Effect of a Novel Transition Program on Disability After Stroke

Key Points Question Is a novel enhanced rehabilitation transition program, Community Participation Transition After Stroke (COMPASS), more effective at improving community participation and daily activity performance and reducing environmental barriers among stroke survivors than an equivalent dose of attentional control? Findings In this phase 2b, single-blind, parallel-group, randomized clinical trial, 180 adults who have had ischemic or hemorrhagic strokes will be studied as they transition home from inpatient rehabilitation. Meaning Removing environmental barriers faced by stroke survivors as they transition home may improve daily activity performance as well as home and community participation.


Glossary of Terms
Environmental barriers Features of the built environment which makes activity performance difficult or unsafe (e.g. lack of hand support by toilet to hold onto while transferring) Environmental modifications Changes to the built environment which remove the barrier and compensate for loss in function that a person might experience after a stroke (e.g. installing grab bars on each side of the toilet to give individual hand hold while transferring).
Strategy training Strategy training is an iterative rehabilitation intervention that promotes goal setting, planning and self-monitoring used to help individuals identity and overcome difficulties on their own.

Tailoring
Tailoring is the process the interventionist uses to adapt the treatment intervention to match and compensate for each participant's pattern of functional loss and unique home and community environments.

Motivational interviewing
Motivational interviewing is a method used to help individuals resolve ambivalent feelings and insecurities to find the internal motivation needed to change behavior. It is a practical empathetic process that takes into consideration the difficulty people face when making life changes, especially after a significant health event like a stroke.
Booster Visit A visit which occurs months after the initial intervention that is completed to reinforce the strategies learned during the intervention as well as help the participant problem solve any new performance problems which might have arisen after the initial intervention was completed.
Community participation An individual's involvement his/her own community, as the individual sees fit and meets his/her individual desires and needs. Version : We will compare the efficacy of a novel enhanced rehabilitation transition program (COMPASS) and an equivalent dose of attentional control for significant improvements in community participation and ADL performance and a reduction in environmental barriers in the home and community after stroke.
Specific Aim 2 (Secondary): We will evaluate alternative primary outcome measures of participation, function, patient-reported quality of life, and caregiver burden that permit comparison to other stroke clinical trials and confirm the safety of COMPASS.

Specific Aim 3 (Secondary):
We will evaluate process outcomes such as reach, cost, fidelity, and adherence to aid interpretability of the trial and future implementation.  6 and test-retest reliability (Interclass correlation range 0.7-0.92), 6 and discriminate across four Rankin levels of stroke severity (P ≤ 0.01). 6 The SIS ADL domain demonstrated a moderate (0.44) effect size in our pilot.
The In-Home Occupational Performance Evaluation (I-HOPE) to measure ADL performance and magnitude of environmental barriers in the home. The I-HOPE is a performance-based, multistep, clientcentered assessment that evaluates the performance of older adults doing 44 activities in the home. Four subscales measure limitations in daily activities, self-reported performance and satisfaction with performance of problematic activities, and the magnitude of environmental barriers that influence performance. I-HOPE assessment is commonly used in occupational therapy (OT) to provide home modification interventions. The instrument yields four subscales with high internal reliability (α = 0.77-0.78) and interrater reliability (α = 0.94-1.0) 7 and is sensitive to change in environmental support.

Exploratory endpoints:
Our goal is to optimize the design of a phase III multicenter trial; therefore, we will examine additional endpoint measures to evaluate their utility in the trial setting. These measures include the Caregiver Inventory 8 (CGI; caregiver burden), Perceived Stress Scale, 9 Barthel Index 10 (BI; ADL performance), and Patient-Reported Outcomes Measurement Information System (PROMIS) Physical and Mental Health Scales 11 (health-related quality of life).

Time Points of Interest:
Baseline, immediately post-intervention, 6 and 12 months post-stroke Responsibilities include: Statistical design of study data verification, developing and implementing data management procedures including the data flow and procedures for data entry, error identification and correction, and preparing quarterly reports on enrollment, participant status (e.g., withdrawals), AEs, and independent safety monitoring body reports.

Data and Safety Monitoring Committee:
Susan Stark, PhD, OTR/L Responsibilities include: reviewing and evaluating the study data to ensure participant safety, study conduct, progress, and efficacy, and making recommendations regarding the continuation, modification, and termination of the trial.

