
, ele
ctor
alpo
litic
al p
artic
i-pa
tion,
and
non
-el
ecto
ral p
oliti
cal
part
icip
atio
n, f
or-
mal
gro
upin
volv
emen
t/rel
i-gi
ous
grou
pin
volv
emen
t/gi
ving
and
vol
un-
teer
ing
Com
mun
ity
leve
l:th
ree
soci
al c
api-
tal s
ubsc
ales
base
d on
sam
e
Dic
hoto
mou
sIn
divi
dual
leve
l:ag
e, g
ende
r,ra
ce/ e
thni
city
,m
arita
l sta
tus,
inco
me,
edu
ca-
tion,
pro
xim
ity to
core
urb
an a
reas
Com
mun
ity
leve
l:m
ean
age,
pro
-po
rtio
n w
ith lo
win
com
e, p
ropo
r-tio
n w
ith lo
wed
ucat
ion,
sta
teco
mm
unity
3) H
igh
form
albr
idgi
ng:
OR
�1.
07, 9
5%
CI
�0.
87-1
.31
4) H
igh
info
rmal
brid
ging
:O
R �
0.99
, 95%
C
I �
0.91
-1.0
85)
Hig
h so
cial
trus
tO
R �
0.54
, 95%
C
I �
0.49
-0.5
9
1) S
ocia
l tru
st:
OR
�0.
56, 9
5%
CI
�0.
52-0
.62
2) I
nfor
mal
soc
ial
inte
ract
ions
: O
R �
0.96
, 95%
CI
�0.
88-1
.05
3) E
lect
oral
par
tici-
patio
n:O
R �
0.78
, 95%
C
I �
0.71
-0.8
64)
Div
ersi
ty o
ffr
iend
ship
s:O
R �
0.98
, 95%
C
I �
0.90
-1.0
75)
Non
-ele
ctor
alpa
rtic
ipat
ion
OR
�1.
18, 9
5%
CI
�1.
06-1
.31
raci
al/e
thni
c ca
tego
ry th
anam
ong
Whi
tes
Subs
cale
1(s
ocia
l tru
st, i
nfor
mal
soci
al in
tera
ctio
ns, e
lect
oral
polit
ical
par
ticip
atio
n): O
R �
1.00
, 95%
CI
�0.
93-1
.06
Subs
cale
2(f
orm
al g
roup
par
tici-
patio
n, r
elig
ious
gro
up p
artic
i-pa
tion,
giv
ing
and
volu
ntee
ring
):O
R �
0.94
, 95%
C
I �
0.89
-0.9
9Su
bsca
le 3
(div
ersi
ty o
f fr
iend
-sh
ips,
non
-ele
ctor
al p
oliti
cal
part
icip
atio
n): O
R �
0.91
,95
% C
I �
0.85
-0.9
8H
igh
on a
ll th
ree
subs
cale
s: O
R�
0.82
, 95%
CI
�0.
69-0
.98
TA
BL
E8.
2. (
Con
tinu
ed)
Sam
ple
size
,A
utho
rs,
popu
latio
n/So
cial
cap
ital
Form
of
self
-rat
edIn
divi
dual
-lev
elA
rea-
leve
l ye
arse
tting
Age
ran
gem
easu
rehe
alth
mea
sure
Cov
aria
tes
effe
ct e
stim
ate
effe
ct e
stim
ate
010203040506070809101112131415161718192021222324252627282930313233343536373839404142434445
Book_Kawachi, Subramanian & Kim_0387713107_proof3_180707
Poor
tinga
, 200
6c
Poor
tinga
, 200
6b
Fran
zini
et a
l., 2
005
2 U
K s
ampl
es:
1) 7
,988
adu
lts in
4,78
7 ho
useh
olds
in 3
60 s
ampl
ing
poin
ts/p
osta
l cod
ese
ctor
s2)
7,3
94 a
dults
in4,
332
hous
ehol
dsin
720
pos
tal
code
sec
tors
14,8
36 a
dults
in 7
20po
stal
cod
e se
c-to
rs in
the
UK
3,15
1 ad
ults
in 1
00ne
ighb
orho
ods
inTe
xas
Soci
al c
apita
l mea
s-ur
e &
hea
lth
outc
ome
mea
sure
: 16�
y
Soci
al c
apita
l mea
s-ur
e &
hea
lth o
ut-
com
e m
easu
re:
16�
y
Soci
al c
apita
l mea
s-ur
e &
hea
lth
8 in
dica
tors
as
atin
divi
dual
leve
l
Indi
vidu
al le
vel:
soci
al s
uppo
rt,
soci
al tr
ust,
civi
cpa
rtic
ipat
ion
Com
mun
ity
leve
l:so
cial
trus
t, ci
vic
part
icip
atio
n, r
ec-
ipro
city
Indi
vidu
al le
vel:
soci
al s
uppo
rt,
soci
al tr
ust,
civi
cpa
rtic
ipat
ion,
re
cipr
ocity
Nei
ghbo
rhoo
d le
vel:
soci
al c
apita
l
Dic
hoto
mou
s
Dic
hoto
mou
s
Con
tinuo
us
Indi
vidu
al le
vel:
age,
gen
der,
phys
-ic
al a
ctiv
ityH
ouse
hold
leve
l:so
cial
cla
ss,
hous
ehol
d te
nure
Indi
vidu
al le
vel:
age,
gen
der,
mar
i-ta
l sta
tus,
soc
ial
clas
s, u
nem
ploy
-m
ent s
tatu
s,ho
useh
old
tenu
re,
acce
ss to
am
eni-
ties,
pre
senc
e of
loca
l soc
ial p
rob-
lem
s, u
rban
re
side
nce
Indi
vidu
al le
vel:
age,
gen
der,
race
/
6) F
orm
al g
roup
invo
lvem
ent/
reli
giou
s gr
oup
invo
lvem
ent/
givi
ngan
d vo
lun-
teer
ing
OR
�0.
68, 9
5%
CI
�0.
62–0
.75
1) S
ever
e la
ck o
fso
cial
sup
port
: O
R �
2.21
, 95%
C
I �
1.76
–2.7
8 2)
Hig
h so
cial
trus
t:O
R �
0.75
, 95%
CI
�0.
62–0
.90
3) H
igh
civi
c pa
rtic
-ip
atio
n:O
R �
0.62
, 95%
C
I �
0.51
–0.7
7
1) S
ever
e la
ck o
fso
cial
sup
port
: O
R �
1.64
, 95%
C
I �
1.42
–1.9
0 2)
Hig
h so
cial
trus
t:O
R �
0.74
, 95%
CI
�0.
67–0
.82
3) H
igh
soci
al p
ar-
ticip
atio
n:
OR
�0.
67, 9
5%C
I �
0.60
–0.7
64)
Hig
h re
cipr
ocity
:O
R �
0.82
, 95%
CI
�0.
73–0
.93
–
1) H
igh
soci
al tr
ust:
OR
�0.
39, 9
5%
CI
�0.
22–0
.71
2) H
igh
civi
c pa
rtic
ipat
ion:
OR
�0.
90, 9
5%C
I �
0.53
–1.5
43)
Hig
h re
cipr
ocity
OR
�0.
52, 9
5%
CI
�0.
33–0
.83 –
ß >
0, p
< 0
.05
(Con
tinu
ed)
010203040506070809101112131415161718192021222324252627282930313233343536373839404142434445
Book_Kawachi, Subramanian & Kim_0387713107_proof3_180707
Yip
et a
l., in
pre
ss
IND
IVID
UA
L-
LE
VE
L
STU
DIE
S:R
ose,
200
0
1,21
8 ad
ults
in 4
8vi
llage
s in
the
Shan
dong
prov
ince
, Chi
na
1,90
4 ad
ults
in th
eR
ussi
an
Fede
ratio
n
outc
ome
mea
sure
: 18–
94 y
Soci
al c
apita
l m
easu
re &
he
alth
out
com
em
easu
re:
16-8
0 y
Soci
al c
apita
l m
easu
re &
hea
lthou
tcom
e m
easu
res:
18�
y
(bas
ed o
n 2
sub-
scal
es o
f so
cial
trus
t, re
cipr
ocity
)
Indi
vidu
al le
vel:
Soci
al tr
ust,
part
ym
embe
rshi
ps,
volu
ntar
y or
gani
zatio
nm
embe
rshi
ps
Vill
age
leve
l:
Soci
al tr
ust,
part
ym
embe
rshi
ps,
volu
ntar
y or
gani
zatio
nm
embe
rshi
ps
Gen
eral
soc
ial t
rust
,so
cial
sup
port
Dic
hoto
mou
s
Self
-rat
ed p
hysi
cal
heal
th (
cont
inu-
ous;
hig
her
�be
tter
heal
th)
ethn
icity
, fam
ilyin
com
e-to
-nee
dsra
tioN
eigh
borh
ood
leve
l:co
llect
ive
effi
-ca
cy, c
hild
-rel
ated
proc
esse
s, d
isor
-de
r, fe
ar, r
acis
m
Indi
vidu
al le
vel:
age,
gen
der,
mar
i-ta
l sta
tus,
occ
upa-
tion,
edu
catio
nH
ouse
hold
leve
l:in
com
e, a
sset
s,si
ze
Age
, gen
der,
inco
me,
edu
ca-
tion,
sub
ject
ive
soci
al s
tatu
s
1) S
ocia
l tru
st:
OR
�0.
71, 9
5%C
I �
0.61
-0.8
32)
Par
ty m
embe
r-sh
ips:
O
R �
0.62
, 95%
CI
�0.
43-0
.90
3) V
olun
tary
org
ani-
zatio
n m
embe
r-sh
ips:
O
R �
0.81
, 95%
CI
�0.
50-1
.32
1) G
ener
al s
ocia
ltr
ust:
ß >
0, p
< 0
.05
4) S
ocia
l sup
port
: ß
> 0
, p <
0.0
5
1) S
ocia
l tru
st:
OR
�0.
76, 9
5%
CI
�0.
51-1
.13
2) P
arty
mem
bers
hips
:O
R �
0.96
, 95%
C
I �
0.13
-7.1
03)
Vol
unta
ry o
rgan
izat
ion
mem
bers
hips
: O
R �
1.15
, 95%
C
I �0.
04-3
5.30 –
TA
BL
E8.
