High-intensity home health physical therapy among older adult Veterans: A randomized controlled trial

Abstract

Background:

Older adult Veterans are at high risk for adverse health outcomes following hospitalization. Since physical function is one of the largest potentially modifiable risk factors for adverse health outcomes our purpose was to determine if progressive, high-intensity resistance training in home health physical therapy (PT) improves physical function in Veterans more than standardized home health PT and to determine if the high-intensity program was comparably safe, defined as having a similar number of adverse events.

Methods:

We enrolled Veterans and their spouses during an acute hospitalization who were recommended to receive home health care on discharge because of physical deconditioning. We excluded individuals who had contraindications to high-intensity resistance training. A total of 150 participants were randomized 1:1 to either 1) a progressive, high-intensity (PHIT) PT intervention or 2) a standardized PT intervention (comparison group). All participants in both groups were assigned to receive 12 visits (3 visits/week over 30 days) in their home.

The primary outcome was gait speed at 60 days. Secondary outcomes included adverse events (rehospitalizations, emergency department visits, falls and deaths after 30 and 60-days), gait speed, modified Physical Performance Test, Timed Up-and-Go, Short Physical Performance Battery, muscle strength, Life-Space Mobility assessment, Veterans RAND 12-item Health Survey, Saint Louis University Mental Status exam, and step counts at 30, 60, 90, 180 days post-randomization.

Results:

There were no differences between groups in gait speed at 60 days, and no significant differences in adverse events between groups at either time point. Similarly, physical performance measures and patient reported outcomes were not different at any time point. Notably, participants in both groups experienced increases in gait speed that met or exceeded established clinically important thresholds.

Conclusions:

Among older adult Veterans with hospital-associated deconditioning and multimorbidity, high-intensity home health PT was safe and effective in improving physical function, but not found to be more effective than a standardized PT program.

INTRODUCTION

Hospital-associated deconditioning (HAD) contributes to precipitous declines in physical function in older adults, leading to development of functional deficits,¹ loss of independence,² and increased risk of rehospitalization.³ Older Veterans represent a growing population⁴ at elevated risk for adverse outcomes after an acute medical event, as this population has lower physical function,⁵ more physical and mental health comorbidities,^6-9 and decreased health-related quality of life⁷ compared to their age-matched peers.

In institutionalized settings with frail populations, high-intensity resistance training has been shown to improve physical function such as gait speed more than low-intensity resistance training.¹⁰ A Cochrane review demonstrated high-intensity resistance training leads to improvements in gait speed, chair rise, muscle strength, and pain.¹¹ High-intensity resistance training using an 8-repetition maximum load (inability to perform additional repetitions due to muscle fatigue),¹² has not been evaluated for those recently discharged from the hospital and receiving home health rehabilitation. This is a critical gap, as utilization of home based care continues to increase after medical hospitalizations¹³ but few guidelines for rehabilitation in these settings exist for those with HAD.¹⁴

The purpose of this study was to determine if progressive, high-intensity home health physical therapy (PT) improved physical function more than standardized PT for older adult Veterans with hospital-associated deconditioning. A secondary aim was to evaluate the adverse events. We hypothesized that progressive, high-intensity PT would lead to greater improvements in gait speed (primary outcome), and similar safety profiles when compared to standardized PT at 60-days after hospital or skilled nursing facility (SNF) discharge to home health.

METHODS

Study Design

This was a randomized, single blinded (assessors) controlled trial to evaluate the effectiveness and safety of high-intensity PT for older adult Veterans with multimorbidity referred to home health PT following hospitalization. The two arms were a 1) progressive, high-intensity (PHIT) PT intervention and 2) a standardized PT intervention (comparison group).