Abstract
Stroke is a leading cause of disability in the US. Most stroke survivors have difficulty performing daily activities and participating in the community. Efficacy of interventions that address the chronic needs of stroke survivors has been identified as a high priority for stroke research. A gap in care exists at the point of transition from inpatient rehabilitation (IR) to home, when survivors encounter new environmental barriers due to the cognitive and sensorimotor sequelae of stroke. Resolving these barriers and improving independence in the community have potential to significantly improve stroke survivors' long-term morbidity. The proposed study investigates the efficacy and safety of a novel, enhanced rehabilitation transition program to reduce environmental barriers and improve daily activity performance and community participation. Community Participation Transition after Stroke (COMPASS) uses two complementary evidence-based interventions: home modifications and strategy training delivered in the home. Home modifications provide environmental support to compensate for impairments. Strategy training enables patients to identify and prioritize ADL problems, identify barriers to performance, and develop strategies to resolve barriers. The complementary effects of these therapies at a novel point of care offer a robust intervention for a current service gap. The primary aim of this phase IIb, single-blind, parallel-group, randomized controlled trial is to investigate the efficacy of COMPASS during the transition from IR to home. Participants will be randomized to receive either COMPASS or attentional control. We hypothesize that COMPASS participants will demonstrate significant improvements in community participation and ADL performance and a significant reduction in environmental barriers versus control. In preparation for a phase III multicenter trial, we will explore additional candidate study endpoints and establish intervention safety by examining the rates of falls and hospital readmissions. Finally, we will conduct a process evaluation examining outcomes such as reach, cost, fidelity, and adherence to aid interpretability of the trial and implementation. The aims fill critical gaps in stroke rehabilitation evidence by investigating the efficacy, safety, and implementation of an intervention targeting the transition from an acute to a chronic condition. The results of this trial will provide important information about the long-term participation and environmental barriers of stroke survivors. The project has the potential to resolve the significant unmet need of disability after stroke.

Background
Stroke is highly prevalent, costly, and disabling. Stroke is a leading cause of serious long-term disability in the US. 12 Half of stroke survivors are dependent on caregivers to perform their ADLs. 13,14 Unless a solution is identified to improve the long-term outcome of stroke survivors, annual US costs attributed to stroke are projected to increase to $240.67 billion by 2030. 15 Patients report "waiting" at home 6-12 months after discharge for "recovery" before attempting to resume participation in daily activities. 16 The transition from IR to home is an important window of opportunity for intervention. 17,18 Resumption of previous activities immediately after discharge, 19 at a time when people with stroke report struggling to reestablish daily routines, 18 can improve immediate and long-term community reintegration.
Transition home from inpatient rehabilitation (IR) is understudied, but face validity exists for treatments that reduce environmental barriers and improve daily activity performance. A paucity of evidence exists for interventions targeting the transition period from IR to home. In a systematic analysis of the Guidelines for Adult Stroke Rehabilitation and Recovery, 20 we mapped COMPASS elements to the evidence recommendations. Transition interventions including environmental modifications (provision of adaptive equipment) and problem-solving approaches like those proposed in COMPASS are rated as having high face validity and low risk but lack rigorous evidence of efficacy. COMPASS also directly addresses goals in the NIH Plan for Rehabilitation Research. 21 The plan has prioritized research that provides an understanding of environmental barriers and individual participation outcomes in real-world settings and specifically calls for research using self-management strategies like strategy training that can be implemented in community settings to achieve patient independence and improve caregiver outcomes.

Reducing environmental barriers improves daily activity performance but is unproven among stroke survivors.
Providing environmental support is a potent and immediate strategy to improve daily activity performance. In our systematic review 22 and home modification practice guideline 23 (including 35 articles with levels I-III evidence), strong evidence shows that home modifications improve daily activity performance [24][25][26][27][28][29][30][31][32][33][34][35][36] and reduce falls 37-43 among adults and older adults with disabling conditions. Notably, during this review, we did not discover a study of the efficacy of environmental modifications for stroke. That may explain why environmental modification programs are not part of IR, despite face validity that environmental interventions might be an effective approach to reduce excess disability for this population.

