Model for the analysis of binary time series of respiratory symptoms

Environmental epidemiologic research on respiratory symptoms presents unique types of data, typically requiring simultaneous analysis of both time- and person-varying factors. In this paper, the authors propose a new, simple model that incorporates such factors and controls for each person's prior history of symptoms. The Yale Mother and Infant Health Study was undertaken to investigate the effects of ambient pollutant concentrations, meteorologic changes, and demographic variables on daily respiratory symptoms in both mothers and infants. This analysis was restricted to 673 mothers followed in southwestern Virginia from June 10 to August 31, 1995. Of the person-varying factors, husband's level of education, nested within marital status, and having pets in the home were related to an increased likelihood of new episodes; however, neither was related to duration of symptoms. Interestingly, women who were unmarried were least likely to have new episodes of respiratory symptoms, while those with the most highly educated spouses were most likely to have new episodes. Having pets in the home increased the likelihood of a new episode. Having a history of allergies and having children in day care were found to be related to the symptom of a runny or stuffy nose, in terms of both incidence and duration. The level of coarse particles was related to the incidence of new episodes of runny or stuffy nose, and a higher level prolonged the duration of symptoms. Sulfate level was not related to the incidence of new episodes but was associated with the duration of the episodes.