Recent studies suggest that there has been an increase in risky sex behaviours associated with the transmission of sexually transmitted infections (STIs) among the general population.¹ In men having sex with men (MSM) in the UK, there have been reports of increased high-risk sexual behaviour^2,3 and a rise in the incidence of Neisseria gonorrhoeae and Chlamydia trachomatis.⁴ Furthermore, there have been marked increases in reported cases of infectious syphilis^5,6 and lymphogranuloma venereum.⁷

Previous studies on MSM have shown prevalence rates of 1–4% and 4–7% for pharyngeal and rectal C trachomatis, respectively, using culture or direct immunofluorescence.^{8,9,10,11,12,13,14} Studies using nucleic acid amplification tests (NAATs) have shown improved detection of C trachomatis and N gonorrhoeae from pharyngeal and rectal specimens.^15–19

Little data are available regarding the association between sexual behaviours of MSM and the presence of either pharyngeal or rectal C trachomatis and N gonorrhoeae. Demonstration of factors associated with an increased prevalence of infections in pharyngeal, urethral and rectal sites may assist the development of testing recommendations in MSM. By improving the detection of these organisms and treating the infected people, the transmission of infections will be reduced, and consequently, the transmission of HIV.

AIMS

This study aimed to determine the association between the presence or absence of symptoms and signs, and to establish behavioural factors associated with an increased prevalence of C trachomatis and N gonorrhoeae at pharyngeal, urethral and rectal sites using NAATs in MSM.

METHODS

Study population characteristics

MSM aged ⩾18 years attending for routine STI testing were eligible; exclusion criteria were declining to participate or receipt of any antibiotics in the previous 4 weeks. Patients completed a questionnaire on history of STIs, the number of male sexual partners in the preceding month, year and lifetime, and the number of protected and unprotected specific sexual practices undertaken with men in the previous month. Clinicians recorded presenting symptoms and signs (additional information is available at the website at http://sti.bmj.com/supplemental)

Microbiology

Samples from the pharynx, urethra and rectum (obtained at proctoscopy) were collected. Randomisation of specimens for each test was ensured by using pre-prepared sealed envelopes.

Urethral samples were subjected to LCR for C trachomatis and N gonorrhoeae, and microscopy and culture for N gonorrhoeae.

Pharyngeal and rectal samples

Culture specimens for C trachomatis, were stored in transport medium (2-sucrose phosphate) and inoculated into tissue culture using cycloheximide treated McCoy cells (in the first 270 patients only).

Microscopy (rectal only) and culture for N gonorrhoeae were carried out using Thayer Martin selective medium (Oxoid, Basingstoke, Hampshire, UK) in 10% CO₂, incubated at 37°C for 48 h should be part of tests for pharynx and rectum.

LCR for C trachomatis and N gonorrhoeae

Discrepant results were further analysed by direct immunofluorescence (Syva MicroTrak, Behring, Cupertino, California, USA) performed on the sample for culture for C trachomatis (available in the first 270 patients only).

After an interim analysis showing poor performance of culture for C trachomatis, the protocol was amended after the first 270 patients were recruited. SDA was performed on pharyngeal and rectal samples for C trachomatis and N gonorrhoeae, retrospectively on the specimens for culture of C trachomatis in the first 270 patients and prospectively thereafter.

LCR and SDA tests were carried out according to the respective manufacturer’s package insert instructions.

Interpretation of microbiological results

A result was considered to be positive if

• LCR was positive for C trachomatis or N gonorrhoeaeⁱ from urethral samples

• Positive culture for either C trachomatis or N gonorrhoeae was obtained from any site

or

• Two or more tests were positive for the same organism from the same site by different testing methods.

Sample size

Assuming a 2.5% prevalence of urethral C trachomatis infection, a sample size of 600 allows a 95% confidence interval (CI) of mean (standard deviation (SD)) width 2.5% (1.25%). Assuming that the prevalence of pharyngeal and rectal C trachomatis and the prevalence of N gonorrhoeae are similar, this study estimates the prevalence of both infections at these sites with similar accuracy. CIs were calculated using STATA V.7.0.

Statistical methods

Unadjusted odds ratios (ORs) and 95% CIs were used to examine the association between factors and a current diagnosis of either infection. Age and the behavioural factors are ordinal, relating to numbers of sexual episodes or partners, and the associations between these factors and infection were tested using the Mann–Whitney U test. Logistic regression was used to examine and test for the simultaneous associations of multiple risk factors with infection. For these ordinal factors, the primary test for association with infection is a test for a trend, obtained by fitting a linear term in the regression model. Although the p values quoted relate to these tests for trend, ORs for categories of these factors are presented for ease of interpretation. When obtaining a test for trend for the highly skewed behavioural factors, the log of the factor was used as the linear term in the model. All analyses were performed using STATA V.7.

RESULTS

In all, 613 MSM were recruited in a 25-month period between August 1999 and September 2001. Of these, 14 were excluded from the analysis because samples were unavailable from every site.

