What are the associations between lifestyle, socioeconomic status, and COVID-19 mortality in UK Biobank?

Talk Code: 
3B.2
Presenter: 
Hamish Foster
Twitter: 
Co-authors: 
Dr Fred Ho, Prof Jason Gill, Dr Carlos Celis-Morales, Prof Frances Mair, and Prof Catherine O’Donnell
Author institutions: 
Institute of Health and Wellbeing, University of Glasgow

Problem

Lifestyle factors (e.g., smoking, low physical activity, obesity) and low socioeconomic status (SES) are each associated with COVID-19 mortality. However, it is unknown whether and how SES influences the association between combinations of unhealthy lifestyle factors and COVID-19 outcomes. Improving our understanding of this will inform population risk stratification and help protect groups more vulnerable to COVID-19. We examined how SES moderates the association between a lifestyle score (LS) and COVID-19 outcomes in a large prospective UK cohort.

Approach

We used data from UK Biobank - 502,505 participants recruited in 2006-2010. Variables were collected by baseline self-report; outcomes ascertained via routine registry linkage. We examined an unweighted LS comprising smoking, alcohol, physical activity, television time, sleep time, intake of fruit/vegetables, oily fish, and red/processed meat. We assigned 1 point for each unhealthy (guideline-based) lifestyle factor and categorised individuals as healthy (LS 0-2), moderate (3-5) and unhealthy (6-9). SES was measured by Townsend deprivation index (main analysis) and by income and education level (sensitivity analyses). Outcome measures were severe COVID-19 (infection diagnosed in hospital) and COVID-19 mortality (COVID-19 given as main cause of death). COVID-19 test data were available from 16 March to 17 June 2020. Mortality analyses were censored from 30 April 2020 or date of death if earlier. Using Poisson regression models, we examined associations between LS and COVID-19 outcomes and between SES and the same outcomes. Then, we examined associations between LS and SES combined. Outcomes were examined using a single reference group of healthy LS and high SES. Models adjusted for 1) sex, age, ethnicity; 2) +SES/LS; 3) +multimorbidity.

Findings

450,962 participants were included; 892 developed severe COVID-19 and 286 died. Both unhealthy LS and high deprivation had, independently, significant linear associations (fully adjusted) with COVID-19 outcomes. Each LS point was associated with risk ratio (RR) of 1.08 (95%CI 1.03-1.13) for severe COVID-19 and 1.12 (1.03-1.21) for COVID-19 mortality. Each deprivation point (1 SD) was associated with RR 1.10 (1.07-1.12) for severe COVID-19 and 1.13 (1.09-1.17) for COVID-19 mortality. Similar associations were seen using income or education for SES. Combined associations of LS and SES showed highest risks for COVID-19 outcomes in those with an unhealthy LS and low SES. There was a significant LS-income interaction, where associations between LS and COVID-19 outcomes were stronger in those with low income.

Consequences

The combination of unhealthy lifestyle and low SES is associated with a greater risk of poor COVID-19 outcomes. Strategies to reduce the impact from COVID-19 should consider the interaction between lifestyle and SES. Public health policy targeting lifestyle factors in poorer communities could incorporate COVID-19 related risk.

Submitted by: 
Hamish Foster
Funding acknowledgement: 
HF is funded by a Medical Research Council (MRC) Clinical Research Training Fellowship (grant reference number MR/T001585/1). The MRC played no role in any part of this study. FM, COD and JG are supervisors of the MRC Fellowship. FH and CCM received no specific funding for this work. All authors declare no conflicts of interest.