SETTING: Djibouti, 1994 and 2001. OBJECTIVE: To estimate the prevalence of tuberculosis (TB) and average annual risk of TB infection (ARTI) and trends, and to test a new method for calculations. METHODS: Tuberculin surveys among schoolchildren and sputum smear-positive TB patients. Prevalence of infection was calculated using cut-off points, the mirror image technique, mixture analysis, and a new method based on the operating characteristics of the tuberculin test. Test sensitivity was derived from tuberculin reactions among TB patients and test specificity from a comparison of reaction size distributions among children with and without a BCG scar. RESULTS: The ARTI was estimated to lie between 2.6% and 3.1%, with no significant changes between 1994 and 2001. The close match of the distributions between children tested in 1994 and patients justifies the utilisation of the latter to determine test sensitivity. This new method gave very consistent estimates of prevalence of infection for any induration for values between 15 and 20 mm. Specificity was successfully determined for 1994, but not for 2001. Mixture analysis confirmed the estimates obtained with the new method. CONCLUSION: Djibouti has a high ARTI, and no apparent change over the observation time was found. Using operating test characteristics to estimate prevalence of infection looks promising.
SETTING: Djibouti, 1994 and 2001. OBJECTIVE: To estimate the prevalence of tuberculosis (TB) and average annual risk of TB infection (ARTI) and trends, and to test a new method for calculations. METHODS: Tuberculin surveys among schoolchildren and sputum smear-positive TB patients. Prevalence of infection was calculated using cut-off points, the mirror image technique, mixture analysis, and a new method based on the operating characteristics of the tuberculin test. Test sensitivity was derived from tuberculin reactions among TB patients and test specificity from a comparison of reaction size distributions among children with and without a BCG scar. RESULTS: The ARTI was estimated to lie between 2.6% and 3.1%, with no significant changes between 1994 and 2001. The close match of the distributions between children tested in 1994 and patients justifies the utilisation of the latter to determine test sensitivity. This new method gave very consistent estimates of prevalence of infection for any induration for values between 15 and 20 mm. Specificity was successfully determined for 1994, but not for 2001. Mixture analysis confirmed the estimates obtained with the new method. CONCLUSION: Djibouti has a high ARTI, and no apparent change over the observation time was found. Using operating test characteristics to estimate prevalence of infection looks promising.