Table 1

Description of the Fried frailty criteria and other exclusion criteria

Recruitment

To recruit frail community-dwelling older adults, the TIBaR model of recruitment for ‘hard-to-reach’ older people for research will be adopted36 (see table 2). This model consists of four basic principles: providing trustworthy companions (T), offering meaningful experiences or incentives (I), removing barriers (B) and being responsive to the needs of the older adults (R). Flemish older adults will be recruited through local newspapers, pharmacies, general practitioners, (home) nurses, occupational therapists, informal caregivers and local government. Flyers and e-mails will be distributed, but personal contact with local professionals and the target population is preferred.

Screening and enrolment

After registration, the blinded assessor contacts the older adult to make an appointment for the eligibility screening and, if eligible, baseline assessment at the home of the participant. Before the screening, the ICF is explained. If older adults agree and written consent is given, eligibility screening can start. Older adults who are eligible will be randomised within 1 week of baseline assessment, and intervention starts within 2 weeks. Enrolment in the study and randomisation in three groups is visualised in figure 1.

Randomisation and blinding

A computer algorithm will be used to generate the random allocation sequence. Participants will be randomised in a 1:1:1 ratio, using permuted block randomisation with blocks of variable sizes (3-6-9) into one of the three parallel groups. Randomisation will be done centrally by an interactive web response system (IWRS). Once the participants are found eligible for the trial, they can be randomised via the IWRS by the principal investigator (PI) or qualified person to whom this study task has been delegated. This will generate a unique study randomisation number and the treatment arm label (A professionals, B volunteers, C control) to which the participant is randomised. An accountability log with the corresponding label will be kept by the PI or delegated person to ensure that allocation was successfully performed. Participants will be scheduled to receive their randomisation within 1 week of eligibility assessment.

The statistician will be blinded to group allocation. Outcome assessors will be blinded to the maximal extent possible. Participants will of course be aware of the intervention received. Regarding this, participants will be asked not to communicate with the assessors about the intervention received. The professional or volunteer supervisors will not be involved in the assessments, and vice versa.

Outcome measures

Outcomes were chosen in line with WHO’s vision on healthy ageing (functional ability and intrinsic capacity)43 44 and the geriatric ICF core set reflecting health-related problems in community-living older adults aged 75 years and older without dementia.33 Older adults will receive assessment of primary, secondary and explanatory measures at baseline (T0), after the intervention of 24 weeks (T1) and at 48 weeks (T2). A blinded assessor will administer four physical performance tests and nine questionnaires at T0, T1 and T2 (see table 3).

TCS

In the TCS test, participants are asked to stand upright from a seated position in a chair (height 43 cm) with their arms folded across their chest and return to a seated position as many times as possible within 30 s.45 The TCS measures lower body strength and relates to important ADL like walking up stairs and getting out of a chair or bed.45 46 The test is frequently used in older adults with fall risk or frailty and is a reliable and valid instrument (reliability: ICC 0.84–0.92, r=0.93).47 48

TMST

In the TMST, participants need to march in place as many times as possible within 2 min, lifting the knees to a benchmark on the wall placed at the height of the exact middle between the patella and the iliac crest. The score is the number of times the right knee passes the benchmark in 2 min.45 The test measures aerobic capacity and relates to ADL, eg, walking in and around the house, climbing stairs and shopping. The test is a valid and sensitive alternative to the 6 min walk test with a test-retest reliability coefficient of 0.90.45 49

TUG

The TUG test is measured by the number of seconds the participant needs to get up from a seated position, walk eight feet (2.44 m), turn and return to the seated position. The purpose of the TUG test is to assess functional mobility, more specifically agility and dynamic balance. These are important in tasks that require quick displacement such as getting on and off public transport or getting up to answer the phone.45 50 The TUG has been proven to be reliable (ICC ≥0.83)51 and valid instrument to test functional mobility and balance in frail older adults.50 52 53

Fatigue resistance grip strength test

In the fatigue resistance grip strength, the Martin Vigorimeter is used to measure muscle fatigability.54 The subject is seated with both feet on the ground, the shoulder in neutral position and the elbow flexed at 90°. The participant holds the large bulb of the Martin Vigorimeter in the dominant hand by enclosing it on one side with the thumb and on the other side with four fingers. The participant is instructed to push the large pear as hard as possible and to maintain this force as long as possible.55 A good to excellent inter- and intra-observer reliability is observed (ICC values ranging from 0.77 to 0.94). Hand grip strength is seen as an indicator of overall body strength for frail older adults and other populations.54–56 Lower body strength leads to adverse health outcomes, eg, falls, disability and a higher risk of frailty.57

