Exploring the links between health and housing: The limitations of population health surveys

While it is intuitively attractive to link health status and the quality of housing, it is methodologically complex to identify the relative importance of housing because individual characteristics and environmental variables act as confounders. A secondary issue is that the data sets which contain detailed data on health status and environmental variables often contain only limited data on housing. Similarly, housing surveys rarely contain detailed data on health and environmental variables. Respondents from Montreal, Toronto and Vancouver in the public use micro-data files of the 1996/97 cycle of the National Population Health Survey make up the data sets for a series of logistic regression models where health status, allergies, asthma, arthritis, migraines and bronchitis/emphysema act as the dependent variables. Sets of variables which measure housing, the environment (smoking and alcohol consumption) and the socio-economic status of individuals are used as independent variables. The issues of confounders and the limitations of the data are illustrated by the results of the analysis. If we are to go beyond local area surveys and carry out broader analyses of the links between health status and the quality of housing, researchers and data collection agencies will need to re-think the artificial divisions between these two critical aspects of peoples' lives.