Protocol for a Randomized Controlled Trial of Low-intensity Physical Activity for Frail Older Adults: Promoting Seniors’ Health with Home Care Aides (Pro-Home)

Abstract

Regular participation in physical activity benefits older adults physically and mentally. However, the availability and assessment of physical activity programs that are safe and appropriate for homebound older adults at risk for nursing home admission are limited. Here we describe the protocol for a randomized controlled trial that examines the effectiveness of a gentle physical activity program. Delivered by home care aides who regularly help hard-to-reach older home care clients with housekeeping and routine personal care services in the home, this program is implemented in a real-world context of caregiver-client dyads in a Medicaid-funded home care program. The trial uses a two-group repeated measures design (baseline, Month 4, and Month 8) with 300 pairs of eligible home care clients and their home care aides. The results from this trial could provide evidence and guidelines for a new model of home care, which would facilitate the working together of older home care clients and their home care aides to maintain or improve the functional status of nursing home-eligible older adults.

1. Introduction

Regular physical activity (PA) participation benefits older adults physically and mentally [1–4]. However, community-dwelling older adults who have difficulty with walking, bathing and other basic daily activities often have limited PA. Evidence is limited regarding the benefits of PA programs that are safe and appropriate for this population of nursing home-eligible older adults [2, 5–7]. Medical professionals (e.g., physical therapists) can deliver PA programs that are beneficial for this population [2], but such programs are too expensive for wide dissemination, especially for Medicaid home care programs. Home care aides (HCAs) who regularly provide housekeeping and personal care have the potential to promote their clients’ PA and health. However, HCAs’ abilities have been underutilized in the health care system. To address this opportunity, “Promoting Seniors’ Health with Home Care Aides: A Randomized Controlled Trial” (Pro-Home RCT) tests the effectiveness of a gentle, low-intensity PA program delivered by HCAs for community-dwelling frail older adults with institutional care-level needs in a Medicaid-funded home care program, building on multiple pilot projects [8–12].

Healthy Moves for Aging Well (Healthy Moves) is a low-cost PA program, specifically designed to safely enhance the PA level of nursing home-eligible Medicaid older clients [13]. Healthy Moves consists of two major components: a brief motivational enhancement and a PA intervention (3 chair-bound moves). Healthy Moves was originally delivered by case managers and trained lay coaches. We adopted Healthy Moves because (1) it was among the few PA programs targeted to frail community-dwelling older Medicaid recipients with nursing home-level care needs, (2) it has theoretical [14–16] and empirical bases [17–21], (3) the program information was publicly available at the time of our adoption, and (4) it is simple enough so that we could adapt it appropriately for HCAs to learn and deliver to their clients. We piloted Healthy Moves with 54 HCA-client dyads in a Medicaid-funded home care program. The pilot study demonstrated the program feasibility and produced promising results, including increases in self-reported and performance-based daily function [11, 21]. During the pilot, HCAs expressed interest in doing other activities, especially word puzzles. Active Mind for Aging Well (Active Mind), an attention control condition, is a thinking activity developed around word puzzles by the Pro-Home team to parallel Healthy Moves.

This RCT is conducted in a real-world intervention delivery setting [22–24], consistent with the long-term goal of Pro-Home to enhance the quality of health care. This RCT is characterized as a Stage III and Stage IV trial according to the National Institute of Health Stage Model for principle-driven behavioral intervention development [25]. This is a Stage III trial in that it assesses a behavioral change intervention in a community setting while maintaining as high a level of control as possible to establish internal validity. It is also a Stage IV trial in that it examines the intervention in a community-based home care setting with community providers (HCAs) while maximizing external validity. This paper describes the research protocol of the Pro-Home RCT.

2. Materials/methods

2.1. Research aims and study hypothesis

Our long-term goal is to develop a sustainable health promotion program delivered by HCAs that can be used by community-based organizations, such as home care agencies, and state units on aging. The objective of this research study is to assess the effectiveness of a gentle, low-intensity physical activity program, Healthy Moves, delivered by HCAs for home care clients who need nursing home-level care.

Pro-Home has two aims. The first aim is to determine whether Healthy Moves delivered by HCAs better preserves function and well-being among older home care clients compared with clients who receive the attention control. Our hypothesis is that home care clients receiving the Healthy Moves intervention will maintain physical function, whereas the function of clients receiving Active Mind will decline. The second aim is to understand how Healthy Moves delivered by HCAs works in a real-world Medicaid-funded long-term service and support setting, specifically (a) for whom the program is beneficial, (b) the extent to which the program can reach the target population, (c) the extent to which participants drop out of the program, (d) the extent to which program participants maintain behavioral changes introduced by Healthy Moves, and (e) the program’s cost effectiveness.

This study is implemented in a real-world home care setting that cannot provide fully controlled conditions (see 2.4, Study context and participants). Efficacy refers to the performance of the intervention under fully controlled conditions. In a setting with absence of control, this study investigates whether the intervention works. In this regard, this is an effectiveness study, though effectiveness subsumes efficacy and we assess the claim of efficacy using the evidence on effectiveness.

2.2. Study sponsorship and Institutional Review Board approval

This work is supported by the National Institute on Aging of the National Institutes of Health under Award Number R01AG053675. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. This trial is registered (ClinicalTrials.gov Identifier: NCT03301116) and approved by the Institutional Review Board of the University of Illinois Chicago.

2.3. Overview and study design

A RCT with a two-group repeated measures design (baseline, Month 4, and Month 8) is being conducted with 300 pairs of eligible clients and their HCAs. The experimental group receives Healthy Moves for Aging Well, and the control group receives Active Mind for Aging Well. In the latter condition, HCAs are trained to deliver easy-to learn thinking activities (word puzzles) that are popular among older adults as a leisure activity. HCAs deliver the program to their clients at the first home visit after the training. After the first day, HCAs provide their client-participant with reminders and encouragement to do the activity for the next 4 months. The primary outcome is older adults’ daily activity function (see below, section 2.15, Primary outcomes).

