Removing environmental barriers to independent living: A feasibility randomised controlled trial targeting people ageing with long-term physical disabilities

Abstract

Introduction:

People ageing with long-term physical disabilities face unique challenges to maintaining independence. We developed a novel, evidence-based intervention – Removing Environmental Barriers to Independent Living – to support participation and reduce falls.

Methods:

To examine Removing environmental barriers to independent living’s feasibility and estimate its magnitude of efficacy, we conducted a single-blinded randomised controlled trial. Fifty participants who were aged 45–65 years with a physical disability for at least 5 years were randomised to intervention (n = 24) or control (n = 26). Activity performance, environmental barriers to performance, community participation, environmental fall hazards and fall prevention behaviours were assessed at baseline and at 6-month follow-up. Dose, adherence, fidelity and cost were examined. Linear mixed-effects models for repeated measures were used to assess the effects of time, group and group by time interaction on the continuous outcomes.

Results:

Removing environmental barriers to independent living had high fidelity and adherence and a low implementation cost. There were trends for greater improvements in activity limitations and activity performance among intervention versus control participants.

Conclusion:

Removing environmental barriers to independent living is a feasible intervention with high fidelity and adherence. Findings trended towards greater improvements in activity performance for intervention participants. Results could inform future planning for a larger pragmatic trial.

Introduction

Individuals with physical disabilities are living longer and facing unique challenges associated with ageing (LaPlante, 2014). They often experience substantial limitations in activity/community participation, have higher risk of falls and fall-related injuries and develop fear of falling. Given the potential adverse effects of falls on activity participation for people ageing with long-term physical disabilities (PAwLTPD), as well as the lack of available evidence-based interventions to address the issues of falls and participation in this population, we have developed an evidence-based intervention, Removing Environmental Barriers to Independent Living (REBIL) (Chen et al., 2023), to bridge this gap in services and support ageing in place for this population. The REBIL intervention was developed using a community-engaged approach and was informed by two evidence-based occupational therapy interventions, Community Participation Transition after Stroke (COMPASS; Somerville, 2019) and the Home Hazard Removal Program (HARP; Stark, 2021). The purpose of this study was to examine the feasibility and estimate the magnitude of efficacy of this new intervention, REBIL, on the primary outcomes of daily activity performance, community participation and environmental fall hazards, as well as exploratory outcomes related to fall rates, activity participation and fall prevention behaviours.

Literature review

PAwLTPD are a diverse group with distinct disability trajectories. This population consists of people with congenital or early-onset conditions (e.g. spina bifida, cerebral palsy, muscular dystrophy, traumatic brain injury) who continue to live into mid- and later life (Kinne, 2004), as well as those who acquire disabilities in midlife due to health issues (e.g. stroke, diabetes-related amputation) and are increasingly surviving into their middle and later years (Vos, 2015). It is estimated that roughly 12 million individuals over the age of 65 are ageing while dealing with disabilities that began early in life (Iezzoni, 2001), and people aged 45 and above with disabilities report a gradual decline in their independence as they grow older (Courtney-Long, 2015).

In conjunction with one’s primary disability, common age-related chronic conditions (e.g. hypertension, diabetes, arthritis, cognitive deterioration, depression) are often part of the ageing experience for PAwLTPD (Kinne, 2004). These chronic conditions can occur at an earlier stage of life for PAwLTPD than for their peers without disabilities and often substantially impact their functional capabilities and independence (Molton, 2014). Moreover, PAwLTPD might simultaneously encounter secondary conditions linked to their primary disability, such as pain (McColl, 2004), fatigue (Jensen, 2007; Patti, 2002) and pressure injuries (Whiteneck, 1992). These shifts in abilities due to common ageing and secondary conditions can profoundly impact independence and community participation for PAwLTPD and can be even more incapacitating than the primary disability itself (Gray, 2008).

Individuals aged 50–65 with long-term physical disabilities demonstrate notably lower participation levels than those aged 18–49. Remarkably, their participation trends closely resemble those of adults aged 65 and above for various daily activities, such as dining out, visiting friends, volunteering, working and shopping (Gray, 2008). Moreover, individuals ageing with conditions such as multiple sclerosis, muscular dystrophy and spinal cord injuries have shown fall rates that are more than double those observed in community-dwelling older adults (Vos, 2015). Substantial limitations in activity participation, including community engagement, are a result of higher rates of falls among people ageing with disabilities (Vos, 2015). Our research team has also discovered a high risk of falls, fall-related injuries and fear of falling among PAwLTPD (Dashner et al., 2023). However, feedback from service providers in focus groups indicates a lack of fall prevention programmes tailored to address the specific needs of this population. An intervention that can target these issues is warranted.

