Background The spectrum of sexual practices that transmit Neisseria gonorrhoeae in men who have sex with men (MSM) is controversial. No studies have modelled potential Neisseria gonorrhoeae transmission when one sexual practice follows another in the same sexual encounter (‘sequential sexual practices’). Our aim was to test what sequential practices were necessary to replicate the high proportion of MSM who have more than one anatomical site infected with gonorrhoea (‘multisite infection’).
Methods To test our aim, we developed eight compartmental models. We first used a baseline model (model 1) that included no sequential sexual practices. We then added three possible sequential transmission routes to model 1: (1) oral sex followed by anal sex (or vice versa) (model 2); (2) using saliva as a lubricant for penile–anal sex (model 3) and (3) oral sex followed by oral–anal sex (rimming) or vice versa (model 4). The next four models (models 5–8) used combinations of the three transmission routes.
Results The baseline model could only replicate infection at the single anatomical site and underestimated multisite infection. When we added the three transmission routes to the baseline model, oral sex, followed by anal sex or vice versa, could replicate the prevalence of multisite infection. The other two transmission routes alone or together could not replicate multisite infection without the inclusion of oral sex followed by anal sex or vice versa.
Conclusions Our gonorrhoea model suggests sexual practices that involve oral followed by anal sex (or vice versa) may be important for explaining the high proportion of multisite infection.
- gay men
- incidence studies
- sexual behaviour
Data availability statement
All data relevant to the study are included in the article or uploaded as supplementary information. The data were all collected from published articles and reports. Data are available in the supplementary materials.
Statistics from Altmetric.com
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.
The incidence of STIs in men who have sex with men (MSM) has been increasing in most countries for at least a decade.1–3 Gonococcal infection is particularly common in MSM populations,4 ,5 and the growing number of cases increases the risk of antimicrobial resistance developing. Both the WHO and US Centers for Disease Control consider Neisseria gonorrhoeae to be an urgent global public health issue and recommend interventions to reduce N. gonorrhoeae transmission.6 Developing such interventions requires a detailed understanding of N. gonorrhoeae transmission routes in MSM, particularly given that condom use in most MSM populations is falling.
Until recently, it was considered that most cases of gonorrhoea in MSM were transmitted from the penis. Recently, however, it has been proposed that the oropharynx and saliva may play a role in gonorrhoea transmission in MSM.7–9 It is suggested that N. gonorrhoeae may be transmitted from one man’s oropharynx to his partner’s oropharynx, anorectum or urethra, in part through saliva.10–12 The potential transmission through saliva is supported by the finding that N. gonorrhoeae can be cultured in the saliva of MSM who have oropharyngeal gonorrhoea.13 A cross-sectional survey reported that using saliva as a lubricant for anal sex was associated with anorectal gonorrhoea infections in MSM.14
If the hypothesis that gonorrhoea can be transmitted via saliva is correct, then a detailed understanding of sexual practices that involve saliva is required in MSM.15 Sexual practices in MSM are complex, and understanding the potential transmission routes requires considering a variety of sexual practices during the same sexual encounter (hereafter referred to as ‘sequential sexual practices’). For example, a man with oropharyngeal gonorrhoea who performs oral sex on his partner and then receives condomless penile–anal sex from his partner’s penis that is covered with his own saliva may have gonorrhoea transmitted from his own oropharynx to his own anorectum. A survey reported that the majority of MSM had more than one sexual practice in the same sexual encounter.16 Understanding the order of sexual practices involving saliva, genital and anorectal sites could help clarify the transmission of gonorrhoea. However, to date, no study has investigated the role of sequential sexual practices during the same sexual encounter in the transmission of gonorrhoea. Undertaking such research would be challenging and require a complex and detailed questionnaire.
