‘Magic’ number of treadmill sessions needed to achieve meaningful change in gait speed post stroke: a systematic review

Abstract

The purpose of this systematic review was to determine the number of treadmill training sessions needed to make a meaningful change in gait speed for chronic stroke survivors. Relevant databases were searched up through February 2020. Articles were included if they fit the following criteria: stroke onset greater than six months, intention to treat with traditional treadmill training, and gait speed included as an outcome. Change in gait speed post intervention was utilized to classify treadmill groups as responders (at least 0.1 meters/second change) or nonresponders (less than 0.1 meters/second change). 17 articles met our criteria, resulting in a total of 19 intervention groups. Ten groups were classified as responders and completed a mean of 30.5 sessions within 6 weeks, while nonresponders completed 20.4 sessions within 10 weeks, indicating that at least 30 treadmill sessions (preferably in a period of 10 weeks and at least 40 minutes per session) is necessary to reach a meaningful change in gait speed. Although these trends were noted between the responder and nonresponder groups, no firm conclusions can be drawn regarding the ‘magic’ number of sessions chronic stroke survivors should perform given the low correlation between number of sessions and change in gait speed.

Introduction:

Stroke is a leading cause of disability in the United States and typically leads to reduced ambulation in more than 50% of the individuals affected post stroke.¹ Due to the high prevalence of impaired mobility and gait after stroke, physical rehabilitation is critical for recovery of motor function. The use of treadmill training has become a commonplace intervention for motor rehabilitation after stroke, especially for those individuals with moderate to high levels of recovery. This method of training employs a variety of Kleim & Jones’ principles of neuroplasticity such as specificity, repetition, and intensity.² Treadmill training has also proven to be an effective and feasible approach to implement gait training and assist with motor recovery after stroke.³

The use of treadmill training in individuals after stroke has been widely studied. Treadmill training has been found to improve gait speed in individuals after stroke both immediately after training and beyond the training period.⁴ Studies and reviews have delved into ideal dosage parameters such as treatment duration, session duration and session frequency. Aerobic exercise prescription recommendations for individuals after stroke include completing aerobic exercise three to five times per week for 20 to 40 minutes to accumulate health benefits, however no specific treadmill exercise recommendations have been provided.⁵ The 2017 Cochrane Review by Mehrholz and colleagues investigated frequency and duration of treadmill training, however no specific recommendations regarding exercise dosage were reached.⁶ The authors concluded that the majority of studies that incorporate treadmill training for individuals after stroke implement a treatment frequency or intensity of three to five times per week and a total treatment duration of four weeks or more. The parameter of volume (total number of training minutes) was explored in the meta-analyses by Abbasian et al. (2018), and the authors determined that training volume was the most crucial element of treadmill training.⁷ This study found that high volume (>500 minutes of total training) at a low intensity (meters/second (m/s)) resulted in the most benefit on motor function outcomes for individuals after stroke.

Much of the literature, as seen above, presents large ranges for exercise dosage parameters such as duration of sessions, total training duration and training frequency. However, none so far have explored the number of treadmill training sessions that individuals after stroke should participate in, specifically to improve mobility related function. The American Heart Association and American Stroke Association adult stroke recovery guidelines recommend that exercise programs be individually tailored and structured around general exercise guidelines for stroke survivors since no specific exercise recommendations have been formulated specifically for the stroke population.⁸ Furthermore, the above guidelines also do not provide specific guidance regarding frequency for outpatient therapy services and merely state that the treatment plans have to be reviewed and approved by the referring physician on a regular basis. In contrast, the Canadian stroke guidelines recommend that individuals who are participating in outpatient or home rehab receive therapy two to five days per week for a minimum of eight weeks in this setting, which would result in completion of 16 to 40 therapy sessions.⁹ Medicare allows a cap of $2,080 per year to be utilized for both physical and speech therapy combined.¹⁰ The cost per physical therapy visit can vary, however on average the cost is approximately $100 per visit which would result in a total of 20 allowed visits for both speech and physical therapy.¹¹ Medicare allows individuals to exceed this threshold, but care has to be proven to be medically necessary. Therefore, if a stroke survivor had Medicare and required both outpatient speech and physical therapy there is a high likelihood that they would not be able to receive the 16 to 40 physical therapy visits recommended by the international stroke guidelines.⁹ A study by Godwin et al. further investigated outpatient costs within the first year after stroke and found that the utilization of rehabilitation services decreased over the year.¹² The authors were unable to determine a clear cause of this decreased utilization of rehabilitation services but postulated that this could be due to “changes in insurance reimbursement or to the individuals reaching a plateau in their functional improvement and being denied rehabilitation.”¹² However, stroke survivors can still improve their mobility with aerobic exercise after the subacute phase¹³ and they may be hitting a plateau in their improvement due to not receiving sufficient dosage or volume of gait training to demonstrate a change in their function. Therefore, the aim of this systematic review is to provide both clinicians and researchers a recommendation for the ideal number of treadmill training sessions required to see a meaningful improvement in gait speed in order to allow for the possibility of flexibility in training regimens, while still providing a concrete dosage recommendation that does not have a large range. By providing clinicians and researchers a specific recommendation for the number of treadmill training sessions to achieve the minimal clinically important difference (MCID) in gait speed in chronic stroke survivors, it may help to ensure that these individuals are receiving the sufficient dosage needed to avoid hitting a plateau in their recovery.^14,15