2.4. Study context and participants

Pro-Home is being conducted in the context of in-home services for older adults within the Illinois Department on Aging Community Care Program (CCP) [26]. CCP provides home and community-based services to approximately 76,000 older adults aged 60+ with assets ≤$17,500. This state program is one of the Medicaid 1915(c) waivers for home and community-based services, which is the main mechanism through which states finance home and community-based services for low-income seniors and people with disabilities, and is rapidly transitioning into managed care [27]. HCAs provide personal care (e.g., cleaning, preparing meals, shopping, dressing and bathing). Qualifications for HCAs include a high school diploma or general education diploma, 1 year of relevant experience (employment in a comparable human service capacity, care for a dependent child or adult family member), or demonstration of continued progress toward a general education diploma [28]. Older CCP clients can receive in-home services from their relatives, if they qualify and become home care agency employees. The state requires HCAs, including family HCAs, to attend 24 hours of initial pre-service training before employment, and thereafter, a minimum of 12 hours per calendar year of interactive in-service training approved by the provider agency. We recruit study participants (HCAs and clients) through two of the largest CCP home care providers in Illinois. HCAs in large home care providers are unionized and the labor union (SEIU Healthcare Illinois and Indiana) facilitates our HCA recruitment. Care Coordination Units (CCUs) are community-based organizations contracted by the State to provide case management services. CCUs’ care managers assess each senior’s care needs and develop care plans, which dictate the kind and amount of services that each older person receives. Thus, their involvement is critical for our program’s success. The research takes place in the Greater Chicago Illinois Area.

2.5. Eligibility

Older adults aged 60+ are eligible for this study if they are eligible for CCP [26], receive home care from one of our partner home care agencies through the Illinois Department on Aging CCP or through managed care plans [27], are English or Spanish speaking, have a cognitive status sufficient to follow directions and respond to survey questions, are assessed to be able to perform Healthy Moves and Active Mind (i.e., be able to sit in a chair independently), and have a HCA who is willing to participate in the study. HCAs are eligible for this study if they are employed by one of our partner home care agencies, have one or more clients who are eligible for this study, English or Spanish-speaking, willing and able to implement Healthy Moves or Active Mind routine with their eligible client(s) for the full 4 months, and intend to be an HCA for the next 12 months. Table 1 lists detailed inclusion and exclusion criteria for study participants.

Table 1.

Pro-Home inclusion and exclusion criteria

Inclusion criteria	Exclusion criteria
Home care client
• Age 60+ • Receiving home care from a partner home care agency through the Illinois Department on Aging Community Care Program or managed care plans. • English or Spanish-speaking • Cognitive status sufficient to follow directions and respond to survey questions as determined by the Six-Item Screener (Callahan, et al., 2002) (the instrument used in the phone screening) and the Modified Mini-Mental State (3MS) Examination (the instrument used in the in-home screening). • Able to sit in a chair independently for >=15 minutes (it is OK if the person needs help with transferring to a chair); assessed to be able to participate in either intervention program. • Receiving in-home services for a home care aide eligible for the study • Willing to have UIC notify their study participation to their care manager (Care Coordination Unit) and to their primary care physician • Willing to be assigned to either intervention program	• Having a legal guardian appointed • Receiving hospice care or having a terminal diagnosis • Self-reported inability to read large print books due to vision issues or poor eye sight. • Having participated in the Pro-Home pilot • Unable to perform the intervention program activities for the full 4 months.
Home care aide
• Aged 18 and older • Employed by a collaborating agency • Caring for a client eligible for the study • English or Spanish-speaking • Willing and able to implement the intervention routine with their eligible client(s) for the full 4 months • Willing to be assigned to either intervention program • Intend to be a home care aide for the next 12 months	• Self-reported inability to read

2.6. Recruitment procedures – Overall

In collaboration with our partner agencies, we identify HCA-client pairs who potentially meet the eligibility criteria. We plan to enroll 300 pairs in the Healthy Moves or Active Mind arms. We assume 75% retention to retain 224 pairs at the end of the 4-month intervention (see 2.21 Power analysis and sample size). We assume the most likely scenario is that each participating HCA has 1 participating client, except for 10% who have 2. We recruit both family and non-family HCA-client pairs (see 2.3, Overview and study design) from a large pool (10,000+) of English or Spanish speaking HCAs employed at partner home care agencies in the Greater Chicago Illinois Area. In collaboration with selected agencies and SEIU, we advertise research opportunities for HCAs and clients via announcements and brochures (English and Spanish) with a QR code or link to a brief recruitment video. We use multi-pronged recruitment strategies, starting with the research team informing HCAs of the Pro-Home research opportunities at in-service training sessions that home care agencies provide for their HCA employees. The state mandates that HCAs satisfy at least 12 hours of in-service training per year (e.g. 3 hours/quarter X 4 quarters). We invite HCAs to sign up or contact the Pro-Home research team in person or via telephone.

2.7. Challenges in the participant recruitment process

Recruiting research participants poses challenges in publicly-funded home care settings. The greatest challenge is to obtain potentially eligible participants’ contact information. Our partner home care agencies have the information, but the State requires agencies to obtain written consent to release contact information to the research team. This requires the agency to send a home care supervisor to clients’ homes to obtain clients’ signatures for the release of information form. However, Home care aides are not allowed to obtain their clients’ signatures. Alternatively, if the client receives the Pro-Home information from the agency or from brochures/letters brought by HCAs and then voluntarily contacts the research team, we can obtain contact information directly from the client. However, for frail home care clients who are interested, initiating a phone call to contact the research team is difficult. Another major challenge is that both members of the HCA-client dyad need to be interested in participating.

