Self-Reported PrEP Use and Risk of Bacterial STIs Among Ontarian Men Who Are Gay or Bisexual or Have Sex With Men

INTRODUCTION

Bacterial sexually transmitted infections (STIs), including chlamydia, gonorrhea, and syphilis, have increased since 2010, with cases of gonorrhea and syphilis commonly reported among men in Ontario, Canada.¹ In 2014, an estimated 40% of gonorrhea cases and 85% of syphilis cases occurred among gay, bisexual, and other men who have sex with men (GBM).² This disproportionate burden is of concern as untreated bacterial STIs can cause serious sequelae such as epididymitis, orchitis, prostatitis, urethral strictures, and infertility.^3,4 Various factors contribute to the increasing incidence of bacterial STIs including expanded extragenital testing,⁵ substance use,⁶ and a reduction in condom use, partly attributed to the availability of HIV pre-exposure prophylaxis (PrEP). These associations have raised concerns about sexual risk compensation (eg, an increase in condomless sex or number of partners), given the concurrent increase in PrEP use and bacterial STI diagnosis among GBM.^7-11

It is hypothesized that PrEP use could reduce condom use and increase the risk of bacterial STI transmission among GBM.^10,11 Research on the associations among PrEP use, sexual behavior, and bacterial STIs in this population has netted inconsistent findings, however.^12,13 Some studies have found associations between PrEP use, condomless anal sex, and bacterial STIs,^14-18 whereas others have not.^19,20 These disparate findings may reflect a methodologic challenge, that is, the bidirectional relationships of PrEP uptake, sexual behavior, and bacterial STI incidence. For example, a history of bacterial STI is often an indicator for initiating PrEP²¹ because it may reflect sexual networks with higher bacterial STI prevalence and greater risk of HIV exposure, or an indicator for increased testing as part of routine PrEP care (Supplemental Figure 1). Moreover, condomless anal sex may both confound future PrEP uptake and bacterial STI incidence and mediate past exposure–outcome associations. Given this complexity, traditional regression analysis approaches may not suffice, necessitating the use of marginal structural models to account for the time-varying effect of PrEP uptake on bacterial STI incidence.

To better understand this complex relationship, we considered whether the trajectory of PrEP uptake over 3 months functions as a determinant of bacterial STI incidence using observational data from an online study of Ontarian GBM.²² We hypothesized that PrEP uptake would affect bacterial STI incidence, whereby those taking PrEP would have a higher incidence rate than those not taking PrEP after accounting for baseline and time-varying covariates.

METHODS

Study Design and Participants

The iCruise study is a longitudinal, online, mixed-methods study designed to assess the sexual health-seeking behaviors of GBM in Ontario, Canada.²³ Participants were purposively sampled with respect to 8 GBM subgroups (aged >50 years old, living with HIV, immigrants, ethnoracial minority, transgender, rural resident, straight-identified, and those encountered online by outreach from sexual-health workers) between July 2017 to April 2018 using sociosexual networking sites, mobile apps, and a community-based e-mail listserv. Eligible participants were men who were aged 14 years or older; identified as gay, bisexual, 2-spirit, straight, queer, or questioning; had sex with another man in the past 12 months; and lived in, worked in, or visited Ontario at least 4 times in the past year.

A total of 907 eligible men enrolled and 659 were randomly selected to participate in a weekly diary survey for up to 12 weeks after baseline (Figure 1). During follow-up, participants were asked to complete at least 10 of 12 weekly electronic diary surveys; those who missed 4 or more consecutive diary surveys were considered lost to follow-up. For our analysis, we further restricted the sample to HIV-negative, cisgender GBM aged 18 years or older.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Recruitment and retention in the iCruise study.

Note: Participants were defined as not completing follow-up if they missed ≥4 consecutive diary surveys.

Participants provided informed consent and received compensation of up to $80 for completing the surveys. Ethical approval was granted by the Research Ethics Board of the University of Toronto.

