Dyad interventions for health-related quality of life, activity, and participation after stroke: A systematic review

Abstract

Purpose:

To characterize the effects of dyad interventions on stroke survivor and caregiver health-related quality of life (HRQOL), participation, and activity outcomes.

Methods:

Five databases were searched using: stroke AND social support/dyad relationships AND intervention. No date or language restrictions were applied. Title/abstract/full-text review and data extraction were conducted by two independent raters. Studies that tested between-group or within-group effects of an intervention that engaged both dyad members together in at least one intervention session and measured at least one outcome of interest (HRQOL, participation, or activity) in both dyad members were included.

Results:

Among 64,988 records, 401 full-text articles were reviewed, and 17 studies were included. Three dyad intervention types were identified: stroke education and caregiver training, joint psychosocial interventions, and caregiver-mediated exercise. Effects on HRQOL were mixed. Among five studies that favored the intervention group on HRQOL, effects ranged from moderate to large among stroke survivors (Cohen’s d=0.51 to 7.03) and caregivers (Cohen’s d=0.66 to 7.90). Measures of caregiver activity and participation were rarely included.

Conclusions:

There is emerging evidence for certain types of dyad intervention after stroke. Future research should examine the effectiveness and mechanisms of these dyad interventions and include measures of caregiver activity and participation.

INTRODUCTION

People with stroke and their caregivers experience detrimental changes in health-related quality of life (HRQOL) [1-3], social and community participation [4], and activity engagement [5,6]. Low HRQOL, participation, and activity engagement persists up to 2 years after stroke [1,2] and is associated with mortality risk [7] and health services utilization [8]. Further, caregivers experience loss of valued activities that may contribute to poor HRQOL and mental well-being [5]. Although stroke survivor-caregiver dyads often navigate post-stroke recovery together [9,10], traditional intervention models prioritize the needs of either the stroke survivor or the caregiver. Dyad interventions can engage both dyad members to address their individual and/or shared needs [11]. These approaches may be promising to address shared needs related to health-related quality of life, participation, and activity among stroke survivor-caregiver dyads.

Dyad interventions are present in the stroke literature [12-16]. However, recent systematic reviews identified inconsistent and conflicting outcomes across a diverse set of outcome measures which often represented only one member of the dyad [12-16]. Two meta-analyses detected small pooled effects of dyad interventions on stroke survivors’ health-related quality of life [12,13], one detected null effects [15], and a third review [14] reported that some, but not all dyad interventions were effective for this outcome. Further, one review reported that some dyad interventions effectively improved caregivers’ health-related quality of life [14], while two meta-analysis detected no effects on caregivers’ health-related quality of life [13,15]. In addition, inconsistent findings were detected across stroke survivor functional, affective, and cognitive outcomes and caregiver depression and burden outcomes [12-16]. The effects of dyad interventions on participation outcomes were not described in the existing reviews. Collectively, these reviews provide a conflicting and incomplete picture of dyad interventions efficacy for stroke survivors and their caregivers.

The varied findings described above may be a result of the broad conceptualizations of dyad interventions employed in these reviews. While the two meta-analyses [12,13] selected studies based on theoretical underpinnings of the intervention content (e.g., psychoeducational and problem-solving interventions), the criteria to be considered a dyad intervention was not precisely defined. This means that although stroke survivor-support person dyads were recruited to these studies, some interventions were delivered to only one individual separately (e.g., psychosocial intervention delivered to the caregiver without involving the stroke survivor [17]). This is further evidenced when comparing the literature across these five most recent reviews of dyad interventions after stroke [12-16]. A total of 53 articles were included across all reviews; however, only 13 articles (24.5%) appeared in 2 reviews and 1 article (1.9%) appeared in 3 reviews (supplementary file 1). This indicates an important difference in key elements of study inclusion and the authors’ approach to defining “dyad” interventions. Therefore, to advance the development of dyad interventions with robust effects, it is critical to clearly define dyad intervention models that may contribute to these effects.

The fields of health promotion and psychology provide models that aid in precisely characterizing how dyads are engaged in behavioral or psychosocial interventions [11,18,19]. For example, the Scholz (2020) model of dyad behavior change interventions describes four dyad intervention models that encompass individual to joint dyad interventions. Individual interventions are those where one dyad member is the focus of intervention while the other is only passively involved. Parallel interventions are those where the intervention aims to address the same individual outcome for each individual dyad member. Cross-over interventions actively engage both dyad members in an intervention designed to address one member’s needs. Joint interventions actively engage both dyad members in an intervention designed to address a shared outcome. Studies included in prior reviews of dyad interventions in stroke included a combination of individual interventions where one dyad member was passively involved in the intervention sessions (e.g., education directed toward the caregiver with passive involvement of the stroke survivor, and caregiver outcomes prioritized [13,20]) and cross-over interventions where both dyad members were actively engaged in intervention with a focus on one member’s outcomes (e.g., stroke self-management training that included both dyad members and focused on the stroke survivors’ outcomes [12,21]). Analyses of effects across interventions that encompass diverse dyad models may contribute to the mixed findings in prior reviews. The next step toward understanding the effects of these interventions is to more precisely characterize the dyad interventions that are being employed among stroke survivor-caregiver dyads.

The purpose of this systematic review was to examine the effects of dyad interventions on stroke survivor-caregiver health-related quality of life, participation, and activity outcomes. Dyad interventions were defined as those that: (1) involved at least one therapeutic interaction that included the stroke survivor and caregiver simultaneously, and (2) measured one of the outcomes of interest for both dyad members. Health-related quality of life was pre-specified as the primary outcome of interest, with participation and activity considered secondary outcomes.

METHODS

This systematic review was registered in PROSPERO (CRD42022365685) and is reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) [22,23]. There is not a published protocol paper.

Search string

The search string was developed in collaboration with two health science librarians. Because the concept of dyad interventions is not succinctly captured within the structure of current Medical Subject Headings (MeSH), the string was designed to identify literature that reported on: stroke AND social support AND intervention. The stroke concept included terms for ischemic and hemorrhagic stroke and excluded terms for transient ischemic attack. The social support concept included search terms for non-professional social support, such as family, friends, peers, kin, and caregivers. The search string was developed, tested, and tuned in PubMed and then translated to four other databases: Embase, Web of Science, PsycInfo, Cumulated Index to Nursing and Allied Health Literature (CINAHL), and the search engine Google Scholar. The first 200 results from Google Scholar were included. Database search strings were constructed with the aid of a search string development tool [24]. Search strings for all databases are available in online supplementary file 2.

