P3.110 How should we monitor chlamydia control programme effectiveness? comparing performance indicators using evidence synthesis to estimate local incidence and prevalence from surveillance data

Introduction Understanding patterns of chlamydia incidence and prevalence is important for addressing inequalities, planning cost-effective control programmes and defining performance indicators. Population-based surveys are costly; the best data for England come from the Natsal surveys which are only available once per decade, and are nationally representative but not powered to compare localities. Estimates at finer spatial and temporal scales are required. Methods We present a method for estimating local incidence and prevalence by modelling the infection, testing and treatment processes. Parameters describing natural history and treatment-seeking behaviour are informed by the literature or calibrated using national prevalence estimates. By combining them with local-level surveillance data on numbers of chlamydia tests and diagnoses in England, we estimate local screening rates, incidence and prevalence. Results There is substantial local-level variation in infection burden. Highest infection rates are in the most-deprived areas?– but deprivation is a poor predictor of prevalence, with large variation within each deprivation quintile. Importantly, positivity is not a reliable proxy for prevalence. Most localities that meet the current performance target of 2300 annual diagnoses per 1 00 000 population have higher incidence and prevalence than most that do not, and the target may be unrealistic for many localities. Conclusion Our approach provides local estimates of chlamydia incidence and prevalence from surveillance data, which can be used to inform analysis of local variation and assess local control programmes. Many localities are unlikely to be able to meet the current annual diagnosis rate target, and successful prevention interventions like condom promotion make the target harder to reach. A better performance indicator could be the proportion of incident infections that are treated, as estimated by our model, since a higher value is always better for public health and other prevention activities make a higher value easier to achieve.