Abstract
Objective
To investigate the effectiveness of rehabilitation interventions on increasing leisure participation in adult survivors of stroke.
Data Sources
Databases (Medline, EMBASE, CINAHL, and Cochrane Central Register of Controlled Trials) were searched from inception to September 2025.
Study Selection and Data Collection
Included studies were randomized and nonrandomized controlled trials investigating leisure rehabilitation for community-living adults with stroke, which include a measure of leisure participation. Data extraction was performed using a predetermined data extraction form. The PEDro Scale was used to assess the risk of bias.
Findings
A total of 6,813 records were retrieved, 46 were assessed at full-text review, and 8 studies were included. The meta-analysis included four studies, and another four studies were synthesized narratively. Leisure participation was higher in the intervention groups at the end of the intervention compared with the control (standardized mean difference = 0.80, 95% confidence interval 0.03 to 1.56). There was low certainty for this evidence due to the small number of studies and high heterogeneity. Subgroup analysis indicated possible benefits when intervention is provided earlier poststroke, and that leisure rehabilitation increases the number of leisure activities, but not the regularity of participation. It remains unclear whether any one intervention approach is more effective.
Conclusions and Relevance
Leisure rehabilitation may improve leisure participation; however, the quality of evidence is low. Further well-designed trials are needed to address leisure participation following stroke, investigating the optimal timing of leisure rehabilitation and identifying the effective elements of the intervention.
Plain-Language Summary
Survivors of stroke participate in fewer leisure activities, which can reduce their quality of life. This study reviewed the research on rehabilitation interventions designed to support survivors of stroke to engage in leisure activities. This review found that leisure rehabilitation may increase the number of leisure activities people do, but they don’t always make participation more regular. It was unclear whether one type of leisure rehabilitation works better than another.
Keywords
Introduction
Stroke is a leading cause of disability worldwide (Stark et al., 2021). This disability affects activity participation, including leisure (Mayo et al., 2002). After a stroke, leisure participation typically declines and shifts toward sedentary, home-based activities (Northcott et al., 2016; Yi et al., 2015), and survivors identify leisure engagement as a key rehabilitation goal (Kuipers et al., 2003). Despite this, current clinical guidelines lack clear recommendations to guide leisure rehabilitation (Mountain et al., 2020; Stroke Foundation, 2024; Winstein et al., 2016).
Leisure rehabilitation refers to therapeutic interventions provided to support people to participate in leisure activities after illness or injury. Acknowledging that leisure itself can be a therapeutic intervention, this study is focused on rehabilitation interventions, where leisure participation is the outcome of interest. This is reflected in the use of the term “leisure rehabilitation.” In occupational therapy, leisure is considered important for health and well-being (Hunt & McKay, 2015). Rehabilitation may seek to remediate the underlying impairments limiting leisure participation, teach leisure-related skills or provide opportunity for practice (i.e., performance; Boop et al., 2020). The Leisure Ability Model describes these functions as three “categories of service”: functional interventions, leisure education, and recreation participation (Stumbo & Peterson, 1998). Functional interventions aim to remediate impairments limiting activity participation. Leisure education supports the development of leisure-related knowledge and skills. Recreation participation provides opportunities to practice leisure skills in the context of the actual activity.
To date, research outcomes have focused on amount; specifically, how often a person does leisure activities (Drummond & Walker, 1995), the number of leisure activities they do (Jongbloed & Morgan, 1991; Parker et al., 2001), or time spent on leisure (Desrosiers et al., 2007). Previous reviews have reported that targeted occupational therapy can increase leisure participation after stroke (Dorstyn et al., 2014; Graven et al., 2011). However, their broad scope limits understanding of the benefit of occupational therapy interventions. Although these reviews grouped studies that reported leisure outcomes, they did not distinguish between the constructs being measured; results were summarized across constructs such as regularity of leisure, hours of participation, number of activities, and satisfaction. The reviews lacked a meta-analysis of data, reporting only a narrative synthesis (Dorstyn et al., 2014) or a “best evidence” approach (Graven et al., 2011).
