P11.02 Representing men who have sex with men (msm) in britain: evidence from comparative analyses of the latest convenience and probability surveys

Abstract

Introduction Convenience samples of MSM are typically skewed towards gay-identified and community-attached men who do not necessarily represent all MSM, a key population for STI/HIV transmission. Comparing convenience samples with a probability sample of MSM enables us to assess their representativeness. Methods We compared 148 MSM in Britain's third National Survey of Sexual Attitudes and Lifestyles (Natsal-3), with men in Britain's key convenience surveys of gay men: 15,500 in the European MSM Internet Survey (EMIS), 1,012 in the Gay Men's Sexual Health Survey London (GMSHS-London), 1,234 in the Gay Men's Sexual Health Survey Scotland (GMSHS-Scotland), all undertaken 2010-2012. Men were aged 18-64, resident in Britain, and reported >=1 male sexual partners (past year). Comparisons include demographic, health characteristics and sexual behaviours, using identically-worded questions. Multivariable analyses accounted for socio-demographic differences between the samples. Results Convenience samples had significantly younger and better educated men than Natsal-3, and a larger proportion of men identifying as gay (88%-95% vs. 62%). Partner numbers were larger and same-sex anal sex more commonly reported by convenience samples but no significant differences were observed in age at first sex or reporting unprotected anal intercourse. Compared to Natsal-3, men in convenience samples were more likely to report HIV testing (past year) (all adjusted odds ratios (AORs) >=2.30, 95% CI lower bounds >=1.49) and gonorrhoea diagnoses (all AORs >=7.99, 95% CI lower bounds >=1.11). However, fewer differences between samples were observed when analyses focused on MSM who identified as gay. Conclusion Participation bias may mean convenience samples of MSM demographically misrepresent and over-estimate sexual activity and HIV testing for the entire MSM population. However, they may be more representative of gay-identified MSM, capture a broader range of behaviours and achieve larger samples, relative to general population surveys. Methods are needed to triangulate data from different surveys to strengthen the evidence- base for MSM.