Abstract
OBJECTIVES: The aim of the study was to evaluate concurrent changes in physical activity and self-rated health during retirement transition over 4 years by multivariate trajectory analysis and to examine whether sociodemographic and lifestyle factors predict the probability of being classified to a certain subgroup of observed changes. DESIGN: Prospective cohort study. SETTING: Public sector employees. PARTICIPANTS: 3550 participants of the Finnish Retirement and Aging study. PRIMARY AND SECONDARY OUTCOME MEASURES: Participants estimated on a yearly questionnaire their weekly hours of different types of activities converted to metabolic equivalent of task-hour/week. Self-rated health was assessed on a 5-point Likert-like scale from poor to excellent and dichotomised as suboptimal and optimal. Multivariate trajectory analysis was used to distinguish different subgroups of trajectories. Multinomial regression analysis was used to describe the associations between covariates and the probability of being classified to a certain trajectory group. RESULTS: Three trajectory groups were identified, all displaying increasing activity during retirement with a simultaneous decrease in perceived suboptimal health. Physical activity peaked at 18 months after retirement and then slightly decreased, except for initially highly physically active participants (9%) with good self-rated health, who demonstrated a constant high level of physical activity. Male gender, professional occupation, being married or cohabiting, body mass index <30 kg/m(2), not smoking and using alcohol below risk levels were associated with higher physical activity and better self-rated health. CONCLUSION: Changes in physical activity and perceived health during retirement transition were interconnected. Both were improved during retirement transition, but the change was temporary. Longer follow-up studies are required to assess the changes over a longer period after retirement.