Effect of Enhanced Medical Rehabilitation on Functional Recovery in Older Adults Receiving Skilled Nursing Care After Acute Rehabilitation

Key Points Question Does enhanced medical rehabilitation, a technique for systematically engaging and motivating patients in their physical and occupational therapy, result in better functional outcomes? Findings In this randomized clinical trial of 229 older adults receiving physical and occupational therapy in skilled nursing facilities, those receiving enhanced medical rehabilitation showed a higher percentage of active time during therapy sessions and a 25% greater functional recovery compared with those receiving standard-of-care therapy. Meaning Systematically engaging older adults in their therapy after acute rehabilitation may result in a better functional outcome.


Introduction
Older adults who experience a disabling medical event, such as hip fracture, require physical and occupational therapy (PT/OT) in postacute care settings, such as skilled nursing facilities (SNFs). The use of postacute care has grown as frail and medically complex older adults survive medical events but with such functional incapacity that they are unable to return home and function independently. 1,2 In 2016, Medicare paid approximately $60 billion for postacute care, including 2.4 million SNF stays. 3 For such patients, postacute rehabilitation is a window of opportunity to regain functional ability. The alternative is persistent disability, which comes with considerable human costs, as well as high health care costs, 4 much of it resulting from rehospitalizations. [5][6][7][8] To date, efforts to improve rehabilitation outcomes of older adults have met with limited success in randomized clinical trials. 9,10 Increasing intensity by providing more PT/OT time in SNF therapy has modestly improved functional gains [11][12][13][14] and increased the rate of discharge to community settings. 15 Yet, a further increase in the use of therapy is likely not an option owing to cost-containment considerations. 16,17 Instead, postacute rehabilitation could be optimized by improving therapists' engagement with patients and the intensity of the therapy sessions, resulting in greater patient active time or patient activity per minute of PT/OT. 4,[18][19][20][21][22][23][24] Such engagement efforts must account for patient factors, such as depression, cognitive impairment, and multiple medical comorbidities, that can undermine motivation. [25][26][27][28] Therefore, enhanced medical rehabilitation (EMR), a set of techniques and tools for therapists to engage patients in therapy, was developed. 29,30 Enhanced medical rehabilitation is a systematic and standardized approach based on behavior change principles 31-34 to enhance patient engagement and intensity to promote optimal functional outcomes. Enhanced medical rehabilitation was designed for real-world therapists and uses short, intuitive motivational messages and simple tools to help the therapist link PT/OT activities, effort, and progress to attainment of the patient's selected and personally meaningful goals ( Table 1). 35 Preliminary research showed that PT and OT therapists could be trained to use EMR, resulting in higher patient active time and better functional outcomes. 30 The purpose of the present study was to compare EMR with standard-of-care therapy for older adults receiving postacute rehabilitation in 2 area SNFs. We hypothesized that patients randomized to receive PT/OT from EMR-trained therapists would have better functional outcomes than those receiving standard-of-care therapy. We also examined whether patient characteristics-depression, cognitive impairment, and medical burden-influenced therapy outcome.

Participants
From July 29, 2014, to March 22, 2018, patients were recruited on admission to 2 SNFs in the St Louis, Missouri, metropolitan area. These facilities were selected based on their willingness to participate in the study and the number and diversity of patients admitted who were receiving postacute care.
Participant inclusion criteria were age 65 years or older, admitted from an acute care hospital ( Table 2), and requiring 2 or more weeks of rehabilitation with the potential to return to the community; individuals already residing in long-term care facilities before hospitalization were excluded. Other exclusion criteria were language, visual, or hearing barriers to participation; medical illness preventing study participation (including metastatic cancer, ongoing cancer treatment, hemodialysis, hospice care, or highly unstable cardiac illnesses with anticipated rehospitalization); moderate to severe cognitive impairment (demonstrated by medical record diagnosis of dementia and/or Short Blessed Test 36 score >13); or psychotic disorder. The study was completed July 13, 2018.
This study followed the Consolidated Standards of Reporting Trials (CONSORT) reporting guideline for randomized clinical trials. This study was a randomized clinical trial with 2 parallel groups (EMR vs standard of care) and blinded outcome assessments. The study was approved by the Washington University Institutional Review Board. Potential participants were enrolled after providing university-approved written informed consent. The trial protocol is available in Supplement 1.

EMR Therapist Selection
Therapists were selected for EMR by measuring their pretraining patient active time and number of motivational messages and creating EMR and standard-of-care therapist groups who were equated in these variables ( Table 3) as well as years and level of experience. Eleven therapists (occupational therapists, physical therapists, and certified therapy assistants) were trained in EMR, including 7 at the larger facility and 4 at the smaller one. Eighteen therapists were not trained and performed standard-of-care therapy. Therapist selection was conducted in collaboration with therapy managers to ensure adequate staffing coverage in each group (EMR and standard of care). Therapists had the opportunity to decline participation in the training and thus be in the standard-of-care group. To   prevent bias or contamination between groups, the research team did not provide the standard-ofcare therapists with information about EMR and discouraged the EMR-trained therapists from

JAMA Network Open | Physical Medicine and Rehabilitation
sharing it with them.

