Sexually transmitted disease risk assessment used among low-risk populations in east central Africa: a review.

In sub-Saharan Africa where sexually transmitted diseases (STDs) are prevalent but health resources are scarce, algorithms that act as STD/HIV risk assessment (RA) tools for family planning (FP) and maternal/child health (MCH) clinic attenders may be useful. These RA algorithms generally comprise questions about a person's social, economic, demographic, behavioural and clinical background. The answers can be scored or scaled to indicate a higher probability of current STD/HIV infection and thus the need for testing or treatment. RA can be done alone, or can be combined with clinical examination where laboratory diagnosis is not feasible. But how accurate are RA tools? We reviewed six studies at African sites that examined the use of RA combined with a clinical algorithm for STD screening of clinic attendees. Researchers reported the standard measures of test accuracy: sensitivity (per cent of infected women found positive by the tool); specificity (per cent of uninfected women found negative by the tool), and positive predictive value (per cent of women found positive by the tool who are truly infected). Most algorithms had low sensitivity (< 50%) and low positive predictive value (< 20%). The addition of clinical examination, and of leucocyte esterase (LE) testing, improved accuracy. But algorithms that performed well in one site were less accurate at other sites. For illustrative purposes, we considered a hypothetical population of 1,000 women attending a FP clinic. With a prevalence of cervical gonorrhoea of five per cent, algorithm sensitivity of 40%, and algorithm specificity of 75%, the positive predictive value of the RA tool would be 7.7%. In other words, 20 (7.7%) women diagnosed as infected would actually have gonorrhoea, and 237 false positive women would be treated needlessly, with the attendant costs and risks. However, with a disease prevalence above 50%, as might prevail for vaginal infections, and the same sensitivity and specificity figures, the positive predictive value may reach 70-80%. Risk assessment appears to be a relatively easy way to integrate STD management into family planning and MCH practice. However, there is no evidence to show its effectiveness for screening low-risk populations. We urge improvement of these tools, including information on the behaviour of male partners, and addition of simple screening tests such as leucocyte esterase dipsticks.