Search strategy

Studies were included if they met the following criteria: (1) the study design was a randomized or non-randomized controlled trial, (2) dyads included in the study were people with stroke and a person with whom they had a relationship prior to the study (e.g., caregiver, support person, family member, friend), (3) tested an intervention that involved one or more interactions with both dyad members, (4) measured at least one outcome of interest among both dyad members. Language and publication date criteria were not imposed. There were no exclusion criteria. The primary outcome of interest was health-related quality of life (HRQOL). HRQOL was defined as the perceived impact of health on an individual’s overall sense of well-being [25]. Measures of HRQOL include questionnaires such as the Medical Outcomes Study 36-Item Short Form Survey (SF-36) or EuroQol-5D-5L. Secondary outcomes of interest were activity and participation. Activity was defined as the ability to complete daily living activities and included measures such as the Barthel Index, Stroke Impact Scale-Mobility or Activities of Daily Living/Instrumental Activities of Daily Living subscales, the Frenchay Activities Index, or the Functional Independence Measure [26]. Participation was defined as level of engagement in the activities that an individual desires to engage in and included measures such as the Stroke Impact Scale-Participation subscale or PROMIS Social Participation Short Form [26]. No date or language restrictions were imposed in the search strategy.

The search was last run on May 9, 2024. All records identified using the search string were uploaded to Covidence (2023, Veritas Health Innovation, Melbourne, Australia) and duplicates were removed. Review team members were trained to title/abstract reviews and full-text reviews through two rounds of training for each level of review. Team members who achieved inter-rater reliability of Kappa ≥ 0.75 completed screening. Each title/abstract was then reviewed independently by two team members. Full texts were then retrieved, uploaded to Covidence, and reviewed independently by two members. Those published in a language other than English or Persian were translated using Deepl (Deepl SE, Cologne, Germany, available at https://www.deepl.com/translator). Persian translation is unavailable in Deepl. Two full-text articles that were published in Persian were read by a team member who is fluent in Persian and discussed with the first and senior author. Conflicts flagged by Covidence at each stage of the review were resolved by discussion with the first and senior author. At the full-text stage of the review, if clarification was required to determine whether the intervention met criteria, author inquiries were attempted via electronic mail. Hand searches were conducted to identify papers reporting trial results of published protocol papers, full reports of conference abstracts, the parent study results of secondary analyses, and the reference lists of review papers.

Risk of bias

The Cochrane Risk of Bias Assessment Tool (version 1.0) was used to assess the risk of bias within studies [27]. Each study was assigned a rating of high risk, low risk, or unclear risk of bias based on 7 items: sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcomes assessment, incomplete outcomes data, selective reporting, and other sources of bias. Examples of other sources of bias include problems with dose matching, unbalanced allocation, and poor intervention fidelity. The tool was applied by two independent reviewers. Results were compared and discrepancies were resolved through discussion between the two reviewers.

Data extraction and synthesis

Two reviewers extracted data from each study. Discrepancies were resolved by discussion among the two reviewers. Authors were contacted as needed to request data necessary to compute effect sizes. Study characteristics (author name, date, country, setting, study design, sample size), participant characteristics (age, sex, race/ethnicity, stroke chronicity, stroke severity), intervention descriptions (dose, content, description of dyad interaction, delivery modality, control condition), and intervention effects (name of outcome measure, group means [SD]) were extracted. Interventions were grouped by content by the first and senior author. Effect sizes were computed for studies that reported a statistically significant between-group effect (α=.05) and that provided the data required for this calculation. Cohen’s d effect sizes were considered small (d=0.2), moderate (d=0.5), and large (d=0.8) [28]. Because of the few studies with relevant outcomes and inconsistency across outcome measures, narrative synthesis was conducted [29].

RESULTS

Study selection

A total of 64,988 records were identified across databases. After duplicates were removed, 56,410 titles and abstracts were screened. The full-text from 418 studies were reviewed for inclusion. Among these, 401 were excluded, culminating in 17 studies included in the review (see Figure 1). The most common study design reasons for exclusion were: the intervention did not include a dyad interaction (n=119), the publication was a conference abstract or protocol paper (n=101), outcomes were only reported for the stroke survivor (n=63), and that there were no relevant outcomes reported (n=53). A list of full-text articles reviewed and reasons for exclusion are included in Supplementary Table S1.

Figure 1. PRISMA Flowchart.

PRISMA chart showing that 64,988 titles and abstracts were screened, 418 full-text articles were reviewed for inclusion and 17 studies were included in the final review.

A flowchart illustrating that 64,988 references were identified, 56,410 studies were screened, 418 studies were assessed for eligibility, and 17 studies were included for review.

Study characteristics

Study characteristics are described in Table 1. Included studies were published between 2003 and 2024 and were primarily conducted in Asia (China, Hong Kong, South Korea, Taiwan, Turkey; 7 studies [30-36]) and Europe (Sweden, Netherlands, United Kingdom; 5 studies [37-41]). Most studies were parallel randomized controlled trial design (82.4%; 14 studies [30-34,36,37,39-44]). Sample sizes ranged from 3 to 300 dyads, with 41.2% of studies including 50 to 99 dyads (7 studies [30,32,33,36,40,42,44]) and 29.4% of studies including 100 or more dyads (5 studies [31,34,37,41,43]). Participants were recruited from inpatient (11 studies; 64.7% [30,31,33,36-42,44]) and community-based (6 studies; 35.5% [32,34,35,43,45,46]) settings.

Table 1.

Study characteristics

Total studies: N=17	N (%)
Study location
Australia	2 (11.8)
Asia	7 (41.2)
Europe	5 (29.4)
North America	3 (17.6)
Study design
Parallel Randomized Controlled Trial	14 (82.4)
Non-Randomized Controlled Trial	1 (5.9)
Single Arm Pre-Post-Test	2 (11.8)
Study setting*
Inpatient	11 (64.7)
Community-based	6 (35.3)
Sample size (dyads)
<10	1 (5.9)
10 to 49	4 (23.5)
50 to 99	7 (41.2)
≥100	5 (29.4)

^*Study setting based on participant recruitment site

Risk of Bias

The risk of bias across studies is depicted in Figures 2 and summarized in Figure 3. High or unclear risk of bias was detected on three or more items on the Cochrane Risk of Bias Tool in nine studies [30,31,35,36,38,39,41,43,46]. The most common sources of high or unclear risk of bias were performance bias (blinding of participants and personnel; 16 studies), selection bias (allocation concealment; 11 studies), and other bias (10 studies; see table S4 in supplementary file 2).