2. (
Con
tinu
ed)
Sam
ple
size
,A
utho
rs,
popu
latio
n/So
cial
cap
ital
Form
of
self
-rat
edIn
divi
dual
-lev
elA
rea-
leve
l ye
arse
tting
Age
ran
gem
easu
rehe
alth
mea
sure
Cov
aria
tes
effe
ct e
stim
ate
effe
ct e
stim
ate
010203040506070809101112131415161718192021222324252627282930313233343536373839404142434445
Book_Kawachi, Subramanian & Kim_0387713107_proof3_180707
Vee
nstr
a, 2
000
Hyy
ppä
& M
äki,
2001
534
adul
ts in
the
prov
ince
of
Sask
atch
ewan
,C
anad
a
2,00
0 ad
ults
inm
unic
ipal
ities
inFi
nlan
d
Soci
al c
apita
l m
easu
re &
hea
lthou
tcom
e m
easu
re: 1
8�y
Soci
al c
apita
l m
easu
re &
hea
lthou
tcom
e m
easu
re: 1
6�y
Civ
ic p
artic
ipat
ion,
part
icip
atio
n in
club
s; p
oliti
cal
trus
t, tr
ust i
nne
ighb
ors,
trus
t in
com
mun
ity m
em-
bers
, tru
st in
mem
bers
of
part
of p
rovi
nce,
gen
-er
al s
ocia
l tru
st;
freq
uenc
y of
soci
aliz
atio
n w
ithco
-wor
kers
, will
-in
gnes
s to
turn
toco
-wor
ker
in ti
me
of n
eed,
relig
ious
ser
vice
atte
ndan
ce
Soci
al m
istr
ust,
asso
ciat
iona
l par
-tic
ipat
ion,
rel
i-gi
ous
grou
p
Dic
hoto
mou
s
Dic
hoto
mou
s
Sele
cted
out
com
es:
inco
me,
ed
ucat
ion
Age
, inc
ome,
sm
ok-
ing,
bod
y m
ass
inde
x, u
rban
res
i-de
nce,
mig
ratio
n,
Civ
ic p
artic
ipat
ion:
ß no
t rep
orte
d,
p >
0.0
5Pa
rtic
ipat
ion
incl
ubs:
ß no
t rep
orte
d,
p >
0.0
5Po
litic
al tr
ust,
trus
tin
nei
ghbo
rs, t
rust
in c
omm
unity
mem
bers
, tru
st in
mem
bers
of
part
of p
rovi
nce,
gen
-er
al s
ocia
l tru
st:
ß no
t rep
orte
d,
p >
0.0
5A
lso
adju
sted
for
inco
me,
ed
ucat
ion:
Freq
uenc
y of
soci
aliz
atio
n w
ithco
-wor
kers
: ß
< 0
, p >
0.0
5W
illin
gnes
s to
turn
to c
o-w
orke
rin
tim
e of
nee
d:ß
< 0
, p >
0.0
5R
elig
ious
ser
vice
atte
ndan
ce:
ß <
0, p
< 0
.05
1) 1
–SD
* hi
gher
soci
al tr
ust:
OR
�0.
69, 9
5%C
I �
0.43
–1.1
6(m
en);
– –
(Con
tinu
ed)
010203040506070809101112131415161718192021222324252627282930313233343536373839404142434445
Book_Kawachi, Subramanian & Kim_0387713107_proof3_180707
part
icip
atio
n,co
mm
unity
pa
rtic
ipat
ion
com
orbi
dity
;an
alys
es s
trat
ifie
dby
gen
der
OR
�0.
64, 9
5%
CI
�0.
39–1
.06
(wom
en)
2) 1
–SD
* hi
gher
asso
ciat
iona
lpa
rtic
ipat
ion:
O
R �
0.74
, 95%
C
I �
0.47
-1.1
7(m
en);
OR
�0.
80, 9
5%
CI
�0.
54–1
.19
(wom
en)
3) 1
–SD
* hi
gher
relig
ious
gro
uppa
rtic
ipat
ion:
OR
�0.
42, 9
5%
CI
�0.
21-0
.85
(men
);O
R �
0.68
, 95%
C
I �
0.41
–1.1
3(w
omen
)4)
1-S
D*
high
erco
mm
unity
part
icip
atio
n:O
R �
1.02
, 95%
C
I �
0.55
–1.8
8(m
en)
OR
�0.
83, 9
5%
CI
�0.
44–1
.59
(wom
en)
TA
BL
E8.
2. (
Con
tinu
ed)
Sam
ple
size
,A
utho
rs,
popu
latio
n/So
cial
cap
ital
Form
of
self
-rat
edIn
divi
dual
-lev
elA
rea-
leve
l ye
arse
tting
Age
ran
gem
easu
rehe
alth
mea
sure
Cov
aria
tes
effe
ct e
stim
ate
effe
ct e
stim
ate
010203040506070809101112131415161718192021222324252627282930313233343536373839404142434445
Book_Kawachi, Subramanian & Kim_0387713107_proof3_180707
Hyy
ppä
& M
äki,
2003
Polla
ck &
K
nese
beck
,20
04
Vee
nstr
a et
al.,
2005
b
2,00
0 ad
ults
inm
unic
ipal
ities
inFi
nlan
d
608
adul
ts in
the
US
and
682
adul
ts in
Ger
man
y
1,50
4 ad
ults
in th
eci
ty o
f H
amilt
on,
Can
ada
Soci
al c
apita
l mea
s-ur
e &
hea
lth o
ut-
com
e m
easu
re:
16-6
5 y
Soci
al c
apita
l m
easu
re &
hea
lthou
tcom
e m
easu
re: 6
0�y
Soci
al c
apita
l &he
alth
out
com
em
easu
res:
18�
y
Ass
ocia
tiona
l par
tici-
patio
n, fr
iend
ship
netw
orks
, rel
igio
usgr
oup
part
icip
a-tio
n, h
obby
gr
oup
part
icip
atio
n
Soci
al tr
ust,
reci
proc
ity, a
sso-
ciat
iona
lm
embe
rshi
ps
Mem
bers
hip/
invo
lve-
men
t in
volu
ntar
yas
soci
atio
ns
Dic
hoto
mou
s
Dic
hoto
mou
s
Dic
hoto
mou
s
Age
, gen
der,
lang
uage
, mig
ra-
tion,
edu
catio
n,in
com
e, e
mpl
oy-
men
t sta
tus,
smok
ing,
dr
inki
ng, b
ody
mas
s in
dex,
com
orbi
dity
Age
, gen
der,
inco
me,
ed
ucat
ion
Age
, gen
der,
inco
me,
ed
ucat
ion,
ne
ighb
orho
od
1) 1
–SD
* hi
gher
asso
ciat
iona
lpa
rtic
ipat
ion:
OR
�0.
84, 9
5%
CI
�0.
71-1
.00
2) 1
–SD
* hi
gher
frie
ndsh
ip n
et-
wor
k:
OR
�0.
80, 9
5%
CI
�0.
69-0
.92
3) R
elig
ious
gro
uppa
rtic
ipat
ion:
OR
�0.
75, 9
5%
CI
�0.
64-0
.89
4) H
obby
gro
uppa
rtic
ipat
ion:
O
R �
1.09
, 95%
CI
�0.
89-1
.33
1) H
igh
soci
al tr
ust:
OR
�0.
5, 9
5%
CI
�0.
3-0.
82)
Hig
h re
cipr
ocity
OR
�0.
4, 9
5%
CI
�0.
2-0.
63)
Hig
h as
soci
a-tio
nal m
embe
r-sh
ips
OR
�0.
7, 9
5%
CI
�0.
4-1.
2
Hig
her
volu
ntar
yas
soci
atio
nin
volv
emen
t:O
R �
0.92
, p
�0.
20
– – –
(Con
tinu
ed)
010203040506070809101112131415161718192021222324252627282930313233343536373839404142434445
Book_Kawachi, Subramanian & Kim_0387713107_proof3_180707
Roj
as &
Car
lson
,20
061,
794
adul
ts in
Taga
nrog
, Rus
sia
Soci
al c
apita
l m
easu
re &
hea
lthou
tcom
e m
easu
re: 2
0+ y
Mem
bers
hip
intr
ade
unio
n/po
litic
al o
rgan
iza-
tions
, in
othe
ror
gani
zatio
ns,
cont
act w
ithne
ighb
ors
Con
tinuo
us (
high
er�
bette
r he
alth
)A
ge, g
ende
r, m
arita
lst
atus
, inc
ome,
educ
atio
n
1) M
embe
rshi
p in
trad
e un
ion/
po
litic
al
orga
niza
tions
: ß
> 0
, p <
0.0
12)
Mem
bers
hip
inot
her
orga
niza
tions
: ß
> 0
, p �
0.01
3) C
onta
ct w
ithne
ighb
ors:
ß >
0, p
�0.
10
–
TA
BL
E8.
2. (
Con
tinu
ed)
Sam
ple
size
,A
utho
rs,
popu
latio
n/So
cial
cap
ital
Form
of
self
-rat
edIn
divi
dual
-lev
elA
rea-
leve
l ye
arse
tting
Age
ran
gem
easu
rehe
alth
mea
sure
Cov
aria
tes
effe
ct e
stim
ate
effe
ct e
stim
ate
*1–
SD =
1–st
anda
rd d
evia
tion.
010203040506070809101112131415161718192021222324252627282930313233343536373839404142434445
8. Social Capital and Physical Health 167
Book_Kawachi, Subramanian & Kim_0387713107_proof3_180707
odds ratios after adjustment for individual-level social capital indicators. Sinceperceptions of social cohesion among individuals are arguably shaped by socialcohesion at higher spatial levels, the contextual effect of social cohesion afteradjustment for individual-level variables may be considered “lower bound” esti-mates for the odds ratios and confidence intervals.
