Association of modifiable and non-modifiable personal characteristics with future physical activity status: data from the UK Biobank. (November 2017)

• Record Type: Journal Article
• Authors: Warr, William; Foster, Charlie; Doherty, Aiden
• Is Part Of: Lancet. Volume 390(2017)Supplement 3
• Page Start: S93
• Publication Date: 2017-11
• Subjects: Medicine -- Periodicals; Medicine -- Periodicals; Medicine; Medicine; Electronic journals; Periodicals; 610.5
• DOI: 10.1016/S0140-6736(17)33028-3
• Languages: English
• ISSNs: 0140-6736

Abstract

Abstract: Background: Physical inactivity causes 5 million premature deaths per year worldwide, but current evidence relies on self-reported data, which are subject to recall and social desirability biases. Moreover, the small number of objective studies that do exist are limited by sample size and heterogeneous methods of analysis. We aimed to discover the modifiable and non-modifiable personal characteristics associated with physical activity outcomes using the world's largest dataset for objectively measured physical activity. Methods: We used the UK Biobank dataset in which 502 619 participants filled in touch screen questionnaires between 2007 and 2010; data included 106 personal characteristics, both non-modifiable (eg, age, sex, ethnicity) and modifiable (eg, stairclimbing frequency or alcohol consumption), and 30 prevalent diseases defined by Hospital Episode Statistics (HES). A subset of 103 712 participants then wore an accelerometer to measure physical activity status between 2013 and 2015. We extracted the raw accelerometer data to a summary physical activity variable, following procedures developed by the UK Biobank expert working group, including a minimum 72 h wear time. A phenome-wide association study with ordinary least squares linear regression was used to associate factors with physical activity adjusted for age, sex, and other confounding factors, with Bonferroni correction for multiple testing. Findings: 95 918 participants (53 714 women [56%], 88 245 white [92%]) were included in our final analysis with physical activity measured a mean of 5·7 years later (SD 1·1 years). We identified 56 characteristics (eg, self-rated health, ease of getting up in the morning) that were strongly associated (p<0·0004) with objective physical activity outcomes, 15 of which were previously unidentified characteristics (eg, shift work). 12 of these characteristics could be modifiable via policy changes; three were novel characteristics: feeling miserable (p<1 × 10 −5 ), number of vehicles in a household (p<1 × 10 −15 ), and ease of getting up in the morning (p<1 × 10 −67 ). 11 HES-defined diseases were also identified as statistically significant, but not novel. Interpretation: This study has identified new potential modifiable and non-modifiable characteristics associated with future objectively measured physical activity status. Although this cohort is not representative of the UK population, our findings suggest that further research should examine the association between these characteristics, and their gene–environment interactions with other objective measures of physical activity. Funding: None.