Figure 2. Risk of bias plot.

Cochrane Risk of Bias plot showing high, unclear, or low risk of bias ratings across 7 items for each study. The 7 items were: random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective reporting, and other bias.

Risk of bias plot that uses red, green, and yellow circles to depict high, low, and unclear risk of selection, performance, detection, attrition, reporting, and other sources of bias for each of the 17 included studies.

Figure 3. Risk of bias summary.

Cochrane Risk of Bias summary showing the proportion of studies with high, unclear, or low risk of bias ratings. The most common sources of high or unclear risk of bias across studies were blinding of participants and personnel, allocation concealment, and other sources of bias

Risk of bias stacked bar plot that uses red, green, and yellow to depict the risk of selection, performance, detection, attrition, reporting, and other sources of bias across included studies. The Over 50 percent of studies had high risk of performance bias and other sources of bias.

Participant characteristics

Participant demographic characteristics are presented in Table 2. Age and sex of both dyad members were reported in 11 studies [30,33,35,36,38-41,45,46]. Three studies reported these demographic characteristics only for the person with stroke [31,37,44], and one study reported only the partner’s demographic characteristics [32]. The mean age ranged from 52 years [39] to 79 years [41] among persons with stroke, and 42 years [33] to 71 years [42] for partners. The proportion of male participants ranged from 30% [41] to 81% [43] among persons with stroke and 14% [45] to 67% [38] among partners. Race or ethnicity were reported in two studies that were conducted in the USA [43,45]. Both studies reported a majority White sample. Stroke chronicity at study enrollment was reported in ten studies and ranged from 6.5 days post-stroke [33] to 2.9 years post-stroke [46]. One study reported that 35% of their sample was less than six months post-stroke [30]. Baseline stroke severity was reported in nine studies using the Barthel Index [30,31,37], Stroke Impact Scale – Mobility [40,45], Communication [40], and Memory and Thinking [45] subscales, National Institutes of Health Stroke Scale [33,35,36], ³³ and a narrative description [38].

Table 2.