We have developed an intervention, REBIL, for the purpose of addressing activity/community participation barriers and decreasing falls in PAwLTPD concurrently. REBIL was developed based on two evidence-based interventions – COMPASS and HARP. While COMPASS focuses on enhancing daily activity performance, the primary objective of HARP is to mitigate the risk of falling. Both interventions employ a combination of environmental modifications and strategy training. As people with physical disabilities age, they often experience a misalignment between their ability and their environment. Acquiring skills to effectively address new performance challenges and environmental barriers could empower these individuals to continue engaging in meaningful activities safely as they age within their communities. The primary purpose of this study was to test the feasibility of the REBIL intervention.

Method

Trial design and setting

This study was approved by the institutional review board at Washington University in St. Louis. We used a single-blind randomised controlled trial (RCT) to assess the feasibility of REBIL and estimate its efficacy in enhancing the participation of PAwLTPD within their homes and communities while also preventing falls. The study protocol has been published (Chen et al., 2023). Briefly, all participants received an initial in-home evaluation (75 minutes) and then were randomised into either the REBIL treatment group or the waitlist attention control group. Treatment was provided by trained occupational therapists who received 4 hours of online/in-person didactic training and self-guided reading before being certified. Participants in the control group received remote life interview sessions regarding factors such as their neighbourhood environment and their activity and social participation during the COVID-19 pandemic. The interviews were carried out by an occupational therapist or a trained occupational therapy student. The control group had a similar number of sessions and visits to the intervention group. Follow-up for both groups included monthly phone calls to monitor falls and healthcare utilisation and a 6-month in-home follow-up evaluation. Members in the waitlist control group received the REBIL intervention following the completion of their 6-month follow-up. Trained raters who were blinded to the group assignments conducted all follow-up assessments. The trial was conducted in the greater St. Louis area. The first participant received their initial evaluation in November 2020, and the last participant received their 6-month follow-up in June 2023. The report of this study follows the Consolidated Standards of Reporting Trials (CONSORT) reporting guidelines.

Participants

People who met the following criteria were included: (1) age 45–65 years, (2) self-report of difficulty with two or more daily activities using the Older Americans Resources and Services (OARS) activity of daily living (ADL) scale, (3) self-reported onset of a physical disability (e.g. cerebral palsy, spinal cord injury, postpolio syndrome, stroke, amputation, arthritis) at least 5 years prior to participation, and (4) lived within 60 miles of St. Louis. Those who were institutionalised at the time of the study were excluded. We recruited potential participants through (1) referrals from a statewide community-based research network consisting of community-based organisations such as Area Agencies on Aging and Centres for Independent Living (Minor et al., 2020),(2) research participants from a previous longitudinal survey study and (3) word of mouth. Potential participants were provided with detailed information about the study using a phone script, responses to any questions and an eligibility screening conducted by our trained research team members. Study staff obtained signed electronic consent from all participants before the initial evaluation.

Randomisation and blinding

After the initial evaluation was completed using instruments listed in the outcomes section, the occupational therapist initiated the randomisation process using a computerised probability model to allocate participants in a 1:1 ratio through block randomisation sequences generated a priori by the study biostatistician. The allocation ratio was maintained regularly, and groups were balanced for race and sex. Allocation concealment was ensured using Research Electronic Data Capture (REDCap), a secure, web-based tool designed to facilitate data collection for research studies. Trained raters completing follow-up assessments remained unaware of the group assignments. However, due to the nature of the intervention, it was not feasible to blind interventionists and participants to their respective group allocations.

Intervention

REBIL is a manualised intervention informed by data from our longitudinal cohort survey, focus groups and key informant interviews (Chen et al., 2023). These data were used to adapt two evidence-based interventions, COMPASS (Somerville, 2019) and HARP (Stark, 2021). Treatment in REBIL is guided by the environmental press model (Lawton, 1973) and uses a set of effective treatment strategies to address barriers to participation in home and community settings and remove hazards to prevent falls for PAwLTPD. An occupational therapist conducts the intervention, consisting of a total of five 75-minute sessions across an 8-week period. Sessions are spaced 1–2 weeks apart.

REBIL is composed of two essential components: (1) the removal of environmental barriers and fall hazards in both home and community settings and (2) strategy training. Throughout the intervention, the participant identifies activities in their home and the community that are difficult to complete. In collaboration with an occupational therapist, the participant then identifies environmental barriers that increase activity difficulty and contribute to fall risks. The occupational therapist uses strategy training to facilitate the participant’s ability to recognise environmental barriers and fall hazards and to apply compensatory strategies, such as assistive devices, architectural modifications, or task adaptation, to mitigate risks in the home and community. Examples of ways to reduce barriers and hazards include adding grab rails to the wall for increased safety when getting in and out of the shower or providing a kitchen trolley for the individual to move heavy pots and pans in order to save energy.