MOCA

The MOCA is a global cognitive test developed to screen for mild cognitive impairment.58 It can also be used as a repeatable measure to evaluate treatment response. In our study, the MOCA will be used to evaluate the effect of the intervention on cognitive function. The total score for the MOCA ranges from 0 to 30 points assessing the following domains: visuo-constructional/executive function skills, verbal memory, attention/working memory, language/executive function, conceptual thinking and orientation.59 The MOCA is a reliable and valid instrument demonstrating adequate to excellent internal consistency and construct validity, an adequate to excellent test-retest reliability, and adequate to excellent inter-rater and intra-rater reliability in community-dwelling older adults.59 Feeney et al. observed a SE of measurement of 1.5 points, with estimates of minimum detectable change of four points at 95% level.60 This is supported by Kopeček et al., indicating that changes of at least four points are necessary to assume a reliable change.61

TMT

The Trail Making Test is a widely used neuropsychological test, originally part of the Army Individual Test battery62 and then included in the Halstead–Reitan Neuropsychological Battery.63 The TMT consists of two parts, namely, TMT-A and TMT-B. In TMT-A, the subject is asked to connect the encircled numbers (1 to 25) in consecutive order as quickly as possible. In TMT-B, encircled numbers and letters must be connected in a numerical and alphabetical order, but alternating between numbers and letters (1, A, 2, B, 3, C etc). The score represents the time to complete each part.64 The TMT-A score is associated with measures of visual scanning, graphomotor speed, and visuomotor processing speed, while TMT-B performance reflects more executive functions such as working memory and inhibition control.65 66 The TMT has an adequate test-retest reliability (r=0.70)67 and has also demonstrated adequate construct validity.63 66

RAVLT

RAVLT is a neuropsychological assessment designed by Rey68 to evaluate verbal episodic memory.68 69 A list of 15 words (LIST A) is presented orally, after which the participant is asked to recall as many words as possible. This procedure is repeated over five trials. Then, the participant listens to an interference list of 15 different words (LIST B) and is asked to recall as many words as possible from list B. Immediately after the interference list, the participant should recall the words from LIST A again. After a 20 min delay, the participant is once again asked to retrieve the words from list A.70 Outcome measures used are the total number of words retrieved in five trials (=sum trial 1–5), and the number of words recalled immediately and following the 20 min delay. Different studies have demonstrated the construct, criterion, convergent and divergent validity71–73 and reported a good internal consistency (Cronbach’s alpha=0.80).72 When alternate forms are used, the RAVLT is a reliable instrument for repeated neuropsychological assessment.74 The total score (= the sum of scores across the five acquisition trials) and the delayed recall score are the most reliable measures.70 74 75

MOS SF-36 / SF-36

Health-related QoL will be measured by the Medical Outcomes Study 36-item Short Form Health Survey (SF-36).76 The SF-36 questionnaire is widely used to monitor general population health status, evaluate the efficacy of interventions, monitor health status in patients with chronic disease and determine the relative burden of various diseases.77 Also, a strong inverse association between (pre)frailty and QoL in community-dwelling older people was detected by the SF-36.78 The SF-36 includes a multi-item scale that assesses eight health concepts: limitations in physical activities due to health problems (one item), limitations in social activities because of physical or emotional problems (two items), limitations in usual role activities because of physical health problems (four items), bodily pain (two items), general mental health due to psychological distress or low well-being (five items), limitations in usual role activities because of emotional problems (three items), vitality due to low energy and fatigue (four items) and lastly general health perceptions (five items). In addition to the scale scores, a Physical Component Score and a Mental Component Score can also be calculated. The SF-36 is suitable, reliable and valid for use in frail older adults.79–81 The SF-36 also allows the calculation of utilities (with 1=perfect health and 0=dead) allowing the calculation of Quality-Adjusted Life Years (QALYs). People with mental or physical problems and/or aged over 75 may need the assistance of an interviewer.82

DJGLS

The DJGLS is an appropriate scale for the assessment of loneliness among older adults.83 This scale consists of 11 items, six items are formulated negatively and five items are formulated positively. Participants can respond with five possible answers (‘yes!’, ‘yes’, ‘more or less’, ‘no’, ‘no!’). Cut points can be used to classify people into not lonely, moderately lonely or strongly lonely.84 The scale’s developers note that this scale can be used as a unidimensional measure for overall loneliness or can provide information about two subscales: the emotional and social loneliness.84 Most studies confirm this two-factor structure79 85; nevertheless, some studies conclude that the scale is unidimensional.86 87 The 11-item DJGLS proved to be a valid and reliable instrument for overall, emotional and social loneliness with Cronbach’s alpha of >0.8.79 84 88

