Effects of High-Intensity Interval Training After Stroke (The HIIT Stroke Study) on Physical and Cognitive Function: A Multicenter Randomized Controlled Trial

Objective: To assess the effects of high-intensity interval training (HIIT) on physical, mental, and cognitive functioning after stroke. Design: The HIIT Stroke Study was a single-blind, multicenter, parallel-group randomized controlled trial. Setting: Specialized rehabilitation units at 3 Norwegian hospitals. Participants: Adult stroke survivors (N=70) 3 months to 5 years after a first-ever stroke. Mean age was 57.6§9.2 years and 58.7§9.2 years in the intervention and control groups, respectively. Interventions: Participants were randomized to standard care in combination with 4£4 minutes of treadmill HIIT at 85%-95% of peak heart rate or standard care only. Outcomes: Outcomes were measured using physical, mental, and cognitive tests and the FIM and Stroke Impact Scale. Linear mixed models were used to analyze differences between groups at posttest and 12-month follow-up. Results: The intervention group showed a significant treatment effect (95% confidence interval [CI]) from baseline to posttest on a 6-minute walk test of 28.3 (CI, 2.80-53.77) meters (P=.030); Berg Balance Scale 1.27 (CI, 0.17-2.28) points (P=.025); and Trail Making Test Part B (TMT-B; 24.16 [CI, 46.35 to 1.98] s, P=.033). The intervention group showed significantly greater improvement on TMT-B at the 12-month followup (25.44 [CI, 49.01 to 1.87] s, P=.035). The control group showed significantly greater improvement in total Functional Independence Measure score with a treatment effect of 2.37 (CI, 4.30 to 0.44) points (P=.016) at 12-month follow-up. No significant differences were identified between groups on other outcomes at any time point. Conclusions: HIIT combined with standard care improved walking distance, balance, and executive function immediately after the intervention compared with standard care only. However, only TMT-B remained significant at the 12-month follow-up. Archives of Physical Medicine and Rehabilitation 2021;102:1683−91 2021 The Authors. Published by Elsevier Inc. on behalf of The American Congress of Rehabilitation Medicine. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/) Clinical Trial Registration No.: NCT02550015. Supported by grants from the Liaison Committee for Education, Research and Innovation in Central Norway.

There is significant evidence that exercise and repetitive task training improve fitness, balance, and walking capacity after a stroke. 1,2 Despite being a key outcome, the benefit of exercise on poststroke cognitive function remains underinvestigated. 3,4 In a meta-analysis, Oberlin et al 5 demonstrated the beneficial effect of physical activity on cognition, with the most promising results for attention and processing speed. It has also been shown that exercise may be a potential treatment for preventing or reducing depressive symptoms in individuals after stroke. 6 However, more research on different training modalities is needed. 3 High-intensity interval training (HIIT) has emerged as a popular and potent exercise modality. 7,8 This intervention has been shown to be safe and feasible for use in the population with stroke with promising findings on both peak oxygen consumption (VO 2 peak) and mobility. 7,[9][10][11][12][13] To better understand the effectiveness of HIIT in stroke rehabilitation, high-quality randomized controlled trials assessing the effect of HIIT on important clinical outcomes including physical, mental, and cognitive functions are warranted. 12,13 Recently, Steen Krawcyk et al 14 reported the results of a home-based HIIT intervention program and found no effects on cardiorespiratory fitness, cognitive domains, or general well-being in a sample of patients who suffered mild strokes. However, these findings might be explained by a lack of control over exercise protocols based on indirect heart rate calculations. In the HIIT Stroke Study, we found that an 8-week treadmill intervention, in addition to standard care with supervised HIIT and individually set exercise intensity and progression, was superior to standard care with respect to VO 2 peak immediately after the intervention. However, improvement was not maintained long term. 15 The overall aim of this study was to investigate the effect of the intervention on secondary outcomes of the HIIT Stroke Study, such as physical, mental, and cognitive functions.

Study design, setting, and participants
The HIIT Stroke Study was approved by the Regional Committee of Medical and Health Research Ethics (REC No. 2015/563) and conducted in accordance with the institutional guidelines at each participating hospital. The conduct and reporting of this study were guided by the CONSORT guidelines. 16 Based on Norwegian regulations and conditions for informed consent, the data set is not publicly available. The study was registered at Clinicaltrials.gov before the inclusion of participants (NCT02550015).

