Background Increasing evidence indicates that Mycoplasma genitalium (Mgen) is a sexually-transmitted infection that can lead to pelvic inflammatory disease (PID) and possibly infertility. Resistance to azithromycin, which has been the first-line treatment, has been widely reported. To develop optimal testing and treatment guidelines, it is necessary to understand the natural history of Mgen and the burden of associated disease. Several observational studies have provided valuable data, but no study has synthesized the available evidence to estimate the population burden of Mgen-associated disease.

Methods The POPI study was a chlamydia screening trial recruiting sexually active female students aged ≤27 years in London, 2004–2006. Women provided vaginal samples at baseline and follow-up, and were assessed for one-year incidence of PID by genitourinary doctors using participant questionnaires and medical records. Mgen infections were identified retrospectively from stored samples, using NAATs. We used the published data on Mgen prevalence, persistence of infection over median 16 (range 12–21) months’ follow-up, and one-year incidence of PID in women infected or not infected with Mgen at enrollment. We conducted a Bayesian evidence synthesis using a simple (Susceptible-Infected-Susceptible) mathematical model of infection, with uninformative priors on all parameters.

Results In the POPI trial, 6.26% (1.82, 15.11)% (posterior median; 95% credible interval) of Mgen infections led to PID. We estimate that there were 1.96 (0.16,6.27) new Mgen-related PID cases per 1000 women per year, and a total of 6728 (537,21547) cases per year in 16–27-year-old English women. 10.8% (0.9,33.0)% of the current burden of PID is caused by Mgen infection.

Conclusion Our model synthesises different types of data to understand the burden of Mgen infection and PID. Further data will be included to increase the precision of estimates, which are currently subject to wide uncertainty. We recommend studies in men, with urethritis as the disease outcome, which could be analysed with a similar model.

Disclosure No significant relationships.