These challenges are a consequence of this being a research endeavor, because research requires voluntary participation among study participants. If this program were offered as part of the usual home care program, the home care agency could have HCAs offer the program to interested clients as part of their job responsibilities.

2.8. Screening procedures

Potentially eligible participants receive advance letters (one from the partner home care agency and one from the research team) with information on the opportunity to participate in the Pro-Home research study. A research staff member conducts telephone screening interviews with potentially eligible participants (clients and HCAs) and answers any questions they may have about the study. For clients who are screened in via telephone, the research staff schedules an in-home screening visit. A research interviewer screens clients by observing the clients’ ability to sit independently and to follow instructions.

2.9. Research interviewer training

Four interviewers, including those who are bilingual in English and Spanish, receive a 20-hour hands-on in-class training, including role playing exercises. The training covers PA and safety in older adults, communicating with older adults with disabilities, protection of human subjects in research, obtaining informed consent, scheduling and conducting interviews and assessments using research instruments, and using technology (i.e., using iPads and internet hotspot devices to capture responses electronically). In-class training is followed by on-the-job training. This includes shadowing experienced interviewers, and supervised interviewing. The interviewer is assigned to a geographic area and is expected to visit the same client for all the assessments in order to increase rapport and efficiency.

2.10. Consent procedures

Standard procedures are followed to obtain signed informed consent forms for the two groups of participants. For home care clients, trained research interviewers obtain consent to participate in the study from those who meet the study eligibility criteria at in-home screening visits. If the client meets the eligibility criteria, the interviewer will explain the study and obtain informed consent. Interviewers are encouraged to schedule in-home screening visits at regular HCAs’ home care to reduce clients’ anxiety and to involve his/her HCA in the study process. UIC informs the clients’ primary care physician via fax of the client’s participation in the Pro-Home program. As requested by the State of Illinois, UIC sends the list of consented clients to the appropriate CCU, which has contracted with the State to conduct needs assessments and care coordination for CCP in-home service recipients who reside in designated geographic areas. CCUs review the list of Pro-Home client participants and file copies of the research consent forms signed by the participants. For HCAs, the research staff obtain consent at the beginning of the Pro-Home in-service training (see 2.12, Intervention).

For all study participants we stress throughout the recruitment and consent process that participation is voluntary so that both home care clients and HCAs will not feel coerced to participate. For home care clients, we emphasize that their participation (or non-participation) will not affect their relationship with their home care agency, their case manager, the Illinois Department on Aging, and the University. Similarly, for HCAs who are the home care agency’s employees, we keep home care agency partners aware of the voluntary nature of participation. UIC and agency partners emphasize that that participation (or non-participation) will not affect the HCA’ relationship with the home care agency, SEIU Healthcare Illinois and Indiana, or any other party involved in this study.

Furthermore, we ensure that potential HCA participants will know that their participation will depend on their client’s participation and that potential client participants will know that that their participation will depend on their HCAs’ participation. Typically, HCAs and their clients have close relationships, and in fact, approximately one half of them are related (e.g., HCAs may be their daughters or nieces). Thus, discussion on whether to participate in this health promotion intervention between HCAs and their clients is healthy and important. Nonetheless, we pay special attention to minimize coercion from each other, especially because of the interdependent relationships between the two.

2.11. Randomization

Clients who complete baseline assessment interviews and whose HCAs meet the study eligibility criteria are randomly assigned to two groups. Clients are stratified by (1) whether the number of chronic conditions is above 5 (the sum of checked items over the list of relevant chronic condition; 0 = No, 1 = Yes), (2) whether the client’s HCA is a family member (0=No, 1=Yes), and (3) client’s preferred language for the conduct of interviews (0=English, 1=Spanish). Using a SAS macro built into REDCap [29] that generates a binary sequence of balanced blocks of sizes 2 and 4 in random order, the randomization algorithm is applied to the first client enrolled for a particular HCA. The first client’s assignment is also the HCA’s assignment to the experimental condition.

We expect that the majority of HCA participants will have one eligible client participating. For a minority of HCAs who have multiple eligible and interested clients, we allow up to 2 clients to participate in Pro-Home. This is to help prevent excessive burden on each HCA. The HCA’s second client participant is assigned to the same group as the first client to prevent group contamination. The HCA’s third client participant, if any, is saved in the potential “pool” of participants in case circumstances arise that prevent the first two clients from participating in the study (e.g., health changes). The research staff member who manages the REDCap randomization module and conducts randomization is not directly involved in participant assessment or interventions.

2.12. Intervention

The intervention starts with HCA training. A half-day training program led by UIC researchers starts with an informed consent and baseline survey session. The training includes (1) introduction to the program (Healthy Moves or Active Mind), (2) background of the program, (3) demonstration of the program on the first home care visit, which includes brief motivational enhancement and the program activity, (4) safety precautions, (5) subsequent reminder and encouragement, and (6) role playing exercises (e.g., HCAs work in pairs to play the role of HCA or client and go through the motivational enhancement and the program activity). The Healthy Moves and Active Mind groups receive almost identical training except for the content of the program, but at a different time to prevent group contamination. The training has been approved by the state to satisfy the state mandated in-service training hours for HCAs. As participant recruitment evolves, multiple sessions are offered for each home care agency to accommodate agency-specific training requirements as well as HCA personal schedules. These arrangements enhance attendance and allow small-group interactions.

On the first home visit after the training, HCAs deliver the program to their client participant(s), starting with the motivational enhancement component and followed by the activity component of Healthy Moves or Active Mind. More specifically, HCAs assess clients’ readiness for the activity and have their clients set personally meaningful goals. Following motivational enhancement, the HCAs will teach the activity (see below). This process will take 15–40 minutes. Clients are asked to do the activity, either in the presence of their HCA on a visit day or by themselves on non-visit days.