Study design
This was a parallel-group, multicenter randomized controlled trial with blinded outcome assessments. The study was conducted in collaboration with 3 rehabilitation hospitals in Norway: Trondheim University Hospital, Sunnaas Rehabilitation Hospital, and A lesund Hospital. Individuals who agreed to participate underwent an initial baseline assessment, received information about physical activity recommendations after stroke, and underwent a posttest at 8 weeks and a follow-up assessment 12 months after inclusion. Inclusion began in September 2015 and data collection ended in December 2017. Groups of 10-22 participants were included at one time.

Participants
Inclusion criteria were adult stroke survivors aged >18 years diagnosed with first-ever stroke (ischemic or hemorrhagic), able to walk independently with or without an assistive device, minimum 3 months and maximum 5 years poststroke, and living near testing and training locations. In addition, approval to participate was required from the study's responsible medical doctor as well as a score of 0-3 on the modified Rankin Scale (mRS). Exclusion criteria were unstable cardiac conditions (ie, serious rhythm disorder, valve malfunction), poorly controlled resting blood pressure (>180/100), other conditions where VO 2peak was contraindicated, subarachnoid hemorrhage, and participation in another ongoing intervention study.

Randomization and data collection
After completing baseline assessments, participants were randomized into the control group (receiving standard care) or the intervention group (receiving HIIT in addition to standard care). Randomization was stratified according to hospital site and performed by a web-based randomization and data collection system developed and administered by the Unit for Applied Clinical Research, Faculty of Medicine and Health Sciences, Norwegian University of Science and Technology, Trondheim, Norway. All outcome assessments were standardized and performed in the same order at each test time and test site.

The HIIT intervention
Each exercise session began with 10 minutes of warmup to increase heart rate to approach the individually calculated intensity training zone. The HIIT protocol comprised 4-minute intervals at 85%-95% of peak heart rate (HRpeak) separated by 3 minutes of active breaks (reduced walking intensity) at heart rate 50%-70% of HRpeak. Intensity was tracked with heart rate monitors. a An experienced physical therapist (PT) helped adjust treadmill speed and inclination to maintain heart rate in the prescribed training zone. Training was performed 3 times per week for 8 weeks, for a total of 24 training sessions.
As part of the intervention, participants attended 3 follow-up meetings in groups of 5-11 participants (at 1 month, 4 months, and 8 months after completing the intervention). The meetings were led by the PT who administered the HIIT sessions at each site.  The purpose of these meetings was 2-fold: to motivate participants to continue to be physically active after the intervention period and to provide instructions for how to complete a weekly training diary.

Standard care
At baseline testing, all participants allocated to the control group were given information about the benefits of high levels of physical activity, in line with this study's hypothesis and with the Norwegian guidelines on physical activity. The control group did not receive any further follow-up other than contact before each test session to schedule a testing time point.

Outcome measures
Physical assessments The 6-minute walk test (6MWT) was used to assess walking distance. We adhered to the protocol approved by the American Thoracic Society, 17 but for testing at 1 hospital (n=36), we used a 25meter walkway and not the recommended 30-meter. The test was performed once for each time point.
We assessed maximal gait speed across 10 meters (10-Meter Walk Test) with a flying start/stop. The mean time (in meters per second) from 2 trials was used for analysis. 18 Dynamic and static balance were assessed with the Berg Balance Scale (BBS). The BBS includes 14 items; each item is ranked on a 4-point scale for a total score of 56. 19 Functional mobility was assessed using the Timed Up and Go (TUG) test. This measures the amount of time it takes for participants to rise from an armchair, walk 3 meters at their own pace, turn around, walk back to the chair, and sit down. 20 TUG was performed twice on each occasion and the mean of the 2 trials was used.

