ABSTRACT
BACKGROUND: The outcome of anti-tuberculosis treatment varies according to patient factors. OBJECTIVE: To retrospectively identify risks related to the extension of time to negative sputum culture (Tn) and to determine their clinical significance. DESIGN: Patients with bacilli susceptible to isoniazid and rifampicin who received initial standard treatment without cessation were recruited into the study. A total of 630 consecutive in-patients were included in the risk development analysis (development cohort) and another 611 consecutive in-patients in the risk validation analysis (validation cohort). RESULTS: Univariate analysis showed that Tn was related to sex, body mass index (BMI), white blood cell count (WBC), serum albumin, fasting blood sugar, haemoglobin A1c, C-reactive protein and total cholesterol levels and sputum smear positivity (SSP). Multivariate analysis showed that BMI, WBC and SSP were significant risk factors related to extended Tn. Optimal cut-offs of BMI and WBC for predicting good (Tn < 46 days) and poor responders (Tn ⩾ 46 days) according to each risk were determined by receiver operating characteristics analysis. Risks were verified with the validation cohort. Tn increased according to the number of risks; the median Tn for patients with three risks was 21 days longer than that of patients with none. CONCLUSION: The nutritional state of a TB patient can be used to predict Tn.