For clinical implementation, several questions remain: (1) which leisure participation constructs (regularity, amount, satisfaction, performance) are improved by leisure rehabilitation, (2) does time poststroke affect intervention effectiveness, and (3) which components are delivered in effective leisure rehabilitation? Therefore, this review appraises all relevant controlled trials of rehabilitation interventions targeting leisure participation after stroke, with findings intended to inform clinical practice guidelines and set research priorities. Specifically, we ask: what is the effectiveness of leisure rehabilitation in increasing leisure participation in adult survivors of stroke, considering time poststroke and intervention context?
Methods
Registration
Developed according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement (Page et al., 2021), the review was prospectively registered in the International Prospective Register of Systematic Reviews (PROSPERO; ID: CRD42024547133; May 15, 2024).
Eligibility Criteria
Population
Adults (18 years or older) diagnosed with stroke, and living in the community. Studies involving mixed age groups were included if data from participants aged ≥18 years could be extracted, or where ≥75% of participants were aged ≥18 years. Studies with mixed diagnostic populations were included where data of participants with stroke could be isolated, or where ≥75% of participants had a diagnosis of stroke. This 75% cut-off was set a priori (McKenzie et al., 2024), and was decided in consideration that community-based rehabilitation can be provided by generalist services and therefore relevant studies may include a mixed cohort.
Intervention
Rehabilitation interventions delivered with the aim of improving leisure participation and provided by rehabilitation professionals, including but not limited to occupational therapists, leisure therapists, physiotherapists, speech and language therapists, nurses, social workers, or allied health assistants. Consistent with the interventions captured within the Leisure Ability Model (Stumbo & Peterson, 1998), leisure rehabilitation may include functional intervention, leisure education, and/or recreation participation. Interventions could be delivered in either an individual or group format, or a mix of both.
Comparators
Active comparisons such as usual care or a nonleisure intervention, passive control arms, waitlist controls, or A- versus B-type interventions where at least one arm addressed leisure participation, were included.
Outcome
All studies must include a measure of leisure participation, as assessed by any outcome measurement tool. As the importance of improving leisure participation relates to its impact on quality of life and mental well-being (Caldwell, 2005), quality of life was included as a secondary outcome. Adverse events and resource use outcomes (e.g., economic evaluations) can inform feasibility and implementation for clinical practice (Hwang et al., 2020), and were therefore extracted where available.
Study Design
Full publications of pretest/posttest randomized or nonrandomized controlled trials were included.
Search Strategy
A systematic search was conducted in Medline, Embase, CINAHL, and Cochrane Central Register of Controlled Trials from inception to September 2025 (Supplemental Appendix 1). One author independently screened the retrieved records for eligibility based on title and abstract. All excluded studies were reviewed by a senior researcher (LJ) to ensure no studies were inadvertently excluded. Full-text publications of potentially eligible studies were reviewed by two authors independently (SA and NAL) to determine eligibility. Disagreements were resolved by a third author (SK). Reference lists of included studies and relevant systematic reviews were manually screened for potentially eligible studies.
Data Extraction
Data extraction was performed by one author (SA) and verified by a senior author (LJ) using a predetermined form including participant characteristics, study methods, sample size, country of study, intervention details (extracted against TIDieR checklist items; Hoffmann & Walker, 2015), outcomes, and results. Means and standard deviations at the end of the intervention were extracted for the control and intervention groups. Where study data were missing, the corresponding authors were contacted to request the data.
Statistical Analysis/Data Synthesis
For continuous outcomes, the between-group treatment effect at the end of the intervention was reported as a mean difference and 95% confidence interval (CI), where studies used the same outcome scales. Standardized mean difference (SMD) was reported where studies used different scales to assess the same outcome construct. All meta-analyses were performed using RevMan (v 8.14) software using a random-effects model. Statistical heterogeneity was assessed using a chi2 analysis and the I2 statistic, where p < .05 and I2 > 80% were considered indicative of significant heterogeneity. Where there was significant heterogeneity, sensitivity analyses were completed considering clinical, statistical, and methodological causes for heterogeneity.
