The cause of most cases of non-gonococcal urethritis (NGU) remains unknown. Chlamydia trachomatis accounts for 30–40%^1,2 and Mycoplasma genitalium is now known to cause a smaller proportion,^2,3 while other established causes are uncommon.⁴Ureaplasma urealyticum, Gardnerella vaginalis, Haemophilus species, and Streptococcus spp have been associated with NGU but their causal role is unproved.^4,5

In Australian general practice men are often treated when they present with symptoms of NGU before laboratory confirmation of urethritis on urethral smear. It is therefore useful to establish the prevalence of urethral pathogens in all men with urethral symptoms regardless of the presence of urethritis on the urethral smear. We conducted a case-control study of men with symptoms of NGU and controls examining the prevalence of known and suspected pathogens and associations with sexual practices.

METHODS

The study was conducted at Melbourne Sexual Health Centre, Australia, from December 2002 to March 2003 with ethics committee approval. Men were included if they had urethral symptoms (discharge, dysuria, or urethral discomfort) for less than 1 month. Men were ineligible if they had HIV infection, urethral gonorrhoea on Gram stain, male sexual partners, a current episode of genital herpes, or had taken antibiotics in the previous month. Controls were men without urethral symptoms attending the clinic during the same period, with the same exclusion criteria. Cases had a urethral smear collected and all participants provided a first pass urine specimen and completed a behavioural questionnaire.

Polymorphonuclear leucocytes on Gram stained urethral smear were counted in five oil immersion ×1000 polymorphs per high power fields (phpf). Urethritis was defined as the presence of 5 polymorphs/hpf or more. First stream urine from cases and controls was tested for streptococci, Haemophilus species, Gardnerella vaginalis, and Ureaplasma urealyticum by culture and Chlamydia trachomatis by ligase chain reaction (LCx, Abbott Diagnostics, Abbott Park, IL, USA) and Mycoplasma genitalium and Ureaplasma urealyticum by polymerase chain reaction (PCR).

A volume of 1 ml of urine was centrifuged and resultant pellet extracted using the automated MagNA Pure LC (Roche Molecular Biochemical, Mannheim, Germany) with the DNA Isolation Kit I protocol. Each extracted DNA was amplified by PCR for two targets utilising 5 μl aliquots of the DNA. Detection of M genitalium was performed as described previously.⁶ Detection of U urealyticum was performed using adaptation of a previously described assay targeting amplification of a 429 base pair sequence of urease structural gene to real time PCR using a Taqman probe.⁷

Data were analysed on SPSS for Windows Version 11, Chicago, IL, USA. Crude odds ratios and 95% confidence intervals were calculated for categorical variables and numerical variables were analysed using a t test or non-parametric equivalent. Two adjusted analyses were performed using logistic regression. In the first, all identified organisms were modelled to determine which were independently associated with NGU after adjusting for the primary behavioural risk factor, unprotected vaginal sex, and past urethral symptoms. In the second, odds ratios for the association of selected behavioural characteristics with NGU symptoms were assessed after adjusting for the main risk factor, unprotected vaginal sex. This study had a power of 80% to detect an odds ratio of 2.6 if the exposure variable was present in 30% of the controls (p = 0.05 two sided).

RESULTS

The 79 controls (mean age 30.2 years) were younger than the 80 cases (mean 35.8 years). Forty nine cases (61%) had 5 polymorphs/hpf or more and 19 (23%) had fewer than 1 p/hpf. Detection of Chlamydia trachomatis, Streptococcus spp, and Gardnerella vaginalis was statistically associated with urethral symptoms in the crude and adjusted analysis (table 1). When only cases with urethritis were included, both the crude and adjusted odds ratios remained essentially unchanged (data not shown). Mycoplasma genitalium was detected in five cases and one control, but this was not statistically significant.

Among cases the presence of 5 polymorphs/hpf or more in the urethral smear did not predict the presence of Chlamydia (OR 1.7 (95% CI: 0.5 to 5.4)) or any other organism (p>0.15). The presence of urethral discharge on history or examination was statistically associated with the detection of chlamydia in cases, OR 3.9 (95% CI: 1.1 to 15) and 3.9 (95% CI: 1.2 to 14) respectively as was a history of previous NGU symptoms (OR 2.4 (95% CI 1.1 to 5.4). No other symptoms or signs were associated with the detection of other organisms (p>0.2).

The behavioural factors associated with NGU are shown in table 1. In the adjusted analysis unprotected vaginal sex with more than one casual partner and unprotected anal sex with a regular partner were significantly associated with NGU, but unprotected oral sex was not.

DISCUSSION

In this case-control study of heterosexual men with urethral symptoms, cases were more likely to have Chlamydia trachomatis, Gardnerella vaginalis, and any streptococci isolated from their urine and to report past urethral symptoms, unprotected vaginal sex with a casual partner, and unprotected anal sex with their regular partner. These results are particularly applicable to health services where most men with urethral symptoms are treated in primary care without access to on-site microscopy.

Gardnerella vaginalis was isolated from significantly more cases (14%) than controls (3%) indicating that organisms associated with bacterial vaginosis (BV) may cause urethral symptoms in men. G vaginalis has been isolated from men with urethritis,⁵ and NGU has also been associated with the presence of BV in female partners.⁸

In this study unprotected anal sex with a regular partner was reported three times more commonly in cases than controls. It is possible that urethral irritation could result from exposure to rectal flora during anal sex.

Among men with urethral symptoms, urethritis by microscopic criteria did not predict the presence of Chlamydia trachomatis, whereas discharge was predictive. Several studies have also found that microscopic urethritis did not predict C trachomatis,^9,10 while Horner et al found discharge to predict C trachomatis.¹ The value of urethral symptoms over microscopy in predicting those likely to have C trachomatis, is supported by our finding that the prevalence of C trachomatis among urethritis negative cases (five of 31) was higher than among asymptomatic controls (one of 79; p = 0.002).

It is possible that our sample size, and the inclusion of cases without microscopic urethritis, prevented us from detecting weaker associations, such as those between individual organisms, microscopic urethritis, or certain behaviours. Ureaplasma urealyticum, Haemophilus influenzae, or any particular streptococcal species were not associated with urethral symptoms.

Acute urethral symptoms in a male population with a low prevalence of bacterial STIs may be due to other causes, including G vaginalis and anal sex. Our results support our practice of offering treatment for C trachomatis and partner notification to all men attending our clinic with symptoms of urethritis and a behavioural risk factor regardless of the presence of urethritis on their urethral smear.