RESULTS

Overall, 535 participants in the iCruise study were included in our analytic sample, most of whom were gay and White (Table 1). The mean (SD) age at baseline was 34.0 (12.8) years and about one-half of participants (46.7%) had had sex partners in the past 3 months, with 18.3% and 11.1% having had more than 5 anal and oral sex partners, respectively. Nearly one-fifth (18.1%) had engaged in condomless sex in the past 3 months. The large majority (85.2%) had access to a primary care clinician. Past STI testing (36.1%) and lifetime STI diagnosis (45.2%) were fairly common. Participants reported an average of 3.4 anal sex partners during the week before initiating PrEP and 1.2 anal sex partners after PrEP initiation. The prevalence of condomless anal sex was similar during the week before (57.1%) and the week after (54.3%) PrEP initiation.

During follow-up, 86.4% of participants remained in the study. Participants were followed up for a median of 3.3 months (interquartile range, 2.9-3.4). During 1,623.5 person-months (135.3 person-years) of follow-up, 13.1% of participants took PrEP (Table 2) and 1.1% discontinued PrEP. In the cohort overall, the crude incidence rates per 100 person-months were 5.5 (95% CI, 4.5-6.8) for any bacterial STI, 2.5 (95% CI, 1.8-3.4) for chlamydia, 1.8 (95% CI, 1.2-2.6) for gonorrhea, and 1.2 (95% CI, 0.8-1.9) for syphilis.

After accounting for confounding and selection bias using the inverse probability of exposure and attrition weights, the incidence rate of any bacterial STI was higher among iCruise participants who took PrEP compared with those who did not (incidence rate ratio [IRR] = 1.86; 95% CI, 0.86-4.04; Table 3); however, this estimate was imprecise and not statistically significant. The effect of PrEP uptake on the incidence rate varied across the bacterial STIs, with an adjusted IRR of 1.22 (95% CI, 0.53-2.81) for chlamydia, 4.00 (95% CI, 1.67-9.58) for gonorrhea, and 1.05 (95% CI, 0.28-3.92) for syphilis.

Sensitivity analyses indicated that associations between PrEP uptake and bacterial STI incidence were robust to varying assumptions about the timing of PrEP uptake (Supplemental Figure 4). For gonorrhea, the E-value was about 7.5, suggesting an unmeasured confounder would need to be associated with both PrEP uptake and bacterial STI incidence with an IRR of at least 7.5 beyond measured covariates to explain away the results (Table 3). Quantitative bias analysis indicated a median IRR of 1.55 (95% simulation interval, 0.76-3.01) for any bacterial STI and 2.36 (95% simulation interval, 1.08-5.06) for gonorrhea in PrEP users compared with nonusers (Figure 2). Thus, nondifferential outcome misclassification biased the results away from the null with an artificially inflated association between PrEP uptake and gonorrhea incidence.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Unadjusted and multibias-adjusted density of incidence rate ratios among GBM by bacterial sexually transmitted infection.

GBM = gay, bisexual, and other men who have sex with men; STI = sexually transmitted infection.

Note: Incidence rate ratios compare GBM with vs without uptake of pre-exposure prophylaxis. Unadjusted density is based on 1,000 bootstrapped replications of the unadjusted model. Multibias-adjusted density is based on the simulated distribution of the true incidence rate while accounting for confounding and losses to follow-up using inverse probability of exposure and attrition weights, respectively. Results are not shown for chlamydia and syphilis because of the null findings in unadjusted models. Furthermore, because of the low incidence rate for syphilis, simulated incidence rate ratios in the multibias-adjusted model resulted in unstable estimation for that STI.

DISCUSSION

In this online longitudinal study, PrEP uptake was associated with a higher incidence rate of gonorrhea among GBM after adjusting for confounding, attrition, and nondifferential outcome misclassification. We did not find sufficient evidence to indicate higher incidence rates of chlamydia and syphilis among men using PrEP during follow-up. The magnitude of risk for gonorrhea among those taking PrEP suggests a need to focus on ancillary services (eg, STI testing) that can support the sexual health needs of GBM taking PrEP and potentially novel interventions for preventing transmission such as doxycycline post-exposure prophylaxis (doxy-PEP).^32,33 For example, there is growing recognition that comprehensive STI prevention requires the presence of primary care,³⁴ and in many cases, family medicine practitioners are well positioned to serve as the main contact point for GBM, offering valuable information about PrEP, prescribing these medications, and supporting ancillary services to ensure long-term retention in care.

