Partnership-based sexual behaviours among adults in Singapore: a cross-sectional population survey

Abstract

Introduction

Sexual behaviour data are essential for monitoring sexual behavioural trends and designing effective sexual health interventions. In Singapore and other Southeast Asian countries, where research has largely focused on key populations, this study aimed to characterise sexual behaviour patterns in the general adult population and their associations with socio-demographics, engagement with sex workers and sexually transmitted infection (STI) positivity.

Methods

We conducted a nationwide cross-sectional survey (September 2023 to July 2024) using voluntary response sampling among adults aged ≥21 years, collecting data on socio-demographics and sexual behaviours in the past year, 6 months and 3 months. Partner numbers and sexual activity frequency were modelled using a Lomax distribution across partnered activities (mutual masturbation, oral, vaginal sex and anal sex), and associations with socio-demographics, sex worker engagement and self-reported STI positivity were evaluated using logistic regression.

Results

Among the 2297 eligible participants, 51.5% were male (1183) and 95.6% were aged 21–60 years (2195). Individuals reporting same-sex partners were more likely to have five or more same-sex partners in the past year. The model-estimated prevalence of engaging with sex workers among individuals aged 21–60 years in Singapore was 4.7% (95% CI 3.7 to 6.2) for females and 18.1% (95% CI 16.1 to 20.4) for males, with higher odds among those with at least five partners in the past year (eg, vaginal sex AOR=7.40, 95% CI 4.03 to 13.58). Self-reported STI positivity was associated with having at least five partners in the past year (eg, vaginal sex AOR=5.03, 95% CI 1.44 to 17.61) and ever engaging with sex workers (eg, vaginal sex AOR=3.16, 95% CI 1.67 to 5.99), but not with barrier protection frequency.

Conclusions

Patterns varied by sex, sexual orientation and risk behaviours, underscoring the need for context-specific behavioural data to inform STI modelling and public health interventions.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Sexual behaviour data are essential for understanding sexually transmitted infection (STI) transmission and guiding public health interventions, yet in Singapore and Southeast Asia such research has largely focused on key populations, with the last general population survey in Singapore conducted in the 1990s.

WHAT THIS STUDY ADDS

• This study provides the first contemporary, population-based survey of sexual behaviours among adults in Singapore, describing frequency and diversity of partnered sexual activities and identifying associations with socio-demographic characteristics, engagement with sex workers and self-reported STI positivity.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• These findings offer updated baseline data to inform targeted sexual health interventions, improve STI transmission modelling and guide context-specific public health policy in Singapore and the wider Southeast Asian region.

Introduction

Understanding sexual health and behaviour is crucial for promoting overall public health, as it has profound implications on both physical¹ and mental health,² and it shapes health outcomes for sexually transmitted infections (STIs).³ However, cultural norms, societal values and religious beliefs in many parts of Asia often inhibit open dialogue about sexual matters.^{4 5} Against this backdrop, research efforts have often concentrated on specific key populations, while the general adult population remains comparatively understudied, particularly in terms of the depth of detailed sexual behaviour data.⁶

This pattern is also evident in Singapore, where research on adult sexual health and behaviour has predominantly focused on key populations, such as men who have sex with men (MSM),^{7 8} sex workers9,11 and heterosexual male clients of sex workers.^{12 13} These studies have largely focused on STIs and participation in risky sexual behaviours. High-risk adolescents have also been studied, particularly in the context of sexual education and STI prevention.^{14 15} In contrast, sexual behaviour within the general population remains underexplored. The last published population-level sexual behaviour survey was conducted in the 1990s.¹⁶ Although more recent studies have included general population samples, they have largely focused on changes in sexual behaviour during the COVID-19 pandemic or psychosocial factors such as stress and fatigue.17,19

Taken together, this highlights the need for contemporary, in-depth sexual behaviour data in Singapore. As cultural norms evolve, sexual attitudes and practices likely vary across age groups, with younger populations potentially being more sexually active.^{20 21} Sexual behaviours may also differ by other demographic factors, such as ethnicity and sexual orientation, but the patterns remain under-characterised. Moreover, detailed information on sexual behaviour patterns, such as frequency of specific partnered sexual activities (eg, mutual masturbation, oral, vaginal and anal sex) and barrier protection use across different sexual acts, is limited.

