The aims of this controlled pretest-posttest study were twofold: 1) examine the effects of a lifestyle program on device-measured moderate-to-vigorous physical activity and objective and subjective sleep (i.e., objective sleep outcomes: sleep efficiency, total sleep time, WASO and number of awakenings and subjective sleep quality (PSQI)) in adults aged 55+ years; and 2) examine if the effects differed between good and poor sleepers. In what follows, we will discuss possible reasons for the fact that although the program successfully increased MVPA levels, only few significant and clinically relevant findings were observed for sleep.
Effects of ‘Lekker Actief’ on MVPA and sleep
The program was effective in increasing MVPA levels with 43 min per day in the fully adjusted models. This is not only statistically significant, but also a relevant increase. Recent PA guidelines for older adults indicate that a minimum level of 150 min MVPA per week is needed for health benefits [29, 30]. If we translate this to a daily level to put the effects of ‘Lekker Actief’ into perspective, this means that the recommendation of 21 min of MVPA (150 min/7 days) is almost achieved twice after participation in this program. Despite these significant increased MVPA levels, there were no significant effects of ‘Lekker Actief’ on sleep. Thus, even though more controlled studies, as described in a recent review, previously established positive effects of PA on WASO, sleep quality, sleep latency and sleep disturbances , these findings could not be confirmed in our intervention study in a real-life setting.
Given the growing life expectancy and the increased prevalence of sleep problems at higher age, it is crucial to detect which programs could benefit sleep for good and poor sleepers separately. Only two interactions between program and sleep condition were significant: one when predicting wake after sleep onset and one when predicting the number of awakenings. Yet, these interactions were only significant in the crude models and not in the partially or fully adjusted model. Furthermore, the simple slopes where either not significant (regarding wake after sleep onset), or clinically not relevant. In general, we can therefore conclude that the intervention effect did not differ between good and poor sleepers. Interestingly, in contrast to our study, previous research reported a positive effect of regular exercise on sleep in older adults that were considered poor sleepers [27, 28]. More specifically, a study by Yang et al. (2012) reviewed studies that examined the benefits of exercise training programs for sleep in middle aged and older adults with sleep problems measured by PSQI and polysomnography . These authors concluded that sleep latency, sleep medication and subjective sleep quality were improved after following a moderate to high intensity exercise training program. A more recent review by Lowe et al. (2019) examined the effects of daily exercise sessions on sleep in adults with insomnia . This review showed that after following an exercise intervention, subjective sleep quality, sleep latency and sleep efficiency was improved, regardless of the exercise intensity. Even though these reviews focused on exercise programs, and not on MVPA as was the case in our study, our finding that ‘Lekker Actief’ did not affect sleep outcomes is in contrast with the conclusion by Lowe et al. (2019). Potential reasons for the discrepancy in findings between those studies and the present study will be discussed below.
With respect to good sleepers, Vanderlinden et al. (2020) reviewed the effects of PA programs on sleep in older adults without sleep problems and concluded that quality of sleep, sleep latency, sleep disturbances, WASO, sleep duration and sleep efficiency were improved after following a PA program in more controlled settings . Our present intervention study did not confirm these previously found positive effects for these sleep outcomes.
Possible reasons for the few significant and clinically limited findings in terms of sleep
This study found only few statistically significant and clinically relevant results in terms of sleep. Three possible reasons that we will discuss below include: the real-life context of the ‘Lekker Actief’ program, the use of accelerometry to measure MVPA and sleep and the 24 h continuum.
The real-life context of the ‘Lekker Actief’ program
The real-life context of this program represents a natural setting, which might have caused a dilution of the effects on sleep outcomes that were found in more controlled studies. This explanation is in line with the conclusions of a previously published review on the translation of program effectiveness to real-life programs in diabetes type 2 patients . The authors of this review identified three key factors that might explain the difference between results from more controlled, lab-based studies and real-life studies: (1) participation levels in the program, (2) intensity and frequency of the program (3); implementation fidelity.
