DOT or not? Direct observation of anti-tuberculosis treatment and patient outcomes, Kerala State, India.

SETTING The Pathanamthittha District of Kerala State, India, where the directly observed treatment, short-course (DOTS) programme was started in October 1994. OBJECTIVE To determine the frequency with which direct observation actually occurred within a district-level DOTS programme, and the association of treatment observation with treatment outcome under programme conditions. DESIGN This retrospective study included 200 consecutive, newly-detected, smear-positive patients registered under the project between February 1995 and February 1996 at the District Tuberculosis Centre, as well as health workers responsible for providing directly observed treatment (DOT) who were separately and confidentially interviewed. Treatment outcomes were identified from results of sputum smear examinations for acid-fast bacilli. RESULTS Although all patients were recorded as having received DOT, more than a quarter of patients (26.5%) did not actually receive it. The 53 patients who were not directly observed were much more likely to have treatment failure or relapse, as compared to those who had received DOT (45% vs 3%, relative risk 16.6, 95% confidence intervals 6-46, P < 0.001). Women were somewhat less likely than men (61% vs 76%, P = 0.06) to receive DOT. Non-receivers of DOT accounted for 86% (24/28) of treatment failures or relapses. CONCLUSION Patients treated without direct observation have a substantially higher risk of adverse outcome than those treated under direct observation. To be maximally effective, the DOTS programme must be both confidential and convenient.