Measurement of Therapists' Techniques in the Study
Research assistants observed study therapists and measured the following for both the EMR and standard-of-care groups: (1) therapists' fidelity (adherence) to EMR 37 (ie, whether the therapist carried out a correct technique at the appropriate time in therapy); (2) rehabilitation engagement of the patient with the therapist, using the Pittsburgh Rehabilitation Participation Scale, 39 which is a 1-item Likert-type scale that assesses how actively the patient participates and engages in the therapy session, ranging from 1 (refused entire therapy session) to 6 (participated 100% and was actively engaged throughout therapy session); and (3) patient active time, 40 which measures the percentage of time during the therapy session in which the patient actively performs a therapeutic activity (eg, walking or practicing an activity of daily living) as opposed to sitting and resting.
All of these techniques or processes were measured in a random sample of 5 therapy sessions per therapist before training to confirm that therapists' skills were equivalent before training. A sample of 4 therapy sessions (2 OT and 2 PT) were then conducted for each randomized participant during their SNF stay to demonstrate that EMR active ingredients were delivered in a manner distinct from standard-of-care therapy (including demonstrating the absence of contamination of EMR into the standard-of-care arm). and therapy notes that were redacted to hide participant identity or treatment assignment. Secondary outcomes were self-reported function, performance-based gait measures, discharge to community settings, and rehospitalization. Self-reported function after discharge was assessed using a patient self-report version of the Barthel Index at days 30, 60, and 90 after randomization administered by telephone; we also ascertained whether rehospitalization occurred after completion of SNF rehabilitation up to day 90 after randomization. Performance-based measures obtained at admission and discharge from the facility were gait speed using 1 trial of the 10-m walk test 44,45 and a 6-minute walk test (number of feet walked in 6 minutes). 46 Gait assessments were videotaped and scored by a blinded assessor. Disposition from the SNF was dichotomized as return to the community setting (ie, to a private residence, group home, or assisted living) vs not returned (ie, to further skilled nursing rehabilitation, long-term care, or a hospital).

JAMA Network Open | Physical Medicine and Rehabilitation
We collected 3 baseline variables for prespecified moderator analyses: the Montgomery-Äsberg Depression Rating Scale for evaluating the severity of depressive symptoms (range, 0-60; higher scores indicate more depressive symptoms; no cutoff was used in this study) 38 ; the Short Blessed Test, a brief assessment of orientation, registration, and attention 36 (higher scores indicate worse impairment, with scores of 5-9 consistent with mild cognitive impairment and Ն10 consistent with dementia); and the Cumulative Illness Rating Scale for Geriatrics score to quantify chronic illness burden from the medical records (higher scores indicate greater burden). 47

Statistical Analysis
The statistical analysis plan is available in Supplement 1. Intention-to-treat analysis was used. Study data were managed using REDCap, version 7. 48 Analyses were performed using R, version 3.5.0 (R Foundation) or SPSS, version 24 (SPSS Inc). The primary outcome was change in Barthel Index score; the secondary outcomes were the 6-minute walk test and gait speed values at SNF discharge, as well as discharge disposition, rehospitalizations, and self-reported function at days 30, 60, and 90. The discharge point was the sole focus for the 6-minute walk test and gait speed test, as most participants were unable to complete these tasks at baseline and, therefore, there was no variability in these data at baseline. To test the primary hypothesis that EMR participants showed greater not done and the level of significance, determined with 2-tailed testing, was set to 5% for functional change.

Results
Of 3265 patients screened for study eligibility, 2909 were ineligible (eg, age <65 years, severe cognitive impairment, or not scheduled to receive Ն2 weeks of therapy), 127 refused to participate, and 229 were randomized (Figure).

Process Data in Therapists Before and After Training
Before any training was conducted, the EMR therapists were similar to the standard-of-care therapists in terms of fidelity to the EMR intervention (quantified as number of engagement/ motivational techniques consistent with EMR per therapy session), mean patient active time per therapy session, and patient engagement as measured by the Pittsburgh Rehabilitation Participation Scale (Table 3). After training (ie, during the randomized clinical trial), EMR therapists used a median (interquartile range) of 24.4 (21.0-37.3) motivational messages per therapy session compared with 2.3 (1.1-2.9) for nontrained therapists (P < .001). EMR patients were active during a mean (SD) of 52.5% (6.6%) of the therapy session time vs 41.3% (6.8%) for nontrained therapists (P = .001). (Table 3). Thus, therapists trained and coached in EMR were conducting this intervention with good fidelity that was clearly differentiated from standard of care, and these differences were attributable to the EMR training and coaching and not preexisting therapist differences. Table 3 also presents the mean (SD) durations of therapy sessions, showing that EMR sessions were not longer than standard-of-care sessions.  Discharge disposition data (discharge to home vs institution) were available on 220 participants.