Participant characteristics and interventions

Study	Country	Setting	Stroke Survivor	Caregiver	Intervention	Control
Crossover Caregiver Training and Stroke Education
Kalra 2004	UK	Inpatient and Community	N=149 control; 151 interventiona Age  Control: 76 (70-82)  Intervention: 76 (70-80) Sex, male  Control: 74 (50%)  Intervention: 86 (57%) Chronicity: NR	N=149 control; 151 intervention Age: NR Sex, male: NR	Dose: 30-45 minute sessions, 3 to 5 sessions based on caregivers' needs plus 1 "follow-through" session at home Content: Conventional care plus 2 components: (1) instruction on common stroke-related problems, and (2) hands-on training in lifting and handling techniques, mobility and transfers, continence, assistance with ADLs and communication tailored to patients' needs Mode: In-person Dyad Interaction: Hands-on caregiver training during 3 to 5 sessions lasting 30 to 45 minutes each	Conventional care
Kim 2010	South Korea	Inpatient	N=26 control; 25 interventionb Age  Control: 61.5 (16.5)  Intervention: 59.9 (12.8) Sex, male  Control: 15 (58%)  Intervention: 18 (72%) Chronicity, n<6 months post-stroke  Control: 8  Intervention: 10	N=36 control; 39 interventionb Age  Control: 54.4 (14.9)  Intervention: 53.2 (13.8) Sex, male  Control: 9 (30%)  Intervention: 10 (34%)	Dose: One week (5 days) training in each specialty area during the patients' treatment time. Each training session lasted 20 to 30 minutes, and the total supervised training time per day ranged from 70 to 120 minutes. Content: Hands-on training of the caregivers under supervision of professionals in the following areas: (1) doctor/nursing: medical needs including blood pressure, diabetes, pressure ulcers, bladder, nutrition, positioning; (2) physical therapy: walking, stretching, strengthening exercises; (3) occupational therapy: maximize independence in ADLs; (4) speech therapy: facilitate communication Mode: In-person during inpatient treatment Dyad Interaction: Hands-on caregiver training 70 to 120 minutes per day.	5 days of caregiver education on stroke prevention and management, nursing care, ADLs, community information and referrals to support groups, inclusion in goal-setting
Shyu 2010	Taiwan	Inpatient and Community	N=86 control; 72 intervention Age  Control: 74.2 (6.5)  Intervention: 74.1 (6.5) Sex, male  Control: 45 (52%)  Intervention: 28 (39%) Chronicity: NR	N=86 control; 72 intervention Age: NR Sex, male: NR	Dose: 30 minutes, 4 to 5 visits (during hospitalization); 30-45 minutes, 1 visit (telephone consult); 30 minute, 2 visits (home visits) Content: Individual discharge needs, consultation and individualized discharge health education and referrals, detect and resolve problems or skills demonstration Mode: In-person visits, telephone consultation, home visits delivered by trained nurses Dyad Interaction: Both members of the dyad participated in the study visits together	Routine hospital discharge services
Johansson 2013	Sweden	Inpatient	N=3 cAge: 65 to 80 Sex, male: 1 (33%)  Chronicity, days: 14 to 53	N=3 cAge: 70 to 85 Sex, male: 2 (67%)	Dose: Six weekly 45-minute treatment sessions Content: Three components: (1) emotional support (active listening, positive psychology, encouragement of self-care), (2) information on aphasia, (3) conversation partner training using the SCA approach Mode: All in person, three significant other only sessions, three dyad sessions. Persons with Aphasia were receiving usual services during the study Dyad Interaction: Three dyad sessions focused on education and communication skills training	No control group
Ostwald 2014	USA	Community	N=80 home-based; 79 mailed Age  Home-based: 66.98 (9.04)  Mailed: 65.75 (9.26) Sex, male  Home-based: 55 (69%)  Mailed: 64 (81%) Chronicity: NR	N=80 home-based; 79 mailed Age Home-based: 63.61 (11.02) Mailed: 61.34 (9.77) Sex, male Home-based: 25 (31%) Mailed: 15 (19%)	Dose: Planned dose not reported. Enacted dose averaged 16 visit, 70 min each (mean total 36.7 hours per dyad) over the first 6 months post-discharge Content: Psychoeducation tailored to needs of the dyad, following 39 protocols in 7 categories: stroke recovery, stress of stroke, healthy lifestyle, special problems, therapeutic skill training, coping strategies, community networks Mode: Home visits delivered by advance practice nurses, occupational and physical therapists and mailed monthly packages Dyad Interaction: All psychoeducation sessions were held with dyads together	Monthly packages containing educational information on stroke, a small gift, and a subscription to the Stroke Smart magazine
Vluggen 2021	Netherlands	Inpatient and Community	N=99 control; 91 intervention Age  Control: 79.0 (6.5)  Intervention: 78.9 (7.0) Sex, male  Control: 45 (49%)  Intervention: 30 (30%) Chronicity: NR	N=82 control; 90 intervention Age  Control: 60.5 (13.5)  Intervention: 61.0 (13.5) Sex, male  Control: 24 (29%)  Intervention: 37 (41%)	Dose: Module 1: Intro session, two home visits, Module 2: Specific dose not stated. Module 3: 4-session course (2 hours/session). Total participation duration ranged from 2-6 months. Content: Modules consisted of: treatment team, self-management training for dyad and education on stroke for dyad. Mode: Module 1: In-person in an inpatient setting by multidisciplinary team; Module 2: In-person in home setting led by stroke care coordinator; Module 3: Face-to-face course led by neuropsychologist Dyad Interaction: Introductory goal setting meeting, training during home visits, and 2 educational sessions were conducted together	Usual care
Baykal 2022	Turkey	Community (recruited to the study during inpatient stay)	N=20 control; 20 web; 19 booklet Age  Control: 65.25 (14.61)  Web: 61.00 (16.96)  Booklet: 59.37 (17.89) Sex, male  Control: 14 (70%)  Web: 9 (45%)  Booklet: 8 (42%) Chronicity, days  Control: 7.25 (3.71)  Web: 6.65 (4.63) Booklet: 6.47 (4.19)	N=20 control; 20 web; 19 booklet Age  Control: 47.55 (11.56)  Web: 42.35 (13.10)  Booklet: 45.16 (14.89) Sex, male  Control: 4 (20%)  Web: 5 (25%)  Booklet: 4 (21%)	Dose: 45-50 minutes of tailored training, received weekly telephone calls Content: Information about stroke, stroke rehabilitation, and reintegration to normal living. The web group also had access to an online forum for caregiver interaction Mode: In person tailored training and follow-up telephone calls. Encouraged to use webpage or booklet (based on group assignment) Dyad Interaction: In person tailored training with the dyad, and both members encouraged to use the website or booklet (based on group assignment)	Usual care
Joint psychosocial interventions
Clark 2003	Australia	Inpatient and Community	N=33 control; 35 intervention Age  Control: 71.2 (8.8)  Intervention: 73.3 (8.5) Sex, male  Control: 19 (63%)  Intervention: 19 (59%) Chronicity, days  Control: 11.7 (3.8) Intervention: 10.3 (4.5)	N=NR Age  Control: 69.3 (5.2)  Intervention: 71.3 (6.7) Sex, male: NR	Dose: 60-minute visits, 3 visits at 3-weeks, 2-months, and 5-months after hospital discharge Content: 2 components: (1) Educational information about stroke and its consequences, reducing the risk of further stroke, coping, and community services and support structures, and (2) family counselling (Family Systems Theory) Mode: Information packet, in-person home visits with social worker Dyad Interaction: 3, 60-minute sessions of family counselling	No intervention
Lamas 2016	Sweden	Community (recruited to the study during inpatient stay)	N=7 control; 7 intervention Age  Control: 52 (3)  Intervention: 58 (6) Sex, male  Control: 4 (57%)  Intervention: 6 (35%) Chronicity: NR	N=14 control; 10 intervention Age  Control: 48 (16)  Intervention: 44 (14) Sex, male  Control: 11 (52%) Intervention: 10 (59%)	Dose: 60-minute family conversations. 3 conversations total at a frequency of 1 every 2-3 weeks Content: Family Health Conversations. Focused on the family structure, jointly prioritize problem(s) to discuss, explore and reflect on narratives shared by family members, give commendations and acknowledge suffering, and a verbal and written reflection of the experience Mode: In-person conversations at family's homes with 2 nurses Dyad Interaction: 3, 60-minute conversations that included multiple family members	No intervention
Mei 2018	China	Community	N=28 control; 25 couples; 22 spouses Age: NR Sex: NR Chronicity: NR	N=28 control; 25 couples; 22 spouses Age  Control: 69.80 (5.58)  Couples: 69.67 (2.35)  Spouses: 70.00 (2.00) Sex, male  Control: 7 (25%)  Couples: 5 (20%)  Spouses: 9 (41%)	Dose: 8 sessions 1 time per week lasting 45-60 minutes Content: 8 specialized themed weeks covering topics from stroke information, previous positive experiences, experience of caring and looking to the future. Prompts were materials like photos, songs, or written material. Mode: Delivered in the home, psychologist led, prompts were materials from the couples’ past. Dyad Interaction: Both were present during sessions if assigned to the couples group	Waitlist control
Terrill 2018	USA	Community	N =11 Age: 56.0 (18.1) Sex, male: 5 (45%) dChronicity: 2.9 (0.6 to 5) years	N=11 Age: 55.9 (16.6) Sex, male: 6 (54%)	Dose: 20-minute training, then 8 week self-administered intervention that included at least 2 activities/week alone and 2 as a couple Content: Positive psychology activities. Examples: expressing gratitude, working toward a shared goal, fostering relationships Mode: In-person training, self-administered via booklet and calendar Dyad Interaction: Training conducted together. Dyad members completed 2 positive psychology intervention activities together each week.	No control group
McCarthy 2021	USA	Community	N=8 control; 7 intervention Age  Control: 67.37 (15.70)  Intervention: 73.71 (10.08) Sex, male  Control: 3 (38%)  Intervention: 4 (57%) Chronicity, months  Control: 7.75 (2.55) Intervention: 9.43 (3.78)	N=8 control; 7 intervention Age  Control: 66.62 (15.94)  Intervention: 60.71 (23.25) Sex, male  Control: 5 (62%)  Intervention: 1 (14%)	Dose: 8 recommended sessions, but dyads could terminate sessions earlier if they felt they met their goals. Planned dose not reported. Average session length was 57 minutes (43-76) Content: Relationship topics (e.g.: maintaining a strong relationship, reaching agreement, working together) selected and prioritized based on the results of a questionnaire completed separately by each dyad member. Mode: "Tip sheets", at least 1 in-person session with social worker, remaining sessions either in-person or using Zoom based on dyad's preference Dyad Interaction: All sessions held with both dyad members together (average 57 minutes, up to 8 sessions)	Attention control: Information, Support, and Referral consisted of 2 pamphlets about stroke, up to 8 weekly sessions (minimum 1 in-person) of active listening and referrals to services as needed
Chow 2023	Hong Kong	Community	N=46 control; 54 intervention Age  Control: NR  Intervention: NR Sex, male  Control: NR  Intervention: NR Chronicity, months  Control: NR Intervention: NR	N=46 control; 54 intervention Age  Control: NR  Intervention: NR Sex, male  Control: NR  Intervention: NR	Dose: 2 hours weekly for 8 weeks Content: Narrative therapy with sessions focused on: where are you coming from, dialoguing with stroke, co-constructing the narrative, planning for the future, and celebrating the new narrative Mode: In-person group sessions Dyad Interaction: Both dyad members participated in all sessions together	Usual care
Lin 2024	Taiwan	Community	N=15 Age: 55.7 (17.0) Sex, male: 11 (73.3%) Chronicity: 4.0 (1.9) months	N=15 Age: 53.1 (18.5) Sex, male: 5 (33%)	Dose: 12 sessions delivered 2 times per week over 6 weeks. Sessions lasted 45 minutes Content: Dyad strategy training, which included shared decision-making, shared goal-setting, shared self-evaluation of performance, strategy development and implementation, and therapeutic guided discovery Mode: In-person sessions delivered at home or in a quiet space at a local hospital Dyad Interaction: Both dyad members participated in all sessions together	No control group
Tsai 2024	Taiwan	Inpatient	N=42 control; 40 intervention Age  Control: 61.79 (14.52)  Intervention: 59.35 (14.08) Sex, male  Control: 29 (69%)  Intervention: 20 (50%) Chronicity, months  Control: 3.60 (1.83) Intervention: 3.93 (1.75)	N=42 control; 40 intervention Age  Control: 49.90 (12.68)  Intervention: 49.28 (10.79) Sex, male  Control: 15 (36%)  Intervention: 14 (35%)	Dose: Two, 30 to 40-minute sessions during the first 4 weeks of hospitalization and access to the Specific Thematic Nursing Care Action Modules online during inpatient hospital stay (STNC-AM) Content: Therapeutic conversations that focused on expectations for caregivers, patients, and nurses, encouragement to share experiences, and were tailored to family needs. The STNC-AM modules provided education on stroke-related topics (e.g., emotional management, physical activity, caregiver training) Mode: In-person sessions and engagement with an online platform Dyad Interaction: Both dyad members participated in nurse counselling together	Usual care
Crossover Caregiver-Mediated Exercise
van den Berg 2016	Australia	Inpatient and Community	N=32 control; 31 intervention Age  Control: 70.1 (12.4)  Intervention: 64.5 (18.5) Sex, male  Control: 21 (66%)  Intervention: 19 (61%) Chronicity, days  Control: 13.9 (7.9) Intervention: 22.4 (13.3)	N: NR Age: NR Sex, male: NR	Dose: 8-week caregiver mediated exercise program, 8 chosen exercises, 5 times a week for 30 minutes Content: App that contained 37 standarized exercieses to improve mobility and transfers. Participants were also given a tracker Mode: Programming began in the hospital and continued post-discharge in the home with tele-rehab services and weekly home visits. Physiotherapists delivered the intervention. Participants also wore a Fitbit Zip. Dyad Interaction: Initial training and exercises were completed with both members together	Usual care
Vloothuis 2019	Netherlands	Inpatient and Community	N=34 control; 32 intervention Age  Control: 59.26 (15.01)  Intervention: 60.53 (14.82) Sex, male  Control: 20 (59%)  Intervention: 21 (66%) aChronicity, days  Control: 37 (26 to 55)  Intervention: 36 (28 to 57)	N (34 control, 32 intervention) Age  Control: 54.00 (12.26)  Intervention: 53.91 (14.90) Sex, male  Control: 13 (38%)  Intervention: 9 (28%)	Dose: 60-minute initial session, then 30-minute weekly sessions with a physical therapist for 8 weeks. Caregiver-mediated exercises 30 minutes, 5 times/week over 8 weeks Content: Physical therapists created a patient-tailored, progressive training regimen for the patient using an e-health application. Patient-caregiver couples were trained in the regimen and instructed to perform the exercises 5 times/week between sessions. Mode: 1 in-person session weekly. The physical therapist was available via telehealth as needed between the weekly sessions. Exercises were presented in an e-health application. Dyad Interaction: Patient-caregiver couples attended weekly physical therapy sessions together and completed the exercise regimen together at least 5 times/week for 30 minutes	Usual care

