Tracking of voluntary exercise behaviour over the lifespan

Abstract

Background: The aim of many physical activity interventions is to develop life-long habits of regular exercise and sports activities in leisure time. Previous studies that assessed tracking (i.e. the stability of a trait over the lifespan) of leisure time exercise behaviour across various parts of the life span have treated it as a uniform construct by summing all types of leisure time exercise activities into a single summary score for the total volume of exercise. This study provides new insight by additionally determining tracking across leisure time exercise activities in six different domains: (1) team-based versus solitary activities, (2) competitive versus non-competitive activities, and (3) externally paced versus internally paced activities. We also assessed which of the domains of exercise activities best predicted total volume of exercise at follow-up.

Methods: A large dataset (N = 43,889) from the Netherlands Twin Register (NTR) was used to analyse the tracking of exercise behaviour over time. Using this dataset, we were able to examine tracking as a function of baseline age (8 to 80 years) and tracking duration (2 to 22-year follow-up), taking into account sex differences, using generalized estimating equations.

Results: Two-year tracking coefficients are moderate to high for total volume of exercise across ages at baseline, ranging from .38 to .77 with a median of .57. Tracking coefficients tend to decrease as the distance to follow-up increases, down to a median of .38 for the 22-year tracking coefficients. The patterns of tracking were largely domain-independent and were largely similar for solitary, competitive, non-competitive, externally and internally paced activities. With the exception of team-based activities, tracking was seen to increase as a function of baseline age. Cross-domain tracking did not favour any specific domain of exercise activity as the best predictor for total volume of exercise behaviour and this was true at all baseline ages.

Conclusion: We conclude that exercise behaviour is moderately to highly stable across the life span. In particular in adulthood, where the tracking of exercise mimics that of a classical behavioural trait like personality. This stability reinforces existing evidence that exercise habits are hard to change, but at the same time suggests that successful intervention leading to the adoption of exercise habits will tend to last.

Keywords: Behavioural trends; Competitive exercise; Leisure time physical activity; Lifespan; Longitudinal stability; Team exercise.

Fig. 1

Mean volume of exercise in males (top panel), and females (bottom panel) over time, total volume and volume of exercise in all three dimensions (competitive & non-competitive, team & solitary, and externally & internally paced)

Fig. 2

Correlations (r) between exercise domains as a function of age

Fig. 3

Tracking coefficient heatmap of total volume of exercise. The inset in the bottom right corner represents sample size (N) in each cell, cut off at N = 300 for visual purposes. Colours in the main panel represent the value of the tracking coefficient (grey to blue to red from low to high), colours in the inset represent number of samples (grey to yellow to red from low to high). To demonstrate; the cell in the full circle represents the tracking coefficient (0.81) of age 48 at baseline, and age 62 at follow-up in females. The cell in the striped circle represents the tracking coefficient (0.57) of age 22 at baseline, and age 26 at follow-up in males

Fig. 4

Tracking coefficients for total volume of exercise behaviour (top panel), and team-based exercise (bottom panel) as a function of age at baseline