Examining Three Potential Drivers of the Increase in Legionnaires’ Disease Incidence: 1) Individual Susceptibility to Infection, 2) Environmental Influences on Infection, and 3) Population-level Factors of Reported Incidence
Date of Award
Fall 2023
Document Type
Dissertation
Degree Name
Doctor of Philosophy (PhD)
Department
Public Health
First Advisor
Weinberger, Daniel
Abstract
Background: Legionnaires’ disease (LD) is a severe form of pneumonia and incidence of the disease has increased dramatically across the US and many European countries since the early 2000s. In the US, incidence has doubled during this time period, reaching 3 cases per 100,000 in 2018. The primary agent of infection, Legionella pneumophila, grows in natural and man-made water systems and is transmitted via inhalation of contaminated aerosols. Multiple hypotheses exist for the increase in LD incidence, including 1) susceptibility to infection, 2) greater concentration and/or exposure to Legionella in the environment, and 3) secular trends population and diagnostic uptake that affect reported incidence. We chose to address these three hypotheses through five studies using data from clinical trials, Connecticut department of public health and Danish national surveillance for Legionnaires’ disease. Methods & Results: In Chapter 1A, we provide a comprehensive study of Legionnaires’ disease reinfection. We characterize the recurrence intervals of all episodes of Legionella reinfection since 2000 in Denmark. Of the 21 Danish residents with recurrent Legionella infection, four had a second infection due to a different sequence type or serotype of Legionella pneumophila. The median time between first and second infection was 637 days. In Chapter 1B, we compared individuals randomized to treated and untreated arms within TNF inhibitor clinical trials. We assessed risk according to intent-to-treat assumptions, and through nested case-control and time-to-event analyses to adjust for the high rate of treatment switching. We found that all cases of LD occurred among patients randomized to the treatment group. The treatment arm also had an increased risk of pneumonia, injection site reaction, gastrointestinal infection (among participants treated with Golimumab specifically). In Chapter 2, we used a Bayesian distributed lag model to provide precise estimates of the association between daily weather and rates of Legionnaires’ disease. The model incorporated the known LD incubation period distribution, as well as distributed lags of rainfall and relative humidity. We also conducted sensitivity analyses to ensure that the association between rainfall and disease was not spuriously linked to the incubation period distribution or rainfall after the expected date of exposure. We found that a one standard deviation increase in rainfall 3 days prior to estimated date of exposure increased LD risk by 15% (95%CI: 2, 28%). When rainfall occurred with high relative humidity (>81%) 2 days prior to exposure, risk increased 70% (95% CI 17, 131%). Then, in Chapter 3A, in order to evaluate whether the increase in LD in Denmark and the difference in incidence across provinces were due to its population structure, we estimated age- and sex-standardized incidence rates for Denmark’s provinces between 2015 and 2018. Provincial differences in incidence were more apparent after adjusting for age and sex. Furthermore, the distinction between 2015 and 2016 incidence and 2016 and 2017 incidence was more distinct and universally increasing across the provinces, indicating a nation-wide shift in incidence. To build on these findings, in Chapter 3B, we examined the relationship between Legionella diagnostic test ordering, pneumonia hospitalizations, and LD incidence over time in Denmark. We showed that the cases of LD per diagnostic test ordered remained relatively stable over time within each region despite a dramatic increase in diagnostic test ordering. The Capital Region had a 155% increase in total tests and Southern Denmark had a smaller increase of 55% over the eight years of study which outpaced trends in pneumonia hospitalizations. Conclusions: In total, our approach of assessing each hypothesis separately allowed us to pair traditional epidemiologic methods with higher level statistics, and match each hypothesis to the optimal dataset for the study question at hand. Healthcare providers and public health professionals should be aware of the possibility for LD recurrence if high-risk individuals return to the location likely responsible with their initial infection. Additionally, rare and opportunistic infectious adverse events should be considered more explicitly in clinical trial design, in addition to the seasonality of pathogen spread when known immunosuppressants under trial. In Chapter 2, we were able to contribute to the growing evidence for a public water mediated link between weather and LD risk. Finally, we explored hypotheses for the sharp increase in LD in Denmark and conclude that the most likely explanation is the adoption of Legionella PCR and large uptake in its use over the past decade.
Recommended Citation
Cassell, Kelsie, "Examining Three Potential Drivers of the Increase in Legionnaires’ Disease Incidence: 1) Individual Susceptibility to Infection, 2) Environmental Influences on Infection, and 3) Population-level Factors of Reported Incidence" (2023). Yale Graduate School of Arts and Sciences Dissertations. 1109.
https://elischolar.library.yale.edu/gsas_dissertations/1109
