Background Screening programmes aim to detect and treat asymptomatic Chlamydia trachomatis (chlamydia) infections to prevent ongoing transmission and reduce the incidence of pelvic inflammatory disease (PID). The timing of development of PID after chlamydia infection might affect the potential impact of screening to interrupt ascending infection. We used mathematical models to study three hypothetical processes for the timing of progression from chlamydia infection to PID and determine which was most consistent with empirical data.

Methods We used data from the Prevention of Pelvic Infection (POPI) randomised controlled trial (RCT) that was designed to examine the effect of chlamydia screening on PID incidence over a follow-up period of 1 year. We developed a compartmental model that simulates the RCT results in a Susceptible-Infected-Susceptible framework and tracks the number of PID episodes in the screened and control groups. The model is fitted to the incidence of 1.9% in the absence of screening (POPI control group) and predicts PID rates when screening is implemented. The hypothesised processes were—PID develops at the beginning of an infection with chlamydia; PID can develop throughout the course of a chlamydia infection at a constant rate; and PID happens at the end of chlamydia infection before spontaneous clearance. We predicted the incidence of PID with each process and compared these with the observed cumulative incidence of 1.3% (95% CI 0.7 to 2.1%) in the POPI screened group. We took into account baseline chlamydia prevalence, screening uptake during the RCT, duration of infection and treatment failure.

Results The mathematical models suggested that the process by which PID develops during the course of a chlamydia infection was closest to the observed cumulative incidence in the screened group, but the process with PID at the end of chlamydia infection was also compatible with the empirical data. The process where PID develops at the very beginning of a chlamydia infection predicted a higher incidence of PID in the screened than the control group. Our model also allowed us to estimate the proportion of chlamydia infections that develops PID and the reduction in PID incidence due to long-term screening.

Conclusion This study suggests that the development of PID can happen during or, possibly, at the end of a chlamydia infection. This implies that screening for chlamydia might have a direct effect on PID prevention by interrupting ascending infection.