Associations between social capital and HIV stigma in Chennai, India: considerations for prevention intervention design.

Abstract

Stigma against persons living with HIV/AIDS (PLHA) is a barrier to seeking prevention education, HIV testing, and care. Social capital has been reported as an important factor influencing HIV prevention and social support upon infection. In the study, we explored the associations between social capital and stigma among men and women who are patrons of wine shops or community-based alcohol outlets in Chennai. We found that reports of social capital indicators were associated with reduced fear of transmission of HIV/AIDS, lower levels of feelings of shame, blame and judgment, lower levels of personal support and perceived community support for discriminatory actions against PLHA. Specifically, when participants reported membership in formal groups, perception of high levels of collective action toward community goals, high norms of reciprocity between neighbors and residents in daily life, and presence of trusted sexually transmitted disease care providers, all levels of measures of stigma were lower. Although we defined social capital rather narrowly in this study, our findings suggest that seeking partnerships with existing organizations and involving health care providers in future interventions may be explored as a strategy in community-based prevention interventions.

Full text

Page 1

AIDS Education and Prevention, 21(3), 233–250, 2009
© 2009 The Guilford Press
233
SIVARAM ET AL.
SOCIAL CAPITAL AND STIGMA IN CHENNAI
associations BEtwEEn social
caPital and Hiv stigma
in cHEnnai, india:
considErations for
PrEvEntion intErvEntion dEsign
Sudha Sivaram, Carla Zelaya, A.K. Srikrishnan,
Carl Latkin, V.F. Go, Suniti Solomon and David Celentano
Stigma against persons living with HIV/AIDS (PLHA) is a barrier to seeking
prevention education, HIV testing, and care. Social capital has been report-
ed as an important factor influencing HIV prevention and social support
upon infection. In the study, we explored the associations between social
capital and stigma among men and women who are patrons of wine shops
or community-based alcohol outlets in Chennai. We found that reports of
social capital indicators were associated with reduced fear of transmission
of HIV/AIDS, lower levels of feelings of shame, blame and judgment, lower
levels of personal support and perceived community support for discrimi-
natory actions against PLHA. Specifically, when participants reported
membership in formal groups, perception of high levels of collective action
toward community goals, high norms of reciprocity between neighbors and
residents in daily life, and presence of trusted sexually transmitted disease
care providers, all levels of measures of stigma were lower. Although we
defined social capital rather narrowly in this study, our findings suggest that
seeking partnerships with existing organizations and involving health care
providers in future interventions may be explored as a strategy in communi-
ty-based prevention interventions.
Worldwide there are over 25 million adults infected with HIV, the virus that causes
AIDS (UNAIDS, 2008). Of these, 2.5 million live in India (National AIDS Control
Organization, 2007). Although concerted prevention education programs have
increased awareness of prevention and transmission in India, there remain several
challenges (Steinbrook, 2007). A key challenge lies in motivating individuals at risk
Sudha Sivaram S, Carla Zelaya, V.F. Go, and David Celentano are with the Infectious Diseases Pro-
gram, Department of Epidemiology, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD.
A.K. Srikrishnan and Suniti Solomon are with the YRG Center for AIDS Research and Education (YRG
CARE), Voluntary Health Services, Taramani, Chennai, India. Carl Latkin is with the Department of
Health, Behavior and Society, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD.
The study was funded by U.S. National Institute of Mental Health Grant U10MH681543-01.
Address correspondence to Sudha Sivaram, Infectious Diseases Program, Department of Epidemiology,
The Johns Hopkins Bloomberg School of Public Health, 615 N. Wolfe Street, Rm E6547, Baltimore, MD
21205; e-mail: ssivarama@jhsph.edu

Page 2

234 SIVARAM ET AL.
to get tested and in managing HIV and treating HIV among those who are infected
(Celentano, 2008). Awareness of testing remains low; counseling and testing servic-
es at antenatal clinics, key points of prevention, based on a report in rural India was
only reported by 3% of women (Celentano, 2008; Singh et al., 2008). Low rates
of HIV prevention and care delivery are further supported by studies of health care
providers that suggest poor clinical management of persons living with HIV/AIDS
(PLHA) and inadequate adherence to guidelines of universal precautions (Datye et
al., 2006). Low awareness of services coupled with poor quality of these services
might explain why individuals get tested very late in infection and typically when
they are sick. Another explanation outside of the health care delivery system might
lie in the role of stigma in HIV/AIDS in India (Mawar, Saha, Pandit, & Mahajan,
2005).
Stigma is experienced as a result of social norms that render an attribute or a
condition to be inferior or inappropriate (Link & Phelan, 2001). It has also been
described as a reaction to fear or perceived threat (Ogden & Nyblade, 2005). Stud-
ies have outlined that stigma may be a social process related to power, control, or
morality (Alonzo, 1995). Although diseases such as leprosy and tuberculosis have
been stigmatized as a result of a combination of these processes, the role of perceived
morality is prominent in HIV stigma. In India, knowledge of a PLHA has resulted
in refusal of care by physician (Kielmann et al., 2005), discriminatory treatment by
health providers (Sheikh et al., 2005), and reports of physical violence (Rogers et al.,
2006). Further reports of suicide of entire families upon receiving an HIV-positive
diagnosis of one member (Kulkarni SS), high-profile law suits arising from marriages
arranged without disclosing HIV status of the prospective groom, and instances of
withholding disclosure of a son’s HIV status until a grandchild is conceived by the
daughter-in-law suggest complex social and cultural dimensions of stigma in India
(Taraphdar, 2007). Perhaps anticipating discriminatory treatment in various walks
of life, individuals hesitate to adopt prevention and care-seeking behaviors. This
is evidenced from studies that show that stigma influences communicating about
prevention, negotiating condom use and seeking an HIV test (Madru, 2003; Roth,
Krishnan, & Bunch, 2001).
The mechanisms of how these prevention and care behaviors are affected by
stigma are poorly understood, although there is evidence that the type of stigma
may influence these behaviors. The literature discusses several sources and forms of
stigma. Fear of transmission by a PLHA is a primary source of stigma. Forms of
stigma include enacted stigma, community stigma, and internalized stigma (Berger,
Ferrans & Lashley, 2001; Steward et al., 2008). Enacted stigma is one that measures
how respondents would act toward PLHA (Swendeman, Rotheram-Borus, Comu-
lada, Weiss, & Ramos, 2006); community stigma refers to perceived community
norms about and behaviors toward PLHA (de Bruyn, 2002); internalized stigma
occurs when an uninfected individual either believes in stigmatizing PLHA, or when
a PLHA believes that s/he deserves to be stigmatized (Thomas, 2005). The fear of
infection by PLHA, of unsolicited disclosure to family or community members, and
of a sexual partner’s reaction (which includes physical and sexual violence) upon
disclosing HIV-positive status have been found by many studies worldwide to be
barriers to seeking HIV counseling-and-testing services (Herek, Capitanio, & Wida-
man, 2003; Lawyers’ Collective, 2004; Keusch, Willentz, & Klienman 2006). Other
barriers include unsupportive community norms and fear of stigma and discrimina-
tion by health providers (Chandrasekaran, Dallabetta, Loo, Rao, Gayle, & Alexan-
der, 2006).

