BACKGROUND: During 2000–2014, Legionnaires’ disease (LD) cases increased 358% nationwide and 378% in New York City (NYC). In NYC, LD incidence is associated with high poverty areas, and >95% of cases have no known exposure source. In July 2015, NYC had the second-largest outbreak of community-acquired LD in US history. A cooling tower (CT) was confirmed as the source, resulting in a local law mandating CT disinfection and establishing a citywide CT registry. We examined whether LD rates were associated with neighborhood CT density.
METHODS: Confirmed LD cases among NYC residents diagnosed during January 2013–October 2016 were obtained from passive surveillance. “Definite” hospital-acquired cases and cases diagnosed during the 2015 outbreak were excluded. LD cases and CTs registered as of December 2016 were geocoded and assigned to neighborhood tabulation areas (NTAs; n=189 populated NTAs in NYC). Incidence rates were calculated using denominators from the American Community Survey 2010–2014 and age-adjusted to the US 2000 standard population. NTAs were categorized by CT density using natural breaks (low, <14 CTs/square mile (SqMi); medium, 14–48 CTs/SqMi; and high, ≥49 CTs/SqMi) and by poverty (low, <10% of residents living below the federal poverty level; medium, 10–<20%; high, 20–<30%; and very high, ≥30%). NTA-level associations between age-adjusted LD counts and CT frequency and poverty were assessed using negative binomial regression, with the log of NTA population as an offset term.
RESULTS: During the study period, 952 LD cases were diagnosed, and 6,087 CTs were registered. Mean CT density across NTAs decreased with increasing area-based poverty from 82.9 to 45.8 to 19.4 to 10.6 CT/SqMi in low, medium, high and very high-poverty NTAs, respectively. In bivariate analysis, LD rates were associated with increasing CT density (medium vs. low: relative risk [RR] =1.46, 95% confidence interval [CI]: 1.09–1.96; high vs. low: RR=1.45, 95% CI 1.03-2.03). In multivariable analysis, LD rates remained associated with increasing CT density (medium vs. low: adjusted RR [aRR]=1.99, 95% CI: 1.47-2.71; high vs. low: aRR=1.99, 95% CI: 1.46–2.71) and increasing area-based poverty (medium vs. low: aRR=1.73, 95% CI: 1.24–2.44; high vs. low: aRR=1.91, 95% CI: 1.33–2.74; and very high vs. low: aRR=3.94, 95% CI: 2.70–5.76).
CONCLUSIONS: LD incidence was associated with both increasing CT density and neighborhood poverty. To further explain variation in LD incidence across NYC neighborhoods, future analyses should examine CT characteristics and patient comorbidities and account for spatial autocorrelation.