Note: Values reported as mean (standard deviation) unless otherwise indicated.

^aMedian (Interquartile range)

^bMedian (range)

^cRange

^dMean (range)

Intervention characteristics

Intervention dose, content, mode of delivery, and the dyad interaction are reported in Table 2. Three types of intervention content were identified: (1) crossover caregiver training and stroke education, (2) joint psychosocial intervention, and (3) crossover caregiver-mediated exercise.

Seven studies examined crossover caregiver training and stroke education interventions. These interventions primarily focused on providing information about stroke and hands-on training in skills for caregiving after stroke. Study sessions were held in the inpatient setting [30,38], the community setting [33,43], and in a combination of inpatient and community settings [31,37,41]. Both dyad members were involved in all crossover caregiver training and stroke education sessions in five studies [30,31,33,37,43], and at least one, but not all sessions in two studies [38,41].

Eight studies examined joint psychosocial interventions. These interventions focused on enhancing dyad members’ shared coping processes through counseling, positive psychology, or skills development. Study sessions were held in the inpatient setting [36], community setting [32,34,35,39,45,46] and in a combination of inpatient and community settings [42]. Both dyad members were involved in all joint psychosocial intervention study sessions in all studies [32,34-36,39,42,45,46].

Two studies examined crossover caregiver-mediated exercise interventions. These interventions provided caregiver training specifically to deliver exercise sessions to the stroke survivor. Study sessions were held in a combination of inpatient and community settings in both studies [40,41]. Both dyad members were involved in all crossover caregiver-mediated exercise study sessions in both studies [40,41].