Page 3

SOCIAL CAPITAL AND STIGMA IN CHENNAI 235
From these studies it is clear that any intervention to reduce stigma needs to
consider factors that operate beyond the individual’s control. A possible interven-
tion strategy may lie in developing supportive social norms about PLHA, mobilizing
community support, and strengthening health care systems for HIV prevention and
care delivery (Gilmore & Somerville, 1994; Parker & Aggleton, 2003). We posit
that this strategy would build social capital. Social capital, in a very broad sense,
refers to the systems within a community that increase the ability of community
residents to work together for mutual benefit (Kawachi, 2006). First postulated by
Durkheim (1951) as the strength of a group over individual members of the group,
this concept was later shaped by Bourdieu, Coleman, and Putnam (2000) to suggest
that individual health and development outcomes are often influenced by factors
operating outside his or her realm of influence; and it is by community participa-
tion and neighborhood-based action that several of these goals might be achieved
(Lomas, 1998; Welshman, 2006). Some elements of social capital are the strength of
formal and informal organizations and social networks within a community, collec-
tive action and social support toward common goals, and norms of interaction and
reciprocity, which refers to the relationships between members of a community that
can work to mutual benefit and growth (Chavez, Kemp, & Harris, 2004; Yamaoka,
2008). Herein, the role of neighborhood organization and community participation
are key (Muntaner, Lynch, & Smith, 2001). Social development researchers have
also discussed the role of trust in government and other civic leadership and empow-
erment as essential components of social capital (Mohseni & Lindstrom, 2008).
However, although there is general agreement in the literature about the impor-
tance of building social capital, there is absent a consensus in the literature about
the manner of its application in public health. As a latent variable, its interpretation
and measurement are also debated. HIV research papers have either measured social
capital as an aggregate variable at the macro level (Holtgrave & Crosby, 2003), or
used few measures such as volunteerism, community group participation or social
networks to measure it at the individual level (Campbell, Williams, & Gilgen, 2002).
Although we recognize this debate, we also in our research seek to explore new ap-
proaches for HIV prevention in contexts where control efforts are most needed such
as in India.
To realize better prevention and care outcomes for HIV, particularly to reduce
stigma in India, social capital may be a relevant strategy for many reasons. First, the
source of stigma is often community or health systems based (Mawar et al., 2005).
Second, there is evidence in the literature that when the social network and social
support aspects of social capital are strengthened, desirable HIV related outcomes
are achieved (Campbell et al., 2002; Gregson, Terceira, Mushati, Nyamukapa, &
Campbell, 2004). Studies in South Africa and Tanzania have shown that partici-
pation in voluntary activities and membership in community groups were factors
associated with lower HIV prevalence (Campbell et al., 2002). Macro analysis of
data from U.S. youth has suggested that social capital measures are associated with
higher rates of protective sexual behaviors (Crosby, Holtgrave, DiClemente, Win-
good, & Gayle 2003).
There is indirect evidence from HIV research in India that suggests a possible
role for social capital in realizing HIV prevention goals. In the Sonagachi project
targeting female sex workers in eastern India, the data show that community or-
ganization and social support generated by this project has resulted in female sex
workers gaining information about HIV prevention and advocating for health care.
Their formal organization allow for dialogue and discussion between women and

