Identifying cervical infection among pregnant women in Nairobi, Kenya: limitations of risk assessment and symptom-based approaches.

OBJECTIVES To examine characteristics of pregnant women associated with cervical infection, and to evaluate the accuracy of symptom-based and risk assessment systems which have been developed for identifying cervical infection in antenatal women. METHODS Interviews were conducted and physical examinations performed on 291 consecutive antenatal clinic attenders in nairobi, Kenya. Vaginal, cervical, urine and blood specimens were also obtained for analysis. RESULTS The following disease prevalences were observed: candidiasis 26.2%; trichomoniasis 19.9%; bacterial vaginosis 20.6%; any vaginal infection 53.8%; chlamydial cervicitis (CT) 8.8%; gonococcal cervicitis (GC) 2.4%; any cervical infection 10.8%. The only statistically significant association with GC and/or CT cervical infection was the presence of cervical friability (OR = 2.1, P = 0.05). There were trends towards associations with the presence of endocervical mucopus (OR = 2.6, P = 0.06), reporting a new sex partner in the past 3 months (OR = 2.2, P = 0.16) and reporting that a sex partner had an STD-related symptom (OR = 4.4, P = 0.13). There were no associations with other demographic, behavioural or medical characteristics. Risk scores previously developed for detecting GC/CT cervicitis in developing country antenatal populations generally performed poorly. CONCLUSIONS The prevalences of vaginal and cervical infection observed were extremely high among these "low risk" women. Owing probably to high levels of vaginal infection and to behavioural characteristics of this urban population, factors which elsewhere have been associated with cervical infection were not found to be so in this setting. Further work on symptom-based approaches and risk assessment for STD case detection in pregnant women is required before STD management recommendations can be generalised.