The Healthy Moves’ activity component involves arm curls (up to 15 arm curls for each arm 2 times a day with 1 lb weight supplied by the project, or 1 lb soup can or water bottle), ankle point and flex up to 30 seconds on each foot 3 times a day, and seated step-in-place up to 1 minute once a day. (See 1. Introduction for more information on Healthy Moves). Clients and HCAs fill out daily logs and mail them using prepaid envelopes supplied by UIC. For each subsequent visit, HCAs remind their clients of their personal PA goals and routines and the exercise log. HCAs continue to encourage the client, re-assess readiness, and re-examine the goals as needed (informally while cleaning, for example). The frequencies and the hours of HCA visits vary by clients’ needs assessed by the Illinois Determination of Needs instrument (typically 2–3 times/week or 8 hours/week). Analysis will take this variation into consideration.

Active Mind’s activity component involves thinking activities, specifically “word searches”, that are unlikely to impact clients’ physical function in 4 or 8 months [30, 31]. “Word searches” involve recognizing and marking words written horizontally, vertically, and diagonally in a grid of seemingly jumbled letters. “Word searches” are alternative cognitive activities to the more commonly utilized crossword puzzles documented in the literature [31]. Studies have shown that those with lower education engage in cognitive activities less frequently [32] [33]. Word searches are utilized in lieu of crossword puzzles as they require significantly less skill, are available in both English and Spanish, are available in large print, and are a popular choice among older adults as a leisure activity. HCAs receive training on cognitive problems in older adults [34] and how to deliver thinking activities to clients.

After the 4-month intervention period, HCAs receive a half-day follow-up training for Healthy Moves or Active Mind, which will also count towards the state-mandated in-service training hours. This session provides HCAs with additional training on the assigned program. This includes safety issues, appropriate activity levels, effective goal setting, and tips for maintaining activities. HCAs have the opportunity to share their successes and challenges and learn from each other’s wisdom. HCAs receive a certificate of completion from the Pro-Home team at the conclusion of the follow-up training. To accommodate HCA turnovers (estimated to be less than 10%), HCAs who replace original HCAs are invited to this follow-up training so that participating clients will continue to receive HCAs’ support for Healthy Moves or Active Mind. The 4-month follow-up training was well received in our Healthy Moves pilot.

2.13. Intervention fidelity

HCAs are asked to report the result of their first day of the intervention (e.g., client’s readiness and goal) to the UIC team immediately after delivering the program to their clients. The UIC research staff make brief telephone calls to all the clients within 48 hours of the start of the intervention to assess the fidelity (e.g., whether the HCA delivered the program as trained, whether the client has identified a personally meaningful goal). The fidelity calls also help the research team identify potential issues in the study implementation and make corrections as appropriate. Direct observation of intervention activities (i.e., visiting each client’s home at the time HCAs deliver the intervention) is not practically possible.

The research team reviews the monthly activity logs submitted by HCAs and clients to monitor what HCAs and clients are doing. For any problems identified (e.g., no form submitted, indicating that the participant has any questions or comments about the log or the program), UIC research staff will call the HCA or the client to understand the situation and provide further assistance as appropriate. We document the extent to which the interventions are being implemented as intended. The intervention and control groups receive comparable fidelity checks.

2.14. Assessments

Clients are assessed by trained interviewers three times: before receiving the intervention (baseline), and after starting the program with their HCA (Month 4 and Month 8). A large number of in-home interviews and assessments occur during the RCT phase. We rigorously train in-home interviewers and establish an efficient system to schedule visits in geographic clusters, to be coordinated with HCAs’ regular home visits, while making scheduling flexible to allow clients’ convenience and occasional need for multiple visits to accommodate any fatigue or health changes. The interviewers are blinded to the group assignment. The clients’ perception and evaluation of the assigned program (Healthy Moves or Active Mind) and their program activity status are assessed by the research staff on the telephone when they schedule a follow-up interviewer visit. See Table 2 for a complete schedule of client assessments.

Table 2.

Pro-Home assessment schedule

	Phone screening	In-Home screening	Baseline	Randomization	Month 4	Month 8
Eligibility	✓	✓
Activity motivation		✓
Primary outcomes
HM6			✓		✓	✓
IADL4			✓		✓	✓
Secondary outcomes
BADL			✓		✓	✓
Depression			✓		✓	✓
Bodily pain			✓		✓	✓
Falls			✓		✓	✓
Fear of falling			✓		✓	✓
Self-rated health			✓		✓	✓
Physical performance			✓		✓	✓
Process outcomes
Exercise-related social support from HCA			✓		✓	✓
Physical activity levels			✓		✓	✓
Client covariates
Client-HCA relationship quality			✓		✓	✓
Family/non-family HCA			✓
Duration of receiving care from HCA			✓
Living arrangement		✓
Marital status		✓
Family support			✓		✓	✓
Comorbidities			✓		✓	✓
Events hindering participation			✓		✓	✓
Demographics		✓
HCA covariates
Demographics			✓
Job tenure			✓
Work full time or part time			✓
PA levels			✓		✓
Outcome expectancy			✓		✓
Program delivery					✓

Notes: HM6 is a scale consisting of 6 daily activities specifically targeted by Healthy Moves