HRQOL Outcomes

Fifteen studies measured HRQOL (Table 3). HRQOL measures included: SF-6D [39], SF-8 [30], SF-12 [36], SF-36 [31,42,43], EuroQOL Visual Analog Scale [30,37,39], EuroQOL-5D-5L [39], Stroke Impact Scale-Recovery Subscale [44,46], Carer Quality of Life Scale [40,41,44], Satisfaction with Life Scale [32], Older People’s Quality of Life Questionnaire [46], Bakas Caregiving Outcomes Scale [45], Stroke-Specific Quality of Life Questionnaire [41], and Personal Well-Being Index [34] (Supplementary Table S2). Twelve studies reported on the same measure for both dyad members [30-34,36,37,39,42,43,45,46], while two reported on different measures of HRQOL for each dyad member [41,44], and one reported on HRQOL of the caregiver but not the stroke survivor [40]. Most studies reported HRQOL separately by dyad role; however, one study reported HRQOL as one sample combined [39].

Table 3.

Health-Related Quality Of Life, Activity, and Participation Outcomes

Study	Stroke Survivor Participant			Caregiver Participant
	HRQOL	Activity	Participation	HRQOL	Activity	Participation
Crossover Caregiver Training and Stroke Education
Kalra 2004 N=300 SS N=300 CG Inpatient & Community, UK	EuroQOL VAS BL to 3M: 0.86 BL to 12M: 0.51	FAI: No inferential statistics computed BI: No effects	No measures	EuroQOL VAS BL to 3M: 0.82 BL to 12M: 0.80	No measures	No measures
Kim 2010 N = 51 SS N = 75 CG Inpatient, South Korea	EuroQOL VAS BL to 3M: Data unavailable SF-8 MCS BL to 3M: Data unavailable	Modified BI BL to 3M: Data unavailable	No measures	EuroQOL VAS BL to 3M: Data unavailable SF-8 MCS BL to 3M: Data unavailable	No measures	No measures
Shyu 2010 N = 158 SS N = 158 CG Inpatient & Community, Taiwan	No effects	No effects	No measures	SF-36 Social BL to 3M: −1.27	No measures	No measures
Johansson 2013 N = 3 SS N = 3 CG Inpatient, Sweden	No measures	No inferential statistics computed	No measures	No measures	No inferential statistics computed	No measures
Ostwald 2014 N = 159 SS N = 159 CG Community, USA	SF-36 SRHS BL to 3M: −0.65 BL to 6M: −0.83	No effects	No effects	SF-36 SRHS BL to 3M: −0.26 BL to 6M: −0.55	No measures	No measures
Vluggen 2021 N = 190 SS N = 172 CG Inpatient & Community, Netherlands	No effects	No effects	IPA: Outdoors BL to 6M: −0.47	No effects	No measures	No measures
Baykal 2022 N = 59 SS N = 59 CG Community, Turkey	SF-12 PCS: BL to 3M: 1.57 (booklet) BL to 3M: 1.85 (webpage)	No measures	RNL BL to 3M: 0.68 (booklet) BL to 3M: −0.16 (webpage)	SF-12 PCS: BL to 3M: 2.75 (booklet) BL to 3M: 0.66 (webpage) SF-12 MCS: BL to 3M: 1.62 (booklet) BL to 3M: 1.53 (webpage)	No measures	No measures
Joint psychosocial interventions
Clark 2003 N = 68 SS CG N not reported Inpatient & Community, Australia	No effects	BI BL to 6M: 0.86 AAP: Domestic BL to 6M: −2.10	AAP: Social BL to 6M: 1.28	No effects	No measures	No measures
Lamas 2016 N = 14 SS N = 24 CG Community, Sweden	EuroQOL VAS BL to 6M: 1.22 SF-6D BL to 6M: 0.99 Effects include PwS and CG combined	No measures	No measures	Effects not divided by role; see participants with stroke effect	No measures	No measures
Mei 2018 N = 75 SS N = 75 CG Community, China	SWLS BL to 8W: 7.03 BL to 12W: 6.94 BL to 14W: 5.70	No measures	No measures	SWLS WL to 8W: 7.90 BL to 12W: 7.60 BL to 14W: 6.36	No measures	No measures
Terrill 2018 N = 11 SS N = 11 CG Community, USA	Single arm design; no between-group comparisons	Single arm design; no between-group comparisons	Single arm design; no between-group comparisons	Single arm design; no between-group comparisons	No measures	No measures
McCarthy 2021 N = 15 SS N = 15 CG Community, USA	Pilot study design; no inferential statistics reported	No measures	No measures	Pilot study design; no inferential statistics reported	No measures	No measures
Chow 2023 N = 100 SS N = 100 CG Community, Hong Kong	Between-group difference not reported; Used Actor-Partner Interdependence Models to model joint outcomes	No measures	No measures	Between-group difference not reported; Used Actor-Partner Interdependence Models to model joint outcomes	No measures	No measures
Lin 2024 N = 15 SS N = 15 CG Community, Taiwan	No measures	Single arm design; No between-group comparisons	Single arm design; No between-group comparisons	No measures	No measures	Single arm design; No between-group comparisons
Tsai 2024 N = 82 SS N = 82 CG Inpatient, Taiwan	No effects	No measures	No measures	No effects	No measures	No measures
Crossover caregiver-mediated exercise
van den Berg 2016 N = 63 SS CG N not reported Inpatient & Community, Australia	No effects	SIS – Memory BL to 3M: 2.34	No effects	No effects	No measures	No measures
Vloothuis 2019 N = 66 SS N = 66 CG Inpatient & Community, Netherlands	Not reported	No effects	No effects	No effects	No measures	No measures

SS = Stroke survivor participant; CG = Caregiver participant; M = Month; W = Week; BL = Baseline; USA = United States of America; UK = United Kingdom; AP: Domestic = Adelaide Activities Profile: Domestic Chores; AAP: Social = Adelaide Activities Profile: Social Activities; BI = Barthel Index; FAI = Frenchay Activities Index; IPA = Impact on Participation and Autonomy; MCS = Mental Component Scale; PCS = Physical Component Score; VAS = Visual Analog Scale; RNL = Reintegration to Normal Living Index; SIS = Stroke Impact Scale; SRHS = Self-Reported Health Status; SWLS = Satisfaction with Life Scale

Among the six crossover caregiver training and stroke education studies, three reported significant effects favoring the intervention group [33,37], two reported significant effects favoring the control group [31,43], and one study reported no significant effects for either dyad member [41]. Among studies that favored the intervention group, effect sizes ranged from moderate to large among stroke survivors (Cohen’s d=0.51 to 1.85) and caregivers (Cohen’s d=0.66 to 2.75) [33,37]. One study did not provide sufficient data to compute effect sizes [30].

