Article Text

Download PDFPDF

Original article
Risk factors for oropharyngeal gonorrhoea in men who have sex with men: an age-matched case–control study
  1. Vincent J Cornelisse1,2,
  2. Sandra Walker1,2,
  3. Tiffany Phillips1,
  4. Jane S Hocking3,
  5. Catriona S Bradshaw1,2,
  6. David A Lewis4,5,
  7. Garrett Paul Prestage6,
  8. Andrew E Grulich6,
  9. Christopher K Fairley1,2,
  10. Eric P F Chow1,2
  1. 1 Melbourne Sexual Health Centre, Alfred Health, Carlton, Victoria, Australia
  2. 2 Central Clinical School, Faculty of Medicine, Nursing and Health Sciences, Monash University, Melbourne, Victoria, Australia
  3. 3 Centre for Epidemiology and Biostatistics, Melbourne School of Population and Global Health, The University of Melbourne, Melbourne, Victoria, Australia
  4. 4 Western Sydney Sexual Health Centre, Western Sydney Local Health District, North Parramatta, New South Wales, Australia
  5. 5 Marie Bashir Institute for Infectious Diseases and Biosecurity, Sydney Medical School, The University of Sydney, Westmead, New South Wales, Australia
  6. 6 The Kirby Institute, UNSW Australia, Kensington, New South Wales, Australia
  1. Correspondence to Dr Vincent J Cornelisse, Melbourne Sexual Health Centre, Alfred Health, Carlton, VIC 3053, Australia; vcornelisse{at}mshc.org.au

Abstract

Objectives Oropharyngeal gonorrhoea is common among men who have sex with men (MSM). We aimed to clarify which oral sex practices were independent risk factors for oropharyngeal gonorrhoea: tongue kissing, receptive oro-penile sex (fellatio) or insertive oro-anal sex (rimming), and whether daily use of mouthwash and recent antibiotic use was protective.

Methods In 2015, we conducted an age-matched case–control study of MSM who attended the Melbourne Sexual Health Centre. Cases had tested positive for oropharyngeal gonorrhoea by nucleic acid amplification testing, and controls had tested negative. Questionnaire items included tongue kissing, oral sex practices, condom use, recent antibiotic use, mouthwash use and alcohol consumption.

Results We identified 177 cases, age matched to 354 controls. In univariable analyses, cases were 1.90 times (95% CI 1.13 to 3.20) more likely than controls to have had casual sexual partners (CSP) in the preceding 3 months, were 2.17 times (95% CI 1.31 to 3.59) more likely to have kissed CSP and were 2.04 times (95% CI 1.26 to 3.30) more likely to have had receptive oro-penile sex with CSP. Oropharyngeal gonorrhoea was not associated with insertive oro-anal sex or mouthwash use. The number of CSP for tongue kissing and receptive oral sex and total CSP were highly correlated, and in multivariable analysis neither kissing nor receptive oro-penile sex was significantly associated with having oropharyngeal gonorrhoea, after adjusting for total number of CSP.

Conclusions The finding that oropharyngeal gonorrhoea was associated with a higher number of sexual partners but not specific sexual practices highlights the need for further research in the area of gonorrhoea transmission to define the probability of transmission from specific sex acts.

  • neisseria gonorrhoea
  • gay men
  • oral sex
  • homosexuality
View Full Text

Statistics from Altmetric.com

Introduction

In Australia, the number of notified cases of gonorrhoea is rising rapidly. In 2009, there were 8000 notifications, and by 2015 these reports had risen by 131% to 18 500 cases. The vast majority of cases in Australian urban areas occur in gay and other men who have sex with men (MSM).1 While some of the increase in notifications has resulted from increased testing with nucleic acid amplification tests (NAAT),2 there has also been a real increase in the prevalence of infection, as indicated by increasing culture positivity.3 The increase in gonococcal infections also increases the probability of the emergence of antimicrobial resistance.4 This highlights the need for a better understanding of gonorrhoea transmission in order to improve gonorrhoea control.

