Aggregation of cancer in relatives of National Population Health Survey respondents with metabolic syndrome-associated diseases
Colorectal, breast, and prostate cancer share aetiologic risk factors with metabolic syndrome diseases, including diabetes and hypertension. Therefore, an underlying process may link these conditions. One way to elucidate potential links is to examine the aggregation of diseases in relatives. The present study assessed familial cancer in 1998 National Population Health Survey respondents with and without diabetes or hypertension, using logistic regression. Only two of 16 estimates were statistically significant after controlling for proband age, sex, and body mass index. Two explicit processes for assessment of interaction and confounding were undertaken. Each process identified similar variables; however, controlling for these variables did not provide any further insight. Monte Carlo sensitivity analyses were conducted to assess possible bias by reporting error. Even if a moderate association existed, realistic levels of proband reporting would have biased results towards the null, under non-differential misclassification, or in either direction under differential misclassification.