EMAS

The EMAS is a 12-item questionnaire with statements about the meaningfulness of activities in one’s life.89 Each item can be scored with rarely (=1), sometimes (=2), usually (=3) and always (=4). The sum of the 12 items ranges from 12 to 48. People perceive their activities as low, moderate or high when they score respectively less than 29, between 29 and 41 or more than 41. SD for the EMAS in community-dwelling older adults are 36.4 (6.2). EMAS is a valid instrument, available in Dutch, with moderate reliability.90–92

SMAS-S

The SMAS-S is a questionnaire designed to measure six self-management abilities in older adults based on five dimensions of well-being specified in the social production function theory.93 94 These consist of the ability to ensure multifunctionality, maintain variety in resources, keep a positive frame of mind, invest in resources for longer-term benefits, self-efficacy and taking initiative.95 SMAS-S is a shorter version of the originally developed SMAS-30 and consists of 18 items.94–96 Having a shorter instrument makes it more feasible to assess self-management abilities in a broader number of people, especially among frail older adults.96 SMAS-S has a good internal consistency, all subscales are 0.63 or higher (Cronbach’s alpha) and total reliability of the SMAS-S is 0.90.95

National HIS

Every 5 years a national HIS is held in Belgium. This cross-sectional population survey includes questions on health status, lifestyle, use of health services and socio-demographic characteristics and is administered in a representative sample of the Belgian population.4 In the ACTIVE-AGE@home study, questions regarding activities of daily living (ADL) are adopted from the HIS.

The respondent is asked to indicate if they have difficulties performing daily activities and/or household tasks. The seven daily activities and seven household tasks are read out loud, answers are given by selecting a predefined four-point scale ranging from ‘no, no difficulties’, ‘yes, a few difficulties’, ‘yes, a lot of difficulties’ or ‘not possible to perform’. When a question is not applicable, ‘does not apply’ can be noted as well.

ADL are getting in and out of bed, sitting down or standing up from a chair, getting (un)dressed, taking a bath or shower, washing your face or hands, eating with knife and fork or using the toilet. Household tasks are preparing meals, using the telephone, doing groceries, taking medication, doing light chores, doing heavy chores and handling money.

iMTA-MCQ

To measure medical costs experienced by frail older adults, the Dutch version of the iMTA-MCQ will be used. The iMTA-MCQ is not related to a specific disease and measures medical consumption through 36 questions covering the past 3 month period. Formal (physician, dietician, …) and informal (family, …) care services are registered as well as out-of-pocket expenses.97 For the purpose of this study, one question was excluded from the iMTA-MCQ regarding the consultations with an occupational physician. Despite the widespread use of the iMTA-MCQ in older adults98–102 and other populations,103–106 the validity and reliability of the iMTA-MCQ still need to be established.97 104

During the intervention, volunteers, professionals and participants are also required to fill in a log after each training session to signal adverse events and measure adherence, motivation and fidelity to the intervention protocol.

Sample size

Physical capacity, measured by the timed chair stand (TCS) is the primary outcome of this study. Based on the major clinically important improvement (ie, 2)109 (intervention group improves by two more rises from a seated position to a standing position than the control group), and a population SD of 3.2, coming from our pilot study with process evaluation, a sample size of 54 participants per study arm provides 90% power at the 0.05 level of significance. To allow for dropouts, it is planned to recruit 65 participants per study arm, resulting in 195 frail community-dwelling older adults to be recruited.

Data management

The assessor will register data from baseline, T1 and T2 measurements in a secure web application for databases called REDCap. A detailed and pseudonymised Case Report Form (CRF) will be created per participant so blinded extraction and analysis can be carried out. The assessor will be adequately trained to collect and register sociodemographic data and administer the four physical performance tests and nine questionnaires.

Participants and the deliverers of the intervention (professional trainers, volunteers and volunteer coordinator) also register proceedings of every training session in their own pseudonymised log on REDCap. The CRF of the assessor and the logs of the participants and deliverers are strictly separated by disabled user rights. Participants and volunteers first register their findings in a paper log which is collected at 12 and 24 weeks. A designated and blinded researcher will then enter the data in REDCap.

