Optimal Intensity and Duration of Walking Rehabilitation in Patients With Chronic Stroke

This randomized clinical trial compares vigorous vs moderate walking exercise to assess the optimal intensity and training duration to maximize improvement in walking capacity among stroke survivors with walking limitations.

Key Points

Question

For walking exercise in patients with chronic stroke, what is the optimal intensity (vigorous vs moderate) and minimum duration (4, 8, or 12 weeks) to maximize immediate improvement in walking capacity?

Findings

In this randomized clinical trial that enrolled 55 stroke survivors, 6-minute walk test gains after 12 weeks of training were 71 m with vigorous training intensity vs 27 m with moderate training intensity, a significant difference. Within the vigorous intensity group, walking capacity significantly increased after each 4-week training block.

Meaning

The findings suggest that optimal dosing for walking exercise for patients with chronic stroke may include at least 12 weeks of training at vigorous intensity.

Abstract

Importance

For walking rehabilitation after stroke, training intensity and duration are critical dosing parameters that lack optimization.

Objective

To assess the optimal training intensity (vigorous vs moderate) and minimum training duration (4, 8, or 12 weeks) needed to maximize immediate improvement in walking capacity in patients with chronic stroke.

Design, Setting, and Participants

This multicenter randomized clinical trial using an intent-to-treat analysis was conducted from January 2019 to April 2022 at rehabilitation and exercise research laboratories. Survivors of a single stroke who were aged 40 to 80 years and had persistent walking limitations 6 months or more after the stroke were enrolled.

Interventions

Participants were randomized 1:1 to high-intensity interval training (HIIT) or moderate-intensity aerobic training (MAT), each involving 45 minutes of walking practice 3 times per week for 12 weeks. The HIIT protocol used repeated 30-second bursts of walking at maximum safe speed, alternated with 30- to 60-second rest periods, targeting a mean aerobic intensity above 60% of the heart rate reserve (HRR). The MAT protocol used continuous walking with speed adjusted to maintain an initial target of 40% of the HRR, progressing up to 60% of the HRR as tolerated.

Main Outcomes and Measures

The main outcome was 6-minute walk test distance. Outcomes were assessed by blinded raters after 4, 8, and 12 weeks of training.

Results

Of 55 participants (mean [SD] age, 63 [10] years; 36 male [65.5%]), 27 were randomized to HIIT and 28 to MAT. The mean (SD) time since stroke was 2.5 (1.3) years, and mean (SD) 6-minute walk test distance at baseline was 239 (132) m. Participants attended 1675 of 1980 planned treatment visits (84.6%) and 197 of 220 planned testing visits (89.5%). No serious adverse events related to study procedures occurred. Groups had similar 6-minute walk test distance changes after 4 weeks (HIIT, 27 m [95% CI, 6-48 m]; MAT, 12 m [95% CI, −9 to 33 m]; mean difference, 15 m [95% CI, −13 to 42 m]; P = .28), but HIIT elicited greater gains after 8 weeks (58 m [95% CI, 39-76 m] vs 29 m [95% CI, 9-48 m]; mean difference, 29 m [95% CI, 5-54 m]; P = .02) and 12 weeks (71 m [95% CI, 49-94 m] vs 27 m [95% CI, 3-50 m]; mean difference, 44 m [95% CI, 14-74 m]; P = .005) of training; HIIT also showed greater improvements than MAT on some secondary measures of gait speed and fatigue.

Conclusions and Relevance

These findings show proof of concept that vigorous training intensity is a critical dosing parameter for walking rehabilitation. In patients with chronic stroke, vigorous walking exercise produced significant and meaningful gains in walking capacity with only 4 weeks of training, but at least 12 weeks were needed to maximize immediate gains.

Trial Registration

ClinicalTrials.gov Identifier: NCT03760016

Introduction

While the majority of stroke survivors eventually regain the ability to walk without physical assistance from another person, most in the chronic phase of recovery still lack sufficient walking capacity (ie, speed and endurance) to resume normal daily activities.^1,2 Both the neuromotor impairment and the aerobic deconditioning underlying these limitations in walking capacity can be targeted simultaneously with locomotor exercise (ie, task-specific walking practice at sufficient intensity to challenge the neuromotor and cardiopulmonary systems).^3,4,5 With this intervention, training intensity appears to be a critical dosing parameter associated with outcomes.⁶ However, the optimal intensity for improving walking recovery remains unknown.