Hyppaa & Maki (men), 2001
Hyppaa & Maki (women), 2001
Subramanian et al., 2002
Pollack & Knesebeck, 2004
Veenstra, 2005a
Kim et al., 2006a
Kim et al., 2006b
Poortinga, 2006a
Poortinga, 2006b
Poortinga, 2006c
Poortinga, 2006d
Yip et al., in press
Stud
y A
utho
rs a
nd Y
ear
of P
ublic
atio
n
.3 .4 .5 .6 .7 .8 .9 1 1.5 2
Odds Ratio and 95% Confidence Interval
FIGURE 8.1. Studies of Individual-Level Trust and Fair/Poor Self-Rated Health (Dichotomous)
Subramanian et al., 2002
Poortinga, 2006a
Poortinga, 2006c
Yip et al., in press
Stud
y A
utho
rs a
nd Y
ear
of P
ublic
atio
n
.3 .4 .5 .6 .7 .8 .9 1 1.5 2
Odds Ratio and 95% Confidence Interval
With Adjustment for Individual-Level Social Capital
FIGURE 8.2A. Studies of Area-Level Trust and Fair/Poor Self-Rated Health (Dichotomous)
010203040506070809101112131415161718192021222324252627282930313233343536373839404142434445
168 Kim et al.
Book_Kawachi, Subramanian & Kim_0387713107_proof3_180707
In the multilevel studies, it is also noteworthy that the studies that were null(i.e., with 95% confidence intervals that included the null value) were mainlybased on study samples in relatively more egalitarian countries (for individual-level social trust, in Finland; and for individual-level associational memberships,in Finland, China, and Canada) (Figures 8.1 and 8.3). In the two studies that used
Kawachi et al., 1999
Subramanian et al., 2002
Poortinga, 2006a
Poortinga, 2006c
Yip et al., in press
Stud
y A
utho
rs a
nd Y
ear
of P
ublic
atio
n
.3 .4 .5 .6 .7 .8 .9 1 1.5 2
Odds Ratio and 95% Confidence Interval
Without Adjustment for Individual-Level Social Capital
FIGURE 8.2B. Studies of Area-Level Trust and Fair/Poor Self-Rated Health (Dichotomous)
Hyppaa et al. (men), 2001
Hyppaa et al. (women), 2001
Hyppaa et al., 2003
Lindstrom, 2004
Pollack & Kneseback, 2004
Veenstra, 2005a
Kim et al., 2006b
Poortinga, 2006a
Poortinga, 2006b
Poortinga, 2006c
Poortinga, 2006d
Yip et al., in press
Stud
y A
utho
rs a
nd Y
ear
of P
ublic
atio
n
.3 .4 .5 .6 .7 .8 .9 1 1.5 2
Odds Ratio and 95% Confidence Interval
FIGURE 8.3. Studies of Individual-Level Associational Memberships and Fair/Poor Self-Rated Health (Dichotomous)
010203040506070809101112131415161718192021222324252627282930313233343536373839404142434445
8. Social Capital and Physical Health 169
Book_Kawachi, Subramanian & Kim_0387713107_proof3_180707
composite indices constructed from multiple social capital indicators (Kim &Kawachi, 2006b; Mellor & Milyo, 2005), significant associations were found,and were stronger than for any given subscale in the study by Kim & Kawachi(2006b), suggesting that measurement error in studies that utilized single-itemmeasures of social cohesion may have downwardly biased the effect estimates.
Poortinga, 2006a
Poortinga, 2006c
Yip et al., in press
Stud
y A
utho
rs a
nd Y
ear
of P
ublic
atio
n
.3 .4 .5 .6 .7 .8.91 1.5 2
Odds Ratio and 95% Confidence Interval
With Adjustment for Individual-Level Social Capital
Kawachi et al., 1999
Poortinga, 2006a
Poortinga, 2006c
Yip et al., in press
Stud
y A
utho
rs a
nd Y
ear
of P
ublic
atio
n
.3 .4 .5 .6 .7 .8.9 1 1.5 2
Odds Ratio and 95% Confidence Interval
Without Adjustment for Individual-Level Social Capital
FIGURE 8.4A. Studies of Area-Level Associational Memberships and Fair/Poor Self-RatedHealth (Dichotomous)
FIGURE 8.4B. Studies of Area-Level Associational Memberships and Fair/Poor Self-RatedHealth (Dichotomous)
010203040506070809101112131415161718192021222324252627282930313233343536373839404142434445
170 Kim et al.
Book_Kawachi, Subramanian & Kim_0387713107_proof3_180707
8.4. Social Capital and Cardiovascular Disease
Seven studies of social capital and cardiovascular disease (incidence or mortality)were included in our review (Table 8.3). Two of these studies were multilevel, whilefour were ecological, and one was conducted solely at the individual level. Both ofthe multilevel studies and the individual-level analysis were prospective.
All studies explored the associations for single indicators of social capitalincluding social trust, associational membership, and reciprocity (aggregated tothe area level), as well as the percentage of the labor force with union member-ships. Most ecological studies examined age-standardized cardiovascular mortal-ity rates (spanning all ages, and specific to gender), with one study focusingon cardiovascular mortality in those of middle age (45–64 years). One multilevelanalysis (Blakely et al., 2006) analyzed the risk of mortality from cardiovasculardiseases [i.e., coronary heart disease (CHD) and stroke], while the other multilevelstudy (Sundquist, Johansson, Yang, & Sundquist, 2006) and the individual-levelanalysis (Sundquist, Winkleby, Ahlen, & Johansson, 2004) examined the risk offirst incident non-fatal CHD events requiring hospitalization and fatal CHD.
Adjustment for key potential confounders in ecological studies was variable.Both multilevel studies controlled for multiple individual-level characteristicsincluding age, gender, and income or education. However, control at the area levelwas either absent or confined to area-level socioeconomic deprivation. In ecologi-cal studies, area-level effect estimates were either non-significant (or significant inthe opposite direction, suggesting worse health with higher social cohesion) at thecountry level and in one regional-level study in Australia (Siahpush & Singh,1999). Both multilevel studies found some evidence of modest significant associa-tions between lower electoral participation (Sundquist et al., 2006; OR � 1.19,95% CI � 1.14–1.24 in men; OR � 1.29, 95% CI � 1.21–1.38 in women) andvolunteerism (Blakely et al., 2006; RR � 0.87, 95% CI � 0.75–1.02 in women)and the risk of CVD events, although none of these studies adjusted for individual-level social capital. In an individual-level analysis, Sundquist et al. (2004)observed a moderate and significant association between low social participationand the risk of non-fatal or fatal CVD (OR � 1.74, 95% CI � 1.24–2.43).
8.5. Social Capital and Cancer
Four studies of social capital and cancer met our inclusion criteria (Table 8.4), andoverlapped with studies that looked at cardiovascular disease. Only one of thesestudies was multilevel (and was additionally prospective) (Blakely et al., 2006),with volunteering measured through aggregation of individual-level measures tothe neighborhood level, while the remaining studies were ecological and cross-sectional, investigating social capital in relation to age-standardized cancer mortal-ity rates at the country, state, and regional levels. One of these studies (Lynch et al.,2001) examined mortality rates for cancer at specific sites (lung, prostate, andbreast).
010203040506070809101112131415161718192021222324252627282930313233343536373839404142434445
Book_Kawachi, Subramanian & Kim_0387713107_proof3_180707
EC
OL
OG
ICA
LST
UD
IES:
Cou
ntry
leve
lLy
nch
et a
l., 2
001
16 c
ount
ries
Soci
al c
apita
l m
easu
res:
18+
yH
ealth
out
com
em
easu
res:
All
ages
Soci
al m
istr
ust,
orga
niza
tiona
lm
embe
rshi
ps,
trad
e un
ion
mem
bers
hips
,vo
lunt
eeri
ng
Gen
der-
spec
ific
age-
stan
dard
ized
mor
talit
y ra
tes
for
each
of
hear
t dis
-ea
se a
nd s
trok
e
GD
P pe
r ca
pita
;an
alys
es s
trat
ifie
dby
gen
der
–1)
Soc
ial m
istr
ust:
Hea
rt d
isea
ser
��
0.63
, p �
0.02
(m
en)
r �
�0.
61, p
�0.
03 (
wom
en)
Stro
ker
��
0.29
, p �
0.33
(w
omen
)r
��
0.15
, p �
0.62
(m
en)
2) O
rgan
izat
iona
l mem
bers
hips
:H
eart
dis
ease
r �
0.30
, p �
0.35
(w
omen
)r
�0.
36, p
�0.
25 (
men
)St
roke
r �
0.02
, p �
0.95
(w
omen
);r
��
0.08
, p �
0.81
(m
en)
3) T
rade
uni
on m
embe
rshi
ps:
Hea
rt d
isea
ser
�0.
46, p
�0.
11 (
wom
en)
r �
0.53
, p �
0.06
(m
en)
Stro
ker
�0.
31, p
�0.
29 (
wom
en);
r �
0.31
, p �
0.30
(m
en)
4) V
olun
teer
ing:
Hea
rt d
isea
ser
��
0.14
, p �
0.67
(w
omen
)r
��
0.11
, p �
0.74
(m
en)
TA
BL
E8.
3. S
ocia
l cap
ital a
nd c
ardi
ovas
cula
r di
seas
e.
Sam
ple
size
,A
utho
rs,
popu
latio
n/So
cial
cap
ital
Hea
lth o
utco
me
Indi
vidu
al-l
evel
Are
a-le
vel
year
setti
ngA
ge r
ange
mea
sure
mea
sure
Cov
aria
tes
effe
ct e
stim
ate
effe
ct e
stim
ate
(Con
tinu
ed)
010203040506070809101112131415161718192021222324252627282930313233343536373839404142434445
Book_Kawachi, Subramanian & Kim_0387713107_proof3_180707
Stat
e or
reg
iona
lle
vel
Siah
push
&Si
ngh,
199
9
Ken
nedy
et a
l., 1
998
Nei
ghbo
rhoo
d le
vel
Loc
hner
et a
l., 2
003
7 st
ates
/terr
itori
es in
Aus
tral
ia in
eac
hof
7 y
ears
(n
�49
)
40 r
egio
ns in
Rus
sia
342
neig
hbor
hood
sin
Chi
cago
in th
eU
S
Soci
al c
apita
l m
easu
res:
15�
yH
ealth
out
com
em
easu
res:
All
ages
Soci
al c
apita
l mea
s-ur
e: 1
6�y
Hea
lth o
utco
me
mea
sure
s: A
llag
es
Soci
al c
apita
l m
easu
res:
18�
y
Perc
enta
ge o
f la
bor
forc
e w
ith u
nion
mem
bers
hips
Mis
trus
t in
loca
l and
in r
egio
nal g
ov-
ernm
ent,
lack
of
soci
al c
ohes
ion
atw
ork,
lack
of
inte
rest
in p
oliti
cs
Tru
st, r
ecip
roci
ty,
asso
ciat
iona
lm
embe
rshi
ps
Age
-sta
ndar
dize
dm
orta
lity
rate
s fo
rea
ch o
f he
art d
is-
ease
and
str
oke
Gen
der-
spec
ific
age-
stan
dard
ized
card
iova
scul
ardi
seas
e m
orta
lity
rate
s
Gen
der
and
race
/et
hnic
ity s
peci
fic
Cal
enda
r ye
ar
Per
capi
ta in
com
e,pr
opor
tion
inpo
vert
y, p
er-
ceiv
ed e
cono
mic
hard
ship
inre
gion
, per
cap
itacr
ime
rate
; ana
ly-
ses
stra
tifie
d by
gend
er
Soci
oeco
nom
ic d
ep-
riva
tion;
ana
lyse
s
– – –
Stro
ker
��
0.55
, p �
0.08
(w
omen
)r
��
0.60
, p �
0.05
(m
en)
Hea
rt d
isea
se:
ß >
0, p
< 0
.05
Stro
ke:
ß >
0, p
< 0
.05
1) M
istr
ust i
n lo
cal g
over
nmen
tß
> 0
, p <
0.0
1 (m
en)
ß >
0, p
�0.
