Community-based Exercise Enhanced by a Self-management Programme to Promote Independent Living in Older Adults

A Pragmatic Randomised Controlled Trial

Pia Øllgaard Olsen; Mark A Tully; Borja Del Pozo Cruz; Manfred Wegner; Paolo Caserott


Age Ageing. 2022;51(7):afac137 

In This Article


This article reports primary and secondary outcomes from a pragmatic trial of the Welfare Innovation in Primary Prevention (WIPP) project funded by the European INTERREG 5a programme to support development and innovations in the German-Danish boarder regions.[23] The trial followed a parallel, two-armed randomised controlled design. Prospective participants received verbal and written information about the random allocation, intervention and testing procedures before agreeing to participate, and they signed written consent to share their anonymised data for scientific purposes. The reporting followed the extended CONSORT statements for pragmatic studies[24,25] and the TIDIER checklist.[26] ClinicalTrials database: reg. no. NCT04531852. Danish Data Protection Agency: reg. no. 10.583

Pragmatic Study Model and Recruitment Pathway

To ensure high ecological validity.[27] we developed a pragmatic study model that enabled effectiveness of the interventions to be evaluated in a naturalistic preventive practice framework.

A detailed description of the model is available in Supplemental Material S1. In summary, the study intervention was co-created with key stakeholders including representatives from academia, primary care providers, community partners and older adult's organisations.

To broaden the reach and obtain a more specific recruitment of the intended at-risk target group, participants were recruited through a well-established nationally regulated preventive pathway since 1996: the Preventive Home-Visits (PVH).[28,29] This pathway dictates that each individual aged 75 years or older (65 for vulnerable subgroups) who does not receive personal care services on a regular basis is entitled to receive a PHV by health care personnel employed in the municipality.[28] The aim of the PHV is to uncover potential problems that may threaten well-being and independent living and provide information about relevant resources in the community, public health care sector and by third-party actors[29] to tackle such problems.

Invitation-letters for the PHV are automatically sent through a secure email system and through phone calls and by post for citizens who do not have such digital systems. Occasionally, secondary pathways (e.g. local media, info-meetings) are used to recruit individuals at high-risk who do not respond to the letters.

All study phases (i.e. recruitment, data collection, intervention) were fully run by the municipalities (i.e. primary care providers), who recruited trainers and assessors among their existing health care personnel, predominantly nurses and occupational therapists.


Over a 2-year period, all citizens who were eligible to the PHV and living in three Danish municipalities received an invitation to WIPP-screening within their routine invitation-letter for the PHV. The WIPP-screening was developed to: (a) early identify citizens at greater risk of functional loss and (b) enable timely and tailored action-plans to modify such risk. It covers physical, mental and social risk factors adopting the International Classification of Health, Functioning and Disability framework.[30] Inclusion criteria included ≥1 of the following risk factors: (1) low PA (moderate to vigorous PA ≤1 day/week while daily sitting time ≥ 8 hs); (2) high fatigability (Pittsburgh Fatigability Score ≥ 15);[31] (3) recurrent falls (≥2 falls over the past year); (4) pain interference with daily activities (Brief Pain Inventory interference score ≥ 20);[32] or (5) low functional capacity (short physical performance battery; SPPB-score ≤ 9).[33,34] Participants with SPPB-score > 10 or physically active ≥3 days/week while daily sitting time < 5 h were excluded irrespective of whether they met one of the other eligibility criteria. Occasionally, ineligible relatives were allowed to participate in interventions if their eligible partner needed support (e.g. transport); however, they were excluded from the analysis.

Data Collection, Randomisation and Blinding Procedures

Eligible participants were invited to a baseline assessment followed by sealed random allocation (Supplemental Material S2, Supplemental Table 1) to: (i) the Complex Lifestyle Intervention (CALSTI) and (ii) usual care enhanced with the self-management intervention (SEMAI). Participants were also handed out a battery of self-report questionnaires to complete at home and be returned to trainers at the following group-session, or by post using pre-paid envelopes. Collection of intervention data was repeated after 12 and 24 weeks. Follow-up assessors were blinded to group-allocation, and participants were instructed not to reveal any information that may unmask their allocation.


CALSTI was designed with two group-based components starting off in parallel: (a) 12-week (24 sessions) progressive exercise component; and (b) 24-week multi-factorial self-management programme (8 sessions) (Figure 1). Interventions took place in diverse community-settings including activity centres, housing organisations and sports clubs. Detailed description of intervention protocols is available in Supplemental Material S3. Briefly, the exercise component aimed to enhance functional capacity primarily through progressive high-intensity explosive-type resistance training following a protocol from a previous study[35] that was modified to fit the diversity availability of exercise equipment. Sessions included also progressive balance and high-intensity aerobic exercises.