Most studies of poststroke locomotor exercise have tested moderate-intensity aerobic training (MAT), typically involving treadmill walking at mean training heart rates (HRs) between 40% and 60% of the heart rate reserve (HRR).^7,8 These studies have found that MAT is associated with significantly greater improvements in walking capacity (6-minute walk distance) and other outcomes than lower-intensity walking practice or nonwalking exercise. Consequently, this approach is currently recommended in stroke rehabilitation guidelines.^3,4,5

Studies also suggest that a more vigorous training intensity (>60% of HRR vs 40%-60% of HRR) could augment outcomes,⁶ but a vigorous intensity can be difficult to achieve and sustain for many persons with stroke.^9,10,11 Thus, high-intensity interval training (HIIT) has emerged as a promising strategy for poststroke locomotor exercise.⁶ This method involves bursts of fast walking alternated with recovery periods and is designed to enable higher sustained intensities¹² at lower perceived exertion than high-intensity continuous exercise.¹³ Initial stroke studies indicated that locomotor HIIT is feasible, sufficiently safe for further study, can improve walking capacity and other outcomes,^{6,14,15,16,17} and has some early signs of potential for eliciting greater improvements than MAT.^14,15

However, to our knowledge, no prior studies have been designed to compare outcomes of HIIT and MAT after stroke. In addition, most prior studies of HIIT after stroke^14,17,18,19 used short 4-week training durations. Therefore, it was unknown whether a longer 12-week training duration typical of many stroke and cardiac rehabilitation trials could yield better outcomes. Further, poststroke HIIT studies have often focused primarily on improving aerobic fitness among persons with mild stroke or transient ischemic attack who had normal, near-normal, or unreported baseline walking function.^15,16 To generate evidence that is generalizable to clinical stroke rehabilitation, it is important to target patients with walking limitations (eg, speed ≤1.0 m/s²⁰).

Therefore, the purpose of this multicenter randomized clinical trial was to assess the optimal training intensity (vigorous vs moderate) and the minimum training duration (4, 8, or 12 weeks) needed to maximize immediate improvement in walking capacity among stroke survivors with chronic walking limitations. We hypothesized that (1) 4 weeks of HIIT would elicit significantly greater improvement in walking capacity compared with 4 weeks of MAT and, (2) compared with 4 or 8 weeks of HIIT, 12 weeks of HIIT would elicit significantly greater improvements in walking capacity and increased benefit compared with MAT.

Methods

This randomized clinical trial was approved by the University of Cincinnati institutional review board and was performed in rehabilitation and exercise laboratories at the University of Cincinnati, University of Delaware, and University of Kansas Medical Center with participant enrollment from January 2019 to April 2022. Written informed consent was obtained from participants prior to participation. The trial was prospectively registered at ClinicalTrials.gov (NCT03760016); the trial protocol is given in Supplement 1 and is published elsewhere.²¹ Additional methods, results, and discussion are provided in the eAppendix in Supplement 2. This trial followed the Consolidated Standards of Reporting Trials (CONSORT) reporting guideline.

Participants

Participants were recruited from the community via outreach to clinicians and support groups, advertisement, existing databases, and health record systems. Screening included a medical history and record review, physical examination, Fugl-Meyer lower limb motor function assessment,²² depression questionnaire,²³ 10-m walk test,²⁰ treadmill screening test,⁸ resting electrocardiography (ECG), and a treadmill-graded exercise test with ECG monitoring.⁸

Inclusion criteria were (1) age of 40 to 80 years at the time of consent, (2) single stroke for which the participant sought treatment 6 months to 5 years prior to consent, (3) walking speed of 1.0 m/s or less,²⁰ (4) ability to walk 10 m over ground with assistive devices as needed and no continuous physical assistance from another person,⁷ (5) ability to walk at least 3 minutes on a treadmill at 0.13 m/s (0.3 mph) or faster,⁸ (6) stable cardiovascular condition (American Heart Association class B²⁴), and (7) ability to communicate with investigators, follow a 2-step command, and correctly answer consent comprehension questions. Exclusion criteria were (1) exercise testing uninterpretable for ischemia or arrhythmia,²⁴ (2) evidence of significant arrhythmia or myocardial ischemia on a treadmill ECG-graded exercise test in the absence of a more definitive negative result from clinical testing,²⁴ (3) hospitalization for cardiac or pulmonary disease within the past 3 months, (4) implanted pacemaker or defibrillator; (5) significant ataxia or neglect (National Institutes of Health Stroke Scale item scores >1),²⁵ (6) severe lower limb spasticity (Ashworth Scale scores >2 of 4 for knee flexion, knee extension, or ankle dorsiflexion),²⁶ (7) recent history of illicit drug or alcohol misuse or significant mental illness, (8) major poststroke depression (Patient Health Questionnaire score ≥10)²³ in the absence of depression management by a health care professional, (9) currently participating in physical therapy or another interventional study, (10) recent botulinum toxin injection to the paretic lower limb (<3 months previously) or planning to have it in the next 4 months, (11) foot drop or lower limb joint instability in the absence of an adequate stabilizing device (eg, ankle foot orthosis), (12) clinically significant neurologic disorder other than stroke, (13) unable to walk outside the home prior to stroke, (14) other significant medical condition likely to limit improvement or jeopardize safety, (15) pregnancy, and (16) previous exposure to fast treadmill walking in the past year.

