Background

Cardiovascular disease is the leading cause of death and disability globally [1]. Cerebrovascular diseases, including stroke and transient ischemic attack (TIA), account for 34% of cardiovascular disease in males and 37% in females [1]. This equates to approximately 15 million people worldwide having a stroke or TIA each year [1]. Due to improvements in acute stroke treatment, survival rates are improving in several parts of the world [1]. However, people who have had a stroke or TIA are at high risk (40% in 10 years) of having a recurrent stroke [2, 3]. Therefore, secondary prevention is vital.

Insufficient levels of physical activity is one of the strongest modifiable risk factors of stroke and recurrent stroke [1, 4, 5]. The World Health Organisation, the American Heart Association and the American Stroke Association recommend 150 min per week of moderate-intensity aerobic activity or 75 min per week of vigorous aerobic activity, or a combination of both, preferably spread throughout the week and preferably performed in bouts of at least 10 min duration [6,7,8]. However, recent studies have shown that the levels of physical activity performed by people with a stroke or TIA do not meet these recommendations and are low compared to the physical activity levels of healthy peers [9,10,11]. Thus, it appears that people with stroke and TIA require additional interventions to support them to improve their level of physical activity.

Several multimodal lifestyle interventions have been developed, incorporating educational, motivational and other psychosocial components with the aim to support behaviour change to reduce risk factors of recurrent stroke, including improving physical activity levels for people after stroke or TIA. Since improving physical activity is recommended in Stroke Clinical Guidelines internationally [12,13,14,15], it is important to know if these lifestyle interventions are effective in order to guide clinical practice. Three earlier similar reviews have been conducted. The first review only included trials published up to 2009 [16], and found insufficient evidence to determine the effects of lifestyle interventions on the levels of physical activity. The second review was also inconclusive [17], both recommend further high quality research [16, 17]. The most recent review [18], including trials published up to May 2015, concluded that a meta-analyses on physical activity was not possible due to diversity in the outcome measures used [18]. A best evidence synthesis including comparison of the intervention effect to controls and weighing the quality of the included trials was not conducted nor was an effect estimate of the interventions provided [18]. It remains unclear if lifestyle interventions are effective in improving the levels of physical activity performed by people with stroke or TIA. Furthermore, the need to include strategies that specifically focus on the levels of physical activity, e.g. supervised exercise, is unclear. A review specifically examining the effects of lifestyle interventions on physical activity after stroke is needed to support physiotherapists’ clinical practice. Therefore, the research question for this systematic review was: What is the effect of lifestyle interventions on the level of physical activity performed by people with stroke or TIA?

Methods

This systematic review was conducted in accordance with the guidelines of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement [19], and is registered with the International Prospective Register of Systematic Reviews (PROSPERO; CRD42018094437).

Eligibility criteria

Trials were eligible for inclusion if:

  1. 1)

    the participants were adults with clinically confirmed stroke or TIA;

  2. 2)

    the intervention was a lifestyle or behavioural intervention, defined as an intervention that incorporated educational, motivational and other psychosocial components with the aim to support behaviour change to reduce risk factors of recurrent stroke;

  3. 3)

    the study design was a randomised clinical trial (RCT) where the lifestyle intervention was compared with ‘no intervention’, ‘placebo’ and/or ‘usual care’;

  4. 4)

    at least one outcome measure of physical activity (any form of light physical activity and/or moderate to vigorous physical activity) was reported;

  5. 5)

    the full text article was available in English or Dutch.

Trials defined in the manuscript as a pilot or feasibility trial were excluded because of likely insufficient power to show effect.

Search

Three electronic databases, Pubmed, Embase and CINAHL, were searched up to August 2018. The search strategy was constructed in Pubmed and adapted for CINAHL and Embase, see supplementary Additional file 1 (‘Search Strategy’) for the search strategy. We also scanned reference lists of relevant previous reviews identified in the initial orientation search and in the systematic search, for any additional relevant citations [16,17,18].

Study selection

All trials identified in the search were first screened by title and abstract, then full-texts reviewed to determine eligibility. The study selection was independently conducted by the 2 authors (WH and LV). Disagreements were resolved by discussion. If no consensus could be reached, a third author (MFP) was consulted.

