Main findings
The findings of this theory-based interview study provide important insights into the perceptions and experiences of PA change among individuals within the transition from employment to retirement. A number of inter-related factors were sensitive to change during the retirement transition. Individuals perceived a reduction in common barriers to PA such as lack of time, energy and goal conflict after retirement. The cost of PA after retirement was perceived as important for some individuals. However, having time to research the most affordable PA classes and taking advantage of concessions may help to mitigate cost concerns. Although individuals generally perceived an increase in the availability of resources which facilitated PA in retirement, the loss of daily structure after leaving work had negative consequences for PA behaviour among some individuals. More specifically, lack of structure led to procrastination and some individuals emphasised the need for a higher level of motivation and self-discipline to ensure that preferences for other, more sedentary pursuits did not distract them from PA. Those who engaged in structured PA sessions before retirement were less likely to experience these negative consequences after retirement. Conscious efforts to create a similar daily structure in retirement to that whilst working may help to maintain PA levels. Retirement appears to be associated with a number of new opportunities, some of which indirectly facilitate or conflict with PA behaviour. For example, the exposure to a different social context and the adoption of roles and responsibilities may help or hinder PA behaviour in retirement for some individuals. Personal aspirations, purpose-seeking and priorities in retirement may determine whether or not new opportunities influence PA behaviour. Finally, some individuals experienced a number of distinct PA phases in retirement which have differential effects on PA.
Strengths and limitations
The study examined the subjective experience of working and retired participants to gain a deeper understanding of how PA is perceived to change during the retirement transition. The study utilised a theoretical framework covering all current hypotheses for behaviour and behaviour change. Therefore, the findings can be interpreted in line with theory. Using theory to inform the development of behaviour change interventions is widely recommended [27]. In contrast to previous studies investigating PA in retirement, this study recruited and explored the views and experiences of individuals who were close to retirement or recently retired. The interview schedule also included open-ended questions which elicited rich, descriptive data. The retirement circumstances reported by participants were heterogeneous and reflected a contemporary picture of retirement [30]. Framework analysis enabled the comparison of sub-groups to determine whether perceived PA changes differed as a function of anticipation or experience of retirement.
However, some limitations of the research are acknowledged. Comparative views between working and retired participants are cross-sectional. In addition, the personal and professional characteristics of the interviewer can influence the collection and interpretation of qualitative data [31]. Thus, research findings have been described as a joint product of the participant, the researcher and the relationship between the two parties [32]. To reduce the potential for researcher characteristics to bias data collection, a reflexive interviewing style was adopted. This involved rephrasing questions when there were issues with comprehension, ensuring participants were given sufficient time to think about their responses, and assuring participants there were no right or wrong responses. These strategies can also reduce the likelihood of participants giving stereotypical or socially desirable answers when they do not have any pre-formulated views on the interview topics. The interviewer informed participants that she was interested in views and experiences of PA during the retirement transition and participants seemed at ease providing the researcher their views on this topic. To address the potential for the researcher’s experience to bias the selection and interpretation of data and to ensure that multiple and alternative interpretations were considered, two data analysis clinics were held where the selection and interpretation of the data was scrutinised by members of the research team. Returning the transcripts to participants for checking and eliciting feedback from participants on the findings may have further reduced potential for bias in the selection and interpretation of data. The value of the factors presented as predictors of PA change may be limited because subjective perceptions of how PA and PA determinants change after retirement may not correspond with actual changes. However, eliciting subjective perceptions can illustrate areas where opportunities exist to encourage individuals to increase or maintain PA behaviour during the retirement transition. The current evidence base for the predictive and potentially causal determinants of PA supports the assumption that interventions effectively targeting predictive cognitions can produce PA changes. The sample was relatively small which may limit the generalisability and transferability of the findings to other individuals or populations. However, the varied sample, including pre-retirement and post-retirement individuals, revealed novel insights into how PA is perceived to change after retirement.
Relationship to previous research
This is the first study to adopt an inclusive theoretical approach to understanding PA change during the retirement transition. Only a limited number of theory-based determinants have been examined as potential candidates explaining PA behaviour in previous studies [15,16]. This is problematic because theories represent the proposed causal determinants of behaviour and therefore provide mechanisms responsible for behaviour that can be targeted during intervention. Furthermore, previous studies have focused on exploring the motivators and barriers to PA in individuals who have retired [15,16] and less on PA determinants which may be sensitive to change during the transition from employment to retirement. As a result, previous studies reveal little about the specific opportunities to promote PA during the retirement transition.
To date, conclusions about PA behaviour in retirement have been based on retrospective reporting in samples of participants who have been retired for some time [13,15,16,33,34]. Retrospective reports may be prone to recall bias and may not accurately capture all important influences of behaviour at the time of the transition. These studies also assume static PA determinants in retirement and fail to capture the dynamic nature of PA during the course of retirement. The findings from the study suggest PA behaviour can change during and up to two years after the initial retirement event. It is important to explore the views of individuals who are approaching retirement and have recently retired. Individuals at later stages of retirement, as those included in previous studies, may have already settled into a more habitual pattern of behaviour and may not be as receptive to programmes aiming to change their PA behaviour. Furthermore, participant samples in previous studies have been biased towards the inclusion of healthy retired individuals who have a high SES, engage frequently in PA behaviour and have exceptionally high levels of motivation for PA behaviour [15,16,33,34]. This study explored a range of views by including a diverse sample of participants at different stages during the retirement transition, with a range of SES and occupational backgrounds and different levels of PA behaviour and motivation for PA behaviour at the time of the interview.
