In this study, we analysed the responses of a large number of European MSM in order to determine what factors were associated with reporting UAI with four or more non-steady male partners in the past 12 months. Most factors identified as being significantly associated with UAI with four or more partners were fairly consistent across both multivariable models, with more and larger differences typically existing when men with fewer non-steady partners were taken to be the comparison group. Therefore, many risk factors in our models increased in impact with an increase in the number of non-steady male partners with whom a man engages in UAI. When comparing the group reporting UAI with 1-3 non-steady partners in the past 12 months with men not reporting UAI with non-steady partners it differs from the respective comparison of the 4+ group. The 1-3 UAI partner group consists of more younger men still in education and men living outside of the large cities having less access to gay sex venues.
Of note, with the exception of Turkey, the country of residence did not greatly impact the probability of reporting UAI with multiple non-steady partners, despite the wide range of legal and social attitudes toward homosexuality in the countries surveyed. However, as laws and social attitudes toward homosexuality are likely to influence some of the individual-level factors in our model, such as experience of physical violence or exposure to HIV-related knowledge, the indirect impact of oppressive laws and general discrimination should not be discounted. Turkey differs from the other countries in the sample by being the only major country with a Muslim majority (the other being Bosnia and Herzegovina, with only 72 respondents), and consequently a different cultural background. Stigmatization and discrimination of MSM in these two countries is high, but this is also the case for several other countries in the sample, e.g. Russia, Bulgaria, and Romania [15, 16]. Thus, we have no ready explanation for this finding, except that there is very little HIV prevention messaging for MSM in Turkey and a very weak “gay community”. Additional research will be necessary to explain this finding, particularly since HIV prevalence – although self-reported – is low in the Turkish sample (3 % among Turkish EMIS participants ever tested for HIV).
Even after exclusion of HIV-positive men with undetectable viral loads and those who consistently practice HIV-positive serosorting, men who reported a previous HIV diagnosis were significantly more likely to report UAI with 4+ partners than men who had never been diagnosed. This finding is consistent with previous research [17–19], and may reflect selection (e.g. use of gay dating sites for recruitment) and self-selection (e.g. men interrupting sexual activity after HIV diagnosis may select not to participate in such a survey) effects in our sample, a continuation of or a return to risky pre-infection sexual behaviour, or may be explained by a lack of perceived relevance of restriction of partner numbers as an HIV prevention strategy once HIV has been diagnosed. To mitigate the high levels of non-seroconcordant UAI reported by men diagnosed with HIV, it is necessary that prevention campaigns continue to target men diagnosed with HIV, and not only presumably uninfected men. As emphasized in the 2013 WHO HIV treatment guidelines [20], the prevention effects of early initiation of antiretroviral treatment should be maximised by prompt diagnosis followed by offering immediate and affordable treatment. Since the START trial provided evidence that early antiretroviral treatment is also beneficial for maintaining the health of the individual [21], public health and individual health benefits are not conflicting goals, which can be emphasized when discussing treatment options with newly diagnosed HIV patients. Policy makers should reconsider policies that recommend waiting until CD4 counts have dropped to a certain level before initiating treatment, and instead recommend treatment immediately upon diagnosis.
Despite previous evidence suggesting a protective effect of higher age [22, 23], we found that reporting UAI with 4+ non-steady partners is associated with higher age groups when compared to reporting UAI with 1-3 non-steady partners., This probably reflects that the 4+ group is rather a segment of MSM with established gay identity and having adopted a specific type of gay urban lifestyle, while the 1-3 UAI partner group represents more young MSM with a more fragile sexual identity, who are not yet integrated in the gay subculture. These two groups need probably quite different prevention approaches.
Additionally, when compared to men reporting no UAI with non-steady partners, being in a steady relationship was found to be protective, especially when men also reported being happy with their sex lives. If older men are more likely to be in steady relationships, this relationship may account for the lack of significance of age in this model. While this result is unsurprising, it is important to note that steady relationships are not necessarily free of HIV risk [24]. It is therefore crucial that public health campaigns do not ignore men with steady partners, and that a range of Safer Sex Messages for these men, including negotiated safety [25], are addressed. Furthermore, it is important to note that, among men who were not in a steady relationship, the probability of reporting 4+ partners was greater for men reporting sexual happiness in both models. Future public health campaigns could emphasize the possibility of having a full, satisfying sex life while still protecting oneself and others from HIV and other STIs.
In agreement with previous research [26, 27], we find that men who visit sex-on-site venues are more likely to report more UAI with non-steady partners. Strategies such as distribution of condoms and safe sex information at sex-on-site venues may be capable of playing a significant role in reducing UAI.