Data extraction

Data extraction included descriptive data, demographics of study populations, sample sizes, the content of the intervention and the control, duration of the intervention, outcome measures on physical activity, time points of measurement and the study results. Data were extracted by one author (WH) and checked by a second author (LV) with disagreements resolved by discussion. If no consensus could be reached a third author (MFP) was consulted.

Quality appraisal

The PEDro scale for RCTs and controlled clinical trials was used to determine the methodological quality of the included trials [20]. The PEDro scale consists of 11 ‘yes’ or ‘no’ statements with regards to domains like randomisation, blinding, attrition and reporting of results (see supplementary Table S1, PEDro scale). Points are only awarded when a criterion is clearly satisfied [20]. The highest possible score is 10 points (item 1 is not scored) [20]. Trials with a total score of 6 or higher are considered to be of high quality [21]. The quality appraisal was independently completed by 2 authors (WH and LV). The results were compared to see if there were any differences. If so, these were discussed. If no consensus could be reached a third author (MFP) was consulted.

Best evidence synthesis

A meta-analysis was the preferred synthesis method. However, due to heterogeneity of outcome measures in the different trials, this was not possible. Instead, a best evidence synthesis was conducted, based on the available results from the included trials. We used the best evidence synthesis method from the Grading of Recommendations, Assessment, Development and Evaluations (GRADE) Working Group [22,23,24,25]. This method combines the consistency of the findings with the quality of the included trials. The domains for high quality evidence are [22,23,24,25]:

  1. 1)

    At least 75% of the RCTs with no limitations of study design have consistent findings,

  2. 2)

    Direct data, (this refers generalisability, the extent to which the people, interventions and outcomes in the trials are comparable to those defined in the inclusion criteria of the review).

  3. 3)

    Precise data, (this refers to a sufficient number of participants and events and the width of the confidence intervals).

  4. 4)

    No known or suspected publication biases.

For each domain for ‘high quality evidence’, that is not met, the level of evidence is downgraded [22,23,24,25]:

  • High quality evidence: At least 75% of the RCTs with no limitations of study design have consistent findings, direct and precise data and no known or suspected publication biases;

  • Moderate quality evidence: 1 of the above domains is not met;

  • Low quality evidence: 2 of the above domains are not met;

  • Very low quality evidence: 3 of the above domains are not met.

Effect size of the intervention and subgroup analyses

To determine the effect size of the interventions, the standardized mean difference (SMD), including the 95% confidence intervals, was calculated where possible for the between group differences at follow-up [26]. A SMD of ≥0.2 was considered a small effect, ≥0.5 a moderate effect, and ≥ 0.8 a large effect of exercise therapy as stated by Cohen et al. [27]. Subgroup analyses were performed based on the content of the intervention, i.e. the inclusion of specific strategies targeting improving the level of physical activity in people with stroke or TIA.

Results

Flow of trials through the review

A total of 8245 articles were identified in the literature search. When duplicates were removed, 7986 articles remained. After screening the titles and abstracts, 35 articles progressed to full text review, of which 11 trials were included (Fig. 1, ‘PRISMA Flow diagram’).

Fig. 1
figure 1

PRISMA Flow diagram

Full size image

Characteristics of participants and trials

Characteristics of included trials are reported in Table 1, ‘Summary of included trials’. The 11 included trials reported data from n = 2403 participants (n = 1205 intervention and 1198 control). The mean age ranged from 57 to 72 years. In all trials, stroke or TIA was clinically diagnosed in a hospital [28,29,30,31,32,33,34,35,36,37,38], and most had a mild stroke or TIA [28,29,30,31, 33,34,35,36,37], and enrolled in the trials after returning home [29,30,31,32,33,34,35,36,37,38]. There was a wide range in the sample sizes, ranging from 29 to 283 per trial arm. Most trials (73%) targeted multiple risk factors without a specific focus on improving the levels of physical activity [29,30,31,32,33, 35, 37, 38]. Three trials (27%) specifically targeted improving physical activity [28, 34, 36].