The intervention encompasses three active ingredients: tailoring, motivational enhancement (ME), and active practice. First, the occupational therapist adapts the compensatory strategies to fit the individual’s specific pattern of functional decline and their distinct home and community environments. During the process, shared decision-making is utilised to engage participants and to discuss expected outcomes and a final prescription based on the participant’s choices. Subsequently, the therapist uses ME, a person-centred directive technique aimed at enhancing motivation for effecting behavioural changes, to formulate a plan for change. Finally, the therapist demonstrates the use of assistive devices, modifications and fall prevention strategies. Following this demonstration, the participant engages in the practice of everyday activities under the therapist’s guidance. If applicable, caregiver training may also be integrated.

A detailed description of the rationale in designing the essential components and active ingredients is provided in the published protocol (Chen et al., 2023). The intervention manual, which includes the conceptual framework, assessments, treatment goals, intervention elements and delivery format, is available upon request from the authors.

Feasibility

Feasibility was informed by process measures. We examined dose (i.e. number of visits and total time spent with participants during the intervention), reasons for delayed intervention, adherence to home modifications (i.e. proportion of recommendations implemented at follow-up), and the reasons for discontinuing the recommended modifications. Fidelity was examined as the proportion of required intervention elements delivered by the interventionist, based on a self-reported checklist completed by the interventionist. Cost for each participant was summed as the total cost of the equipment and the total cost of the contractor to install equipment. Then the median cost was calculated across participants to avoid the effect of extreme values.

Outcomes measures

The prespecified primary outcomes were assessed at baseline and at 6-month follow-up. They included ADL/instrumental ADL (IADL) performance, environmental barriers to ADL/IADL performance, community participation and environmental fall hazards. ADL/IADL performance and environmental barriers were assessed using the In-Home Occupational Performance Evaluation (I-HOPE; Stark, 2010), a performance-based assessment of 44 activities in the home. The I-HOPE focused on the performance of and barriers to participation in the participant’s 10 most problematic activities. The I-HOPE had 4 subscales with high internal reliability (Cronbach α = 0.77–0.78) and interrater reliability (Intraclass correlations (ICC) = 0.94–1.0) (Stark, 2010). The subscales consisted of limitations in daily activities, self-reported performance, satisfaction with one’s performance of problematic activities, and the magnitude of environmental barriers that were affecting performance. A higher I-HOPE activities score indicated a greater proportion of activities (across 44 total activities) that the participant did not have difficulty performing. I-HOPE performance and satisfaction scores consisted of mean ratings across up to 10 participant-prioritised activities on a scale of 1–5 (performance scale: 1 = unable to perform the activity at all, 5 = able to perform the activity without difficulty; satisfaction scale: 1 = not satisfied at all with their performance of the activity, 5 = very satisfied with their performance of the activity). The I-HOPE barrier severity score was a total sum of barrier ratings (0 = independent with/without a device, 1 = standby assistance needed, 2 = minimal assistance needed, 3 = moderate assistance needed, 4 = maximum assistance needed, 5 = no activity) for all identified barriers across all prioritised activities.

Community participation was measured using the Reintegration to Normal Living Index (RNLI; Daneski, 2003; Wood-Dauphinee, 1987), which contained 11 self-report questions (scored 0–100) regarding participation in self-care, functional mobility, work and school, leisure activities and travel in the community. Higher scores indicated greater community integration.

Environmental fall hazards were assessed using the Westmead Home Safety Assessment (WeHSA) short form, a performance-based assessment of fall hazards in different spaces of the home (e.g. kitchen, bathroom, hallway, bedroom). WeHSA scores indicated the total number of environmental fall hazards in the home. Content validity and interrater reliability have been established (Clemson, 1999).

Exploratory efficacy outcomes included falls, healthcare utilisation, activity participation and fall prevention behaviours. Falls (i.e. an unexpected event in which a person comes to rest on the ground, floor, or a lower level) and healthcare utilisation (i.e. the number of emergency department and doctor visits, number of hospitalisations, and number of days in therapy) were collected monthly for 6 months via phone or online survey. Activity participation and fall prevention behaviours were assessed at baseline and at 6-month follow-up. We used the Participation Scale to measure activity participation. This scale included 25 diverse activities across 4 participation domains: routines, recreation, responsibilities and relationships (Yorkston, 2008). Participants rated their frequency of activity participation, the importance of the activity and their self-efficacy in completing the activity using a 5-point Likert-type scale. We calculated the ratio of the number of activities the participant rated as important/very important that they could perform very much/as much as they wanted (numerator) to the total number of activities they rated important/very important (denominator). This scoring was modified from a similar participation scale with a slightly different structure of response options (Plow, 2015). This activity participation measure was not a duplicate of the I-HOPE, as the I-HOPE only measured ADLs/IADLs that happen at home, while the Participation Scale mainly focused on activities that take place in the community. We included this activity participation measure as an exploratory outcome because we wanted to test whether resolving participants’ participation difficulty and fall hazards at home would lead to an increase in community activity participation.