02 (
wom
en)
2) M
istr
ust i
n re
gion
algo
vern
men
tß
> 0
, p �
0.01
(m
en)
ß >
0, p
�0.
04 (
wom
en)
3) L
ack
of s
ocia
l coh
esio
n at
wor
kß
> 0
, p �
0.58
(m
en)
ß >
0, p
�0.
72 (
wom
en)
4) L
ack
of in
tere
st in
pol
itics
ß >
0, p
�0.
046
(men
) ß
> 0
, p �
0.10
(w
omen
)
1) T
rust
:W
hite
wom
enß
< 0
, p <
0.0
5
TA
BL
E8.
3. (
Con
tinu
ed)
Sam
ple
size
,A
utho
rs,
popu
latio
n/So
cial
cap
ital
Hea
lth o
utco
me
Indi
vidu
al-l
evel
Are
a-le
vel
year
setti
ngA
ge r
ange
mea
sure
mea
sure
Cov
aria
tes
effe
ct e
stim
ate
effe
ct e
stim
ate
010203040506070809101112131415161718192021222324252627282930313233343536373839404142434445
Book_Kawachi, Subramanian & Kim_0387713107_proof3_180707
MU
LTIL
EV
EL
STU
DIE
S:N
eigh
borh
ood-
or
regi
onal
-lev
el s
ocia
lca
pita
lSu
ndqu
ist e
t al.,
2006
1,35
8,93
2 m
en a
nd1,
446,
747
wom
enag
ed 4
5–74
yea
rsin
9,6
67 s
mal
l
Hea
lth o
utco
me
mea
sure
s: 4
5–64
y
Soci
al c
apita
l mea
s-ur
e: 1
8�y
Hea
lth o
utco
me
mea
sure
s: 4
5–74
y
Are
a le
vel:
loca
l ele
ctor
alpa
rtic
ipat
ion
hear
t dis
ease
mor
talit
y ra
tes
Firs
t hos
pita
lizat
ion
for
a fa
tal o
r no
n-fa
tal c
oron
ary
hear
t dis
ease
even
t
stra
tifie
d by
rac
e/et
hnic
ity a
nd
gend
er
Indi
vidu
al le
vel:
age,
cou
ntry
of
birt
h, m
arita
l sta
-tu
s, e
duca
tion,
hous
ing
tenu
re;
–
Whi
te m
enß
< 0
, p <
0.0
1B
lack
wom
enß
< 0
, p >
0.0
5B
lack
men
ß <
0, p
> 0
.05
2) R
ecip
roci
ty:
Whi
te w
omen
ß <
0, p
> 0
.05
Whi
te m
enß
< 0
, p <
0.0
5B
lack
wom
enß
> 0
, p >
0.0
5B
lack
men
ß <
0, p
> 0
.05
3) A
ssoc
iati
onal
mem
bers
hips
:W
hite
wom
enß
< 0
, p <
0.0
1W
hite
men
ß <
0, p
< 0
.01
Bla
ck w
omen
ß <
0, p
> 0
.05
Bla
ck m
enß
< 0
, p >
0.0
5
Low
ele
ctor
al p
artic
ipat
ion:
OR
�1.
19, 9
5% C
I �
1.14
–1.2
4 (m
en)
OR
�1.
29, 9
5% C
I �
1.21
–1.3
8 (w
omen
) (Con
tinu
ed)
010203040506070809101112131415161718192021222324252627282930313233343536373839404142434445
Book_Kawachi, Subramanian & Kim_0387713107_proof3_180707
Bla
kely
et a
l., 2
006
IND
IVID
UA
L-
LE
VE
L
STU
DIE
S:Su
ndqu
ist e
t al.,
2004
area
mar
ket s
ta-
tistic
s in
Sw
eden
A
ll 25
–74
year
-old
sin
1,6
83 C
ensu
sar
ea u
nits
in N
ewZ
eala
nd
6,86
1 m
en a
ndw
omen
in
Swed
en
Soci
al c
apita
l m
easu
re: 1
5�y
Hea
lth o
utco
me
mea
sure
: 25–
74 y
Soci
al c
apita
l &he
alth
out
com
em
easu
res:
35–
74 y
Vol
unte
erin
g
Soci
al p
artic
ipat
ion
(der
ived
fro
m 1
8ite
ms
on in
form
also
cial
in
tera
ctio
ns a
ndas
soci
atio
nal
mem
bers
hips
)
Car
diov
ascu
lar
mor
talit
y
Dea
th d
ue to
cor
o-na
ry h
eart
dis
ease
or f
irst
hos
pita
l-iz
atio
n fo
r a
non-
fata
l cor
onar
yhe
art d
isea
seev
ent
anal
yses
str
atif
ied
by g
ende
rIn
divi
dual
leve
l:ag
e, r
ace/
ethn
icity
, mar
ital
stat
us, i
ncom
e,ed
ucat
ion,
car
acce
ss, e
mpl
oy-
men
t sta
tus,
urba
n re
side
nce
Nei
ghbo
rhoo
d le
vel:
soci
oeco
nom
icde
priv
atio
nA
naly
ses
stra
tifie
dby
gen
der
Age
, sex
, edu
catio
n,ho
usin
g te
nure
–
Low
soc
ial
part
icip
atio
n:H
R �
1.74
, 95%
C
I �
1.24
–2.4
3
Low
vol
unte
eris
m:
RR
�1.
00, 9
5%
CI
�0.
90–1
.12
(men
)R
R �
0.87
, 95%
C
I �
0.75
–1.0
2 (w
omen
)
–
TA
BL
E8.
3. (
Con
tinu
ed)
Sam
ple
size
,A
utho
rs,
popu
latio
n/So
cial
cap
ital
Hea
lth o
utco
me
Indi
vidu
al-l
evel
Are
a-le
vel
year
setti
ngA
ge r
ange
mea
sure
mea
sure
Cov
aria
tes
effe
ct e
stim
ate
effe
ct e
stim
ate
010203040506070809101112131415161718192021222324252627282930313233343536373839404142434445
Book_Kawachi, Subramanian & Kim_0387713107_proof3_180707
EC
OL
OG
ICA
LST
UD
IES:
Cou
ntry
leve
lLy
nch
et a
l., 2
001
16 c
ount
ries
Soci
al c
apita
l m
easu
re: 1
8�y
Hea
lth o
utco
me
mea
sure
s: A
llag
es
Soci
al m
istr
ust,
orga
niza
tiona
lm
embe
rshi
ps,
trad
e un
ion
mem
-be
rshi
ps,
volu
ntee
ring
Age
-sta
ndar
dize
dm
orta
lity
rate
s fo
rea
ch o
f lu
ng,
pros
tate
, and
brea
st c
ance
r
GD
P pe
r ca
pita
;an
alys
es s
trat
ifie
dby
gen
der
–1)
Soc
ial m
istr
ust:
Lun
g ca
ncer
r �
�0.
07, p
�0.
83 (
men
)r
��
0.44
, p �
0.13
(w
omen
)Pr
osta
te c
ance
r (m
en)
r �
�0.
16, p
�0.
60B
reas
t can
cer
(wom
en)
r �
�0.
21, p
�0.
492)
Org
aniz
atio
nal m
embe
rshi
ps:
Lun
g ca
ncer
r �
0.33
, p �
0.30
(m
en)
r �
0.17
, p �
0.59
(w
omen
)Pr
osta
te c
ance
r (m
en)
r �
0.48
, p �
0.12
Bre
ast c
ance
r (w
omen
)r
�0.
37, p
�0.
233)
Tra
de u
nion
mem
bers
hips
:L
ung
canc
err
��
0.34
, p �
0.26
(m
en)
r �
�0.
06, p
�0.
84 (
wom
en)
Pros
tate
can
cer
(men
)r
�0.
52, p
�0.
07B
reas
t can
cer
(wom
en)
r �
0.20
, p �
0.50
4) V
olun
teer
ing:
Lun
g ca
ncer
r �
0.27
, p �
0.43
(m
en)
TA
BL
E8.
4. S
ocia
l cap
ital a
nd c
ance
r.
Sam
ple
size
,A
utho
rs,
popu
latio
n/So
cial
cap
ital
Hea
lth o
utco
me
Indi
vidu
al-l
evel
Are
a-le
vel
year
setti
ngA
ge r
ange
mea
sure
mea
sure
Cov
aria
tes
effe
ct e
stim
ate
effe
ct e
stim
ate
(Con
tinu
ed)
010203040506070809101112131415161718192021222324252627282930313233343536373839404142434445
Book_Kawachi, Subramanian & Kim_0387713107_proof3_180707
Stat
e or
reg
iona
lle
vel
Siah
push
&
Sing
h, 1
999
Ken
nedy
et a
l., 1
998
7 st
ates
/terr
itori
es in
Aus
tral
ia in
eac
hof
7 y
ears
(n
�49
)
40 r
egio
ns in
Rus
sia
Soci
al c
apita
l m
easu
res:
15�
yH
ealth
out
com
em
easu
res:
All
ages
Soci
al c
apita
l mea
s-ur
e: 1
6�y
Hea
lth o
utco
me
mea
sure
s: A
llag
es
Perc
enta
ge o
f la
bor
forc
e w
ith u
nion
mem
bers
hips
Mis
trus
t in
loca
l and
in r
egio
nal g
ov-
ernm
ent,
lack
of
soci
al c
ohes
ion
atw
ork,
lack
of
inte
rest
in p
oliti
cs
Age
-sta
ndar
dize
dca
ncer
mor
talit
yra
tes
Age
-sta
ndar
dize
dca
ncer
mor
talit
yra
tes
Cal
enda
r ye
ar
Per
capi
ta in
com
e,pr
opor
tion
inpo
vert
y, p
er-
ceiv
ed e
cono
mic
hard
ship
inre
gion
, per
cap
itacr
ime
rate
– –
r �
0.53
, p �
0.10
(w
omen
)Pr
osta
te c
ance
r (m
en)
r �
0.07
, p �
0.84
Bre
ast c
ance
r (w
omen
)r
��
0.22
, p �
0.51
ß >
0, p
< 0
.05
1) M
istr
ust i
n lo
cal g
over
nmen
tß
> 0
, p �
0.06
(m
en)
ß >
0, p
�0.