The findings from this study suggest that, for the majority of individuals, retirement may be associated with favourable circumstances which promote the adoption or increased engagement of PA. Based on these findings, PA levels may be predicted to increase around the retirement period for many people; however, studies assessing objectively measured PA in discrete age groups do not appear to provide evidence to confirm these findings but rather a gradual decline in PA is identified with retirement [5]. Previous research has shown that levels of regular, sustained PA decreases during late adulthood [35]. It is possible that the perception of favourable changes within the transition from employment to retirement, are temporal and short-lived. For example, the perception of time availability, increased energy levels, decreased goal conflict and new opportunities for PA may diminish rapidly as soon as an individual has filled up their post-retirement schedule.
It appears that the inconsistent findings regarding the direction of PA change in previous studies may be explained by the considerable intra-individual variability in retirement-related lifestyle changes. Previous research has identified that SES indicators, such as the distinction between manual or non-manual occupations, may be key moderators of the relationship between retirement and PA levels [9,12,36]. In this study, some individuals with sedentary occupations anticipated a decrease in PA levels, which is not in line with previous predictions [12]. The findings from this study suggest that several positive and negative inter-related factors may influence PA levels after retirement, which are not explained by previous occupation alone. For example, being exposed to a different social context after retirement can have an impact on PA behaviour, particularly in respect to the attitudes and preferences of family members. Similar findings have been reported in a recent study exploring how couples influence each other’s physical activity behaviour in retirement [37].In addition, the structure of pre-retirement PA may be a key moderator of the relationship between retirement and PA.
Interviewing participants who were at different stages of the retirement transition revealed that retirement may involve a series of different phases which have differential effects on PA behaviour. This finding complements Atchley’s conceptual model describing the adjustment to retirement [38], where retirement is conceptualised as a transitional process which involves different distinct phases. More specifically, the model suggests that some individuals are quick to establish a retirement routine after leaving work, whereas others enter either a ‘honeymoon’ phase where they initiate new activities and opportunities or a ‘rest and relaxation’ phase, where they enjoy a temporary period of calmness after leaving work. It appears that these distinct phases may also shape PA patterns during the retirement transition and individuals may find that they take several months to settle into their post-retirement PA patterns.
The findings from this study support the idea that contemporary routes to retirement are highly individualistic [39]. As a result, PA changes are likely to vary considerably from one individual to another. The study revealed that perceptions of PA change during retirement can be extremely diverse and appear to be determined by individual circumstances including the timing of retirement, the reason for retirement and the voluntary or involuntary nature of retirement. The impact of retirement on PA also appears to depend on exclusive preferences and responsibilities after retirement and the way in which individuals prioritise PA against other aspirations for their retirement.
Implications
The findings from this study offer understanding of how interventions delivered during the retirement transition can be optimised. Retirement may represent an ideal opportunity to promote PA behaviour in older adults because individuals may be particularly motivated and receptive to attempts to change behaviour due to a perceived reduction in common barriers to PA. Attempts to emphasise changes in time availability, energy, increased opportunities through physically active social or voluntary roles and the availability of free or reduced PA types may be most effective to ensure individuals engage in sufficient PA behaviour after retirement. The potential short-term window of opportunity to promote the positive changes facilitating PA behaviour after retirement has important implications for the timing of future PA interventions. PA interventions may need to commence when individuals are approaching retirement and are in the process of planning their post-retirement schedule. At this stage it may be beneficial to emphasise to individuals that they may experience an increase in energy levels to engage in PA after retirement and to encourage individuals to consider how PA fits in with their retirement roles and goals. Previous research has shown that interventions targeting PA behaviour in adults of retirement age are viable and may be more effective if they include components that are personalised to the individual and their environments [40,41]. Designing interventions which are tailored to the specific needs of the individual may be particularly important when targeting behaviour change in retirement due to the heterogeneity of routes to retirement. One-to-one, personalised intervention approaches are likely to be costly if delivered face-to-face. PA interventions which adopt digital platforms may be more cost-effective and have the additional advantage of reaching a greater number of individuals [42,43]. The content and interface of web-based interventions can be adapted and personalised to the characteristics and preferences of the individual. Encouraging findings are emerging about the acceptability and effectiveness of web-based interventions in older adults [44-46].
Future research
Future research should explore the extent to which perceived changes in PA during the retirement transition reflect objective changes in PA during this period using methodology which focuses on PA change within individuals. Future research should also explore how individuals perceive sedentary behaviour to change during the retirement transition and how sedentary interests fit into their retirement routines and interpret this in line with the findings from the small number of studies which have examined changes in sedentary behaviour during the retirement transition [6,10,11]. It remains unclear how post-retirement PA is maintained into older adulthood. Future research should investigate the maintenance of PA habits developed during the first two years of retirement into older adulthood. Future research should concentrate on developing and evaluating a range of theory- and evidence-based interventions to promote and maintain PA in and beyond the retirement transition. Intervention research should focus on cost-effective personalised delivery modes (e.g. using digital media).