Fall prevention behaviours were assessed with the Falls Prevention Strategy Survey (FPSS), designed to evaluate 11 protective behaviours regarding fall risk in adults (Finlayson, 2009). Respondents indicated the frequency of their engagement in 11 distinct fall prevention behaviours using response options that included ‘never’ (scored 0), ‘sometimes’ (scored 1) and ‘regularly’ (scored 2). To accommodate the population of PAwLTPD, we slightly modified the FPSS by adding 1 additional item: ‘I use the mobility device (e.g. cane, walker, wheelchair) that is most likely to reduce my fall risk during the activity I will be doing’. With this modification, a total FPSS score ranged from 0 to 24, with a higher score indicating greater use of fall prevention strategies.

Demographic information, including age, race and ethnicity, sex, education level, marital status, primary disability, use of mobility device and living arrangement (i.e. living alone or with others) was collected. Covariates included demographic characteristics and depressive symptoms, which were measured using the Patient-Reported Outcomes Measurement Information System (PROMIS) Depression Scale-Short Form 8-item. Uncorrected t-scores were calculated by the PROMIS scoring system, with higher scores indicating a higher level of depressive symptoms. A score of 50 is the average in the US general population, with a standard deviation of 10 (Schalet, 2016).

Sample size

Although this was primarily a feasibility study and statistical significance was not the primary goal, a sample size calculation was conducted a priori. A sample size of 50 (25 in each group) was necessary to provide 80% power to detect differences in changes in I-HOPE barrier scores between groups (expected change in intervention group 26.3, SD 18.0; expected change in control group 10.1, SD 22.3) with a significance level of 5%. More information on the sample size calculation can be found in our article detailing the study protocol (Chen et al., 2023).

Statistical analyses

Baseline characteristics were compared between the REBIL intervention group and the waitlist control group using Pearson’s chi-square tests or Fisher’s exact tests for categorical variables and independent samples t-tests for continuous measures. These tests were also used to compare demographic characteristics between those who were eligible and randomised versus those who were eligible but declined or dropped out before randomisation. Because this was primarily a feasibility study, for all analyses of outcomes, we report results of statistical tests but focus the interpretation on the size and direction of effects between REBIL and control participants. Linear mixed-effects models appropriate for repeated measures data were used to assess the effects of time (6-month follow-up versus baseline), group (REBIL vs control), and the interaction of group and time on continuous primary and exploratory outcomes. All models were adjusted for age, sex, race and depression score at baseline. Models for participation-related outcomes (primary outcome: RNLI, exploratory outcome: Participation Scale) were additionally adjusted for employment status. Estimated marginal means along with comparisons of change over time within each group were requested from the models. Mean differences over time within each group were reported with 95% confidence intervals (CIs). The exploratory outcomes of rate of falls, emergency department visits, hospitalisations, doctor visits and therapy visits (all calculated as the number per 1000 observed days) were compared between groups using negative-binomial regression models. Models for exploratory outcomes were also adjusted for age, sex, race and depression score at baseline. IBM SPSS version 29.0 and SAS version 9.4 were used for statistical analyses.

Results

Participant flow

Among 147 individuals assessed for eligibility, 89 were eligible. Fifty-eight did not meet inclusion criteria. Among those who did meet eligibility criteria, 26 declined (13 were not interested or did not see a benefit to participation, 4 thought the study would be too much of a time commitment, 3 had COVID-19 concerns, 2 had other health concerns, 1 felt too overwhelmed, 1 lived with another study participant and thus declined, 1 did not believe the study provided enough compensation and 1 did not provide a reason). In addition, research staff were unable to contact 9 eligible participants after screening. Finally, 4 dropped out after consenting but before randomisation (1 unable to contact, 1 not enough time, 1 privacy concerns and 1 perceived no benefit to participating). A total of 50 were randomised, with 24 assigned to the REBIL intervention group and 26 to the waitlist control group (Figure 1). When comparing those who were eligible and randomised (n = 50) with those who were eligible but declined or dropped out prior to randomisation (n = 35; 4 did not have demographic data available), there were no significant differences by sex or race (both ps > 0.05). Those who were eligible but declined/dropped out were slightly older (mean = 59.8, SD = 5.4) than those who were randomised (mean = 57.6, SD = 5.1), but this did not quite reach statistical significance (mean difference = 2.2, 95% CI = −0.1 to 4.5, p = 0.06). After randomisation, the study team was unable to reach 1 participant in the REBIL intervention group and two participants in the waitlist group; thus, no follow-up assessments were available for these individuals. The final sample size for analysis was 23 in the REBIL intervention group and 24 in the waitlist control group.

