The Rhythm of Risk: Sexual Behaviour, PrEP Use and HIV Risk Perception Between 1999 and 2018 Among Men Who Have Sex with Men in Amsterdam, The Netherlands

Current HIV risk reduction strategies include condom use, maintaining undetectable viral load (for individuals living with HIV), regular HIV testing, and use of pre-exposure prophylaxis (PrEP). These strategies are effective means of reaching the goal of zero HIV infections among men who have sex with men (MSM). Recent research showed that MSM who had a higher perceived risk of acquiring HIV were more interested in taking PrEP [1‐3], showed higher PrEP adherence [4], and were more likely to get tested for HIV [5, 6]. Conversely, unrealistically low risk perception was found to be an important barrier for successful PrEP implementation [7] and HIV testing [8, 9], and was one of the main reasons why patients with late HIV diagnosis did not get tested earlier [10]. These studies show that in order for MSM to take preventive measures, a realistic perception of HIV risk is crucial. To improve uptake of preventive strategies, further understanding is needed as to what factors influence risk perception.

Sexual behaviour and other risk factors for acquiring HIV, including drug use and history of sexually transmitted infections (STI), have been found to be associated with higher risk perception in MSM, although specific findings have been mixed [1, 7, 11, 12]. For example, Wilton et al. found that high risk perception was associated with past history of bacterial STI and a high score on the HIRI-MSM, an HIV risk screening tool based on number of partners, frequency of condomless anal intercourse (AI) and use of drugs [7]. Biello et al. found that high risk perception was associated with a high number of receptive AI partners, but not with an STI diagnosis in the past year [1]. Kesler et al. found that high risk perception was associated with low condom use with a regular partner living with HIV and with use of poppers, but, in contrast to Wilton et al. [7], was not associated with high HIV risk (i.e. high score on the HIRI-MSM) [11]. These inconsistent findings may be explained by different definitions of risk perception, i.e. perceived risk of having acquired HIV [12], current perceived risk [1, 7] or perceived risk to acquire HIV in the future [11]. Also, differences in sexual behaviour measures and study population characteristics may have played a role. Importantly, these studies were cross-sectional and did not take into account possible longitudinal changes in risk perception and the mutual relationship with sexual behaviour. Little is known about how risk perception among MSM has evolved in the past decades, especially in light of the developments in treatment and prevention, fluctuations in HIV incidence, and changes in sexual behaviour within the population.

Since the introduction of ART in 1996, MSM in Western countries have been more likely to have condomless AI [13, 14] and to have more casual sex partners [15, 16]. Risk behaviours have continued to increase [17, 18] and have been accompanied by increasing STI incidence [15, 18, 19]. From 2000 onwards, the number of new HIV diagnoses increased among MSM in Western countries [20, 21], yet, in the past decade, numbers of diagnosed HIV infections have been decreasing in the Netherlands [18, 22], most likely due to increased testing and early treatment. The evolution of sexual behaviour, STI and HIV incidence, and HIV treatment and prevention may have affected perceived risk to acquire HIV among MSM and in turn, could influence the willingness of MSM to partake in preventive measures.

The aims of the current study were to examine (1) the course of HIV risk perception among HIV-negative MSM over the past 20 years, and (2) how sexual behaviours, recent history of STI, and PrEP use are related to risk perception and whether these associations changed over time. Risk perception was defined as the perceived likelihood of acquiring HIV in the past 6 months. We aimed to study several behavioural characteristics, including partner type, numbers of partners, condom use, chemsex, and AI during group sex. This study was performed using longitudinal data from the Amsterdam Cohort Studies (ACS) among MSM collected between 1999 and 2018.

Among MSM participating in the Amsterdam Cohort Studies (ACS), we found that perceived risk of HIV has fluctuated at the population level in the past 20 years. Based on relative differences in risk perception between waves and the overall grand mean, we distinguished five alternating periods of relatively low and high perceived risk of HIV. Throughout these time periods, increasing number of receptive and insertive AI partners and having condomless AI with casual partners were consistently associated with higher risk perception. In contrast, condomless receptive AI with a partner living with HIV was no longer associated with perceived HIV risk in 2017–2018. MSM who had condomless AI with a steady partner initially were more likely to have a higher perceived risk, but since 2011 it was associated with lower risk of acquiring HIV.

Our study has major strengths. We used data from the ACS, a prospective cohort study including a large group of MSM with long periods of follow-up. We were able to study the course of risk perception over a time span of 20 years, covering a period in which important developments occurred. Extensive behavioural information was collected in the ACS, allowing us to study the individual contribution of a range of sexual behaviours associated with perceived risk of HIV. This information was reinforced with STI testing results, which were available since 2008. Despite these strengths, our findings must be interpreted in the context of some limitations. In the ACS, highly educated MSM are overrepresented. Additionally, participants were recruited by convenience sampling and chain referral sampling. By participating in a study on HIV, the participants may, by definition, be more aware of risk of acquiring HIV than MSM in the general population. This may limit generalizability to the entire Dutch MSM population. However, behavioural outcomes from the ACS have been previously found to be similar to larger nation-wide monitoring studies, such as the Schorer Monitor [25]. Furthermore, repeatedly filling in questionnaires covering HIV related topics may have raised awareness of developments in the HIV field. The time taken before MSM in the general population become aware of these developments, and their ability to impact the perceived risk of HIV, could be longer. Another possible limitation is that the ACS is an open cohort wherein participants can leave and enter the cohort over time. We cannot completely rule out the influence of the changing study sample on longitudinal changes in risk perception. We attempted to minimize this bias by controlling for age and year of study entry in our analyses. We also reran our analysis among MSM who participated in the cohort for at least 6 years, therefore contributing to both a period of relatively low and relatively high risk perception, and identified the same pattern of change in risk perception. Finally, ordinal logistic regression requires that the effect estimates are equal across categories of risk perception. This proportional odds assumption did not hold for the models presented here, indicating that the strength of the sexual behaviour effect estimates differed across the risk perception scale. However, applying a binary logistic regression approach using different thresholds to dichotomize risk perception [26], resulted in the same conclusions.