Race and ethnicity data were ascertained by self-report and were included in the analysis to describe the study sample. Race categories included American Indian, Asian, Black/African American, and White, and ethnicity categories included Hispanic/Latinx and not Hispanic/Latinx.

Randomization

After screening and baseline testing, eligible participants were randomized to HIIT or MAT in a 1:1 ratio using the web-based REDCap randomization module. Randomization was stratified by site and baseline walking limitation severity (severe, <0.4 m/s; mild or moderate, 0.4-1.0 m/s)¹⁷ and used randomly permuted block sizes of 2 or 4. The study statistician (J.C.K.) generated the random allocation sequence and kept it confidential to ensure concealed allocation. When participants arrived for the first training session, the treating therapist or study coordinator initiated the randomization within REDCap to irreversibly assign the participant to a group before revealing the assignment.

Interventions

Target training volume for both groups was 45 minutes 3 times per week for 12 weeks. Participants were not physically assisted with stepping but were guarded and assisted for injury prevention as needed by a physical therapist. During overground training bouts, participants walked back and forth in a corridor. When training on a treadmill, participants wore a fall-protection harness without any weight support and used a handrail for balance support.⁸ Training HR zones were calculated using the HRR method: (peak HR – resting HR) × percent HRR target + resting HR,²⁴ where peak HR was the highest HR obtained in any previous exercise test and resting HR was the resting HR while standing from the current day. Each session, the training protocol for both groups included a 3-minute warm-up of overground walking at 30% to 40% of the HRR, a 10-minute bout of overground HIIT or MAT, a 20-minute bout of treadmill HIIT or MAT, another 10-minute bout of overground HIIT or MAT, and a 2-minute cool down at 30% to 40% of the HRR.

HIIT and MAT Protocols

The HIIT group used a short-interval HIIT protocol that was specifically developed for locomotor exercise after stroke.^{14,17,27,28,29} It involved repeated 30-second bursts of walking at maximum safe speed, alternated with 30- to 60-second passive recovery periods (standing or seated rest as tolerated), targeting a mean aerobic intensity above 60% of the HRR. Following current best-practice guidelines for stroke rehabilitation,^3,4 the MAT group performed continuous walking practice with speed adjusted to maintain an initial target HR of 40% ± 5% of the HRR, progressing by 5% of the HRR every 2 weeks up to 60% of the HRR as tolerated.⁸

Adverse Event Monitoring

At each study visit, participants were asked about adverse events (AEs) in general and then were specifically queried about falls, injuries, pain, and fatigue. During exercise testing and training, safety monitoring included HR, blood pressure, and continuous observation for other signs or symptoms of cardiorespiratory insufficiency, worsening neurologic impairments, or orthopedic injury using accepted stopping criteria.²⁴ Adverse events were systematically categorized and graded using Common Terminology Criteria for Adverse Events³⁰ by a centralized physician (O.O.A.) blinded to group randomization.

Outcome Assessment

Outcomes were assessed by blinded raters before randomization (baseline) and after 4, 8, and 12 weeks of training. The primary outcome measure was walking capacity, measured by distance walked during the 6-minute walk test.³¹ Secondary measures included self-selected and fastest speeds, measured by the 10-m walk test²⁰; self-reported fatigue over the past 7 days, measured by the Patient Reported Outcomes Measurement Information System (PROMIS) Fatigue Scale³²; and aerobic capacity, measured by oxygen consumption rate (VO₂) at the ventilatory threshold during a treadmill-graded exercise test.³³

Statistical Analysis

Statistical analyses followed a prespecified plan,²¹ used an intent-to-treat approach, and were done with SAS, version 9.4 (SAS Institute Inc). The primary study statistician (J.C.K.) remained blinded to treatment groups until after the primary analysis.

