Chlamydia and gonorrhoea in pregnant Batswana women: time to discard the syndromic approach?

Abstract

Background: Chlamydia and gonorrhoea are major causes of morbidity among women in developing countries. Both infections have been associated with pregnancy-related complications, and case detection and treatment in pregnancy is essential. In countries without laboratory support, the diagnosis and treatment of cervical infections is based on the syndromic approach. In this study we measured the prevalence of chlamydia and gonorrhoea among antenatal care attendees in Botswana. We evaluated the syndromic approach for the detection of cervical infections in pregnancy, and determined if risk scores could improve the diagnostic accuracy.

Methods: In a cross-sectional study, 703 antenatal care attendees in Botswana were interviewed and examined, and specimens were collected for the identification of C trachomatis, N gonorrhoeae and other reproductive tract infections. Risk scores to identify attendees with cervical infections were computed based on identified risk factors, and their sensitivities, specificities, likelihood ratios and predictive values were calculated.

Results: The prevalence of chlamydia was 8%, and gonorrhoea was found in 3% of the attendees. Symptoms and signs of vaginal discharge did not predict cervical infection, and a syndromic approach failed to identify infected women. Age (youth) risk factor most strongly associated with cervical infection. A risk score with only sociodemographic factors had likelihood ratios equivalent to risk scores which incorporated clinical signs and microscopy results. However, all the evaluated risk scores were of limited value in the diagnosis of chlamydia and gonorrhoea. A cut-off set at an acceptable sensitivity to avoid infected antenatal care attendees who remained untreated would inevitably lead to considerable over-treatment.

Conclusion: Although in extensive use, the syndromic approach is unsuitable for diagnosing cervical infections in antenatal care attendees in Botswana. None of the evaluated risk scores can replace this management. Without diagnostic tests, there are no adequate management strategies for C trachomatis and N gonorrhoeae in pregnant women in Botswana, a situation which is likely to apply to other countries in sub-Saharan Africa. Screening for cervical infections in pregnant women is an essential public health measure, and rapid tests will hopefully be available in developing countries within a few years.

Figure 1

Receiver Operating Characteristics (ROC) curves for three levels of risk scores (sociodemographic, clinical and microscopy risk scores). The risk scores are based on multiple logistic regression analyses and used as a screening tool to identify N gonorrhoeae and C trachomatis. The risk scores are applied retrospectively on 703 antenatal care attendees in Gaborone, Botswana.