Chlamydia trachomatis

We found an overall prevalence (resolved positive) of 11% (66/599) for C trachomatis (95% CI 8.6% to 13.8%), including 1% (6/599) of men having more than one site infected.

The prevalence at each site was

• pharyngeal: 1.2% (95% CI 0.5% to 2.4%);

• urethral: 4.3% (95% CI 2.9% to 6.3%); and

• rectal: 6.5% (95% CI 4.7% to 8.8%).

Neisseria gonorrhoeae

We found an overall prevalence (resolved positive) of 16% (96/599) for N gonorrhoeae (95% CI 13.2% to 19.2%), with 4.8% (29/599) of men having more than one site infected.

The prevalence at each site was

• pharyngeal: 7.3% (95% CI 5.4% to 9.7%);

• urethral: 7.2% (95% CI 5.0% to 9.2%); and

• rectal: 7.3% (95% CI 5.4% to 9.7%).

In all, 21 men were coinfected: 21.9% of all those with N gonorrhoeae and 31.8% of all those with C trachomatis (table 1; supplementary material available online at http://sti.bmj.com/supplemental).

Performance testing methodologies

The sensitivity and specificity of LCR for the detection of C trachomatis and N gonorrhoeae from the pharynx and rectum exceeded 95.5%. The sensitivity and specificity of SDA for the detection of C trachomatis and N gonorrhoeae from the pharynx and rectum exceeded 86.6%. The predictive value of a positive LCR from the pharynx for C trachomatis and N gonorrhoeae was 58.3% and 78.6%, respectively. The predictive value of a positive LCR from the rectum for C trachomatis and N gonorrhoeae was 83% and 89.4%, respectively. The predictive value of a positive SDA result from the pharynx for C trachomatis and N gonorrhoeae was 100% and 60.9%, respectively. The predictive value of a positive SDA result from the rectum for C trachomatis and N gonorrhoeae was 92.3% and 89.6%, respectively. We found no significant differences in the sensitivity of SDA to detect pharyngeal and rectal C trachomatis and N gonorrhoeae for samples tested prospectively compared with those tested retrospectively (Fisher’s exact test; available as supplementary material at http://sti.bmj.com/supplemental). Culture was less sensitive when compared with LCR or SDA for the detection of C trachomatis and N gonorrhoeae from the pharynx (N gonorrhoeae only) and rectum (table 1).

Associations with C trachomatis and N gonorrhoeae infection

Associations between history of STIs, age, symptoms, signs and sexual behaviour with each infection were tested, adjusting for other risk factors.

Some parameters were missing from the questionnaires.

A high proportion of men reported having had previous STI (75.8%); 8.4% (50/599) reported themselves as being HIV positive. No significant association was found between ever having had any STI or previously having had C trachomatis or N gonorrhoeae and currently having either C trachomatis or N gonorrhoeae.

The mean (SD) age was 34.8 (8.4) years (range 18–71 years). We found no significant association between age and having C trachomatis at any site, although there was some evidence of declining prevalence of urethral C trachomatis with increasing age after adjusting for other risk factors (p = 0.09). A declining prevalence of N gonorrhoeae at rectal sites with older age was found after adjustment for other risk factors (p = 0.002; tables 2–4).

A total of 92 of 593 (15.5%) men reported pharyngeal symptoms or signs. These were not predictive of pharyngeal infection (supplementary material available online at http://www.stijournals.com/supplemental). If pharyngeal testing was restricted to men with symptoms or signs, only 1 of 6 (16.7%) and 8 of 44 (18.2%) of those with pharyngeal C trachomatis or N gonorrhoeae, respectively, would have been identified. In all, 36 of 501 (7.2%) men without pharyngeal symptoms or signs had pharyngeal N gonorrhoeae (table 1).

Urethral and rectal symptoms and urethral signs were strongly associated with having urethral or rectal C trachomatis or N gonorrhoeae. Only 8 of 402 (2.0%) and 3 of 402 (0.8%) men without urethral symptoms or signs had urethral C trachomatis or N gonorrhoeae, respectively. Men reporting any urethral symptom (OR 5.40, 95% CI 2.30 to 12.7) or having any urethral sign (OR 8.30, 95% CI 3.69 to 18.7) were more likely to have urethral C trachomatis compared with asymptomatic men without any urethral symptoms or signs. This relationship was even stronger for N gonorrhoeae; urethral symptom: OR 36.2, 95% CI 11.0 to 119 and urethral sign: OR 79.5, 95% CI 29.8 to 212 (supplementary material).

Men reporting any rectal symptom, 42 of 594 (7.1%), were more likely to have rectal C trachomatis (OR 3.95, 95% CI 1.69 to 9.3) or N gonorrhoeae (OR 4.76, 95% CI 2.2 to 10.5). The prevalence of rectal C trachomatis or N gonorrhoeae was 29 of 541 (5.4%) and 33 of 541 (6.1%), respectively, among those without rectal symptoms or signs. In these men, after adjusting for other factors, those reporting 1–4 episodes of unprotected receptive anal intercourse compared with those reporting none were more likely to have rectal C trachomatis (OR 3.72, 95% CI 1.56 to 8.86) or N gonorrhoeae (OR 2.92, 95% CI 1.28 to 6.66).

