Development of a case fatality prognostic score for HIV-associated histoplasmosis

Abstract

ObjectivesThe burden of histoplasmosis is as great as that of tuberculosis in Latin America and the attributable mortality is even higher. A better assessment of severity could help reduce mortality. MethodsFrom the French Guiana HIV-histoplasmosis database, we attempted to identify factors associated with 30-day death after antifungal drug initiation and constructed a prognostic score. We evaluated its discrimination performance using several resampling methods. ResultsOf the 415 patients included, 56 (13.5%) died within 30 days of treatment. The fatality-associated factors were performance status ≥3, altered mental status, dyspnea, C-reactive protein ≥75 mg/l, hemoglobin <9 g/dl and/or a platelet <100000/ml, and an interstitial lung pattern on chest X-ray. We constructed a 12-point prognostic score. A threshold ≥5 classified patients as alive or dead at 30 days with a sensitivity of 84%, a specificity of 81%, a positive predicted value of 40%, and a negative predicted value of 97%. The area under the curve of the receiver operating characteristic curves from the different resamples were stable between 0.88 and 0.93. ConclusionThe histoplasmosis case fatality score, which is easy and inexpensive to perform, is a good tool for assessing severity and helping in the choice of induction therapy. An external validation remains necessary to generalize these results.