Are a lack of social relationships and cigarette smoking really equally powerful predictors of mortality? Analyses of data from two cohort studies

G David Batty,Paola Zaninotto,Marko J Elovainio,Christian A Hakulinen

doi:10.1016/j.puhip.2021.100140

Abstract

ObjectiveThe suggestion from cross-review comparison that lower levels of social integration (social isolation, loneliness) and cigarette smoking are equally powerful predictors of premature mortality has been promulgated by policy organisations and widely reported in the media. For the first time, we examined this assertion by simultaneously comparing these associations using data from two large cohort studies. Study designIndividual-participant analyses of two large prospective cohort studies. MethodsParticipants in UK Biobank and the English Longitudinal Study of Ageing reported loneliness, social-isolation and smoking behaviours using standard scales at baseline. Cause-specific mortality was ascertained via linkage to national registries. We used Cox regression analyses to compute a relative index of inequality to summarise the relation between these baseline characteristics and mortality experience. ResultsMean age at baseline was 56.5 years in the 466,876 (273,452 women) Biobank participants and 66.1 years in the 7505 (4123 women) English Longitudinal Study of Ageing members. In Biobank, a mean duration of mortality surveillance of 6.6 years gave rise to a total of 13,072 deaths, while in the English Longitudinal Study of Ageing, 1183 deaths occurred after a mean of 7.7 years. In ascending magnitude, loneliness, social isolation then cigarette smoking were associated with an increased risk of mortality from all-causes and all cancers combined. When cardiovascular disease mortality was the endpoint of interest, both smoking and social isolation, though not loneliness, revealed similar relationships. ConclusionsContrary to cross-review comparisons, in the present datasets it appears that poor social integration is in fact less strongly linked to total mortality than cigarette smoking.

Highlights

The reviews on which these associations are based, posit that the magnitude of this association is the same as that for risk factors known to be causally linked to mortality, such as cigarette smoking [1, 2]
For the first time to our knowledge, we compared the magnitude of individual associations of social support, loneliness, and cigarette smoking with total and cause-specific mortality using individual-participant data from two prospective cohort studies
Ratings were summed (ELSA range: 0–5; Biobank range: 0–3), with a higher score denoting greater isolation. In both studies, smoking behaviour was based on standard enquiries and mortality, including cause of death, was ascertained by linkage of study members with a national registry

Summary

Introduction

The reviews on which these associations are based, posit that the magnitude of this association is the same as that for risk factors known to be causally linked to mortality, such as cigarette smoking [1, 2]. For the first time to our knowledge, we compared the magnitude of individual associations of social support, loneliness, and cigarette smoking with total and cause-specific mortality using individual-participant data from two prospective cohort studies. We used data from UK Biobank [3,4] (baseline 2006–10) and the English Longitudinal Study of Ageing (ELSA) (baseline 2002–03) [5,6], on-going prospective cohort studies with detailed and comparable in formation on social isolation, loneliness, and cause-specific mortality.

Results

Conclusion