Statistical Procedures for Estimating the Community Probability of Illness in Family Studies: Rhinovirus and Influenza

Longini I M (Department of Epidemiology, School of Public Health, University of Michigan, Ann Arbor, Michigan 48109, USA), Monto A S and Koopman J S. Statistical procedures for estimating the community probability of illness in family studies: rhinovirus and influenza. International Journal of Epidemiology 1984, 13: 99–106. A statistical method is presented for determining acquisition rates of illness from community sources of infection and for distinguishing among viruses which are associated with different epidemic seasons. The new method consists of a computationally simple procedure for estimating the community transmission parameter and the standard error of the estimator. This method, as well as a previously developed maximum likelihood procedure, is applied to illness data as a means for distinguishing among broad patterns of illness acquisition. The periods evaluated are the rhinovirus, influenza A and influenza B seasons in Tecumseh, Michigan, for the years 1976–1980. Tecumseh households are stratified into exposure groups depending on age–group composition. Estimates are found for the risk differences of illness acquired from the community for households with different age–group distributions. Analysis of the Tecumseh data reveals that members of households with children are more apt than members of households without children to acquire illness associated with rhinovirus, influenza A and influenza B from the community. Members of households with just preschool children (and adults) are more apt than members of house–holds with just school children (and adults) to acquire illness associated with rhinovirus and influenza B from the community. In contrast, members of households with just preschool children (and adults) are just as likely as members of households with just school children (and adults) to acquire illnesses associated with influenza A from the community.