Control of tuberculosis in Uganda: a tale of two districts.

SETTING: Rakai and Mbarara districts, south-western Uganda. OBJECTIVE: To compare compliance and other treatment outcomes with tuberculosis (TB) treatment for Rakai and Mbarara districts, and to elucidate factors associated with the disparity. DESIGN: A retrospective cohort analysis of tuberculosis treatment outcomes for the period 1992-1996. Predictors for compliance were examined for 1995 using univariate and multivariate methods. RESULTS: Of the surviving patients, 57% finished treatment in Mbarara compared to 92% in Rakai. Enhanced compliance on univariate analysis was associated with being smear positive; previous TB treatment; use of short-course chemotherapy; return for repeat smear; no change of health unit; no change of district; registration in Rakai; and not being hospitalised. On multivariate analysis, being smear positive, previous TB treatment, hospitalisation and registration site were not significant. The highest risk for default was unit change after the intensive phase of treatment, with an adjusted odds ratio of 17.53. The highest differences in the two districts were for the hospitalisation and unit change rates, with corresponding odds ratios of 52 and 0.06. CONCLUSIONS: Initial hospitalisation of TB patients is not necessary for subsequent completion of treatment. Use of one health unit for both the intensive and continuation phase of treatment may improve compliance.