Mathematical models provide an alternative method to investigate the role that sequential sexual practices may play in explaining variations in the prevalence of N. gonorrhoeae at different anatomical sites. Two published mathematical models explored the possibility of transmission through kissing (Zhang et al 10 and Spicknall et al 12). Zhang et al found that the inclusion of transmission from the oropharynx to either the oropharynx or anorectum was necessary to replicate the existing prevalence data. However, this study not only considered infections that occurred at a single anatomical site but also infections at two or three anatomical sites simultaneously in MSM (multisite infection).7 17 The study by Zhang et al also did not consider the sequence in which sexual practices occurred. The second published model by Spicknall et al included multisite infections but did not include sequential sexual practices. Spicknall et al also indicated that his models could not replicate multisite infection.12 Here, we aim to extend these existing models to include the sequence of sexual practices to determine what sexual practices are required to replicate the prevalence of N. gonorrhoeae at all anatomical sites, including multisite infection.
We constructed a susceptible–infectious–susceptible compartmental model to examine the transmission routes of gonorrhoea in MSM (online supplemental figure S1) and based this model on the three previous models.10 12 18
Baseline transmission routes
N. gonorrhoeae can infect the oropharynx, urethra and anorectum.7 17 Our baseline model included the same seven transmission routes as the two previous models.10 12 These seven transmission routes were part of four sexual practices, that is, anal sex (urethra to anorectum and anorectum to urethra), penile–oral sex (oropharynx to urethra and urethra to oropharynx), rimming (oropharynx to anorectum and anorectum to oropharynx) and kissing (oropharynx to oropharynx). We did not consider urethra-to-urethra or anorectum-to-anorectum as possible transmission routes given the absence of data or work to support these (figure 1). In the baseline model, multisite infections occurred when a man with a single site infection had sexual practices with another infected man (figure 1).
Additional transmission routes
We added three transmission routes (figure 1) to the baseline model described previously to investigate potential transmission via three sexual practices that ‘connect’ anatomical sites: the first sexual practice is oral sex followed by anal sex (or vice versa). In this route, saliva that remained on a partner’s penis after oral–penile sex could pass infection into their own anorectum through penile–anal sex. If anal sex preceded oral sex, then N. gonorrhoeae that remained on a partner’s penis after anal sex could pass into their own oropharynx through penile–oral sex. In both acts, the penis acts as a mediator and carries N. gonorrhoeae to the respective site.
The second sexual practice is when a man uses saliva as a lubricant for penile-anal sex. In this route, his saliva that is used as a lubricant could pass the infection from his oropharynx to his own urethra and thereby cause multisite infection of the oropharynx and urethra.
The third sexual practice is when oral sex is followed by oral–anal sex (rimming) or vice versa. In this route, N. gonorrhoeae remaining in the oropharynx after oral sex could pass infection into their anorectum through oral–anal sex. Similarly, N. gonorrhoeae remaining in the oropharynx after oral–anal sex could pass infection to the urethra. In both acts, the oropharynx acts as a mediator and carries N. gonorrhoeae to either the urethra or anorectum. We assumed that the mediators (oropharynx or urethra) of the partner is not infected during these additional sequential sexual practices (figure 1).
Our models assumed: (1) a ‘closed’ population where no individuals entered or left; (2) the members of the MSM population mixed homogeneously; (3) a multisite N. gonorrhoeae infection in the same individual could occur either by from an infected partner or through sequential sexual practices that connected anatomical sites in the same man as described above.
Definition of models
We developed eight models based on the transmission routes we have described above. Model 1 was the baseline model that included no sequential sexual practices. All subsequent models included the sexual practices that were in model 1. We then added three possible transmission routes to model 1: (1) oral sex followed by anal sex (or vice versa) (model 2); (2) using saliva as a lubricant for penile–anal sex (model 3) and (3) oral sex followed by oral–anal sex (rimming) or vice versa (model 4). Models 5, 6 and 7 included the three combinations of any two of the three sequential sexual practices. Model 8 included all three types of sequential sex practices during the same sexual encounter (figure 1).
