Development and validation of a tuberculosis prognostic score for smear-positive in-patients in Japan.

BACKGROUND: No scoring system has ever been used to estimate the prognosis of individual tuberculosis (TB) patients. OBJECTIVE: To develop and validate a tuberculosis prognostic score. METHODS: This retrospective cohort study conducted in Japan comprised the development (n = 179; mean age 65.9 ± 18.8 years) and validation (n = 244; mean age 64.3 ± 20.1 years) of a tuberculosis prognostic score among patients with newly diagnosed smear-positive non-multidrug-resistant pulmonary tuberculosis without human immunodeficiency virus infection. The score (raw score) was defined by modifying a logistic regression formula using known risk factors as independent variables and in-patient death as a dependent variable. RESULTS: The raw score was calculated as follows: age (years) + (oxygen requirement, 10 points) - 20 × albumin (g/dl) + (activity of daily living: independent, 0 point; semi-dependent, 5 points; totally dependent, 10 points). The raw scores were grouped into risk groups 1 (raw score < -30) to 5 (raw score ≥ 60) using 30-point intervals. Every increase in risk group was equivalent to a 7.3-fold increase in the odds ratio for in-hospital death (P < 0.001). The area under the receiver operating characteristics curve by risk group for in-patient death was 0.875 (P < 0.001). CONCLUSIONS: In this study we were able to develop and validate a tuberculosis prognostic score.