Biological and behavioral risks for incident Chlamydia trachomatis infection in a prospective cohort.

Abstract

To identify biological and behavioral risks for incident Chlamydia trachomatis among a prospective cohort of young women followed frequently. Our cohort of 629 women from two outpatient sites was seen every 4 months (October 2000 through April 2012) for behavioral interviews and infection testing. C trachomatis was tested annually and any time patients reported symptoms or possible exposure using commercial nucleic acid amplification tests. Analyses excluded baseline prevalent C trachomatis infections. Risk factors for incident C trachomatis were assessed using Cox proportional hazards models. Significant risks (P<.10) from bivariate models were entered in a multivariate model adjusted for four covariates chosen a priori (age, race or ethnicity, condom use, study site). Backward stepwise elimination produced a final parsimonious model retaining significant variables (P<.05) and the four adjustment variables. The 629 women attended 9,594 total visits. Median follow-up time was 6.9 years (interquartile range 3.2-9.8), during which 97 (15%) women had incident C trachomatis. In the final multivariate model, incident C trachomatis was independently associated with human papillomavirus at the preceding visit (P<.01), smoking (P=.02), and weekly use of substances besides alcohol and marijuana (P<.01) since the prior visit. Among 207 women with available colpophotographs (1,742 visits), cervical ectopy was not a significant risk factor (P range=.16-.39 for ectopy as continuous and ordinal variables). Novel risks for C trachomatis include preceding human papillomavirus, smoking, and substance use, which may reflect both biological and behavioral mechanisms of risk such as immune modulation, higher-risk sexual networks, or both. Improved understanding of the biological bases for C trachomatis risk would inform our strategies for C trachomatis control.