In MSM, gonorrhoea can occur at the urethra, rectum and oropharynx, with the oropharynx being the most common site in some studies.2 5 6 Conventional wisdom in the sexual health professions assumes that oropharyngeal gonorrhoea is primarily acquired as a result of performing receptive oro-penile sex (‘fellatio’). However, the vast majority of urethral gonorrhoea is symptomatic7 8 and therefore the urethra seems an unlikely source for a sexual partner’s oropharyngeal gonorrhoea, if we assume that men with symptomatic urethral gonorrhoea are more likely to seek treatment than to seek insertive fellatio. This assertion is supported by findings from the Health in Men (HIM) study in Australia, which found that oropharyngeal gonorrhoea in MSM is not associated with receptive fellatio after adjusting other risk factors.9 Some authors have hypothesised that oropharyngeal gonorrhoea may be acquired from a partner’s oropharynx by tongue kissing, and that onwards transmission from the oropharynx may occur to the oropharynx, urethra and anus of sexual partners through tongue kissing, receptive fellatio and insertive oro-anal sex (‘rimming’), respectively.10 11 This hypothesis is supported by findings that Neisseria gonorrhoeae can be grown from expectorated saliva from a substantial proportion of patients with oropharyngeal gonorrhoea,12 13 and that MSM with oropharyngeal gonorrhoea diagnosed by culture have substantial bacterial loads in expectorated saliva,14 suggesting a potential for transmission. In addition, insertive fellatio is an independent risk factor for urethral gonorrhoea,15–17 and saliva use as a lubricant for anal sex has been identified as a risk factor for ano-rectal gonorrhoea.18 However, it remains unclear to what extent kissing contributes to the transmission of N. gonorrhoeae. Early case reports described transmission of N. gonorrhoeae by kissing,19 20 but the HIM study, an Australian cohort study of 1427 HIV-negative MSM in which 65 men tested positive for oropharyngeal gonorrhoea by NAAT, did not confirm this risk factor. While ‘wet kissing’ was significantly associated with oropharyngeal gonorrhoea in univariable analysis, this association did not persist after adjusting for confounding factors such as age, known contact with gonorrhoea, number of sexual partners and other sexual practices.9 Younger MSM are more likely to have oropharyngeal gonorrhoea,3 9 and they more commonly tongue kiss sexual partners than do older MSM,21 so the inclusion of participants’ age may have resulted in collinearity in their multivariable model and may have obscured an association between kissing and oropharyngeal gonorrhoea. To assess this further, we conducted an age-matched case–control study to assess associations between oropharyngeal gonorrhoea and specific oral practices such as tongue kissing, insertive rimming and receptive fellatio, and whether the daily use of mouthwash or recent use of antibiotics was protective.

Methods

Study design, setting and participants

We conducted an age-matched 1:2 case–control study of MSM who attended the Melbourne Sexual Health Centre (MSHC), Australia, for STI screening between 1 March 2015 and 30 September 2015. All men of age 16–80 years who had sex with a man in the past 12 months and had an oropharyngeal swab for gonorrhoea were invited to complete our study questionnaire. The MSHC screening protocol specifies that all MSM are offered an oropharyngeal swab for gonorrhoea, regardless of whether they report a history of oral sex, in line with the Australian STI testing guidelines for MSM.22

Laboratory methods

Gonococcal testing was performed using the Aptima Combo 2 (AC2. Hologic, Marlborough, Massachusetts, USA) assay for screening, and the Aptima GC (AGC. Hologic) test for confirmation of all positive samples, on the Panther Platform (Hologic).23

Matching and sample size

Cases were defined as MSM who tested positive for oropharyngeal N. gonorrhoeae by NAAT, and controls were MSM who tested negative for oropharyngeal N. gonorrhoeae by NAAT, regardless of gonorrhoea infections at other anatomic sites. Two controls were age-matched to each case, being no more than 2 years older or younger than cases. Controls were randomly selected from MSM who completed the survey during the study period and who tested negative for oropharyngeal gonorrhoea. Sample size calculations were based on the assumption that risk factors for oropharyngeal gonorrhoea would be present in 30% of controls; hence, a total of 147 cases and 294 controls would be required to detect a minimum odds ratio (OR) of 1.8 (assuming a power of 80% and a two-sided significance level of 5%).9 This study included 177 cases and 354 controls, and it had 87% power to detect a difference using the significance level of 5%.