Therefore, we conducted a nationwide cross-sectional survey of the general adult population in Singapore to understand how sexual behaviours vary across population subgroups and their associations with sex work engagement and self-reported STI positivity. Specifically, we aimed to: (i) describe sexual behaviour patterns by parameterising distributions of sexual partner counts and activity frequency across different partnered sexual activities, stratified by sex, sexual partner preference and recall period (1 year, 6 months and 3 months); and (ii) evaluate the relationship between sexual behaviours, socio-demographic characteristics, engagement with sex workers and self-reported STI positivity. These findings may help inform targeted sexual health interventions and public health messaging in Singapore and the broader Southeast Asian region.

Methods

Survey design and data collection

Between September 2023 and July 2024, a cross-sectional online survey on sexual behaviour was conducted in English among adults in Singapore using voluntary response sampling. Eligible participants were aged 21 years old and above (as required by ethics), had resided in Singapore within the past 6 months and provided key demographic information, including age and sex assigned at birth.

The questionnaire was developed based on existing sexual health surveys22,24 and was further reviewed by the study team for content validity and clarity. Quality control included reCAPTCHA verification at survey entry to prevent automated or non-human responses. However, as the survey collected only discrete demographic and behavioural variables (details in subsequent subsection), internal consistency metrics (eg, Cronbach’s alpha) were not applicable.

Anonymous survey responses were collected through multiple recruitment strategies to reach diverse demographic groups. To engage younger participants, the survey was shared on social media platforms that are widely used by younger demographics (Facebook and Telegram). To increase the representation of individuals with diverse sexual orientations, the survey was promoted in collaboration with Pink Dot SG, a non-profit movement supporting the LGBTQ (individuals who are lesbian, gay, bisexual, transgender or queer) community, via their Instagram account and at the offline Pink Dot 16 event. To reach older residents, responses were also gathered through the survey company Ipsos. All recruitment streams operated simultaneously.

Variables and measures

In the survey, respondents first answered multiple-choice questions on their socio-demographics, such as age, sex assigned at birth (male/female), ethnicity, religion, resident status within Singapore and sexual identity. They were also asked to disclose the sex assigned at birth of their past and current sexual partners.

Sexual behaviour measures were collected from respondents who reported having at least one sexual partner in their lifetime. These included the number of sexual partners, frequency of sexual activity and barrier protection usage for each of the four partnered sexual activities: mutual masturbation, oral sex, vaginal sex and anal sex. Additional behavioural variables included sex toy usage and engagement with sex workers, defined as individuals who received money or goods in exchange for sexual services. These measures were assessed over three recall periods: past year, past 6 months and past 3 months.

In addition, the survey also assessed STI-related factors, including STI positivity, testing frequency and symptoms and treatments received for those claiming to be STI-positive. Please refer to the online supplemental information for the full survey.

Statistical analysis

We summarised the socio-demographic characteristics and sexuality for all eligible participants using counts and percentages. Individuals who reported having sexual partners but indicated no engagement in partnered sexual activities were recoded as having no sexual partners. For the frequency of sexual behaviours, responses such as ‘Not sure’, ‘Too many to count’ or ‘Lost track’ were treated as invalid and missing. When a range of values was provided, the mean of the upper and lower bounds was used. If only a single value was given, it was recorded as is.

To characterise sexual contact patterns in Singapore, we focused on the four sexual activities (mutual masturbation, oral sex, vaginal sex and anal sex), estimating the proportion of individuals aged 21–60 who engaged in each activity. Meanwhile, among those who reported having at least one sexual partner, we approximated the number of sexual partners and the frequency of sexual activities utilising a Lomax distribution to account for the heavy-tailed nature of the observations. Specifically, let $N\S amp;gt;0$ denote the number of partners and $n$ the realisation. The probability of having more than $n$ partners (ie, at least $(n+1)$ partners) is expressed as

$$\begin{array}{c}P\left(N\ge n+1\right)=P\left(N\S amp;gt;n\right)=1-F\left(n\right)={\left(1+\frac{n}{n_0}\right)}^{-\alpha },\#\left(1\right)\end{array}\mathrm{}$$

where $F(\cdot )$ is the cumulative density function of $N$, and $n_0$ and $\alpha$ are positive parameters to be estimated. In addition, the distribution of the number of times a survey participant engaged in sexual activities was also modelled in the format of (1). We defined four partnership types based on the sex assigned at birth of both the survey participants and their partners, including female respondents with female partners, female respondents with male partners, male respondents with female partners and male respondents with male partners. Individuals with partners of both sexes were classified into two categories simultaneously. We fitted the model to all combinations of the four partnership types, three recall periods and four sexual activity types included in the survey. We also estimated the prevalence of sex worker engagement and the distribution of the number of sex workers one engaged with among individuals who reported engaging with them. The detailed estimation approach can be found in the online supplemental information.

