Using Prospective Illness Surveillance Data to Quantify, Characterize, and Mitigate the Risk of Childcare-Associated Respiratory Disease

Abstract

Young children experience high rates of morbidity and mortality associated with acute respiratory illness (ARI) and are noted contributors of ARI risk to their households as well. It is also well established that ARI risk is higher in children who attend out-of-home childcare, compared to children cared for at home. Using prospective illness surveillance data collected regionally in southeast Michigan, the four Aims of this Dissertation add pertinent information to better quantify, characterize, and mitigate childcare-associated ARIs in both children and their families.

Aim 1 investigates the association between household childcare use and illness risk for family members. Adjusted mixed-effects Poisson regression models reveal no association between whether or not a young child concurrently attends childcare and the incidence rate of reported ARI in family members. Likewise, we find no evidence for any relevant effect modification of the association by household or individual characteristics. Contrary to our original hypothesis, out-of-home childcare does not appear to be linked to higher illness incidence in associated families; several reasons are offered to explain these null findings.

In Aim 2, we shift our focus to the children themselves. Previous research has shown that childcare attendees are ill more often than their counterparts cared for at home—we were interested in whether childcare-associated ARI was also characteristically different in terms of illness etiology and severity. In this analysis, we compare molecular test results from ARI episodes of children who do and do not attend childcare. Adjusted mixed-effects logistic regression models reveal higher odds of both adenovirus and human metapneumovirus infection in childcare attendees, yet lower odds of rhinovirus infection; we found no evidence for a difference in illness severity. Furthermore, we statistically show that the pool of viruses causing childcare illness is significantly more diverse than that of homecare illnesses. Our results support the idea that childcare children are sick more than homecare children because they are exposed to a wider array of viruses early in life.

Aim 3 steps back to assess surveillance data in the context of the larger community. Three prospective illness surveillance networks of southeast Michigan operate in distinct population subgroups; we wanted to determine how different surveillance settings influenced observed epidemic patterns. Using influenza as our outcome of interest, we find that a household, ambulatory clinic, and hospital network all capture similar epidemic trends across six surveillance seasons. Annually, all three systems report comparable distributions of influenza A and B and record similar timing in epidemic activity start and peak activity. Broad agreement between network epidemic curves should be reassuring for local public health departments that may only rely on one system of active influenza surveillance.

Aim 4 focuses on the role of illness surveillance in childcare programs. Epidemiologic data serve as an important tool in timely outbreak intervention; we wanted to understand how illness surveillance in childcare could be more effectively leveraged to mitigate childcare disease outbreaks. Using results from a series of semi-structured focus groups among childcare providers, we find that providers are eager to use epidemiologic data in their programs to corroborate their subjective experiences and policy decisions. Our findings offer a framework by which surveillance data could be collected and packaged in a more beneficial way for providers and parents, and—ideally—help to prevent childcare associated illness outbreaks.