Risk of Bias
A risk of bias assessment using the PEDro Scale (Maher et al., 2003) was performed for all included studies by one author (SA) and verified by a second author (LJ). Disagreements in appraisal scores were resolved by consensus. A PEDro score of <4 indicated high risk of bias/poor study quality (Cashin & McAuley, 2020). Studies excluded from the meta-analysis due to risk of bias are reported descriptively.
Definitions and Constructs
Leisure participation outcome constructs included in this study were: (1) amount; (2) satisfaction or confidence; and (3) performance. Amount can be measured as how often a person participates in a leisure activity (i.e., regularity), the amount of time spent participating in leisure, or the number of leisure activities a person does. Satisfaction is the fulfillment of needs or expectations gained through leisure engagement (Beard & Ragheb, 1980). Measures of confidence aim to quantify a person's self-belief in their capability for a given activity or goal achievement (Poulsen et al., 2014). Performance measures the person's ability to independently, safely and/or proficiently carry out the required actions or tasks of an activity (Holm et al., 2017). Where able, each of these constructs was analyzed separately. The content of leisure rehabilitation described in the included studies was categorized using the Leisure Ability Model (Stumbo & Peterson, 1998); interventions could include one or more of the categories of service.
Subgroup Analysis
A subgroup analysis for time since stroke was preplanned, categorized as interventions provided at 0–3 months poststroke (early-subacute) versus >3 months poststroke (subacute and chronic or any mixed time poststroke).
Results
Study Selection
The PRISMA diagram (Figure 1) shows the flow of identified publications through the search and study selection process. After title and abstract screening, 46 full-text publications were retrieved and screened, resulting in 8 eligible studies (Desrosiers et al., 2007; Drummond & Walker, 1995; Fernández-Solano et al., 2020; Jongbloed & Morgan, 1991; Logan et al., 2003, 2004; Nour et al., 2002; Parker et al., 2001). Risk of bias assessment of the included studies did not identify studies with a high risk of bias as a potential reason for heterogeneity (Table 1).
PRISMA diagram.
Risk of Bias Outcomes.
Note. Y = yes; N = no. Scoring for overall risk of bias assessment is as follows: <4 = poor; 4–5 = fair; 6–8 = good; 9–10 = excellent. Table format adapted from Maher et al. (2003).
Description of Included Studies
Of the eight included studies, seven were randomized controlled trials (Desrosiers et al., 2007; Drummond & Walker, 1995; Jongbloed & Morgan, 1991; Logan et al., 2003, 2004; Nour et al., 2002; Parker et al., 2001), and one was a nonrandomized controlled trial (Fernández-Solano et al., 2020); the total participant population was n = 650 recruited from three countries (Canada, United Kingdom, and Spain; Table 2). One study (Logan et al., 2003) was a posthoc analysis of another included study (Parker et al., 2001), so this sample is not included in the total number of participants. These two studies analyzed and reported different outcome measures and were therefore both included in the systematic review.
Evidence Table.
Note. Bold type = papers included in meta-analysis; RCT = randomized controlled trial; nRCT = nonrandomized controlled trial; OT = occupational therapist; HRQoL = health-related quality of life; ADL = activities of daily living.
van Straten, A., de Haan, R. J., Limburg, M., Schuling, J., Bossuyt, P. M., & van den Bos, G. A. M. (1997). A stroke-adapted 30-item version of the Sickness Impact Profile to Assess Quality of Life (SA-SIP30). Stroke (1970), 28(11), 2155–2161. https://doi.org/10.1161/01.STR.28.11.2155
Katz, M. M., & Lyerly, S. B. (1963). Methods for measuring adjustment and social behaviour in the community: Rationale, description, discriminative validity and scale development. Psychological Reports, 13, 503–535.
