This study used a prospective cohort design to investigate individual and environmental/organisational factors as predictors of attendance to group exercise offered to an older population resident in LTC. We used an efficient nested cohort study design to select adults aged 75 to 107 years, participating in a cluster-RCT of a physical activation programme which included group exercise versus a control intervention incorporating depression awareness training for care home staff. This is the first large study of exercise adherence within residential and nursing homes in the UK.
Attendance
Overall percentage attendance to the exercise groups in the residential and nursing homes was just over 50% in both settings. For the most part this is slightly lower than in previous trials where attendance rates to group exercise in LTC has ranged from 42.5% to 100% [
28]. In the process evaluation of the OPERA study, Ellard et al. [
33] conclude that this was a low level of attendance, particularly amongst those who were depressed, partly due to issues such as frailty, cognitive impairment and lack of staff time to assist in getting residents to the groups. Thus, overall exposure to adequate intensity of exercise was low, subsequently resulting in an overall negative effect of the intervention on depression [
21].
However, for residents attending regularly, continued attendance may have been due to the physiological and psychological beneficial effects of exercise [
28]. Potentially, participants profiting from exercise or even perceiving that they are benefitting will have attended more groups, compared to those perceiving no benefit. Plausibly, therefore, although short term, these improvements in physical and psychological well-being and feelings of enjoyment and achievement linked to self-efficacy and mastery may have encouraged attendance [
15,
16].
It is also argued that living in LTC can be a passive experience, typified by increasing dependency and lack of control. However, exercise interventions similar to that used in OPERA have been shown successfully to give participants the opportunity to regain some control and sense of self-worth [
10] and therefore may have been another possible explanation for participant attendance. The staff and residents in the intervention homes were very positive about the exercise groups and during observations undertaken during the process evaluation, the exercise intervention did appear to make a short term difference to residents’ mood and physical abilities [
33].
Predictors of attendance
Of the nine individual and home level characteristics variables investigated in this study, six were associated with group attendance in univariate analysis in the residential home population, four having a positive effect, and two, a negative effect. However, after adjustment, only two variables were associated with attendance in the residential homes. None of the predictor variables were significantly associated in either univariate or multivariate analysis with attendance in the nursing homes.
Despite health status/physical limitations including pain, fear of falling and number of chronic conditions being common barriers and motivators to exercise and physical activity [
6,
10-
12], number of co-morbidities as a measure of health status was not significantly predictive of attendance in LTC. Potentially this was because all residents had co-morbidities and there was lack of variation amongst this age group. However, this predictor is important, because health issues have consistently been shown to be a common barrier and motivator to attendance at exercise and number of illnesses, although a crude indicator on its own, may be a significant indicator of frailty. But perhaps frailty as a distinct health state is crucial, and needs to be considered as more than just number of health problems.
In our study, a large proportion of our sample would have been classified as ‘frail older adults’ based on their SPPB results for lower limb function, i.e. median baseline scores and 12 month follow up scores were zero in nursing homes and one in residential homes. Frailty, including symptoms of muscle weakness, slow walking speed and low physical activity [
34] can result in non-compliance and low participation in physical activity and structured exercise programmes. This is common amongst older people and even more so those resident in LTC [
8,
10,
35].
Despite this, at initial assessment, lower limb function was not significantly associated with attendance at the exercise groups. However, it could be argued that the SPPB was an inappropriate measure to use in this population, as it was originally designed for community dwelling adults and lacks sensitivity in this particular group. This is highlighted by the obvious floor effect in this current sample [
21].
Thirdly, fear of falling was not associated with attendance at the exercise groups. Numbers of participants reporting fear of falling was lower than anticipated, possibly because of the lack of activity being undertaken by some residents i.e. they were not fearful of falling because they were not active to the extent that falling might be a risk or concern. Instead, and in line with the idea of passivity, participants were moved around the homes and often brought to groups in wheelchairs [
10]. Interestingly, fear of falling declined over time. Alongside this, residents may not have been fearful because they know that carers are nearby should they need help, compared to older people living alone in the community who may not have anyone close by to help them if they do fall.
Although data shows that two in every five older people living in care are depressed [
36], the negative effect of elevated depression scores (suggesting more depressive symptoms) were only predictive of attendance to the exercise groups in the residential home sample. This is reiterated by studies reporting that depressed older people are less likely to attend or participate in exercise [
33,
35]. An implication of this is that those engaged in the life of the home would be more likely to join in with group exercise. However, exercise groups do not necessarily attract everybody, and an alternative approach to engage those with depression in exercise may be beneficial. Additionally, early detection, diagnosis and treatment of depression in this population is vital.