Page 4

236 SIVARAM ET AL.
their influencers (brokers, brothel madams and clients) and has added more value
and acceptability to HIV prevention efforts (Jana, Basu, Rotheram-Borus, & New-
man, 2004). In addition to community organization for HIV prevention, there is
also evidence of the role of social networks in promoting HIV prevention knowledge
and care. Our earlier work has shown that communication about sex and sexual
health among men occurs among members of close personal networks that not only
inform but also positively influence sexual decisions about care seeking (Sivaram et
al, 2005).
However, this evidence of strength in organization and peer networks around
HIV prevention is more the exception than the norm in India. Among persons and
families living with HIV/AIDS studied here however, a lack of social support remains
a predominant finding. Lower levels of social support, higher feelings of isolation,
and loss of control are expressed by PLHA. Female PLHA experience stigma at
much higher levels and are more likely than men to be blamed for their infection
(Alert, 2002; Solomon, Chakraborty, & Yepthomi, 2004). Another source for lack
of social support is in the health service sector. In some studies physicians them-
selves fear touching PLHA, and often blame them for their illness (Kielmann et al.,
2005). Advocates for HIV-positive individuals and their families emphasize the need
for social support not only from physicians but also from community and families
in order to better manage disease in India (Indian Network of Positive Living with
HIV/AIDS, 2003).
This article aims to identify measures of social capital based on ethnography
and to explore the statistical association between these social capital measures and
HIV-related stigma. We describe domains of social capital that are both positively
and negatively associated with stigma. To our knowledge such an analysis has not
been reported in the literature from India. However, as public health scientists and
practitioners move toward developing new approaches and interventions to address
stigma in the community, we anticipate that our findings might help highlight areas
of focus for future interventions to reduce HIV stigma.
mEtHods
The study was conducted in Chennai city in the south India state of Tamil Nadu.
The data was collected as part of the NIMH HIV/STD Prevention Trial, a commu-
nity-based cluster randomized trial to test the efficacy of HIV prevention messages
disseminated through community popular opinion leaders, or CPOLs (National In-
stitute of Mental Health [NIMH, 2007). The study was implemented in wine shops
or bars in Chennai and CPOLs were selected among men and women who patronize
these venues. The trial began with ethnographic in-depth interviews of 41 men
and women who partronize wine shops. These interview helped develop the survey
questionnaire and intervention content. The trial assessments consisted of a baseline
survey of behavioral and biological risk followed by CPOL training in intervention
venues (control venues received intervention after a 2-year lag). Follow-up assess-
ments of the baseline cohort were conducted at 18 and 34-months after the base-
line. At the 34 month assessment in 2006, we conducted an additional assessment of
HIV/AIDS stigma and social capital. Men were recruited using systematic random
sampling in the venue and women associated with the wine shop were also recruited.
To be eligible, participants had to be 18-40 years of age, patronize bars at least
thrice weekly, anticipate living in Chennai in the next year, and be lucid and capable

Page 5

SOCIAL CAPITAL AND STIGMA IN CHENNAI 237
of providing voluntary informed consent at the time of the interview. Recruitment
was conducted in the wine shops, and informed consent procedures followed by
assessment was conducted in the trial offices. Where needed, participants were pro-
vided transportation to the trial office for assessment. This study was initiated after
ethical review and approval from the institutional review boards of YRG Center for
AIDS Research and Education in Chennai, India and the Johns Hopkins University
Bloomberg School of Public Health in Baltimore, MD.
mEasurEs
stigma
Stigma was measured by a scale that had been previously piloted and validated
in a similar population, with four domains (Zelaya et al., 2008). The first domain,
fear of transmission, measured causes of stigma by asking questions on individual’s
fears of contracting HIV by their interaction with PLHA. The second domain, as-
sociation with shame blame and judgment, measured participants’ attitudes toward
PLHA. Although these two domains measured the causes of stigma, the third do-
main and fourth domain measured enacted and perceived stigma. The third do-
main, personal support of discriminatory actions or policies against PLHA, sought
responses on how participants’ would themselves treat PLHA or how they perceive
PLHA should be treated. The fourth and final domain, perceived community sup-
port of discriminatory actions or policies against PLHA, assessed participants, per-
ception of how the community treats and perceives PLHA. All stigma domains were
the main outcomes (latent variables) in this study.
social caPital
Social capital measures were developed considering two sources of information. The
first was our literature review and second was review of existing instruments (World
Bank, 2009). Following this, within the protocol of the NIMH trial, we developed
ethnographic guides to gather information on (a) social networks, (b) sources of
social support for HIV prevention and care, (c) health-care seeking behaviors, and
(d) community life and interaction. We conducted 41 ethnographic interviews. All
ethnographic interviews were tape-recorded in the local language, translated and
transcribed in English.
Development of Social Capital Variables. In the ethnographic analysis, participation
in community groups and the role of friends emerged as valuable in providing and
discussing information about HIV/AIDS, shaping decisions to drink and to have sex.
Participants reported being part of several formal and informal community groups.
Among men, these included youth groups, movie star fan clubs, and trade groups
such as auto-rickshaw driver associations. Among women, respondents reported
belonging to community-based women’s groups that discussed health and nutrition
issues. Friends encouraged each other to drink in order to gain weight, in order to
get the courage to have sex, to relive minor aches and pains, or simply to be “jolly,”
a term that reflects a stress-free state of mind. In addition to knowledge, friends also
supported each other by providing money to have sex and negotiating rates with a
sex worker as this quote suggests:

Page 6

238 SIVARAM ET AL.
When we want to have sex, four of us friends go together to the sex worker. Some of us
may not have enough money but still want to have sex. So we negotiate from 500 rupees
and at last settle down for 300 rupees and have sex (Unmarried male, aged 28)
Participants also reported getting support from friends and neighbors in several
ways: money loaned when needed, assistance with getting a job, support talking
about concerns and problems at home.
In the domain of health care seeking, we found several instances of HIV preven-
tion and care information coming from health care providers. One male respondent
reported learning from a doctor that the primary sources of sexually transmitted
diseases (STDs) are women. Another doctor provided information the following on
sexual behavior:
After drinking, I will have the sexual urge and want to masturbate. But the doctor told
me that masturbation will damage the penis. So, in order to fulfill my urge, I visit a sex
worker--Unmarried male, age 24.
Based on this analysis, our earlier work on community norms and HIV communica-
tion among social networks and by referring to questions on social capital from the
literature, we measured two domains of social capital--groups and networks and
collective action. Groups and networks were measured by assessing (a) participants’
membership in formal and informal groups in the community, (b) reported number
of close friends and (c) availability of financial assistance in times of need, and (d)
trustworthiness of community members. Collective action was measured using
three indicators: norms of reciprocity, likelihood of collaboration for mutual benefit,
HIV-related support and action. Norms of reciprocity were measured by a series of
questions assessing how participants assist their community and friends and how in
turn they are assisted similarly by community and friends. For instance, we asked if
they received assistance in seeking health care and, conversely, if they got assistance
in seeking health care. Collective action was determined by whether community
members donate time to projects that further common goals in the community. We
developed questions to assess available HIV/AIDS support by assessing the availabil-
ity of a trusted health care provider and by assessing if they have discussed HIV with
friends. All social capital domains served as independent variables in the analysis.
data analysis
Ethnographic transcripts were analyzed using Atlas.ti (Muhr, 1998), a textual analy-
sis software program. The data were reviewed for three main themes or codes as
they were called in the software program: composition and characteristics of social
networks, the content of communication in these networks and the association of
communication with sexual behavior. Text that matched these codes were retrieved
and reviewed. Matrices were developed for each code to enable organization of the
data and to understand similarities and contrasts across related themes.
In the quantitative analysis, our main outcome variables were the four domains
of stigma. Predictor variables were the four domains of social capital. Domains of
social capital were dichotomous or continuous variables. Some questions in collec-
tive action and available HIV support were measured using a Likert scale of items.
We first conducted a descriptive analysis of the variables. Following this, we used
structural equation modeling (SEM) to investigate the relationship between social