First, in terms of the participation levels in our study, it was not possible for us to register the session attendance of participants because of two reasons. First, OKRA SPORT+, who organized the program, wanted to keep the participation burden for the local organizers as low as possible. Second, given the self-determination theory, which formed the theoretical backbone of this program, controlling for sessions attendance might have a negative effect on the self-perceived autonomy of the participants. Consequently, we were not able to control for program adherence in our analyses. It may have been the case that there would be an effect on sleep in people with higher participation rates, as has previously been shown in PA intervention studies [56, 57]. Second, with regard to the intensity, frequency and type of exercise, the ‘Lekker Actief’ program was developed and organized by the socio-cultural organization (OKRA SPORT+). Therefore, we did not have a possibility to change or control the content of the program itself. The different components in this real-life lifestyle program (walking, muscle strengthening and healthy nutrition) were not offered in each meeting point in the same standardised way or dosage which could have impacted the effects on MVPA or sleep. In addition, the 12-week duration of the program might have been too short to establish more and larger effects on sleep, considering that PA programs with a duration of six months and more have shown larger effects on sleep outcomes . In addition to this second key factor, the attained frequency (session attendance) and intensity (compliance with MVPA) of the program was not monitored by use of exertion scales or heart rate measurement during the program. Monitoring intensity may have provided participants with feedback and the possibility to adjust their intensity levels if they did low intensity rather than MVPA. We cannot exclude the possibility that the dose of MVPA might not have been high enough to replicate the positive effects on sleep observed in more strictly controlled studies. Furthermore, ‘Lekker Actief’ was intended to increase MVPA levels and this was communicated to the participants. The lack of significant positive effects on sleep could also be based upon the fact that our participants were not focused on improving sleep during the program, as was the case in other studies that did show significant and positive effects on sleep .
Third, in terms of implementation fidelity, no quality assurance was implemented in our study to assess the extent to which the program was delivered in real-life as originally intended by the organization OKRA SPORT+. In order to successfully facilitate translation of established effects from more controlled lab-based studies in to community settings, Miller et al. (2012) emphasized that program fidelity, staff and organizational capacity and engagement as well as a program evaluation constitute important steps in this translation process . Future organisers of real-life lifestyle programs are encouraged to focus on these criteria.
The use of accelerometry to measure MVPA and sleep
Another potential reason for the clinically limited findings is the use of accelerometry, more specifically Actigraphs, to objectively measure sleep. Although these devices are shown to be reliable in measuring sleep in this population group, polysomnography could have provided more in-depth data in terms of sleep stages and might also have been more sensitive to detect smaller changes in sleep compared with the Actigraph. These actigraphs are less expensive and easier to use when compared with polysomnography, which makes this method more realistic to use when assessing sleep in larger samples . Furthermore, polysomnography is performed in a clinical or controlled sleep lab setting, which does not represent a natural sleep environment compared with accelerometry measurement, which can be performed in the participants’ own bed (room). Finally, the placement of the accelerometer (i.e. wrist-worn) could have affected the effects in this study . It is known that hip-worn accelerometers reproduce much smaller acceleration values during walking than wrist-worn accelerometers which in turn could result in a misclassification of LPA as MVPA. Indeed, a recent study by Bammann et al. (2021) in which older participants wore Actigraph GT3x + accelerometers on the ankle, hip and wrist, concluded that there was a higher inter-individual variability of arm movement (wrist-worn accelerometer) compared with core body (hip or ankle-worn accelerometer) during activity . In this study, we used this wrist-worn placement for measures of both MVPA levels and sleep. In order to identify MVPA levels and avoid misclassification of LPA as MVPA, we made use of a tri-axial vector magnitude (VM3) cut-point (≥3268 CPM) for wrist-worn actigraphy in older adults .
Although wrist-worn accelerometry is preferred above waist-worn accelerometry when assessing sleep [60, 61], waist-worn accelerometry is preferred for measuring PA [62, 63]. Previous studies also showed valid results when using wrist-worn accelerometers for measuring physical activity [21, 36, 41].
