Spatial mapping of acute diarrheal disease using GIS and estimation of relative risk using empirical Bayes approach

Abstract

Background/objectivesDiarrheal disease is one among the top five causes of death in low- and middle-income countries. It is the second leading cause of death in children under five years of age. Diarrheal disease contributes to the mortality of nearly 1.5 million children and globally there are about two billion cases of diarrheal diseases every year. In the present study, we studied the spatio-temporal pattern of acute diarrheal disease (ADD) ward-wise and to estimate and compare two widely used Bayesian models in the study of measuring relative risk of ADD in Chennai Corporation, Tamilnadu, India. Materials and methodsData on ADD were obtained from Communicable disease hospital, Chennai, Tamilnadu, India from 2009–11. Geographical Information System (GIS) technique was used to map ADD data ward- wise and relative risk was estimated using empirical Bayes approach using Poisson gamma and Poisson log normal models. ResultsOver a period of three years from 2009–11, nearly 7661 cases of ADD were reported in Chennai Corporation. The cumulative incidence rate of diarrhea was 142.6 cases per 100,000 population ranging from minimum of 0 to maximum of 1699.7 cases per year. Males had higher average incidence 147.4 cases than females with 137.7 cases per 100,000 population per year. Also, the cumulative incidence was higher in the age group of 0–4 years (306.8 cases) than that in any other category. Higher incidence was observed during the months of Apr–Jun (55.2 cases) than that in any other seasons. Choropleth map indicates that higher incidence of ADD was more prevailed in northern part of Chennai near coastal area which includes the wards from Tondiarpet, Basin bridge, and Pulianthope zones. The posterior relative risk estimate obtained using empirical Bayes approach identified 23, 30, and 19 wards having relative risk significantly greater than 1 for years 2009, 2010, and 2011, respectively. Fitting standardized morbidity ratio (SMR) and the other two models showed that, consistently Tondiarpet ward had the highest relative risk of ADD in all the three years (relative risk (95% credible interval) based on SMR, Poisson gamma and Poisson log normal models were 13.3 (11.6, 15.2), 13.0 (11.3, 14.8), 13.3 (11.5, 15.1) in 2009 and 8.0 (6.9, 9.2), 7.8 (6.7, 8.9), 8.0 (6.9, 9.1) in 2010 and 19.6 (17.0, 22.3), 19.0 (16.5, 21.7), 19.5 (16.9, 22.1) in 2011, respectively. ConclusionGIS mapping of ADD during the years from 2009–11 identified wards having higher incidence near northern part of Chennai near coastal area. Thus estimation of relative risk using empirical Bayes approach and mapping of disease would help us to estimate accurate incidence despite small observations and to take precautionary measure to prevent and control the spread of the disease.