Figure 1.

CONSORT diagram for the pilot randomised controlled trial to evaluate the efficacy of the REBIL intervention to promote activity participation in home and community and to prevent falls for people ageing with long-term physical disability.

Baseline characteristics

Table 1 presents baseline characteristics for REBIL and control participants. Among the 47 participants, over 60% were White, over 60% were female, and the average age was 58 years. Over half lived with someone else (most commonly their spouse, n = 20), and over half received assistance with their daily activities. Over half were on disability leave, and over one-quarter were working either full- or part-time. Participants self-reported a wide range of conditions that caused their physical disability, including, but not limited to, multiple sclerosis, stroke, cerebral palsy, spinal cord injuries, diabetes and arthritis. The majority used some type of mobility device, the most common being a wheelchair (15), cane (13), or walker/rollator (11). Baseline PROMIS depression scores indicated slightly higher levels of depressive symptoms (mean t-score = 54.1) than the general US population (mean t-score = 50, SD = 10). There were no statistically significant differences between groups for demographic characteristics or baseline values for primary outcomes. There were an average of 11 (SD 5.3, range 1–23) home fall hazards documented among participants in the treatment group, with the most common hazards being bath/shower rails (n = 17, 73.9%), seating surfaces (n = 14, 60.9%), lawns/gardens/grounds (n = 13, 56.5%), shower recesses (n = 12, 52.2%) and toilet grab rails (n = 12, 52.2%). Waitlist participants had an average of 10.8 (SD 4.7, range 2–19) home hazards, with the most common fall hazards being bath/shower rails (n = 18, 75.0%), beds (n = 16, 66.7%), seating surfaces (n = 15, 62.5%), and toilet grab rails (n = 13, 54.2%). REBIL participants needed an average of 11 home modifications (SD 4.7, range 4–21), and the large majority of these were adaptive equipment (mean 8.7, SD 3.9, range 1–16).

Table 1.

Baseline demographics and primary outcomes.

Baseline characteristic	Total (N = 47)	REBIL intervention (n = 23)	Waitlist control (n = 24)	Mean difference or difference in proportions (95% CI)	p
Participant characteristics
Age (years), mean (SD)	57.8 (4.9)	57.3 (4.4)	58.4 (5.3)	−1.1 (−4.0, 1.8)	0.44
Sex, n (%)
Female	31 (66.0%)	15 (65.2%)	16 (66.7%)	−1.5% (−28.6%, 25.6%)	0.92
Male	16 (34.0%)	8 (34.8%)	8 (33.3%)
Race, n (%)
Black/African American	17 (36.2%)	9 (39.1%)	8 (33.3%)	5.8% (−21.6%, 33.2%)	0.65
Hispanic or Latino	1 (2.1%)	1 (4.3%)	0 (0.0%)	4.3% (−4.0%, 12.6%)
White	29 (61.7%)	13 (56.5%)	16 (66.7%)	−10.2% (−37.9%, 17.5%)
Living situation, n (%)
Lives with someone	27 (57.4%)	13 (56.5%)	14 (58.3%)	−1.8% (−30.1%, 26.5%)	0.90
Lives alone	20 (42.6%)	10 (43.5%)	10 (41.7%)
Receives assistance with daily activities, n (%)
Yes	28 (59.6%)	14 (60.9%)	14 (58.3%)	2.6% (−25.5%, 30.7%)	0.86
No	19 (40.4%)	9 (39.1%)	10 (41.7%)
Employment status, n (%)
Working full-time/part-time	13 (27.7%)	6 (26.1%)	7 (29.2%)	−3.1% (−28.7%, 22.5%)	0.51
Disability leave	27 (57.4%)	12 (52.2%)	15 (62.5%)	−10.3% (−38.4%, 17.8%)
Retired/other	7 (14.9%)	5 (21.7%)	2 (8.3%)	13.4% (−6.7%, 33.5%)
Use of mobility device, n (%)
None	5 (10.6%)	3 (13.0%)	2 (8.3%)	4.7% (−12.9%, 22.3%)	0.69
Wheelchair	15 (31.9%)	6 (26.1%)	9 (37.5%)	−11.4% (−37.8%, 15.0%)
Other device (e.g. cane, walker)	27 (57.4%)	14 (60.9%)	13 (54.2%)	6.7% (−21.5%, 34.9%)
PROMIS depression t-score, mean (SD)	54.1 (9.2)	53.5 (9.6)	54.6 (9.0)	−1.1 (−6.5, 4.4)	0.70
Primary outcomes
I-HOPE score, mean (SD)
Activity	0.73 (0.11)	0.74 (0.12)	0.72 (0.10)	0.02 (−0.05, 0.08)	0.65
Performance	2.6 (0.6)	2.6 (0.6)	2.6 (0.6)	−0.04 (−0.4, 0.3)	0.82
Satisfaction	2.2 (0.7)	2.2 (0.7)	2.1 (0.7)	0.03 (−0.4, 0.5)	0.88
Barrier	51.4 (24.6)	50.0 (23.7)	52.8 (25.7)	−2.9 (−17.4, 11.7)	0.69
Reintegration to Normal Living Index (RNLI), mean (SD) a	62.7 (19.4)	63.5 (20.2)	62.0 (19.1)	1.5 (−10.4, 13.3)	0.81
Westmead Home Safety Assessment score (WeHSA), mean (SD)	10.9 (5.0)	11.0 (5.3)	10.8 (4.7)	0.2 (−2.7, 3.2)	0.89

