Article Text

Download PDFPDF
Original article
High-risk sexual behaviours among gay and bisexual men: comparing event-level casual sex encounters among seroconverters and non-seroconverters
  1. Ian Down1,
  2. Jeanne Ellard2,
  3. Kathy Triffitt1,
  4. Iryna Zablotska1,
  5. Michael Hurley2,
  6. Graham Brown2,
  7. Jack Bradley1,
  8. Garrett Prestage1
  1. 1 The Kirby Institute, University of New South Wales, Sydney, NSW, Australia
  2. 2 Australian Research Centre in Sex Health and Society, La Trobe University, Melbourne, VIC, Australia
  1. Correspondence to Ian Down, Research Associate, The Kirby Institute, University of New South Wales, Sydney NSW 2052, Australia; Australian Research Centre in Sex Health and Society, La Trobe University, 215 Franklin Street, Melbourne VIC 3000, Australia; idown{at}


Background With increasing use of non-condom-based HIV risk reduction strategies by gay and bisexual men (GBM), we compared occasions of condomless anal intercourse with casual partners (CLAIC) that resulted in HIV transmission and similar occasions when HIV transmission did not occur.

Methods We compared two demographically similar samples of Australian GBM. The HIV Seroconversion Study (SCS) was an online cross-sectional survey of GBM recently diagnosed with HIV. The Pleasure and Sexual Health (PASH) study was an online cross sectional survey of GBM generally. Using logistic regression, we compared accounts of CLAIC reported by men in SCS as being the event which led to them acquiring HIV, with recent CLAIC reported by HIV-negative men in PASH.

Results In SCS, 85.1% of men reported receptive CLAIC, including 51.8% with ejaculation; 32.1% reported having previously met this partner and 28.6% believed this partner to be HIV-negative. Among HIV-negative men in PASH reporting recent CLAIC, 65.5% reported receptive CLAIC, including 29.9% with ejaculation; 59.3% reported having previously met this partner and 70.1% believed this partner to be HIV-negative.

Conclusions While both groups of men engaged in CLAIC, how they engaged in CLAIC differed, and the context in which they did so was different. A generic measure of CLAIC conceals the critical elements of HIV risk, particularly the role of receptive CLAIC, among GBM that distinguish those who seroconverted and those who did not. Detailed information about the context and nature of the practise of CLAIC is required for a more complete understanding of HIV risk among GBM.

  • HIV

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Condomless anal intercourse between casual male partners (CLAIC) has been identified as the main predictor of trends in HIV diagnoses among gay and bisexual men (GBM) in Australia and internationally.1 ,2 The proportion of GBM who report CLAIC has been increasing since the early 2000s in Australia and elsewhere,3 ,4 with corresponding increases in HIV incidence among this group.5 ,6 Since the mid-1990s, research has increasingly noted the use of non-condom-based HIV risk reduction strategies (RRS) among GBM seeking to reduce the risk of HIV transmission during condomless anal intercourse (CLAI) such as serosorting, strategic positioning and withdrawal.7–11 Changing perceptions among HIV-negative GBM about the likelihood and consequences of HIV infection may be one important factor in their use of RRS.12

RRS, applied correctly, can offer varying degrees of protection against HIV.13 So, not all CLAIC is equally risky. A Sydney-based cohort study of HIV-negative GBM found that those men who restricted CLAI to only seroconcordant HIV-negative partners (serosorting) were 2.2 times more likely to seroconvert than men who reported no CLAI, whereas those who reported any CLAI with HIV-positive partners were 16.1 times more likely to seroconvert than those who reported no CLAI.14 Those men who practised insertive CLAI only (strategic positioning) were not at significantly higher risk of infection than those who practised no CLAI; while men who engaged in receptive CLAI were 4.8 times more likely to seroconvert than those who practised no CLAI. Withdrawal, the strategy involving the insertive partner withdrawing his penis prior to ejaculation during intercourse, still represented a substantial risk of HIV infection.14

Most GBM in Australia appear to have a good understanding of these varying levels of risk associated with different anal intercourse practises,15 and a sizable minority appears to practise RRS purposefully.16–18 However, the extent to which these strategies are correctly and consistently applied is unclear.18 In Australia, around two-thirds of HIV transmissions among GBM occur in the context of casual sex encounters. To be applied effectively, RRS rely on both partners having accurate knowledge of their HIV status, and on their ability to effectively communicate this to each other. Most men do not routinely disclose their HIV status to casual partners,4 so they often rely on assumptions about their partners' HIV status.17 Prior acquaintance with casual partners has been associated with likelihood to engage in CLAIC, leading some to suggest that familiarity may increase men's risk of HIV.19 Multiple and simultaneous partnering has also been associated with increased risk.20 ,21 CLAIC occurs at elevated rates in the context of group sex, and group sex has been identified as a key factor in HIV infection.22