Page 7

SOCIAL CAPITAL AND STIGMA IN CHENNAI 239
capital and HIV/AIDS stigma. SEM is a system of simultaneous equations that
estimate associations between observed predictor variables and outcome variables
through intervening continuous latent variables. This study uses a multiple cause
and multiple indicators (MMIC) model, a type of SEM in which one or more la-
tent variables intervene between a set of multiple observed predictor and dependent
variables. All analyses were conducted using Mplus, Version 4.2 (Muthen, 2007).
The relationships between social capital variables (predictors) and stigma (latent
variables) while controlling for demographic characteristics are described by multi-
variate linear regression equations within the model.
rEsults
At the 34-month survey, 2,422 participants were interviewed. Of this number, 53
were removed from the analysis as they had previously tested positive for HIV re-
sulting in a final sample of 2,369. Eighty-four percent were male. Median age of the
population was 30 years with women being slightly older than men in this sample.
Overall, participants had an average of 7 years of formal education. Sixty percent
of participants were married or living with a partner and 31.9% were never mar-
ried; relatively more men were married and single in this sample as compared with
women. Among women, fewer than 5% reported no children with over 80% of the
women reporting two or more children. Thirty-three percent of men reported no
children and 50% reported two or more children. Our ethnographic data showed
that most resided in either nuclear families or had a older relative (in-laws, parents)
living with them. Occupations such as auto-rickshaw drivers, day laborers and
those in private sector and government service were represented among men. All
women in the sample were female sex workers. Among women, there were reports
of abusive spouses, separation or abandonment by spouses.
All participants patronized wine shops. We have reported detailed descriptions
of wine shops earlier (Sivaram et al., 2007). Men visited wine shops at least thrice
weekly and represented individuals from all walks of life. Typically, men patron-
ized wine shop with friends--65% of our sample reported visiting wine shops with
friends and saw wine shops as a primary venue for socialization. Women visited
wine shops less frequently but through mediators had male patrons as their clients.
We also explored social network and capital measures reported by the partici-
pants (Table 1). Twenty-seven percent reported membership in a formal group in
the community such as youth group or women’s clubs. Informal group membership
was reported by less than 2%. A majority (53%) of participants reported between
one and four friends, and 31% reported five or more friends. Overall, 55% of par-
ticipants reported that they cannot trust people in the community.
We also assessed norms of reciprocity between friends and neighbors--69%
reported loaning money, 85% reported listening to problems, 42% reported help-
ing neighbors and friends with child care, and 46% reported assisting with seeking
health care. Conversely, 58% reported having receiving money from friends and
neighbors in times of need, 82% reported that they had a listening ear to share per-
sonal concerns, 33% they received help with seeking health care, and 29% reported
that they receive help with child care. Frequency analysis of community partici-
pants showed that 64% reported willingness to contribute time to a project to ben-
efit the community and 40% said that they will contribute money to a community
project. Finally, we examined two variables that indicated HIV related support and

Page 8

240 SIVARAM ET AL.
action. Twenty-five percent reported that a trusted health provider to discuss HIV/
STD was available in their community, and 43% reported that they discussed HIV
prevention with friends.
Table 2 presents the results of the analysis of the association between the domain
of fear and transmission and various covariates measuring social capital. Among
measures of groups and networks, we found that fear of transmission significantly
decreased as membership in formal groups increased. Further, the lesser the likeli-
hood of perceived financial support, the higher reported fear of transmission and
disease among men and women in the multivariate analyses, and the lesser reports
of trust, the higher the likelihood of reports of fear of transmission and disease.
We assessed norms of reciprocity to estimate how individuals help and rely on each
other and the extent of cooperation in the community. We found that when reci-
procity was higher, fear was lower among both men and women. Similarly, higher
likelihood of collective action was associated with lower fear among men and wom-
en in the adjusted analysis. Where there was a trusted STD doctor available, this
availability was significantly associated with lower fear of transmission and disease
TABLE 1. Frequency Distribution of Social Network and Social Capital Measures Reported by
Participants
measure frequency Percentage
groups and networks

Membership in a formal group27
Membership in an informal group1.30
Number of Close Friends
0 16
1-4 friends53
5 or more friends31
Trust
Cannot trust people in the community55
collective action

Norms of reciprocity

Loaned money to friends/neighbors69
Listening to concerns of friends/neighbors85
Helped friends/neighbors with childcare42
Helped friends/neighbors seek health care46
Was loaned money by friends/neighbors58
Friends/neighbors listened to my concerns82
Friends/neighbors help me with child care33
Friends/neighbors helped me seek care29
Collaboration for community’s benefit