Use of nitrite inhalants, drugs associated with sex and parties, and erectile dysfunction drugs was also consistently associated with increased probability of reporting UAI with 4+ non-steady partners. The association between drug use and risky sexual behaviour has been well documented [28–30]. Drugs associated with sex and parties are typically used to enhance sociability and feelings of euphoria. Some of the newer drugs can also trigger feelings of intense sexual arousal. This crucial, additional effect has led to the use of these drugs by MSM in sexual contexts, a behaviour often referred to colloquially as ‘chemsex’ (also called ‘Party and play’), depending on the exact drugs used. The drugs are frequently combined and are typically consumed during prolonged sexual sessions, which can involve multiple sexual partners [31]. Drugs associated with sex and parties are also often combined with the more commonly used sexual performance drugs, which include both nitrite inhalants and erectile dysfunction drugs and are used to enhance and maintain erection and, in the case of nitrite inhalants, to relax the anal sphincter. Considering these contexts, the association of multiple UAI partners with use of these specific drugs is not surprising.
Our results indicate that selling and buying sex are generally associated with higher numbers of UAI partners. However, as a detailed report of the associations between transactional sex and associated health outcomes in the EMIS dataset is currently in preparation, we do not expand upon the implications of this finding here.
Additionally, we find that experiencing physical violence, and perhaps even verbal abuse and intimidation, due to sexual attraction to men was associated with reporting UAI with multiple non-steady partners. It is possible that individuals engaging in UAI with more non-steady partners may tend to visit gay sex venues more frequently, and are therefore more likely to be recognized as MSM and attacked. The more sexual partners one engages with, especially in sex-associated venues as shown in our data, the higher the likelihood to be seen outside gay venues and identified as gay by passers-by. Very few studies have previously examined the link between physical violence and HIV-related sexual risk behaviour, although Wheeler et al. found that experiencing physical violence was associated with both reporting multiple partners and STI diagnosis [32], and Santos et al. found increased risk of both UAI and HIV infection among MSM who had experienced violence due to their homo- or bisexual orientation [33]. Future research should look more closely at the association between experienced physical violence, verbal abuse, and risk behaviour.
Our consistent finding of a protective effect of higher education was in line with previous research [22]. Being exposed to information on HIV or STIs specifically designed for MSM was also associated with lower risk of reporting UAI with 4+ non-steady partners compared to no UAI partners, indicating the continuing importance of HIV education campaigns in reducing risk behaviours in this population. However, results also suggest that knowing that ART reduces HIV transmissibility is associated with higher probability of reporting UAI with 4+ non-steady partners. But this effect was very small, probably because men diagnosed with HIV and having an undetectable viral load had been excluded from our analysis. This association has been observed before, and it has been suggested that knowledge and opinions of ART are more influential than actual treatment status [34]. Still, it remains important that individuals with new HIV infections are promptly identified and given access to treatment. It should also be noted that the EMIS data were collected in 2010, and that treatment practices in many European countries likely have evolved towards earlier treatment start.
Men reporting UAI with 4+ partners were more likely to feel lonely. Meanwhile, engaging in UAI with many partners may help to temporarily reduce feelings of loneliness and enhance feelings of intimacy [35]. Thus, it may be worthwhile to invest in approaches which enable men to improve the quality of their sexual relations instead of increasing the quantity. Finally, being out to all or most of the people a participant knew was positively associated with reporting 4+ UAI partners in both models, possibly because individuals who are more out are also more comfortable seeking out sexual partners.
Rather than focusing only on whether or not an individual reports any UAI, we analyse here the factors associated with engaging in UAI with four or more non-steady partners specifically. While this makes it difficult to compare our findings to previous research, this approach is a major strength of our study in light of research showing that simple measures of “any UAI” do not accurately capture sexual risk behaviour [3, 36]. Additionally, our large dataset allowed us to consider a great number of variables in our models, and the wide geographic range of our data allowed us to analyse both individual- and country-level factors.
Limitations
However, we also acknowledge various drawbacks of our study. Because our study was exploratory in nature and considered such a large number of variables, it is difficult to state with certainty that these variables will still be important indicators of sexual risk in other datasets or populations. Additionally, the large number of variables analysed increases the risk that some apparently significant results may actually be due to Type I errors. Furthermore, after the removal of HIV-diagnosed men with undetectable viral loads or consistent serosorting behaviour, we did not have the statistical power to ascertain how and to what extent specific variables may impact HIV diagnosed men in a different way. The change of behaviour after subjects became aware of their HIV status may not uniformly occur in the interval between the time of diagnosis and the time for interview. The analysis does not capture the dynamic changes of behaviour in different time intervals. We also note that our survey contained only very few items to assess mental health, which has been previously found to be associated with sexual risk behaviour and HIV infection [33, 37]. Due to the cross-sectional nature of our survey, we are also unable to elucidate any causal information, and temporal ambiguity bias cannot be eliminated. Finally, as with all self-reported data, there is likely to be some degree of recall bias present, and the large number of questions results in some inconsistency in responses. We also note that, since the survey was conducted online, men without internet access were unable to participate, likely leading to selection bias.