Participants

Eligible Veterans were greater than 55 years of age with at least 3 comorbid conditions who were referred for home health PT with deconditioning due to an acute hospitalization or COVID stay-at-home policies.¹⁵ Exclusion criteria included inability to walk 10 feet without human assistance or gait speed greater than 1.0 m/s at baseline, contraindications to high-intensity resistance training,¹² acute lower extremity fracture with weight-bearing restriction, recent joint replacement surgery, active treatment for cancer, current dialysis treatment, acute cardiac involvement, acute stroke, lower extremity amputation, severe cognitive impairment indicating inability to provide informed consent, hospice care referral, or significant medical complication that precluded safe participation in PT. Veterans’ spouses who were referred to home health PT for deconditioning and met the above criteria were also eligible for enrollment.

Recruitment and Randomization

Of 5022 Veterans and spouses assessed for eligibility, 4809 (96%) did not meet inclusion criteria; of the 213 who met criteria, 135 Veterans and 15 spouses (total 70% of eligible) agreed to participate. Eligible Veterans (n =135) and spouses (n=15) were recruited during an acute care hospitalization from the Rocky Mountain Regional VA Medical Center or UCHealth (University of Colorado Health). Most were enrolled immediately following a hospital stay (n=111), but some Veterans requiring a SNF stay following hospitalization were also enrolled (n=29). Beginning on 08/01/20, Veterans who were comparably deconditioned related to COVID stay-at-home orders were referred by their VA provider (n=10). Following informed consent and within 72 hours of discharge home or referral, the study team conducted baseline assessments in participants’ homes. Following baseline assessment, study coordinator randomized participants to PHIT or comparison group. The 1:1 randomization scheme with random block sizes of four and six¹⁶ was prepared by the study biostatistician (JEF). Allocation to assignment was transferred to concealed, sealed envelopes until needed for randomization. A blinded assessor completed follow-up assessments in the home at 30-, 60-, 90-, and 180-days following baseline assessment. Only therapists conducting the home visits, study coordinators, and the home health liaison were aware of group assignment.

Intervention

Trained physical therapists initiated study intervention within 48 hours of randomization. All participants received strengthening exercise, gait training, and functional training. While exercise intensity differed between study arms, prescribed frequency of visits was identical: 12 PT visits (3 visits/week) over 30 days. Brief descriptions are provided below. See supplemental material for more detailed intervention descriptions as guided by the Template for Intervention Description and Replication (TIDieR) checklist (Appendix A).

PHIT Intervention

The strengthening exercise component consisted of 3 sets of progressive high-intensity resistance training (8 repetition maximum load) that targeted lower extremity muscle groups. On non-therapy days, participants were instructed to complete: three body-weight exercises, 10 or 20-minute walking bouts, and log adherence to their home exercise program (Appendix B).

Standardized PT Intervention (Comparison group)

The strengthening exercises did not include weights and participants completed 10 repetitions for any given exercise. On non-therapy days, therapists instructed participants to perform two unweighted open chain exercises for 10 repetitions and to complete a home exercise program log (Appendix C).

Fidelity

All therapists received in-person, intervention-specific training with written and hands-on skill assessments initially and annually thereafter. To ensure protocol adherence, the study team performed quarterly on-site observations, performed documentation audits, and conducted regular conference calls to review issues relating to study protocol.

Outcomes

The primary study outcome was self-selected gait speed as measured by the four-meter walk test (4MWT) at 60 days. We chose 60 days as it coincided with Medicare-based episode of care at trial initiation. Gait speed was selected because it 1) predicts risk of disability, higher health care utilization, and increased mortality;^17-19 2) is a meaningful outcome measure in older persons;²⁰ 3) is valid and reliable;^21,22 and 4) is easily performed in the home.²³ Also, in Veteran populations, improvements in gait speed after hospitalization are correlated with reduced disability and health care costs.²⁴ We collected the following descriptive measures at baseline: age; patient-reported gender, race, ethnicity, marital status, height, weight, educational level, Functional Comorbidity Index (FCI)²⁵; and the Patient Activation Measure (PAM-10).^26,27

We collected the following secondary outcome measures at baseline, 30, 60, 90 and 180 days following hospitalization: adverse events, the modified Physical Performance Test (mPPT)^28,29, the Timed Up-and-Go (TUG)³⁰, Short Physical Performance Battery (SPPB)¹⁸, and muscle strength (knee extensors and grip strength). Life Space Mobility assessment (LSM)³¹ was collected at all times except baseline. Veterans RAND 12-Item Health Survey (VR-12)³² and Saint Louis University Mental Status exam (SLUMS)³³ were collected at 60 and 180 days.