Behavioural data source

Participants were asked to complete a questionnaire on kissing and oral sex practices with regular (RSP) and casual (CSP) sex partners in the preceding 3 months, including the number of sexual partners and the number of sex partners with whom participants engaged in tongue kissing, receptive fellatio (‘oral sex’, specifying whether with or without ejaculation) and insertive rimming. The questionnaire also asked about condom use for receptive fellatio, antibiotic use within the month prior to testing for oropharyngeal gonorrhoea, the frequency of mouthwash use and habits of alcohol consumption, including the frequency of consumption and the type of alcoholic beverages consumed (beer, wine or spirits). Mouthwash use categories were ‘daily’, ‘weekly’, ‘monthly’, ‘less than monthly’ and ‘less than 6 monthly’, but this variable was recoded to a binary variable (‘daily’ vs ‘less than daily’) for analysis.

Statistical methods

In our unadjusted analyses, using conditional logistic regression analyses, we calculated ORs and 95% confidence intervals (CIs) for associations with oropharygeal gonorrhoea with each of the following binary variables: antibiotic use (yes vs no), frequency of mouthwash use (daily vs less than daily), RSP (yes vs no), CSP (yes vs no), CSP tongue kissed (yes vs no), CSP for receptive fellatio (yes vs no) and CSP for insertive rimming (yes vs no). The continuous variables of age, ‘number of CSP’, ‘number of CSP tongue kissed’, ‘number of CSP for receptive fellatio’ and ‘number of CSP for insertive rimming’ all had distributions that were highly right skewed, and hence for continuous variables the difference between cases and controls was assessed by Mann-Whitney tests, and the median and interquartile range (IQR) were reported.

We investigated the correlation between several risk factors using non-parametric Spearman’s rank correlation. This showed strong correlation between several independent variables, which resulted in multicollinearity in our multivariable model, as signified by large variance inflation factors (VIFs). To correct this variance inflation, we removed variables with VIFs greater than 4,24 but this reduced the model to containing only variables on the use of mouthwash and antibiotics, and variables on whether participants had an RSP and their number of CSP in the preceding 3 months (log transformed, x=ln(y+0.1)).

To attempt to overcome multicollinearity, we constructed binary variables for the proportion of CSP whom participants kissed, and for whom participants performed receptive fellatio with ejaculation, receptive fellatio without ejaculation and insertive rimming. The cut-offs for these new binary variables were based on the quartile nearest to the beta coefficient of the linear regression analysis of the natural log-transformed input variables. For example, linear regression of the independent variable ‘natural log of number of CSP’ and the dependent variable ‘natural log of number of CSP kissed’ produces a beta coefficient of 0.83 (x=0.13+0.83 y); hence, the new binary variable of ‘proportion of CSP kissed’ was split at the third quartile (>75%  or ≤75%), as this is the quartile nearest to the beta coefficient. We then conducted a multivariable conditional logistic regression analysis that included these binary variables to assess the difference between cases and controls in terms of the proportion of CSP they kissed, and for whom they performed receptive fellatio and insertive rimming.

Statistical analyses were conducted using STATA (V.13.1).

Results

Participants

A total of 3371 MSM attended the clinic during the study period, of whom 1150 (34%) were invited to participate, and 989 (86%) MSM completed the questionnaire. A total of 2221 men (66%) were not invited to participate for service delivery considerations. Of the men who completed the survey, we identified 177 cases with a positive NAAT for oropharyngeal gonorrhoea, who were age matched to 354 controls. Of the 177 cases, 54 (46%) tested positive for rectal gonorrhoea by NAAT and 2 (1%) tested positive for urethral gonorrhoea by NAAT. Of the 354 controls, 8 (2%) tested positive for rectal gonorrhoea by NAAT and 3 (1%) tested positive for urethral gonorrhoea by NAAT. The median age of cases and controls was 29 years (IQR 25 to 34; P=0.995 for difference between cases and controls). Seventeen cases had previously been diagnosed with HIV, and a further two cases were diagnosed with HIV during the study. One control had previously been diagnosed with HIV and another two controls were diagnosed with HIV during the study.