Hypothesis Testing

Primary hypothesis testing used a linear model with the 6-minute walk test as the dependent variable and fixed effects for treatment group, testing time point (baseline, 4 weeks, 8 weeks, and 12 weeks), group-by-time interaction, study site, study site–by-time interaction, baseline walking limitation severity, and baseline walking limitation severity–by-time interaction, with unconstrained covariance between repeated testing time points within the same participant and a significance threshold of 2-sided P < .05. Secondary outcomes were tested using the same model, with false discovery rate correction to the significance threshold to control for multiple testing time points (change at 4 weeks, 8 weeks, and 12 weeks) and the 5 primary or secondary outcome measures. These analyses handled any missing data with the method of maximum likelihood, which assumes that data were missing at random. A sensitivity analysis assessed how much the results depended on the missing at random assumption (eAppendix in Supplement 2).

Sample Size

This study was powered to detect a between-group difference of 20 m in 6-minute walk test change³⁴ using the software GLIMMPSE.³⁵ Covariance parameters and MAT group changes were estimated from pilot work.¹⁴ Calculations indicated a total target sample size of 40. To account for up to 20% attrition, we initially planned to randomize 50 participants. However, after having to withdraw 4 participants due to a COVID-19 pandemic shutdown, we opted to increase the enrollment target to 55. This decision was made before any analysis of outcome data.

Results

Recruitment and Retention

We consented and screened 74 participants to find 55 who were eligible for randomization (Figure 1). The mean [SD] age was 63 [10] years; 19 participants (34.5%) were female, and 36 (65.5%) were male. One participant (1.8%) was American Indian, 3 (5.5%) were Asian, 11 (20.0%) were Black/African American, 1 (1.8%) was Hispanic/Latinx, and 40 (72.7%) were White. The mean (SD) time since stroke was 2.5 (1.3) years. The mean baseline 6-minute walk test distance was 239 (132) m, and the mean baseline self-selected gait speed was 0.63 (0.31) m/s. Twenty-seven participants were randomized to HIIT and 28 to MAT. Baseline characteristics were similar between groups (Table 1). In total, participants attended 1675 of 1980 planned treatment sessions (84.6%; HIIT, 800 of 972 [82.3%]; MAT, 875 of 1008 [86.8%]), with individual participants attending a mean (SD) of 30.5 (8.4) sessions (HIIT, 29.6 [8.3]; MAT, 31.3 [8.5]). Altogether, participants also attended 197 of 220 planned testing sessions (89.5%), leaving 23 of 220 outcome assessments missing (10.5%; HIIT, 13 of 108 [12.0%]; MAT, 10 of 112 [8.9%]).

Figure 1. CONSORT Flow Diagram.

AE indicates adverse event; HIIT, high-intensity interval training; and MAT, moderate-intensity aerobic training.

^aPrior myocardial infarction with peri-infarct ischemia on most recent perfusion scan.

^bFlare-ups of preexisting back pain.

^cRecurring hamstring strain or soreness.

Table 1. Baseline Participant Characteristics.

