BACKGROUND: Since the first human cases of West Nile virus (WNV) infection in Chicago were diagnosed in 2002, vector control initiatives have been implemented and annual case counts have dramatically decreased. However, cases continue to occur each season. Identifying factors that predict residential locations of cases would be useful for prevention efforts. In this exploratory analysis, we evaluate whether demographic, socioeconomic, or built-environment factors are predictive of where cases reside in a large urban center.
METHODS: From 2012 to 2016, the Chicago Department of Public Health (CDPH) received 173 reports of human cases of WNV in Chicago through the Illinois National Electronic Disease Surveillance System (I-NEDSS). Address at diagnosis for each case was geocoded to obtain the cumulative number of cases within each census tract. Descriptive information about each census tract was acquired from the American Community Survey (ACS). Potential predictors of WNV case rates were selected, including age, sex, race/ethnicity, income, education, insurance status, nationality, and housing factors. A multivariable Poisson regression model was developed to identify significant predictors and compare expected case rates based on census tract attributes.
RESULTS: Age, race/ethnicity, and housing factors were found to be significantly associated with case rates at the census tract level; income, education, insurance status and nationality were not. According to the Poisson regression model, expected rates of reported cases were greater for census tracts with a higher median age (relative risk [RR], 1.48, 95% confidence interval [CI]: 1.14-1.93, for each 10 year increase) and a higher percentage of Non-Hispanic (NH) white persons (RR, 1.17, 95% CI: 1.10-1.24, for each 10 percentage point increase). Each 10 percentage point increase in proportion of housing structures with 20 or more units in a census tract was associated with a twenty-four percent decrease in case rate (RR, 0.76, 95% CI: 0.69-0.84). Additionally, each 10 percentage point increase in proportion of housing units vacant was associated with a thirty-four percent increase in case rate (RR, 1.34, 95% CI: 1.08-1.67), after adjustment for other factors.
CONCLUSIONS: Vacant housing and housing structures with fewer units are predictive of higher reported WNV case rates within census tracts. Further research is necessary to determine whether large structures in urban areas influence weather patterns (e.g., wind) or limit mosquito breeding locations and whether the presence of vacant housing creates more ideal conditions for mosquito proliferation. These findings will help to inform vector control interventions and public awareness campaigns.