2.15. Primary outcomes

The primary outcome is older adults’ self-reported physical function. Daily activity difficulties and dependency are the complementary components of the disability continuum in frail older adults [35, 36]. Based on basic activities of daily living (BADL) [37, 38] and instrumental activities of daily living (IADL) scales often used in the gerontological literature [39], we developed a scale consisting of 6 items that are specifically targeted by Healthy Moves: (1) pulling or pushing large objects like a living room chair or lifting or carrying things that weigh over 10 pounds like a large bag of groceries or laundry, (2) pouring a drink from a carton, (3) picking up your feet to avoid tripping on rugs, steps, or curbs, (4) walking up one step or up onto a side walk, (5) walking from room to room, (6) getting to the toilet. These items were chosen theoretically (i.e., activities targeted by Healthy Moves; items unlikely to be affected by Healthy Moves were excluded [e.g., ability to use telephone, ability to handle finances]). Following Gill et al.[40], two questions are asked for each of the above-mentioned items: “At the present time, how much difficulty (none/some/a lot), on average, do you have doing the task without help from another person?” and “Do you need help (yes/no) from another person to (do the task)?” Each task will be scored 0=no difficulty/no help; 1= difficulty but no help; 2=need help (summed score, range:0–12) [2]. This scale, HM6, together with the commonly used measure of IADL (IADL4, a scale of 4 items measuring activities likely to be affected by Health Moves [39]), are used as the primary outcomes. Empirically, our Healthy Moves pilot data indicated improvement in HM6 and IADL4 [11] that was statistically significant. Alternative scores (number of dependent activities, number of difficult activities, sum of degrees of difficulties) will also be calculated.

2.16. Secondary and process outcomes

Secondary outcomes include an alternative measure of self-reported physical function (BADL [37, 38]); depression (Center for Epidemiologic Studies Depression Scale, 10 item, α=.80) [41, 42]; bodily pain (SF36[43], 2 items, α=.80–.90[44]); number of falls in the past 4 months; fear of falling; and self-rated health. These self-reported measures are complemented by physical performance tests, including ankle range of motion, upper body extremity, and the Short Physical Performance Battery (SPPB) that includes side-by-side test, semi-tandem test, tandem test, 3-meter gait speed test, single chair stand test, and multiple chair stand test [45]. The performance tests are conducted by trained research interviewers (see 2.9).

Process outcomes include exercise-related social support from HCA [46, 47] (e.g., “During the past 4 weeks, how often did your home care aide give you helpful reminders to exercise?”; 3-point scale: hardly ever or never, sometimes, a lot), PA levels [48] including time spent on different physical activities in the past week (strengthening or stretching, walk or steps for exercise), and intervention-related activity frequencies (days/week).

2.17. Client covariates

The client-HCA relationship quality may modify Healthy Moves’ impact. Parallel questions are asked to clients and HCAs. We adapted the Dyadic Relationship Scale [49], an 11-item scale that captures negative strain (care-recipient α=.84, caregiver α=.89) and positive interaction (care-recipient α=.86, caregiver α=.85) in family care, and the Better Jobs Better Care project questionnaire [50]. Other covariates include the client’s relationship to the HCA (whether client is cared for by a family or non-family HCA, duration of the relationship in years and months), living arrangement (e.g., living alone), marital status, and family support availability. Self-reported comorbidities (e.g., coronary artery disease, congestive heart failure, COPD). Also included are events that would hinder Healthy Moves (e.g., hospitalization, death), and demographics including age, gender, race/ethnicity, education.

2.18. HCA covariates

HCA covariates include age, gender, race/ethnicity; job tenure (years), working full- or part-time; PA levels [48], outcome expectancy (e.g., “ How likely is it that your client will have improved function in the next 4 months?”), HCAs’ program delivery assessed by the first day information provided by HCAs, activity logs, and post-intervention survey questions.

2.19. Monitoring adverse events, and data and safety monitoring

Adverse events and serious adverse events are being carefully monitored and managed according to the guidelines of the National Institute on Aging and the UIC Institutional Review Board, and will be reported in future publications. This study involves frail older adults, who may experience hospitalization and death independently of the study intervention. Precautions for unlikely safety-related incidents include safety monitoring protocols and timely response procedures, training of HCAs and research staff, and having clinically trained research staff (e.g., geriatrician, kinesiologist) available for timely decision-making. Clients who experience any of the “red flags” (e.g., radiating pain) are asked to stop exercising immediately and call their doctor or 911. We carefully document reasons for attrition. The geriatrician on the research team provides medical perspectives to the day-to-day data and safety monitoring process. An independent Safety Officer has been appointed by NIA and reviews the data and safety monitoring report every 6 months to discuss study progress and safety.

With the intervention built into HCAs’ jobs and routine home care services, relatively high retention rates are expected. Based on a prior Healthy Moves observational study [20, 21], we aim at 75% retention. This target is between the rate that was actually attained (65.3%) and highest possible rate (82.4%, excluding participants who left the study due to reasons that were not under the control of the researchers, i.e., participants’ death, health problems, institutionalization and discharges from the home care program). For participants who discontinue the intervention activity (e.g., due to health or life situations), we follow up with them to document important factors (e.g., duration of and reasons for the interruption) and take them into account in analysis. Those who temporarily discontinue the activity will be advised to resume it when deemed safe by the study geriatrician. For participants who drop out of the study, we obtain reasons for dropouts.

2.20. Data management and quality control

We have established an effective scientific workflow to make our data capture, management, and analysis replicable [51]. REDCap [52], a secure web-based data collection, entry and management system, is used to capture data generated at each research step described above. The data to be obtained include client and HCA tracker files of all the participants recruited and contacted; client assessment data at baseline, at Month 4, and at Month 8, intervention implementation data (HCA training date and time, initial program delivery date and time, etc.), client and HCA program activity monthly calendar log reports; fidelity check call reports; interviewer field notes of each assessment, and others (e.g., records of contacts made with HCAs, clients, key stakeholders; incident reports). Other data, such as research staff field notes, are electronically stored and managed using personal computers linked into a secure local area network. This data system helps us detect issues and monitor for events that may threaten internal validity.

Focus groups are conducted among pretest and enrolled HCAs. To keep participants’ privacy, we ask HCAs not to use last names (both HCA and clients) in the discussion. The audio recordings, research notes, and transcriptions are stored in a directory with restricted access on a secure network. Access is granted to key research personnel only. Transcriptions for data analysis will be de-identified.

