RESUMO
SETTING: A tuberculosis (TB) referral hospital in the state of Bahia, north-eastern Brazil. OBJECTIVE: To evaluate the association between serum albumin at admission and in-hospital deaths due to TB. DESIGN: A prospective cohort study of 373 patients hospitalised with TB. Samples were systematically collected from all patients at hospital admission for measurement of serum albumin levels. In-hospital death was the study endpoint. Univariate analysis was used to evaluate the association between serum albumin at admission and mortality from TB, and multivariate logistic regression was used to adjust for potential confounding variables. RESULTS: The average age of the 373 patients was 41.1 years (+/-15.2); 77.2% (288/373) were male; 45 eventually died in hospital (in-hospital death rate 12.1%, 95%CI 9.0-15.7). Multivariate analysis showed that the presence of low serum albumin levels (< or =2.7 g/dl) was strongly and independently associated with in-hospital death due to TB (adjusted OR 3.38, 95%CI 1.51-7.59; P = 0.001), even after adjusting for potentially confounding variables identified in the univariate analysis (age, comorbidities, TB-human immunodeficiency virus [HIV] coinfection and history of previous default from treatment). CONCLUSION: Our results highlight the importance of measuring serum albumin levels at admission to hospital as an indicator of the prognosis for patients hospitalised with TB.