Will contribute time to work on a community project64
Will contribute money to help a community project40
HIV-related support and action

Availability of a trusted health provider25
Discussion about HIV with friends43

Page 9

SOCIAL CAPITAL AND STIGMA IN CHENNAI 241
OUTCOME: FEAR OF TRANSMISSION AND DISEASE
UNIVARIATE MALES
MULTIVARIATE MALES*
UNIVARIATE FEMALES
MULTIVARIATE - FEMALES*
Estimate
(LCI, UCI)
Estimate
(LCI, UCI)
Estimate
(LCI, UCI)
Estimate
(LCI, UCI)
# informal groups where participant is a member
0.049
(-0.024, 0.122)
0.09
(-0.008, 0.188)
0.05
(-0.160, 0.260)
0.114
(-0.139, 0.367)
# formal groups where participant is a member
-0.111
(-0.180, -0.042)
-0.117 (-0.186, -0.048)
-0.326 (-0.583, -0.069)
-0.301
(-0.585, -0.017)
# of close friends
-0.008
(-0.016, 0.000)
-0.002 (-0.012, 0.008)
-0.013 (-0.027, 0.001)
-0.011
(-0.027, 0.005)
People available to help if you need money urgently
0.118
(0.059, 0.177)
0.079
(0.018, 0.140)
0.228
(0.097, 0.359)
0.178
(0.033, 0.323)
(Definitely=1, Probably=2, Unsure=3, Probably not=4)

People can be trusted (1)
REF
REF
REF
REF
You have to be careful (2)
0.325
(0.198, 0.452)
0.304
(0.182, 0.426)
0.625
(0.345, 0.905)
0.617
(0.882, 0.882)
Helping others (Continuous Latent Variable, higher value
higher reciprocity)
-0.538
(-0.646, -0.430)
-0.488 (-0.596, -0.380)
-0.924 (-1.665, -0.183)
-0.979
(-1.649, -0.309)
Others helping you (Continuous Latent Variable, higher
value higher reciprocity)
-0.275
(-0.338, -0.212)
-0.261 (-0.324, -0.198)
-0.593 (-1.071, -0.115)
-0.472
(-0.854, -0.090)
Likelihood that people will cooperate to solve water supply problem (1=Very likely to 5=very unlikely)
0.205
(0.113, 0.297)
0.202
(0.108, 0.296)
0.126
(0.010, 0.242)
0.121
(0.001, 0.241)
Perception of safety at home (1=very safe to 5=very
unsafe)
0.134
(0.034, 0.234)
0.104 (-0.004, 0.212)
0.197
(0.060, 0.334)
0.183
(0.056, 0.310)
Trusted STD doctor available
No
REF
REF
REF
REF
Yes
Collective Action
-3.437
(-0.713, -0.195)
-0.426 (-0.691, -0.161)
0.028 (-0.231, 0.287)
0.068
(-0.212, 0.348)
Have you discussed HIV with friends
No
REF
REF
REF
REF
Yes
-0.46
(-0.597, -0.323)
-0.419 (-0.566, -0.272)
-0.792 (-1.041, -0.543)
-0.808
(-1.102, -0.514)
* Controling for age (in years); education (years of school completed); marital status (married, never married, widowed, and separated/divorced); income (do you regularly earn money-yes/no); when
you visit the wine shop, do you come with friends (yes/no); religion (hindu, christian, muslim, other)
Groups and
Networks
Collective Action
(HIV related)
Table 2: Linear regression coefficients and 95% confidence intervals of social capitol domains as predictors of fear of transmission and disease (domain of HIV/AIDS stigma) among high
risk men and women in Chennai, India

Page 10

242 SIVARAM ET AL.
among men. However, among women the presence of a trusted doctor was associ-
ated with more fear of HIV transmission. Among both men and women, discussing
HIV with friends was significantly associated with lower fear.
Following fear of transmission, we explored social capital variables associated
with feelings of shame blame and judgment (Table 3). Here we found that among
men, as membership in informal groups increased, there was more association with
shame, blame and judgment. The less likely there were people available to help in
financial need, the more the association with shame, blame, and judgment among
men and women. Among women, the higher the number of close friends, the lower
the association of shame, blame, and judgment. There was no significant association
between close friends and the outcome among men. Among men and women, high-
er norms of reciprocity were associated with lower associations with shame, blame
and judgment. When women reported less likelihood of collective action, there was
more association with shame, blame, and judgment. When we analyzed the role
of availability of a trusted STD physician, we found that among men who reported
having availability, there was more association with shame, blame and judgment.
Among both men and women, discussing about HIV with friends predicted less as-
sociation with shame, blame, and judgment.
We next examined the associations between social capital and personal sup-
port for discriminatory actions and policies against individuals affected by HIV/
AIDS (Table 4). Membership of men in a higher number of formal groups; higher
norms of reciprocity, availability of a trusted doctor, and discussing HIV with friends
were associated with lower personal support for discriminatory actions and policies.
Among women, higher number of close friends, high levels of norms of reciprocity,
availability of a trusted STD doctor, and discussing HIV with friends (among men)
were associated with less personal support for discriminatory actions and policies.
More personal support of discriminatory actions were associated among men and
women with unsafe neighborhoods.
Finally, we examined the associations between social capital variables and
perceived community support for discriminatory actions or policies toward PLHA
(Table 5). Here, we found that more individuals perceived community support for
PLHA discrimination when they (men) also reported lesser availability of financial
help, more close friends (women), lower reciprocity and unsafe neighborhoods. In-
dividuals perceived lesser community support for discrimination when there was a
trusted STD doctor and when men reported discussing HIV with friends.
discussion
In India, stigma is an important barrier faced by PLHA to seek care and maintain
a desired quality of life. In this study, we explored the role of social capital as a
possible strategy to reduce HIV stigma and our findings suggest that this may be a
viable approach. The two domains of social capital that we measured were groups
and networks and collective action. We discuss the findings under each social capital
domain and outline suggestions of how these findings might translate to intervention
design.
grouPs and nEtworKs
Membership in groups and networks are integral to community life in the study
area. Informal meetings in the neighborhood to discuss a range of issues, formal

