OBJECTIVE

To study the validity and performance of a number of rapid indicators for the diagnosis of sexually transmitted infections (STIs) in female prostitutes in Dakar, Senegal; characteristics of these indicators were rapidly obtainable, easy to perform, accurate, useful at district level, and reasonable cost.

METHODS

An STI prevalence study in female prostitutes (n = 374) seen at the STD clinic in Dakar, Senegal was done; a history, clinical examination, simple laboratory tests, and "gold standard" microbiological tests were performed. For a number of sociodemographic data, actual or past symptoms of STI, clinical signs, and rapid laboratory tests, validity variables, performance characteristics, and likelihood ratios for detection of gonococcal or chlamydial cervical infection were determined.

RESULTS

Cervical infection (chlamydial or gonococcal) was present in 24.9% of prostitutes; 46% had trichomoniasis and 29.4% had syphilis. Young age, abnormal vaginal discharge, endocervical mucopus, a positive leucocyte esterase test on urine, and 10 or more leucocytes in Gram stained smears of vaginal, cervical, or urine samples were significantly associated with cervical STI. Some of the rapid indicators had high sensitivity, others high specificity but none had acceptable overall validity. None of the indicators had at the same time a sensitivity above 50% and a positive predictive value above twice the background prevalence of cervical infection. 10 or more leucocytes in the cervical smear had a likelihood ratio of 1.83 increasing pretest probability of 24.9% to post-test probability of 38%, the best result obtained by any of the rapid indicators.

CONCLUSIONS

Rapid indicators of cervical STIs are insufficiently valid, which largely restricts their usefulness to high STI prevalence situations for instance, in prostitute populations and in STD patient management.