Early treatment response evaluated by a clinical scoring system correlates with the prognosis of pulmonary tuberculosis patients in Ethiopia: A prospective follow-up study

Background: In resource-limited settings the monitoring of tuberculosis (TB) patients is challenging, and early identification of TB patients with a high mortality risk is important. The aim of this study was to investigate prospectively whether early changes in a clinical scoring system (TB score) can predict treatment outcome in Ethiopian patients with pulmonary tuberculosis. Method: TB patients (n = 250) and blood donors (n = 82) were recruited prospectively at Gondar University Hospital, Ethiopia. Clinical scoring was performed using an interview-based questionnaire and clinical examination. Results: Among TB patients (53.6% of whom were HIV co-infected) the median TB score declined from week 0 to week 2 (8 (interquartile range (IQR) 6-9) vs 4 (IQR 2-6)) and dropped to a low level at week 8, which was still significantly higher than that found in blood donors (2 (IQR 1-4) vs 0 (IQR 0-1), p < 0.0001). Patients who died had a significantly higher TB score at week 0, week 2, and week 8 than survivors. Mortality was associated with a failure to achieve a decrease greater than 25% in the TB score at 2 weeks. Baseline CD4 + cell counts (< 200 cells/mm(3)) were associated with mortality but not with initial TB score results. Conclusions: The TB score was increased during the first 2 months of treatment among patients who died. Failure to achieve a greater than 25% decrease in TB score after 2 weeks of treatment was associated with increased mortality. Repeated clinical scoring during the intensive phase of TB treatment could be useful to identify high-risk patients.