Previous studies of recently HIV-diagnosed GBM in Australia have examined the high-risk event the men believe led to their infection.16 In one analysis of recently HIV-diagnosed GBM, evidence of attempts to use RRS was noted for one in three men, at the time their infection occurred.16 Incident-level comparisons of RRS have been rarely reported, particularly comparisons of men who were subsequently diagnosed with HIV and men who remained uninfected. We hypothesised that the risk behaviour in a sample of recent seroconverters would show little evidence of the use of RRS compared with that in a sample of non-seroconverters who engaged in similar risk behaviours to those of the seroconverters. This paper examines the detailed accounts of CLAIC in which HIV transmission occurred, and compares these with accounts of CLAIC among men who remained HIV-negative, to assess how more specific measures of CLAIC may be more informative of the context of risk and therefore more useful for prevention purposes.


This paper compares data from two studies: the HIV Seroconversion Study (SCS) and the Pleasure and Sexual Health (PASH) study. These two studies were developed and implemented by the same study team using similar recruitment and interview methods. SCS commenced in 2007, while PASH was conducted concurrently during 2009. Many of the same questions were asked in both studies, and the same community stakeholders were associated with both studies. Ethics approval for both studies was obtained from the University of New South Wales (SCS approval number HREC 10305, PASH approval number HREC 07207) and La Trobe University Human Ethics Committees (SCS approval number FHEC07-193, PASH approval number FHEC07-170).

HIV Seroconversion Study

The SCS included an online survey of people in Australia who had recently been diagnosed with HIV. Eligibility criteria for the study were: living in Australia; being over 16 years of age and having been diagnosed HIV-positive within the 2 years prior to enrolment. Eligible participants were referred to the study website by diagnosing doctors, HIV community organisations or by direct online self-referral.

The questionnaire collected demographic characteristics, details of participants' diagnosis with HIV and sexual behaviours. Participants were asked to identify an event that they believed had resulted in their HIV infection, and to describe the sexual practises that occurred at this event as well as information about the partner they believed to be the source of their infection. Data collection began in 2008 and is ongoing. The study methods have been described in more detail elsewhere.23 Between 2008 and 2013, 506 recently HIV-diagnosed GBM were enrolled into the study.

Pleasure and Sexual Health Study

PASH involved an online survey of 2306 Australian GBM conducted in 2009, exploring their sexual behaviour and beliefs about HIV and the risks of HIV transmission; the methods are described elsewhere.12 Men in PASH were asked to describe their most recent episodes of both condomless and condom-protected anal intercourse with casual partners.


Both SCS and PASH involved online survey questionnaires. Men in SCS were asked to describe their relationship to the person they believed to be the source of their infection, with a response option being ‘casual partner’. Men defined the concept of casual partner themselves. In PASH, men were asked if they had an occasion of CLAI with a casual partner in the previous 6 months. Again, men defined the concept of casual partner themselves. In both questionnaires, the men were asked about their prior acquaintance with that partner, and their belief about his HIV status at that time: ‘what did you believe was his HIV status at that time?’ with response categories: ‘HIV-positive’, ‘HIV-negative’ and ‘I did not know’. The men were asked about the location of the event they were describing, and if it occurred in the context of group sex, as well as being asked to describe the sexual practises that occurred, in terms of receptive and insertive anal intercourse, as well as whether ejaculation occurred in the rectum.

The sample

Of the 506 GBM enrolled in the SCS, 473 (93.5%) were able to identify a high-risk event they believed led to their infection with HIV. Of those men, 257 (54.3%) describe the source of the infection at that high-risk event as a casual partner. Of those latter men, 200 (77.8%) describe engaging in CLAIC on that occasion. Among the 2306 men who participated in PASH, 1738 reported that they were HIV-negative, including 338 (19.4%) who described an occasion of CLAIC that had occurred within the previous 6 months. A larger proportion of men in PASH were not gay identified (32 identified as ‘bisexual’, 2 as ‘other’ and data were missing for 19 men) than in SCS (10 identified as ‘bisexual’, 1 as ‘other’, with data missing for 1 participant). Given these differences, and that gay and non-gay men often have different behaviours that may have affected the results,21 we restricted our analyses to gay-identified men in each sample. There was a clustering of men aged 60 years and over in the PASH sample (16 compared with two in SCS). Also, 96 men in PASH and 18 men in SCS did not provide their age. So, to improve comparability, we restricted the sample to those who specified their age as being under 60 years. After these exclusions, 169 men in SCS and 194 men in PASH were included in the analyses. These exclusions ensured that as far as was practicable, we were comparing samples with similar characteristics.