Characteristic	Participantsa
	HIIT (n = 27)	MAT (n = 28)
Age, mean (SD), y	63.8 (9.9)	61.5 (9.9)
Sex
Female	11 (40.7)	8 (28.6)
Male	16 (59.3)	20 (71.4)
Race
American Indian	0	1 (3.6)
Asian	0	3 (10.7)
Black/African American	4 (14.8)	7 (25.0)
White	23 (85.2)	17 (60.7)
Hispanic/Latinx ethnicity	1 (3.7)	0
Side of paresis
Left	14 (51.9)	14 (50.0)
Right	13 (48.1)	14 (50.0)
Stroke type
Ischemic	15 (55.6)	19 (67.9)
Hemorrhagic	12 (44.4)	9 (32.1)
Stroke chronicity, mean (SD), y	2.7 (1.4)	2.2 (1.2)
Aphasia present	7 (25.9)	7 (25.0)
Comorbid medical conditions
Any	26 (96.3)	27 (96.4)
History of myocardial infarction	1 (3.7)	2 (7.1)
Atrial fibrillation	2 (7.4)	5 (17.9)
History of lightheadedness and/or syncope	4 (14.8)	5 (17.9)
COPD	3 (11.1)	1 (3.6)
Diabetes
Any	8 (29.6)	9 (32.1)
Insulin dependent	2 (25.0)	2 (22.2)
Arthritis	8 (29.6)	10 (35.7)
History of cancer	4 (14.8)	3 (10.7)
Prescribed medication
Antihypertensive
Any	23 (85.2)	20 (71.4)
β-Blocker	7 (25.9)	11 (39.3)
Statin	22 (81.5)	23 (82.1)
Antispasmodic	11 (40.7)	12 (42.9)
Antidepressant	20 (74.1)	13 (46.4)
Pain reliever	11 (40.7)	11 (39.3)
Reported falling in previous 3 mo	8 (29.6)	7 (25.0)
BMI, mean (SD)	28.9 (5.3)	28.7 (4.6)
Functional ambulation categoryb
Mean (SD)	3.3 (0.6)	3.4 (0.8)
2	2 (7.4)	5 (17.9)
3	14 (51.9)	8 (28.6)
≥4	11 (40.7)	15 (53.6)
Orthotic and/or assistive device use	18 (66.7)	19 (67.9)
Orthotic device use	12 (44.4)	12 (42.9)
Solid ankle foot orthosis	4 (14.8)	2 (7.1)
Articulated or flexible ankle foot orthosis	7 (25.9)	7 (25.0)
Assistive device use
Any	16 (59.3)	18 (64.3)
Single-point cane	8 (29.6)	11 (39.3)
Narrow-based quad cane	5 (18.5)	1 (3.6)
Wide-based quad cane	3 (11.1)	5 (17.9)
Front-wheeled walker	0	1 (3.6)
Fugl-Meyer lower limb motor score, mean (SD)c	23.8 (5.1)	22.8 (5.1)
Self-selected gait speed
Mean (SD), m/s	0.65 (0.29)	0.62 (0.33)
Mean (SD), % normative value	50.5 (23.3)	47.3 (25.1)
<0.4 m/s	7 (25.9)	7 (25.0)
6-min Walk test distance
Mean (SD), m	248 (136)	230 (130)
Mean (SD), % normative value	48.5 (26.3)	44.3 (26.8)
Ventilatory threshold VO2
Mean (SD), mL/kg/min	12.1 (3.9)	11.6 (3.9)
Mean (SD), % normative value	93.1 (24.4)	82.5 (26.2)
Peak VO2
Mean (SD), mL/kg/min	14.3 (4.4)	14.0 (4.8)
Mean (SD), % normative value	60.4 (15.9)	55.1 (19.3)

Abbreviations: BMI, body mass index (calculated as weight in kilograms divided by height in meters squared); COPD, chronic obstructive pulmonary disease; HIIT, high-intensity interval training; MAT, moderate-intensity aerobic training; VO₂, oxygen consumption rate.

^a Data are presented as the number (percentage) of participants unless otherwise indicated.

^b Category 2 indicates dependent, level 1 (requires continuous or intermittent light touch to assist balance or coordination from no more than 1 person during ambulation on level surfaces to prevent falling); category 3, dependent, supervision (can physically ambulate on level surfaces without manual contact of another person but requires standby guarding of no more than 1 person because of poor judgment, questionable cardiac status, or the need for verbal cuing to complete the task); and category ≥4, independent, level surfaces (can ambulate independently on level surfaces).

^c Range, 0 to 34; higher scores indicate less motor impairment.

Treatment Fidelity and Adverse Events

Compared with MAT, HIIT involved significantly higher training speed, HR, blood lactate level, and perceived exertion with significantly lower step count (Figure 2 and eAppendix in Supplement 2). For example, the peak training speed was 161% of the baseline fastest 10-m speed in the HIIT group and 96% in the MAT group. The mean bout steady state HR (compared with true resting HR and baseline peak HR) was 75% of the HRR in the HIIT group and 59% in the MAT group. The mean session step count was 2847 in the HIIT group and 3532 in the MAT group. There were no serious AEs related to study procedures and no significant between-group differences in any AE categories (eAppendix in Supplement 2).

Figure 2. Treatment Intensity.

Values are model estimates for each bout of each session, averaging data across participants. Filled circles indicate overground 1; open circles, overground 2; and crosses, treadmill. A, Speed data are peak values within a bout. B, Steady state mean heart rate (HR) excludes the first 3 minutes of the bout. For treatment monitoring, the percentage of HR reserve (HRR) was relative to the standing resting HR and the most current peak HR, which could increase after sessions 12 and 24. Shading indicates aerobic intensity zones: orange, moderate; gray, vigorous. HIIT indicates high-intensity interval training; MAT, moderate-intensity interval training.

