Improving the accuracy of syndromic diagnosis of genital ulcer disease in Malawi

Abstract

Objectives/goal: Most resource-poor settings rely on syndromic criteria to diagnose genital ulcer disease (GUD). However, the etiologic pathogens of GUD vary temporally and geographically, and current criteria may not reflect changes in the prevalence of specific pathogens.

Study: In 1999, we estimated the prevalence of Treponema pallidum (Tp), herpes simplex virus (HSV), and Haemophilus ducreyi (Hd) in Malawi. We then used regression coefficients of independent correlates of HSV and Hd to develop weighted diagnostic algorithms, in which weights were beta-coefficients corresponding to each factor.

Results: Overall, a decrease in the proportion of sexually transmitted disease attributable to GUD was noted in 7 years. Thirty-five percent were attributable to HSV, 30% to H. ducreyi, and 4% to T. pallidum. Areas under the receiver operating characteristic curves for weighted and unweighted HSV diagnostic algorithms were 67.6% and 66.5%, respectively. There was no significant difference in the explanatory performance of the weighted and unweighted algorithms.

Conclusions: Unweighted algorithms can therefore be used to improve diagnostic accuracy of GUD.