If so, these symptoms
were added separately as independent continuous variables to the
adjusted regression models. Resulting changes in beta-coefficients
for event categories were considered measures of explanation. We
beforehand decided that changes more than 10% indicated the
presence of explanation