Although there are a variety of gait variables that could improve after treadmill training in stroke, gait speed is one of the more commonly utilized parameters as it can be easily assessed in a clinical setting and requires minimal equipment.¹⁶ Gait speed has been widely studied in individuals after stroke and has come to be known as the “sixth vital sign” given its ability to predict an individual’s mobility level in the community and fall risk.^16–18 Due to the clinical applicability of gait speed, we aimed to objectively determine the number of treadmill training sessions necessary to have a significant clinical change in gait speed, which has been determined to be 0.1 m/s in individuals after chronic stroke.¹⁹

Treadmill training has been implemented with a variety of additional measures and adjuncts such as body weight support, mirror therapy, various forms of feedback (visual, auditory, manual), peripheral nerve stimulation, cortical stimulation and exoskeleton assisted training. However, treadmill training combined with adjuncts has not yet been demonstrated to be superior to traditional treadmill training for chronic stroke survivors.^6,20–23 For the purposes of this review, treadmill training without the inclusion of other adjunct interventions was investigated to keep this review focused on treadmill training intervention alone. In this systematic review we sought to answer the following question: How many treadmill training sessions are needed to achieve a clinical change in gait speed for individuals with chronic stroke?

Methodology

Data Sources and Searches

A search of the literature through February 2020 was performed in the following databases: Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE, PubMed, Web of Science and Physiotherapy Evidence Database (PEDro). Combinations of the following terms were used: stroke, gait, walking, gait speed, walking speed, treadmill, exercise, and 10 meter walk test (10MWT). There were no limitations to the years considered but the search was limited to articles written in English. An example of a search conducted in PubMed includes the following terms: stroke AND gait AND treadmill, along with the filters of human (see Supplementary Search Table for other search strategies). The study conforms to PRISMA guidelines for systematic reviews and reports the required information accordingly (see Supplementary Checklist).²⁴ The protocol was not published or registered prospectively.

Study Selection

Two authors (MB and SM) screened and assessed all articles based on titles, abstracts and full texts. Studies were selected based on the following inclusion criteria: intention to treat, included five or more human participants, intervention of treadmill training, individuals with chronic stroke (onset greater than six months), and an outcome measure of gait speed. The following were exclusion criteria for the review: subjects were not humans, participant receiving concurrent physical therapy treatment or any other intervention apart from treadmill training, use of split-belt or nonmotorized treadmill, use of adjuncts with treadmill training (i.e. central or peripheral stimulation, body weight support, mirror therapy, visual or tactile feedback, etc.), gait speed was not assessed before and after treadmill training, and outcome measures did not include gait speed. All study titles and abstracts were first screened utilizing the above inclusion and exclusion criteria. When a study could not be included or eliminated from the review based on the title or abstract alone, a full text review was performed. Whenever there was any uncertainty regarding inclusion or exclusion of a study, this was discussed between the 2 authors (MB and SM) until consensus was achieved. The authors came to a 100% agreement on articles to be included for this review.

Data Extraction and Quality Assessment

Study characteristics (i.e. author, year of publication, sample size), participant characteristics (i.e. stroke onset), characteristics of intervention (i.e. number of sessions, duration and intensity of treatment), and results were extracted by author (MB) from each article.

