HIV Infection and Legionnaires’ Disease in New York City, 2006–2015

Background: Legionnaires’ disease (LD) is caused by inhalation or aspiration of water contaminated with Legionella bacteria. Traditional LD risk factors include older age, smoking, lung disease, residing in high poverty areas, and immune system compromise. The incidence of reported LD in New York City (NYC) has increased over the past 15 years. During an outbreak of LD in the South Bronx (SB) in 2015, we observed a high rate of comorbidities, particularly HIV. We assessed associations between HIV status and LD diagnosis and death.

Methods: Confirmed and suspected cases of community-acquired LD among NYC residents diagnosed during 2006–2015 were matched to the HIV surveillance registry. Logistic regression was used to assess the association between LD diagnosis and HIV status in the NYC population. Interaction with HIV status and SB residency was used to assess modification by outbreak zone residence. Multivariable logistic regression was used to measure the association between LD death and HIV status. HIV prevalence among 2015 outbreak cases and community-acquired cases residing in the outbreak zone during 2006-2014 was compared with a chi-square test.

Results: During 2006–2015, 2041 community-acquired cases of LD were diagnosed in NYC and 233 (11%) had diagnosed HIV infection. SB residency modified the association between HIV status and LD (p=0.04): people living with HIV/AIDS (PLWHA) in the SB had 11.9 (95% CI 8.7-16.4) times the odds and, in the rest of NYC, 8.3 (95% CI 7.2-9.7) times the odds of LD compared with HIV-uninfected persons. Among LD cases, HIV infection was not associated with death (p=0.9), controlling for age, sex, and census tract-based poverty level. LD cases in the SB outbreak had a smaller percentage of PLWHA than historically (18% vs. 31%, p=0.02).

Conclusion: HIV infection was associated with LD diagnosis, but LD case fatality was not higher than among HIV-uninfected cases. The lower proportion of PLWHA in SB outbreak cases compared with historical cases suggests HIV was not a major driver of the outbreak. Lack of detailed population-level data on comorbidities precluded further confounder adjustment. LD is likely underdiagnosed, and our estimates reflect risk for diagnosed disease. Providers should consider testing LD cases for HIV.