Ellis, G. D., & Witt, P. (1986). The leisure diagnostic battery: Past, present and future. Therapeutic Recreation Journal, 20, 31–47.
Population Characteristics
Fifty-six percent (n = 365) of participants were male and the mean age ranged from 53.6 to 74.0 years across studies. Seven of the included studies recruited only participants with a diagnosis of stroke, with one study (Fernández-Solano et al., 2020) recruiting a mixed cohort with 83% (n = 10) of participants diagnosed with stroke. Mean time from the onset of stroke was reported in only three studies (Desrosiers et al., 2007; Drummond & Walker, 1995; Logan et al., 2004): 27.4 days, 10.5 months, and 28.3 months; while three studies included a maximum time since stroke in their study inclusion/exclusion criteria (Jongbloed & Morgan, 1991; Logan et al., 2004; Parker et al., 2001), ranging from <6 months to <36 months since stroke.
Intervention
The duration of interventions ranged from 5 weeks to 6 months. Three studies (n = 394 participants) investigated occupational therapy interventions tailored to the individual's leisure goal (Drummond & Walker, 1995; Jongbloed & Morgan, 1991; Parker et al., 2001). These interventions addressed underlying impairments, such as practicing transfers or addressing related activities of daily living. One study (Logan et al., 2003) completed further analysis using data from one of these previously reported studies investigating different outcome measures. One study provided an outdoor mobility program provided by an occupational therapist (Logan et al., 2004). As these interventions primarily addressed underlying impairments, they were classified under “functional intervention” as per the Leisure Ability Model. Two studies (n = 76 participants) investigated a protocolized education program provided by an occupational therapist and recreation therapist (Desrosiers et al., 2007; Nour et al., 2002). One further study (Fernández-Solano et al., 2020) provided a structured group program that progressed through four thematic modules. These education programs addressing leisure skills were categorized as “leisure education.” All of the included study interventions also included “recreation participation”; none provided an intervention that only included this category of service. Supplemental Appendix 2 demonstrates how the study interventions can be described using the Leisure Ability Model. Supplemental Appendix 3 provides the results (summary statistics) of individual studies organized by intervention content.
Comparators
An active comparator intervention was used in three studies (Desrosiers et al., 2007; Jongbloed & Morgan, 1991; Nour et al., 2002): a home visit completed by an occupational therapist or recreation therapist where topics related or unrelated to leisure were discussed. One study provided a one-off occupational therapy session with education and information provision as the comparator intervention (Logan et al., 2004). The remaining three studies used usual care as the control (Drummond & Walker, 1995; Fernández-Solano et al., 2020; Parker et al., 2001).
Outcomes
The timing of outcome measurement was at the end of the intervention for all except for one study, which completed outcome assessment at 1 month following the end of the intervention (Logan et al., 2004). To measure amount of leisure, one study used the Katz Adjustment Index (Jongbloed & Morgan, 1991), three studies used the Nottingham Leisure Questionnaire (Drummond & Walker, 1995; Logan et al., 2004; Parker et al., 2001), and one study used participant diaries/activity logs (Desrosiers et al., 2007). To assess satisfaction with leisure, two studies used the Leisure Satisfaction Scale (Desrosiers et al., 2007; Nour et al., 2002), and one study used the Katz Adjustment Index (Jongbloed & Morgan, 1991). None of the studies measured confidence. To measure performance, one study used the Leisure Diagnostic Battery (Nour et al., 2002), and one study that reported performance did not describe their method (Logan et al., 2003). Supplemental Appendix 4 provides the results of individual studies organized by outcome construct.