All of the included studies were appraised and assessed for quality using the PEDro tool.²⁵ The PEDro tool is a scale that is commonly used to assess and rate the methodological quality of randomized controlled trials.²⁶ The PEDro tool consists of 11 items, however the first item on the tool does not count towards the final score. Each study was rated and given a PEDro score, which can range from 0 to 10, to be categorized as poor (<4), fair (4–5), good (6–8) or excellent (9–10) quality.²⁷ PEDro scores were determined by the author MB, however when confirmed scores were available on the PEDro website, the author checked for agreement between self-rated and already confirmed PEDro scores.

Data Synthesis and Analysis

The results for the included studies focused on gait speed and change in gait speed and were either obtained from the studies or calculated by the authors based on data provided. The studies varied in how gait speed was collected and included the following methods: 10MWT, two minute walk test (2MWT), “straight-away-walk” (SAW) during a get up and go mobility task, or the GAITRite. All changes in gait speed were converted to m/s to allow for comparison between studies.

The Spearman’s rank correlation was utilized to test the association between change in gait speed and the number of treadmill training sessions. The association between change in gait speed and target session duration (minutes) and total training length (weeks) was also performed. All statistical tests were run at α = 0.05 level of significance and SAS software 9.4 (SAS Institute Inc., Cary NC) was utilized for statistical data analysis. Correlations greater than 0.8 were considered very strong, between 0.6 an 0.8 as moderately strong, between 0.3 and 0.5 as fair and below 0.3 as poor.^28,29

For further exploratory observations, selected studies were stratified into one of two categories based on their change in walking speed after the treadmill training intervention. The treadmill training groups within each study were categorized as those that achieved the MCID for change in gait speed (responders) and those that did not achieve the MCID (nonresponders). For the purpose of this review, the MCID utilized for change in gait speed was 0.1 m/s.¹⁹ A few stroke studies have used 0.16 m/s as the MCID for gait speed change, however this MCID is based on stroke survivors within their first 60 days of recovery³⁰ and as the focus of this review is chronic stroke survivors, we decided to use the MCID guidelines provided by Perera el al.¹⁹

Role of the Funding Source

The funders played no role in the design, conduct, or reporting of this study.

Results

Study and Participant Characteristics

Seventeen articles met the inclusion criteria (Figure 1). The studies by Park et al. and Boyne et al. had more than one traditional treadmill training group as a part of their study design, therefore the total number of intervention groups we included in our analyses was 19 rather than 17.^31,32 The study by Boyne et al. compared high intensity interval treadmill training against moderate intensity continuous treadmill training and the gait velocity outcomes were presented separately for each intervention group.³² The study by Park et al. presented its data separately for the two participant subgroups who were either placed in the slow or fast gait velocity group according to their baseline gait speed.³¹

Figure 1: Study selection process.

Flowchart illustrating study selection process using PubMed, PEDro, CENTRAL, MEDLINE, and Web of Science databases. The number of results yielded in the database search and the filters utilized are shown, as well as number of studies remaining after inclusion and exclusion criteria were applied.

Participant characteristics are detailed in Table 1 and training characteristics are outlined in Table 2. The total PEDro score was calculated for each study and 6 out of 17 studies were rated at good to excellent quality (Figure 3). The total pooled sample size of all stroke participants from all of the articles included in this review was 509 individuals regardless of the type of intervention that was performed, while the total pooled sample size of participants in the standard treadmill training groups specifically was 289. Participants mostly were greater than 6 months from stroke onset except for the study by Lee et al., in which the participant’s stroke onset was not specified.³³ Also, Bonnyaud et al. published two applicable studies with varying protocols, however the data for the treadmill only intervention group was identical, therefore one of the studies by Bonnyaud et al. was not included.³⁴

Table 1: Participant Characteristics.

Numbers reported as mean (standard deviation). Abbreviations: M: Male, F: Female, L: Left, R: Right, HIT: High-intensity interval Training, MCT: Moderate-intensity, continuous aerobic training, S: slow group, F: fast group.