I-HOPE activity score: a higher score indicates fewer limitations in daily activities; I-HOPE performance score or satisfaction score: a higher score indicates better performance or satisfaction; I-HOPE barrier score: a higher score indicates more barriers to activity participation; RNLI: A higher score indicates a better normal living; WeHSA: a higher score indicates more environmental hazards.

^aRNLI n = 45 (two missing data in the intervention group missing data).

Intervention feasibility

REBIL intervention participants received on average of 303.0 minutes (SD = 110.9) over 4.2 visits (SD = 1.2) with the interventionist. Some intervention participants experienced a delay in intervention visits (n = 9), most commonly due to equipment or contractor delays (n = 7), followed by medical reasons (n = 3), work schedules (n = 2) and travel or holidays (n = 2). Waitlist participants received an average of 135.2 minutes (SD = 42.9) over 4.0 (SD = 0.2) visits for life interviews. Fidelity was calculated as the number of required intervention elements implemented by the interventionist divided by the number of required intervention elements designed in REBIL. Average fidelity across treatment participants was 89.9% (SD = 8.3%).

Median total cost, including equipment and contractor costs to implement home modifications, was $614.96 (US dollars; range $93.19–$1523.20). The majority of costs were for equipment, as the median equipment cost was $421.69 (range $93.19–$886.69) and the median contractor cost was $115.00 (range $0–$875.00). Average adherence to home hazard removal strategies (number of recommendations still used at 6-month follow-up divided by number of recommendations initially implemented) was 94.5% (SD 8.2%). Reasons for no longer using strategies were it was not working how the participant wanted it to, they used something else instead, they decided they did not need it, and the item broke.

Primary outcomes

Because this was a feasibility study and not powered to detect significant differences, we focused our interpretation on size and direction of effects. Table 2 presents results of linear mixed models examining the effects of group, time and group by time interaction on each primary outcome, adjusting for demographic covariates and depression scores. Estimated marginal means for the outcomes from the models are graphed in Figure 2. While I-HOPE activity scores changed little over time, the interaction of group and time trended towards significance, suggesting greater improvement among REBIL versus control participants (β = 0.06, 95% CI −0.005, 0.13). While activity scores tended to increase for the REBIL intervention group from baseline to follow-up (mean difference = 0.03), scores tended to decrease for the control group (mean difference = −0.03). For I-HOPE performance scores, the interaction of group and time also trended towards significance, suggesting greater improvements among REBIL versus control participants (β = 0.39, 95% CI −0.02, 0.80). Performance scores tended to increase about twice as much for the REBIL intervention group (mean difference = 0.74) than for the control group (mean difference = 0.35). Both groups improved in I-HOPE satisfaction and barrier severity scores and, although improvements were slightly greater among REBIL participants, improvements were relatively similar for both groups.

Table 2.

Results from linear mixed models predicting primary outcomes.

	REBIL intervention			Waitlist control			Group main effect		Time main effect		Group by time
Outcome	Baseline Mean (SE)	6-month follow-up Mean (SE)	Difference (95% CI)	Baseline Mean (SE)	6-month follow-up Mean (SE)	Difference (95% CI)	F	p	F	p	F	p
I-HOPE activity score	0.73 (0.03)	0.76 (0.03)	0.03 (−0.02, 0.08)	0.72 (0.03)	0.69 (0.03)	−0.03 (−0.08, 0.01)	2.19	0.15	0.01	0.91	3.50	0.07
I-HOPE performance score	2.49 (0.15)	3.23 (0.15)	0.74 (0.45, 1.03)	2.52 (0.15)	2.87 (0.15)	0.35 (0.06, 0.64)	0.81	0.37	28.56	<0.01	3.64	0.06
I-HOPE satisfaction score	2.12 (0.17)	2.86 (0.17)	0.74 (0.37, 1.10)	2.09 (0.18)	2.74 (0.18)	0.65 (0.29, 1.01)	0.13	0.72	29.59	<0.01	0.13	0.72
I-HOPE barrier score	50.28 (4.78)	21.93 (4.78)	−28.35 (−36.42, −20.28)	51.83 (4.82)	27.79 (4.82)	−24.04 (−31.94, −16.14)	0.38	0.54	87.26	<0.01	0.59	0.45
RNLI	65.59 (3.56)	70.08 (3.52)	4.49 (−1.83, 10.81)	66.10 (3.72)	68.90 (3.72)	2.80 (−3.14, 8.75)	0.01	0.93	2.88	0.10	0.16	0.70
WeHSA	10.52 (1.05)	7.04 (1.05)	−3.48 (−6.16, −0.80)	9.94 (1.05)	6.94 (1.05)	−3.00 (−5.62, −0.38)	0.09	0.76	12.12	<0.01	0.07	0.80