Descriptive data and univariable analyses

There was no difference between cases and controls in terms of consistent use of condoms for receptive fellatio (4.2% of cases, 3.8% of controls, P=1.000), antibiotic use in the preceding month (16.2% of cases, 15.3% of controls, P=0.965) or the frequency of mouthwash use (29.0% of cases and 30.9% of controls used mouthwash daily, P=0.682) (table 1). There were no differences in alcohol consumption habits. For example, for the consumption of spirits, coded as a binary variable of ‘daily’ versus ‘not daily’ consumption, there was no difference between cases and controls (OR 0.86, 95% CI 0.22 to 3.31).

Table 1

Univariable conditional logistic regression analyses of oropharyngeal gonorrhoea risk factors between cases with oropharyngeal gonorrhoea and controls without oropharyngeal gonorrhoea, consisting of men who have sex with men attending the Melbourne Sexual Health Centre in 2015

Cases were less likely than controls to have one or more RSP (OR 0.56, 95% CI 0.39 to 0.81, P=0.002) (table 1), but there was no difference between cases and controls in terms of whether they kissed, fellated or rimmed their RSP (P>0.2). Reported numbers of RSP were inversely correlated with reported numbers of CSP (rs=−0.22, P<0.001) (table 2).

Table 2

Spearman’s rank correlation of risk factors for oropharyngeal gonorrhoea as independent continuous variables

Cases were 1.90 times (95% CI 1.13 to 3.20) more likely than controls to have CSP in the preceding 3 months (median 5 in cases vs 3 in controls, P<0.001), were 2.17 times (95% CI 1.31 to 3.59) more likely to have kissed CSP (median 4 in cases vs 2.5 in controls, P<0.001) and were 2.04 times (95% CI 1.26 to 3.30) more likely to have fellated CSP (median 5 CSP in cases vs 3 CSP in controls, P<0.001). There was no association between oropharyngeal gonorrhoea and having rimmed CSP (binary), although cases had rimmed slightly more CSP than controls (median 1 vs 1, P=0.035) (table 1). There were no differences between cases and controls in terms of the duration of tongue kissing or rimming during a typical episode of sex with CSP, but cases estimated that they spent slightly longer performing fellatio (median 10 min and mean 14 min for cases, median 10 min and mean 10 min for controls, P=0.038). There were no differences in the time since last act of tongue kissing, receptive fellatio and insertive rimming (table 1).

Multivariable analyses and assessment of collinearity

Spearman’s rank correlation analysis showed strong correlation between several variables, including between number of CSP and number of CSP for receptive fellatio (rs=0.88, P<0.001), number of CSP and number of CSP tongue kissed (rs=0.91, P<0.001), number of CSP kissed and number of CSP for receptive fellatio (rs=0.87, P<0.001), and several other significant correlations of weaker magnitude (table 2). To reduce multicollinearity, we removed variables with a VIF greater than 4, after which a multivariable analysis found that only the number of CSP was significantly associated with having oropharyngeal gonococci (adjusted OR (aOR) 1.25, 95% CI 1.10 to 1.43) (table 3).

Table 3

Multivariable conditional logistic regression analysis of oropharyngeal gonorrhoea risk factors between cases with oropharyngeal gonorrhoea and controls without oropharyngeal gonorrhoea, consisting of men who have sex with men attending the Melbourne Sexual Health Centre in 2015. Variables with a VIF greater than 4 not included in the analysis.

After excluding the variable "number of CSP", another multivariable regression analysis of kissing and oral sex practices with CSP as binary variables showed a non-significant difference between cases and controls in proportion of CSP kissed, with a larger percentage of cases having kissed more than 75% of their CSP, compared with controls (aOR 1.55, 95% CI 0.96 to 2.49). No other sex practices neared significant difference (table 4). Low VIFs confirmed a lack of collinearity between sexual practices in this regression model.

