This paper presents the largest community-based estimate of chlamydia prevalence among young Australian women and Australia's first MG prevalence survey. Consistent with previous international reports, we found the chlamydia prevalence (4.9%) was higher than the MG prevalence (2.4%) among young women [9–11]. We also found some important clinical and epidemiological differences between chlamydia and MG in this cohort, suggesting different transmission dynamics between the two infections.
Firstly, it is possible that MG is less infectious than chlamydia requiring a greater "exposure" or direct genital or cervical contact to acquire MG. This is supported by the 100 fold lower organism load among samples from women with MG compared with chlamydia, and the finding that MG was more strongly associated with unprotected sex than chlamydia. Clearly, further partner studies are needed to investigate the transmission dynamics for MG and chlamydia to determine if and how transmission dynamics differ.
The clinical features associated with MG and chlamydia also differed, MG was associated with vaginal discharge, but chlamydia showed no associations with any reported symptoms. Studies of the association between MG and specific genital symptoms have been somewhat conflicting with some studies determining an association between MG and genito-urinary symptoms including vaginal discharge and dysuria , and other studies finding no association with symptoms . Overall, published data suggests that MG appears to be somewhat similar to chlamydia [10, 29–31]. Further to this, no associations were found between organism load and reported symptoms for either chlamydia or MG, which was also consistent with the other studies .
Younger women were more likely to have a prevalent chlamydia infection which is consistent with other research , although younger age was not associated with MG infection.
Antibiotic use in the two months prior to being tested demonstrated a protective effect against chlamydia but not for MG. This is most likely because chlamydia has been shown to be sensitive to a number of commonly prescribed antibiotics , and MG is less likely to be sensitive to the same prescribed antibiotics [23, 32–34].
We also found that MG but not chlamydia was associated with Indigenous status (Australian Aboriginal and Torres Strait Islander women). The number of Indigenous women in our study (n = 25) limited further exploration in the analysis. Nonetheless, these are the first prevalence estimates for MG in Indigenous Australian women and given that STI rates are generally higher in Indigenous women in Australia this is not a surprising MG finding .
Unlike other studies, we did not find any associations between chlamydia organism load and age or past history of chlamydia infection [30, 36]. We did find evidence to suggest that chlamydia serovar was associated with organism load. However, this was based on a small number of cases. Nevertheless, given that others have not found any association between serovar and organism load , and uncertainty remains as to whether serovar is associated with disease severity, further studies with larger sample sizes are needed to investigate serovar and organism load.
There were a number of limitations to our study. Firstly, our sample had a higher proportion of Australian-born, well-educated and sexually-active women than the general background population in Australia for the same age [26, 27], however, these are common findings in similar research studies investigating sexual health issues [2, 37]. It is difficult to assess the impact this may have had on our prevalence estimates because increased number of partners is often associated with increased prevalence [2, 38] and higher education levels tend to be associated with reduced prevalence. We also were unsuccessful in recruiting 34% of the eligible women who were approached in the clinics, and while there were no associations between age and participation, we have no other information about the women we were unable to recruit. Nevertheless, this participation compares favorably with other chlamydia prevalence surveys [39–41].
Another limitation was relying on self-reported genital symptoms; these have been found to be highly subjective, non-specific and frequently poorly associated with cervical STIs. Self-reporting of genital symptoms on questionnaires also do not always correlate well with clinician elicited symptoms . We audited the clinical notes of a sub-set of 100 women and found very poor correlation with genital symptoms reported in the clinical notes (data not shown). Women were far more likely to self-report symptoms on their study questionnaire than were recorded by their clinicians in their clinical notes at the time of recruitment.
There were limitations to the organism load analysis. Samples were self-collected and therefore the equal efficiency of sampling could not be assured, and as positive samples were subjected to a number of assays, the mean organism loads were not able to be normalized to number of cells per sample. However, we did find that the serovars detected in our study were consistent with those reported in international data (serovar E followed by serovar F) [30, 43].
There were a number of strengths to this study including the large sample size, the high participation rate of 66% and the broad range of geographical locations and socio-economic areas from where the women were recruited. Also, considering 66% of the women were recruited from general practice, and between 80 to 90% of young Australian women visit a general practice clinic each year , the study method chosen was likely to provide a broadly representative sample.
The prevalence of chlamydia in our study was higher (4.9%) compared with the only other Australian population-based chlamydia prevalence study for women in the same age range (3.7%)  but was similar to a small community-based study , and other studies involving young women [44, 45]. Importantly our data suggest chlamydia prevalence is still somewhat lower in Australia than some other countries, most notably the UK .
These are the first population data on MG prevalence in Australia and our findings are very similar to the population data to date from international studies[2.3% (95%CI:1.3, 3.2)] , and [3.4% (95% CI:2.7, 4.3)] , but are somewhat higher than a study in the U.S [0.8% (95% CI:0.4, 1.6)] . As increasing evidence supports a role for MG in PID and tubal factor infertility, MG is emerging as an important treatable STI in women. Worryingly, consistent with other published studies, we found that 1 g of azithromycin appears to be 85% effective at best for uncomplicated MG [24, 33]. Further, MG is less responsive to the doxycycline and cefoxitin based regimens used in the presumptive treatment with in women with PID . Clearly our data provide evidence that MG is not uncommon in young women in Australia, and impetus is needed for the commercialization of a diagnostic assay to improve the management of MG. This study also contributes to our understanding of the MG organism load in clinical samples. However, further studies are needed to be done to understand this compared with other STIs such as chlamydia and if there is any relationship between copy number and pathogenicity.