We further constructed logistic regression models to investigate the relationship between socio-demographics, sexual behaviour patterns and STI positivity among individuals aged 21–60. As with the previous analysis, the respondents were also classified based on the sex assigned at their birth. For those who reported having sexual partners in the past year, their partnership-based sexual orientation was determined by the disclosed sex assigned at birth of their partner(s), while for those without sexual partners, the variable was determined based on their sex and the self-reported sexual orientation. In total, we included seven distinct partnership-based sexual orientations: men who have sex with men only (MSMO), men who have sex with women only (MSWO), men who have sex with both sexes (MSMW), women who have sex with men only (WSMO), women who have sex with women only (WSWO), women who have sex with both sexes (WSMW) and others. We considered three binary variables as the outcome variable in the regression, including (i) sexual partnership status (ie, whether to engage in a specific type of sexual activity with one or more sexual partners) in the past year, (ii) engagement with sex workers and (iii) reporting any STI positivity. The predictors encompassed a subset of age, ethnicity, partnership-based sexual orientation, number of sexual partners in the last year, engagement with sex workers, STI positivity and barrier protection usage frequency. See the online supplemental information for detailed descriptions of the models (online supplemental table 1). Adjusted ORs were estimated using the survey package in R.²⁵

All data were assumed to be missing at random, with no imputation applied. Sex was defined based on the sex assigned at birth unless otherwise stated. For the analyses of sexual contact patterns and their association with socio-demographics and sexual behaviour patterns, we applied sample weights based on the age and sex composition of Singaporean residents²⁶ to improve the representativeness of our estimates for the general population in Singapore. Due to the cross-sectional design of this survey, all results only suggested descriptive associations and did not imply temporal or causal relationships. All the analyses were performed using the R software.²⁷

Results

Participant characteristics

A total of 2297 eligible participants were included in the analysis after exclusion (579 [20.1%] excluded; details in Methods section), with their socio-demographics and sexuality summarised in table 1. Approximately half of the participants were males (1183; 51.5%). The majority were Singapore citizens (1938; 84.4%), aged 21–60 (2195; 95.6%), of Chinese ethnicity (1368; 59.6%) and heterosexual (1776; 77.3%). A higher proportion of female respondents reported having no sexual partners compared with males (table 1). While the response rates for the socio-demographic questions were generally high, those for certain sensitive questions fell below 80% in specific age groups (online supplemental table 2). For example, responses to the STI positivity question were missing for 30.3% (256) of participants aged 21–30 years and 27.4% (61) of those aged 51–60 years.

Table 1. Socio-demographics and sexuality of all eligible participants, stratified by sex assigned at birth.

	Males		Females
	51.5% (n=1183)		48.5% (n=1114)
	%	N	%	N
Age
21–30	16.2	373	20.5	472
31–40	15.6	359	17.3	397
41–50	9.9	227	6.3	144
51–60	6.3	145	3.4	78
61+	3.4	79	1.0	23
Gender
Men	50.7	1165	0.9	21
Women	0.4	9	46.9	1078
Others	0.4	9	0.7	15
Ethnicity
Chinese	35.5	815	24.1	553
Malay	6.9	159	11.5	265
Indian	7.2	165	12.4	284
Others	1.6	36	0.5	11
Unknown	0.3	8	0.0	1
Residential status
Singapore citizen	44.0	1010	40.4	928
Permanent resident	4.9	113	6.5	149
Work pass holder	2.1	48	1.0	24
Others	0.3	7	0.4	11
Unknown	0.2	5	0.1	2
Religion
Non-religious*	8.7	199	6.7	154
Buddhism	12.3	282	9.5	218
Christianity	11.1	256	9.6	221
Hinduism	4.4	102	7.6	175
Islam	7.3	168	11.1	255
Taoism	3.0	68	2.2	51
Others	0.3	6	0.1	2
Unknown	4.4	102	1.7	38
Monthly salary
No income	7.1	162	6.8	156
SGD0–2000	4.4	101	6.6	151
SGD2000–4999	14.8	341	17.5	401
SGD5000–7999	12.1	277	9.8	225
SGD8000–9999	4.7	108	4.2	96
SGD10000+	6.8	157	3.1	71
Unknown	1.6	37	0.6	14
Sexual identity
Heterosexual	35.7	820	41.6	956
Homosexual	9.7	223	0.8	19
Bisexual	4.0	92	3.3	76
Questioning	0.7	17	1.0	22
Others	0.3	8	0.7	15
Queer	0.2	5	0.7	17
Unknown	0.8	18	0.4	9
Partner’s sex
Female	29.9	687	2.9	66
Male	10.4	238	30.4	699
Both	2.8	65	1.5	34
No sexual partners†	8.4	193	13.7	315

