Are the religiously unaffiliated less likely to engage in health-related behaviours compared to Christians or Catholics? Evidence from the family cohort