Author (Year)	Total participants	Participants in Treadmill group	Gender (M/F)	Age (years)	Time since stroke onset (months)	Stroke type (Ischemic/Hemorrhagic)	Lesion Side (L/R)
Boyne (2016)	18	HIT 11 MCT 5	HIT: 7/4 MCT: 2/3	HIT: 59 (9) MCT: 57 (12)	HIT: 45.6 (34.8) MCT: 75.59 (24)	HIT: 9/11 MCT: 2/5	HIT: L 3/8 MCT: L 4/1
Kim(b) (2017)	30	15	7/8	50.73 (13.5)	11.27 (4.1)	6/9	7/8
Globas (2012)	38	18	14/4	68.6 (6.7)	60.2 (46.6)	18/0	14/4
Ivey (2011)	38	19	11/8	61 (8)	> 6	18/1	13/6
Macko (2005)	61	32	22/10	63 (10)	35 (29)	32/0	14/18
Lee(b) (2014)	30	15	12/3	64.3 (4.8)	-	-	-
Park (2013)	40	S: 20 F: 20	S: 7/3 F: 5/5	S: 51.9 (8.9) F: 53.4 (7.1)	S: 23.4 (6.4) F: 19.4 (7.6)	S: 7/3 F: 5/5	S: 4/6 F: 5/5
Lee(a) (2011)	40	11	4/7	61.9 (11.26)	>/=6	-	-
Meester (2019)	50	24	11/13	62.25 (15.53)	25.71 (32.70)	13/10 Both: 1	6/13 Midbrain: 5
Kim(a) (2018)	26	13	7/6	56.15 (10.82)	11.46 (3.80)	8/5	5/8
Silver (2000)	5	5	5/0	60.4 (2.7)	26 (4.6)	5/0	4/1
Mohammadi (2017)	10	10	10/0	60.60 (8.49)	33.10 (51.88)	-	6/4
Patterson (2008)	39	39	25/14	64 (8)	20.55 (64)	-	24/15
Yoon (2016)	30	9	5/4	61.2 (13.0)	17.1 (8.4)	5/4	5/4
Bonnyaud (2013)	26	13	8/5	52.46	49.79	-	5/8
Madhavan(a) (2016)	11	11	4/7	58 (8.95)	108 (71.64)	7/4	7/4
Madhavan(b) (2019)	16	16	10/6	57 (9.77)	76.44 (54.36)	13/3	11/5

Table 2: Training Protocol Characteristics.

Abbreviations: UE: Upper extremity, HIT: High-intensity interval Training, MCT: Moderate-intensity, continuous aerobic training, HRR: Heart Rate Reserve, HR: Heart Rate, BWS: Body Weight Support

Author (Year)	Training	Number of Sessions	Frequency (Sessions/Week)	Training Duration (weeks)	Session duration (minutes)	Treadmill Assist or UE support
Boyne (2016)	HIT: 30-sec bursts at maximum-tolerated speed alternated with 30–60 sec rest MCT: continuous walking with speed adjusted to maintain 45%±5% HRR. After 2 weeks target HR progressed to 50%±5%	12	3	4	25	Harness, elastic band, assist if needed to prevent injury, handrail on overhead support or least supportive handhold
Kim(b) (2017)	Speed and intensity were determined according to subjects’ comfort level, speed increased by 0.1 km/hr per week	20	5	4	30	No BWS, PT behind subject to assist with weight shifting
Globas (2012)	Training began 40–50% max HRR and progressed to 60–80% of the maximum HRR.	39	1 to 3	12	10–20--> 30–50	Handrail/forearm support, hip protection device (as needed)
Ivey (2011)	Training began at 40–50% HRR and gradually progressed to 60–70% HRR	78	3	26	10–20 --> 40	Handrail and harness support
Macko (2005)	Training started at low intensity 40–50% HRR and progressed to target intensity of 60–70% HRR	72	3	24	10–20 --> 40	-
Lee(b) (2014)	Began with minimum speed 0.1 km/hr & then subjects maintained their individual maximum speed	24	3	8	30	-
Park (2013)	Started at the lowest velocity and increased so subjects could walk at a comfortable speed	10	5 (2x day)	1	30	Handrail support, no BWS
Lee(a) (2011)	Determined gait speed using the self-paced capability of the treadmill and subjects’ gait speed at will	18	3	6	30	-
Meester (2019)	Target intensity was 55–85% age-predicted max HR for 30 min of walking	20	2	10	45	-
Kim(a) (2018)	When a subject was able to walk for more than 20 sec stably, the velocity increased by 0.1 km/hr per session	20	5	4	30	Harness/suspension device (no BWS)
Silver (2000)	Intensity was advanced as tolerated to 40 min at 60–70% max HRR	36	3	12	40	handrail support ad lib
Mohammadi (2017)	Walked at +40% treadmill self-selected speed which was re-calculated every week	12	3	4	32	Handrail support ad lib, no harness or BWS,
Patterson (2008)	Training started at 40–50% HRR and progressed to max intensity of 60–70% max HRR	72	3	24	10–20 --> 40	Handrail support
Yoon (2016)	Walked at favorite speed in week 1, increased speed by 5% in week 2, 5% increase in week 3, no increase in week 4	20	5	4	30	-
Bonnyaud (2013)	Walked continuously and increased speed progressively to match comfortable speed (determined during clinical exam) and then kept constant	1	1	1	20	Handrail support
Madhavan(a) (2016)	2 min fast walking phase alternated by recovery. During fast phase speed set to highest speed subject could walk safely and was held for 10 sec. If subject maintained the speed during 10 sec at end of 1st interval, then increased speed 10% during next interval. New speed was held for 10 sec at the end of 2nd interval followed by recovery.	1	1	1	40	Harness (no BWS)
Madhavan(b) (2019)	Same as Madhavan (2016) study above	12	3	4	40	Harness (no BWS), handrail support, person behind in case of trip or fall