Strategy training shows promise in acute IR for stroke but is unproven at the point of transition home.
Strategy training enables patients to identify and prioritize daily activity problems and to identify barriers to performance and strategies to resolve the barriers. The approach leads to generalization and learning rather than skill attainment. 44,45 OT scientists have demonstrated the efficacy of guided strategy training interventions for patients with neurological impairments, including acute stroke. [46][47][48][49][50][51][52] Strategy training is consistent with new models of inpatient stroke care such as those tested by Skidmore 46-48 but is unproven in transition programs for stroke survivors.
We will provide evidence of the efficacy and safety of a transition program designed to increase the independence of stroke survivors. COMPASS is a manualized intervention that includes pre-and postdischarge visits in the home to remove barriers through home modification and strategy training. The intervention targets community reintegration, or the ability to resume daily activities in the home and community. Our preliminary data provide evidence of acceptability and feasibility for delivery during IR (versus home health, for which regulatory policies and practice patterns preclude a focus on community participation). 53,54 If effective, this program will reduce disability in daily activity performance and improve participation outcomes. We will explore anticipated downstream effects of the intervention including rehospitalization rates and caregiver burden.
The conceptual model of this intervention ( Figure 1) is the World Health Organization's International Classification of Functioning, Disability and Health (ICF). 55 The ICF describes the mechanism of COMPASS. Survivors of IS experience decreased cognitive, sensory-motor, and emotional function. After stroke, the home and community environments of survivors pose barriers that prevent successful performance of daily activities and participation. Intervening to remove barriers and enable survivors to use problem-solving strategies to overcome barriers will improve daily activity performance and participation outcomes. This intervention is designed to augment current rehabilitation practice focused at the body structure and function domains. This conceptual model is empirically supported by our recent work exploring the role of environmental barriers in function. 56

Research Design Overview
The objective of this phase IIb RCT is to determine the efficacy and safety of a compensatory intervention designed to improve daily activity performance and participation outcomes in preparation for a definitive trial. We will test the hypothesis that a manualized, theoretically grounded, and empirically supported transition intervention (COMPASS) is superior to an equivalent dose of attentional control (AC) for daily activity performance and participation outcomes with high adherence. The hypothesis is based on our previous work, 35,36 the work of others, 27,57,58 and our preliminary data. We will recruit 180 patients currently undergoing IR for IS or focal ICH and randomize them to receive five in-home sessions (one pre-discharge visit and four post-discharge visits) and two booster contacts of the enhanced rehabilitation transition program or AC. Our primary endpoint of participation and secondary outcome of ADL performance will be assessed by certified blinded raters at baseline, immediately after intervention, and at 6 and 12 months after stroke. Measures of mobility, cognition, and depression, will be used to characterize the clinical phenotype or body structure and function of the participants as important covariates. We will also examine important process outcomes such as cost, satisfaction, adherence, safety, and caregiver burden.

Study Participants
Stroke patient inclusion criteria: (1) aged ≥50 years, (2) acute IS or focal ICH diagnosis, (3) independent ADLs prior to stroke (premorbid Modified Rankin Scale Score ≤2), (4) plan to discharge to home. Stroke patient exclusion criteria: (1) severe terminal systemic disease that limits life expectancy to < 6 months, Blue boxes represent intervention. Blue lines and green boxes represent the hypothesized mechanism of action. By reducing environmental barriers, activity and participation outcomes will improve. We will measure covariates in the domains of body function, personal factors, and health condition.
(2) previous disorder (e.g., dementia) that makes interpretation of the self-rated scales difficult or SBT score of 10 or less (indicating significant cognitive impairment), (3) moderate-to-severe aphasia as determined by the NIHSS Best Language rating of 2 or more, (4) reside in congregate living facility.
Caregiver inclusion criteria: (1) primary informal (family or friend) caregiver of a stroke patient enrolled in the COMPASS II study, (2) aged ≥18 years Caregiver exclusion criteria: (1) non-English speaking

Recruitment
We will recruit participants through Barnes Jewish Hospital (BJH) and TRISL.
At BJH, potential participants will be screened and referred to the study by the Stroke Patient Access Core. At TRISL, rehabilitation staff and physicians will identify potential participants daily and, with permission, refer to the study team. We will target survivors of IS and focal ICH but not subarachnoid hemorrhage. Outcomes for IS and focal ICH are generally better, but those with subarachnoid hemorrhage are associated with considerably increased mortality and decreased function after stroke. 59

Informed Consent
Stroke Survivors: A study team member will visit all patients/caregivers who meet the inclusion criteria and invite them to participate in the study. The informed consent form will be administered to all patients interested in participating in the study. During the consent process, participants will: (1) have what the study is about and what is expected of them explained in detail, (2) discuss potential problems that could interfere with participation, (3) have their questions answered, and (4) receive a summary of the study and contact information for the PI and study coordinator. Written informed consent to participate in the study will be obtained before any test or measurements are performed. The consent form will be signed by a witness and will be stored in the office of the PI under double locks. Written informed consent to participate in the study will be obtained before any tests or measurements are performed. Participants will be advised in the consent form that there is a possibility that their medical research record, including identifying information, may be inspected and photocopied by officials of federal or state government agencies and the Washington University Human Research Protection Office (HRPO).
Caregivers: We are also recruiting the primary informal caregivers of enrolled participants. Caregivers must be over 18 years of age and speak English in order to participate.