Analysis and Synthesis of Study Results
Four studies reported the data required to include in a meta-analysis of the primary outcome, leisure participation (Desrosiers et al., 2007; Drummond & Walker, 1995; Logan et al., 2004; Parker et al., 2001). The other four studies, which met the inclusion criteria, did not provide data for inclusion in meta-analysis (Fernández-Solano et al., 2020; Jongbloed & Morgan, 1991; Logan et al., 2003; Nour et al., 2002). The authors were contacted, however two were unable to provide the requested data due to the time since the studies were completed (Jongbloed & Morgan, 1991; Logan et al., 2003), and one did not respond to our request (Fernández-Solano et al., 2020). One further study completed leisure participation outcome measures for the experimental group only, and only at the postintervention time point (Nour et al., 2002).
Leisure Participation
Analysis was performed on data from the first assessment postintervention, which was at the end of the intervention for all four studies included in the meta-analyses. When including all of these four studies in a meta-analysis of the primary outcome (leisure participation), regardless of which outcome construct they measured, the SMD was 0.80 (95% CI 0.03 to 1.56, p = .04) in favor of leisure rehabilitation, with high heterogeneity (I2 = 84%; Supplemental Appendix 6). The four studies that could not be included in the meta-analysis reported mixed results. Two studies analyzed each leisure activity individually (Jongbloed & Morgan, 1991; Logan et al., 2003) and reported no statistically significant differences between the control and intervention groups. One study (Fernández-Solano et al., 2020) completed a qualitative analysis of leisure participation, and another (Nour et al., 2002) completed outcome measures for the intervention group only, and only at the follow-up time point. These two studies reported that the intervention was effective in improving leisure participation.
Possible clinical and methodological factors that explain the statistical heterogeneity results of the meta-analysis were identified, including differences in outcome measure constructs and intervention content. Therefore, posthoc analyses were undertaken to explore emerging trends in the different constructs and to support hypothesis generation for future studies. We analyzed these constructs, grouping studies by the following characteristics: (1) regularity of leisure participation (n = 3), (2) number of leisure activities per participant (n = 2), and (3) intervention content (n = 3; Supplemental Appendix 5).
Amount
Meta-analysis for regularity of leisure participation, measured using the Nottingham Leisure Questionnaire (Drummond et al., 2001), showed that results were not statistically significant (MD 4.78, 95% CI −0.58 to 10.14, p = .08; Figure 2). One further study (n = 40 participants), which also measured regularity but was unable to be included in the meta-analysis, measured this outcome for each type of leisure activity separately using the Katz Adjustment Index (Jongbloed & Morgan, 1991). Consistent with the meta-analysis, the authors reported that no statistically significant differences were observed between the two groups for any of the 26 listed leisure activities. For number of leisure activities, results were statistically significant (SMD 1.34, 95% CI 0.35 to 2.33, p = .008) in favor of leisure rehabilitation (Figure 2). The number of leisure activities was measured using the Nottingham Leisure Questionnaire (Drummond et al., 2001) or participant diaries. One other study reported qualitative results regarding the number of leisure activities using participant-written reflections (diaries) and focus groups (Fernández-Solano et al., 2020). The authors reported that participants in the intervention group had either begun new leisure activities or were participating in leisure activities more frequently.
Secondary analyses.
Satisfaction
Only one study (n = 62 participants) measuring satisfaction with leisure participation provided change scores for both the control and intervention groups. Satisfaction was measured using the Leisure Satisfaction Scale (Beard & Ragheb, 1980). The results were statistically significant (MD 11.9, 95% CI 4.2 to 19.5, p = .0003), favoring the intervention group (Desrosiers et al., 2007). Two further studies also reported outcomes for satisfaction. Also using the Leisure Satisfaction Scale, the study (n = 13 participants) which measured outcomes only for the experimental group at the postintervention time point, concluded that the intervention achieved its purpose as the participants were satisfied with their leisure (Nour et al., 2002). In contrast to these studies, the study which measured outcomes for individual leisure activities (n = 40 participants) used the Katz Adjustment Index, and reported no statistically significant differences in satisfaction between groups for any of the 26 listed leisure activities (Jongbloed & Morgan, 1991).