We have created guidelines for assessment, data checking and data backup to ensure the quality and the security of the data. There is a 120-day window to complete baseline assessment before randomization. The 120-day window is used to balance the need for timely baseline assessment with the potential time lags created by HCA screening requirements and occasional switching of HCAs as well as with our concerns about respondent burden of getting reassessed after a short period of time. If more than 120 days have passed since baseline assessment, a client will be reassessed for eligibility before randomization. The more recent assessment becomes the baseline data. Month 4 assessment is completed at 120 ± 7 days (i.e., 113 to 127 days) after the first day when the HCA delivers the program to the client. Month 8 assessment is completed within 240 ± 7 days (i.e., 233 to 247 days) after the first day. If a Month 4 or Month 8 assessment is completed outside the time windows, we flag the case for data analysis purposes. Interviewers are required to double check the assessment data to ensure that all responses are recorded. Research staff members are also assigned to check the data upon completion of an assessment and the self-check by interviewer. Our REDCap installation is backed up daily. In addition, a separate backup is made bi-weekly by project staff.

Upon completion of entry, quantitative data will be exported to SAS and STATA where labels and formats will be applied. All SAS and STATA scripts will contain header sections documenting author, date, purpose, content, and version number. Data and scripts will be maintained on the file servers that are professionally managed with firewalls, virus detection and removal, daily backups (complete weekly backups stored off site) and are HIPAA-compliant for secure storage of private health information. File hierarchies are used to organize and manage the growing collection of files. A codebook is prepared at an early stage to promote careful data management. A data management procedures manual summarizes operations and staff will receive training on these procedures. We maintain the data in a transparent manner to facilitate data sharing with other investigators. We will make the data and associated documentation available to users under a data-sharing agreement, managed by the National Archive of Computerized Data on Aging (NACDA).

2.21. Power analysis and sample size

The study is a 2 Group × 3 Times repeated measures (split-plot) design, with 4-month intervals between occasions. We employ stratified and blocked randomization [53] of clients into Healthy Moves (Treatment) and Active Mind (Control) conditions. The first stratification factor is a median split on the number of chronic conditions at baseline (5, based on the median from 49 pilot clients). The second stratification factor is whether a client’s HCA is a family member. The third stratification factor is whether the client is Latino. Block size alternates randomly between 2 and 4. The key hypothesis concerns Group × Time interaction: clients in the Healthy Moves group will better preserve function and well-being than will clients in the Active Mind control group at the end of the 4-month intervention and at the end of the 8-month follow-up period. Phrased negatively, clients in the Active Mind control group will experience a greater decline toward disability than will clients in the Healthy Moves intervention group. As mentioned earlier, our power analysis is for our primary aim to establish the impact of the intervention on IADL4. [Note that pilot data at baseline (n=54) showed Cronbach’s alpha of 0.73 for IADL4, whereas HM6 had an alpha of 0.60.]

Pilot data from 49 clients in Healthy Moves treatment showed an effect size of 0.38 for improvement in IADL4 after 4 months (p = 0.001, one-sided) and an effect size of 0.35 for improvement in HM6 after 4 months (p = 0.006). We also found, through an involved discrimination method based on ROC analysis applied to 49 clients in our pilot study [54], that the lower-bound “minimal clinically important difference” (MCID) was 1 point improvement in IADL4, which corresponds to an effect size 0.40 (the upper-bound on the minimum was 2 points, with an effect size of 0.80). As Wolinsky et al.[30] state, “In the gerontology and geriatrics literatures, the development of difficulty with an additional IADL is widely considered clinically meaningful…,” which coincides with our MCID analysis. Based on close examination of these empirical findings, we judge that an effect size of 0.35 (≈ 1/3 SD) is a reasonable expectation for the impact of our planned intervention. In Cohen’s terms [55], we aim to detect a “smallish” effect (between “small” and “medium”). Note that we calculate the sample size needed for a two-sided Group × Time interaction test of the 2 d.f. hypothesis that change in the intervention group differs from change over the same period in the control group, meaning we can detect intervention superiority or inferiority vis-a-vis control. Given that randomization achieves baseline equivalence between groups (i.e., quite similar means), the test of interaction is sensitive to the difference between intervention and control group means regardless of the relative stability or increase in disability. What matters is the final difference, not whether disability trends upward or downward in respective groups.

Calculation of the requisite sample size used the method of Rochon [56]. This method takes into account the cross-time correlation of residuals, which was 0.68 for IADL4 in preliminary data. Our calculation used a cross-time correlation of 0.65, which yields a somewhat conservative sample size. In order to detect an interaction effect size of 0.35 at two-sided alpha 0.05 with power 0.80 (beta 0.20) and with cross-time correlation of residuals rho 0.65, 112 clients are needed in each group. The Group main effect requires 100 per group; the Time main effect requires 59 across both groups. Assuming further a plausible retention rate of 75% of subjects through the completion of intervention, the required sample size is 112/0.75 = 149, or a total across two groups of 298. As mentioned earlier, the most likely scenario is that each participating HCA will have 1 participating client. Even in an unlikely situation where all HCAs have 2 clients and the intra-class correlation is 0.05, the sample size requirement would increase only by 5% (300×1.05 = 315). Only about 10% of HCAs have 2 clients, so recruiting 0.10×15 ≈ 2 more clients corrects for this. We aim to recruit 300 pairs of eligible clients and their HCAs.