All digital data will be securely stored in a secure cloud-based repository at Vrije Universiteit Brussel, which is encrypted and password-protected and can only be accessed by the PI and/or its representatives. Offline copies of data and the ICFs will be separately archived in a locked cabinet in a room only accessible by the PI (access with key card). The collected data from the target population applies to physical and mental health and, following European law, it will be kept in secure storage for 25 years after final contact with the research participant. All other data will be kept for at least 5 years.

Quality assurance

An independent research partner with no link to the collection of the data will be responsible for fidelity and quality assurance. The intervention groups will be monitored at different times throughout the study in both intervention arms (arm 1: intervention provided by professionals and arm 2: intervention provided by trained volunteers) by means of video recordings and live observations. The recordings and the observations will be evaluated by two researchers independently from each other by completing a checklist with predefined quality indicators. The quality assurance plan is compiled out of three major steps: (a) developing a set of critical indicators of the intervention in a focus group110 with the original developers, (b) collecting data to measure the indicators by video recordings and life observations and (c) evaluating the indicators on a five-point Likert scale. When a score lower than four on the Likert scale is marked, feedback will be given to the trainer.

The researchers adhere to the data management guidelines of the funding partner,111 a formal data-monitoring committee is not requested, but the blinded statistician and independent quality assurance partner will execute data monitoring tasks.

Data regarding the quality assurance will be securely stored in a secure cloud-based repository at the University of Ghent, which is encrypted and password-protected and can only be accessed by the responsible parties of the quality assurance. Offline copies of data will be archived in a locked cabinet in a room only accessible by the responsible researcher at the University of Ghent (access with personal key).

Health economic evaluations

The health economic evaluations concern incremental analyses in terms of incremental costs over incremental effects between the alternatives. Therefore, cost-utility analyses will be conducted. The thresholds suggested by the Belgian Healthcare Knowledge Centre will be applied, expressed in Gross Domestic Product per capita ≈40 000 €/QALY. This threshold represents a willingness-to-pay, as society, for one QALY gained. There is, however, no ‘hard’ cut-off in determining interventions to be cost-effective or not. The investigators will conduct threshold analyses which look for a tipping point for one or more specific input parameters that lead to an incremental cost-effectiveness ratio above or below the threshold. Additionally, probabilistic sensitivity analyses will be conducted and presented in cost-effectiveness acceptability curves indicating at each possible threshold the likelihood of whether one of the intervention arms is cost-effective compared with the alternative. Both trial and model-based economic evaluations will be made. First, a within-trial health economic evaluation will be conducted, with a time horizon aligned with the follow-up time (48 weeks). Next, a model-based approach will link the results at 48 weeks on the Fried frailty criteria with reported healthcare costs (healthcare perspective) and QoL in order to predict over a lifetime horizon the impact of the intervention.

Trial-based evaluation

In the trial-based economic evaluation, the effects are expressed in utilities, derived from the national values of the MOS-SF-36. QALYs will be calculated using the area under the curve method. The cost-effectiveness of the intervention will be expressed in incremental cost per QALY. The incremental cost per QALY will be calculated as a ratio of (Expected Cost ACTIVE-AGE@home - Expected Cost standard care) / (Expected Outcome ACTIVE-AGE@home - Expected Outcome standard care). The robustness of the results will be analysed by probabilistic sensitivity analyses on the cost as well as on the outcome. Bootstrapping with replacement will be employed using @Risk and MS Excel, using a minimum of 1000 iterations to obtain 2.5% and 97.5% of the incremental cost-effectiveness ratio (ICER) distribution. All bootstrapped ICERs will be presented on a cost-effectiveness plane to determine the robustness of the ICER and to determine the probability that ACTIVE-AGE@home is cost-effective at various willingness-to-pay thresholds. A cost-effectiveness acceptability curve will be used to depict the probabilities of acceptable ICERs.

Model-based evaluation

In addition to the trial-based evaluation, a model-based evaluation will be performed to account for the expected costs and health outcomes in both intervention and control groups beyond the follow-up period of the trial. A probabilistic Markov model will be developed compliant with Belgian guidelines for health economic evaluations.112 The rationale of the model is based on the association between frailty and healthcare resources and cost and on the expected outcome of reduced frailty in the intervention arms compared with the control arms. The investigators assume a cycle of 1 year in the model and apply a lifetime horizon. Lifetime incremental costs and QALYs will be the input for the ICER calculation. Discount rates of 3% for costs and 1.5% for utilities will be applied, which is in line with the Belgian guidelines. Non-parametric bootstrapping will be applied for both costs and outcomes to test the robustness of the results. These iterations will be presented in cost-effectiveness planes. Probabilities to be cost-effective for the different willingness-to-pay thresholds will be presented in cost-effectiveness acceptability curves.