Outcome Changes

After 4 weeks of training, there was no significant difference in primary outcome measure (6-minute walk test) changes between groups (HIIT: 27 m [95% CI, 6-48 m]; MAT: 12 m [95% CI, –9 to 33 m]; mean difference, 15 m [95% CI, −13 to 42 m]; P = .28). The HIIT group improved significantly more than the MAT group after 8 weeks (58 m [95% CI, 39-76 m] vs 29 m [95% CI, 9-48 m]; mean difference, 29 m [95% CI, 5-54 m]; P = .02) and 12 weeks (71 m [95% CI, 49-94 m] vs 27 m [95% CI, 3-50 m]; mean difference, 44 m [95% CI, 14-74 m]; P = .005) of training (Figure 3 and Table 2).

Figure 3. Primary Outcome of 6-Minute Walk Test Changes During 12 Weeks of High-intensity Interval Training (HIIT) or Moderate-intensity Aerobic Training (MAT) in Patients With Chronic Stroke.

Values are model estimates; error bars, 95% CIs; and shading, clinically important difference in distance (20-50 m).

^aP < .05 for the false discovery rate in between-group differences in change from baseline.

^bP < .05 for within-group change from baseline.

^cP < .05 for within-group change between consecutive time points.

Table 2. Primary and Secondary Outcome Changes During 12 Weeks of HIIT or MAT Among Patients With Chronic Stroke^a.

Outcome, time point	Mean change vs baseline (95% CI)		Mean difference in change between HIIT and MAT (95% CI)	P value
	HIIT (n = 27)	MAT (n = 28)
Primary outcome
Walking capacity, 6-min walk distance, m
Baseline, mean (SD)b	196 (98)	177 (99)	19 (−31 to 69)	.44
4 wk	27 (6 to 48)	12 (−9 to 33)	15 (−13 to 42)	.28
8 wk	58 (39 to 76)	29 (9 to 48)	29 (5 to 54)	.02c
12 wk	71 (49 to 94)	27 (3 to 50)	44 (14 to 74)	.005c
Secondary outcomes
Self-selected gait speed, 10-m walk test, m/s
Baseline, mean (SD)b	0.52 (0.20)	0.49 (0.21)	0.03 (−0.07 to 0.13)	.56
4 wk	0.11 (0.06 to 0.15)	0.02 (−0.02 to 0.07)	0.08 (0.02 to 0.15)	.009c
8 wk	0.14 (0.08 to 0.20)	0.06 (0.00 to 0.12)	0.08 (0.00 to 0.15)	.04
12 wk	0.19 (0.13 to 0.25)	0.06 (0.00 to 0.12)	0.13 (0.05 to 0.20)	.003c
Fastest gait speed, 10-m walk test, m/s
Baseline, mean (SD)b	0.70 (0.32)	0.62 (0.32)	0.08 (−0.08 to 0.24)	.33
4 wk	0.22 (0.16 to 0.28)	0.01 (−0.05 to 0.07)	0.21 (0.13 to 0.29)	<.001c
8 wk	0.24 (0.17 to 0.32)	0.09 (0.01 to 0.17)	0.15 (0.05 to 0.25)	.003c
12 wk	0.28 (0.19 to 0.37)	0.09 (−0.01 to 0.18)	0.20 (0.08 to 0.32)	.002c
PROMIS Fatigue Scale T scored
Baseline, mean (SD)b	52.4 (8.6)	50.8 (8.7)	1.7 (−2.7 to 6.0)	.45
4 wk	−1.7 (−4.0 to 0.6)	0.0 (−2.3 to 2.3)	−1.7 (−4.8 to 1.3)	.26
8 wk	−3.0 (−5.5 to −0.5)	1.0 (−1.6 to 3.6)	−4.0 (−7.3 to −0.7)	.02c
12 wk	−1.1 (−3.7 to 1.5)	−0.1 (−2.7 to 2.5)	−1.0 (−4.4 to 2.5)	.57
VO2, mL/kg/min
Baseline, mean (SD)b	11.5 (4.0)	10.9 (4.0)	0.6 (−1.4 to 2.6)	.55
4 wk	1.3 (0.4 to 2.2)	0.4 (−0.5 to 1.3)	0.9 (−0.4 to 2.1)	.16
8 wk	2.0 (0.7 to 3.3)	1.7 (0.3 to 3.1)	0.3 (−1.4 to 2.1)	.72
12 wk	1.9 (0.4 to 3.4)	1.4 (−0.2 to 3.0)	0.5 (−1.5 to 2.6)	.60

Abbreviations: HIIT, high-intensity interval training; MAT, moderate-intensity aerobic training; PROMIS, Patient Reported Outcomes Measurement Information System; VO₂, oxygen consumption rate.