Performance
Two studies measured the performance of leisure activities with differing results. The study which measured outcomes only for the experimental group at the postintervention time point, used the Leisure Diagnostic Battery and reported that the participants were autonomous with their leisure activities (Nour et al., 2002). The other study (n = 309 participants) measured performance by recording whether a participant was independent in a particular leisure activity; it was not specified which outcome measure was used to rate independence. The study analyzed the number of participants who were independent in seven listed leisure activities, comparing results for each individual activity. The authors reported there was no statistical difference between the groups for any of the listed leisure activities (Logan et al., 2003).
Time since Stroke
A subgroup analysis was performed by time since stroke: 0–3 months (n = 1; Drummond & Walker, 1995), and mixed time since stroke (n = 3; Desrosiers et al., 2007; Logan et al., 2004; Parker et al., 2001; Figure 2). A statistically significant improvement in leisure participation was found in the 0–3 months subgroup (SMD 1.39, 95% CI 0.72 to 2.05, p ≤ .0001). The effect was not statistically significant for the mixed time since stroke subgroup (SMD 0.51, 95% CI −0.21 to 1.23, p = .16), with time post-stroke possibly explaining some of the heterogeneity (I2 = 74%).
Intervention Content
Meta-analysis comparing intervention content showed a statistically significant improvement in leisure outcome for protocolized education programs (SMD 0.86, 95% CI 0.31 to 1.41, p = .002), but not for occupational therapy tailored to the individual's leisure goal (SMD 0.70, 95% CI −0.56 to 1.96, p = .28; Figure 2). The outcome measure used for the education program was the number of leisure activities, whereas the tailored occupational therapy interventions measured the regularity of leisure. Two further studies, not included in the meta-analysis, also provided one of these two intervention approaches. Consistent with the meta-analysis, the study which also investigated tailored occupational therapy by measuring the regularity of participation, reported there was no statistically significant difference between groups for any of the 26 listed leisure activities (Jongbloed & Morgan, 1991). The study which measured outcomes for the intervention group at the postintervention time point, provided a protocolized education program. As reported earlier, this study concluded that the participants who received the intervention were satisfied and autonomous with their leisure activities (Nour et al., 2002); however, this study did not report outcomes for amount, as the other studies in this subgroup analysis did.
Secondary Outcomes
Two studies (Desrosiers et al., 2007; Nour et al., 2002) measured quality of life using the Stroke-adapted Sickness Impact Profile, with one study indicating the intervention resulted in improved quality of life (Desrosiers et al., 2007) and the other indicating worsened quality of life (Nour et al., 2002). These same two studies assessed depression, again providing conflicting results. Two studies (Desrosiers et al., 2007; Parker et al., 2001) assessed general health with results not being statistically significant (SMD 0.15, 95% CI −0.06 to 0.36, p = .15).
None of the included studies described adverse events, resource use, or provided economic evaluations.
Discussion
This review offers new insights into which outcome constructs, intervention timing, and intervention types may influence intervention effectiveness and warrant further investigation. Subgroup analyses indicated a trend toward an increased number of leisure activities and greater effects when intervention commenced within the first 3 months of stroke. However, these results should be interpreted with caution. Considered alongside the narrative synthesis, uncertainty remains, given the small sample sizes and variability in study methodologies and results.
Consistent with previous systematic reviews, when all studies were pooled, meta-analysis showed a statistically significant improvement in leisure participation, favoring leisure rehabilitation. While this pooled estimate was statistically significant, it was imprecise with a wide confidence interval (CI = 0.03 to 1.56), with the lower bound lying near no effect. Considered together with results of studies that could not be included in meta-analysis, half reported that the intervention did positively influence the primary leisure outcome (Desrosiers et al., 2007; Drummond & Walker, 1995; Fernández-Solano et al., 2020; Nour et al., 2002), and half reported it did not (Jongbloed & Morgan, 1991; Logan et al., 2003, 2004; Parker et al., 2001). Previous reviews have not examined why results differ. To move beyond simply confirming uncertainty, this review conducted secondary exploratory analysis of outcome measurement construct, timing of intervention, and intervention content; each of which is discussed below with specific implications for advancing the field.
