AIDS surveillance in Africa

Abstract

Surveillance of the AIDS pandemic in Africa has always posed formidable problems for epidemiologists. Diagnostic accuracy-according to the case definitions for AIDS used in industrialised countries-is impossible to achieve in all but a few places with the right diagnostic facilities. Responding to the urgent need for surveillance, the World Health Organisation drew up a clinical case definition (the WHO/Bangui definition), which depended on clinical criteria without the need for serological verification.'2 Judged by its use, the WHO/Bangui definition has been successful-52 African countries have reported cases ofAIDS using mainly this definition.3 Some countries have modified it to fit local circumstances, removing a defining symptom here, adding the need for an extra sign there, and many now accept or encourage a positive result of an HIV test as supportive evidence. (At least one, COte d'Ivoire, requires such a result.2) Inevitably the definition has its limitations, and two papers in this week's journal discuss these at length (p 11852, p 11894). Because of limited laboratory facilities published evaluations of the WHO/Bangui definition have been mainly restricted to groups of sick patients using HIV positivity as the reference standard. The definition's sensitivity and specificity have been calculated as being between 60% and 90%2 5- useful for purposes of surveillance, but leaving uncertainty over whether this surveillance tool is intended to monitor trends in cases of AIDS or HIV infection. Other problems exist with the WHO/Bangui definition. Because many doctors lack diagnostic facilities they use the definition for diagnosis. The title "clinical case definition" encourages this confusion. The misuse is disturbing as the probability that a patient who fulfils the WHO/Bangui definition tests positive for HIV may fall well below 50% when seroprevalence is low.5 Another problem of using the definition is the delayed and incomplete picture that it gives of the spread of infection.6 Far preferable for surveillance of infection is the unlinked anonymous testing for HIV of sentinel groups attending health services67 (such as pregnant women and people with sexually transmitted diseases), which has now begun in several African countries89 using the same methods as in industrialised countries.'° Where does this leave the WHO/Bangui definition? De Cock and colleagues rehearse the overwhelming case for AIDS reporting to continue and suggest a thoughtful redesign of the definition, which includes the requirement for a positive HIV test result.3 Insisting on positive test results in all circumstances, however, is impractical: HIV tests are already limited and are lioely to become more so as AIDS funding to Africa inevitably falls. As a provisional solution to the problem of surveillance the WHO/Bangui definition has been useful, but the time has come for its reappraisal.