Integrating Sleep, Physical Activity, and Diet Quality to Estimate All-Cause Mortality Risk: A Combined Compositional Clustering and Survival Analysis of the National Health and Nutrition Examination Survey 2005–2006 Cycle
Borja del Pozo Cruz, Duncan E. McGregor, Jesús del Pozo Cruz, Matthew P. Buman, Javier Palarea‐Albaladejo, Rosa M. Alfonso‐Rosa, Sébastien Chastin
Abstract
We aimed to compare all-cause mortality risk across clusters of adults ≥50 years of age (n = 1,035) with common lifestyle behaviors patterns, enrolled in the US National Health and Nutrition Examination Survey (2005-2006). Log-ratio coordinates of 24-hour movement pattern and z scores of diet quality were used as input into a model-based clustering analysis. A Cox regression model was fitted to ascertain the all-cause mortality risk associated with each cluster. Participants were clustered into 4 groups: 1) a group characterized by a better physical activity profile and longer sleep duration coupled with an average diet quality (cluster 1); 2) a group with the poorest activity profile and shortest sleep but also the best diet quality (cluster 2); 3) another group featuring lower levels of activity of either intensity and higher levels of sedentary behavior and also a poor diet quality score (cluster 3); and 4) a group with an average diet quality and the best activity profile in the sample (cluster 4). A combination of a poorer diet and activity profile increased the prospective risk of all-cause mortality. Our findings emphasize the importance of considering the combination of diet quality and 24-hour movement patterns when developing interventions to reduce the risk of premature mortality.