We used previously published behaviour and infection progression data, expert opinion and assumption for the parameters in our models (online supplemental table S1). A total of 26 parameters were needed to parameterise the eight N. gonorrhoea models (online supplemental table S1): two parameters related to condom use, four parameters related to sexual practices, five parameters related to sequential sexual practices, six parameters related to the N. gonorrhoea infection recovery process, three parameters related to screening for sexually transmitted infections and seven parameters were site-specific infection prevalence data. Details are provided in the supplementary materials. The collected parameters were sampled within the CI by the Latin hypercube sampling method and repeated 300 times. For every simulation, we calculated the sum of squared errors to estimate the goodness of fit between seven simulation output prevalence data and epidemiology data. According to the goodness of fit to select ‘optimal runs’, we sorted the simulation outputs in descending order, and 10% of 300 simulations were regarded as the calibrated model estimates to the epidemic trend and used to generate the N. gonorrhoeae models outputs with 95% CIs. A t-test was performed to examine the differences in the sum of squared errors between two models. Statistical analyses and simulations were performed in MATLAB R 2019a.
Our N. gonorrhoeae model was calibrated to anatomic site-specific prevalence of gonorrhoea data from 4873 MSM (2555 in 2018 and 2318 in 2019) who visited Melbourne Sexual Health Centre (MSHC) for the first time and were tested for N. gonorrhoeae at all three anatomic sites. The proportion of MSM who had more than one anatomical site infected with N. gonorrhoeae was high (41.6%=223/536). At MSHC, MSM was routinely tested for N. gonorrhoeae at the oropharynx, urethra and anorectum using nucleic acid amplification tests (NAATs).13 To further validate our model findings, we also used six other similar studies with multisite infection data where MSM were tested using NAAT. These six studies were from four continents12 19–23 (details in the online supplemental tables S2–3).
The prevalence of urethral gonorrhoea infection in the community at any given time would be much lower than that among men attending STI clinics. Based on the previously published method,10 we calibrated the prevalence of single-site infections (oropharynx only, anorectum only and urethral only) and multisite infections (oropharynx and urethra; oropharynx and anorectum, urethral and anorectum and all three sites) to the community level (details in the online supplemental materials).
The proportion of N. gonorrhoeae incidence
We used our models to estimate the proportion of N. gonorrhoeae incidence that occurs at the oropharynx, anorectum and urethra. Our models also estimated the proportion of incidence caused by sexual practices (details in the online supplemental materials).
Uncertainty and sensitivity analysis
We conducted sensitivity analyses for the models over the duration of infection, frequency of sexual practices and the proportion of men with sequential sexual practices (details in the online supplemental materials).
Replication of site-specific infection for N. gonorrhoeae without sequential sexual practices
Figure 2 shows that the baseline model (model 1) could replicate oropharyngeal, urethral and anorectal prevalence at single anatomical sites but underestimated the prevalence in men with multisite infections at both the oropharynx and urethra or at both the oropharynx and anorectum.
Replication of site-specific infection for N. gonorrhoeae with sequential sexual practices
The key route of transmission that was necessary to replicate multisite infection was oral sex followed by anal sex (or vice versa) (transmission route 1), and this route of transmission was included alone in model 2 (figure 1). Saliva use as a lubricant for anal sex (transmission route 2) alone was not able to replicate multisite infection, and this route alone was included in model 3. Oral sex followed by oral–anal sex (rimming) or vice versa (transmission route 3) alone was not able to replicate multisite infection and was included in model 4.
We then combined the three different sequential transmission routes to determine if they provided additional benefit when combined. These combinations are show in models 5–8, and it is clear from figure 1 that the key transmission route that was necessary to replicate multisite infection was oral sex followed by anal sex (or vice versa). Specifically, the absence of this transmission route was not able to replicate the prevalence in men with multisite infection. The reduced percentages of the sum of squared errors varied from 0.8% to 99.0% across six models (models 2–6 and 8) compared with model 1, but model 7 increased 1.5% of the sum of squared errors compared with model 1. Model 8 had the least sum of squared errors, but the reduced percentages of the sum of squared errors of model 8 (99.0%) were very similar to model 2 (97.8%), model 5 (98.3%) and model 6 (97.5%) (online supplemental table S4).
We used six other datasets to calibrate our models separately (details in the online supplemental materials). All datasets produced similar results for calibration of single-site infection and multisite infection to the data of 4873 MSM attending MSHC in 2018 and 2019. Similarly, the inclusion of oral sex followed by anal sex (or vice versa) was necessary to explain the prevalence of multisite infection at the oropharynx and anorectum. Online supplemental figure S2a–f shows the calibration results of six other datasets.