Baseline Home Visit (T1) for All Participants
Prior to randomization, baseline activity assessment will be conducted in the home before discharge by an OT using methods established in our pilot study. The intervention is designed to augment rather than replace usual care, so both groups will receive the in-home assessment (considered best practice). Medically stable participants who are eligible for a therapeutic pass will participate in the home visit. The OT provider will meet the participant and their family at TRISL for car transfer training before departing for the participant's home. The visit will be scheduled to avoid disrupting ongoing inpatient therapy and to accommodate family schedules. If the participant is not able to transfer safely into a family vehicle or if no family or vehicle is available, we will charter a wheelchair-accessible taxi. If the entrance to the home is not accessible, a temporary ramp will be installed. We were able to install ramps within 48 hours of request in our pilot trial. Onsite, we will use I-HOPE to identify environmental barriers in the home and to establish baseline activity levels. Results of the T1 home assessment (for participants in each arm) will be shared with the rehabilitation health care team, because home assessment is considered best practice for patients whose disposition is home. 60

Randomization and Blinding
Age is a negative marker for functional recovery after stroke, and functional status is a strong predictor of recovery. 61,62 The Functional Independence Measure (FIM) will be used to determine functional status poststroke. Randomization will be balanced using the participant's age and FIM score.
Participants will be allocated using a 1:1 ratio via adapted randomization sequences generated a priori by the study statistician, Dr. Yan, using a computerized formal probability model. The allocation ratio will be maintained at periodic intervals. Randomization sequence concealment will be achieved by query of the REDCap (Research Electronic Data Capture) system. 63 After baseline assessment (T1) is securely uploaded and locked and stratification variables are entered, the interventionist will elicit the treatment assignment in the field in real time using a secure data connection to REDCap. This will permit prioritization of problems for participants assigned to the treatment group.
Upon randomization, participants will be assigned to an interventionist by the treatment coordinator. Although the results of the home assessments will be shared with the rehabilitation staff, IR staff will be blinded to the group placement so that they do not modify their inpatient or discharge treatment plan. All outcomes will be assessed by a blinded rater.

Phone Assessment T1b
To allow time for participants to adjust and personally assess their participation, T1 participation assessments will be conducted for both groups by telephone 2 days after discharge from IR. A rater who is blinded to allocation will conduct the baseline assessment for the primary, secondary, and exploratory endpoints.
Upon completion of T1b phone assessment, the participant will receive visits 2-6 over an 8-week period. The home modification group will progress from daily activity performance to community participation.

Primary Intervention
COMPASS is a complex intervention that combines two evidence-based treatment strategies at a new point of care (transition from IR; Table 1). The objective of home visits by an OT interventionist is to remediate barriers in the home and community that influence daily activities and community participation. The COMPASS manual fully defines and justifies each element of the intervention and has been iteratively revised during the pilot phase. The treatment will include a set of one pre-discharge 64,65 and four 75-minute post-discharge 66 visits ( Table 1). The intervention is followed by two booster sessions. Data from the T1 home barrier assessment, demographic assessments, and rehabilitation staff assessment of functional abilities (e.g., transfers) will be used by the OT interventionist to develop an environmental modification intervention plan. Environmental modifications will be installed before discharge if possible. The intervention (not reimbursable) will be provided at no cost to the participant. Problem areas addressed are participant-specific (tailored), but the process to identify and address the target area is systematic. All participants will receive identical intervention components. The standardized components include assessment, identification of five problematic activities (and environmental barriers), identification of three solutions (for each problem), implementation of a solution set selected by the participant, training, and active practice of daily activities in one's own home and community.
Defining the Treatment (Table 1) The treatment theory guiding the intervention is a competence-press model that posits that removing environmental barriers (e.g., grab bars near the toilet, using accessible transportation) matched with the patient's pattern of functional loss will improve the outcomes of daily activity performance and participation. 67 The two essential components 68 of COMPASS are (1) home modification and (2) strategy training. Both address barriers in the participant's own home and community environment. 69 The intervention is tailored [70][71][72] by the interventionist to each participant's pattern of functional loss and unique home and community environments. Tailoring is necessary, given the heterogeneity of environments. 22 The participant will engage in active practice with the OT provider. Home modifications coupled with active practice of daily activities improves functional performance of persons with disabilities. 8,25,27,69,73 Timing: Pre-discharge Environmental Modification Intervention. Using standardized assessment and manualized procedures, environmental modifications will be installed before discharge. The intervention (not reimbursable) will be provided at no cost to the participant. Post-discharge Activity and Participation Intervention. On completion of T1b phone assessment, the participant will receive visits 2-5 over an 8-week period. Interventions will progress in complexity from daily activities to community participation based on the participant's goals. A booster session will be provided at 4 and 5 months after intervention over the phone; participants may request a home visit.