Data were analysed in SPSS software (IBM SPSS Statistics V.22.0, IBM, Armonk, New York, USA). Significant differences between seroconverters (men in SCS) and non-seroconverters (men in PASH) were determined by univariable χ2 tests or analysis of variance tests. Univariable logistic regression was used to determine significant differences between seroconverters and non-seroconverters on items previously found to be associated with HIV risk behaviour and infection, including: group sex21–22; prior acquaintance with the sexual partner19; knowledge of partners' HIV serostatus14–15 ,17 and sexual position.9–10 ,14 Univariable and multivariable logistic regression was used to determine associations between the key independent variables and subsequent HIV seroconversion. Multivariable models were developed using forward stepwise techniques. We present unadjusted ORs or adjusted ORs (aOR), and their corresponding 95% CIs for these associations.


The demographic profiles of the two samples were broadly similar: the mean age in both was around 34 years (SCS 33.3, SD 8.47 vs PASH 35.0, SD 10.2, p=0.091), and half had a university education (SCS 89/168 vs PASH 99/194, p=0.712). The men who seroconverted were more likely to be of Anglo-Australian background than the PASH participants (SCS 135/168 vs PASH 123/194, p<0.001) (table 1). PASH participants were asked if they had tested for HIV since the CLAIC event they described: one-third (32.5%) reported having been tested, and that they had tested HIV-negative, while the remainder had not tested since the event. We compared the demographic and behavioural characteristics of these men with those who had not tested for HIV between the time of the CLAIC event and completing the survey and found no difference between the two groups (data not shown).

Table 1

Comparing the men in SCS and PASH, and their CLAIC* events

The men who seroconverted were more likely to have met their CLAIC partner for the first time at the time of the event (SCS 114/168 vs PASH 79/194, p<0.001) (table 1). When asked what they believed the partner's HIV status to be at the time, the seroconverters were less likely to report that they believed this partner to be HIV-negative than were the men who remained HIV-negative (SCS 48/168 vs PASH 133/194, p<0.001).

The location in which the sexual risk event occurred differed between the two groups (table 2): among the seroconverters, less than half of the CLAIC events occurred in a private home (most commonly that of the casual partner), while for more than two-thirds of the PASH participants the event occurred in a private home, and, more commonly, their own home (p<0.001). More than a third of the men who seroconverted describe their CLAIC as occurring in a public sex environment, while this was the case for one in six of the PASH participants (SCS 64/168 vs PASH 33/194, p<0.001). The CLAIC events among seroconverters were much more likely to have occurred in the context of group sex (SCS 62/168 vs PASH 17/194, p<0.001) (table 1).

Table 2

Location of CLAIC* event

When describing the sexual practises that occurred during the event, the men who seroconverted were more likely to have been the receptive partner during CLAIC (SCS 143/168 vs PASH 127/194, p<0.001), and more likely to have had their partner ejaculate inside them (SCS 87/168 vs PASH 58/194, p<0.001), while the PASH participants were more likely to have restricted CLAIC to the insertive position only (SCS 25/168 vs PASH 67/194, p<0.001) (table 3).

Table 3

Anal sex practises at the CLAIC* event


Our study compares similar event-level data between men who acquired HIV and those who did not in two comparable and concurrent studies. There have been few previous such studies, particularly with detailed contextual information about the circumstances of the event or the relationship to the sex partners involved. The RRS described in these analyses are not equally effective,14 and the men in PASH who remained HIV-negative appear to have more often applied those strategies that are more effective at preventing HIV transmission. In comparing the CLAIC events, there is little evidence that many men who subsequently seroconverted were employing RRS at the event they attribute to their infection. Specifically, the men who seroconverted were far more likely to have engaged in receptive CLAIC than were the PASH participants. Receptive CLAI plays a considerably greater role in HIV infection among GBM than does insertive CLAI.13

Although the two studies were conducted concurrently and the participants were broadly similar to other samples of Australian gay men,12 ,14 these samples may not be representative of all homosexually active men in Australia and were restricted by age and sexual identity. There may be differences between the Australian context and other equivalent countries. The men in PASH may have subsequently seroconverted, following their CLAIC event, although one-third reported having tested HIV-negative since that event. We cannot account for the men's knowledge and consideration of the testing window period. The data may be subject to recall bias: the men in SCS described an event they believe led to their HIV infection, and so may have reflected on their partner's HIV status differently than the men in PASH. Also, these analyses occurred prior to the introduction of pre-exposure prophylaxis and widespread recognition of the role of treatment as prevention, which are significantly changing the landscape of sex and HIV risk.

From these data, we are unable to determine if the sexual positioning described is evidence of risk reduction being applied intentionally, or whether the men were simply acting out their preferences for either the receptive or insertive roles—with that partner at that time. Nonetheless, PASH participants who were the receptive partner more often reported their partner withdrawing prior to ejaculation than was the case among the seroconverters, further suggesting that the PASH participants applied strategies to their CLAIC that may have helped to reduce their risk of HIV infection. Withdrawal has not, however, been demonstrated to be an effective risk reduction strategy.14 We could not determine from our data whether withdrawal was being used as a deliberate HIV-risk reduction strategy, or was simply a preference for where ejaculation occurs.