2.22. Data analytic plan – Aim 1

We will conduct a group-by-3 wave mixed model analysis. Our primary hypothesis is that changes from Month 0 to 4 will differ between groups, which will be tested via an interaction contrast within the linear mixed model framework. The model also permits testing the maintenance question, namely, whether changes in function from Month 4 to 8 differ by treatment group. We will conduct our primary, confirmatory analysis under the mixed model for repeated measures (MMRM) specification. The MMRM uses all available data for estimation under the missing at random (MAR) assumption [57], but no imputations. This approach satisfies the intention-to-treat (ITT) rules [58–60], whereby subjects must be analyzed with the group to which they were randomized even if they dropped out midway or were reassigned to a different treatment.

It is reasonable to expect that data will be missing at random (MAR) for the current study, meaning that missingness will relate to observable covariates, which can be included in mixed models to achieve unbiased estimates under direct likelihood estimation; however, it is not possible to rule out missing not at random (MNAR), given that we have no access to the unobserved data. We will conduct the following sensitivity analyses to rule out MNAR missingness [61]: (1) multiple imputation via chained equations by group [62]; (2) pattern mixture model [63] (which does not assume MAR) based on dropout pattern; and (3) local influence technique [64].

The modified intention-to-treat (mITT) approach [65] allows the exclusion of participants post-randomization. In the current study, some participants may never start the intervention after they are randomized to an intervention group (e.g., HCA absent of training, participant deemed ineligible due to health change). We will follow the mITT approach to exclude these participants and analyze dataset consisting of participants who start the intervention following randomization.

Supplementary analyses will extend the foregoing primary analyses as follows: (1) To enhance group mean equivalence and reduce unexplained variance and thus to enhance power, we will incorporate variables used in the stratified randomization (Section 2.11). We will also include variables considered to have potential to strongly affect the mechanism of action of the interventions on primary outcomes, based on prior evidence or theory. (2) Similarly, potentially unbalanced HCA characteristics will be included as covariates (Section 2.18, HCA covariates). Such characteristics include HCAs’ delivery of motivational enhancement, delivery of the planned program with fidelity (Section 2.13, fidelity calls, or log sheets), frequencies of HCA visits, years of in-home service experience, and other HCA characteristics such as age, gender, race/ethnicity. (3) The clustering of clients within HCAs, although expected to be rare, will be statistically controlled for. These analyses will enhance confidence in results by eliminating competing explanations of results.

To interpret the clinical significance of our outcomes, we will define small, medium, and large changes using global ratings of change as the anchors (e.g., averaging the ADL disability change scores of those patients choosing “minimal” change responses to calculate the “minimal clinically important difference, MCID”) [66–68]. Evidence of established MCID is scarce for the commonly used ADL instruments among nursing home-eligible community-dwelling seniors [69]. We only found one study that identified anchor-based MCID of the (I)ADL scores among Dutch older adults [70]. We will also present percentages of people who declined, maintained, and improved in function in the intervention and control groups, using the above-mentioned definition of changes. Given the downward trajectory of nursing home-eligible persons’ function, “no change” (preserving function) is considered an intervention benefit. Using “patient” global rating results as anchors is the most common way of interpreting patient-reported outcomes, but there are some concerns (e.g., a recall bias for the present state) [66, 71–73]. Therefore, we will triangulate the results with other approaches, including the use of HCAs’ ratings of global change and distribution-based methods. We will also explore the relationships between changes in client-reported function and changes in more “objective” outcomes (e.g., physical performance test results).

We expect that home care clients receiving the Healthy Moves intervention will better maintain function in comparison with clients receiving an attention control, whose function will decline. Regardless of the results of our primary hypothesis testing, we will analyze, and report results for secondary outcomes (e.g., IADL4, BADL, depression, bodily pain, falls, self-rated health, and physical performance tests).

2.23. Data analytic plan – Aim 2

Our Aim 2 is to understand how Healthy Moves delivered by HCAs works in a home care setting, specifically (a) for whom the program is beneficial, (b) the extent to which the program can reach the target population, (c) the extent to which participants drop out of the program, (d) the extent to which program participants maintain the behavioral change introduced by Healthy Moves, and (e) what the program’s cost-effectiveness is.

We will use mixed methods to address Aim 2. Closed- and open-ended questions will be asked to both clients and HCAs to evaluate program implementation outcomes (e.g., feasibility, acceptability). Post-intervention interviews with partner organizations will be conducted to assess the program’s feasibility, acceptability, and cost-effectiveness as perceived by all the key parties involved.

For quantitative data, we will extend the primary and supplementary analyses described in Aim 1 by incorporating moderating variables. For qualitative analysis, we will use data from interview questions for study participants and stakeholders and focus group to supplement field notes to produce reports and inform next steps. More specific data analysis plans are presented below:

Our goal here is to determine the types of clients for whom the program is beneficial, especially in terms of clients’ relationship with HCAs (i.e., cared for by a family HCA, relationship quality), living arrangement (living alone or not), gender, and race/ethnicity. This is to provide useful information to older adults and providers, and to support later meta-analyses (and thus not powered to test the heterogeneity within the study) [74]. We will test interaction terms that involve the client subgroup factors and the program (Healthy Moves or Active Mind). For example, we expect clients with a better relationship with the HCA will feel more motivated and supported by HCAs to engage in physical activity and thus are more likely to preserve function [75]. We expect that the likelihood of preservation of function is not greater among clients who are cared for by family HCAs than those who are cared for by non-family HCAs. Family HCAs are likely to have more time with clients to engage in Healthy Moves activities. However, clients may resist behavioral changes brought by family members even more than those brought by non-family HCAs.

Further, to determine the extent to which intervention (Healthy Moves and Active Mind) reaches the target population, we will calculate program participation rates among HCAs and clients contacted by the research team for program participation. To identify the characteristics of HCAs who are interested in Healthy Moves, and who participate in and successfully complete the program, we will examine how HCAs’ age, gender, race/ethnicity, and years of HCA experience are associated with the likelihood of program participation. To understand the representativeness of client participants, we will compare the profile of client participants in this study with the profile of all in-home service clients in the CCP in Chicago and Illinois (to be obtained from the Illinois Department on Aging).

