# Changes in physical activity across retirement: a compositional data analysis approach in a Swedish cohort study

## TL;DR

This study explores how retirement affects physical activity, sleep, and sedentary behavior in a Swedish cohort, finding that pre-retirement activity levels influence post-retirement changes.

## Contribution

The study introduces compositional data analysis to examine time-use changes during retirement, revealing how pre-retirement physical activity levels affect post-retirement movement behaviors.

## Key findings

- High pre-retirement occupational physical activity is linked to increased sleep-to-wake time ratio after retirement.
- Physically active time to sedentary behavior ratios converge across pre-retirement activity groups post-retirement.
- Tailored interventions may be needed for retirees based on their pre-retirement physical activity levels.

## Abstract

Retirement is a major life transition that can alter patterns of movement behaviors (physical activity, sedentary behavior and sleep). While some studies indicate an increase in physical activity post-retirement, others report a rise in sedentary behavior. However, evidence is lacking on how individuals re-allocate time among movement behaviors, particularly using analytical approaches that account for the co-dependence of 24-hour time-use data. Furthermore, little is known about how pre-retirement occupational physical activity (OPA) levels influence physical activity after retirement. This study examined changes in the relative time spent in sleep, sedentary behavior (SB), light physical activity (LPA), and moderate-vigorous physical activity (MVPA) over retirement, and how these changes vary by pre-retirement OPA levels.

Data were drawn from the Swedish Retirement Study, which followed 112 participants (47 men, 65 women; age: 60–72) at three timepoints during the retirement transition. Movement behavior and sleep data were collected over a week-long period using thigh-worn accelerometers and wrist-worn actigraphs. Compositional data analysis (CoDA) was employed to account for the co-dependent nature of 24-hour time-use data. Multivariable linear mixed models, adjusted for sociodemographic and health covariates, were used to evaluate the associations between retirement, OPA tertiles, and movement behaviors.

In the overall sample, changes in movement behaviors mainly involved sleep. However, substantial variation was observed across OPA tertile groups. The sleep-to-wake time ratio increased in the high OPA group and, to a lesser extent, in the medium OPA group. Regarding physically active and sedentary time, a convergence between the high and low OPA groups was observed, as pre-retirement differences diminished. Specifically, the ratio of physically active time to SB decreased in the high OPA group and increased in the low OPA group.

The findings indicate that pre-retirement OPA is a significant factor in understanding changes in movement behaviors during the retirement transition. The reduction in post-retirement physical activity among high-OPA workers may represent a healthier rebalancing rather than a decline, which aligns with the “physical activity paradox” and the “Sweet-Spot Hypothesis”. This evidence highlights the need for tailored interventions for retirees, particularly those from physically demanding occupations.

The online version contains supplementary material available at 10.1186/s11556-025-00395-6.