Overestimation of complication rates in evaluations of Chlamydia trachomatis screening programmes--implications for cost-effectiveness analyses.

BACKGROUND: Cost-effectiveness analyses of screening programmes for asymptomatic Chlamydia trachomatis infection suggest that screening at low prevalences in the population is cost-effective. However, the decision models in these studies are based on assumptions about the risk of complications, which are derived from the literature. Incorrect assumptions may lead to under- or overestimation of the effectiveness of screening. The first objective of this paper is to evaluate the assumptions about the probability of complications after an asymptomatic C. trachomatis infection. The second objective is to calculate alternative rates by using available data on the incidence of complications. METHODS: We identified cost-effectiveness studies via Medline, and evaluated these for the evidence for the quoted probabilities. In addition, the probability of complications was calculated for Amsterdam from available registration data. RESULTS: In the three studies that were identified, the assumptions for the rates of pelvic inflammatory disease (PID) (clinical and subclinical) after C. trachomatis infection varied from 15% to 80%, and for ectopic pregnancy, tubal factor infertility, and chronic pelvic pain after PID from 5-25%, 10-20%, and 18-30%, respectively. The assumptions were based on data from high-risk populations, case-control data, and data not accounting for misdiagnoses. Using data obtained from local registrations, we estimated the probability of a clinical PID (0.43%), ectopic pregnancy (0.07%), and tubal factor infertility (0.02%) for women with a current infection. These estimates were consistently lower than the estimates based on the literature. CONCLUSIONS: We argue that an overestimation of the current complication rates is likely. The effect of overestimation is potentially the greatest in populations with a low prevalence, since the currently assumed cost savings associated with screening may disappear when using more realistic estimates for complications