Estimated marginal means with standard errors are shown. Differences in estimated marginal means are shown for 6-month follow-up minus baseline. Models were adjusted for age, gender, race and ethnicity and baseline depression score. The model for RNLI was additionally adjusted for employment status.

Figure 2.

Estimated marginal means for I-HOPE activity (2a), performance (2b), satisfaction (2c), barrier severity (2d), RNLI (2e), and WeHSA (2f) scores at baseline and 6-month follow-up for participants randomised to REBIL intervention and control groups.

Bars represent 95% confidence intervals. I-HOPE activities score: Higher scores indicate a greater proportion of activities that a participant does not have difficulty performing. I-HOPE performance and satisfaction scores: Higher scores indicate more positive outcomes. I-HOPE barrier severity score: Higher scores indicate lower independence. WeHSA: Higher scores indicate more fall hazards. RNLI: Higher scores indicate greater community integration.

Both groups also improved in community participation as measured by the RNLI, and while improvements were slightly higher for REBIL participants, the increase was relatively similar to that of control participants. Environmental hazards in the home measured by the WeHSA decreased for both groups to a relatively similar extent.

Exploratory outcomes

Baseline fall prevention behaviours, measured by the FPSS, did not significantly differ between REBIL participants (M = 13.3, SD = 4.4) and control group participants (M = 14.0, SD = 3.8; mean difference −0.7, 95% CI = −3.1–1.7, p = 0.57). Similarly, baseline activity participation scores measured by the Participation Scale did not significantly differ between REBIL participants (M = 0.52, SD = 0.25) and control participants (M = 0.52, SD = 0.19; mean difference 0.0, 95% CI = −0.14 to 0.13, p = 0.94). According to linear mixed models (Supplemental Table S1), although not statistically significant, improvements in FPSS scores and participation scores tended to be about twice as high among the REBIL group than the control group.

Fall rates and healthcare utilisation rates for each group are shown in Supplemental Table S2. During follow-up, the fall rate of REBIL participants (15.0 falls per 1000 observed days) was approximately 15% lower than that of the control group (17.6 falls per 1000 observed days). In general, rates of healthcare utilisation during follow-up were lower among REBIL participants than the control group, with the largest differences being for emergency department visits and hospitalisations. None reached statistical significance.

Discussion and implications

This study examined the feasibility and estimated the magnitude of efficacy of a novel home modification intervention, REBIL, which was informed by 2 evidence-based interventions (HARP and COMPASS) and aims to improve participation and decrease falls for PAwLTPD. Findings suggest that REBIL is a feasible intervention with high fidelity and high adherence to the suggested home modifications. The high fidelity (89.9%) indicates that the 4-hour didactic training and self-guided reading are sufficient for therapists to implement the REBIL intervention with the required intervention elements. Based on our clinical experience, 4 hours of training for clinicians is considered a manageable and reasonable amount of time. The high adherence rate (94.5%) reflects that therapists’ suggestions are highly relevant in addressing participants’ difficulties with activity participation and removing fall hazards, so participants continued to use them 6 months after the intervention. Twenty-three REBIL intervention participants received the full intervention (average 303 minutes over 4.2 visits), and none dropped out during the intervention, even with some delays in intervention. Only 1 person was lost to follow-up after the intervention. This indicates that the REBIL intervention has good acceptability in terms of its content and required time. Our findings also show that REBIL is a low-cost intervention to implement (median $615, range $93–$1523). This is much lower than the medical expenses for treating fatal and non-fatal fall-related injuries in the United States (Burns et al., 2016). As an example, according to a study which analysed Medicaid expenditure in the US state of California, people with disabilities spend, on average, $738 per month on in-home personal assistance services to help with daily activities (Newcomer et al., 2012). Compared to this number, the one-time cost of REBIL is lower, and the intervention is likely to help reduce the cost of personal assistance.

