The first aim of this study was to examine whether 2-year changes in specific physical activities and sedentary behaviors differed between retiring adults (i.e. adults who retired between baseline and follow-up) and recently retired adults (i.e. adults who were retired for at least 6 months and maximum 5 years at baseline). Our findings showed that leisure-time cycling increased in retiring adults, but decreased in recently retired adults. Furthermore, (voluntary) work-related walking and MVPA decreased strongly in retiring adults, while a slight increase over time was found in the recently retired group. Regarding sedentary behaviors, passive transport decreased more strongly in recently retired adults than in retiring adults, and computer use increased more strongly in retiring adults compared with recently retired adults.
These findings are partly in line with previous results of Menai and colleagues [13], who also found that retiring adults’ leisure-time PA increased, while retired adults’ leisure-time PA decreased. However, they also found an increase in household PA in retiring adults, while our findings could not confirm this. Concerning TV viewing and computer use, Menai and colleagues [13] concluded that the increases were larger in retiring adults than in retired adults. No other types of sedentary behavior were included in their study, nor was (voluntary) work-related PA. The strong decrease in (voluntary) work-related walking and MVPA found in retiring adults in the present study is very logical. The fact that voluntary work-related PA increases in recently retired adults is promising as it means that some of them fill up their free time with active voluntary work. Furthermore, our results, as well as the previous findings of Menai and colleagues [13] confirm the hypothesis that adults increase specific types of PA (leisure-time cycling) at the start of retirement, probably because the transition to retirement induces reduced time constraints and because many adults perceive the start of retirement as a moment to take up healthy habits and to replace their working-day routine with new routines [21, 22]. Nonetheless, a decrease in cycling was identified in adults who were already retired at baseline. Based on these findings, it seems like for specific PA types, adults tend to lapse into old habits once they are accustomed to retirement. Previous qualitative research showed that individuals subdivide ‘retirement’ in different temporary phases, each with distinct PA patterns [23]. So, later phases of retirement may imply decreased levels of PA. However, in order to firmly confirm this suggestion, changes in PA patterns of retiring adults should be followed up during a longer period (e.g. from before retirement until 5 or more years after retirement).
Concerning sedentary behaviors, our findings are not very straightforward. A positive change (i.e. decrease) was identified for passive transport, while computer use increased, and for both behaviors changes were more pronounced in retiring adults. Most previous studies only included TV viewing time when examining changes in sedentary behavior [10, 11], but our results confirm that computer use is becoming highly prevalent in older adults, and should certainly be included as a sedentary behavior to target in future studies. The identified decrease in passive transport in retiring adults is very positive, not only for health, but also for the environment (e.g. pollution), and is probably mainly related to quitting work. The continued decrease in passive transport in recently retired adults might have less positive consequences too: old age is often associated with increased isolation and decreased mobility, so it might be the case that retired adults are more ‘tied’ to their home [24]. In future research the reasons for this decrease in passive transport should be examined in more detail. Overall, our results confirm the need to include several sedentary behaviors in future studies, instead of focusing only on TV viewing time or overall sedentary time.
The second aim was to examine if educational level and gender moderated the effects of retirement status (retiring versus recently retired) on the 2-year changes in PA and sedentary behaviors. Concerning the moderating effects of educational level, our findings showed that low-educated recently retired adults had the strongest decrease in walking for transport, while low-educated retiring adults’ work-related walking decreased most strongly. Previous studies showed that low-educated adults (who often have a physically demanding job, which was also the case in our study [results not shown]) have stronger decreases in PA during the transition to retirement than high-educated adults, on the one hand because they usually do more work-related PA before retirement, and on the other hand because they often believe retirement is a period of well-deserved rest [10, 14]. Our findings on transport-related walking confirm that also after retirement, low-educated adults are an important at-risk group for having low PA levels. Furthermore, TV viewing time and computer use increased most strongly in low-educated retiring adults, which confirms the susceptibility for low-educated adults to lapse into an unhealthy lifestyle around and after the transition to retirement. Nonetheless, no moderating effects of educational level were found for most other PA and sedentary behaviors, so it seems that future interventions may be most effective when both high- and low-educated adults are targeted.
Regarding the moderating effects of gender, our results showed that work-related walking and MVPA decreased in recently retired men, while increases were identified in recently retired women; MVPA during gardening decreased strongly in recently retired men, and increased in retiring men, retiring women and retired women. However, retiring women showed strong increases in passive transport and computer use while retiring men’s passive transport decreased and computer use increased moderately. So, these findings demonstrate that retiring women are inclined to increase MVPA during gardening, but also their passive transportation and computer use, more than retiring men. Almost no previous studies examined gender differences, but those that did found post-retirement increases in PA to be slightly higher in men than in women [8], which conflicts with our results.
Previous studies carefully suggested that interventions aiming to increase PA around the age of retirement, should target individuals who retired recently or who are planning to retire in the near future [9–12]. However, those studies only compared retiring adults with adults who continued to work, and did not include recently retired individuals. So, based on our study, these recommendations can be somewhat adapted and fine-tuned. Our findings suggest that interventions should focus not only on retiring, whose overall PA tends to decrease because the decline of work-related PA is not sufficiently compensated by increases in other domains. Future research should also pay attention to long-term follow-up as the changes in behavior seem to be dependent on the specific stage of retirement adults are facing. Furthermore, interventions should also focus on sedentary behaviors, with specific attention for sedentary time during computer use in both retiring and recently retired adults. Low-educated retiring and recently retired adults seem to be particular at-risk groups for insufficient transport- and voluntary work-related walking and high levels of TV viewing and computer use, so they should receive specific attention in interventions. Nonetheless, since no moderating effects of educational level were found for the other behaviors included in the study, it can be recommended to target both high- and low-educated adults in interventions. No straightforward results were identified regarding the moderating role of gender, so currently, tailoring of interventions towards retiring/recently retired men or women cannot be recommended.
Strengths of this study firstly include that a broad range of PA and sedentary behaviors were examined. Previous studies mainly focused on changes in leisure-time PA and TV viewing time, or overall PA and sedentary time. Secondly, this study had a strong (longitudinal) design. Furthermore, some limitations should be acknowledged. First, a relatively small sample of retiring adults (n = 105) participated, which limits the power of our analyses and the generalizability of our findings. The three-way interaction analyses were particularly underpowered with approximately 50 participants in each group of retiring adults (i.e. retiring men, retiring women, retiring low-educated adults and retiring high-educated adults). Second, follow-up time was limited to 2 years. Future studies on health behaviors over the retirement transition should include more participants and should follow the same cohort of participants over a longer period (e.g. from before retirement to 5 years post retirement). Third, selection bias was probably present as only 446 of the 7500 adults who received an invitation letter for the study, completed both baseline and follow-up measurements. It may thus be the case that mainly motivated and active adults responded to the invitation letter. Fourth, only self-reported PA and sedentary behaviors were included. As the use of questionnaires is subject to several biases (e.g. recall bias, social desirability), it is recommended to combine this with objective assessments of PA and sedentary time in future studies.