Environmental tobacco smoke and childhood asthma: comparing exposure metrics using probability plots

This study demonstrates the use of probability plots for assessing exposure-disease relationships in case-control studies. We reanalyze data from a study of childhood asthma and exposure to environmental tobacco smoke, comparing alternative measures of smoking intensity. Previous analyses showed a higher mean urinary cotinine-to-creatinine ratio (CCR) in cases as compared with controls and a statistically significant odds ratio with exposure defined as 30 ng/mg CCR or higher. Using probability plots, we compare alternative exposure metrics and exposure cutpoints with respect to their ability to detect a statistically significant exposure-disease effect. A strong exposure-disease response is evident, overall, and the results are fairly robust to the choice of exposure cutpoint.