23 (
wom
en)
2) M
istr
ust i
n re
gion
al g
over
n-m
ent
ß >
0, p
�0.
18 (
men
)ß
> 0
, p �
0.70
(w
omen
)3)
Lac
k of
soc
ial c
ohes
ion
atw
ork
ß >
0, p
�0.
13 (
men
)ß
> 0
, p �
0.89
(w
omen
)4)
Lac
k of
inte
rest
in p
oliti
csß
> 0
, p �
0.2
9 (m
en)
ß >
0, p
�0.
91 (
wom
en)
TA
BL
E8.
4. (
Con
tinu
ed)
Sam
ple
size
,A
utho
rs,
popu
latio
n/So
cial
cap
ital
Hea
lth o
utco
me
Indi
vidu
al-l
evel
Are
a-le
vel
year
setti
ngA
ge r
ange
mea
sure
mea
sure
Cov
aria
tes
effe
ct e
stim
ate
effe
ct e
stim
ate
010203040506070809101112131415161718192021222324252627282930313233343536373839404142434445
Book_Kawachi, Subramanian & Kim_0387713107_proof3_180707
MU
LTIL
EV
EL
STU
DIE
S:N
eigh
borh
ood-
leve
lso
cial
cap
ital
Bla
kely
et a
l., 2
006
All
25-7
4 ye
ar-o
lds
in 1
,683
Cen
sus
area
uni
ts in
New
Zea
land
Soci
al c
apita
l m
easu
re: 1
5�y
Hea
lth o
utco
me
mea
sure
s: 2
5–74
y
Vol
unte
erin
gG
ende
r-sp
ecif
ic a
ll-ca
ncer
mor
talit
yIn
divi
dual
leve
l:ag
e, r
ace/
ethn
icity
, mar
ital
stat
us, i
ncom
e,ed
ucat
ion,
car
acce
ss, e
mpl
oy-
men
t sta
tus,
urba
n re
side
nce
Nei
ghbo
rhoo
d le
vel:
soci
oeco
nom
icde
priv
atio
nA
naly
ses
stra
tifie
dby
gen
der
–L
ow v
olun
teer
ism
:R
R �
0.98
, 95%
C
I �
0.88
–1.1
0 (m
en)
RR
�1.
00, 9
5%
CI
�0.
89–1
.12
(wom
en)
010203040506070809101112131415161718192021222324252627282930313233343536373839404142434445
178 Kim et al.
Book_Kawachi, Subramanian & Kim_0387713107_proof3_180707
As with the health outcomes already reviewed, all studies in this group ana-lyzed associations for single indicators of social cohesion (trust, associationalmembership, and reciprocity), as well as the percentage of the labor force withunion memberships. With the exception of one study that was confined to adults(Blakely et al., 2006), studies examined cancer mortality rates across all agegroups (summarized through age-standardization).
Adjustment for key potential confounders in ecological studies was variable.The single multilevel analysis controlled for multiple individual-level characteris-tics including age, gender, income, and education, as well as neighborhood-levelsocioeconomic deprivation.
As observed for cardiovascular disease, area-level effect estimates were non-significant or significant in the opposite direction (i.e., suggesting increased harmfrom social cohesion) at the country level (e.g., for prostate cancer in Lynch et al.,2001), and at the regional level in Australia (Siahpush & Singh, 1999). However,in contrast to the findings in the regional-level ecological study on social capitaland cardiovascular disease in Russia, associations between social cohesion (e.g.,mistrust in local and regional government) and cancer mortality rates in the samestudy were predominantly non-significant. Likewise, the sole multilevel analysis(Blakely et al., 2006) showed null associations between low neighborhood-levelvolunteerism and individual risk of cancer mortality in women (RR � 1.00, 95%CI � 0.89–1.12), whereas for cardiovascular disease as earlier indicated, it wasmarginally non-significant for women.
8.6. Social Capital and Obesity and Diabetes
We identified only four studies of social capital and obesity or diabetes to date(Table 8.5). One study that examined US state-level social capital in relation toadult obesity and diabetes prevalence rates was ecological (Holtgrave & Crosby,2006), while the remaining studies [one of which was prospective (Kim,Subramanian, Gortmaker, & Kawachi, 2006c)] applied multilevel analysis andexamined social capital in relation to individual-level obesity status (body massindex, BMI, �30 kg/m2).
Studies ranged from those investigating single indicators of social capital, tothose applying indices or scales which combined multiple social capital indica-tors. All studies were based on primarily adult populations.
The only ecological study (Holtgrave & Crosby, 2006) adjusted for the stateproportion in poverty, and found statistically significant inverse associationsbetween the Putnam state-level social capital index and obesity and diabetesprevalence rates (the latter which were not explicitly age-standardized). Themultilevel analyses controlled for several individual-level characteristics includ-ing age, gender, and income and/or education, although only one of these studies(Kim et al., 2006c) controlled for multiple potential contextual confounders. Thatstudy found a modest marginally significant association between higher state-level social capital and lower individual risk of obesity (OR � 0.93, 95% CI �0.85–1.00), but no association for county-level social capital (OR � 0.98,
010203040506070809101112131415161718192021222324252627282930313233343536373839404142434445
Book_Kawachi, Subramanian & Kim_0387713107_proof3_180707
EC
OL
OG
ICA
L
STU
DIE
S:St
ate
leve
lH
oltg
rave
&C
rosb
y, 2
006
MU
LTIL
EV
EL
STU
DIE
S:St
ate-
or
coun
ty-
leve
l soc
ial
capi
tal
Kim
et a
l., 2
006c
48 U
S st
ates
2 sa
mpl
es: 1
01,1
98ad
ults
in 4
13co
untie
s in
48
US
stat
es/
Dis
tric
t of
Col
umbi
a;18
1,20
0 ad
ults
in48
US
stat
es/
Dis
tric
t of
Col
umbi
a
Soci
al c
apita
l &he
alth
out
com
em
easu
res:
18+
y
Soci
al c
apita
l &he
alth
out
com
em
easu
res:
18+
y
Putn
am s
ocia
l ca
pita
l ind
ex(d
eriv
ed f
rom
14
indi
cato
rs)
Cou
nty
leve
l:2
subs
cale
s (b
ased
on f
ive
indi
cato
rs)
corr
espo
ndin
g to
form
al g
roup
and
attit
udin
al/in
for-
mal
soc
ializ
ing
form
s
Obe
sity
and
di
abet
es p
reva
-le
nce
rate
s
Obe
sity
(d
icho
tom
ous)
Prop
ortio
n in
pove
rty
Indi
vidu
al le
vel
(bot
h se
ts o
fan
alys
es):
age,
gen
der,
race
/et
hnic
ity, m
arita
lst
atus
, inc
ome,
educ
atio
nC
ount
y-le
vel
anal
ysis
:St
ate-
leve
l Gin
ico
effi
cien
t and
prop
ortio
n of
Bla
ck r
esid
ents
;co
unty
-lev
elm
ean
hous
ehol
din
com
e
– –
Obe
sity
:ß
< 0
, p �
0.02
Dia
bete
s:ß
< 0
, p <
0.0
1
Cou
nty-
leve
l ana
lysi
s:H
igh
in s
ocia
l cap
ital o
n at
leas
t one
(vs
. nei
ther
) of
the
2 su
bsca
les:
OR
�0.
98, 9
5%C
I �
0.93
-1.0
3St
ate-
leve
l ana
lysi
s:H
igh
in s
ocia
l cap
ital o
n at
leas
t one
(vs
. nei
ther
) of
the
2su
bsca
les:
OR
�0.
93, 9
5%C
I �
0.85
-1.0
0
TA
BL
E8.
5. S
ocia
l cap
ital,
obes
ity, a
nd d
iabe
tes.
Sam
ple
size
,A
utho
rs,
popu
latio
n/So
cial
cap
ital
Hea
lth o
utco
me
Indi
vidu
al-l
evel
Are
a-le
vel
year
setti
ngA
ge r
ange
mea
sure
mea
sure
Cov
aria
tes
effe
ct e
stim
ate
effe
ct e
stim
ate
(Con
tinu
ed)
010203040506070809101112131415161718192021222324252627282930313233343536373839404142434445
Book_Kawachi, Subramanian & Kim_0387713107_proof3_180707
Nei
ghbo
rhoo
d- o
rre
gion
al-l
evel
soc
ial
capi
tal
Poor
tinga
, 200
6b
Vee
nstr
a et
al.,
2005
b
14,8
36 a
dults
in 7
20po
stal
cod
e se
c-to
rs in
the
UK
1,50
4 ad
ults
in th
eci
ty o
f H
amilt
on,
Can
ada
Soci
al c
apita
l &he
alth
out
com
em
easu
re: 1
6�y
Soci
al c
apita
l &he
alth
out
com
em
easu
res:
18�
y
Stat
e le
vel:
2 su
bsca
les
(bas
ed o
n 10
indi
-ca
tors
) co
rres
pon-
ding
to a
ttitu
dina
l/in
form
al
soci
aliz
ing/
form
algr
oup
and
form
alci
vic
and
polit
ical
part
icip
atio
n fo
rms
Indi
vidu
al le
vel:
soci
al s
uppo
rt,
soci
al tr
ust,
civi
c pa
rtic
ipat
ion,
re
cipr
ocity
Mem
bers
hip/
invo
lvem
ent
in v
olun
tary
as
soci
atio
ns
Obe
sity
(d
icho
tom
ous)
Bod
y m
ass
inde
x>
27 k
g/m
2(d
icho
tom
ous)
Stat
e-le
vel a
naly
sis:
Stat
e-le
vel G
ini
coef
fici
ent,
mea
nho
useh
old
inco
me,
pro
por-
tion
of B
lack
re
side
nts
Indi
vidu
al le
vel:
age,
ge
nder
, mar
ital
stat
us, s
ocia
l cla
ss,
unem
ploy
men
tst
atus
, hou
seho
ldte
nure
, acc
ess
toam
eniti
es, p
res-
ence
of l
ocal
soci
al p
robl
ems,
urba
n re
side
nce
Age
, gen
der,
inco
me,
educ
atio
n,
neig
hbor
hood
1) S
ever
e la
ck o
fso
cial
sup
port
: O
R �
1.01
, 95%
C
I �
0.88
–1.1
7 2)
Hig
h so
cial
trus
t: O
R �
0.86
, 95%
C
I �
0.78
–0.9
53)
Hig
h so
cial
part
icip
atio
n:
OR
�1.
