Published in:
Open Access 01-12-2010 | Research article
Bayesian mapping of pulmonary tuberculosis in Antananarivo, Madagascar
Published in: BMC Infectious Diseases | Issue 1/2010
Login to get accessAbstract
Background
Tuberculosis (TB), an infectious disease caused by the Mycobacterium tuberculosis is endemic in Madagascar. The capital, Antananarivo is the most seriously affected area. TB had a non-random spatial distribution in this setting, with clustering in the poorer areas. The aim of this study was to explore this pattern further by a Bayesian approach, and to measure the associations between the spatial variation of TB risk and national control program indicators for all neighbourhoods.
Methods
Combination of a Bayesian approach and a generalized linear mixed model (GLMM) was developed to produce smooth risk maps of TB and to model relationships between TB new cases and national TB control program indicators. The TB new cases were collected from records of the 16 Tuberculosis Diagnostic and Treatment Centres (DTC) of the city from 2004 to 2006. And five TB indicators were considered in the analysis: number of cases undergoing retreatment, number of patients with treatment failure and those suffering relapse after the completion of treatment, number of households with more than one case, number of patients lost to follow-up, and proximity to a DTC.
Results
In Antananarivo, 43.23% of the neighbourhoods had a standardized incidence ratio (SIR) above 1, of which 19.28% with a TB risk significantly higher than the average. Identified high TB risk areas were clustered and the distribution of TB was found to be associated mainly with the number of patients lost to follow-up (SIR: 1.10, CI 95%: 1.02-1.19) and the number of households with more than one case (SIR: 1.13, CI 95%: 1.03-1.24).
Conclusion
The spatial pattern of TB in Antananarivo and the contribution of national control program indicators to this pattern highlight the importance of the data recorded in the TB registry and the use of spatial approaches for assessing the epidemiological situation for TB. Including these variables into the model increases the reproducibility, as these data are already available for individual DTCs. These findings may also be useful for guiding decisions related to disease control strategies.