^a All models were adjusted for study site, study site–by-time interaction, baseline walking limitation severity, and baseline walking limitation severity–by-time interaction.

^b Baseline rows show the mean (SD) for each group and the mean difference (95% CI) in baseline values between groups.

^c Statistically significant after false discovery rate correction across time points and measures (primary and secondary).

^d Score range, 33.1-77.8; lower scores indicate less fatigue.

Among the secondary outcome measures, both groups exhibited significant increases in self-selected gait speed, fastest gait speed, and ventilatory threshold VO₂ at various time points relative to baseline (Table 2), with the HIIT group showing significantly greater increases than the MAT group in self-selected and fastest gait speeds. Only the HIIT group had significantly decreased PROMIS Fatigue Scale T scores compared with the MAT group and only at the 8-week time point (HIIT, −3.0 [95% CI, –5.5 to −0.5]; MAT, 1.0 [95% CI, –1.6 to 3.6]; mean difference, −4.0 [95% CI, −7.3 to −0.7]; P = .02). Given the significant decrease in fatigue scores in the HIIT group at the 8-week time point but not the 12-week time point, we also examined HIIT group PROMIS Fatigue Scale changes between the 8- and 12-week time points and found no significant difference (T-score increase of 1.9; 95% CI, –0.5 to 4.3; P = .12).

Discussion

Results of this multicenter randomized clinical trial give proof of concept that the vigorous training intensity of HIIT is a crucial dosing parameter for locomotor exercise in patients with chronic stroke. Findings also indicated that locomotor HIIT can produce significant and meaningful gains in walking capacity in 4 weeks but that a training duration of at least 12 weeks is needed to maximize immediate gains in walking capacity with this intervention. Our primary hypothesis that HIIT would improve walking capacity significantly more than MAT after 4 weeks of training was not supported. However, the gap between groups continued to widen over time, and HIIT elicited significantly greater improvement in walking capacity than MAT after 8 weeks and 12 weeks of training in the prespecified primary analysis, supporting the secondary hypotheses. Among secondary outcome measures, HIIT also elicited significantly greater improvements than MAT in gait speed (self-selected and fastest at all time points) and fatigue (after 8 weeks of training) after adjustment for multiple comparisons (Table 2). In addition, intensive safety monitoring was successful at identifying expected AEs, and the results suggest that poststroke HIIT is sufficiently safe for further study.

Within the HIIT group, the mean 6-minute walk test gain after 4 weeks of training was comparable to that found in prior single-site 4-week pilot testing,¹⁷ confirming successful multicenter implementation. As hypothesized, gains in walking capacity and other outcomes were even larger after 8 and 12 weeks of HIIT, with no apparent plateau. These findings suggest that most prior poststroke HIIT studies have used insufficient training durations and that HIIT durations of at least 12 weeks should be considered in the future. Outpatient therapy durations longer than 12 weeks (36 sessions) are not likely to be routinely feasible in the current model for clinical rehabilitation due to reimbursement constraints^36,37 and adherence issues.³⁸ Thus, a 12-week duration for a clinician-led HIIT program may be an optimal standard to target. Future studies might also consider transitioning to home- or community-based training after clinician-led HIIT to test longer training durations.

Both groups showed significant improvement in aerobic capacity (ventilatory threshold VO₂). The lack of significant between-group differences in aerobic capacity gains suggests that the greater improvement in walking capacity with HIIT vs MAT may have been primarily driven by neuromotor rather than cardiopulmonary adaptations. However, it is also possible that this lack of significant between-group differences was due to greater measurement error for VO₂ measurement vs other outcomes or confounding effects of gait efficiency changes on VO₂ during exercise testing (improved gait efficiency decreases measured VO₂).^33,39,40,41

The intensity-dependent improvement (decrease) in PROMIS Fatigue Scale T scores after 8 weeks of training (HIIT, −3.0; MAT, 1.0; P = .02) is especially noteworthy because identifying a treatment for poststroke fatigue has been highlighted as a research priority.⁴² This finding is consistent with prior neurophysiologic evidence showing that poststroke fatigue is associated with lower corticomotor excitability⁴³ and that HIIT is associated with significantly greater acute increases in corticomotor excitability than MAT.²⁹ Interestingly, participants randomized to HIIT reported a substantial decrease in fatigue during a period when they were going to the research sites multiple times per week for training sessions that commonly induced temporary fatigue. One possible explanation for this apparent paradox is that the acute central motor activation effects of poststroke HIIT counterbalanced or exceeded the effects of fatigue.²⁹ It is also possible that the greater longitudinal increases in walking capacity and exercise tolerance with HIIT may have reduced fatigue with everyday activities, especially when coupled with significant reductions in the metabolic cost of gait.

