Background Chlamydia trachomatis (chlamydia) is an important cause of pelvic inflammatory disease (PID). Preventing PID is a main objective of chlamydia screening. There are many uncertainties about how and when bacteria spread from lower to upper genital tract. The potential impact of screening and treatment, which could interrupt ascending infection, might be affected by the timing of development of PID. Models are often used to investigate the potential impact of screening strategies on PID and should therefore include information about the timing of progression. We conducted a systematic review to determine how the progression from chlamydia to PID is described in mathematical models.
Methods We searched four electronic databases using search terms related to mathematical models and PID from the earliest date to 19 October 2009 without language restrictions. Eligible publications included progression from chlamydia to PID either using a decision tree or a mathematical model. We extracted information about how authors conceptualised the dynamics of chlamydia infection and the development of PID, and assumptions about rates of progression.
Results We identified 41 unique publications about chlamydia infection; 28 of these included PID in a static decision tree. The average percentage of women developing PID in decision analyses was 22.9% (range 10–35%, n=26). For five publications it was not clear how the described model worked. The other eight publications described progression from chlamydia infection to PID dynamically. Of these, two models incorporated PID as a state in a Markov-chain model, four used compartmental models and two used individual-based models. Explicit statements about model structure included the possibility that PID can occur uniformly during a woman's infection, that tubal damage occurs in the second half of the chlamydia infection, and that the model had the ability to vary PID development time. Twenty-eight publications did not mention the stage during a chlamydia infection that progression to PID happens.
Conclusion Most modelling studies do not consider dynamic aspects of C trachomatis transmission and the timing of progression to PID. The mechanisms proposed in studies that made explicit statements could be compared to examine the impact of screening. We suggest that explicit statements about the timing and rates of progression would help improve understanding of the pathogenesis of chlamydial complications and the potential effects of screening.