Our findings on PrEP uptake and risk compensation are consistent with some previous reports from observational studies in high-resource settings that showed an association between PrEP uptake and bacterial STIs.^13,16-18 For instance, in a longitudinal study using propensity score–matched historical controls, investigators found that the incidence rate of bacterial STIs over time increased by 2.4-fold in participants who initiated PrEP at a sexually transmitted disease clinic in the United States.¹⁷ The investigators also reported associations between PrEP initiation and each bacterial pathogen, including chlamydia, gonorrhea, and syphilis. Similar results were also reported among Australian GBM and indicated a higher incidence rate of bacterial STIs after PrEP initiation among those who did not have any prior experiences with PrEP.¹⁸

We did not have sufficient evidence, however, that PrEP uptake was associated with an increased incidence rate of any bacterial STI, chlamydia, or syphilis. This difference may reflect a higher prevalence of gonorrhea within sexual networks in Ontario during the time in which our data were collected. According to the 2017 European Men-who-have-sex-with-men Internet Survey, the prevalence of gonorrhea (24.6%) was higher than that of chlamydia (22.4%) and syphilis (13.7%) among GBM in Ontario that year,³⁵ with data suggesting rising rates in Canada during the past few years.^7-9 Moreover, differences in the transmission probability, incubation period, and intensity of symptoms of gonorrhea vs other bacterial STIs^3,36,37 may also contribute to the increased reporting of gonorrhea in the shorter follow-up period. These data indicate that the potential decrease in condom use by iCruise participants while using PrEP may further contribute to the proliferation of gonorrhea transmission among Ontarian GBM.

Although more research with longer follow-up is needed to understand the long-term impact of PrEP uptake on gonorrhea, our findings support calls to further integrate STI risk-reduction counseling and testing services within PrEP programs,^21,38 ensuring that PrEP adopters understand its benefits and limitations, and have access to care that addresses their sexual health needs. Furthermore, recent evidence from the US DoxyPEP trial shows a 65% (95% CI, 54%-73%) reduction in bacterial STI incidence over a 3-month period³³ and similar effectiveness for people living with HIV, whether taking PrEP or not.³⁹ Restricting doxy-PEP to people taking HIV PrEP therefore may not be warranted. Our results provide some support for this recommendation for more lenient prescribing as transmission rates for any bacterial STI, chlamydia, and syphilis were similar regardless of PrEP uptake. Ongoing surveillance is essential, however, to address tetracycline resistance concerns in gonorrhea.⁴⁰

Our study has several limitations. It relied on a shorter follow-up period with data collected from 2017 to 2018, during which the landscape of PrEP uptake changed. We made important assumptions to obtain valid estimates, including consistency, exchangeability, positivity, and no model misspecification.²⁵ We could include only confounders available in the iCruise study data, however, potentially omitting other important ones. We did not differentiate between partner role (insertive vs receptive) when including condomless anal sex as a time-varying covariate in our models, possibly contributing to residual confounding. Although we assumed nondifferential misclassification, it is possible that differential misclassification may be present as participants’ PrEP status may affect their probability of receiving STI testing and subsequent diagnosis. We were unable to evaluate this potential bias given the lack of external validation data within PrEP uptake strata. Furthermore, if validation data do not accurately represent the iCruise study population, this may further bias the findings. Increasing the statistical uncertainty of the sensitivity and specificity parameters may help mitigate this limitation. Although our analysis accounted for outcome misclassification, there is also the potential of exposure misclassification of reported PrEP uptake due to social desirability bias. The use of inverse probability of attrition weights strengthens the validity of the estimates, but also relies on the same assumptions as exposure weights, which might not fully adjust for selection bias. Finally, recruitment was primarily through sex-seeking social media sites, which introduces potential self-selection bias.

CONCLUSIONS

Our findings elucidate the early impact of PrEP uptake on reported bacterial STI incidence among Ontarian GBM. As bacterial STI rates rise, understanding the etiologic mechanisms will be crucial. Our findings, however, do not detract from PrEP’s role in HIV prevention. Rather, they underscore the importance of investing in STI testing, risk reduction, secondary prevention measures such as doxy-PEP, surveillance, and partner management—alongside PrEP—to reduce bacterial STI transmission effectively.