Page 11

SOCIAL CAPITAL AND STIGMA IN CHENNAI 243
OUTCOME:ASSOCIATIONS WITH SHAME BLAME AND JUDGEMENT
UNIVARIATE MALES
MULTIVARIATE MALES*
UNIVARIATE FEMALES
MULTIVARIATE - FEMALES*
Estimate
(LCI, UCI)
Estimate
(LCI, UCI)
Estimate
(LCI, UCI)
Estimate
(LCI, UCI)
# informal groups where participant is a member
0.064
(0.001, 0.127)
0.083
(0.018, 0.148)
-0.083 (-0.336, 0.170)
-0.100
(-0.398, 0.198)
# formal groups where participant is a member
-0.017
(-0.046, 0.012)
-0.014 (-0.041, 0.013)
-0.127 (-0.280, 0.026)
-0.094
(-0.304, 0.116)
# of close friends
-0.008
(-0.018, 0.002)
-0.001 (-0.011, 0.009)
-0.023 (-0.035, -0.011)
-0.024
(-0.038, -0.010)
People available to help if you need money urgently
0.196
(0.123, 0.269)
0.172
(0.092, 0.252)
0.178
(0.055, 0.301)
0.132
(0.010, 0.254)
(Definitely=1, Probably=2, Unsure=3, Probably not=4)
People can be trusted (1)
REF
REF
REF
REF
You have to be careful (2)
-0.023
(-0.111, 0.065)
-0.034 (-0.116, 0.048)
0.798
(0.490, 1.106)
0.817
(0.523, 1.111)
Helping others (Continuous Latent Variable, higher value higher
reciprocity)
-0.234
(-0.369, -0.099)
-0.180 (-0.303, -0.057)
-1.257 (-2.233, -0.281)
-1.387
(-2.398, -0.376)
Others helping you (Continuous Latent Variable, higher value
higher reciprocity)
Collective Action
-0.113
(-0.201, -0.025)
-0.086 (-0.176, 0.004)
-0.838 (-1.402, -0.274)
-0.692
(-1.135, -0.249)
Likelihood that people will cooperate to solve water supply problem
(1=Very likely to 5=very unlikely)
0.073
(-0.017, 0.163)
0.057 (-0.039, 0.153)
0.127
(0.002, 0.252)
0.137
(0.006, 0.268)
Perception of safety at home (1=very safe to 5=very unsafe)
-0.168
(-0.264, -0.072)
-0.198 (-0.294, -0.102)
0.329
(0.198, 0.460)
0.323
(0.194, 0.452)
Trusted STD doctor available
No
REF
REF
REF
REF
Yes
Collective Action
-0.272
(-0.417, -0.127)
-0.243 (-0.380, -0.106)
-0.109 (-0.297, 0.079)
-0.093
(-0.277, 0.091)
Have you discussed HIV with friends
No
REF
REF
REF
REF
Yes
-0.330
(-0.438, -0.222)
-0.286 (-0.400, -0.172)
-0.298 (-0.580, -0.016)
-0.290
(-0.608, 0.028)
Groups and
Networks
(HIV related)
* Controling for age (in years); education (years of school completed); marital status (married, never married, widowed, and separated/divorced); income (do you regularly earn money-yes/no); when you
visit the wine shop, do you come with friends (yes/no); religion (hindu, christian, muslim, other)
Table 3: Linear regression coefficients and 95% confidence intervals of social capitol domains as predictors of 'associations with shame, blame and judgment (domain of HIV/AIDS stigma)
among high risk men and women in Chennai, India

Page 12

244 SIVARAM ET AL.
OUTCOME:PERSONAL SUPPORT FOR DISCRIMINATORY ACTIONS OR POLICIES TOWARDS PLHA
UNIVARIATE MALES
MULTIVARIATE MALES*
UNIVARIATE FEMALES
MULTIVARIATE - FEMALES*
Estimate
(LCI, UCI)
Estimate
(LCI, UCI)
Estimate
(LCI, UCI)
Estimate
LCI
UCI
# informal groups where participant is a member
0.016
(-0.029, 0.061)
0.033 (-0.016, 0.082)
-0.035 (-0.258, 0.188)
-0.034
(-0.277, 0.209)
# formal groups where participant is a member
-0.032
(-0.050, -0.014)
-0.036 (-0.056, -0.016)
-0.189 (-0.452, 0.074)
-0.152
(-0.417, 0.113)
# of close friends
0.001
(-0.005, 0.007)
0.006
(0.002, 0.010)
-0.018
(-0.03, -0.006)
-0.016
(-0.028, -0.004)
People available to help if you need money urgently
0.094
(0.053, 0.135)
0.072
(0.027, 0.117)
0.192
(0.104, 0.280)
0.173
(0.093, 0.253)
(Definitely=1, Probably=2, Unsure=3, Probably not=4)
People can be trusted (1)
REF
REF
REF
REF
You have to be careful (2)
0.064
(-0.020, 0.148)
0.044 (-0.038, 0.126)
0.378
(0.145, 0.611)
0.376
(0.155, 0.597)
Helping others (Continuous Latent Variable, higher value
higher reciprocity)
-0.411
(-0.491, -0.331)
-0.364 (-0.444, -0.284)
-1.229 (-2.256, -0.202)
-1.324
(-2.255, -0.393)
Others helping you (Continuous Latent Variable, higher
value higher reciprocity)
-0.222
(-0.265, -0.179)
-0.204 (-0.249, -0.159)
-0.799 (-1.367, -0.231)
-0.696
(-1.127, -0.265)
Likelihood that people will cooperate to solve water supply problem (1=Very likely to 5=very unlikely)
0.146
(0.095, 0.197)
0.137
(0.086, 0.188)
0.233
(0.145, 0.321)
0.234
(0.142, 0.326)
Perception of safety at home (1=very safe to 5=very
unsafe)
0.234
(0.177, 0.291)
0.213
(0.152, 0.274)
0.212
(0.112, 0.312)
0.203
(0.099, 0.307)
Trusted STD doctor available
No
REF
REF
REF
REF
Yes
-0.271
(-0.414, -0.128)
-0.246 (-0.393, -0.099)
-0.435 (-0.658, -0.212)
-0.427
(-0.654, -0.200)
Have you discussed HIV with friends
No
REF
REF
REF
REF
Yes
-0.271
(-0.347, -0.195)
-0.232 (-0.310, -0.154)
-0.411 (-0.627, -0.195)
-0.416
(-0.671, -0.161)
Groups and
Networks
Collective Action
Collective Action
(HIV related)
* Controling for age (in years); education (years of school completed); marital status (married, never married, widowed, and separated/divorced); income (do you regularly earn money-yes/no); when you
visit the wine shop, do you come with friends (yes/no); religion (hindu, christian, muslim, other)
Table 4: Linear regression coefficients and 95% confidence intervals of social capitol domains as predictors of 'personal support for discriminatory actions and policies towards PLHA' (domain
of HIV/AIDS stigma) among high risk men and women in Chennai, India