This table presents unweighted (raw) sample demographics. Unless otherwise stated, results reported elsewhere in the paper are based on weighted analyses.

^*Including individuals who were non-religious, atheist, agnostic or freethinkers.

^†Including individuals who had not participated in any of the four partnered sexual activities.

Self-reported sexual activity

Compared with females, males were more likely to report having engaged in mutual masturbation and oral sex. Vaginal sex was relatively uncommon among MSM. Only one of the six participants who reported ever engaging in vaginal sex had at least one such partner in the past year. In contrast, this group had the highest likelihood to engage in anal sex (48.6%, 95% CI 43.7 to 55.3, adjusted for age) across different subpopulations. Women who have sex with women (WSW) were the least likely to engage in all the sexual activities except for vaginal sex, with fewer than 25% reporting one or more sexual partners for the respective activities (figure 1, online supplemental figures 1–3).

Figure 1. Estimated proportion of individuals with at least one sexual partner, stratified by the sex assigned at birth of survey respondents and their partners (for the specified sexual activity), type of sexual activity and recall period. Dots represent point estimates and lines indicate the corresponding 95% CIs. Female or male respondents with partners of both sexes were classified into two categories simultaneously.

Nevertheless, among individuals with at least one sexual partner, WSW were the most likely to have high numbers of partners compared with other subpopulations. For instance, based on the age-adjusted fitted distributions, the estimated probability of WSW having five female partners or more in the past year was 28.4% for mutual masturbation, 27.8% for oral sex, 34.6% for vaginal sex and 71.2% for anal sex. The age-adjusted distribution for the number of male partners among MSM was also heavy-tailed, with the model-based probability estimates of having five partners or more being 36.2% for mutual masturbation and 34.5% for oral sex (figure 2). Compared with the number of partners, differences in the frequency of sexual activities across various partner sex combinations were less pronounced, especially for mutual masturbation and oral sex in the past year (online supplemental figures 4–7).

Figure 2. Fitted distribution of the number of partners (N) among individuals with sexual partners, stratified by the sex assigned at birth of survey respondents and their partners, type of sexual activity (columns) and recall period (rows). Female or male respondents with partners of both sexes were classified into two categories simultaneously. The distribution for the number of male partners among male respondents who engaged in vaginal sex with males was not fitted or presented due to the small sample size. Both the x- and y-axes are displayed on a logarithmic scale.

Survey responses also revealed that engagement in each of the four types of sexual activities varied across age groups (online supplemental figure 8). Compared with 21–30, individuals aged 31–40 years were more likely to have vaginal sex (adjusted odds ratio (AOR): 1.57, 95% CI 1.22 to 2.02). In contrast, people aged over 40 years were less sexually active in the other three types of sexual activities. Among those aged 41–50 years, the adjusted OR of having at least one sexual partner in the past year was estimated at 0.55 (95% CI 0.41 to 0.73) for mutual masturbation, 0.75 (95% CI 0.56 to 1.00) for oral sex and 0.57 (95% CI 0.37 to 0.87) for anal sex. Even lower levels were estimated for the 51–60 group, with adjusted ORs of 0.31 (95% CI 0.22 to 0.44), 0.42 (95% CI 0.30 to 0.59) and 0.31 (95% CI 0.17 to 0.59), respectively, for the aforementioned three sexual activities. These two groups also had a substantially lower likelihood to have at least five partners across the three types of sexual activity in the past year (online supplemental figure 9).