Table 1 Summary of included trials
Full size table

All 11 interventions included a form of education, motivation and/or guidance to support the participants in changing their lifestyle. Regular supervised exercise was included in 2 of the trials that specifically targeted improving physical activity [28, 34], and on an ad hoc basis in the third [36]. In 3 of the included trials a physiotherapist was involved in the intervention [28, 36, 38]. In the other 7 trials the intervention was delivered by either a case manager, a general health care professional, a general practitioner, a nurse, an exercise practitioner, or it was not stated.

The type of outcome measures used to determine the level of physical activity varied. Only one trial used an objective outcome measure to measure steps and minutes spent in low, moderate and high intensity activity time per day [28]. The other 10 trials (91%) used self-reported outcome measures [29,30,31,32,33,34,35,36,37,38]. Two trials used a standardized, validated questionnaire [34, 36], and 8 trials used general non-validated questionnaires [29,30,31,32,33, 35, 37, 38].

Methodological quality

The quality assessment of the included trials is reported in Table 2, ‘PEDro scores’. Initial agreement among the 2 authors was 95% with full consensus reached through discussion. The PEDro scores ranged from 4 to 8 points (Table 2, ‘PEDro scores’). No study achieved a full score of 10 points due to lack of blinding of the participants (question 5, supplementary Table S1, PEDro scale) and the professionals responsible for the treatment (question 6, supplementary Table S1, PEDro scale), which is not possible in these types of interventions. Eight studies had a score of 6 or higher and were therefore considered to be of high quality.

Table 2 PEDro scores
Full size table

Results of individual trials

Five out of the 11 trials found significant differences in the level of physical activity in favour of the intervention [28,29,30, 32, 36]. The effect size of the intervention could be determined by calculating the SMD (see Table 3, ‘Results individual studies’) in three trials only [28,29,30], and this ranged from 0.29 to 0.98.

Table 3 Results individual studies
Full size table

As described above some of the trials specifically targeted improving physical activity levels and included either a standard or ad hoc supervised exercise component. Subgroup analyses of these 3 trials that included specific physical activity coaching and/or supervised exercise [28, 34, 36], showed that 2 trials found a significant difference in the levels of physical activity in favour of the intervention [28, 36]. For one of these trials the effect sizes of the intervention could be determined by calculating the SMD (see Table 3, ‘Results individual studies’), which were 0.73 and 0.98 [28].

Best evidence synthesis

Based on PEDRO scores, 8 trials overall were considered to be of high quality and were included in the best evidence syntheses [28, 31, 33,34,35,36,37,38]. Two of these trials (25%) found a significant difference in favour of the intervention [28, 36], and 6 found (75%) no significant between group difference [31, 33,34,35, 37, 38], therefore the domain of consistent findings (≥75%, see methods) is met. The domain of precise data (see methods) is not met because in 38% of the trials the sample size was equal or below 35 for each treatment arm. Overall, this means there is moderate-quality evidence that lifestyle interventions do not lead to significant improvements in the level of physical activity in people with stroke or TIA, compared to usual care.

A subgroup best evidence synthesis including only trials with interventions that specifically targeted physical activity shows low quality evidence that such interventions are effective to improve the level of physical activity in people with stroke or TIA, compared to usual care. This is based on three high quality trials, of which two (67%) found a significant difference in favour of the intervention [28, 36]. One trial (33%) found no significant between group difference [34], therefore the domain of consistent findings (≥75%, see methods) is not met. The domain of precise data (see methods) is not met because in 67% of the trials the sample size was equal or below n = 35 for each treatment arm.

When only general lifestyle interventions were included in a best evidence syntheses there was high quality evidence they do not lead to significant improvements in the level of physical activity in people with stroke or TIA, compared to usual care. Of the five high-quality trials included in this this analysis, all (100%) show no significant between group difference [31, 33, 35, 37, 38]. This means that the domain of consistent findings (≥75%, see methods) is met.

Discussion

This review found low-quality evidence that lifestyle interventions overall do not lead to significant improvements in the level of physical activity in people with stroke or TIA, compared to usual care, with only 2 (25%) of the 8 high-quality trials demonstrating positive findings. The results of the subgroup analyses suggest that only lifestyle interventions that include specific strategies targeting physical activity have a positive effect on the levels of physical activity. However, sample sizes were small, and in the majority of trials the levels of physical activity was a secondary outcome measure. Therefore, it is possible that some of the included trials were insufficiently powered to determine the effectiveness of the interventions on physical activity.

