Clinical predictors for death in HIV-positive and HIV-negative tuberculosis patients in Guinea-Bissau

To assess easily monitored predictors for tuberculosis mortality. Risk factors for tuberculosis mortality were assessed during the 8-month treatment in 440 men and 269 women diagnosed with confirmed or presumed intrathoracic tuberculosis included prospectively in Guinea-Bissau from May 1996 to April 2001. A civil war occurred in the study area from June 1998 to May 1999. 12% were HIV-1 positive, 16% HIV-2 positive and 7% were HIV dually infected. Case fatality rates for HIV positive were higher during (35% [22/63]) and after the war (29% [27/92]) compared to before the war (17% [15/88]). The war did not have an effect on the case fatality rate in HIV negative (10% [13/135] before the war). HIV-1-infected patients had higher mortality than HIV-2 infected, mortality rate ratio (MRR) = 2.28 (95% confidence interval 1.17-4.46). Men had higher mortality than women but only among the HIV negative (MRR = 2.09 [0.95-4.59]). Hence, the negative impact of HIV infection on mortality was stronger in women (MRR = 6.51 [2.98-14.2]) than in men (MRR = 2.64 [1.67-4.17]) (test of homogeneity, p = 0.051). Anergy to tuberculin was associated with death in HIV positive (MRR = 2.77 [1.38-5.54]) but not in HIV negative (MRR = 1.14 [0.52-2.53]). Signs of immune deficiency, such as oral candida infection or leukoplakia (MRR = 4.25 [1.92-9.44]) and diarrhea (MRR = 2.15 [1.29-3.58] was associated with mortality in HIV positive. Tendencies were similar among HIV negative. HIV-positive relapse cases were at increased risk of dying (MRR = 2.42 [1.10-5.34]). Malnutrition, measured through mid-upper arm circumference (MUAC), increased the risk of death. Easily monitored predictors for mortality in tuberculosis patients include clinical signs of immune deficiency and low MUAC.