Engagement with sex workers

Partnership-based sexual orientation was inferred to be highly correlated with engagement with sex workers, where women who have sex with men were consistently the least likely to engage with sex workers across all four partnered sexual activities (figure 3). Overall, females were significantly less likely than males to engage with sex workers. Based on age-adjusted model estimates, 4.7% (95% CI 3.7 to 6.2) of females and 18.1% (95% CI 16.1 to 20.4) of males reported such engagement in the past year (online supplemental figure 10). These estimates slightly decreased with a shorter recall period. Females were also less likely to engage with high numbers of sex workers compared with males (online supplemental figure 11). Apart from sex, a strong association was also observed between engagement with sex workers and sexual activity level, quantified through the number of sexual partners in the past year, across all sexual activity types. Within each group reporting mutual masturbation, oral sex, vaginal sex and anal sex within the past year, engagement with sex workers was more commonly reported among individuals who had at least five sexual partners compared with those with no sexual partners. The adjusted ORs were 4.67 (95% CI 2.83 to 7.73), 6.96 (95% CI 4.16 to 11.64), 7.40 (95% CI 4.03 to 13.58) and 11.37 (95% CI 5.50 to 23.53), respectively (figure 3). Similar trends were observed for these factors in their impacts on frequent engagement with sex workers (online supplemental figure 12).

Figure 3. Adjusted OR for socio-demographics, partnership-based sexual orientation defined by sex assigned at birth and number of sexual partners in the past year, stratified by sexual activity type (columns). These include point estimates (dots) and corresponding 95% CIs (lines), quantifying the factors’ impacts on engagement with sex workers. Extreme estimates due to insufficient samples (eg, vaginal sex among MSMO) are not shown. Please also note the x-axes are presented on a logarithmic scale. MSMO, men who have sex with men only; MSMW, men who have sex with both sexes; MSWO, men who have sex with women only; WSMO, women who have sex with men only; WSMW, women who have sex with both sexes; WSWO, women who have sex with women only.

STI positivity

In total, 218 (13.1%) respondents aged 21–60 years who answered the STI positivity question reported being diagnosed or tested positive for at least one STI, with nearly half (105) aged 21–30 years and a negative correlation between positivity and age (online supplemental table 3). However, after adjusting for testing frequency and partnership-based sexual orientation, the association between STI positivity and age was not statistically significant.

While men who have sex with women tended to have lower odds of reporting STI positivity than other groups when other conditions were adjusted for, higher odds were associated with having at least five sexual partners, engaging with sex workers and undergoing STI testing once every three or 6 months. Particularly, having at least five sexual partners over the past year was correlated with an increase in the odds of STI positivity 2.60 (95% CI 1.22 to 5.54) times for mutual masturbation, 4.04 (95% CI 1.55 to 10.54) times for oral sex, 5.03 (95% CI 1.44 to 17.61) times for vaginal sex and 4.52 (95% CI 2.22 to 9.20) times for anal sex among individuals ever tested for STIs and engaged in respective sexual activities. Similarly, having ever engaged with sex workers led to 3.99 (95% CI 2.14 to 7.43), 3.61 (95% CI 1.94 to 6.72), 3.68 (95% CI 1.80 to 7.50) and 3.16 (95% CI 1.67 to 5.99) times’ increase in the odds of STI positivity for the respective sexual activities (figure 4). Nevertheless, our analysis did not suggest a statistically significant association between STI positivity and barrier protection usage frequency (online supplemental figure 13).

Figure 4. Adjusted OR of STI positivity status for socio-demographics, partnership-based sexual orientation, number of sexual partners in the past year, engagement with sex workers and STI testing frequency, stratified by sexual activity type (columns). These include point estimates (dots) and corresponding 95% CIs (lines). Extreme estimates due to insufficient samples (eg, vaginal sex among MSMO) are not shown. Individuals who had never tested for STIs were excluded from the analysis. Rare STI testing is defined as testing once a year and frequent testing refers to testing once every three or 6 months. Please also note the x-axes are presented on a logarithmic scale. MSMO, men who have sex with men only; MSMW, men who have sex with both sexes; MSWO, men who have sex with women only; STI, sexually transmitted infection; WSMO, women who have sex with men only; WSMW, women who have sex with both sexes; WSWO, women who have sex with women only.