Figure 3. PEDro Scores.

Physiotherapy evidence database (PEDro) tool and eligibility criteria. Each of the included studies was scored using the PEDro eligibility criteria. A green circle shows that the specific criteria was met, while a red circle signifies that the criterion was not met. The total PEDro score is shown at the bottom of the figure.

Training Protocols

The treadmill training protocols varied between the selected studies. Training duration ranged from 10 to 50 minutes per session, session frequency ranged from 1 to 5 times per week, and total intervention length ranged from 1 to 26 weeks. The intensity of treadmill training also varied significantly. Some studies performed high intensity interval treadmill training, while others implemented a progressive treadmill training program that modified either total training duration time, speed, or intensity (based on heart rate reserve or maximum heart rate) on a week to week basis.

The number of treadmill training sessions and change in gait speed

The number of treadmill training sessions also varied significantly; ranging from 1 to 78 sessions. The average number of sessions completed was 28.

18 out of 19 treadmill intervention groups had an improvement in gait speed after the intervention period, however the improvement was not necessarily statistically and/or clinically significant in all studies. 15 treadmill groups reported statistical significance in change in gait speed (p ≤ 0.05) and we observed that 10 groups reached the MCID of 0.1 m/s.¹⁹ The only treadmill group that did not demonstrate a change in gait speed, was the slow gait velocity group in the Park et al. study, who included only individuals with stroke with a slow gait velocity (< 0.5 m/s).³¹ The average change in gait speed for all of the studies after treadmill training was 0.12 m/s, however there was considerable variability among the studies with change in gait speed ranging from 0 to 0.56 m/s. Furthermore, the trends of the pooled data show that studies who performed between 12 to 36 sessions appeared to gain the most benefit from treadmill training and achieved the MCID for change in gait speed (Figure 2).

Figure 2. Change in gait speed based on number of treadmill training sessions completed.

Scatter plot demonstrates each group’s change in gait speed based on the number of treadmill training sessions that were completed. X-axis corresponds to the average change in gait speed from the beginning to the end of treadmill training for each group. Y-axis corresponds to total number of sessions that were completed for each group. Each data point represents each study. For those studies that included two treadmill training intervention groups each particular group is specified. The study by Park et al. (2013) had two treadmill training groups, one with slow baseline gait speed (denoted as Park-Slow) and one with fast baseline gait speed (denoted as Park-Fast).³¹ The study by Boyne et al. (2016) also included two treadmill intervention groups, High-intensity interval Training (HIT) and Moderate-intensity, continuous aerobic training (MCT).³² For those studies where there was another primary author with the same last name this was differentiated in the legend. Madhavan (a) corresponds to the 2016 study, while Madhavan (b) corresponds to the 2019 study.^42,47 Lee (a) represents the 2011 study, while Lee (b) represents the 2014 study.^33,45 Lastly, Kim (a) depicts the 2018 study and Kim (b) represents the 2017 study.^43,46 The dashed line at 0.1 m/s represents the MCID for change in gait speed for individuals after stroke (Perera et al, 2006).¹⁹ Those treadmill intervention groups that achieved the MCID have data points that are at or to the right of the dashed line and were classified as responders. The treadmill intervention groups that did not achieve the MCID have data points that are to the left of the dashed line and were classified as nonresponder.