ActivPAL^™ (PAL Technologies, Glasgow, Scotland) was used to measure physical activity, including step counts.^34,35 Participants wore activPAL^™ for 24 hours/day for 7-10 days at baseline and again at 60-days.

Safety was assessed by evaluating the number of adverse events including rehospitalizations, emergency department visits, falls, and deaths at all timepoints. We collected information on adverse events from the treating therapist’s and outcome assessor’s reports which were audited against medical records.

Power and Sample Size

Statistical power estimates suggested a sample size of 56 participants per group would provide 90% power to detect a between group difference (2-sided, alpha = 0.05 level 2-group t-test). More details are provided in Supplemental Materials.

Statistical Analyses

All analyses assumed a two-sided test of hypothesis, a significance level of 0.05 and were run in SAS v9.4 (SAS Institute, Cary, NC). Demographic and clinical characteristics and outcome measures were compared at baseline using t-tests, chi-square tests and nonparametric tests, as appropriate. The primary analysis was an intent-to-treat (ITT) comparison of the differences between treatment groups in change in gait speed 60 days following initiation of treatment. Statistical inference regarding the difference between treatment groups was based on the estimated coefficient for a treatment group indicator variable in an ANCOVA model with 60-day change in walking speed as the response variable, and explanatory variables that include an indicator of treatment group and baseline gait speed. A similar approach was used for all secondary and tertiary outcomes. Demographic or clinical characteristics that were found to be statistically different between the groups and could plausibly have a confounding effect on the relationship between group and the outcome were included in the models. Cumulative adverse events (falls, ED visits, and hospitalizations) were calculated at 30, 60, 90 and 180 days. The median cumulative number and occurrence (yes/no) of each adverse event at each timepoint were compared using Wilcoxon rank-sum and chi-square tests, respectively. The median cumulative deaths were calculated at 180 days and compared using a Wilcoxon rank-sum test.

RESULTS

Between 4/15/16 and 3/30/21, 150 eligible Veterans were enrolled and randomized to PHIT (n=75) or comparison group (n=75) (Appendix D). The average gait speed for the PHIT group and the comparison group were not significantly different (comparison group= 0.49 m/s, SD=0.19; PHIT=0.50 m/s, SD=0.17, p=0.75). The PHIT group had significantly more female participants (comparison group=8%; PHIT=23%, p=0.01). Baseline characteristics did not significantly differ between groups otherwise (Table 1, Supplementary Table S1).

Table 1.

Demographics and Clinical Measures at Baseline

Characteristic	Comparison Group (N=75)	PHIT (N=75)	p-value
Age	76.7 (8.8) 76.0 (56, 95)	76.6 (9.0) 77.0 (59, 93)	0.90
Gender
Female	6 (8%)	17 (23%)	0.01
Male	69 (92%)	58 (77%)
Race (N=149)
Caucasian	58 (78%)	50 (67%)	0.23
Black	10 (14%)	18 (24%)
Other	6 (8%)	7 (9%)
Hispanic (N=143)	7 (10%)	14 (19%)	0.12
Education (N=149)
Less than High School	5 (7%)	3 (4%)	0.53
High School Diploma	17 (23%)	23 (31%)
Some College	21 (28%)	26 (35%)
Associate’s	10 (14%)	8 (11%)
Bachelor’s/Master’s/ Doctoral	21 (28%)	15 (20%)
Marital Status (N=140)
Married/Cohabitating	29 (44%)	33 (45%)	0.84
Single	6 (9%)	9 (12%)
Widowed	13 (20%)	16 (22%)
Divorced/Separated	18 (27%)	16 (22%)
Height (cm) (N=143)	174.2 (8.1)	171.6 (9.7)	0.08
Weight (kg) (N=144)	88.1 (24.5)	84.9 (24.6)	0.44
Functional Comorbidity Index (N=147)	4.5 (2.2) 4 (0, 10)	4.8 (2.1) 5 (0, 10)	0.40
Patient Activation Measure-10 Score (N=140)	60.8 (13.0) 59.3 (40.9, 100)	62.1 (14.0) 59.3 (40.9, 100)	0.57
Patient Activation Measure-10 Level (N=140)	2.7 (0.83) 3 (1, 4)	2.8 (0.86) 3 (1, 4)	0.21