Estimating composition of N. gonorrhoeae incident cases
Based on the data from 4873 MSM attending MSHC in 2018 and 2019, our model estimated oropharyngeal infection accounted for 38.8%–60.6% of all new cases, followed by urethral infection (17.2%–46.5%) and anorectal infection (14.7%–22.5%), across all eight models (figure 3A). We used our models to estimate the proportion of incident gonorrhoea of three anatomical sites in six other datasets. Oropharyngeal infection accounted for the majority of new infection cases, followed by urethral infection and anorectal infection, across all seven datasets (details in the online supplemental materials).
Based on the data from 4873 MSM attending MSHC in 2018 and 2019, we used our models to estimate the proportion of incidence caused by the different sexual practices (figure 3B). The proportion of all incident infections caused by anal sex only was 8.8%–18.5% (anorectum-to-urethra: 7.9%–15.6%; urethra-to-anorectum: 0.5%–2.9%). Transmission through rimming only contributed to 16.5%–26.4% of incident cases (anorectum-to-oropharynx: 8.3%–13.4%; oropharynx-to-anorectum: 7.1%–13.5%). Oral sex only contributed to 17.5%–34.7% (urethra-to-oropharynx: 7.7%–16.0%; oropharynx–to-urethra: 8.6%–27.0%). The proportion of incident infections through kissing only (oropharynx-to-oropharynx) accounted for 10.5%–20.8% of incidental cases. The proportion of incidence caused by oral sex followed by anal sex (or vice versa) was 32.3%–33.6% (oral sex followed by anal sex: 3.6%–8.0%, anal sex followed by oral sex: 25.7%–29.2%). The proportion of incident infections caused by using saliva as a lubricant for penile–anal sex was 10.9%–17.8% (saliva on own penis: 0.9%–12.4%, using saliva as a lubricant for anal sex: 5.4%–10.6%). The proportion of incident infections caused by oral sex followed by oral–anal sex (rimming) was 0.8%–4.1% (oral sex followed by oral–anal sex: 0.02%–0.4%, anal–oral sex followed by oral sex: 0.7%–3.8%). Our models estimated the proportion of incident infections caused by sexual practices in six other datasets (details in the online supplemental materials).
Sensitivity analyses showed that varying key model outcome indicators did not alter our conclusions related to sequential sexual practices. Oral sex followed by anal sex (or vice versa), saliva use as a lubricant for anal sex and oral sex followed by oral–anal sex (rimming) or vice versa was essential for replicating multisite infection. Our sensitivity analyses did not alter our conclusion that oropharyngeal infections contributed to the majority of incident infections (details in the online supplemental materials).
To our knowledge, this is the first model of gonorrhoea transmission to explore the role of sequential sexual practices. Our model suggests that including these sequential sexual practices is necessary to replicate the observed prevalence of single and multisite infections at the urethra, oropharynx and anorectum in MSM. In particular, our findings suggest that including the transmission of gonorrhoea from oral sex followed by anal sex (or vice versa) is necessary to explain the high proportion of gonorrhoea infection at both the oropharynx and anorectum. Spicknall et al 12 proposed that future work should determine what additional features could be added to their models to replicate the observed rates of site-specific infection. We addressed some of these issues raised by Spicknall in our current models that investigated three additional types of sequential sexual practices involving the oropharynx and saliva. We acknowledge that observational studies to investigate these routes of transmission are in their infancy and that these studies are very complicated to undertake. We further acknowledge that our suggestion that the oropharynx plays a major role in transmission is debated, although there is now emerging evidence for its validity.7 9 In this context, we argue that the role of mathematical models is to explore complex questions such as the ones we pose in order to stimulate research to prove or disprove findings. If our findings were confirmed by empirical studies, this would provide additional potential interventions for the prevention of gonorrhoea at a time when transmission rates in MSM are rising rapidly and condom rates are falling.24
Our study has some limitations that need to be considered. First, there are few data on multisite infection available. To address this, we searched widely among the research community to identify seven data sets from four continents that included testing at all sites in MSM. We then repeated the calibration of our model using the data sets separately and demonstrated that the results are similar across all data sets. Second, we made a number of assumptions in our models when published data were not available. An example of this includes the proportion of men who may have both oral and anal sex during the same sexual encounter and then the order of the sexual acts among these men. According to a survey among men who identified as gay or bisexual in the USA,16 people may use oral sex as part of foreplay before anal intercourse,25 and our clinical experience