Spearman’s correlation analyses showed a fair correlation between change in gait speed and number of sessions (R=0.33), but this relationship was not statistically significant (p = 0.17). Similarly, training length (in weeks) had a fair correlation (R=0.44) with change in gait speed and was not statistically significant (p=0.06). In contrast, target session duration (in minutes) had a moderate and statistically significant association with change in gait speed (R=0.62, p=0.005). Individuals with stroke who participated in longer training sessions showed a greater change in gait speed.

Response characterization

We characterized participant groups into subgroups based on the group average meeting the MCID for change in gait speed after training or not. 52.6% (10/19 groups) achieved the MCID for change in gait speed and were classified as responders,^{31,32,35–42}while the rest were classified as nonresponders.^{31–34,43–47}

The number of sessions for the responders ranged from 10 to 72 treadmill training sessions (average 30.5,standard deviation (SD) 24, median 20), resulting in an average change in gait speed of 0.19 m/s. The average number of sessions per week was approximately 4, the average length of the treadmill training intervention was 9.9 weeks, and the average total time spent performing treadmill training was 37.2 minutes. We noticed that studies utilizing target Heart Rate Reserve (HRR) to progress intensity reported a target HRR ranging from 55–85% (moderate-vigorous intensity) during training.^{31,32,35,37,39} Majority of the training groups in this category had to complete their training at a designated intensity as opposed to walking at a comfortable or self-paced speed.^35–38,40

For the nonresponders, number of sessions ranged from 1 to 78 sessions (average 20, SD 23, median 18). This subgroup had an average gait speed change of 0.04 m/s after the intervention. The average number of sessions per week was approximately 4, the average length of the treadmill training intervention was 5.9 weeks, and the average total time spent performing treadmill training was 30.6 minutes per session. Participants in the study by Ivey et al. completed the highest number of treadmill sessions (78) with an improvement in gait speed of 0.09 m/s, however the change did not meet the MCID of 0.1 m/s.⁴⁴ Studies that utilized target HRR to progress intensity reported a target HRR that ranged from 50–70% (moderate intensity) during training.^32,44 It is also important to note that both single session studies were categorized in this subgroup. Also, many of the studies allowed participants to walk on the treadmill at comfortable or self-paced speeds.^31,34,45

Discussion

This systematic review aimed to determine the ideal number of treadmill training sessions required to achieve a clinically significant change in gait speed for chronic stroke survivors. We found that studies that performed a moderate number of sessions (average 30.5), were the most successful in achieving the MCID of 0.1 m/s for change in gait speed. These sessions were performed within a period of 10 weeks and averaged approximately 40 minutes per session. In addition, we noted that a longer duration of training (within each session) was associated with a positive change in gait speed. In contrast, number of training sessions and number of training weeks were not associated with a greater change in gait speed. Since training duration seemed to be the primary dosage component that had a significant association with change in gait speed, we cannot exclude the possibility that change in gait speed was more influenced by training duration than the number of training sessions.

When comparing the training of responders and nonresponders, i.e. those who met the MCID for change in gait speed and those who did not, the biggest differences we observed was in the number of training sessions, length of training (total weeks), and the total target duration (in minutes) of each treadmill training session. The responders completed an average of 30.5 treadmill training sessions, while the nonresponders completed an average of 20 sessions. The studies that did not achieve the MCID for change in gait speed performed an average of 6 weeks of training, while the studies that did achieve the MCID completed an average of approximately 10 weeks of training. The additional length of treadmill training may be one of the possible reasons why these participants met clinical significance. We also noted that the responders completed nearly 7 more minutes of walking on the treadmill at each session as compared to the nonresponder group. More training sessions, longer training length (weeks) and session duration (minutes) would allow for more time for cardiovascular and neural adaptations to occur. A meta-analysis by Hannan et al. observed that the most improvements in cardiorespiratory fitness occurred in programs with a duration greater than six weeks, with the largest improvements seen in programs with a duration of 7 to 12 weeks.⁴⁸ In contrast, neural adaptations due to exercise have been shown to require a shorter timeframe as compared to cardiovascular adaptations and may be seen after only 3 to 6 weeks of training.^49,50The 2013 study by Reisman et al. speculated that the rapid changes in neural activation may be responsible for the improvements in gait biomechanics that were seen after only 4 weeks of gait training in individuals with chronic stroke.⁵¹ In this same study walking speed improved after 4 weeks, however it continued to improve with more training and participants did not achieve the MCID for walking speed until they completed 36 sessions or 12 weeks of training.⁵¹ These findings suggest that chronic stroke survivors would benefit more from longer training durations in order to achieve clinically meaningful changes in gait speed. Since the responder group received more training sessions, more weeks of training and spent more time walking during each session resulting in more walking practice overall and greater training intensity, more time for cardiovascular and neural adaptations could have resulted in a more significant change in gait speed.

