Objective: To generate simple algorithms for the diagnosis of cervical infection with Neisseria gonorrhoeae or Chlamydia trachomatis in female sex workers in Abidjan, Côte d'Ivoire and to evaluate their validity.
Methods: From October 1992 to the end of June 1993, female sex workers were interviewed and clinically examined at a confidential clinic. N gonorrhoeae was cultured on modified Thayer-Martin medium and C trachomatis was detected by polymerase chain reaction. The associations of gonococcal or chlamydial cervical infection with sociodemographic, behavioural, clinical, and biological factors were assessed and three algorithms were generated. The validity parameters of these diagnostic algorithms were calculated and compared to those of standard algorithms and mass treatment.
Results: Among 683 women, cervical infection was present in 239 (35%). The sensitivity an algorithm incorporating sociodemographic and behavioural factors and symptoms, of an algorithm incorporating clinical signs and simple laboratory tests, and of a combined algorithm was 83%, 86%, and 79% respectively while the specificity was 32%, 44%, and 54%, and the positive predictive value 40%, 46%, and 48% respectively. A standard algorithm incorporating only the symptom vaginal discharge, and a standard algorithm requiring both the symptom vaginal discharge and the presence of an endocervical mucopurulent discharge on examination had a sensitivity of 44% and 18%, a specificity of 75% and 95%, and a positive predictive value of 49% and 67% respectively.
Conclusions: The algorithms generated in this study may be useful for the control of cervical infections in female sex workers in resource poor settings in the absence of rapid, inexpensive, and accurate laboratory tests for the diagnosis of cervical infections.