Analysis of population-level determinants of legionellosis: spatial and geovisual methods for enhancing classification of high-risk areas

Abstract

Abstract Background Although the incidence of legionellosis throughout North America and Europe continues to increase, public health investigations have not been able to identify a common exposure in most cases. Over 80% of cases are sporadic with no known source. To better understand the role of the macro-environment in legionellosis risk, a retrospective ecological study assessed associations between population-level measures of demographic, socioeconomic, and environmental factors and high-risk areas. Methods Geographic variability and clustering of legionellosis was explored in our study setting using the following methods: unadjusted and standardized incidence rate and SaTScan™ cluster detection methods using default scanning window of 1 and 50% as well as a reliability score methodology. Methods for classification of “high-risk” census tracts (areas roughly equivalent to a neighborhood with average population of 4000) for each of the spatial methods are presented. Univariate and multivariate logistic regression analyses were used to estimate associations with sociodemographic factors: population ≥ 65 years of age, non-white race, Hispanic ethnicity, poverty, less than or some high school education; housing factors: housing vacant, renter-occupied, and built pre-1950 and pre-1970; and whether drinking water is groundwater or surface water source. Results Census tracts with high percentages of poverty, Hispanic population, and non-white population were more likely to be classified as high-risk for legionellosis versus a low-risk census tract. Vacant housing, renter-occupied housing, and homes built pre-1970 were also important positively associated risk factors. Drinking water source was not found to be associated with legionellosis incidence. Discussion Census tract level demographic, socioeconomic, and environmental characteristics are important risk factors of legionellosis and add to our understanding of the macro-environment for legionellosis occurrence. Our findings can be used by public health professionals to target disease prevention communication to vulnerable populations. Future studies are needed to explore the exact mechanisms by which these risk factors may influence legionellosis clustering.

Keywords

• Legionellosis
• Legionnaires’ disease
• Cluster analysis
• GIS
• SaTScan™
• Ecological study