Background
Smoking is often colloquially considered “social”. However, the actual relationship of smoking with current and future social isolation and loneliness is unclear. We therefore examined these relationships over a 12-year follow-up.

Methods
In this cohort study, we used a nationally representative sample of community dwelling adults aged 50 years and over from the English Longitudinal Study of Ageing (N=8780) (45% male, mean(SD) age 67(10) years. We examined associations of self-reported smoking status at baseline assessment, with social isolation (low social contact, social disengagement, domestic isolation), and loneliness (3-item UCLA loneliness scale), measured at baseline, and follow-up at 4, 8 and 12 years, using ordinary least squares regression models.

Findings
At baseline, smokers were more likely to be lonely (coef.=0·111, 95% CI 0·025 – 0·196) and socially isolated than non-smokers, having less frequent social interactions with family and friends (coef.= 0·297, 95%CI 0·148 – 0·446), less frequent engagement with community and cultural activities (coef.= 0·534, 95%CI 0·421 – 0·654), and being more likely to live alone (Odds Ratio =1·400, 95%CI 1·209 – 1·618). Smoking at baseline was associated with larger reductions in social contact (coef.=0·205, 95%CI 0·053 – 0·356, to 0·297, 95%CI 0·140 – 0·455), increases in social disengagement (coef.=0·168, 95%CI 0·066 – 0·270, to coef.=0·197, 95%CI 0·087 – 0·307), and increases in loneliness (coef.=0·105, 95%CI 0·003 – 0·207), at 4-year follow-up) over time. No association was found between smoking and changes in cohabitation status. Findings were independent of all identified confounders, including age, sex, social class and the presence of physical and mental health diagnoses.

Interpretation
Smoking is associated with the development of increasing social isolation and loneliness in older adults, suggesting smoking is detrimental to aspects of psychosocial health. The idea that smoking might be prosocial appears a misconception.