Increased familiarity with casual partners has been found to be associated with an increased likelihood to engage in CLAIC.19 Such findings have informed education campaigns designed to discourage men from trusting what their partners tell them about their HIV status.24 Our finding that men who seroconverted were less likely to have had prior acquaintance with their CLAIC partners than were those who did not seroconvert suggests that familiarity may, however, offer a somewhat protective role for some men. A more nuanced approach to the value of trust based on demonstrated knowledge and familiarity with partners may be warranted. The issue of familiarity raises questions about the concept of ‘casual partner’ and its use in research; in some cases, a casual partner might be well known, while in other cases they may be someone completely unknown.

The PASH participants were also more likely to engage in CLAIC in their own home, where they may have felt in greater control, than was found among the men who had seroconverted. ‘Home’ may provide an additional familiarity factor. This may also explain why PASH participants were more likely to have used RRS in those circumstances. Consistent with previous findings, the men who seroconverted were also more likely to have engaged in group sex.22

Consistent use of condoms for anal intercourse with casual partners has been the hallmark of HIV prevention education aimed at GBM in Australia. Negotiated safety has been endorsed and promoted in Australia as a strategy for men to use in the context of a regular relationship. For sex with casual partners, occasional examples of other RRS have been acknowledged, although usually discouraged.24 Recently, there has been a shift at the policy level to recognise the need for a sustained engagement with RRS.25 ,26 Some recent health promotion resources address RRS in a more nuanced way to discuss relative risk, informing men about the risk associated with different anal sex acts and RRS options.27

In a rapidly changing HIV prevention landscape, the ability to accurately measure and report levels of risk, and evaluate the effectiveness of health promotion that engages with RRS, may be compromised by an over-reliance on a single, undifferentiated concept of CLAIC as the general measure of HIV risk. Recently, members of our team have argued that this single direct measure of CLAIC does not account for HIV risk in the context of recent developments in HIV prevention.28 Our data indicate that receptive CLAI may contribute substantially more to the circumstances that lead to HIV infection than has tended to be implied by HIV prevention work to this point. Also, whether ejaculation occurs in the rectum, the location where the CLAIC occurs, familiarity with partner and group sex appear to play a key role in the circumstances that result in HIV infection among GBM. A more nuanced consideration of the role of CLAIC as a marker of risk is needed. The sexual partnerships and networks in which CLAIC events take place are highly determinative of any associated risks.

It is crucial to the control of HIV that we maintain the ability to accurately measure and report levels of risk. With an increasing minority of GBM now choosing to occasionally forgo condoms in particular circumstances, some men appear to be seeking to manage their relative risk, while others are not. Clearer information on the effective use of non-condom-based risk reduction, and the contexts in which that can be managed better, may help some men protect themselves and their partners from possible HIV infection. Our findings suggest likely contextual factors predicting HIV acquisition in the context of CLAIC among GBM. Longitudinal studies that compare such event-level data between recent seroconverters and appropriately selected controls are needed.

Key messages

  • Condomless anal intercourse with casual partners, on its own, is an insufficient measure of sexual risk behaviour among gay and bisexual men (GBM).

  • Sexual positioning and other key contextual information need to be incorporated in analyses of sexual risk behaviour among GBM.

  • Seroconcordance between casual male partners may be more reliably established when they have some prior acquaintance with each other.

  • More than a third of recently HIV-diagnosed men report engaging in group sex at the time they believe they acquired their infection.


The authors thank the participants of both studies for giving their time. We also thank the many staff in community organisations and clinics who helped in the recruitment of participants to both studies.



  • Handling editor Jackie A Cassell

  • Contributors GP was the Principal Investigator of both the HIV Seroconversion Study (SCS) and the Pleasure and Sexual Health (PASH) study. GP, JE, KT, GB, ID and JB designed the study protocols and data collection instruments for SCS. GP, MH, GB, ID and JB designed the study protocols and data collection instruments for PASH. ID analysed the data, with support from GP and IZ. ID wrote the first draft of this manuscript. All authors contributed to and approved the final manuscript.

  • Funding The HIV Seroconversion Study was funded by the Health Departments of: New South Wales, Victoria, Queensland, Western Australia, South Australia, Tasmania, the Northern Territory and the Australian Capital Territory. The Pleasure and Sexual Health study was funded by the Health Departments of: New South Wales, Victoria, Queensland, Western Australia and South Australia.

  • Competing interests None declared.

  • Ethics approval UNSW Human Ethics Review Committee.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement Data can be made available on request.