01, 9
5%
CI
�0.
90–1
.14
4) H
igh
reci
proc
ity:
OR
�1.
07, 9
5%
CI
�0.
95–1
.19
Hig
her
volu
ntar
yas
soci
atio
nin
volv
emen
t:O
R �
0.91
, p
�0.
03
– –
TA
BL
E8.
5. (
Con
tinu
ed)
Sam
ple
size
,A
utho
rs,
popu
latio
n/So
cial
cap
ital
Hea
lth o
utco
me
Indi
vidu
al-l
evel
Are
a-le
vel
year
setti
ngA
ge r
ange
mea
sure
mea
sure
Cov
aria
tes
effe
ct e
stim
ate
effe
ct e
stim
ate
010203040506070809101112131415161718192021222324252627282930313233343536373839404142434445
8. Social Capital and Physical Health 181
Book_Kawachi, Subramanian & Kim_0387713107_proof3_180707
95% CI � 0.93–1.03). Evidence from the two other studies that applied a multilevel framework was somewhat mixed, with one study observing highindividual-level social trust to be significantly inversely associated with obesityrisk (OR � 0.86, 95% CI � 0.78–0.95), but no associations for other social capitalmeasures (social support, social participation, and reciprocity) (Poortinga, 2006b).Meanwhile, the other study (Veenstra et al., 2005b) found higher voluntary asso-ciation involvement to be significantly associated with a 9% lower risk of a higherbody weight (BMI � 27 kg/m2).
8.7. Social Capital and Infectious Diseases
We identified three studies of social capital and infectious diseases, all of whichwere ecological (Table 8.6). One of these studies was cross-national and cross-sectional (Lynch et al, 2001), while the other two studies were conducted at theUS state level and were prospective (Holtgrave & Crosby, 2003, 2004).
The cross-national study (Lynch et al., 2001) applied single indicators of socialcapital including social trust, organization and trade union membership, and vol-unteering (based on surveys among adults), while the two state-level studiesemployed the Putnam social capital index. All studies included individuals of allages in the calculation of case rates and mortality rates.
The cross-national study (Lynch et al., 2001) adjusted for GDP per capita,stratified the analyses by gender, and controlled for age composition through age-standardization of the mortality rates. Findings from this study were mixed, withnon-significant weak to moderate correlations between each of country-levelsocial mistrust and trade union memberships in the anticipated direction with age-standardized mortality rates from all infectious diseases in men and in women.Associations for organizational memberships in both sexes were null, and therewere weak to moderate positive correlations between volunteering and infectiousdisease mortality rates in men and women, respectively. By contrast, associationsin the two studies that examined the Putnam state social capital index in relationto state case rates from each of gonorrhea, syphilis, Chlamydia, AIDS, and tuber-culosis (controlling for income inequality for the latter two outcomes) were allsignificantly inverse, although neither of these studies controlled for area-levelsocioeconomic deprivation (Holtgrave & Crosby, 2003, 2004).
8.8. Summary and Synthesis
8.8.1. Summary of Findings
Our review of the literature found fairly consistent associations between trust asan indicator of social cohesion and better physical health. The evidence for trustwas stronger for self-rated health than for other physical health outcomes, andstronger for individual-level perceptions than for area-level trust. Associational
010203040506070809101112131415161718192021222324252627282930313233343536373839404142434445
Book_Kawachi, Subramanian & Kim_0387713107_proof3_180707
EC
OL
OG
ICA
L
STU
DIE
S:C
ount
ry le
vel
Lync
h et
al.,
200
1
Stat
e le
vel
Hol
tgra
ve &
Cro
sby,
200
3
Hol
tgra
ve &
Cro
sby,
200
4
16 c
ount
ries
48 U
S st
ates
48 U
S st
ates
Soci
al c
apita
l m
easu
re: 1
8�y
Hea
lth o
utco
me
mea
sure
s: A
llag
es
Soci
al c
apita
l m
easu
re: 1
8�y
Hea
lth o
utco
me
mea
sure
s: A
llag
es
Soci
al c
apita
l m
easu
re: 1
8�y
Hea
lth o
utco
me
mea
sure
s: A
llag
es
Soci
al m
istr
ust,
orga
niza
tiona
lm
embe
rshi
ps,
trad
e un
ion
mem
bers
hips
,vo
lunt
eeri
ng
Putn
am s
ocia
l cap
i-ta
l ind
ex
(der
ived
fro
m 1
4in
dica
tors
)
Putn
am s
ocia
l ca
pita
l ind
ex(d
eriv
ed f
rom
14
indi
cato
rs)
Gen
der-
spec
ific
,ag
e-st
anda
rdiz
edm
orta
lity
rate
s fo
ral
l inf
ectio
us
dise
ases
Gon
orrh
ea, s
yphi
lis,
chla
myd
ia, A
IDS
case
rat
es
Tub
ercu
losi
s ca
sera
tes
GD
P pe
r ca
pita
;an
alys
es s
trat
ifie
dby
gen
der
Inco
me
ineq
ualit
y(f
or a
naly
sis
ofA
IDS
case
rat
eson
ly)
Inco
me
ineq
ualit
y
– – –
1) S
ocia
l mis
trus
t:r
�0.
30, p
�0.
32 (
men
)r
�0.
26, p
�0.
39 (
wom
en)
2) O
rgan
izat
iona
l mem
bers
hips
:r
��
0.06
, p �
0.85
(m
en)
r �
0.01
, p �
0.96
(w
omen
)3)
Tra
de u
nion
mem
bers
hips
:r
��
0.42
, p �
0.16
(m
en)
r �
�0.
39, p
�0.
19 (
wom
en)
4) V
olun
teer
ing:
r �
0.24
, p �
0.48
(m
en)
r �
0.33
, p �
0.32
(w
omen
)
Gon
orrh
ea c
ase
rate
s:r
��
0.67
, p <
0.0
1Sy
phili
s ca
se r
ates
: r
��
0.59
, p <
0.0
1C
hlam
ydia
cas
e ra
tes:
r �
�0.
53, p
< 0
.01
AID
S ca
se r
ates
:ß
< 0
, p �
0.01
ß <
0, p
< 0
.01
TA
BL
E8.
6. S
ocia
l cap
ital a
nd in
fect
ious
dis
ease
s.
Sam
ple
size
,A
utho
rs,
popu
latio
n/So
cial
cap
ital
Hea
lth o
utco
me
Indi
vidu
al-l
evel
Are
a-le
vel
year
setti
ngA
ge r
ange
mea
sure
mea
sure
Cov
aria
tes
effe
ct e
stim
ate
effe
ct e
stim
ate
membership as an indicator of cohesion was also consistently associated with bet-ter self-rated health at the individual level, although reverse causation cannot beexcluded (see discussion below). On the other hand, the evidence was weak thatassociational membership at the area level is associated with self-rated health (ineither direction).
8.8.2. Social Cohesion in Egalitarian versus Inegalitarian Social Contexts
In a recent systematic review of forty-two published studies, Islam, Merlo,Kawachi, Lindstrom, & Gerdtham, (2006a) found that an association betweensocial capital and health was much more consistently reported in inegalitariancountries i.e., countries with a high degree of economic inequality; whereas anassociation was either not observed or was much weaker in more egalitariansocieties. Economic inequality was assessed by the country’s Gini coefficient(based on disposable income) and by the country’s public share of social expen-diture. Regardless of the type of study (individual, ecological, or multilevel) orthe country’s degree of egalitarianism, the authors found generally significantpositive associations between social capital and better health outcomes.
Moreover, from the multilevel studies that were identified in this review byIslam et al. (2006a), there was also evidence to suggest that the between-areavariation in health (i.e., the random effect) was considerably lower in more egali-tarian countries (such as Canada and Sweden) as compared to more unequalcountries (such as the United States). For example, the intraclass correlation(ICC, corresponding to the percent of variation in health explained at the arealevel) was approximately 7.5% in a US study of neighborhood influences on vio-lent crime and homicide, whereas the ICCs ranged from 0–2% for studies inCanada and Sweden (Islam et al., 2006a). Likewise, a recent multilevel analysisof 275 Swedish municipalities found a modest fixed effect association betweenvoting participation and health-related quality of life, with 98% of variation inhealth attributed to the individual level, and only 2% to the municipality level(Islam et al., 2006b).
One potential explanation for this pattern (of generally null findings from mul-tilevel studies of social capital and self-rated health in more egalitarian countries)is that in egalitarian societies characterized by strong provision of safety nets andspending on public goods (such as health care, education, unemployment insur-ance), social capital may be less salient for the health of its residents, by contrastto highly unequal and segregated societies such as the United States.
8.8.3. Limitations of Studies
Our review of the literature has highlighted increasing methodological sophistica-tion in study design over time, progressing from the earlier ecological studies ofsocial cohesion and health, to the more recent multilevel study designs. Nonethe-less, our review also points to a number of gaps in the existing literature. As the
010203040506070809101112131415161718192021222324252627282930313233343536373839404142434445
8. Social Capital and Physical Health 183
Book_Kawachi, Subramanian & Kim_0387713107_proof3_180707
tables demonstrate, many studies continue to rely on secondary sources of data toconstruct “indicators” of social cohesion. As pointed out by Harpham in chapter 3,proxy indicators of social cohesion – such as trade union membership, volunteer-ing, and social participation – can be construed as either precursors or consequencesof social capital, but they are not part of social capital per se. Accordingly, there isan urgent need to incorporate direct measures of social cohesion into existingnational surveys, taking care to specify the scale of measurement (e.g., neighbor-hoods) as well as making sure to include relevant distinctions such as bondingversus bridging capital, or cognitive versus behavioral measures (see chapter 3 forfurther tips).
Virtually none of the studies have distinguished between the effects of bond-ing versus bridging capital, and few studies have explicitly sought to examinethe deleterious consequences of social cohesion through careful analyses ofcross-level interactions between community cohesion and individual charac-teristics. As the multilevel analysis by Subramanian, Kim, and Kawachi,(2002) suggests, community cohesion can be beneficial for some groups, yetcan be harmful to the health of others. Studies have also been inconsistent withrespect to controlling for potential confounding variables at both the individualand area levels.