Among the seven studies examining joint psychosocial interventions, two reported significant effects favoring the intervention group [32,39], two reported no significant effects for either dyad member [36,42], and two did not compute inferential statistics due to the intervention pilot study design [45,46]. One study did not report between-group differences in change in HRQOL [34]. Rather, this study used actor-partner interdependence models to examine the joint effects of HRQOL on partners’ depressive symptoms [34]. Among the studies that favored the intervention group, effect sizes were large among stroke survivors (Cohen’s d=5.70 to 7.03 [32]), caregiver (Cohen’s d=6.36 to 7.90 [32]) and in a combined stroke survivor-caregiver sample (Cohen’s d=0.99 to 1.22 [39]).

Among the two studies examining crossover caregiver-mediated exercise interventions, no significant effects were detected on the caregiver’s health-related quality of life [40,44]. Only one of these studies measured stroke survivor’s HRQOL and did not detect significant effects [44].

Activity Outcomes

Eleven studies measured activity (Table 3). Activity measures included the Frenchay Activities Index [37,41], Barthel Index (or modified Barthel Index) [30,31,37,42,44], Skattning av Samtalsförmåga (SaS; Conversation Skills) [38], Stroke Impact Scale – Physical Composite Score [40,43], Stroke Impact Scale – Memory [40,43,44], Stroke Impact Scale – Communication [40,43,44,46], Stroke Impact Scale – Daily Activities [40,44,46], Stroke Impact Scale – Mobility [40,44,46], Stroke Impact Scale – Hand [40,44,46], Adelaide Activities Profile (Domestic Chores and Household Maintenance) [42], and Activity Measure for Postacute Care (Basic Mobility, Daily Activity, and Applied Cognition [35] (Supplementary Table S3). Nine studies reported one or more activity measure for the stroke survivor but not the caregiver [30,31,35,40-44,46], and two studies reported on the same activity measure for both dyad members [37,38].

Among the six crossover caregiver training and stroke education studies, one reported significant effects on stroke survivors’ scores on the modified Barthel Index [30] but did not provide sufficient data to compute effect sizes. Four studies reported no significant effects on stroke survivors’ activity measures [31,37,41,43]. Two studies measured activity with the same tool for both the stroke survivor and caregiver; however, neither study computed inferential statistics for these measures [37,38].

Participation Outcomes

Eight studies measured participation (Table 3). Participation measures included: Stroke Impact Scale – Participation [40,43,44,46], Impact on Participation and Autonomy (Outdoors and Social Life) [41], Reintegration to Normal Living Index [33], Adelaide Activities Profile (Service to Others and Social Activities) [42], and Participation Measure-3D4D (Social, Community, and Difficulty Frequency and Difficulty) [35] (Supplementary Table S4). Seven studies reported one or more participation measures for the stroke survivor [33,40-44,46]. Only one study reported participation outcomes for both the stroke survivor and the caregiver [35].

Among the three crossover caregiver training and stroke education studies, one reported significant effects favoring one of the intervention groups (booklet group, Cohen’s d=0.68) [33], two reported significant effects favoring the control group (webpage group versus control group in Baykal, 2022) [33,41] and one reported no significant effects [43]. Among the three joint psychosocial interventions, one reported significant effects favoring the control group (Cohen’s d=1.28) [42] and two did not compute inferential statistics due to the study design [35,46]. Both crossover caregiver-mediated interventions reported no significant effects on participation [40,44].

DISCUSSION

This systematic review identified 17 studies that examined the effects of a dyad intervention on HRQOL, activity, and participation outcomes among stroke survivor-caregiver dyads. Using a precise definition to identify dyad interventions, three types of dyad interventions were identified: crossover caregiver training and stroke education programs, joint psychosocial interventions, and crossover caregiver-mediated exercise interventions. These findings enhance the clarity of the existing body of literature on post-stroke dyad interventions in several ways. First, the observed effects across the types of dyad interventions have implications for future dyad intervention development. Second, the inconsistent measurement of activity and participation outcomes demonstrates a need for outcomes measurement that aligns with dyad intervention models. Third, consideration of the study samples supports a need to further develop dyad interventions tailored to populations not represented in this body of literature (i.e., low- and middle-income countries, members of historically excluded communities). Collectively, these findings provide a clearer direction for the advancement of post-stroke dyad interventions.

Considering the effects across the three intervention types in relation to the Scholz (2020) model of dyad behavior change provides insight into potential mechanisms of dyad interventions. Dyad models of stroke education and caregiver training had mixed effects on stroke survivors’ and caregivers’ HRQOL, and on stroke survivors’ activity and participation outcomes. This is consistent with a prior review that focused on post-stroke joint psychoeducation interventions [12]. While three of the six studies in the present review favored the intervention group for both dyad members on HRQOL [30,33,37], it is notable that some had null effects or favored the control group for both dyad members on HRQOL [31,41,43] and on the stroke survivors’ participation [33,41,43]. The interventions classified as stroke education and caregiver training were complex. They involved multiple training modalities (e.g., education sessions, written materials, hands-on demonstration) and some were delivered across multiple settings. They map onto the Scholz model as cross-over interventions, suggesting that the content is directed at one dyad member to address the other dyad member’s needs [11]. Stroke education and caregiver training programs may prioritize the stroke survivor’s health-related quality of life, unintentionally heightening the demands on the caregiver. This aligns with evidence that caregiver burden is associated with caregiver health-related quality of life [47,48]. Stroke survivors who perceive improved social support through caregiver involvement in stroke education and training may also experience improved health-related quality of life [49,50]. Further exploring the differences between crossover caregiver training and stroke education programs that favored the intervention group [30,33,37] versus those that had null effects [41] or favored the control group [31,43] may provide insight into education content and training delivery modalities that are likely to improve caregivers’ and stroke survivors’ health-related quality of life [51].

