Divergence in smoking and drinking trends: Results from age-period-cohort analytical approach

Smoking has dropped substantially over time while alcohol consumption has increased. Understanding how smoking and drinking relate to age, time-period or birth cohorts may improve public health measures.

The cross-sectional Continuous Household Survey in Northern Ireland provided bi-annual, household-level data on smoking and drinking from 1985 to 2015. An age-period-cohort approach was employed using 10-year age groups and 9 birth cohorts. A descriptive analysis of smoking and drinking by birth cohort was performed and supplemented by a joinpoint analysis. An Intrinsic Estimator model was used to examine behaviour patterns defined separately by age, time-period, and cohort.

Age effects showed that smoking and drinking peaked during adolescence and young adulthood (relative risk (RR) of age 25-34 vs. all ages: RR _smoking = 1.59, RR _drinking = 1.61, p < 0.001), followed by a progressive subsequent decline with age. Distinct secular trends were observed whereby smoking decreased over time but non-abstinence rates in relation to alcohol consumption increased substantially from 1985 to 2015. Cohort effects in smoking and drinking suggested that Generation X (born around 1961-1980) appeared to have a slightly higher likelihood of smoking and drinking but this converged with other cohorts over time.

The absence of similar policy efforts for alcohol as given to smoking may explain the divergence in secular trends of smoking and drinking. Targeting interventions towards adolescents and young adults may offer the best chance of reducing smoking and drinking prevalence. Cohort effects suggest the need for further research about the relationship between drinking and Northern Ireland's political history.