Background A multidimensional construct of social capital was employed to understand the relationship between social capital and migrants' HIV risk at destination place.

Methods The study was undertaken among Rajasthani migrants of age 18 and above in Mumbai and Ahmedabad in India to collect data from 1598 migrants through survey method and from 93 migrants through qualitative methods from January to June 2007. Social capital was measured in terms of three domains: bonding, bridging and linking. HIV risk had three measures: having casual partner in last 12 months in the city; sex with a sex worker in the city in last 12 months; and no or inconsistent condom use with a sex worker in last 12 months.

Results Migrants had substantial risk behaviour at destination: 385 respondents (24.1%) reported having one or more casual partners; 218 migrants (13.6%) had had sex with a sex worker; and 123 respondents (7.7%) reported no or inconsistent condom use with a sex worker in the last 12 months in the city. Migrants reported higher risk in Ahmedabad compared to Mumbai for all the three risk measures: 251 (31.6%) vs 134 (16.7%) reported casual partners; 138 (17.4%) vs 80 (10%) reported having had sex with a sex worker; and 96 (12.1%) vs 27 (3.4%). Bonding and linking domains of social capital had higher values for migrants in Ahmedabad than Mumbai. Bridging social capital was higher in migrants in Mumbai as compared to migrants in Ahmedabad. All the components of bridging social capital had higher values for migrants in Mumbai than Ahmedabad. Bonding and linking social capital behaved differently in Mumbai and Ahmedabad. In Mumbai, migrants had lower HIV risk if they had high linking social capital and had higher risk if they had high bonding social capital. In Ahmedabad, bonding social capital at high levels was associated with lower risk behaviour while linking social capital at high level was associated with higher risk. On the other hand, high levels of bridging social capital and components of bridging social capital were protective of HIV risk in both the cities see Abstract O2-S5.01 table 1.

Conclusion This study was able to explore the mediating effect of social capital on migrants' HIV risk at the domain levels. Bridging kind of social capital with the host community and migrants from other states was associated with lower HIV risk behaviour. Further research should be undertaken in different epidemiological contexts to validate the findings of this study.