Time and type of PA and the circadian rhythm
Previous studies in more controlled settings have already established positive effects of PA programs on sleep in older adults. However, as already discussed in the review by Vanderlinden et al. (2020), it remains unclear how timing and type of exercise might impact sleep .
Although previous studies in more controlled settings have already established positive effects of PA programs on sleep in older adults, it remains unclear how timing and type of exercise might impact sleep . Exercising at different times throughout the circadian rhythm could result in different effects on sleep. With regard to time of PA, previous research has already shown that exposure to bright daylight improves sleep [19, 20]. Furthermore, vigorous intensity physical activity prior to bedtime could worsen sleep outcomes as it would increase bodily temperature, which in turn is not conducive for sleep [64, 65]. Regarding the type of PA, different types of PA (i.e. aerobic exercise versus resistance training) might go along with different effects on sleep, although evidence on the ideal type of exercise for sleep in older adults is still lacking . Finally, individual characteristics of participants (i.e. circadian rhythm, chronotypes) could also have impacted the effects of the program on sleep [19, 20, 67]. Although these factors may have affected the effects of PA on sleep, we did not control for participants that performed other types of exercise on top of the components of this real-life program, nor did we control for the exact time of the day when participants were physically active, or collected data about their actual chronotype.
Strengths and limitations
This study has several strengths and limitations. Strengths include the focus on a lifestyle program in a real-life setting, which favours the ecological validity of the findings; an intervention in a large sample size of older adults and; the device-based measurement of MVPA and the availability of subjective and objective sleep data. There are also several limitations. First, this study was a non-randomised controlled trial, which may have caused bias. Despite this lack of randomisation, we did match participants of the intervention group with the control groups in terms of regions and the number of participants per meeting points. Second, we are not sure that the dose of PA in this program was delivered as initially intended, given that we were not able to register and monitor the session attendance and compliance with intensity of physical activity during the program. Consequently, we could not control for these indicators in the analysis in this study, which might have caused an underestimation of the effects and associations. Third, although we used several objective and subjective sleep outcomes, we performed several analyses in different models. Therefore, the possibility of a type 1 error cannot be excluded. Based upon a set α 0.05, there is a 5% possibility that significant findings might be based on coincidence rather than on significance. Finally, the use of wrist-worn devices is a limitation as this wear-location might lead to an overestimation MVPA time, which is reflected in high MVPA levels in both conditions at baseline [40, 59].
Generalizability and implications
As the lifestyle program ‘Lekker Actief’ was not intended to improve sleep, its components (i.e., walking program, muscle strengthening program and healthy nutrition) were mainly focused on increasing MVPA levels and overall health. Based on findings from more controlled studies examining the effects of PA and exercise programs on sleep, one of the aims of this study was to explore whether real-life lifestyle programs that offer MVPA could also be used to promote sleep in older adults. Only participating meeting points in the real-life lifestyle program ‘Lekker Actief’ were eligible for the inclusion in the intervention group. This selective sampling strategy could have caused selection bias. Finally, when comparing baseline measures of MVPA levels and sleep between conditions and with the general population, it became clear that the older adults in this study were generally more active when compared with the Belgian population of the same age [7, 68]. Moreover, the older adults in the intervention group were significantly more active at baseline when compared with the control group. Therefore, the generalizability of our data to more inactive older adults can be questioned. It could be possible that in older adults with lower MVPA baseline levels, increases in MVPA could result in improved sleep as has been observed in previous studies.
It is clear from this study that the findings from previous more controlled studies do not equally translate to real-life studies like ‘Lekker Actief’ with regard to positive effects on sleep. In order to facilitate this translation to a real-life setting, future lifestyle programs for older adults should include strategies to monitor session attendance and compliance with the intended program intensity of each program session to ensure that the intended dose of MVPA is achieved. Although there are indications for a different effect of physical activity on good and poor sleepers, this was not confirmed in the present study. It is recommended that, in order to increase focus on improving sleep in these lifestyle programs, a sleep enhancing component (i.e. sleep hygiene advice, cognitive behavioural therapy or relaxation), preferably provided by a licensed sleep therapist, might increase the overall impact on sleep [69,70,71,72].