Attention Control Group
The control group will experience the same effects of time and attention in the home but no effect on the outcome of interest. 74 Because this study examines the efficacy of a new intervention, there is no opportunity for a usual care arm with dose equivalency. Because 75% of stroke survivors and 78% of caregivers report that their informational needs were not met in the hospital, 75 we developed and piloted a tailored stroke education program for AC. An OT practitioner will deliver the program in accordance with "Evidence-Based Educational Guidelines for Stroke Survivors after Discharge Home." 76 Topic order is determined by participants. Four 75-minute sessions will be provided. Topics include stroke symptoms, risk factors and preventing stroke recurrence, nutrition, managing emotions, sleep, fatigue, pain, social support, and sexuality. Written materials from the National Stroke Association and the American Stroke Association are provided. Environmental barriers will not be addressed in the educational sessions.

Follow-up Period for All Participants
Stroke and caregivers: Home visits T2 (immediately after intervention), T3 (6 months after stroke), and T4 (12 months after stroke). The follow-up activity, participation, and process assessments will be conducted in the home. All follow-up assessments will be completed by a blinded rater who did not conduct the T1 evaluation, as new home modifications could reveal group assignment. Distal time points will explore the permanency of any treatment effect on participation. If the participant is not available for in-home assessment, the primary endpoint will be collected by phone. Falls and healthcare utilization are collected monthly by phone (using an automated system).
For participants who experience a delay in the study timeline (such as hospital admission, travel, short-term admission to a skilled nursing facility, etc.), the timeline will be adjusted accordingly. For example, if someone returns to the hospital during the intervention period and is in the hospital for 4 weeks, their T2 follow-up will be pushed back 4 weeks.

Outcome Assessments for Baseline & Follow-up Home Visits
In-Home Occupational Performance Evaluation (I-HOPE). The I-HOPE will be used to measure current activity patterns of participants, identify activities that are difficult but important to them, and identify the environmental barriers that influence those activities (person-environment fit). The I-HOPE is a multi-step assessment that evaluates the performance of older adults doing 42 activities in the home. Using labeled pictures of each home activity (e.g., washing dishes, getting in and out of bed), older adults identify activities difficult for them to perform and self-rate their performance and satisfaction with performing each of the activities. A trained rater (e.g., OT interventionist) observes the older adult performing the activity and quantifies the personenvironment fit (or misfit). The instrument yields four subscales: the activity subscale (α = 0.78; 6 items); the performance subscale (α = 0.85; 38 items); the satisfaction subscale (α = 0.77; 8 items); and the environmental barrier subscale (α = 0.77; 6 items). Intraclass correlation coefficients were calculated for the I-HOPE subscales on a sample of 10 participants, and scores ranged from 0.94 to 1.0 for raters. 77 Barthel Index (BI). The BI is a reliable and valid assessment that will be used to assess a person's ability to complete 10 ADLs (feeding, bathing, grooming, dressing, bowel control, bladder control, toileting, chair transfer, ambulation, and stair climbing) and the amount of assistance they need to complete these activities. We will use this information to examine changes in participants' functional performance in this study. There is excellent correlation and agreement for stroke patients between the FIM motor and 10-item Barthel at admission and discharge (r > 0.92; ICC > 0.83). 78 Stroke Impact Scale (SIS). The SIS is a health-related quality of life measure that quantifies the impact of stroke on a participant's life, via questionnaire. Fifty-nine items are measured across eight domains: strength, hand function, mobility, ADLs and instrumental activities of daily living (IADLs), emotion, memory, communication, and participation. Performance is self-reported based on the difficulty the participant experiences with each rater. Each item is scored on a five-point scale, and total domain scores are based on a scale from 0 (poorest outcome) to 100 (best outcome). A final question asks the participant's perception of their recovery post-stroke on a visual analog scale from 0 (no recovery) to 100 (full recovery). Rasch analysis, which is widely used to validate self-report outcome measures, established the SIS's validity and reliability and concluded that it has good psychometric characteristics. 79 Reintegration to Normal Living Index (RNLI). The RNLI is a disability-related quality of life instrument that will be used to measure participants' satisfaction with their home and community participation and has been validated on a population of community-dwelling individuals with chronic conditions. 2 Participants will read 11 statements related to their monthly activity patterns and assign each one a score based on a 10-point ordinal scale with 1 indicating "does not describe my situation at all" and 10 indicating "fully describes my situation." 11 will be used to measure health-related quality of life. PROMIS is a publicly available system of highly reliable, precise measures of patient-reported health status for physical, mental, and social well-being.