Table 4

Multivariable conditional logistic regression analysis of kissing and other oral sexual practices with casual sexual partners between cases with oropharyngeal gonorrhoea and controls without oropharyngeal gonorrhoea, consisting of men who have sex with men attending the Melbourne Sexual Health Centre in 2015.

Binary variables are split around the quartile nearest to the beta coefficient of the linear regression analysis of the natural log-transformed input variables. The input variables were natural log transformed prior to analysis as the untransformed variables were highly right skewed. For example, linear regression of the independent variable ‘natural log of number of CSP’ and the dependent variable ‘natural log of number of CSP tongue kissed’ produces a beta coefficient of 0.83 (x=0.13+0.83 y); hence, the new binary variable of ‘proportion of CSP tongue kissed’ was split at the third quartile (>75%  or ≤75%), as this is the quartile nearest to the beta coefficient.

Discussion

This case–control study identified that the number of CSP was the principal risk factor for oropharyngeal gonorrhoea. We attempted to identify which specific sexual practice with CSP was responsible for this increased risk, but the number of CSP for tongue kissing, receptive fellatio and insertive rimming were so closely correlated with the total number of CSP that we could not assess their independent contributions in a multivariable model. In an attempt to overcome multicollinearity, we further examined differences in kissing and oral sex practices between cases and controls in a multivariable model that did not include the total number of CSP, and found that the only sexual practice with CSP that had a near-significant association with oropharyngeal gonorrhoea was kissing. It is likely that different sexual practices do carry different risk of acquiring oropharyngeal gonorrhoea, but the concurrent nature of oral sex practices makes it difficult to investigate this.

The use of antibiotics in the past month was not associated with a lower rate of oropharyngeal gonorrhoea despite about one in seven men having used them. While this association has not previously been assessed for gonorrhoea infections, recent antibiotic use is known to be associated with a reduction in incident chlamydia infections in young women,25 and countries with higher rates of antibiotic use have lower rates of chlamydia among women.26 The different effect of recent antibiotic use on gonorrhoea compared with chlamydia may be due to differences in antibiotic susceptibility of these organisms. Chlamydia trachomatis is generally sensitive to commonly used antibiotics such as macrolides, doxycycline and beta-lactams. N. gonorhoeae, on the other hand, is increasingly resistant to most classes of antibiotics.27 28 Out of the 79 participants who reported antibiotic use in the preceding month, the majority (58%) did not know what antibiotic they had used. The remainder mainly used amoxicillin (20%), azithromycin (9%) or doxycycline (8%). The four most commonly supplied antibiotics in Australia are amoxicillin (35%), cephalexin (19%), roxithromycin (6%) and doxycycline (6%), together accounting for 66% of supplied antibiotics.29 Current patterns of antimicrobial resistance in gonococcal isolates in Australia28 suggest that most of these antibiotics would not be particularly effective at reducing oropharyngeal carriage of gonococci, except perhaps cephalexin. Furthermore, a recent trial of doxycycline postexposure prophylaxis in MSM suggested that doxycycline does not prevent gonorrhoea.30

We wished to explore factors that may protect against oropharyngeal gonorrhoea. A recent laboratory study and randomised controlled trial found that a common commercially available mouthwash has antigonococcal activity in vitro and may also be effective at reducing oropharyngeal gonorrhoea in vivo when gargled for 1 min and when oropharyngeal swabs were collected 5 min after gargling.31 In our study, mouthwash use on a daily basis was not associated with a lower risk of oropharyngeal gonorrhoea despite about a third of participants using it. There are several possible explanations for failing to find an association in our current study. First, daily use of mouthwash may not reduce oropharyngeal carriage of gonorrhoea. Second, we have no data on exactly which mouthwash brands were used by patients. In the aforementioned laboratory study, only two mouthwashes from a single brand were tested and findings were similar,31 but there are more than 80 commercially available mouthwashes in Australia with different ingredients, and it is likely that only certain types of mouthwash are effective against gonorrhoea. Third, we have previously shown that use of mouthwash reduced the detection of N. gonorrrhoeae at the tonsillar fossea, but does not reduce the detection of N.gonorrhoeae at the posterior oropharyngeal wall even after gargling.31 Clinicians at the MSHC routinely sample both the tonsils and oropharynx when taking a swab for oropharyngeal gonorrhoea,32 using a single swab, so our data do not permit us to differentiate between gonorrhoea of the oropharynx and tonsils. Fourth, we do not know how the participants used mouthwash, the typical duration of use or whether they gargled or rinsed. We are currently conducting a large multisite randomised controlled trial of a mouthwash with antigonococcal activity to assess whether this reduces gonorrhoea at the tonsillar fossae and/or oropharynx.33

