Decentralisation of directly observed treatment in a large African city: evaluation of the experience of Djibouti.

SETTING: Decentralisation of directly observed treatment (DOT) for tuberculosis patients in three public centres in Djibouti city from April 2000. OBJECTIVES: To evaluate decentralisation based on the success rate by site of treatment and according to certain critical variables. METHODS: Comparative evaluation of the success rate of smear-positive patients followed in all treatment centres from 1 May 2000 to 31 March 2001. RESULTS: The success rate was 58% for the main centre, Centre Paul Faure, and 81% for all the peripheral centres together (P < 10.6). It was 80% for the three new centres and 85% for the established centres (P > 0.05). Age under 20 years, female sex and treatment centre were factors linked to success. After stratification, sex was shown to be the confounding variable. Multivariate analysis shows that non-Djibouti nationality is related to treatment success (P = 0.02). In the groups of established and new centres, there is an inverse linear relationship between success rate and workload, with greater capacity in the established centres. CONCLUSIONS: Urban decentralisation of DOT increases the chances of treatment success among smear-positive patients. Another centre for supervised treatment needs to be created in one of the public health centres in Djibouti city.