Patient-Reported Outcomes Measurement Information System (PROMIS) Physical and Mental Health Scales
Perceived Stress Scale (PSS). 9 The PSS will be used to measure perceived stress in the caregiver participants. The PSS is a global measure of perceived stress designed for use in community samples. The participant rates feelings and thoughts during the past month using a Likert scale from 0 (never) to 4 (fairly often). The items on the PSS were designed to assess the degree to which respondents find their lives unpredictable, uncontrollable, and overloading.
Caregiver Inventory (CGI) 80 . The CGI will be used to measure self-efficacy in caregiving. The CGI is a valid and reliable measure consisting of four subscales-managing medical information (three items), caring for the care recipient (seven items), caring for oneself (five items), and managing difficult interactions and emotions (six items)-and will be collected at all follow-up study visits to examine whether the study had an impact on caregiver burden.

Demographic Assessments
Demographics. Questions such as living situation, education level, and employment status will be used to gather demographic information from the participant.

Geriatric Depression Scale, Short Form (GDS-SF).
The GDS-SF will be used to assess depression levels in stoke participants. 81 The GDS-SF is a 15-item self-report questionnaire specifically designed and validated with the older population and correlates highly with the original 30-item GDS (r=.84). 81 Participants answer yes-orno questions about their feelings in the past week. Total scores range from 0 to 15, and scores of 5 or more indicate probable depression.

Intention-to-Treat Analysis
We will perform our analyses using an intention-to-treat paradigm. We will perform exploratory data analysis looking for extreme or otherwise unusual values. Non-normal and heteroscedastic data will be transformed as necessary.

Baseline Analyses
We will use unpaired t tests and chi-square tests to compare TØ and baseline characteristics in the two groups for descriptive information (except when statistical assumptions are not met, in which case we may use Wilcoxon or Fisher exact tests).

Missing Data
We expect missing values in outcome measures because of dropout, death, missed assessment, or nonresponse. Our main analysis, a linear mixed-effects model, accommodates missingness due to treatment, prior outcome, or baseline covariates missing at random. 82 Assuming that missing data occur at random, inferences will be valid even if we have differential dropout by intervention arm. If the missing data mechanism is not ignorable (ie, missing is not at random), then mixed-effects selection models or pattern-mixture models will be used. 83

Primary Analysis
Our primary analysis (testing primary hypothesis) will be based on a linear mixed model using baseline and 12month RNLI s, accounting for the correlation between a participant's repeated measurements over time. The fixed-effect portion of the model will have the form Yit = β0 + β1 × 12 months + β2Group + β3Group × 12month, in which Yit is the RNLI score for participant i at baseline (time 0) and 12 months (time 1), and Group indicates study arm. In this model, the baseline RNLI is modeled as a dependent variable. 27 For improved precision, the model will be adjusted for baseline covariates including race, sex, depression, and length of hospital stay if an imbalance in covariates between arms is observed in baseline analyses. In this model, β0 is the mean RNLI score for the control arm at time 0, and β1 is the change in the mean RNLI from baseline to time 1 for the control arm; β2 is the mean RNLI score for the treatment arm at time 0, and β3 is the change in mean RNLI from time 0 to time 1 for the treatment arm. The primary hypothesis is that the difference in the change in RNLI scores from time 0 to time 1 between arms will be tested by examining β3, which estimates the difference.

Secondary Analyses
For the secondary analyses of the change in SIS score at 1 year and the change in I-HOPE at 1 year, we will use the same approach as for the primary analysis because these two outcomes are also continuous. We have overall type I error control for testing the 1-year change in these three analyses at the design stage. The significance level for testing is 0.016. In addition to comparison of 1-year change, we will extend the model by including scores immediately after intervention and at 6 months to see whether the difference in outcomes i s achieved at those time points. Depending on the form of time variable in the model, we will use appropriate regression coefficients or a linear combination of the regression coefficients to determine the difference in change of these scores between arms at certain time points. Interpretation of these results should be cautious because we do not have the type I error controlled for these analyses. Because it is possible that severity of functional impairment after stroke may impact response to treatment, we will analyze impact of functional impairment on response to treatment. We will examine functional impairment by group interaction to examine possible differential intervention effects of functional impairment on community participation and performance of daily activities.