Mental health
The Hospital Anxiety and Depression Scale (HADS) is a questionnaire including 14 questions: 7 cover anxiety symptoms and 7 cover depressive symptoms. 21 HADS is not a diagnostic tool but can be used to score symptoms of anxiety (HADS-A) and depression (HADS-D). The score of each subscale ranges from 0-21 points. A higher score indicates more severe symptoms. It is recommended that the subscales be used separately. 22

Cognitive assessments
The Montreal Cognitive Assessment test (MoCA) is a valid and reliable screening test of global cognitive function in the population with stroke. 23 MoCA covers 9 cognitive domains including attention, concentration, executive functions, memory, language, visuospatial ability, conceptual thinking, calculations, and orientation. The total score is 30; the minimum score is 0.
The Trail Making Test (TMT) Parts A and B was used to assess executive function. The TMT requires a variety of mental abilities, including letter and number recognition, mental flexibility, visual scanning, and motor function. TMT has been shown to correlate with processing speed and cognitive fluidity. 24 Health-related quality of life The Stroke Impact Scale is a self-report questionnaire that evaluates health-related quality of life after stroke. It covers 8 domains: strength, hand function, mobility, activities of daily living, memory, communication, emotion, and handicap. Each domain is scored on a metric of 0-100 and higher scores indicate better selfreported health. 25

Functional independence
The FIM is an 18-item outcome measure comprising both cognitive (5 items) and motor (13 items) subscales and has shown acceptable reliability in the population with stroke. 26 Each item assesses the level of assistance required to complete an activity of daily living on a 7-point scale. The sum scores range from 18-126, with higher scores indicating greater functional independence. 27

Statistical analyses
The sample size calculation for the HIIT Stroke Study was based on the primary outcome (VO 2 peak) with an expected VO 2 peak of 2.27 §0.45 L/min at inclusion and an expected 10% improvement in the intervention group and a 2% deterioration in the control group, as reported in Gjellesvik et al. 15 Hence, each group required 32 participants to reach a power of 80% with a P value of .05. We also expected a 10% dropout rate, so the final number was set at 70 participants. Linear mixed models were used to evaluate differences between the groups for all outcomes across the 3 time points, with treatment group, time point, their interactions, and hospital as fixed factors. All analyses were adjusted for baseline differences for the outcome of interest. In this model, the coefficients for the interaction terms give the estimated treatment effects at posttest and follow-up. This corresponds to equation (2d) in Twisk et al. 28 The level of significance was set at P<.05. IBM SPSS Statistics, versions 24 and 25 b were used for all statistical analyses.

Results
Of 70 randomized participants, 29 (42%) were women; 36 were allocated to the intervention group and 34 to the control group. Mean age was 57.6 §9.2 and 58.7 §9.2 in the intervention and control groups, respectively. Demographic characteristics were similar between groups at baseline (table 1). In the intervention group,  2; fig 2). The total FIM score showed a significantly lower mean score (114.4 §9.7 vs 119.6 § 8.3 points, P=.016) in the intervention group compared with the control group at the 12-month follow-up (see table 2).
There were no significant differences between the groups for the remaining outcome measures at any time point (tables 2-4; fig 2).