The HIIT group improvements and between-group differences in fatigue (Table 2) were only statistically significant after 8 weeks of training and not after 4 or 12 weeks of training. Yet, all these effect estimates were in the same direction, suggesting that the lack of significance at the 4- and 12-week time points may have been due to insufficient statistical power rather than an exclusive fatigue benefit at only the 8-week time point. Conversely, it is possible that the significant improvement in fatigue observed after 8 weeks of HIIT was due to chance, but such improvement was also observed in a prior 4-week pilot study,¹⁷ suggesting otherwise. It is also possible that fatigue decreases begin returning toward baseline values with more extended HIIT durations, but this tendency in the HIIT group between 8 and 12 weeks of training was not statistically significant (fatigue T-score increase, 1.9; 95% CI, –0.5 to 4.3; P = .12). More definitive fatigue assessment in a larger trial with follow-up testing is warranted.

Strengths and Limitations

A strength of this trial is that it successfully recruited participants with a variety of comorbidities, assistive and orthotic device needs, and walking limitations. Further, study participants had a mean (SD) self-selected gait speed of 0.63 (0.31) m/s, which is representative of patients typically seen in outpatient stroke rehabilitation (eg, 0.59 m/s⁴⁴). Thus, the results should be generally applicable in this setting. Another major strength was our systematic AE data collection with blinded adjudication, which is rare in the rehabilitation literature.

In terms of rigor, this study used randomization, concealed allocation, assessor blinding, and intent-to-treat analysis to minimize risk of bias and only had approximately 10% missing outcome data (HIIT, 12%; MAT, 9%). We also added the rigor of multicenter implementation earlier than usual in the rehabilitation research continuum,⁴⁵ increasing confidence in the reproducibility and generalizability of the results. In addition, we included more rigorous assessment than many clinical trials about how missing data might influence the results. Further, the comparison group received an active, best-practice treatment (MAT),^3,4,5 with mean intensity and volume of walking exercise exceeding previous reports for clinical rehabilitation in outpatient or extended inpatient settings (eg, HRR, 59% vs 24%-47%^46,47; steps per session, 3532 vs 501-886^48,49; and sessions attended, 31 vs 9-25^37,50,51).

A primary limitation of this trial was its relatively modest sample size (N = 55) since it was designed to be an early proof-of-concept study. This meant that the trial was likely underpowered to detect all but very large differences in AEs between HIIT and MAT. The small sample size is also probably why the primary findings were not fully robust to different plausible assumptions about the true values of the missing data (eAppendix in Supplement 2). Thus, a larger trial is needed to provide more definitive evidence to guide clinical practice. The sensitivity analysis also validated the general intuition that larger samples are needed to have more confidence in the results of clinical trials.

Another important limitation was the lack of follow-up testing to assess sustained effects after treatment ended. This also made it more challenging to interpret the results of the PROMIS Fatigue Scale since participants were asked to report on periods in which they were engaged in the study treatment at 4 weeks, 8 weeks, and 12 weeks. We attempted to mitigate this issue by specifically asking participants to answer questions based on their usual activities outside the study, but we are skeptical that this strategy was completely successful.

It is also unclear how well the results might generalize to stroke survivors who need continuous physical assistance from another person to walk (ie, functional ambulation category <2 of 5⁵²). Less than 10% of stroke survivors appear to require this level of assistance after inpatient rehabilitation.⁵³ However, this percentage is higher in earlier subacute stroke (eg, 49%-58%).^53,54 Thus, future studies seeking to adapt the current training protocols for subacute stroke might consider providing the minimum necessary assistance to enable walking practice for participants who need it.⁵⁵

Conclusions

In this randomized clinical trial involving locomotor exercise among stroke survivors with chronic walking limitations, vigorous training intensity appeared to be significantly better than moderate intensity for eliciting immediate improvements in walking capacity and other outcomes. With this vigorous-intensity locomotor exercise, 12 weeks of training produced greater gains than shorter durations.

Supplement 1.

Trial Protocol

Supplement 2.

eAppendix

eReferences

Supplement 3.

Data Sharing Statement