Statistical Inference in Epidemiology - Analysis of Gastro-intestinal Disease in Day-nurseries

Abstract

Summary : This study aimed to give a picture of gastro-enteritis among children in the community as opposed to in the hospital, and also to study the main agents of infection and the amount of asymptomatic carriage of disease. A one-year follow-up study was conducted in two different day nurseries in a socially deprived area of Greater Manchester. A total of 167 children were enrolled in the study. Faecal specimens were collected fortnightly from the children and were tested for certain agents of infection. Three different types of microorganisms and 21 agents were isolated. The presence of any symptoms of illness was recorded. Demographic information on the children and their families was also acquired. A set of simple statistical analyses were conducted on the data, such as calculating means and standard deviations of the number of times each agent was isolated per child, or an episode of a particular symptom was reported Correlation coefficients between the number of times each agent was isolated per child and the occurrence of symptoms were also calculated. Further comprehensive statistical analyses such as fitting logistic regression models and survival analysis were also conducted. This aimed at estimating the attributable fraction of gastro-intestinal disease due to different agents, the force of infection, and the basic reproductive rate (R0), where feasible. This study was useful in two ways : in gaining an understanding of the causes of childhood diarrhea in the community; and Mathematically, some useful techniques for drawing inferences from "messy" epidemiological data have been developed.