Counselling, advice, education, support and encouragement were commonly incorporated into the interventions, however descriptions were sparse. In those trials that included general lifestyle counselling, details about the relative emphasis on physical activity was not provided. Therefore, there is limited information to guide clinical practice regarding lifestyle counselling or physical activity coaching to improve physical activity levels of people with stroke or TIA.

There were more consistent findings of benefit for trials that included specific physical activity coaching and/or supervised exercise. The 2 high quality trials with significant positive findings included an exercise program as a standard part of their intervention or on an ad hoc basis [28, 36]. However, one high quality trial that included an exercise program found no significant between group differences [34]. This study had a sample size of 29 participants per arm (compared to 35 and 186 in the other two), so might have been underpowered [34]. This suggests that including an exercise program in the lifestyle intervention may lead to better results. In 2 of the 3 high quality trials that specifically targeted improving physical activity [28, 36], a physiotherapist was involved in the intervention and both had positive findings [28, 36]. Since a specific focus on physical activity and/or adding an exercise component to a lifestyle intervention might be beneficial, the involvement of experts in physical activity and exercise, such as physiotherapists may be a critical component for success.

The outcome measures used across the included trials were too diverse to conduct meta-analyses in this review. This corresponds to the conclusions of earlier reviews [16,17,18]. All but one study included self-reported physical activity outcome measures. Additionally, several trials measured one aspect of physical activity (e.g. taking exercise walks or participating in exercise sessions), instead of all possible types of physical activity combined. These factors may have influenced the effect estimation. Without an overall, objective measure of physical activity definitive conclusions cannot be drawn. Further high-quality research, using objective outcome measures, is needed. Our results on physical activity are in line with the recently updated Cochrane review on educational and behavioural interventions effects on physiological risk factors of recurrent stroke (e.g. blood pressure), which concluded these interventions did not lead to improvements in physiological risk factors [39].

Limitations

A meta-analysis was not possible and, though a best evidence synthesis was conducted, the limitations to sample sizes and the use of non-objective outcome measures still call for caution when interpreting the results. A systematic review on the use of different self-reported outcome measures of physical activity concluded that measurement properties were insufficiently addressed, specifically content validity [40]. Furthermore, the follow-up period was less than 2 months in 5 of the 8 high quality trials which limits the determination of sustainability of the effects.

The search strategy used was thorough and included three of the most commonly used databases. Though, it is always possible that due to the build of the search string, not including other databases and the exclusion of papers not published in English or Dutch, trials on the subject may not have been identified. We also acknowledge that since the search was conducted it is possible that additional trials have been published on the subject. Though a search in one database (Pubmed) in March 2020 did not reveal new studies.

All studies included in this review were conducted in high income countries [28,29,30,31,32,33,34,35,36,37,38]. However, the World Health Organisation concludes that the middle and low-income countries have the highest incidence and death rates for stroke [1]. Further trials are needed to determine the effectiveness of lifestyle interventions in middle and low-income countries.

Implications and recommendations for future research

Current clinical guidelines emphasise the importance of increasing physical activity levels as part of (secondary) stroke prevention [12,13,14,15]. Clinicians therefore need clear guidance on the best way to improve physical activity levels for their patients. Although a positive trend is seen for trials that include specific physical activity coaching and/or supervised exercise programs, there is currently insufficient evidence to support definitive recommendations. There is also a lack of specific detail on the content and behaviour change techniques used in these interventions which further limits implementation. In light of the fact that sustainable behaviour change has been proven very difficult both in research and clinical practice, this information is crucial [41, 42]. Recommendations for further research include better description of the content of the intervention in particular the behaviour change techniques used, more homogeneous objective outcome measures, adequate sample sizes, and longer follow-up periods [43]. Populations from middle and low income countries should also be included.

Conclusion

The results of this review demonstrate high-quality evidence that general lifestyle interventions seem insufficient to improve the levels of physical activity in people with stroke or TIA. The subgroup analyses indicate that lifestyle interventions specifically targeting the levels of physical activity might be effective. Further research is needed to determine the effectiveness of combining lifestyle interventions that include behaviour change strategies specifically focusing on improving physical activity and/or supervised exercise programs to sustainably improve physical activity after stroke.