Outcomes
Whether a participant was discharged to home or was institutionalized was independent of condition. Rehospitalization data during the 90-day follow-up were available on 221 participants.
Whether a participant was rehospitalized was independent of condition. In terms of adverse events, there were none that were related to the study procedures or interventions.

Moderator Results
We considered 3 variables measured on admission as potential moderators: depressive symptoms (Montgomery-Äsberg Depression Rating Scale score), level of cognitive impairment (Short Blessed Test score) and total amount of medical morbidity (Cumulative Illness Rating Scale for Geriatrics score). Results are presented in Table 4. Contrary to our hypothesis, none of these variables was a

Discussion
This randomized clinical trial evaluated the effects of EMR, an approach to engage and motivate patients in PT/OT. Our main finding is that patients treated by EMR therapists had an estimated 25% greater functional recovery on average during postacute SNF rehabilitation, compared with those who received standard-of-care therapy. With respect to secondary outcomes, however, there were no group differences in gait measures, discharge disposition, or longer-term self-reported function.
These findings are important because they indicate that older patients can achieve better shortterm functional outcomes when treated by therapists who are trained and coached to systematically motivate patients and strive for higher-intensity therapy. Additional strategies are needed to maintain these functional gains after discharge from the rehabilitation facility and affect outcomes such as rehospitalization.
To our knowledge, this is the first full-scale test demonstrating benefits of a standardized method to improve rehabilitation outcomes by increasing engagement and intensity of therapy sessions. Functional recovery of older adults is an important outcome 50 and one that is not always achieved despite postacute rehabilitation services. 51 This finding fits well with the 2008 Institute of Medicine report Retooling for an Aging America, which recommended models of treatment that make older persons more active partners in their own care. 52 This finding is also supported by smaller studies demonstrating the benefits of systematic interventions to increase engagement and intensity in rehabilitation settings. 53 The EMR model does not ask therapists to do anything technically different in their practice, such as specific exercises or therapy protocols. Instead, it integrates communication techniques into therapy, increasing the focus on treatment engagement and intensity, thus providing more potent therapy without more treatment time. As such, EMR could be applied to any rehabilitation setting.
This finding is important, because we found no differences in key secondary discharge outcomes, including disposition (frequency of returning home) or gait measures (gait speed and 6-minute walk test). Therefore, to fully optimize outcomes, it may be necessary to combine EMR with additional components, such as techniques to increase muscle strength and stamina to improve gait performance, 55 and a postdischarge care component to reduce rehospitalization. 6,56 In addition, there were no differences in longer-term self-reported function. Self-reported function may be a different construct than therapist-measured function; furthermore, self-reported function appeared to show a ceiling effect by 60 to 90 days after SNF admission, suggesting that self-report scores were inflated or that participants who were able to be assessed after discharge were also those who regained most or all of their function. Studies should measure both selfreported and observed function to better understand the long-term functional trajectory after rehabilitation.
The effectiveness of EMR was not moderated by baseline levels of depression, cognitive impairment, or medical complexity, which we had estimated would be potential barriers to motivation and recovery. This finding argues against providing EMR only to certain patient groups, such as individuals with depression, in favor of a more universal application of this approach in rehabilitation.

Strengths and Limitations
Strengths of this study included demonstrating that EMR and standard-of-care therapists were equated at baseline prior to training and blinding of outcomes. The study has several limitations. The trial was conducted in 2 facilities in 1 geographic area. Because of logistical challenges, we were unable to control therapy done on weekends and we were only able to assign and randomize participants after 1 to 2 days in the SNF. However, this limitation would not invalidate positive study findings; possibly, EMR would have greater effects if implemented for all therapy sessions. Further studies of EMR are needed to replicate and extend these findings. Another limitation is that, for feasibility reasons, we were unable to directly observe all therapy sessions and we did not have a method of assessing treatment fidelity by therapists other than by direct observation; this lack of continuous assessment could have influenced EMR-trained therapists to carry out more motivational messaging when observed. Other limitations include low statistical power for examining binary secondary outcomes, such as rehospitalization, and a high rate of SNF admissions excluded for reasons such as patients being severely cognitively impaired or not requiring intensive rehabilitation.

Conclusions
This trial's findings suggest that EMR is effective in improving functional recovery for older adults in postacute rehabilitation. Improving outcomes is paramount for the estimated 6.4 million older adults receiving rehabilitation services yearly, 51 and the medical rehabilitation field has urged a greater focus on patient engagement and intensity in medical rehabilitation. [57][58][59]