Page 13

SOCIAL CAPITAL AND STIGMA IN CHENNAI 245
OUTCOME: PERCEIVED COMMUNITY SUPPORT FOR DISCRIMINATORY ACTIONS OR POLICIES TOWARDS PLHA
UNIVARIATE MALES
MULTIVARIATE MALES*
UNIVARIATE FEMALES
MULTIVARIATE - FEMALES*
Estimate
(LCI, UCI)
Estimate
(LCI, UCI)
Estimate
(LCI, UCI)
Estimate
(LCI, UCI)
# informal groups where participant is a member
0.028
(-0.039) 0.095)
0.018 (-0.049, 0.085)
0.051
(-0.022, 0.124)
0.018
(-0.056, 0.092)
# formal groups where participant is a member
0.024
(-0.013, 0.061)
0.034 (-0.001, 0.069)
0.009
(-0.020, 0.038)
0.023
(-0.016, 0.062)
# of close friends
-0.004
(-0.012, 0.004)
-0.001 (-0.009, 0.007)
-0.008 (-0.012, -0.004)
-0.008
(-0.012, -0.004)
People available to help if you need money urgently
0.144
(0.087, 0.201)
0.135
(0.074, 0.196)
0.004
(-0.023, 0.031)
-0.001
(-0.026, 0.024)
(Definitely=1, Probably=2, Unsure=3, Probably not=4)
People can be trusted (1)
REF
REF
REF
REF
You have to be careful (2)
0.114
(-0.011, 0.239)
0.102 (-0.027, 0.231)
0.084
(-0.020, 0.188)
0.059
(-0.029, 0.147)
Helping others (Continuous Latent Variable, higher value
higher reciprocity)
-0.373
(-0.479, -0.267)
-0.344 (-0.450, -0.238)
-0.340 (-0.607, -0.073)
-0.310
(-0.543, -0.077)
Others helping you (Continuous Latent Variable, higher
value higher reciprocity)
-0.226
(-0.304, -0.148)
-0.211 (-0.287, -0.135)
-0.397 (-0.666, -0.128)
-0.294
(-0.504, -0.084)
Likelihood that people will cooperate to solve water supply problem (1=Very likely to 5=very unlikely)
0.032
(-0.033, 0.097)
0.028 (-0.035, 0.091)
-0.025 (-0.052, 0.002)
-0.020
(-0.047, 0.007)
Perception of safety at home (1=very safe to 5=very
unsafe)
0.256
(0.166, 0.346)
0.243
(0.147, 0.339)
0.063
(0.014, 0.112)
0.051
(0.008, 0.094)
Trusted STD doctor available
No
REF
REF
REF
REF
Yes
Collective Action
-0.323
(-0.488, -0.158)
-0.307 (-0.472, -0.142)
-0.099 (-0.154, -0.044)
-0.090
(-0.139, -0.041)
Have you discussed HIV with friends
No
REF
REF
REF
REF
Yes
-0.350
(-0.495, -0.205)
-0.334 (-0.479, -0.189)
0.050
(-0.023, 0.123)
0.036
(-0.029, 0.101)
Table 5: Linear regression coefficients and 95% confidence intervals of social capitol domains as predictors of 'percieved community support of discriminatory actions and policies
towards PLHA' (domain of HIV/AIDS stigma) among high risk men and women in Chennai, India
Groups and
Networks
Collective Action
(HIV related)
* Controling for age (in years); education (years of school completed); marital status (married, never married, widowed, and separated/divorced); income (do you regularly earn money-yes/no); when
you visit the wine shop, do you come with friends (yes/no); religion (hindu, christian, muslim, other)

