The Utility of Measures of Physical Behavior, Function, and Fitness as Predictors of Mortality

TL;DR

This study shows that lifestyle and physical health measures, like walking pace, can better predict mortality risk than traditional factors, especially in people with health conditions.

Contribution

The study demonstrates that physical behavior measures can improve mortality prediction models by replacing traditional clinical factors.

Findings

Replacing traditional risk factors with physical measures improved mortality prediction accuracy in people with health conditions.

Walking pace was the strongest individual predictor of mortality risk when substituting traditional factors.

Improvements in prediction were more significant in men and those with health conditions compared to healthy individuals.

Abstract

To assess whether simple measures of lifestyle and physical health improve mortality risk prediction in healthy adults and those living with health conditions when added to or substituted for traditional risk factors.

Data were from the UK Biobank (N=407,569; median age, 58 years; May 1, 2006 to September 30, 2022), stratified by sex and health status. The base model included the following: age, smoking status, body mass index, systolic blood pressure (SBP), total cholesterol-to-high-density lipoprotein cholesterol ratio (CHR), and material deprivation. Five simple physical measures (leisure time physical activity; sleep duration; resting heart rate; maximum handgrip strength; and walking pace [WP]) were added to or substituted for traditional risk factors (ie, SBP and CHR), both individually and combined. Model performance was assessed using the C-index and net reclassification index (NRI).

Replacing both CHR and SBP with all physical measures improved C-index of 0.022 (95% CI, 0.018-0.026) and NRI by 9.6% (95% CI, 6.5%-12.8%) in women living with a health condition. Corresponding values for men were as follows: C-index: 0.024 (95% CI, 0.020-0.027); NRI: 19.0% (95% CI, 16.6%-21.4%). The improvement of C-index was smaller in healthy women (0.006; 95% CI, 0.004-0.007) and men (0.007; 95% CI, 0.006-0.009). WP alone improved risk prediction and classification when replacing CHR and SBP in individuals with a health condition, with a C-index improvement of 0.015 (95% CI, 0.012-0.019) and an NRI of 11.0% (95% CI, 7.4%-14.6%) in women and 0.014 (95% CI, 0.012-0.017) and 14.0% (95% CI, 11.5%-16.5%) in men, respectively.

Measures of lifestyle and physical health improve mortality prediction and classification and could be used as potential substitutes for traditional clinical factors, particularly in populations with prevalent illness. WP was the strongest individual predictor.

Figures