Objectives: To determine which clinical and immunological features of patients with symptomatic HIV-1 and HIV-2 infection best predict survival in The Gambia.

Methods: All patients presenting to two hospitals in The Gambia between January 1987 and June 1990 with symptoms or signs suggesting chronic HIV infection were tested for HIV-1 and HIV-2 antibodies. Eighteen HIV-1 and 31 HIV-2-infected patients were recruited to the study, investigated intensively on admission and followed up until the end of 1990. Presenting clinical features, such as Karnofsky score, diagnosis of AIDS according to World Health Organization Bangui or Centers for Disease Control criteria and number of associated infections, together with five immunological measurements, as well as type of HIV infection, were related to length of survival using proportional hazard models fitted to Kaplan-Meier plots of survival times.

Results: Karnofsky score and diagnosis of AIDS were the best clinical predictors of survival. Type of HIV infection or number of associated infections did not predict outcome. The most powerful laboratory predictors were loge serum neopterin level, CD4 cell count and loge serum [beta] 2-microglobulin ([beta] 2M) level. The estimated median survival times (90% confidence interval) of the HIV-1 and HIV-2-infected patients were six (4-11) and 13 (9-20) months, respectively. These survival times do not differ significantly.

Conclusions: The Karnofsky score and measurements of serum neopterin or [beta] 2M, which are easier and cheaper to perform than CD4 counts, may prove to be useful guides to prognosis for HIV infection in Africa.