Although a longer training length and session duration seem to be characteristics of treadmill training in the responder group, the element of number of sessions did not follow this trend. Based on the fair and non-significant association between number of treadmill training sessions and change in gait speed, more treadmill training sessions did not always result in a greater change in gait speed. The participants in the study by Ivey et al. completed a total of 78 treadmill training sessions, the most out of any of the other studies, and did not achieve the MCID for change in gait speed.⁴⁴ This study reported no dropouts or issues with compliance with the treadmill program, but they noted their small sample size of 19 individuals in the treadmill group as a possible explanation for not achieving the MCID, probably due to larger variability in response seen with small sample size. Observationally, most of the studies that completed 39 or more treadmill training sessions did not have a greater change in gait speed than those studies that completed between 12 and 36 sessions. Overall, it appears that a moderate number of sessions of around 30 may be more beneficial for individuals who had a stroke more than 6 months ago. These results are similar to the findings by Rose et al., who found improvements in gait speed with 36 sessions of locomotor training with body weight support or strength and balance exercises, however there were only modest improvements seen after the 24^th session.⁵² Also, as far as feasibility of implementing treadmill training interventions in research, completing a large number of treadmill training sessions (i.e. more than 40) can lead to higher dropout rates due to the amount of commitment required from participants and researchers and it may result in poorer compliance to the research protocol. For treadmill training performed in a clinic setting, therapists and patients also have to contend with insurance companies who are becoming more restrictive as far as the number of visits that are allowed for physical therapy each year.⁵³ Therefore, in addition to being more efficient and fiscally responsible, we observe that an individual with stroke can reap similar benefits from receiving 30 treadmill training sessions spread out over a longer period of time as compared to 40 or more treadmill training sessions.

Although the primary outcome of focus in our review was gait speed, there are several additional benefits of treadmill training which we did not pursue in this review. Treadmill training can help improve stroke survivors’ cardiovascular health, endurance, quality of life, and neuroplasticity. A meta-analysis by Pang et al. determined that aerobic exercise similar to treadmill training has a significant effect on peak oxygen consumption and walking endurance.⁵ Aerobic exercise such as treadmill training has also been shown to reduce insulin resistance⁵⁴ and improve both systolic and diastolic blood pressure.⁵⁵ Treadmill training can also have a positive impact on other domains such as depression and social participation.⁵⁶ Gait training on the treadmill has also been shown to induce changes in corticomotor excitability in individuals with chronic stroke and these changes have been shown to have a strong positive correlation with motor performance measures.⁵⁷

Although treadmill training can be a very effective method to promote practice of a high number of step repetitions in a controlled environment, there are many limitations to this form of training. Treadmill training may not always result in direct transfer of learned skills to overground walking due to the difference in training environments and/or walking surfaces.^58,59 Also, treadmill training provides its users with a form of passive assistance to the lower extremities from the belt⁶⁰ and additional support for the lower extremities if the user chooses to use the handrails⁴⁰, which can alter certain walking requirements such as balance and propulsion.^61,62 Although we noted that studies in this review increased overground gait speed after training, failure to meet MCID could be due to these reasons. Since treadmill training may not always result in a direct transfer of improvements to overground walking, stroke survivors may benefit more from treadmill training when a controlled training environment is desired (i.e. safety harness, speed, incline, etc.). Hence, incorporating overground walking training is also important to ensure training specificity and transference of walking skill set to similar surfaces, similar amounts of lower extremity weightbearing, and to allow the individual to get accustomed to an assistive device if applicable.