Discussion

This study provides insights into the diversity and distribution of sexual behaviours across the multi-ethnic Singapore population. By capturing detailed behavioural patterns, including partner numbers, types of sexual activity and engagement with sex workers, across both heterosexual and non-heterosexual populations, the study enhances understanding of sexual behaviours that have been underrepresented in regional studies. In particular, the fitted distributions for the number of sexual partners across different sexual activities, alongside reported frequencies of participation in these activities, offer a more granular characterisation of sexual contact patterns than previously understood. Such localised data offer behavioural insights into transmission risk across sociodemographic groups and can better inform the parameterisation of STI models, enabling them to more accurately reflect behavioural patterns and population heterogeneity in Singapore.

Implications

Due to the lack of contemporary data in Singapore and many other Asian countries, STI modelling studies in Singapore have used partner number distributions and sexual activity assumptions drawn from Western contexts.²⁸ However, the number of partners and the participation in sexual practices likely vary between western countries and countries in Asia. This is corroborated by higher participation in partnered sexual activities in the past year reported by the National Survey of Sexual Attitudes and Lifestyles (NATSAL-3) in Britain, with 82.1% of men and 77.7% of women between 16 and 74 years old who had at least one opposite sex partner over the past year.²⁹ In contrast, our results indicated that the participation in any type of sexual activity (mutual masturbation, oral sex, vaginal sex and anal sex) in the past year was lower in Singapore than in Britain, at 71.7% (95% CI 68.7 to 74.6) and 56.6% (95% CI 52.7 to 60.6) of male and female participants aged 21–60 years old. This highlights the need for localised data when modelling STIs in the region. By providing behavioural estimates in Singapore, this study lays essential groundwork for more contextually accurate STI transmission models and more targeted public health interventions.

Marked sex- and partnership orientation-based differences in reported sexual activity were observed, particularly among individuals with opposite-sex partners. Male participants were more likely than female participants to report recent sexual activity across all four sexual activities (mutual masturbation, oral sex, vaginal sex and anal sex) and reported higher frequencies of engaging in these acts with female partners than female respondents did with male partners. This asymmetry could reflect actual differences in behaviours, but may be influenced by gendered reporting biases, such as social desirability or cultural stigma, that differentially affect willingness to disclose sexual activity, especially among women.^{30 31} In addition, WSW with at least one sexual partner reported lower frequencies of engaging in mutual masturbation, oral sex and vaginal sex across different recall periods (figure 1). However, among those who were sexually active, a notable proportion reported high partner numbers (>5 partners in the past year). A similar pattern was observed among MSM, suggesting substantial variation within these groups: while many report low or moderate sexual activity, a smaller subset is more sexually active.

Engagement with sex workers was significantly more common among males than females, with males, across all partnership-based sexual orientation categories, having higher odds of engaging with sex workers within each of the regression models stratified by the four sexual activities (mutual masturbation, oral sex, vaginal sex and anal sex). While fewer females reported such engagement overall, variations by partnership-based sexual orientation were still apparent, with WSW reporting higher rates of engagement with sex workers than women who have sex exclusively with men (WSMO). These observations challenge prevailing assumptions in Singapore, where research on sex worker clientele has primarily focused on heterosexual men^{12 13} and point to the need for broader recognition of behavioural diversity. Importantly, individuals reporting five or more sexual partners in the past year were significantly more likely to have engaged with sex workers across all sexual activity types. This co-occurrence of different sexual behaviours, such as transactional sex, partner concurrency and high partner numbers, has been documented in various settings globally, including among male factory workers in China,³² substance-using youth in the United States³³ and adults in rural communities in Kenya and Uganda.³⁴ Recognising co-occurrence of different sexual behaviours is important for both surveillance and intervention design. Surveillance strategies that analyse each risk factor independently may underestimate the threat posed by individuals with co-occurring behaviours, and interventions targeting one behaviour at a time may miss those most likely to benefit. Instead, using combined indicators such as partner number and partner type (sex worker or not) can enhance identification of high-transmission subgroups and guide more efficient intervention efforts like targeted STI screening, prevention outreach and linkage to care.