A key finding of this review is that the outcome measures used to assess leisure may be fundamentally misaligned with the goals of leisure rehabilitation. Most studies used amount of leisure as the primary endpoint, yet counting activities alone does not capture whether a person is performing leisure more independently, with greater skill, or with more enjoyment. Reliance on quantity-based measures may explain the mixed results observed across trials: interventions may be producing meaningful changes in how people engage in leisure, but these changes are not being detected by the measures used. Alternatively, measures of performance or independence align with functional assessment in stroke rehabilitation (Barnes & Good, 2012). As leisure is undertaken for enjoyment, satisfaction and quality of life may be important secondary outcomes. Notably, performance, independence, satisfaction, and quality of life were rarely measured in the included studies. We therefore propose that the field adopt a core outcome set for leisure rehabilitation trials that includes measures of performance and satisfaction alongside participation quantity, to ensure that future trials capture the full scope of intervention effects.
Our exploratory subgroup analysis provides preliminary evidence, not previously reported in reviews of this topic, that timing of intervention may influence outcomes. However, only one study enrolled participants within 3 months poststroke (Drummond & Walker, 1995), with the remaining studies including a mixed subacute and chronic cohort, making definitive conclusions premature. To determine whether time poststroke influences the effectiveness of leisure rehabilitation, future trials should prespecify poststroke time ranges in their inclusion criteria and align them with the care context. Limited formal rehabilitation services immediately after hospital discharge have been identified as a barrier to leisure participation (Harrison et al., 2022; Masterson-Algar et al., 2020), and community-based services often involve stroke survivors more than 3 months poststroke (Bonifacio et al., 2022). Aligning eligibility criteria of future trials with the setting will improve interpretability, service planning, and support decisions about when leisure rehabilitation should be provided. Specifically, we recommend that the design of future trials compare early-subacute (within 3 months) and later (beyond 3 months) intervention commencement to directly test the timing hypothesis generated by this review.
Understanding intervention content is central for translating research into practice, and the classification of interventions in this review represents a novel contribution to the literature. Two main types of leisure rehabilitation were identified in the review: occupational therapy tailored to a leisure goal (Drummond & Walker, 1995; Jongbloed & Morgan, 1991; Parker et al., 2001) and structured education programs (Desrosiers et al., 2007; Nour et al., 2002). Our exploratory meta-analysis suggested that education programs may offer greater benefit, but conclusions are limited by the small number of eligible studies, concentrated within two research groups (Desrosiers et al., 2007; Drummond & Walker, 1995; Nour et al., 2002; Parker et al., 2001). Comparisons have been further complicated by differences in outcome measures, with regularity of leisure measured in tailored therapy and number of leisure activities measured in leisure education programs. Some studies excluded from meta-analysis appeared to favor education programs, but methodological limitations weaken confidence in these findings. For example, Fernández-Solano et al. (2020) relied on qualitative data, and Nour et al. (2002) lacked a control group and preintervention measures. Overall, there is currently insufficient evidence to determine whether one approach is more effective, but the distinction between these two approaches provides a testable framework for future head-to-head comparison trials. Reporting of intervention components was limited, which made drawing conclusions difficult. To address this critical barrier to progress, we strongly recommend that future trials use a published framework such as the Leisure Ability Model (Stumbo & Peterson, 1998) to guide intervention design, explicitly map intervention components to the frameworks’ three service categories, and report against the TIDieR checklist (Hoffmann & Walker, 2015). This level of reporting is essential if the field is to build a cumulative evidence base rather than continue producing isolated trials with incomparable interventions.
This review focused specifically on leisure rehabilitation after stroke using an updated search to September 2025. We employed methods to increase rigor, including independent screening, quality assessment, and grading of evidence certainty. We acknowledge limitations: we included English-only studies with the potential omission of relevant trials, and there is inconsistency in how our field reports leisure outcomes, which complicates interpretation. Planned analyses of adverse events, resource use, and economic evaluation were not possible due to a lack of data, which in turn limited conclusions about feasibility and implementation.