Figure 1.

Schedule of interventions and data collection. SMS: self-management strategy programme; Ex: exercise programme; h: hours

The self-management programme aimed at empowering participants to tackle key modifiable risk factors for functional loss and barriers for engaging in PA. This was operationalised by (i) supporting self-management skills (problem solving, decision making, resource utilisation and taking action);[36] (ii) mobilising use of activities (e.g. senior sports clubs, social eating); and (iii) identifying supportive resources in the personal network and local community. Formal and informal (i.e. peer-based) educational strategies (presentations by the instructor, brains-storming techniques) were used to raise awareness about personal risks factors and opportunities to act on them. This facilitated participants to focus on the risk factors they each perceived most relevant, enabling a more individualised programme. The programme content was manualised, and drew from theoretical frameworks based on Social Cognitive Theory,[37] focusing on self-efficacy[38] as the major determinant of successful behaviour-change process. The inclusion and temporal progress of components (e.g. health education, barrier management) was guided by the Health Action Process Approach (HAPA)[33,34] (Figure 2). Trainers led the sessions adopting a goal-oriented and participant-centred counselling and communication style based on motivational interviewing principles.[39]

Figure 2.

Scheme of behaviour-change strategies in the self-management programme and how they map to the HAPA. Centrally: the HAPA model with two temporal phases: the Motivation phase that closes with forming an intention to change, and the Action phase which includes planning, action and maintenance. Ovals illustrate HAPA-determinants of successful Motivation and Action phases, with self-efficacy being the major predictor in both phases (Schwarzer, R., Applied Psychology, 2008 [40, 41]). The toned text boxes list behaviour-change strategies in the self-management programme (e.g. health education) and examples of activities representing each strategy (e.g. information booklets). Stippled arrows illustrate how behaviour-change strategies in the self-management programme map to the elements in the HAPA model (e.g. task self-efficacy). dSMART: Specific, Measurable, Achievable, Realistic, Timely. The scheme was used and modified with permission by R. Schwarzer. This scheme is not covered by the terms of the Creative Commons licence of this publication. For permission to reuse, please contact the rights holder.

The SEMAI intervention included the self-management programme as in CALSTI with four additional sessions dedicated to gain mastery experiences with exercising by (1) visiting local exercise facilities that offer activities for older adults, (2) practising exercises to be performed at home during the sessions, (3) exploring local walking routes and outdoor exercise facilities (2 sessions).


Primary Outcome. The primary outcome of this study was functional capacity assessed by the SPPB[42] which was earlier shown to predict home care service utilisation,[34,43] nursing home admission,[42] disability[44] and mortality.[42,45] The SPPB consists of three tests (gait, chair-rise and balance) each scored from 0 to 4 and summed into a total score of maximum 12 points (best performance). Previous data from Perera and colleagues (2006)[46] indicate that to detect a small (0.5 ± 1.48 point) and substantial (1 ± 1.48 point) meaningful change in SPPB-score, minimum numbers of 138 and 35 participants per group respectively are needed for 80% power in a between-group comparison.

Secondary Outcomes. Secondary outcomes included self-reported function and disability assessed by the sum of difficulty in five ADL/IADL items,[47] and the short form of the Late-Life Function and Disability Instrument (SF-LLFDI).[48] LLFDI consists of two disability subscales measuring frequency of and limitation in participation in major life tasks and social roles, and one function subscale assessing inability to perform discrete physical tasks.[48] Self-reported health state was assessed the EQ-health VAS score.[49] Secondary outcomes are described in detail in Supplemental Material S4.


Key socio-demographic variables were described using mean and standard deviations (SDs) or counts (n) and proportions (%) for continuous and categorical variables, respectively. An intention-to-treat approach in participants with baseline data for the outcome of interest was used to test the hypothesis in this study. A detailed description of the statistical methods is available in the Supplemental Material S5. Briefly, the core principles in the statistical analysis were linear mixed models, and multiple imputation using chained equations. The proportion of imputed data is displayed in Supplementary Table 4. Missing mechanisms were tested, and we found that baseline data were missing at random (Supplementary Tables 2 and 3).

To investigate the validity of the multiple imputation model, the distribution and ranges of the imputed values were checked against the observed data, and sensitivity analysis on observed data was conducted (Supplementary Table 8). All models were adjusted for age, sex and baseline value of the outcome being tested. We used Stata/IC 16.1 for Mac (StataCorp LCC, 2019) and did statistical testing at a two-tailed alpha level of 0.05.