Safety
To determine whether the intervention poses no greater risk than AC, we will examine the differences in numbers of falls and rehospitalizations between groups. The statistical models for count data will be used for analyses of these two outcomes. Using the number of falls as an example, we will fit a Poisson regression model (with overdispersion adjustment if necessary), in which a dummy variable for the intervention arm is used. The regression parameter estimate for this dummy variable is the log of rate ratio of falls for COMPASS versus control arm, and exponentiation of the regression parameter estimate is rate ratio. Using the parameter estimate and its standard error, we can construct a two-sided 95% CI for the rate ratio. We expect the confidence interval for rate ratio will include 1, indicating no significant difference in the fall rate between the two groups.

Process Outcomes
Evaluate acceptability and feasibility to aid in the interpretability of the trial. COMPASS will have high acceptability (80% retention), high fidelity by therapists (95% of elements and 90% of dose delivered), low safety risk (no increased rate of falls or health care use compared with the AC group), and high adherence (80% of modifications in use) at 12 months. We will conduct between-group comparisons of process endpoints collected at each time point (time to first fall, number of injurious falls, health care utilization rate, dosage delivered, and adherence rate) using unpaired t tests or chi-square tests. We will compare the characteristics of patients who complete the assigned intervention with those who do not for differences in stroke severity and comorbidities. Descriptive statistics will be used for costs per participant and adherence.

Sample Size Calculations
The study is designed to have 80% of power to reject three null hypotheses of equal mean changes in the primary and secondary endpoints (RNLI, SIS, and I-HOPE) using a two-sided, two-sample, unequal-variance t test with overall type I error <0.05. Three alternative mean changes (standard deviation We will enroll 180 patients to account for a 30% attrition rate. This magnitude of between-group difference is considered clinically meaningful based on prior relevant literature and is achievable based on our pilot study. Sample size calculation includes correlation between baseline and follow-up measures and is based on analysis of change scores, which is equivalent in efficiency to the proposed analytic model. Based on the number of stroke patient participants, we will enroll an equivalent number of caregivers.

Strengths and Limitations
The COMPASS study has important strengths. The intervention is grounded in sound theory and evidence. The intervention has made a significant impact on the ability of people with disability to improve their ability to complete daily activities. The intervention will now be tested as part of a community transition program to help stroke survivors complete daily activities at home, as well as transition back into the community.
The research proposed in this application is innovative because it represents a departure from the status quo by delivering a transition program targeting environmental barriers and providing skill training in the "real world" at the point of discharge from IR. Our preliminary studies suggest that this approach will be highly effective in improving the performance of daily activities and participation outcomes, which could lead to reduced readmission and improved safety.
An additional strength of this study is that it is being conducted in partnership with TRISL, which is managed by HealthSouth, a national health care organization that specializes in providing rehabilitation services for individuals. If this trial is feasible, it can be rolled out nationally with HealthSouth because that partnership already exists and the mechanisms are already in place for a national, multisite trial.
There are a few limitations that exist for this study. The first is low generalizability of findings with only a single site. However, participants at TRISL are comparable nationally regarding demographics, length of stay, and treatment; the only exception is that our sample has a higher percentage of Black patients. In addition, TRISL and HealthSouth have policies in place that limit the amount of contact outside study staff can have with individuals prior to recruitment. In some cases, the policies are even stricter than institutional review board (IRB) policy and make recruitment difficult.

Benefits
There will be considerable benefits to the participants enrolled in the proposed study. All participants will receive a complete home evaluation provided by a registered and licensed OT interventionist. In addition, participants in the active treatment group will receive free home modifications, while participants in the AC group will have a trained research assistant deliver stroke education.

Risks
Potential risks of research participation-physical, psychological, financial, and legal risks, among others-are considered minimal (45CFR46.404). In-home evaluations and assessments of barriers may result in fatigue or aggravation. In addition, some questions may touch on emotionally sensitive issues that could cause anxiety or other forms of emotional stress. The performance-based testing involves observation of everyday activities, which may result in fatigue or embarrassment. In rare instances, the evaluation protocol could result in a fall. Participants will be told that their involvement in this research study is voluntary and that they may choose not to participate or to withdraw their consent at any time. Withdrawal from the study will not at any time affect the commitment of the clinician to administer care, and there will be no penalty or loss of benefits to which participants are otherwise entitled. Participants who undergo the study visits will be given the option to reschedule the visit or take a break at any time during the study if necessary. There is little legal risk to participating in this research. All research-related information will be kept confidential and accessible only to authorized members of the research team.