Cases were less likely than controls to have an RSP, suggesting that having an RSP is protective against oropharyngeal gonorrhoea. However, this is probably because participants with an RSP were less likely to have CSP, and the risk of oropharyngeal gonorrhoea was directly proportional to the number of CSP. Alternatively, it is possible that having a regular partner may be truly protective against having undetected oropharyngeal gonorrhoea: the regular partner of a man with oropharyngeal gonorrhoea is at risk of urethral gonorrhoea, and urethral gonorrhoea is almost always symptomatic,7 so it is likely that these men would be notified by their regular partner of their gonorrhoea carriage, and hence they would likely receive timely treatment and as such have a lower prevalence of oropharyngeal gonorrhoea. It is likely that MSM are less likely to notify their CSP of a urethral gonorrhoea infection, as they may not have contact details for CSP or may feel less inclined to make contact.

In addition to the limitations discussed, a potential source of selection bias is that not all MSM who attended the MSHC were asked to participate in this study. On most days, the clinic runs at full capacity, and in order to ensure efficient service delivery it is not possible to ask all patients to participate in studies.

Oropharyngeal gonorrhoea may be more important in the transmission of gonorrhoea between men than was previously thought,10 and given that the prevalence of gonorrhoea continues to increase among MSM, it is important that we identify the sexual practices that confer the greatest risk of oropharyngeal gonorrhoea. This information will assist public health practitioners to formulate evidence-based prevention strategies, which may require interventions that specifically target oropharyngeal gonorrhoea.10 Future studies that seek to address this question will need to ask men about all their tongue kissing partners, not just the proportion of sex partners they have kissed,10 and ask about the number of partners who were kissed but with whom they did not have other forms of sexual contact. These studies will likely need nuanced information on partner types, partnership concurrency and group sex activities. For example, recent work has indicated that MSM commonly have concurrent regular partners,34 and that some of these partnerships (termed ‘fuckbuddies’ or ‘sex buddies’) confer greater risk of STIs than other regular partners.35 Such detail will need to be captured in order to accurately assess the contribution of tongue kissing to gonorrhoea transmission.

Key messages

  • It has been hypothesised that kissing may be a significant contributor to oropharyngeal gonorrhoea transmission.

  • Compared with controls, in univariable analysis men with oropharyngeal gonorrhoea were twice as likely to have casual sexual partners.

  • Compared with controls, in univariable analyses men with oropharyngeal gonorrhoea were twice as likely to have engaged in kissing and other oral sex practices.

  • Due to concurrency of practices, we were unable to determine independent contributions of kissing and oral sex practices to the risk of oropharyngeal gonorrhoea.

References

View Abstract

Footnotes

  • CKF and EPFC contributed equally.

  • Handling editor Nigel Field

  • Contributors EPFC and CKF designed the study. VJC conducted the data analysis and wrote the first draft of the manuscript. SW and TP were involved in data collection and management. All authors contributed to the interpretation of the study findings, revised the manuscript critically for important intellectual content and approved the final version of the manuscript.

  • Funding This work was supported by the National Health and Medical Research Council (NHMRC) programme grant (number 568971). EPFC is supported by the Australian NHMRC Early Career Fellowship (number 1091226). VJC is supported by a stipend from the Research Training Program of the Australian Government’s Department of Education and Training.

  • Competing interests DAL’s organisation receives funding for External Consultancy work for GlaxoSmithKline. The other authors have no conflict of interest to declare.

  • Ethics approval Ethical approval was obtained from the Alfred Hospital Ethics Committee (544/14).

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

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.