Functional comorbidity index (FCI) scores range from 0-18, with higher scores representing a greater number of comorbidities.

Patient Activation Measure (PAM) scores range from 0 to 100, with higher scores representing higher self-management and understanding of health conditions. PAM Levels are the corresponding sub-categories from 1-4, wherein a higher Level indicates higher patient activation.

There was no statistically significant difference in the estimated 60-day change in gait speed when controlling for baseline gait speed and gender (Supplementary Table S1). The changes were in the hypothesized direction (improvement), but there were also no differences in 60-day change across all other outcomes (Table 2).

Table 2.

Model Estimates for 60-Day Change

	Comparison Group		PHIT		Estimated Difference in Change (SE)	p- value
Model	N	LS Mean Change (95% CI)	N	LS Mean Change (95% CI)
Gait Speed (m/s)	58	0.08 (0.04, 0.12)	59	0.08 (0.04, 012)	−0.0003 (0.03)	0.99
Short Physical Performance Battery	57	1.04 (0.42, 1.65)	58	0.83 (0.22, 1.44)	−0.20 (0.44)	0.64
Modified Physical Performance Test	54	2.78 (1.57, 4.00)	54	2.72 (1.50, 3.93)	−0.07 (0.87)	0.94
Timed Up and Go	47	−3.20 (−5.85, −0.55)	47	−3.03 (−5.68, −0.38)	0.17 (1.9)	0.93
Knee Extensor Strength, Right (kg)	35	−0.05 (−2.60, 2.50)	41	−0.24 (−2.59, 2.11)	−0.19 (1.76)	0.92
Knee Extensor Strength, Left (kg)	34	−0.06 (−2.51, 2.39)	39	−0.80 (−3.09, 1.49)	−0.74 (1.7)	0.67
Grip Strength (kg)	52	0.88 (−0.88, 2.64)	52	−1.08 (−2.84, 0.68)	−1.96 (1.3)	0.13
Veterans RAND-12 Item Health Survey
Physical Component Score	58	2.89 (0.44, 5.34)	60	2.45 (0.04, 4.86)	−0.44 (1.7)	0.80
Mental Component Score	58	2.77 (−0.17, 5.71)	59	−0.60 (−3.51, 2.31)	−3.37 (2.11)	0.11
St. Louis University Mental Status	57	0.49 (−0.44, 1.43)	58	0.24 (−0.69, 1.17)	−0.25 (0.67)	0.71
Life Space Mobility	49	5.18 (0.51, 9.86)	49	7.33 (2.65, 12.0)	2.14 (−4.53, 8.81)	0.53
Step Count	27	309 (−184, 802)	31	628 (169, 1088)	319 (−368, 1006)	0.36

Note. LS=Least-squares; SE=standard error. Modified Physical Performance Test (mPPT) scores range from 0 (unable to complete) to 4 (performed quickly and easily) with a maximum score of 28.

Short Physical Performance Battery (SPPB) scores range from 0-12, with higher scores indicative of better performance.

Veterans RAND-12 Item (VR-12) is summarized into a Physical Component Score (PCS, range 0-100) and a Mental Component Score (MCS, range 0-100), with a higher score indicating a better outcome.

Life Space Mobility scores range from 0 to 120, with higher scores representing greater mobility.

