Sample size considerations using mathematical models: an example with Chlamydia trachomatis infection and its...

Area of Science:

• Epidemiology
• Mathematical Modeling
• Clinical Trial Design

Background:

• Intervention success for infection complications depends on disease natural history.
• Assumptions on infection-sequelae timing impact predicted effect size and sample size calculations.
• Chlamydia trachomatis infection and pelvic inflammatory disease (PID) serve as a case study.

Purpose of the Study:

• To investigate mathematical modeling's utility in informing sample size calculations for randomized controlled trials (RCTs).
• To analyze the impact of infection natural history on sample size determination for RCTs.

Main Methods:

• A compartmental model was employed, mirroring a published RCT structure.
• Three distinct temporal relationships for PID development post-Chlamydia trachomatis infection were modeled: immediate, continuous, and late.
• The influence of natural history parameters on sample size requirements was examined.

Main Results:

• Assumed event rates and effect sizes in sample size calculations implicitly defined the infection-PID temporal relationship.
• Minor adjustments in assumed PID incidence and relative risk (RR) significantly altered the hypothesized PID development mechanism.
• Both RR and required sample size per group were found to be dependent on Chlamydia's natural history parameters.

Conclusions:

• Mathematical modeling enhances understanding of the temporal dynamics between infection and sequelae.
• Modeling demonstrates how natural history parameter uncertainties affect RCT sample size calculations.