Among individuals who reported ever testing for STIs, those who tested more frequently had higher self-reported positivity. This could reflect risk-aware behaviour, with higher-risk individuals proactively seeking testing, but could also indicate reverse causality, where symptomatic individuals or those previously diagnosed may test more often. This is important for interpreting surveillance data, as higher positivity rates among frequent testers may represent greater detection rather than elevated risk alone. Additionally, higher levels of sexual activity and reported engagement with sex workers were both associated with greater odds of self-reported STI positivity. While these associations align with existing understandings of STI risk, reverse causality remains a possibility, where individuals who test positive may retrospectively report more sexual activity or more carefully recall engagements with sex workers. This underscores the importance of targeting sexual health interventions and counselling toward subpopulations indicating these behaviours, to efficiently reduce transmission risks.

Although STI positivity was not significantly associated with reported barrier protection frequency in our analyses, recall and social desirability bias may have reduced our ability to detect an association. The relatively high non-response rates for questions on both barrier protection use and STI positivity suggest some participants may have felt uncomfortable disclosing these sensitive details, leading to reporting bias (online supplemental table 3, figure 13). Given these limitations, the absence of a clear association should be interpreted with caution. Nevertheless, it highlights the importance of complementing barrier protection education with comprehensive behavioural risk-reduction strategies, including regular STI screening and robust partner notification programmes, rather than relying on protective behaviours. Furthermore, there is potential value in strengthening timely and targeted STI screening through accessible pathways such as local clinic services, primary care and community-based testing programmes. Outreach efforts to improve testing literacy and awareness of asymptomatic infection, particularly among higher-risk networks, could also be effective.

Limitations

Several limitations should be acknowledged when interpreting our findings. First, older individuals were under-represented in our sample, with only 9.7% aged 51–60 years (223) and 4.4% aged 61 years and above (102). In addition, non-response rates for specific sexual activity questions were high among participants aged at least 61 years. Consequently, this age group was excluded from our analyses, limiting the generalisability of our findings to older populations (aged 61+) and our ability to reliably infer their sexual behaviour patterns. Additionally, despite our use of anonymous online surveys, social desirability bias may still have influenced responses, potentially leading to under-reporting of behaviours viewed as socially undesirable or stigmatised. Consequently, actual prevalence rates of higher-risk sexual activities may be underestimated. Equally, individuals may have also self-reported an STI infection without formal diagnosis from a clinician, or may choose not to report. Also, while our analyses assumed missingness to be random, non-response patterns for individual questions may vary by age, especially given the inconsistent direction and magnitude of the inferred age effects across different sexual activity types. In addition, employing specific web communities and online survey methods could have introduced sampling bias, excluding less technologically inclined individuals and thus potentially impacting the representativeness of our findings. While the heavy-tailed Lomax model for partner counts was used to capture heterogeneity, whereby a small proportion of individuals typically report substantially higher numbers of partners, fitted parametric distributions represent a simplified approximation of the empirical counts and may not fully capture subgroup-specific variation or extreme values observed in the raw data. Finally, to ensure sufficient sample sizes for characterising sexual behaviours among sexual minority groups, we actively recruited individuals identifying as MSM and WSW through LGBTQ-focused platforms. While this approach improved representation within these subgroups, the overall sample is not representative of the general population by sexual orientation. Accordingly, findings should be interpreted as reflective within, rather than across, each partnership-based sexual orientation category.

Future research

Future research should aim to address identified limitations by engaging broader demographic groups, particularly older adults and less digitally accessible populations, ensuring more comprehensive representation. Continued surveillance, paired with interventions tailored to behavioural subgroups exhibiting elevated risk, will be essential in effectively mitigating STI transmission in Singapore.

Conclusion

This study examines contemporary sexual behaviours in Singapore, providing critical insights for infectious disease modelling and public health practice. The findings highlight nuanced differences in sexual practices and patterns based on sex, partnership-based sexual orientation and risk behaviours. Importantly, the observed clustering of higher-risk behaviours within specific subgroups points to the need for integrated prevention strategies that consider the co-occurrence of multiple behavioural risks. These insights can inform the design of more targeted, contextually relevant sexual health interventions and support the development of transmission models that are better calibrated to reflect the social and behavioural realities of the local population.

Data availability statement

Data are available upon reasonable request.