St. Louis University Mental Status examination scores range from 0 to 30. Scores of 27 to 30 are considered normal in a person with a high school education. Scores between 21 and 26 suggest a mild neurocognitive disorder. Scores between 0 and 20 indicate dementia.

There were no significant differences in the median number of falls, ER visits, hospitalizations, or deaths (examined cumulatively) between groups at any timepoint, nor between the occurrence of these adverse events (at least one) at any timepoint (Table 3). Fewer participants experiences falls and hospitalizations in the PHIT group versus the comparison group at 30 days, but this difference was not statistically significant (falls: 19 comparison group participants, 11 PHIT participants; hospitalizations: 16 comparison group, 10 PHIT participants).

Table 3.

Adverse Events

Through Timepoint	Event	Comparison group (N = 75)		PHIT (N = 75)		Chi-square and Wilcoxon p-value
		Number of participants with at least one event (%)	Number median (range)	Number of participants with at least one event (%)	Number median (range)
0-30 Days	All falls	19 (25%)	0 (0, 3)	11 (15%)	0 (0, 5)	ALO: 0.10 N: 0.15
	Falls requiring medical attention	5 (7%)	0 (0, 2)	5 (7%)	0 (0, 2)	ALO: >0.99 N: 0.99
	ED visits	11 (15%)	0 (0, 2)	11 (15%)	0 (0, 5)	ALO: >0.99 N: 0.97
	Hospitalizations	16 (21%)	0 (0, 2)	10 (13%)	0 (0, 2)	ALO: 0.20 N: 0.19
0-60 Days	All falls	25 (33%)	0 (0, 3)	20 (27%)	0 (0, 5)	ALO: 0.37 N: 0.36
	Falls requiring medical attention	9 (12%)	0 (0, 2)	10 (13%)	0 (0, 2)	ALO: 0.81 N: 0.79
	ED visits	17 (23%)	0 (0, 2)	19 (25%)	0 (0, 7)	ALO: 0.70 N: 0.56
	Hospitalizations	20 (27%)	0 (0, 3)	20 (27%)	0 (0, 2)	ALO: >0.99 N: 0.88
0-90 Days	All falls	26 (35%)	0 (0, 3)	24 (32%)	0 (0, 5)	ALO: 0.73 N: 0.74
	Falls requiring medical attention	10 (13%)	0 (0, 2)	11 (15%)	0 (0, 2)	ALO: 0.81 N: 0.77
	ED visits	23 (31%)	0 (0, 2)	21 (28%)	0 (0, 7)	ALO: 0.72 N: >0.99
	Hospitalizations	22 (29%)	0 (0, 4)	22 (29%)	0 (0, 2)	ALO: >0.99 N: 0.67
0-180 Days	All falls	35 (47%)	0 (0, 11)	31 (41%)	0 (0, 11)	ALO: 0.51 N: 0.93
	Falls requiring medical attention	15 (20%)	0 (0, 3)	18 (24%)	0 (0, 2)	ALO: 0.55 N: 0.56
	ED visits	33 (44%)	0 (0, 5)	30 (40%)	0 (0, 7)	ALO: 0.62 N: 0.73
	Hospitalizations	30 (40%)	0 (0, 6)	33 (44%)	0 (0, 5)	ALO: 0.62 N: 0.85
	Deaths	7 (9%)	-	2 (3%)	-	0.17a

Note. All subjects are included in all denominators as the total counts are cumulative.

Abbreviation: ALO, at least one.

^aFisher’s exact test.

Fidelity:

Out of 12 possible PT home visits, PHIT participants had a mean of 10.4 (SD 2.9) visits; strengthening exercises and ADL training; motor control training exercises were completed in > 80% of sessions. UC participants had a mean of 9.9 (SD 3.6) visits which included therapeutic exercises, gait training, and review of home exercise programs. ADL training occurred less frequently during sessions (55%). Additional fidelity results are reported in Supplemental Materials.