We will assess the extent to which participants drop out of the program and describe the characteristics of client and HCA dropouts compared with successful completers.

To examine the extent to which program participants maintain the behavioral change introduced by Healthy Moves, we will describe the proportion of Healthy Moves participants who are continuing Healthy Moves activities at Month 8. Logistic regression analysis will identify participant characteristics associated with maintenance of behavioral change.

In addition, we collect detailed data on costs associated with implementing and running the program (e.g., HCA training materials, time spent by trainers and HCAs, handouts for participants, time spent on Healthy Moves by HCAs and the home care agency staff) [76]. Our Healthy Moves pilot indicated that the proposed Healthy Moves program would require no more than 25 dollars of supplies for each older participant (Healthy Moves supplies for clients and HCA training) in addition to up to 60 minutes of additional HCA time used to deliver Healthy Moves on the first day and remind clients of the exercise over the entire program period. Total program costs across settings would depend on a variety of factors (e.g., salary levels of the staff providing training, long-term care policies and regulations that vary across states [77–80] that can affect the extent to which extra resources are required to implement Healthy Moves.) In consideration of these potential sources of variance, our analysis will include a variety of sensitivity analysis accounting for potential variance in time and salary levels. Finally, the cost-effectiveness of the program will be assessed by combining, separately, the cost estimates with the two summary score measures of improvement in activities (ADL4, HM6). Here again potential variance will be assessed both in terms of the cost sensitivity analyses described above and by including data-driven ranges of outcomes of the program (e.g., 95% confidence intervals).

3. Discussion

The clinical trial described above tests the hypothesis that home care clients receiving the Healthy Moves intervention will better maintain function in comparison with clients receiving an attention control, whose function will decline. In addition, a mixed methods approach is used to understand how Healthy Moves delivered by HCAs works in a home care setting. Healthy Moves is a sustainable health-promoting model for home care programs to help their clients maintain, and possibly improve, their function and continue to live in the community. Healthy Moves led by HCAs offers a specific set of practical, easy-to-understand tools for home care providers to incorporate prevention and wellness-oriented services. This gentle, simple, low-cost program requires few extra resources and appeals to practitioners. This is the first RCT of Healthy Moves. Producing evidence for the effectiveness of Healthy Moves is the first step to development of sustainable and translatable home care practice that emphasizes health promotion for a community-dwelling nursing home-eligible older adults.

Strengths of our research design include a RCT conducted in the context of a Medicaid-funded home care setting, an intervention delivered by the home care aide who has had established ongoing caregiving relationships with the client. Despite its strengths, our project involves challenges and limitations that are inherent in clinical trials conducted in a rapidly changing long-term care system. Recruitment of participants in a public home care program is particularly challenging, given that clients’ personal information (e.g., phone numbers) is protected by multiple layers of regulations. For example, the state requires home care agencies to obtain written consent from each client to release clients’ names and contact information to the UIC research team. The requirement that an agency staff person must be the one to obtain signatures from homebound clients has become a major bottleneck for our participant recruitment. Although interviewers are blinded to clients’ group assignment, clients may disclose it during the post-intervention interview. Our office research staff make every effort to reduce such disclosures by discouraging clients from discussing their experiences in the intervention programs with interviewers and by obtaining their program input via phone before interviewers conduct post-intervention in-home assessments and interviews. Group contamination is a potential problem although HCAs work in their clients’ homes and rarely interact with each other. We schedule Active Mind and Healthy Moves training sessions at different times to prevent interactions between HCAs in Active Mind and Healthy Moves groups. We also empirically investigate the extent of contamination by asking participants whether they were exposed to the alternative program in the post-test survey and statistically investigate its effects.

As a clinical trial implemented in a real-world home care setting, Pro-Home lacks control over unexpected public health emergencies or social events that can affect study participants and research operations, as represented by the COVID-19 pandemic. Pro-Home continues to make maximum efforts to minimize threats to the research design and follow the protocol presented in this paper. For example, we document evolving COVID-19 social distancing policies and their effects on our study participants and our stakeholders. This allows us to adjust our research operations (e.g., to maximize participant retention and equivalency between the two experimental groups), while observing social distancing policies at federal, state, and local levels.

A growing number of this nursing home-eligible older population receive home care to remain in their own homes. HCAs are projected to add the most new jobs of any occupation between 2012 and 2022 (580,800 jobs, 49 percent growth) [81]. This project will produce knowledge on whether a gentle, low-intensity physical activity program led by home care aides for community-dwelling nursing home-eligible seniors is effective and how it works in a Medicaid-funded home care program. If this project produces evidence for the program’s effectiveness, the results will facilitate the adoption and implementation of the sustainable physical activity program in real-life home care settings. The outcomes from this study will have important scientific and practice implications for a new model of home care, which would facilitate a cultural shift towards an “active service model” where home care aides and their clients work together on a gentle, easy-to-learn physical activity program to maintain or improve function to remain in the community.

Abbreviations

ADL

Activities of Daily Living

BADL

Basic Activities of Daily Living

CCP

Community Care Program

CCU

Care Coordination Units

HCA

Home Care Aide

IADL

Instrumental Activities of Daily Living

HIPAA

Health Insurance Portability and Accountability Act

ITT

Intention-to-Treat

MAR

Missing at Random

MCID

Minimal Clinically Important Difference

mITT

Modified Intention-to-Treat

MMRM

Mixed Model for Repeated Measures

MNR

Missing Not at Random

PA

Physical Activity

Pro-Home

Promoting Seniors’ Health with Home Care Aides

RCT

Randomized Controlled Trial

SEIU

Service Employees International Union Healthcare Illinois & Indiana

SPPB

Short Physical Performance Battery

UIC

University of Illinois Chicago