Although the depression scores of the nursing home residents were indicative of more depressive symptoms, depression was not significantly associated with attendance at the exercise groups in the nursing homes. These results may have been due to the smaller sample size within the nursing home population or may be due to selection bias. For example, GDS scores were skewed due to the inability of some residents to complete all 15 GDS test questions due to cognitive impairment; therefore, a smaller number or lower proportion (n = 87) of actual valid GDS scores were available for analysis. This again implies that the GDS was perhaps not the most appropriate measure to capture depression in this vulnerable population [
37].
Social engagement and support from others e.g. initiating interaction with other residents and pursuing involvement in the life of the facility can be related to self-efficacy or the residents’ beliefs in their own ability to complete tasks and achieve goals. As in this study, socialising, interaction and support from other residents, carers and family have previously been found to have a positive effect on attendance to physical activity in community dwelling older adults and those living in low level residential care [
12,
29,
38].
However, the two social engagement measures were not found to predict attendance at group exercise in the nursing homes. The MDS tool was designed to be used as a direct observational tool [
21] but in this instance it was used as a proxy measure based upon staff judgement rather than actual observation of social involvement or extent of engagement with others. However, due to the large sample size, it was not possible to observe individual study participants. Staff from homes were asked to complete data collection forms on social engagement for all participants, but this process was often rushed, due to time constraints and staffing issues. However, there was no evidence to suggest differential error (measurement bias) between nursing and residential homes, but we cannot exclude the possibility that the data may not have been always entirely accurate.
Socioeconomic status underlies major determinants of health including health behaviour. Measured by income, education, or occupation, SES has been associated with a wide range of health conditions, including cardiovascular disease, arthritis and diabetes. This is evident in economically deprived groups where engagement with activity interventions and exercise and physical activity levels are consistently lower [
32]. Residents’ SES was a significant predictor of attendance to group exercise but only in the univariate analysis of the residential home sample. This might be explained by the differences in funding; the proportion of fully self-funded participants and therefore, potentially less economically deprived, was double in residential homes (52%) compared to that in the nursing homes (24%).
Similarly, the presence of an activity co-ordinator was only a significant predictor of attendance to group exercise in the residential home sample. Ellard et al. [
33] considered the negative effect of the exercise intervention in terms of whether the intervention changed the culture of physical activity within homes. They report that whilst some activity co-ordinators were observed just performing their own role, many were also assigned other tasks e.g. personal care, helping with meals etc. therefore taking them away from their primary activity role. Thus it is possible that homes with no activity co-ordinator or those spending less time involved in activities may have had lower levels of attendance as the residents were less used to activities of any kind, let alone physical activity. Alternatively, the influence of a motivated, enthusiastic activity co-ordinator may have impacted positively on attendance [
39].
The differences in results between the residential and nursing homes should be interpreted with caution. Despite the overall large sample size, the number of residents completing the intervention in the nine nursing homes was relatively small (n = 72) resulting in lower statistical power in this group. Therefore, the lack of any significant results from the nursing home data may be due to the size of the sample. Additionally, the low R
2 value in the multivariate analysis indicates that the model may not be the best fit for the data. However, as in this instance, lower R
2 values are often expected when predicting human behaviour and conclusions can still be made from the significant coefficients representing changes in the predictor values and their association with changes in the response values [
40].
It is recognised that the older people participating in the study were from a target population, in selected geographical areas and those who gave consent were motivated to volunteer for a 12 month clinical trial involving exercise. For those whom assent was needed, family members may have given agreement for their relatives to participate when it was not particularly appropriate or vice versa.
Attrition rate, perhaps not surprisingly, was a limitation of the study. As expected, the mortality rate in this older population was high, with a total of 11% crude death rate over the 12 months. Additionally, although taken into consideration in the denominator, other uncontrollable reasons for non-attendance i.e. illness, transfers in and out of homes, and ineligibility for attendance at group exercise, resulted in over 6000 potential group attendances being lost. Therefore, this will affect percentage attendance figures and data that may have been incorporated in this analysis.
The primary aim of the trial was to investigate depression rather than predictors of attendance to group exercise, therefore other variables that may have contributed as predictors were not available for inclusion in this analysis i.e. previous exercise/physical activity levels or experience. Analysing the data by type of accommodation may have also had an impact on the results; this approach was taken because we anticipated a difference in resident characteristics within the homes. Therefore further work is needed to examine other factors, including type of accommodation, which may influence attendance to group exercise in this population.
Nevertheless, our findings are based upon a large sample of older adults up to the age of 107 years, incorporating high quality data from validated, standardised measures collected within the framework of a rigorously conducted, high quality trial. Despite the practical and methodological challenges of conducting clinical research with a very frail elderly population, a very large sample of participants from care homes across Coventry, Warwickshire and London were recruited. Our study provides good quality research findings on a vulnerable population resident in LTC in the UK.