Functional fitness and psychological well-being in older adults – BMC Geriatrics

Participants

We sought permission from the management of five nursing homes randomly selected.^{Footnote 1} on Google Maps within Fejér county in Hungary to conduct the study with their inhabitants who are fit for the research and wish to volunteer. Within two weeks, we received a response from two, one in Aba and one in Székesfehérvár. After obtaining permission, we visited the nursing home and verbally presented the research and the tasks participants must complete. Volunteers signed an informed consent form and a General Data Protection Regulation (GDPR) data handling form. In total, 39 older adults have completed the study. They were all 60 or over (M_age = 80.15 years; ± SD = 7.21 years, range 60 to 94 years), and two-thirds (n = 26) were women. Their mean weight was 78.00 ± 12.87 kg, height 1.70 ± 0.12 m, and BMI 27.11 ± 3.38. The institution’s doctor or head nurse screened the volunteers for health, considering the mental and physical conditions of the participants.

The physical criteria for exclusion were the inability to stand or walk, upper or lower limb deficits, medical conditions predisposing the person to dizziness, cardiac risk factors, loss of balance, and untreated hypertension. The psychological criteria for exclusion consisted of previously diagnosed mental and behavioral disorders. All participants had medical clearance for the study. Table 1 presents their demographic characteristics.

Ethics

The study was conducted in December 2022 with approval (permission No. 2022/510) from the Faculty of Education and Psychology Research Ethics Board at Eötvös Loránd University. The work conformed to the ethical guidelines of the British Psychological Society (BPS) Code of Human Research Ethics [21]. Additionally, the protocol followed the research principles with human participants of the Helsinki Declaration [22].

Measures

The dependent measures comprised subjective responses collected with questionnaires and objective functional fitness assessments.

Subjective measures

Apart from demographic questions (see Table 1), mental well-being (MWB) was measured using the Mental Health Continuum-Short Form [23, 24]. The 25-item Connor-Davidson Resilience Scale [25, 26] assessed resilience. Optimism was gauged with the Life Orientation Test (LOT-R) [27, 28], while SWL was measured with the Satisfaction with Life Scale [29, 30]. Finally, the perceived happiness was estimated with the Subjective Happiness Scale [31, 32]. All instruments have good psychometric properties (see their sources) that are not described here for parsimony.

Objective measures

The FFFT evaluates older adults’ functional fitness by assessing upper and lower body strength, upper and lower body flexibility, complex coordination, and endurance. The test consists of tasks, such as walking, lifting weights, reaching, and balancing, which are essential in daily life activities. Research evidence based on measures of stability, reliability, and discriminant validity supports the validity of the FFFT [14]. In this study, we used the FFFT to assess functional fitness [15, 16, 33] through six measures:

1. 1.

   Lower body strength(FL1): 30s chair test, complete stand up and sit down (number of repetitions).

2. 2.

   Upper body strength(FL2): lifting 2 (women) or 3.5 (men) kg dumbbell while sitting on a chair and doing complete arm bends and stretches (number of repetitions in 30 Sect. ).

3. 3.

   Upper body flexibility(FL3): fingers touching behind the back (back scratch) (+/- cm).

4. 4.

   Lower body flexibility(FL4): forward bend from chair to extended leg (chair sit-and-reach) (+/- cm).

5. 5.

   Complex coordination (agility, balance, and walking speed)(FL5) – circling a cone, which is located 2.44 m (8 feet) from the starting position, in the shortest possible time and then returns to the starting position.

6. 6.

   Endurance (physical effort(FL6): 2-minute step test – records the number of whole steps completed in two minutes, raising each knee to the point halfway between the patella (knee-cap) and iliac crest (top hip bone).

The FFFT is safe for both inactive and physically active older adults. Moreover, by using everyday motor patterns, researchers can get an insight into the six functional fitness indices described above [34]. Finally, we calculated BMI by dividing participants’ weight (kg) by height (m) squared.

Procedure

Data collection occurred individually in a quiet room in the participants’ habitual environment. While completing the questionnaires, a researcher was present, but she did not interact with the participants unless they had questions about their tasks. After the participants completed the questionnaires presented in shuffled (random) order and answered the demographic questions, the researcher explained the FFFT before each of the six trials. Next, she demonstrated the correct execution of the upcoming task. Each FFFT trial was performed twice and the better performance on the two trials was recorded. After completing twice each of the six FFFT trials, the participants were debriefed and thanked for participating.

Data analyses

After testing the normal distribution of the results, we examined possible gender differences in the dependent measures using the Mann-Whitney U tests. Subsequently, we calculated Pearson’s correlations between the observed variables. Regression analyses were conducted in R programming language [35], using ‘tidyverse’ [36], ‘glment‘ [37], ‘glmnetUtils’ [38], and ‘relaimpo‘ [39] packages. Five regressions examined the relative importance of FFFT and BMI scores in predicting five psychological measures. Since there was a strong multicollinearity present in each model, it was impossible to distinguish between each predictor’s importance by looking at their coefficients and p-values. For example, the R² was high in all instances, but the individual predictors were insignificant.

Therefore, to unveil the relative importance of each FFFT component and BMI, we ran a series of regularized regressions (elastic net) in addition to ordinary least squares (OLS) regressions [40]. In an elastic net (the combination of ridge and lasso regressions), multicollinearity usually does not present a problem. We can see the predictors’ importance by combining this method with the LMG relative importance metric (see [41]). Although coefficients from regularized regression cannot be easily interpreted as OLS coefficients, they can help identify essential contributors and the direction in which they relate to the outcome measures. Five-fold cross-validation was used to determine optimal values for the hyperparameters alpha and lambda (see [40]). However, for SWL and happiness, the alpha was increased from 0.0 to 0.1 to avoid fitting a complete ridge regression model.

The LMG metric is independent of the order of predictors in the model. It can represent a relative proportion (summing to 1) or relate to the model’s R², with LMG values adding up to the model’s R². In contrast to regression coefficients, which can be distorted and yield reversed signs in multicollinearity, the importance measures derived from LMG are always positive and provide a more suitable decomposition of the model’s R² than standardized regression coefficients. Further, multicollinearity can inflate p-values, indicating nonsignificant predictors, while the overall model’s F-test remains significant [41, 42]. Therefore, we utilized regularized regression models, such as elastic net, to effectively identify the most influential predictors in multicollinear contexts, supplementing our analysis and avoiding misleading results from ordinary least squares (OLS).