Previous reviews have concluded that leisure rehabilitation may be beneficial after stroke, but have not examined sources of heterogeneity in findings. Combining exploratory meta-analysis with a narrative synthesis has allowed this review to move beyond just summarizing mixed results and to articulate a research agenda. This review advances the field by identifying specific issues that may account for the mixed results and proposes three priorities to guide the next generation of trials. First, reliance on quantitative-based outcome measures likely underestimates intervention effects and obscures benefits in performance, satisfaction, and quality of life. Future trials should adopt a core outcome set that includes measures of leisure performance, satisfaction, and quality of life alongside participation quantity, and should incorporate longer-term follow-up to assess maintenance of effects. Second, exploratory subgroup analysis suggests that earlier intervention commencement may be more effective, a hypothesis that has not been formally tested. Trials should stratify or directly compare early (within 3 months) and later (after 3 months) intervention commencement, with eligibility criteria aligned to service delivery contexts. Third, the distinction between individualized occupational therapy approaches and structured leisure education programs provides a testable typology for comparing intervention approaches. Future studies should conduct heat-to-head comparisons of these two interventions, or design and evaluate combined interventions that address all three service categories of the Leisure Ability Model. In each case, transparent reporting using the TIDieR checklist is essential for building a cumulative and comparable evidence base.
Implications for Occupational Therapy Practice
Leisure participation is consistently identified as a rehabilitation need for survivors of stroke (Northcott et al., 2016; Woodman et al., 2014). Occupational therapists are well placed to address this rehabilitation goal. The findings of this systematic review have the following implications for occupational therapy practice:
Occupational therapists should ask survivors of stroke about leisure participation goals early in their recovery, and specify aspects of leisure participation (e.g., quality, amount, frequency). Identifying and addressing leisure participation goals early in stroke recovery may be beneficial. Models such as the Leisure Ability Model can help inform the content of interventions aimed at improving leisure participation. Interventions evaluated in studies to date have included targeting stroke-related impairments that may influence leisure performance skills and providing leisure education. Consider the most relevant or appropriate measure of intervention success for the individual person. Goal-attainment scales can be used to measure whether clients have achieved their leisure rehabilitation goals. Other leisure-specific outcome measures, such as those which measure satisfaction with leisure participation, may also be appropriate.
Conclusion
While this review confirms that the evidence base for leisure rehabilitation after stroke remains limited, it goes beyond previous reviews by identifying specific methodological and conceptual issues that have impeded progress. Three priorities outlined above, adopting multidimensional outcome measures, testing the timing hypothesis, and conducting head-to-head intervention comparisons with transparent reporting, provide a clear path forward. Addressing these priorities will be essential for determining whether and how leisure rehabilitation should be integrated into routine poststroke care.
Supplemental Material
sj-docx-1-aot-10.1177_19437676261464109 - Supplemental material for The Effectiveness of Rehabilitation Interventions on Improving Leisure Participation Following Stroke: A Systematic Review
Supplemental material, sj-docx-1-aot-10.1177_19437676261464109 for The Effectiveness of Rehabilitation Interventions on Improving Leisure Participation Following Stroke: A Systematic Review by Serena Alves-Stein, Natasha A. Lannin, Sharon Kramer, Kylie Wales and Laura Jolliffe in American Journal of Occupational Therapy
Footnotes
Funding
The authors received no financial support for the research, and/or publication of this article. SA was supported by an Australian Government Research Training Program scholarship; NL receives a fellowship from Heart Foundation (#106762).
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Data Availability Statement
The data generated and analyzed during this study will include content extracted from published, peer-reviewed journal articles. Additional data generated and analyzed during the study will be available from the corresponding author upon reasonable request.
Supplemental Material
Supplemental material for this article is available online.
Author Biographies
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