Limitations

We attempted a systematic review to explore the ‘magic’ number of sessions that can improve gait speed in people with stroke. We acknowledge that there are several limitations that could affect interpretation of our results. One limitation is the inclusion of single treadmill training session studies as results from these studies may have greatly skewed data interpretation. We decided to include these single session studies as they met our inclusion criteria which was an intention to treat and assessed the participant’s gait speed both before and after the intervention. Another limitation to this review is that studies that included adjuncts were eliminated and adjuncts such as body weight supported treadmill training, non-invasive brain stimulation, functional electrical stimulation or robotic exoskeletons are commonly utilized methods in research studies. The rationale to eliminate adjuncts was based on the large amount of variability these entail. For example, many studies individualize the amount of body weight support for each individual based on their walking ability and many protocols gradually reduce the amount of support as the individual improves. Studies that include peripheral or cortical stimulation use different stimulation parameters. Future reviews should examine whether addition of an adjunct therapy reduces the time needed to achieve a significant change in gait speed. In addition, we reviewed studies that included only chronic stroke participants. We felt that it would not be wise to review studies with acute or subacute stroke participants for the following reasons: these individuals’ functional status is rapidly evolving as they go through the prime window of recovery following stroke onset, there is significant variability in their recovery level depending on the severity of stroke, and we cannot control for other necessary clinical interventions that they will be receiving at this time. Furthermore, our findings may not be generalizable to those who cannot walk independently on a treadmill. The majority of the studies included in this review had inclusion criteria requiring that the participants had an ability to walk independently. Moreover, we cannot generalize the findings from this review to those stroke survivors with multiple comorbidities as these individuals are not typically included in research studies. The information presented in this review is based on group average data observed in the selected studies, which may not reflect every individual’s response to treadmill training as each individual is different and may have unique needs as it applies to exercise training. Also, only 6 out of 17 of the articles included were rated as good to excellent which may limit the value of the pooled results due to the lower quality of the majority of the articles included in the review, which also indicates the need for more studies in this area. A few of the included studies had very small sample sizes which reduces the power of the results obtained from those respective studies. Also, the average number of sessions calculated for each group were accompanied by wide standard deviation values which demonstrates variability in the implementation of treadmill training programs, making it difficult to make firm conclusions and recommendations from these values. Lastly, the review protocol was not registered prospectively as requirements for protocol registration on PROSPERO changed during the time of data collection. We realize this is a major shortcoming of this review, however we took many steps to reduce sources of bias by setting very clear inclusion criteria and performing careful review of the articles (titles, abstracts and full-texts).⁶³

Overall, although the responder group did achieve the MCID for change in gait speed no conclusions can be drawn regarding the number of treadmill sessions chronic stroke survivors should perform. Rationale for this include the low correlation that was not statistically significant between the number of treadmill sessions and change in gait speed, along with the limitations discussed above. Based on the trends observed in this systematic review, those individuals who have had a stroke at least 6 months ago and participate in treadmill training complete at least 30 treadmill training sessions (within a period of 10 weeks); however the data from this review does not allow us to provide firm recommendations regarding the number of sessions that should be performed. Although identifying other dosage parameters was not the primary aim of this study, based on trends seen in the group who achieved the MCID, participants may be more likely to note a greater change in gait speed if they complete the treadmill training intervention four times per week, for at least 37 minutes, for a total intervention length of at least 10 weeks, and at a moderate to vigorous intensity level. These trends are in alignment with aerobic exercise guidelines for stroke provided in the review by Kim et al. which recommends a frequency of three to five days per week, for 20 to 40 minutes at a moderate intensity.⁶⁴ The 30 training sessions performed in the responder group is in agreement with the Canadian stroke guidelines which recommend 16 to 40 outpatient therapy visits, however these number of sessions are most likely above the allowed therapy cap provided by Medicare in the United States suggesting that many therapy recipients may not be receiving sufficient dosages of therapy after stroke.^9,10 The information we provide in this systematic review is intended to be a basic guideline for clinicians and researchers as training parameters can vary significantly amongst training environments based on the clinician or researcher and/or patient or subject’s availability, the individual’s stroke severity and/or their level of conditioning and available training resources.

Summary:

What is Known:

Following stroke, treadmill training is frequently utilized as a task-specific intervention to improve walking. There is guidance on dosage parameters such as frequency and volume of treadmill training but not for duration.

What is New:

Trends showed that those chronic stroke survivors who achieved the minimal clinically important difference for change in gait speed from the start to the end of the intervention period completed at least 30 training sessions, furthermore the sessions and total training length were longer as compared to those who did not achieve it. However, no firm conclusions can be made regarding a ‘magic’ number of sessions these individuals should complete.