Page 14

246 SIVARAM ET AL.
membership in youth groups or women’s groups have been reported by us earlier
(Sivaram et al., 2005). In this analysis, we found that among men and women, mem-
bership in formal community groups is associated with reduced fear of transmission;
reduced shame, blame, and judgment, and reduced personal support for discrimi-
natory actions against PLHA. This finding merits consideration as HIV education
implemented by governmental and nongovernmental organizations is often targeted
at these formal groups (Godbole & Mehendale, 2005; Pallikadavath, Garda, Apte,
Freedman, & Stones, 2005). This is particularly relevant in the case of female sex
workers in Chennai who receive several targeted HIV prevention interventions (Pan-
chanadeswaran et al., 2008; Uma et al., 2005). A formal group might allow for
structure in presenting and processing information and it is plausible that members
in these formal groups receive relatively more accurate information than informal
groups. This is further supported by our finding of a significant increase of associa-
tion with shame, blame, and judgment as membership in informal groups increased
among men. Although the number of close friends among men was not associated
with any stigma domain except personal support of discriminatory actions where
there was a negative association, among women, higher number of close friends may
be an important point for intervention. Finally, key measures of network support-
-the ability to rely on others for financial help and trustworthiness--were strongly
associated (among men and women) with low levels of stigma. These social sup-
port measures have been reported to facilitate disclosure (Chandra, Deepthivarma,
Jairam, & Thomas, 2003), and access to and adherence to HIV medication (Ku-
marasamy, Safren, Raminani, Pickard, James, Krishnan et al., 2005) in India and
elsewhere (Wolitski, Pals, Kidder, Courtenay-Quirk, & Holtgrave, 2008). In the
Indian context, assessing these factors while planning an intervention for PLHA may
suggest areas of focus and individuals of focus in order to build social capital and
reduce stigma. The factors of community support, trust worthy individuals have
been also reported as key facilitators in the effectiveness of several public health and
development initiatives (Van Rompay et al., 2008). PolioPlus campaigns in India
rely on community effort to motivate and support mothers to get their children vac-
cinated; the trust enjoyed by government health workers for this effort is perhaps a
key factor in the large turnout in national drives to eradicate polio (Balraj, Mukun-
dan, Samuel, & John, 1993). A similar initiatives, the AIDS Support Organization
in Uganda, began as a community-based peer support campaign that is currently
a dominant force against HIV/AIDS stigma in the country’s national campaigns
(Rwemisisi, Wolff, Coutinho, Grosskurth, & Whitworth, 2008).
Based on our findings, future programs that seek to reduce stigma may consider
exploring the feasibility of stable community institutions that meet formally and
seek the participation of individuals who work to better their communities as a first
step in developing the intervention. Further, programs can understand the composi-
tion and number of close social networks of PLHA. This can serve as a catalyst to
help disseminate communication messages that can reduce HIV stigma.
collEctivE action
The norms of reciprocity--the practice of give-and-take and assisting each other-
-that are the hallmark of an ideal community have been applied to several develop-
ment related initiatives. Community-based bed- netting programs for malaria con-
trol in Africa rely on collective action. Studies in the United States and elsewhere
suggest that social isolation and lack of cohesiveness in a community are associ-
ated with poverty, which in turn predicts poor mental health outcomes and unsafe

Page 15

SOCIAL CAPITAL AND STIGMA IN CHENNAI 247
communities (Chavez et al., 2004; Msisha, Kapiga, Earls, & Subramanian, 2008;
Rothenberg, Muth, Malone, Potterat, & Woodhouse, 2005). Our findings that high
norms of reciprocity, higher level of collective action and higher perception of safety
are associated with lowered HIV stigma suggests a more holistic approach to HIV
prevention efforts. We measured collective action by asking individuals about their
views on community cooperation to solve water supply problem, a chronic and
prominent concern for citizens of Chennai. Prevention programs may be limited in
effect if they focus only on individuals without addressing some of the larger com-
munity-level needs. Particularly for outcomes such as HIV stigma that are reliant
on community perceptions and action, interventions that address these factors may
be more relevant to individuals have higher public health significance.
Our findings on the role of HIV specific collective action further illustrate the
sources of stigma in the community and consequently the role of community in-
volvement in reduction of HIV stigma. Among men who informed the study that
there was a trusted doctor in the community, there were lower levels of stigma re-
ported, although the associations were stronger in men than women. This finding
may reflect on the content of doctor-client communication and suggest involvement
of physicians in disseminating HIV prevention messages and promoting positive at-
titudes toward PLHA. Given the significant evidence that suggests a negative role of
physicians--refusal to provide care to PLHA, referrals of PLHA, and misinformation
about HIV--these data show the positive role that physicians can play in prevention
education (Datye et al., 2006).
limitations
Our study has limitations that we would like to outline. First, as it was nested
in a larger trial we measured stigma and social capital as it was relevant to the
Chennai context. This trial sought participation from alcohol users. These par-
ticipants’ behavior and attitudes may not apply to nonalcohol users or other socio-
demographic groups in Chennai. As such, we are unable to make any statements
about generalizability of study findings. However, we believe that the perspectives
of the respondents with regard to social capital measures may not be biased. This
is because these measures were relatively less sensitive than the other items such as
sexual and substance use behavior that were measured during the survey. Second,
we would like to acknowledge that we measured only a few domains of social capi-
tal. Others reported in the literature include volunteerism, homicide rates, commu-
nity participation, to name a few--these were not measured and as a result we may be
presenting a narrow operational of definition of social capital (Gregson et al., 2004).
A more rigorous approach to measuring social capital is to conduct formative re-
search to learn about the relevance of indicators in the Indian context followed by
development and quantitative assessment of these measures. As we conducted this
research as part of an ongoing trial whose primary outcome was not HIV stigma, we
were limited in our scope. Third, we have limitations in our analytical approach. By
modeling social capital covariates with domains of stigma, we were able to observe
only the direction of association and significance of the associations but not the mag-
nitude or relative strength of these associations. As an illustration of this limitation,
in Table 2, although we are able to note that having more close friends is associated
with less fear and having less access financial support leads to more fear, we are
unable to say which element of groups and networks--friends or financial support
is more important. Further, we did not standardize our coefficients. This does not
allow us to compare the values or width of the confidence intervals directly. We will