suggests that oral sex generally precede anal sex in most circumstances. Therefore, we used an estimate of 80% or higher for the proportion of oral sex that preceded anal sex in the absence of any published studies. We performed sensitivity analyses by varying the proportion of acts where oral sex preceded anal sex. Third, there is considerable uncertainty in many indicators (such as the untreated duration of oropharyngeal infection, the duration of asymptomatic urethral infection and bacterial load at various anatomical sites), and these may affect the estimate of transmission.10 We performed univariate sensitivity analyses with varying the duration of asymptomatic oropharyngeal and anal infection and frequency of sexual practices. Fourth, we acknowledge that there may be other sequential sex practices we did not consider that would have made the model even more complex. Fifth, group sex was not included in our models. Sequential sexual practices with more than one sex partner are very complex. Because there is limited published information, our sequential sexual practices included two consecutive sex practices during the same sexual episode with a same-sex partner, and we did not consider this.
Two previously published mathematical models have investigated the multiple anatomical site transmission of N. gonorrhoeae between men. The first model, published by Zhang et al, did not consider multisite infection. This model found that transmission through kissing was a common cause of incident cases.10 We have refined this model to include multisite infection, and our model suggests that infection through kissing accounted for only 10.5%–20.8% of all new infections, although oropharyngeal infection accounted for 38.8%–60.6% of all new infections. Thus, efforts to control infection from or to the oropharynx appear important.
The second model was published by Spicknall et al, and unlike the model by Zhang et al, it did include multisite infection. Spicknall et al suggested that the contribution of transmission from each site was quite uncertain if models allowed all seven basic transmission routes (anal sex (urethra–anorectum and anorectum–urethra), penile–oral sex (oropharynx–urethra and urethra–oropharynx), rimming (oropharynx–anorectum and anorectum–oropharynx) and kissing (oropharynx–oropharynx)). They found that the oropharynx could contribute 0%–100% of all transmissions.12 Spicknall et al indicated that their models slightly overestimated the observed prevalence of single-site infections and slightly underestimated the proportions of multisite infections. Spicknall et al also proposed that future work should determine what additional features could be added to their models to resolve this problem. We addressed some of these issues raised by Spicknall in our current models that investigated three additional types of sequential sexual practices involving the oropharynx and saliva.
Our study has confirmed previous findings that the oropharynx is likely to be a key site for the transmission of N. gonorrhoeae, but our study also suggests that anorectal infection may also play a larger role than had previously been appreciated. Previous studies had indicated that anorectal infection was primarily transmitted to the penis but not common to other sites.8 26 Our multisite infection models, however, suggest that transmission from the anorectum to the oropharynx and from the oropharynx to the anorectum may give rise to multisite infection at both the oropharynx and anorectum. As well as being important in our model, oral exposure to faecal contamination is also a risk of sexually transmitted enteric infections. Shigellosis and hepatitis A outbreaks related to sexual practices have been noted in MSM in Australia, the UK and other developed countries.27 28 Rimming is a very common behaviour in MSM.11 Our model suggests that interventions directed towards limited faecal–oral exposure may be necessary and important to reduce gonorrhoea transmission.29
We specifically found that oral sex followed by anal sex appeared to be particularly important for our model, while the other two sequential transmission routes did not help replicate multisite infection at both oropharynx and anorectum. This may relate to the consistent observation in epidemiological studies that multi site infection at oropharynx and anorectum is the most common multisite infection, while infection at the oropharynx/urethra or anorectum/urethra are less common.12 19–23 It is also possible that the mediator for multisite infection at the oropharynx and anorectum is the penis when oral sex follows anal sex, and in this situation, the amount of saliva or material from the rectum is quite large and spread over a moderate surface area of the penis that has direct contact with the sites it may infect, thereby providing more opportunity for transmission. In contrast, for example, the mouth has less direct contact with the anus when oral sex is following by oral–anal sex, and therefore the opportunity for transmission is less. We hope this work encourages research to explore these possibilities and acknowledge there is very little empirical data at this time.
