Why the mortality penalty of smoking is rising over time

The adverse association between smoking and mortality is well known, yet estimated risks for smokers have risen markedly in recent decades. Explanations range from physiological impacts of changes in smoking intensity to shifts in the social characteristics of smokers. However, these hypotheses have rarely been tested. Using National Health Interview Survey data linked to the National Death Index through 2018, we confirm the mortality penalty associated with smoking increased by nearly 50% from 1987-1990 to 2012-2013, controlling for age. We show that this growing gradient between smokers and nonsmokers is driven largely by changes in the age distribution of smokers. This occurs even as more recent cohorts of smokers exhibit lower education, worse health, and greater psychological distress. These findings illustrate how evolving population composition can shape estimated mortality risks for exposures such as smoking and hold implications for understanding temporal changes in other risk factors like obesity.