Sexual preference is a better guide to HIV risk in gay men than reported behaviour, Chinese study suggests

This article is more than 12 years old. Click here for more recent articles on this topic

A study of men who have sex with men (MSM) in the city of Chongqing in China has found that HIV prevalence was strongly correlated with whether men preferred to take the insertive, or ‘top’, role in anal sex – but was not correlated with the actual risk behaviour they reported, at least with the last three partners.

This apparently paradoxical result – suggesting, contrary to prevailing wisdom in HIV prevention, that what people desire may be a better guide to their HIV risk than what they report doing – echoes the findings of a study in Thai MSM reported earlier this year, which found that an exclusive preference for other men as sex partners was more strongly associated with having HIV than exclusively reporting sex with men.   

Another finding of this study was that, because a majority of men took both roles and because behaviour did not always go with preference, circumcision would be unlikely to play a significant role in HIV prevention with gay men.


voluntary male medical circumcision (VMMC)

The surgical removal of the foreskin of the penis (the retractable fold of tissue that covers the head of the penis) to reduce the risk of HIV infection in men.


The surgical removal of the foreskin of the penis (the retractable fold of tissue that covers the head of the penis) to reduce the risk of HIV infection in men.


A sexually transmitted infection caused by the bacterium Treponema pallidum. Transmission can occur by direct contact with a syphilis sore during vaginal, anal, or oral sex. Sores may be found around the penis, vagina, or anus, or in the rectum, on the lips, or in the mouth, but syphilis is often asymptomatic. It can spread from an infected mother to her unborn baby.

statistical significance

Statistical tests are used to judge whether the results of a study could be due to chance and would not be confirmed if the study was repeated. If result is probably not due to chance, the results are ‘statistically significant’. 


Insertive anal intercourse refers to the act of penetration during anal intercourse. The insertive partner is the ‘top’. 

It also found – against a background of already very high HIV prevalence – that an association between preferred sex role and age was a primary driver of continued HIV transmission, at least in this group of men; younger people were very much more likely to be ‘bottoms’ than older men, but older men, who preferred to be 'tops', were more likely to have HIV. Age mixing has been suggested as a driver of high HIV incidence in other MSM populations, though this study did not specifically look at what age groups the participants has, or preferred to have, sex with.

The study was primarily designed to find out if circumcision could be a significant HIV intervention in Chinese gay men. It used respondent-driven sampling (RDS), in which MSM attending an STI clinic in Chongqing were asked both to join the study and to refer contacts to it. RDS is a useful way of recruiting hidden populations for studies, but its weakness is that its sample is unlikely to be representative of the population studied as a whole. In the case of this study, fewer older men were included: although the oldest man was 65, the median age was 24 and only 14% were aged over 30.

The researchers recruited 491 men in a two-month period in September 2010. In terms of preferred sex role, two-thirds of men said they preferred taking either role: of the one-third who expressed an exclusive preference, 12.5% said they preferred being bottom and 22.5% said they preferred being top.

There was a very strong correlation between preferred role and age. Among exclusive 'bottoms', 80% were under 25; among exclusive 'tops', 57% were over 25. Only 9% of under-21s were 'tops' and only 4% of over-30s were 'bottoms'.

Age predicted most other associations with preferred sex role: for instance, 'bottoms' were more likely to be students, poorer, have started anal sex younger, and to self-identify as gay. Even circumcision status was tied in with this: more 'bottoms' were in fact circumcised, but this is because circumcision has become more common in China in the last 20 years.

The researchers also asked about sex role with up to three partners during the last six months (fewer if men reported only one or two partners; in fact about 40% of men had only had one partner and only a quarter three or more).

Reported sex behaviour with the last three partners generally correlated with preference; 87% of 'tops', 81% of 'bottoms', and 75% of 'versatiles' had had sex concordant with their preferred role. But this left significant minorities of men who had not done: for instance, nearly 10% of ‘exclusive bottoms’ had also been top on at least one occasion and 12.5% of ‘exclusive tops’ had been bottom. Fifty per cent of 'tops', 56% of 'bottoms' and 65% of versatile men had had unprotected sex on at least one of the three occasions; although this was generally within preferred role, 5% of bottoms had had unprotected insertive sex and 6% of tops had had unprotected receptive sex.

'Tops' were more likely to be consistent within their sex role than 'bottoms': 87% had maintained their reported sex role preference in the last six months compared with 81% of 'bottoms' and 75% of versatile men.

HIV prevalence among the group as a whole was 15%. It was strongly negatively associated with preferred role: 19% of men who preferred being bottom, 17% of versatiles and 7% who preferred being top had HIV. It was also strongly correlated with maintaining one’s preferred sex role: HIV prevalence among men who preferred being 'tops' and who exclusively had been, for instance, was also 7% and it was 20% in preferred 'bottoms' who had been exclusively so.

But it was not correlated with the actual sex people reported having during the last six months, even if men reported exclusively taking one role during this period. HIV prevalence in men who said they’d been exclusively bottom was 14%, in men who said they’d been exclusively top was 10%, and in men who’d taken both roles it was 17%.

This was largely driven by a number of men who said they were versatile but who had only taken one role in the last six months – possibly because of a partner’s preference. What this means, as the researchers comment, is that “interventions must assess preference and not simply enacted sex role, as what someone has been doing in the last six months may be no guide to what they generally prefer doing, given the chance".

This also meant that there was no correlation between circumcision status and HIV prevalence: although circumcised men were 28% less likely to have HIV, this was not statistically significant.

Factors that remained significant were lower educational status (with a nearly five-times-higher risk in men who did not get as far as high school), having three or more partners (twice the risk) and preferring the top role (a quarter of the risk). Preferring to be bottom was associated with a 78% raised risk but because of numbers this was not statistically significant. Interestingly, the small number of men who reported no anal intercourse in the last six months were twice as likely to have HIV: although this was not statistically significant, it also demonstrates that recent behaviour may not be a good guide to historical HIV risk.

Finally, by far the most significant risk factor for HIV infection was syphilis infection. Syphilis prevalence was 5%, with no correlation with preferred or actual sex role, and men with syphilis were 9.6 times more likely to have HIV.        

The researchers comment that “interventions will need to address anal sex roles in more sophisticated ways than originally thought.

“Simplistic assumptions that anal sex role is a fixed behaviour undermines interventions such as circumcision amongst MSM.”


Zhou C et al. Anal sex role, circumcision status, and HIV infection among men who have sex with men in Chongqing, China. Arch. Sex. Behav., early online publication, DOI 10.1007/s10508-012-0008-6, 2012.