Inadequate performance of a risk score to predict asymptomaticNeisseria gonorrhoeaeandChlamydia trachomatisinfection among cisgender men who have sex with men

Abstract

ObjectivesEpidemiological treatment of persons who are sexual contacts to partners withNeisseria gonorrhoeae(NG) andChlamydia trachomatis(CT) often results in treatment of uninfected persons, which may increase the risk of antibiotic-resistant infections. We sought to identify the predictors of NG and/or CT infections to develop a risk score that could be used to limit epidemiological treatment to persons most likely to have these infections.MethodsWe included visits to the Public Health - Seattle & King County Sexual Health Clinic by asymptomatic cisgender men who have sex with men (MSM) aged ≥18 who presented as a sexual contact to partner(s) with CT or NG infection between 2011 and 2019. We used logistic regression to estimate the odds of CT and/or NG infections associated with demographic and clinical predictors, selecting the final set of predictors using the Akaike information criteria and obtaining score weights from model coefficients. We used a cross-validation approach to obtain average model discrimination from each of 10 models, leaving out 10% of the data, and evaluated sensitivity and specificity at various score cut-offs.ResultsThe final model for predicting NG or CT infection included seven predictors (age <35 years, HIV status, receptive oral sex in the prior 2 months, CT diagnosis, condomless receptive anal intercourse, condomless insertive anal intercourse and methamphetamine use in the prior 12 months). Model discrimination, as measured by the receiver operating curve, was 0.60 (95% CI 0.54 to 0.66). Sensitivity for detection of infection was ≥90% for scores ≥3, ≥5 and ≥7; specificity for these cut-offs was <16%. At scores ≥9, ≥12 and ≥14, specificity increased but sensitivity decreased to ≤76%.ConclusionsOur risk score did not sufficiently discriminate between asymptomatic MSM with and without NG/CT infection. Additional studies evaluating epidemiological treatment as a standard of care in diverse populations are needed to guide best practices in the management of contacts to NG/CT infection.