Discussion
The secondary results from the HIIT Stroke Study showed that 24 HIIT treadmill sessions in combination with standard care is superior to standard care only for improving walking distance, balance, and executive function immediately after the intervention. The improvement in executive function remained significant at the 12-month follow-up. In contrast, the total FIM score showed a different development from baseline to 12-month follow-up. The HIIT Stroke Study is one of the first randomized controlled stroke trials to assess the effect of HIIT on physical, mental, and cognitive functions after stroke.
As shown in figure 2, the initial 40-meter (8.3%) improvement in the 6MWT in the intervention group was followed by a decline until the end of follow-up, whereas the control group declined slightly over the whole period, resulting in no significant differences between the groups at 12-month follow-up. Previous research in the stroke population also reported benefits for walking capacity after treadmill interventions, 3,9,10,29 indicating that improved cardiovascular fitness and balance might contribute to increased walking distance.
The overall change from baseline to 12-month follow-up in the intervention group was 25 m, which should be considered a clinically important change, because a 23-m improvement on 6MWT is defined as a significant change for patients with coronary artery disease. 30 Although the intervention group was invited to regular meetings every third month to encourage maintaining an active lifestyle, the participants did not adhere well to this part of the   intervention. This was confirmed by a lack of differences in activity levels between the intervention and control groups as measured by the activPAL, as reported in Gjellesvik et al. 15 A closer followup might be needed to maintain recommended activity levels long term after stroke. 31 In contrast to the immediate improvement in 6MWT and all other physical and cognitive outcomes (see fig 2; table 2), the intervention group declined in functional independence by the 12month follow-up, as indicated by the total FIM score. This unexpected finding could be explained by poor interrater reliability of the FIM as shown in a previous study. 32 Despite efforts to have the same test personnel available at each test point, some variability occurred at posttest and 12-month follow-up. The fact that the FIM was assessed by interview rather than by observation could also have contributed to a reduction in the reliability of the results. Normally, variability in interrater reliability would balance out between groups, but because more participants in the intervention group with larger variability in test scores had experienced more severe strokes, this could partially explain the results.
Of the few studies conducted on interval training with a stroke population, 9,10,33 only 1 investigated its effect on cognitive function. 14 According to a review by Cumming et al, 34 the majority of research on cognitive function after physical activity and exercise interventions is limited because of relatively small heterogeneous trials that do not specify cognitive function as the primary outcome. The present study found no benefit of HIIT on cognitive function as measured by the MoCA. However, the time to complete the TMT-B improved from pre-to postintervention and remained significant in regard to the intervention at the 12-month follow-up. One reason for this improvement could be that the intervention increases arousal and wakefulness, as proposed by McMorris et al, 35 and cerebral blood volume, thereby enhancing executive function. 36 This might be a random finding that needs to be investigated in a future study using cognitive function as the primary outcome.
A major strength of the present study was the high-quality multicenter trial with randomization of participants and blinded outcome assessment conducted by experienced clinicians who had no information about previous results. In addition, the relatively balanced proportion between sexes is a major strength because female participants are often underrepresented in most training studies. 37 The modest time commitment required for this training protocol is also a strength because 1 barrier to maintaining high levels of physical activity after stroke is lack of time. 38 Another strength was the absence of adverse effects (except initial muscle soreness in a few participants) and the very good compliance with the training protocol and interval sessions for participants in the intervention group. Each training session was monitored by experienced PTs, treadmill speed and incline and heart rate were monitored during each interval, and active breaks were allowed in addition to rating of perceived exertion after each interval. Collectively, these recordings confirmed that training intensities were in the prescribed range.
The mixed model statistical analyses have strengths compared with methods based on complete cases. Complete case analysis is unbiased if data are missing completely at random, and a linear mixed model also results in unbiased estimates under the less restrictive missing at random assumption. The mixed model uses data from all participants, including those with partially missing data, thus avoiding a loss of statistical power.

Study limitations
This study's primary limitation is the relatively young age and high functional level of participants in both groups, limiting generalizability to the general stroke population. Participants were recruited mainly from 2 specialized hospitals treating patients 67 years or younger, making the sample notably younger than the general Norwegian stroke population. 39 In addition, the scores on the physical tests confirm our participants' high levels of function. Recruitment bias is also likely because potential participants who were already interested in physical activity and exercise were more likely to participate. Additionally, precise information on the activities for the participants in the control group between assessments is lacking. Furthermore, the dropout rate was higher than anticipated for the functional and cognitive outcome measures compared with the primary outcome (VO 2peak ) at the 12month follow-up, reducing the power of our results. Another weakness is that our study lacks data on participants' educational levels, aphasia, visual deficits, and fine motor deficits, which are potential confounders when interpreting the results from the MoCA and the TMT.
The wide variability of time since stroke onset in our sample is also a limitation and a potential confounder in the interpretation of our findings.
In a future exercise trial using cognition as the primary outcome, information from cerebral images about stroke location should be collected because cortical and subcortical strokes show symptoms within different cognitive domains. Such a study should also include a comprehensive battery of cognitive assessments addressing all aspects of cognition. A larger sample size should be included to enable subgroup analysis because ischemic and hemorrhagic strokes represent different pathologies. Choosing an outcome showing good measurement properties and responsiveness within the cognitive domain of interest and being aware that hemiparesis might interfere with cognitive responses are also important to consider. 40

Conclusions
Secondary results from the HIIT Stroke Study showed a significant increase in walking distance, balance, and executive function immediately after 8 weeks of supervised treadmill HIIT intervention. Only the improvement in scores on TMT-B remained significant at follow-up. This may suggest that the HIIT intervention can produce an effect in executive function. However, such a relationship needs to be confirmed in a future trial with cognitive function as the primary outcome of an HIIT intervention.