Minimization of Risks and Confidentiality
To protect against and minimize potential risks, participants will be carefully screened and evaluated for eligibility by the research coordinator. To avoid or minimize symptoms of fatigue, agitation, or emotional distress due to testing, participants will be instructed to notify the rater or interventionist if they experience any discomfort. They will also be periodically questioned about their tolerance for the tests/intervention. Testing and interviews will be terminated if participants develop fatigue, agitation, or emotional distress. Participants will be trained in the use of all modifications by a licensed and registered occupational therapist. When photographing the home, only environmental barriers and changes will be collected. Images of participants will not be collected. An ID number will be assigned to each participant. All data collected from a participant will be labeled with the ID number. All participant electronic and hard-copy data will be kept under double-lock protection. All hard-copy forms that contain personal identifiers (e.g., name, address, phone number) will be stored in a separate, locked file drawer under double-lock protection. No publication or presentation of the study data will uniquely identify or provide sufficient information to uniquely identify participants.
Risks during the home visit will be minimized by having licensed and trained interventionists available to monitor safety during the home assessment. If needed, an additional trained assistant will accompany the interventionist and family on the home visit. Gait belts will be used at all times while in the home. If needed, temporary ramps will be installed to facilitate entry To guard against unauthorized data access, all shared-use computer systems at the Washington University School of Medicine are protected with passwords, which are changed at 4-month intervals. Only individuals with a particular "need to know" status are given access, and system privileges are carefully restricted. All personal computers to be used in the Administrative Unit are located within a secure area, and the system is locked when not in use. SAS and SPSS software packages will be used for data management and analysis. Datasets generated from these programs will be password protected, which will make accessing study data difficult even in the event that unauthorized computer access occurs. Systems connected to the Ethernet are carefully controlled, and all systems without Ethernet access control are insulated from the backbone by bridges or routers. The Ethernet cable itself is routed only through secure passageways.
Data are directly entered into a REDCap database. REDCap servers are securely housed in an onsite, limitedaccess data center managed by the Division of Biostatistics at Washington University. All Web-based information transmission is encrypted. All data are stored on a private, firewall-protected network. All users are given individual user IDs and passwords, and their access is restricted on a role-specific basis. REDCap was developed specifically around HIPAA security guidelines and is implemented and maintained according to Washington University guidelines. Study data will be collected via tablet in the field and managed using REDCap electronic data-capture tools hosted at Washington University. REDCap is a secure, Web-based application designed to support data capture for research studies.
In order to complete the automated phone calls, first names and phone numbers will be shared with Twilio, an online communication software designed to make automated phone call surveys. To complete the call, Twilio will access first names and phone numbers via REDCap. Survey responses will be temporarily stored by Twilio, transferred into Washington University's secure databases (REDCap), and deleted from Twlilio as soon as possible.

Adverse Event Reporting and Safety Monitoring
All SAEs will be reported to the HRPO in the following time frames: (a) death-immediately, (b) life-threateningwithin 7 calendar days, (c) all other SAEs-within 15 calendar days using the Electronic Serious Adverse Event Reporting System. Should an SAE occur that increases the risk to the participants, the study will be stopped, an investigation will be conducted, and a findings report will be generated before the study is resumed.
Dr. Stark will be responsible for reviewing study progress and outcomes including recruitment, data quality, safety, and efficacy. Quarterly reports will be reviewed by the study investigators. Because risk in the proposed study is considered minimal, the data monitoring plan will include continuous, close monitoring by the study investigator with prompt reporting of any AEs. Given the small number of subjects undergoing treatment, problems will become more readily apparent through close monitoring of individual participants. In this study, Dr. Stark will monitor the study for AEs, adherence to the protocol, and safety.

Premature Study Termination
Preliminary study data will be monitored by the data management team for any potentially harmful outcomes. If interim data raise significant safety concerns, the trial will be ended early.

Indemnity
Washington University School of Medicine is responsible for any non-negligent damage incurred as a result of participating in the COMPASS Trial. The indemnity is renewed on an annual basis. Washington University School of Medicine assures that it will continue renewal of the indemnity for the duration of the trial.

Ethics and Dissemination
This protocol and the template informed consent forms will be reviewed and approved by the Washington University IRB with respect to scientific content and compliance with applicable research and human subjects regulations. All study personnel involved in the conduct of this research will receive the required education on the protection of human participant rights.
On publication of the study results, participants will be invited to attend a community meeting, during which the results of the study will be reported. The information will be repeated during three community sessions to be held during daylight hours. Participants will receive a mailing announcing the meetings and summarizing the study findings. The location of the meetings will be in a fully accessible auditorium with accessible parking and access to public transportation. A written report will be distributed, and the results will be presented by the study investigators, followed by a question-and-answer period. Refreshments will be served, and participants and their family members will be thanked for their generous support of the project.