Two studies examined crossover caregiver-mediated exercise interventions [40,44]. Training caregivers and stroke survivors to complete physical therapy exercises together also maps to the Schulz (2020) model of dyadic behavior change interventions as a cross-over intervention designed to meet the needs of the stroke survivor. Although published literature supporting the plausibility of improving stroke survivors’ health-related quality of life through improving physical function [52,53], both studies reported null effects on stroke survivors’ physical function and health-related quality of life [40,44]. Moreover, caregivers’ HRQOL was only measured in one study and demonstrated null effects [40]. Like caregiver training interventions, crossover caregiver-mediated exercise adds responsibilities to the caregiving role. These cross-over dyad intervention models appear to be common in stroke rehabilitation and stroke caregiving intervention literature. Further exploring the impact of these interventions on both dyad members is an important next step.

Psychosocial interventions that leveraged interpersonal processes within dyads were promising to improve both the stroke survivors’ and caregivers’ health-related quality of life. This category included a variety of approaches: family counseling [34,36,39,42,45], reminiscence therapy [32], strategy training [35] and positive psychology [46]. The interventions included in this category map onto the Scholz model of dyad behavior change interventions as joint interventions that are directed at addressing the dyads’ needs as a unit. Two of the four studies that were designed to examine between-group intervention effects detected large effects on both dyad members’ health-related quality of life [32,39]. The third study detected large effects on stroke survivors’ social participation [42]. These findings align with literature suggesting that couples leverage communal coping processes to navigate unexpected medical events or chronic illness [9,54]. The fourth study, which was the only intervention in this category delivered solely in an inpatient setting, did not detect effects on either dyad members’ health-related quality of life [36]. Three of the eight studies included in this category were designed as pilot studies [35,45,46] suggesting that joint dyad interventions are an emerging area of research in the stroke literature. Future studies that build on these models are warranted.

Critical consideration of outcomes assessment is required to build on the existing post-stroke dyad interventions. Most studies included measures of health-related quality of life for both dyad members. Apart from three studies [35,37,38], activity and participation outcomes were only measured for stroke survivors. Indeed, stroke survivors experience changes in activity engagement and participation that stroke survivors experience as a direct result of their stroke [4,6,55]. However, caregivers also experience changes in activity engagement and participation as they take on new responsibilities [4,5]. Outcomes selection must align with the dyad intervention model employed [11]. Measuring caregiver activity and participation in cross-over interventions may play an important role in determining how training and education or psychosocial intervention also affects caregivers’ daily life. An underlying assumption of joint dyad interventions is that the outcomes mutually benefit both dyad members. Our results support this assumption by demonstrating that joint psychosocial interventions improved HRQOL for both dyad members in some of the studies that examined this intervention model using a randomized controlled trial. However, it is notable that these effects were analyzed at the individual level. Only one study analyzed outcomes at the dyad-level using actor-partner interdependence models [34]. Dyad-level analyses require that the same measurements are available for both dyad-members. Only one of the joint psychosocial interventions measured caregivers’ activity or participation [35]. The limited measurement of caregivers’ activity and participation may be attributed to the use of stroke-specific measurement tools included in the studies (e.g., Stroke Impact Scale subscales as measures of activity and participation). Further advancement of these dyad approaches should include the use of measurement tools that are diagnosis-agnostic among stroke survivors and caregivers (e.g., PROMIS measures) and application of analyses at the dyad-level using statistical methods that can account for joint intervention effects.

An additional area for future development is the equitable inclusion of stroke survivors and caregivers from low- and middle-income countries, countries with collectivistic culture, and people who are part of historically excluded communities. The majority of studies included in this review were conducted in high income countries, with only two of the 14 conducted in upper middle-income countries as defined by the World Bank [32,33,56]. Because of limitations to post-acute rehabilitation and access to formal caregivers in low- and middle-income countries, family caregivers often take on much responsibility with limited training [57]. In countries with a collectivistic culture, where families hold assumptions of filial responsibilities and uphold values of familism, family members often undertake the responsibility of assisting the stroke survivor with their daily activities [58]. Dyad intervention models may be a useful model to synergistically address the needs of stroke survivors and caregivers. Reporting of race or ethnicity across studies was rare. This reflects the global nature of this body of literature. Most studies were conducted in countries considered to have broadly homogenous populations, where reporting of race and ethnicity is less common. When race was reported, the sample was predominantly White. Further developing dyad interventions in partnership with communities that are not represented in this body of literature (e.g., low- and middle-income countries, collectivist society, Indigenous) is vital to support long-term effectiveness and implementation.

Limitations

This systematic review takes a critical step toward advancing the science of post-stroke dyad interventions. However, there are limitations to these review methods and within the body of literature that was identified. The concept of “dyad” is not indexed in most library databases. The search string was designed to capture all possible interpersonal relationships that may have been tagged in dyad intervention studies. However, there may be search terms that were not captured. Additionally, statistical synthesis was not conducted due to the heterogeneity of interventions (including content, dose, and delivery modality), study designs (e.g., pilot studies), and outcomes measures. Narrative synthesis was used to provide a summary of promising interventions that accounted for those studies in the earliest phase of intervention development. Further, fully elucidating the effects of dyad interventions on dyad members’ activity and participation was limited by the lack of measurement of these constructs among caregivers. Among studies included in this review, other sources of bias (most commonly an unmatched control group dose), blinding of study personnel, and randomization procedures were the most common sources of bias that reduce the quality of the evidence and impact the interpretation of individual study findings. These methodological limitations may also contribute to mixed results reported across studies.

CONCLUSIONS

Dyad interventions are promising for restoring health-related quality of life, activity, and participation among stroke survivor-caregiver dyads. These findings have research, practice, and policy implications. Research that advances the development of joint and crossover approaches to promoting health-related quality of life, activity, and participation after stroke is warranted. Precisely employing dyad intervention models and aligning outcomes measures with the selected model is important to further disentangle effective versus ineffective delivery approaches. Practitioners who deliver caregiver training interventions should be mindful of the potential effects on both dyad members. At the policy level, there is need to advocate for models of healthcare delivery that enable the implementation of joint intervention models that expand beyond caregiver training. Finally, research and policy that focuses on equitable design and implementation of dyad approaches within historically excluded communities (e.g., LMIC, collectivist society, Indigenous) is needed to advance the science and practice of post-stroke dyad interventions.

Implications for Rehabilitation.

• Dyad interventions that engage people with stroke and their caregivers in psychosocial intervention, stroke education, and caregiver training can impact both individuals’ health-related quality of life.

• The impact of post-stroke dyad interventions on caregivers’ participation and activity should be measured consistently with validated tools.

• Rehabilitation professionals who provide caregiver training after stroke should remain mindful of the potential for both positive and negative effects on caregivers’ and stroke survivors’ health-related quality of life.