DISCUSSION

Overall, our results indicate that both PHIT and standardized PT interventions were associated with clinically significant improvements in physical function, but no group differences in primary or secondary outcomes at 60 days were observed. Reassuringly, the safety profile of both interventions was similar. Overall, our findings indicate high-intensity PT rehabilitation immediately following acute hospitalization is safe to implement among, older adult Veterans with hospital-associated deconditioning and multimorbidity in the home health setting.

Prior work indicates high-intensity approaches are efficacious for older adults.^11,36 The current findings extend this work by showing that safety does not differ between high-intensity and standard PT among older adults with multimorbidity who were recently hospitalized. Both interventions resulted in similar increases in gait speed and SPPB scores, which approached or exceeded established minimally clinically important difference (MCID) values (MCID for gait speed=0.1 meters/second; MCID for SPPB=1.0 point).³⁷

There are several potential explanations for the lack of differences between groups. First, the comparison group received a standardized, multicomponent protocol with equivalent frequency comparable to the high-intensity group, which may have diluted the effects of the PHIT intervention. Additionally, the amount of PT visits may allow for a large enough dose to see differences between high and low-intensity interventions, especially in the absence of robust adherence with unsupervised home exercise. We believe the relatively low overall home exercise adherence could have attenuated the potential gains. In contrast, using a higher volume of high-intensity exercise (daily PT session for at least 60 minutes 5 days per week), with a comparable sample of older adults with multimorbidity in SNF found greater gains in function; this approach may also be needed to achieve superior high-intensity outcomes in the home health setting.³⁸ Finally, we speculate that because both groups received front-loaded PT and a greater number of overall home health rehabilitation visits than typical, these elements may be key for rehabilitation in this population.

Strengths and Limitations

Our study has several strengths. First, this randomized clinical trial was conducted in a real-world home health care setting and was completed with high fidelity despite challenges associated with the COVID-19 pandemic. Furthermore, the target Veteran population included individuals with hospital-associated deconditioning and multimorbidity who are often difficult to study. Our study is not without limitations. First, patients were recruited from a variety of settings (e.g., hospital, SNF, home) which may have captured patients at different phases in their acute illness recovery. The analytic sample includes a small number of participants (n=10) with pandemic related deconditioning which introduces heterogeneity, however these participants were balanced across groups (n=5 comparison group, n=5 PHIT). Second, despite our best efforts and encouragement, home exercises logs were poorly completed by patients, and thus it was difficult to ascertain overall compliance. Last, our sample size is relatively small³⁹ from a single geographic region, which may limit precision and generalizability.

CONCLUSIONS

For Veterans with hospital-associated deconditioning and multimorbidity who received home health PT for deconditioning, high-intensity PT was safe but not more effective than standardized PT in improving function. We also observed potential protective effects of PHIT PT early after hospitalization, which need to be replicated in future studies.

Supplementary Material

SupinfoS2

The supplemental materials include additional details regarding methods and results. For methods, we provide details regarding study design, detailed descriptions of the PHIT intervention and standardized PT intervention (comparison group), and additional details regarding our power and sample size calculation. For results, we present additional fidelity results and a supplementary table detailing outcomes at baseline of the comparison and PHIT groups (Supplementary Table S1).

Key points:

• For older adult Veterans with hospital-associated deconditioning and multimorbidity who received home health PT for deconditioning, high-intensity home health PT was safe and effective in improving physical function, but not more effective than a standardized PT program.

• Participants in both groups experienced increases in gait speed (primary outcome) that met or exceeded established clinically important thresholds.

• Future investigation is warranted as individual patient factors, consolidation of PT visits into one month, higher frequency of PT visits post-hospitalization, and home exercise program compliance may have impacted the current study results.

Why does this matter:

• High-intensity PT can be implemented in home health settings among Veterans with hospital-associated deconditioning and multimorbidity without concerns regarding safety, and with comparable effectiveness to standardized PT. Further work should explore characteristics of responders and non- responders to high-intensity PT to better direct rehabilitation intervention priorities.