Aside from the threat of omitted variable bias, one of the biggest challenges forestablishing causality in this area remains the paucity of longitudinal data. Cross-sectional data are less than ideal for establishing causality. For example, at theindividual level, one could argue that being in good health is a precursor of hav-ing trusting opinions of others, or participating in civic associations (i.e., reversecausation). Ideally, what is needed are data with repeated assessments of bothsocial cohesion and health outcomes; in other words, data of the type that wouldlend itself to analytical strategies such as “difference-in-difference” (DiD) esti-mators (Ashenfelter, 1978; Ashenfelter & Card, 1985). The other major criticismof the research to date is that no studies have adequately dealt with the potentialproblem that community cohesion is endogenous (Kawachi, 2006). For example,some people are likely to choose the communities they live in based on their pref-erences for social interactions with neighbors. To the extent that such preferencesare also correlated with health, we have an endogeneity problem. Solving theendogeneity problem will require study designs in which the exposure (socialcapital) can be manipulated through either natural experiments (instruments) orrandomization (e.g., cluster community trials) (Oakes, 2004) (see also chapter 7fur further discussion of these issues).
8.8.4. Examining Social Capital in Diverse Populations
While many existing studies have sampled populations across a wide range of ages,the investigation of specific effects among elderly populations (e.g., persons over age65) and among children and adolescents (for which behaviors may be more mal-leable; Dietz & Gortmaker, 2001) has been sparse (Drukker, Kaplan, Feron, & vanOs, 2003; Drukker, Buka, Kaplan, McKenzie,& van Os, 2005; Wen et al., 2005).
010203040506070809101112131415161718192021222324252627282930313233343536373839404142434445
184 Kim et al.
Book_Kawachi, Subramanian & Kim_0387713107_proof3_180707
Populations in developing countries further represent an uncharted territory of inves-tigation of the physical health effects of social capital, for which the associationsmight potentially differ due to vastly different political economies, sociocultural con-texts, and patterns of disease than in developed nations.
8.8.5. Mechanisms Linking Social Capital to Physical Health
Although few studies have sought to directly assess the mechanisms linkingsocial capital to health, a variety of hypothesized pathways have been proposedby which cohesion may affect health, including the diffusion of knowledgeabout health promotion, maintenance of healthy behavioral norms throughinformal social control, promotion of access to local services and amenities,and psychosocial processes which provide affective support and mutual respect(Kawachi & Berkman, 2000). These mechanisms could broadly be catego-rized into local behaviorally-mediated mechanisms, and more upstream policy-mediated mechanisms.
On the behavioral front, drawing on the diffusion of innovations theory(Rogers, 2003), we may posit that residents of high social capital neighborhoodsor regions in which healthy behaviors (e.g., engagement in exercise and avoid-ance of foods high in saturated fats) are practiced among some residents may bemore likely to adopt these behaviors through diffusion of knowledge about thebehaviors.
At larger geographical scales (e.g., the county, state, or regional level), socialcapital might also conceivably affect physical health through policy-relatedmechanisms. In his seminal work Making Democracy Work (Putnam, 1993) thepolitical scientist Robert Putnam lends empirical credence to the notion that pros-perous democracies are tied to the presence of civic engagement and social capi-tal. Within the health context, it has been hypothesized that more cohesivesocieties are more apt to cooperate in the provision of health-promoting publicgoods for its residents, such as health care (see also Introduction and chapter 7).Social cohesion at other scales might have contextual effects on individual levelsof social capital through attitudinal/cognitive mechanisms. For instance, trans-parency and the absence of corruption increase public confidence in governmen-tal institutions, which in turn may raise levels of interpersonal trust (Brehm &Rahn, 1997; Levi, 1996).
A number of behavioral risk factors have been established for chronic diseasessuch as cardiovascular diseases (coronary heart disease and stroke), selected cancers(e.g., colon cancer, lung cancer, breast cancer), and diabetes. Several of these riskfactors (e.g., dietary intakes, smoking, and physical inactivity) have themselves beenlinked to community cohesion (see chapter 10 by Lindström). Psychosocial factors(e.g., depression, anxiety) may also affect disease risk, either through directpathways (e.g., through psycho-neuro-immune effects) or indirect pathways(e.g., mediated by behavioral changes), and are putative risk factors for heart disease(Kubzansky & Kawachi, 2000; Kuper, Marmot, & Hemingway, 2002), and to a lesserextent, for cancers and infectious diseases (Cohen, Alper, Doyle, Treanor, & Turner,
010203040506070809101112131415161718192021222324252627282930313233343536373839404142434445
8. Social Capital and Physical Health 185
Book_Kawachi, Subramanian & Kim_0387713107_proof3_180707
2006; Kroenke et al., 2005; Leonard, 2000). Of course, social cohesion can alsoplausibly contribute to greater transmission of infectious diseases through higherperson-to-person contact (Holtgrave & Crosby, 2003).
8.9. Conclusions
The past decade has borne witness to a flourishing epidemiologic and publichealth interest in the investigation of the effects of social capital on physicalhealth outcomes. This inquiry has broadened from an emphasis on overall mortal-ity and self-rated health to include more specific disease diagnoses. Our review ofthe literature to date suggests several points of convergence – for example, themore consistent associations between social cohesion and health in unequalsocieties with weak safety nets compared to egalitarian countries with a strongtradition of public goods provision; the stronger associations between health andtrust (as an indicator of cohesion) compared to associational membership;and stronger associations for the same indicator at the individual compared tocollective level. At the same time, our review also points to several gaps thatthe next generation of research needs to address, in particular, stronger studydesigns that address questions of causality, and deepen our understanding ofcausal mechanisms.
ReferencesAshenfelter, O. (1978). Estimating the effect of training programs on earnings. Review of
Economics and Statistics, 60, 47–57.Ashenfelter, O., & Card, D. (1985). Using the longitudinal structure of earnings to estimate
the effect of training programs. Review of Economics and Statistics, 67, 648–660.Bearman, P., & Moody, J. (2004). Suicide and friendships among American adolescents.
American Journal of Public Health, 94, 89–95.Berkman, L. F., & Glass, T. (2000). Social integration, social networks, social support, and
health. In L. F. Berkman & I. Kawachi (Eds.), Social Epidemiology (pp. 137–173). NewYork, NY: Oxford University Press.
Blakely, T., Atkinson, J., Ivory, V., Collings, S., Wilton, J., & Howden-Chapman, P.(2006). No association of neighbourhood volunteerism with mortality in New Zealand:a national multilevel cohort study. International Journal of Epidemiology, 35, 981–989.
Brehm, J., & Rahn, W. (1997). Individual-level evidence for the causes and consequencesof social capital. American Journal of Political Science, 41, 999–1023.
Browning, C. R., & Cagney, K. A. (2003). Moving beyond poverty: neighborhood struc-ture, social processes, and health. Journal of Health & Social Behavior, 44, 552–571.
Cohen, S., Alper, C. M., Doyle, W. J., Treanor, J. J., & Turner, R. B. (2006). Positive emo-tional style predicts resistance to illness after experimental exposure to rhinovirus orinfluenza a virus. Psychosomatic Medicine, 68, 809–815.
Dietz, W. H., & Gortmaker, S. L. (2001). Preventing obesity in children and adolescents.Annual Review of Public Health, 22, 337–353.
Drukker, M., Kaplan, C., Feron, F., & van Os, J. (2003). Children’s health-related quality oflife, neighborhood socioeconomic deprivation and social capital. A contextual analysis.Social Science & Medicine, 7, 825–841.
010203040506070809101112131415161718192021222324252627282930313233343536373839404142434445
186 Kim et al.
Book_Kawachi, Subramanian & Kim_0387713107_proof3_180707
Drukker, M., Buka, S. L., Kaplan, C., McKenzie, K., & van Os, J. (2005). Social capitaland young adolescents’ perceived health in different sociocultural settings. Social Sci-ence & Medicine, 61, 185–198.
Franzini, L., Caughy, M., Spears, W., & Fernandez Esquer, M. E. (2005). Neighborhoodeconomic conditions, social processes, and self-rated health in low-income neighbor-hoods in Texas: A multilevel latent variables model. Social Science & Medicine, 61,1135–1150.
Friedman, S. R., & Aral, S. (2001). Social networks, risk-potential networks, health, anddisease. Journal of Urban Health, 78, 411–418.
Helliwell, J. F., & Putnam, R. D. (2004). The social context of well-being. Proceedings ofthe Royal Society B: Biological Sciences, 359, 1435–1446.
Holtgrave, D. R., & Crosby, R. A. (2003). Social capital, poverty, and income inequality aspredictors of gonorrhoea, syphilis, chlamydia and AIDS in the United States. SexuallyTransmitted Infections, 79, 62–64.
Holtgrave, D. R., & Crosby, R. A. (2004). Social determinants of tuberculosis case rates inthe United States. American Journal of Preventive Medicine, 26, 159–162.
Holtgrave, D. R., & Crosby, R. (2006). Is social capital a protective factor against obesityand diabetes? Findings from an exploratory study. Annals of Epidemiology, 16,406–408.
Hyyppä, M. T., & Mäki, J. (2001). Individual-level relationships between social capitaland self-rated health in a bilingual community. Preventive Medicine, 32, 148–155.
Hyyppä, M. T., & Mäki, J. (2003). Social participation and health in a community rich instock of social capital. Health Education Research, 18, 770–779.
Islam, M. K., Merlo, J., Kawachi, I., Lindstrom, M., & Gerdtham, U-G. (2006a). Socialcapital and health: Does egalitarianism matter? A literature review. International Jour-nal for Equity in Health, 5, 3. doi:10.1186/1475–9276–5–3.
Islam, M. K., Merlo, J., Kawachi, I., Lindstrom, M., Burstrom, K., & Gerdtham, U-G.(2006b). Does it really matter where you live? A panel data multilevel analysis ofSwedish municipality level social capital on individual health-related quality of life.Health Economics, Policy and Law, 1, 209–235.
Kavanagh, A. M., Turrell, G., & Subramanian, S. V. (2006). Does area-based social capitalmatter for the health of Australians? A multilevel analysis of self-rated health in Tasma-nia. International Journal of Epidemiology, 35, 607–613.
