Study design and participants
This cross-sectional study was a secondary analysis based on data obtained from recruitment for a randomized controlled trial (RCT) comparing two supervised physical exercise interventions of different lengths of the supervised part. (16). The protocol was registered retrospectively under the Australian and New Zealand Clinical Trials Registry with the identifier ACTRN12619000093189 (date of registration: 22/01/2019). The Clinical Research Ethics Committee University Hospital of Araba (2017-021) approved the study protocol, which complied with the revised ethical guidelines of the Declaration of Helsinki (2013 revision). All participants provided informed written consent before enrollment in the study.
RCT enrollment data were obtained from September 2017 to July 2018 in the Departments of Internal Medicine and Neurology at the Santiago University Hospital of Araba (Basque Country, Spain). Both departments have a high prevalence of older people who usually experience rapid deconditioning due to hospitalization. Eligible participants included men and women ≥70 years who scored ≥20 on the Mini-Mental State Examination (MMSE) (17) and were able to stand and walk independently or with assistance for at least 4 m. The MMSE is a 30-point test used in clinical settings to measure cognitive impairment and to screen for dementia. The cut-off point was set at ≥20 to ensure patients could follow the instructions of the physical exercise program. Exclusion criteria were a diagnosis of chronic kidney disease, autoimmune neuromuscular disease, acute myocardial infarction, or bone fracture in the past three months.
During the recruitment period, 2,365 non-surgical patients were admitted to the Departments of Internal Medicine and Neurology of University Hospital of Araba, a tertiary teaching hospital in the Basque Country. After screening medical histories and inclusion criteria, 509 patients were eligible to initiate the physical exercise program at discharge. After signing an informed consent document, a comprehensive geriatric assessment was performed by nurses and physicians of the hospital staff and physiotherapists of the research team. Once assessment of each patient was completed, they were given the option of starting a physical exercise program in the same hospital at discharge. Once per week, we also provided informative sessions and distributed informational leaflets throughout the hospital. Of the evaluated patients, 55 agreed to participate and started the physical exercise program, and 454 refused to participate. Those rejecting participation (n=14) in the program because they were active enough on their own were excluded from analysis. Finally, 495 patients were analyzed (Figure 1).
Characteristics of the physical exercise program
The characteristics of the program were described previously (16). Briefly, subjects who agreed to take part in the intervention were randomly assigned to a short (6 weeks) or long (12 weeks) supervised exercise program. In both groups, participants continued the program at home from either 6 weeks or 12 weeks until 24 weeks after the start of the intervention. The program consisted of 1-hour group sessions (2 days/week in the supervised part and 5 days/week at home) and included strength, balance and walking training. All sessions began with warm-up exercises (5 min) and continued with strength training of upper and lower limbs (35 min) tailored to the individual’s functional capacity. Balance training exercises (15 min) were also practiced, progressing in difficulty during the program: decreasing arm and base of support and increasing the complexity of movements. Finally, participants were recommended to perform walking sessions on their own. The first participant started the exercise intervention in November 2017 and the last participants ended the program in January 2019.
Figure 1. Study flow diagram.
We asked patients who refused participation in the exercise program to provide reasons for not participating. We categorized refusal reasons as internal reasons, external reasons, and those pertaining to a lack of interest in physical exercise. Internal reasons related to the patients themselves reporting poor health. External reasons related to the patients’ social burdens (travel problems, being a family caregiver, or social problems). Uninterested patients simply had no interest in the physical exercise program.
Measurements
Sociodemographic and clinical data were retrieved from the Basque Public Health Service’s database. The following data were collected: patient demographic data (sex and age), days of hospitalization, hospitalizations in the previous year, emergency care admissions, and comorbidity (assessed through the Charlson comorbidity index, a method of categorizing patient comorbidities based on the International Classification of Diseases) (18). We collected the Barthel Index score, which measures performance in basic activities of daily living such as continence and mobility (19). In addition, we collected the Lawton Index score which measures the instrumental activities of daily living necessary for independence in the community (20). Finally, we collected socio-demographic data, including patient educational level, whether or not they live alone, and whether they use assistive devices for walking and home accessibility, such as home entrance-related assistance, e.g. an elevator or lift (21).
Physical function was measured individually in patients’ hospital rooms by the Short Physical Performance Battery (SPPB) (22). SPPB is a battery of tests that combines assessment of balance, gait speed, and lower limb strength. The balance test measured ability to stand for 10 sec in side-by-side, semi-tandem, and tandem stands. To assess gait speed, we measured the time to walk 4 m at the patients’ usual speed twice (participants could use a walking aid if necessary). The best walking time of the two measurements was scored. Finally, lower limb strength was measured by time to perform five repeated chair stands, and the sit-to-stand speed was determined. Each test was scored on a scale of 0–4 points, with a total performance score range of 0–12 points using cut-point criteria established by Guralnik et al. (22). Higher scores indicate better physical function.
Nutritional status was measured using the calf circumference short form of the Mini Nutritional Assessment (MNA-SF) (23). The MNA-SF is a validated screening tool designed by Nestle to identify elderly persons at risk of malnutrition. The test consists of five questions (appetite or eating problems, recent weight loss, mobility impairment, acute illness/stress, dementia, or depression) and measurement of calf circumference (CC). CC was measured at the calf’s greatest circumference with the patient sitting down, resting their feet on the floor, and knees bent 90°. The test provides a maximum score of 14 points. Higher scores indicate better nutritional status.
Frailty was measured according to the Fried phenotype criteria (24). A Spanish language version of the frailty performance criteria was used to measure grip strength, walking speed, weight loss, physical activity, and exhaustion.
Cognitive function was measured by the Spanish validated version of the Short Portable Mental Status Questionnaire (SPMSQ) (25). The SPMSQ includes ten questions to briefly test short- and long-term memory, orientation, and capacity to perform serial mathematical tasks. The total number of errors was counted and one point was subtracted if the patient had a grade school education and one point was added if the patient had a high school education.
Statistical analyses
Continuous variables were expressed as means with standard deviations (SD), and categorical variables were expressed as frequency counts and percentages (%). Sociodemographic characteristics and clinical data were compared between patients who initiated or declined participation in the post-hospitalization physical activity program. Normality of data was assessed using the Kolmogorov-Smirnov test. We used an appropriate statistical test according to the type and distribution of the data: Student’s t test (normal distribution data) or Mann-Whitney U test (non-normal distribution data) for continuous variables, and a chi-squared test for categorical variables considering a P < 0.05 to be statistically significant. Variables with P < 0.05 in univariate analysis were considered eligible for a backward multivariate logistic regression model of participation in the program. The Hosmer-Lemershow test was used to determine the goodness-of-fit of the model and determine whether the observed event rate matched the expected one. A number closer to 1 indicates better goodness-of-fit. Omnibus was used to test whether the explained variance was significantly greater than the unexplained variance; a P value < 0.05 was considered statistically significant. Nagelkerke’s R2 value estimated the proportion of the dependent variable explained by the independent variables. Statistical analysis was performed using SPSS v.21 software (IBM, Chicago, IL).