The periods of low and high risk perception that we distinguished were based on wave specific differences from the grand mean. Between 1999 and 2008, these periods also included waves in which risk perception did not significantly differ from this mean. From 2008 onwards a clearer pattern arose. Here we describe how these fluctuations from 2008 onwards may be interpreted in light of developments in HIV treatment and prevention and patterns of HIV incidence in the Netherlands, keeping in mind that no conclusions regarding causality can be sufficiently drawn. The low risk perception period between 2008 and 2011 may reflect optimism resulting from the first studies indicating that HIV transmission is low when a partner living with HIV has an undetectable viral load, as was reported in the Swiss statement in 2008 [27]. The high risk perception period between 2011 and 2016 may be related to the high annual numbers of new HIV diagnoses among MSM during these years [22]. This may have raised concerns of HIV risk via widespread media and recognition of more HIV diagnoses within a person’s network. Also, promoting regular testing and the implementation of an opt-out HIV testing policy at all STI clinics in the Netherlands in 2010 [28, 29], together with increasing sexual risk behaviour and increasing STI incidence [18], may have increased risk perception among MSM in this period. In 2017–2018, risk perception was relatively low and might have been related to optimism about the impact of biomedical prevention strategies for HIV, including treatment as prevention and PrEP use. In 2016, PrEP was approved in Europe and the Undetectable equals Untransmittable (U = U) consensus statement was issued [30], indicating that people living with HIV on ART with an undetectable viral load in their blood do not sexually transmit HIV. The U = U statement may have made MSM more confident to rely on undetectable viral load of their partner living with HIV. This is supported by our findings that MSM became more likely to have condomless receptive AI with a partner living with HIV and, in 2017–2018, this did not influence their perceived risk. Since 2015, an increasing number of ACS participants reported PrEP use [24]. We showed that PrEP users in the ACS have a lower perceived risk. Lowered perceived risk of acquiring HIV could also stem from individuals having sex with a person using PrEP [31].

The consistent associations between risk perception and numbers of casual AI partners or having condomless AI with a casual partner, which are well known risk factors for HIV [32], support the hypothesis that MSM have, overall, an adequate risk assessment. The finding that having condomless receptive AI with a partner living with HIV is no longer associated with higher perceived risk could also signify ‘adequate’ risk assessment, under the assumption that MSM only have unprotected sex with partners with an undetectable viral load. We also found that, in 2017–2018, a recent STI was not associated with increased risk perception, which can be considered adequate since with PrEP use and viral load sorting, risk behaviour for a bacterial STI does not necessarily coincide with a high risk for HIV. Strikingly, however, condomless AI with a steady partnership has been related to a lower risk perception since 2011. Studies using data collected before 2011 suggest that the steady partner is an important source of infection, with up to two thirds of infections attributed to sex with a steady partner [33‐35]. More recent studies in Western countries suggest a smaller role of the steady partner [36, 37]. An Australian study found that only 10% of recently infected MSM believed they acquired HIV from a steady partner [37]. In the last 15 years, preventive interventions in the Netherlands have promoted negotiated safety, whereby couples agree to always use condoms or PrEP when having AI with casual partners, together with regular HIV testing [38]. Practicing negotiated safety or having a monogamous relationship with an HIV-negative partner may be possible explanations why condomless sex with a steady partner is associated with lower perceived risk. However, MSM who are in a steady partnership and underestimate the risk of their partner’s behaviour may be a target group for prevention.

Currently, we are in a period of low risk perception, most likely influenced by the availability of many risk reduction strategies. If the actual risk of acquiring HIV becomes smaller in coming years, due to successful implementation and scale up of biomedical prevention strategies allowing to reach the 95-95-95 targets set by UNAIDS, risk perception may continue to stay low or decrease even further. Alternatively, low risk perception may make MSM indifferent about acquiring HIV, which may result in lower PrEP uptake, lower adherence, and lower condom use. This could potentially give rise to new HIV infections and subsequently lead to a period of higher risk perception. Monitoring risk perception among MSM will therefore be important for future prevention. To prevent new infections, intervention strategies should target people who underestimate their own risk and continue to engage in risky behaviour. Such approaches should take into account that assessment of HIV risk is complex and condomless AI can be considered as non-risky if biomedical prevention strategies or negotiated safety have been successfully implemented.

Apart from the relationship between individual behaviour and risk perception, which was the focus of the current study, other environmental factors may also influence a person’s risk perception, including attitudes and perceptions of HIV risk within a person’s network and within the larger societal context. Future research may focus on these environmental factors. Another direction for future research may be to examine the mediating role of risk perception in the association between sexual behaviour and risk reduction strategies. In studying this mediating role, factors that may limit risk reduction behaviour despite adequate risk perception will need consideration, including stigma, social barriers, social inequality and social disparity.

In conclusion, our study showed that HIV risk perception among MSM has fluctuated in the past 20 years. U = U and PrEP use may have resulted in lower perceived risk in 2017–2018. As long as MSM adhere to current HIV risk reduction strategies, including condom use, regular HIV testing, viral load sorting, and use of PrEP, low risk perception is in agreement with their actual risk. Awareness of MSM who underestimate their risk and become careless in taking preventive measures is needed to stop further transmission.