This study also examined the outcomes of daily activity performance, community participation, and environmental fall hazards (primary outcomes), as well as fall rates, activity participation, and fall prevention behaviours (exploratory outcomes). Although not statistically significant, improvements in I-HOPE activity and performance scores, participation, and fall prevention behaviours tended to be greater for the intervention group compared to control. Fall rates also appeared lower for REBIL compared to control participants. As REBIL is a newly developed intervention, the study was designed to understand its feasibility and primary efficacy for future larger trials. Therefore, the number of participants in this study was not sufficiently powered for all outcomes, likely explaining the nonsignificant findings.

We found that both intervention and control groups showed improvements in I-HOPE performance, I-HOPE satisfaction, and I-HOPE barrier scores, as well as WeHSA, FPSS, and activity participation scores at the 6-month follow-up. These improvements over time among REBIL participants were expected and indicate that the intervention has the potential to enhance participation and fall prevention behaviour. Planning a phase III trial with a larger sample size may be warranted. Although the control group exhibited a similar trend for some outcomes, this could be because having a rater/therapist assess the participant’s fall prevention strategies and examine the house for fall hazards and environmental barriers could raise clients’ awareness, prompting changes in their fall prevention behaviours and environment (e.g. Hawthorne effect; Becker, 2003; Waters, 2011).

To our knowledge, REBIL is the first home modification programme targeting PAwLTPD. The strength of this study lies in its inclusion of a population with diverse physical conditions (e.g. neurological, orthopaedic, cardiopulmonary) instead of focusing on a single diagnosis. This diversity reflects the real world, wherein physical disabilities can be caused by various medical conditions. This sample also had higher depressive symptoms than the general US population (mean = 50), which is one of the prominent ageing-with-disability symptoms that PAwLTPD face, along with pain and fatigue (Putnam, 2022). The employment rate in this sample was also low, with over half on disability leave. This result is not surprising and aligns with employment status in our PAwLTPD cohort study (Morgan et al., 2022), from which most of this study’s participants were recruited. Some of these PAwLTPD chose to forgo employment opportunities to maintain public health insurance through Medicaid (Morgan et al., 2022). Although the Medicaid Expansion Programs in Missouri, which began enrolment on October 1, 2021 (McBride, 2022), were intended to improve employment rates for people with disabilities, by the enrolment date, our study had completed the majority of participant recruitment and had randomised 44 participants. Consequently, the employment status of our study sample who were on disability leave remained high.

There were a few limitations to the study that may have impacted the findings. One limitation was the single-blind study design, which may have contributed to biased results; however, it was not possible to blind participants and therapists providing the intervention. Also, the control group received the same number of visits but less contact time in total compared to the intervention group. This may have introduced unknown bias to the outcome findings. In a future, larger efficacy trial, this should be prevented from occurring by preparing extra activities in case the control group activity (e.g. interview) ends early. Third, this study was primarily designed as a feasibility trial; therefore, it was not powered to find significant differences between groups for outcomes. However, the intervention was proven to be feasible and low cost.

REBIL is relevant to occupational therapists internationally not only because home modification is one of the practice areas of occupational therapy but also because falls and difficulties participating in daily activities are universal concerns for people with physical disabilities. This emerging population, PAwLTPD, can also be found across developed countries. The cost of the intervention will vary in different countries, as equipment prices and contractor rates are different from those in the United States. The training time required to become certified to implement the REBIL intervention is also considered manageable for international practitioners, as it is similar to the amount of time required for regular continuing education courses that are often required for practicing occupational therapists in most countries. For a future, larger trial, our priorities are to maximise generalisability by continuing to recruit participants from diverse demographic backgrounds as we did in this study and to examine the effectiveness of REBIL.

Conclusions

This RCT found that the manualised REBIL intervention delivered to PAwLTPD was feasible and low cost to implement, had high fidelity, and had high adherence to home modifications. Though not statistically significant due to low power in this feasibility trial, there were trends for greater improvements for in-home activity limitations, performance of participants’ prioritised activities, and fall prevention behaviours for REBIL participants than for control participants. Fall rates also appeared lower for REBIL than control participants. The findings of this study will help us select optimal endpoints for a large, pragmatic trial in which we will consider recruiting more participants with diverse demographic characteristics. REBIL has the potential to address participation issues and prevent falls concurrently, but future studies are needed to establish its effectiveness.

Key findings

• The REBIL intervention was low-cost and feasible with high fidelity and adherence to home modifications.

• Intervention participants tended to have greater improvements for activity limitations and performance compared to control.

What the study has added

REBIL has the potential to address participation barriers and prevent falls concurrently, and this study will help us select optimal endpoints for a large, pragmatic, future trial to establish its effectiveness.