Men recruited to this study were highly sexually active. In the preceding month, year and total lifetime, a median of 2, 10 and 80 male partners, respectively, were reported. After adjusting for other risk factors, there were no significant associations between the number of male partners reported and having either C trachomatis and N gonorrhoeae at any site (tables 2–4).

We found no significant association between those men reporting higher numbers of specific sexual practices in the preceding month and either pharyngeal or urethral C trachomatis. After adjustment for other factors, the number of episodes of unprotected receptive anal intercourse was significantly associated with rectal C trachomatis (p = 0.03). We found a significant association between men reporting higher numbers of specific sexual practices in the preceding month and having N gonorrhoeae at all three sites. After adjusting for other risk factors, we found a significant association between those men reporting more episodes of unprotected insertive oral sex and having pharyngeal N gonorrhoeae (p = 0.005; tables 2–4).

In all, 188 of 558 (33.7%) men reported at least one episode of unprotected anal intercourse in the previous month. HIV-positive men reported significantly more episodes of unprotected anal intercourse in the previous month compared with men of unknown or negative HIV status. Men reporting at least five episodes of unprotected receptive or insertive anal intercourse were more likely to be HIV positive; OR 7.34 (95% CI 3.30 to 16.3, p<0.001) and 4.39 (95% CI 1.88 to 10.2, p<0.001), respectively (more information is available at http://sti.bmj.com/supplemental).

DISCUSSION

Testing in MSM routinely includes a urethral specimen or a first-voided urine sample for C trachomatis (NAAT or enzyme immunoassay), and microscopy and culture for N gonorrhoeae from pharyngeal (culture only), urethral and rectal samples. Currently, NAATs are only licensed for use in men on specimens from the urethra, although several studies demonstrate that LCR performed well for detecting both organisms from the pharynx and rectum.^16–18 LCR is no longer available but SDA performed well for the detection of both infections from the rectum (in addition to the urethra). The use of either test from the pharynx in a population with low prevalence may result in considerable numbers of false positives.

This testing policy, treatment of people with non-specific urethritis and epidemiological treatment for C trachomatis in those with a positive gonococcal culture compared with using LCR from all three sites, failed to identify 20 of 66 (30.3%) and 20 of 96 (20.8%) men with C trachomatis and N gonorrhoeae, respectively. Similarly, by comparison, using SDA from all three sites we identified 19 of 66 (28.8%) and 18 of 96 (18.8%) men with C trachomatis and N gonorrhoeae, respectively, who would not have had their infection diagnosed or treated.

We demonstrate that LCR and SDA performs less well for samples from the pharynx than the rectum for the detection of C trachomatis and N gonorrhoeae. However, 16 of 30 (53.3%) men positive only by LCR for C trachomatis and N gonorrhoeae from the pharynx and rectum were positive for the same infection at another site. By comparison, only 4 of 30 (13.3%) men positive only by SDA for N gonorrhoeae from the pharynx and rectum were positive at another site (supplementary material available online at http://sti.bmj.com/supplemental).

During the study period, 10% of MSM attending our service were recruited to the study. Recruitment required patients to have an additional urethral specimen taken and to undergo proctoscopy to obtain rectal specimens, both of which may have deterred patients to participate in this study. Despite this limitation, the prevalence of urethral C trachomatis using LCR and N gonorrhoeae by culture observed in this study are similar to a retrospective study of all MSM attending our clinic during a 12-month period.²²

Previous C trachomatis or N gonorrhoeae infections is one of the greatest predictors of future infection in heterosexuals.²³ However (after adjusting for behavioural risk factors), HIV status and history of STIs were not markedly associated with an individual having either infection in this population. The large proportion of men reporting previous STIs may account for the lack of an association shown by this study.

Urethral and rectal symptoms and urethral signs seem to be strongly associated with infection, whereas pharyngeal symptoms and signs are not. We found significant associations between some sexual practices and both infections at the relevant site. By using these factors alone to identify whom to test, we showed that a marked number of infections would be missed.

A high proportion (33.7%) of men, particularly those who were HIV positive, reported unprotected anal intercourse in the previous month. Recent studies also show similar levels of ongoing high-risk behaviour among potentially serodiscordant men.³ In this study, the HIV status of sexual partners was unknown—that is, how many and which sexual practices occurred with which sexual partner. Therefore, the significance of high-risk sexual practices can only be inferred.

This study clearly shows a high prevalence of both C trachomatis and N gonorrhoeae, high rates of partner acquisition and, despite safer sex messages, an alarmingly high level of high-risk sexual behaviour among this group of MSM.

Evidence suggests that STIs may enhance transmission of HIV.^24–28 It is plausible that pharyngeal and rectal C trachomatis or N gonorrhoeae may enhance HIV acquisition. Improving detection using NAATs for both infections will reduce the duration of infection and reduce the transmission of these STIs and, consequently, may have an effect on transmission of HIV in this population.