Kavanagh, A. M., Bentley, R., Turrell, G., Broom, D. H., & Subramanian, S. V. (2006b).Does gender modify associations between self rated health and the social and economiccharacteristics of local environments? Journal of Epidemiology & Community Health,60, 490–495.
Kawachi, I., Kennedy, B., Lochner, K., & Prothrow-Stith, D. (1997). Social capital,income inequality, and mortality. American Journal of Public Health, 87, 1491–1498.
Kawachi, I., Kennedy, B., & Glass, R. (1999). Social capital and self-rated health: A con-textual analysis. American Journal of Public Health, 89, 1187–1193.
Kawachi, I., & Berkman, L. F. (2000). Social cohesion, social capital, and health. In L.F.Berkman & I. Kawachi (Eds.), Social Epidemiology (pp. 178–190). New York, NY:Oxford University Press.
Kawachi, I. (2006). Commentary: Social capital and health – making the connections onestep at a time. International Journal of Epidemiology, 35, 989–993.
Kennedy, B., Kawachi, I., & Brainerd, E. (1998). The role of social capital in the Russianmortality crisis. World Development, 26, 2029–2043.
Kennelly, B., O’Shea, E., & Garvey, E. (2003). Social capital, life expectancy and mortal-ity: a cross-national examination. Social Science & Medicine, 56, 2367–2377.
010203040506070809101112131415161718192021222324252627282930313233343536373839404142434445
8. Social Capital and Physical Health 187
Book_Kawachi, Subramanian & Kim_0387713107_proof3_180707
Kim, D., Subramanian, S. V., & Kawachi, I. (2006a). Bonding versus bridging socialcapital and their associations with self-rated health: a multilevel analysis of 40U.S. communities. Journal of Epidemiology & Community Health, 60, 116–122.
Kim, D., & Kawachi, I. (2006b). A multilevel analysis of key forms of community- andindividual-level social capital as predictors of self-rated health in the United States.Journal of Urban Health, 83, 813–826.
Kim, D., Subramanian, S. V., Gortmaker, S. L., & Kawachi, I. (2006c). US state- andcounty-level social capital in relation to obesity and physical inactivity: a multilevel,multivariable analysis. Social Science & Medicine, 63, 1045–1059.
Kroenke, C.H., Bennett, G.G., Fuchs, C., Giovannucci, E., Kawachi, I., Schernhammer, E.,Holmes, M.D., & Kubzansky, L.D., (2005). Depressive symptoms and prospective inci-dence of colorectal cancer in women. American Journal of Epidemiology, 162, 839-48.
Kubzansky, L., & Kawachi, I. (2000). Affective states and health. In L. F. Berkman & I. Kawachi (Eds.), Social epidemiology (pp. 213–41). New York, NY: Oxford UniversityPress.
Kuper, H., Marmot, M., & Hemingway, H. (2002). Systematic review of prospectivecohort studies of psychosocial factors in the etiology and prognosis of coronary heartdisease. Seminars in Vascular Medicine, 2, 267–314.
Leonard, B. (2000). Stress, depression and the activation of the immune system. WorldJournal of Biology & Psychiatry, 1, 17–25.
Levi, M. (1996). Social and unsocial capital: A review essay of Robert Putnam’s MakingDemocracy Work. Politics and Society, 24, 45–55.
Lindström, M., Moghaddassi, M., & Merlo, J. (2004). Individual self-reported health,social participation and neighborhood: A multilevel analysis in Malmö, Sweden. Pre-ventive Medicine, 39, 135–141.
Lynch, J. W., Davey Smith, G., Hillemeier, M. M., Shwa, M., Raghunathan, T., & Kaplan,G. A. (2001). Income inequality, the psychosocial environment and health: comparisonsof wealthy nations. Lancet, 358, 194–200.
Lochner, K., Kawachi, I., Brennan, R. T., & Buka, S. L. (2003). Social capital and neigh-borhood mortality rates in Chicago. Social Science & Medicine, 56, 1797–1805.
Mellor, J. M., & Milyo, J. (2005). State social capital and individual health status. Journalof Health Politics, Policy & Law, 30, 1101–1130.
Milyo, J., & Mellor, J. M. (2003). On the importance of age-adjustment methods in eco-logical studies of social determinants of mortality. Health Services Research, 38,1781–1790.
Mohan, J., Twigg, L., Barnard, S., & Jones, K. (2005). Social capital, geography andhealth: a small-area analysis for England. Social Science & Medicine, 60, 1267–1283.
Oakes, J. M. (2004). The (mis)estimation of neighborhood effects: Causal inference for apracticable social epidemiology. Social Science & Medicine, 58, 1929–1952.
Pollack, C. E., & von dem Knesebeck, O. (2004). Social capital and health among theaged: comparisons between the United States and Germany. Health & Place, 10,383–391.
Poortinga, W. (2006a). Social capital: An individual or collective resource for health?Social Science & Medicine, 62, 292–302.
Poortinga, W. (2006b). Perceptions of the environment, physical activity, and obesity.Social Science & Medicine, 63, 2835–2846.
Poortinga, W. (2006c). Social relations or social capital? Individual and community healtheffects of bonding social capital. Social Science & Medicine, 63, 255–270.
010203040506070809101112131415161718192021222324252627282930313233343536373839404142434445
188 Kim et al.
Book_Kawachi, Subramanian & Kim_0387713107_proof3_180707
Poortinga, W. (2006d). Do health behaviors mediate the association between social capitaland health? Preventive Medicine, 43, 488–493.
Putnam, R. D. (1993). Making democracy work. Princeton, NJ: Princeton University Press.Putnam, R. D. (2000). Bowling alone: The collapse and revival of American community.
New York: Simon and Schuster.Rojas, Y., & Carlson, P. (2006). The stratification of social capital and its consequences for
self-rated health in Taganrog, Russia. Social Science & Medicine, 62, 2732–2741.Rogers, E. M. (2003). Diffusion of innovations (5th ed.). New York: The Free Press.Rose, R. (2000). How much does social capital add to individual health? A survey of
Russians. Social Science & Medicine, 51, 1421–1435.Siahpush, M., & Singh, G. K. (1999). Social integration and mortality in Australia.
Australia & New Zealand Journal of Public Health, 23, 571–577.Skrabski, A., Kopp, M., & Kawachi, I. (2003). Social capital in a changing society: Cross
sectional associations with middle aged female and male mortality. Journal of Epidemi-ology & Community Health, 57, 114–119.
Skrabski, A., Kopp, M., & Kawachi, I. (2004). Social capital and collective efficacy inHungary: Cross sectional associations with middle aged female and male mortalityrates. Journal of Epidemiology & Community Health, 58, 340–345.
Steptoe, A., & Feldman, P. J. (2001). Neighborhood problems as sources of chronic stress:development of a measure of neighborhood problems, and associations with socioeco-nomic status and health. Annals of Behavioral Medicine, 23, 177–185.
Subramanian, S. V., Kawachi, I., & Kennedy, B. P. (2001). Does the state you live in makea difference? Multilevel analysis of self-rated health in the US. Social Science & Medi-cine, 53, 9–19.
Subramanian, S. V., Kim, D. J., & Kawachi, I. (2002). Social Trust and Self-Rated Healthin US Communities: A Multilevel Analysis. Journal of Urban Health: Bulletin of theNew York Academy of Medicine, 79, S21–S34.
Sundquist, K., Winkleby, M., Ahlen, H., & Johansson, S. E. (2004). Neighborhood socioe-conomic environment and incidence of coronary heart disease: a follow-up study of25,319 women and men in Sweden. American Journal of Epidemiology, 159, 655–662.
Sundquist, J., Johansson, S. E., Yang, M., & Sundquist, K. (2006). Low linking social cap-ital as a predictor of coronary heart disease in Sweden: a cohort study of 2.8 millionpeople. Social Science & Medicine, 62, 954–963.
Turrell, G., Kavanagh, A., & Subramanian, S. V. (2006). Area variation in mortality in Tas-mania (Australia): The contributions of socioeconomic disadvantage, social capital andgeographic remoteness. Health & Place, 12, 291–305.
Valente, T. W., Gallaher, P., & Mouttapa, M. (2004). Using social networks to understandand prevent substance use: A transdisciplinary perspective. Substance Use & Misuse,39, 1685–1712.
Veenstra, G. (2000). Social capital, SES and health: An individual-level analysis. SocialScience & Medicine, 50, 619–629.
Veenstra, G. (2002). Social capital and health (plus wealth, income inequality and regionhealth governance). Social Science & Medicine, 54, 849–868.
Veenstra, G. (2005a). Location, location, location: Contextual and compositional healtheffects of social capital in British Columbia, Canada. Social Science & Medicine, 60,2059–2071.
Veenstra, G., Luginaah, I., Wakefield, S., Birch, S., Eyles, J., & Elliott, S. (2005b). Whoyou know, where you live: social capital, neighborhood and health. Social Science &Medicine, 60, 2799–2818.
010203040506070809101112131415161718192021222324252627282930313233343536373839404142434445
8. Social Capital and Physical Health 189
Book_Kawachi, Subramanian & Kim_0387713107_proof3_180707
Wanless, D. (2004). Securing good health for the whole population. London, UK: Crown.Wen, M., Browning, C. R., & Cagney, K. A. (2003). Poverty affluence and income
inequality: neighborhood economic structure and its implications for health. Social Sci-ence & Medicine, 57, 843–860.
Wen, M., Cagney, K. A., & Christakis, N. A. (2005). Effect of specific aspects of commu-nity social environment on the mortality of individuals diagnosed with serious illness.Social Science & Medicine, 61, 1119–1134.
Wilkinson, R. G., Kawachi, I., & Kennedy, B. P. (1998). Mortality, the social environment,crime and violence. Sociology of Health & Illness, 20, 578–597.
Yip, W., Subramanian, S. V., Mitchell, A. D., Lee, D. T., Wang, J., & Kawachi, I. (Inpress). Does social capital enhance health and well-being? Evidence from rural China.Social Science & Medicine, 64, 35–49.
Ziersch, A. M., Baum, F. E., Macdougall, C., & Putland, C. (2005). Neighborhood life andsocial capital: the implications for health. Social Science & Medicine, 60, 71–86.
Zimmerman, F. J., & Bell, J. F. (2006). Income inequality and physical and mental health: test-ing associations consistent with proposed causal pathways. Journal of Epidemiology &Community Health, 60, 513–521.
010203040506070809101112131415161718192021222324252627282930313233343536373839404142434445
190 Kim et al.
Book_Kawachi, Subramanian & Kim_0387713107_proof3_180707