The limited success in getting and keeping older adults physically active shows a great need for knowledge of determinants of physical activity. Even though a variety of possible determinants have been identified (e.g., characteristics of the individual, of the social and physical environment), major life events remained largely unstudied as possible motivators or barriers of physical activity in older adults. The current study examined the association of two major life events, namely ‘widowhood’ and ‘retirement’ with MVPA and sports participation among older adults. MVPA increased in those who retired compared to those who continued employment, but no influence of widowhood was observed. Furthermore, we observed no association between either major life event and sports participation.
The finding of no difference in MVPA or sports participation between those who lost their partner compared to those whose marriage continued was surprising, since we had hypothesized that a major life event, like widowhood could induce a change in daily routine and disrupt physical activity habits and result in a decline in physical activity
[30–32]. A possible explanation is that widowhood may have acute and short-term effects on physical activity, but that these effects had subsided by the time of the follow-up assessment. Wilcox and King
 showed a decline in adherence to an exercise intervention after interpersonal loss, suggesting immediate short-term effects on exercise behaviour. However, interpersonal loss may have a different impact on organised exercise sessions - as observed in the aforementioned intervention study - compared to daily physical activity, like biking or walking for transport
. Alternatively, in the period preceding widowhood, individuals may already adapt their physical activity to a lower level because of caring commitments. Assessing caring commitments and exact time of widowhood could provide additional insights in physical activity patterns. In this study, we were not able to examine such pre-widowhood adjustment. Future research is needed to explore the effects of widowhood on different aspects of physical activity, e.g. supervised versus unsupervised exercise and changes preceding a major life event.
As expected, retirement was significantly and positively associated with physical activity. Retired participants significantly increased their time spent in MVPA compared to employed participants. Contrary to our hypothesis no significant association with sports participation was observed. One previous study, similar to our design, used validated physical activity questionnaires both at baseline and follow-up and assessed sports participation separately
. This study showed an increase in sports participation, as assessed on a 5-point scale. A possible explanation for our contrasting finding may lie in our assessment of sports participation. To study the association between major life event and sports participation we have classified participants as either participating or not participating in sports. This classification allowed us to assess changes in the uptake or termination of sports participation but was not sensitive enough to detect an increase or decrease in time spent in established sports participation. Another explanation could be the cause of retirement. The retirees in our study population were on average 58 years of age, well below the 65 years mandatory retirement age. A recent study showed that early retirees (aged 55–58) were less likely to have a good perceived health, and health reasons may have been the reason for retirement
. This could explain our finding of the stronger association of retirement and MVPA with increasing age. Previous work from Lahti et al.
 supports this view. Their study shows that contrary to older-age retirees, disability retirees did not increase their physical activity from baseline to follow-up. Future research is needed to explore the effects of various reasons for retirement on change in physical activity and sports participation. This may also include the physical burden of work and the economic status of the jobs, for these are factors that could possibly influence the need and opportunity for early retirement. Alternatively, like we proposed for widowhood, individuals who anticipate their retirement may have started to adjust their level of MVPA, so that we could not observe change in our study.
Since we have no data available on work-related physical activity or total physical activity no conclusion can be drawn regarding the net results in physical activity. Therefore, we cannot tell whether the increase in moderate to vigorous physical activity compensated for the loss of work related activity. Touvier et al.
 concluded that the increase in overall leisure time physical activity after retirement did not compensate for the loss of work-related activity. Cozijnsen et al.
 concluded that recent retirees, as compared to older cohorts of retirees, were more likely to report sports participation. More research is needed to systematically assess the effects of retirement on MVPA and sports participation.
The main strength of our study is the use of the longitudinal data. This prospective cohort study allowed us to examine the association of widowhood and retirement with change in physical activity, using the same validated physical activity questionnaire at both time points.
Some limitations should also be mentioned. First, the small number of participants that widowed or retired limited our sample size. A second limitation is the reliance on self-reported data of physical activity. Questionnaires are generally suitable for categorizing or ranking people in levels of physical activity but not accurate enough to assess absolute levels of physical activity
[35, 36]. Since ranked MVPA scores and dichotomized sports participation we were not able to express the difference between groups in minutes per day. Finally, we have studied the independent influence of two major life events on physical activity. Future research is needed to explore the underlying pathways of how and when major life events influence physical activity.
In conclusion, our findings suggest that the association between major life events and MVPA and sports participation varies by the two major life events under investigation. Widowhood was not associated time spent in MVPA or sports participation, while retirement was associated with higher levels of MVPA, but not sports participation. Research is needed to study the underlying pathways of how major life events are associated with physical activity. Future research needs to assess the effectiveness of physical activity interventions targeting older adults just before or after a life-changing event.