Estimating the True Burden of Legionnaires' Disease

Over the past decade, the reported incidence of Legionnaires' disease (LD) in the northeastern United States has increased, reaching 1-3 cases per 100,000 population. There is reason to suspect that this is an underestimate of the true burden, since LD cases may be underdiagnosed. In this analysis of pneumonia and influenza (P&I) hospitalizations, we estimated the percentages of cases due to Legionella, influenza, and respiratory syncytial virus (RSV) by age group. We fitted mixed-effects models to estimate attributable percents using weekly time series data on P&I hospitalizations in Connecticut from 2000 to 2014. Model-fitted values were used to calculate estimates of numbers of P&I hospitalizations attributable to Legionella (and influenza and RSV) by age group, season, and year. Our models estimated that 1.9%, 8.8%, and 5.1% of total (all-ages) inpatient P&I hospitalizations could be attributed to Legionella, influenza, and RSV, respectively. Only 10.6% of total predicted LD cases had been clinically diagnosed as LD during the study period. The observed incidence rate of 1.2 cases per 100,000 population was substantially lower than our estimated rate of 11.6 cases per 100,000 population. Our estimates of numbers of P&I hospitalizations attributable to Legionella are comparable to those provided by etiological studies of community-acquired pneumonia and emphasize the potential for underdiagnosis of LD in clinical settings.