Some may argue that the questions about the transmission of N. gonorrhoeae should be addressed with empirical observational studies rather than hypothetical models. We agree that empirical studies are essential, but they would be very complicated and expensive to carry out. For example, if a case–control study investigated the risks for oropharyngeal gonorrhoea, then in addition to information on exposures to their partners’ penis, anorectum and oropharynx, the order of the exposures would also be required, and this would need to be repeated for every sexual partner over a specified period.30 The analysis of these data would also be complicated by the high degree of collinearity and necessitate a very large number of participants in addition to the concern that these events could be recalled accurately in such detail. Mathematical models could provide some guidance to researchers on what are the key behaviours to focus on.
This is the first model of gonorrhoea transmission to explore the role of sequential sexual practices. Our models may suggest that sequential sexual practices involving the oropharynx and saliva may exacerbate the transmission of N. gonorrhoeae infections at multiple anatomical sites in MSM. Our model suggests interventions on reducing the transmission from sequential sexual practices may be important for controlling the high proportion of multisite infection. Future empirical studies will be needed to confirm our model findings.
Sequential sexual practices involving the oropharynx and saliva may exacerbate the transmission of Neisseria gonorrhoeae infections at multiple anatomical sites in men who have sex with men.
Transmission of Neisseria gonorrhoeae from the anorectum to the oropharynx may give rise to multisite infection at both the oropharynx and anorectum.
Reducing the transmission of Neisseria gonorrhoeae from sequential sexual practices may be important for controlling the high proportion of multisite infection.
Data availability statement
All data relevant to the study are included in the article or uploaded as supplementary information. The data were all collected from published articles and reports. Data are available in the supplementary materials.
This study used secondary data analysis of datasets obtained from previous publications and therefore ethical approval was not required.
We thank Afrizal at the MSHC for extracting the 2018-19 MSHC data and Marjan Tabesh at the MSHC for providing the unpublished MSHC data. We thank Barbara Van Der Pol at the University of Alabama at Birmingham for providing data and insightful discussion. We also thank Dr. van Liere Gafs for providing the unpublished Dutch data.
Handling editor Laith J Abu-Raddad
Twitter @EricPFChow, @drdebwilliamson
Contributors XX, LZ and CKF conceived and designed the study; XX and LZ established the model; XX and LZ did the analysis; MS contributed in checking the analysis and gave overall feedback to the analysis; XX wrote the first draft; CKF provided critical insights for framing of the first draft; EPFC, JJO, CJPAH, DB, MS, FYSK, JSH, CKF and LZ revised the manuscript. All authors reviewed drafts and approved the final manuscript.
Funding EPFC, DW and CKF are supported by an Australian National Health and Medical Research Council (NHMRC) Emerging Leadership Investigator Grant (GNT1172873, GNT1174555 and GNT1172900, respectively). JJO is supported by an Australian NHMRC early career fellowship (APP1104781). MS was supported by the National Natural Science Foundation of China (grant number:11 801 435 (MS)), China Postdoctoral Science Foundation (grant number 2018M631134M631134), the Fundamental Research Funds for the Central Universities (grant number: xjh012019055, xzy032020026) and Natural Science Basic Research ProgramProgram of Shaanxi Province (Grant number: 2019JQ-187). LZ is supported by the National Natural Science Foundation of China (Grant number: 81950410639); Outstanding Young Scholars Support Program (Grant number: 3111500001); Xi’an Jiaotong University Basic Research and Profession Grant (Grant number: xtr022019003, xzy032020032); Epidemiology modeling and risk assessment (Grant number: 20200344) and Xi’an Jiaotong University Young Scholar Support Grant (Grant number: YX6J004).
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.