Profiles of physical activity biographies in relation to life and aging satisfaction in older adults: longitudinal findings

The present study focused on retrospectively reviewed PA biographies across the lifespan of older adults. The full study design is described elsewhere [11]. Ethical guidelines were followed and clearance from the ethics committee of the German Psychological Society (JZ062009). Participant inclusion criteria included being above the age of 60 and not having a medical inability to perform physical exercise (assessed in an open format with self-reporting). Participants were recruited via newspaper announcements. After they sent back the informed consent form, participants received questionnaires in the mail at baseline (i.e., PA biography), 6 months (i.e., current PA), and again at 12 months (i.e., indicators of life and aging satisfaction). Data collection took place in 2010 across Germany, with most participants coming from Berlin and the surrounding area. At baseline, 418 participants completed the questionnaire. The second questionnaire was completed by 340 participants (81% of baseline) and 335 participants completed the third questionnaire (80% of baseline).

The PA biography assessment was adapted from the 2005 study “Tracing the development of athletes using retrospective interview methods” [12] and was assessed with four single items: “In your childhood and adolescence, how frequently were you consistently physically active?”; “In your 30s, how frequently were you consistently physically active?”; “In your 50s, how frequently were you consistently physically active?”; and “In your 60s, how frequently were you consistently physically active?” A 6-point Likert-type response format (never, seldom, sometimes, quite often, mostly, and always) was used.

Current PA was assessed using two items adapted from a brief assessment of the Longitudinal Aging Study Amsterdam [13]: “In a normal week, how often do you perform physical exercise?” and “In the last four weeks, how often have you performed physical exercise?” Internal consistency was high with an item correlation of .85, p < .001 in the present study.

Life and aging satisfaction served as the outcome in this study and was assessed using the 17-item Philadelphia Geriatric Center Morale Scale (PGCMS), which consists of three components: non-agitation, life satisfaction, and aging satisfaction. For each dimension, the items were collapsed into two manifest mean score indicators [14]. Internal consistency was good, as indicated by Cronbach’s Alphas of .82, .58, and .68 for non-agitation, life satisfaction, and aging satisfaction, respectively.

Educational level was self-reported. Those who reported that they had a degree (i.e., from a university or technical college) were considered to have a higher education.

Multimorbidity was assessed with a list of 24 chronic conditions based on the Charlson Comorbidity Index, which has been used in other surveys as well [15, 16]. Participants were given the opportunity to name up to two more conditions. Those who reported five or more separate conditions were considered multimorbid.

Latent class analysis (LCA) refers to modelling with categorical latent variables (classes) that represent subpopulations where population membership is not known but is inferred from the data [17]. The number of latent classes was chosen based on Akaike’s and Bayesian information criterion (AIC, BIC), where lower values indicate better trade-off between the complexity and the goodness of fit of the model. Further, the Lo-Rubin Test (LR), where a significant value indicates that the present model is superior compared to one with less classes was used. In a consecutive step, we applied structural equation modelling to regress current PA and satisfaction on manifest PA biography classes. PA class allocation, sex, age, and current PA were manifest indicators, whereas each dimension of life satisfaction was modelled as a latent factor with two manifest indicators (mean scores of items). Gender and age served as covariates in the model. Missing values were accounted for by applying the Full Information Maximum Likelihood procedure (FIML) using Mplus v7.3, which incorporates all available data into the model estimation, i.e., retaining cases for which missing data are present [18]. The FIML procedure typically yields unbiased estimates even if data is missing at random. Fully standardized coefficients were reported.

This study has strengths and limitations. A strength of the study is the innovative approach to PA, which incorporates a perspective of current behavior and satisfaction with a genuine lifespan perspective using a simple and intuitive biography method. Other strengths include the longitudinal design and the sophisticated statistical analyses, wherein we combined LCA with structural equation modelling. One limitation is the validity of self-reported PA levels. However, the similarities in the results of this study with other studies on the same topic show the strength of the classes developed through this survey. We do not know whether people gave accurate reports of their lifespan PA levels. Retrospectively assessed PA biographies are not equivalent to actual PA across the lifespan. The results may thus be considered to be an individual’s mental representation of PA over their lifespan. While retrospectivly reported PA is related to actual PA, it can be considered a psychological concept in its own right. Our sample was relatively active and well educated, which limits the generalizability of the findings. Future studies should replicate the class structure using a more representative sample. We were not able to link the trajectories to common life events such as transition from primary to secondary school, transition from high school to college or university, marriage, becoming a parent, or retirement [23, 24]. Nor could we link the trajectories to non-normative critical life events such as the death of a spouse or the onset of a chronic disease such as rheumatoid arthritis [25]. These elements should be investigated in future studies.

While our findings have shown that specific PA biography patterns relate with life and aging satisfaction in older adults, future studies should investigate whether this pattern can be replicated in different settings and subgroups of older adults. Moreover, the relationship of PA biographies with indicators of physical functioning, including specific syndromes such as frailty, would be of interest for future studies. This was not included in our study, as we focused our investigation on aging satisfaction only. Finally, more research is needed that applies tailored interventional designs. It could be tested whether certain PA biographies (e.g., consistently inactive versus decreasingly active) are more responsive to information, which are framed to their PA background (e.g., cues that describe activity as pleasure or highlight the functional value versus cues that activating past habits and benefits of an active lifestyle [26]).

Implications for policy and practice include the importance of childhood fitness and PA for the entire life [20]. Although PA patterns across life appear to be stable for many individuals, tailored interventions that consider unique PA biographies may improve the effectiveness of PA promotion interventions for older adults [27‐29]. Negative views on aging and aging satisfaction in individuals with low levels of PA across the lifespan may be target of interventions on an individual level as well as in pubic campaigns [10]. Results from a PA intervention study in older adults have shown that positive views on aging could be changed by interventions, which, in turn, result in changes in PA in that sample [30]. Self-reported PA biographies contribute valuable insight and enable researchers and health care professionals to gain a greater understanding of PA in older adults.