A systematic review and meta-analysis of lifestyle interventions for informal caregivers of older adults: effects on behavioral change, mental health, and physical health

Abstract

Background and Objectives

Informal caregivers of older adults often experience poorer lifestyle behaviors and elevated risks of physical and mental health problems. This systematic review and meta-analysis evaluated the effects of lifestyle interventions on behavioral, mental, and physical health outcomes among caregivers of older adults, and explored whether intervention characteristics influenced outcomes.

Research Design and Methods

Following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines (PROSPERO CRD42023439797), a systematic search was conducted using PsycINFO, MEDLINE, and EMBASE up to July 11, 2025. Randomized controlled trials were included. Pooled estimates were calculated using random-effects meta-analysis. Risk of bias was assessed using the Revised Cochrane Risk of Bias tool.

Results

21 studies (N = 2,099) were identified, with 17 included in the meta-analysis. Most interventions focused on physical activity, few addressed sleep or nutrition and none targeted smoking or alcohol use. Pooled estimates derived mainly from physical activity interventions suggested a small reduction in depressive symptoms (SMD = −0.34, 95% CI −0.47 to −0.20). Subgroup analyses suggested that depressive symptom reductions reached statistical significance only in hybrid physical activity interventions combining in-person sessions with technology-supported monitoring, but not in telephone-based or in-person formats. Most studies were rated as high risk of bias.

Discussion and Implications

Preliminary findings, largely derived from physical activity interventions, suggest a small reduction in depressive symptoms, with no consistent effects for physical activity, sleep quality, caregiving burden, stress, or self-rated health. The evidence is limited in scope and rigor. Future research may target a broader range of lifestyle behaviors.

The global population of individuals aged 60 or above has exceeded one billion (World Health Organization, 2025), with nearly two-thirds requiring assistance with daily living (World Health Organization, 2024). This demographic shift has led to a growing reliance on informal caregivers, primarily spouses, children, and close friends, who provide approximately 80% of long-term care (United Nations, 2022). While caregiving can foster emotional rewards and strengthen bonds between the care dyads (Lloyd et al., 2016), it also imposes substantial physical and emotional challenges to caregivers. On average, informal caregivers spend more than 20 hr per week performing caregiving duties, including personal hygiene assistance, mobility support, medication management, and household chores (National Alliance for Caregiving & AARP, 2020). These responsibilities are associated with stable caregiver burden over time (Ping et al., 2025), affecting caregivers’ physical and mental health (Lacey et al., 2024).

Caring for older adults is accompanied by unique challenges for caregivers. Older adults often present with age-related conditions such as dementia and multiple comorbidities, which intensify caregiver stress. Dementia caregivers, for example, report higher levels of anxiety, mood disorders, and depressive symptoms compared to caregivers of individuals with other conditions (Tremont, 2011). Older caregivers, who often age alongside their care recipients, are more likely to experience frailty, chronic illness, and reduced physical capacity, which can compromise their ability to provide care (Barbosa et al., 2021; Muñoz-Bermejo et al., 2019; Potier et al., 2018). Their risk of malnutrition increases with time (Potier et al., 2018), and those with chronic conditions report fewer perceived benefits from caregiving (Polenick et al., 2020). Middle-aged caregivers face the challenge of managing both paid employment and caregiving duties, compounding stress and time constraints (Gomez-Leon et al., 2019). These pressures can result in reduced work hours or impaired job performance, contributing to financial strain through decreased earnings (Schulz et al., 2020).

A growing body of research highlights the importance of maintaining healthy lifestyle behaviors to support both physical and mental health. Lifestyle medicine identified six modifiable behaviors prominent to health, including nutrition, physical activity, restorative sleep, stress management, positive social connections, and avoidance of risky substances (Lippman et al., 2024). These behaviors are especially relevant for informal caregivers, who often experience elevated health risks due to the demands of caregiving. Nonetheless, caregivers often prioritize the needs of their loved ones over their own, overlooking their own health (Waligora et al., 2019). Studies have shown that caregiving is associated with decreased physical activity and weight gain (Beesley et al., 2011), lower adherence to healthy eating patterns (Mochari-Greenberger & Mosca, 2012), and increased alcohol use (Rospenda et al., 2010). Dementia caregivers also report significantly lower sleep quality and shorter sleep duration compared to age-matched non-caregivers (Gao et al., 2019). These behaviors contribute to long-term health risks and diminish caregivers’ capacity to sustain their caregiving roles.

Interventions focused on social support and stress management have been extensively studied. A meta-review in 2021 identified five systematic reviews and meta-analyses evaluating supporting interventions for informal dementia caregivers, such as support groups and mobilizing informal networks (Cheng & Zhang, 2020). These studies yielded mixed evidence regarding improvements in depression, anxiety, quality of life, and perceived social support. In contrast, there is comparatively less research on behavioral interventions targeting other health-related behaviors among informal caregivers of older adults. To date, only two systematic reviews examined physical activity interventions in this population, with findings indicating mixed effects on psychological and physical health (Baik et al., 2021; Won et al., 2025). Nonetheless, these reviews did not assess intervention effect sizes or explore how delivery formats and settings might influence outcomes, despite revealing diverse intervention characteristics among included studies. Additional evidence from caregivers of adults with chronic illnesses suggests that physical activity interventions may yield small-to-medium improvements in physical and psychological well-being (Epps et al., 2021; Marshall et al., 2022). Another meta-analysis examined behavioral interventions aimed at improving sleep quality among informal dementia caregivers and reported small positive effects (Gao et al., 2019). However, these meta-analyses included non-randomized trials, which may have inflated effect estimates due to uncontrolled confounding. Moreover, few studies have evaluated behavior change outcomes, such as increased physical activity or improved sleep hygiene, which are critical indicators of intervention success (Livia et al., 2016). Existing reviews have primarily focused on physical activity and sleep, overlooking other lifestyle factors that significantly affect caregiver health, including nutrition, smoking, and alcohol use.

To address these gaps, this review is novel in providing a comprehensive synthesis across multiple lifestyle domains, rather than focusing on a single behavior. It extends prior work by including behavioral outcomes alongside mental and physical health outcomes and by examining how intervention characteristics may influence effectiveness and help explain inconsistencies in the literature. This systematic review and meta-analysis aimed to evaluate the efficacy of lifestyle interventions, including those targeting physical activity, sleep hygiene, nutrition, smoking cessation, and alcohol use, in promoting behavioral change, mental health, and physical health among caregivers of older adults. It also examined whether specific intervention characteristics influence intervention effects.

Method

This review was conducted adhering to the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines (Moher et al. 2009) and registered on PROSPERO ([anonymized for review]).

Search strategies

The PsycINFO, MEDLINE, and EMBASE databases were searched for randomized controlled trials examining lifestyle interventions targeting informal caregivers of older adults. Articles up to 11 July 2025 were included without restriction on the duration of the study. Search terms were related to the population (e.g., carer* OR caregiver* AND aged OR older OR old OR oldest OR elder*), presence of an intervention (randomi*ed OR RCT), and the focus and outcomes of the intervention (e.g., “health behavio*r” OR lifestyle OR self-care OR exercise OR “physical exercise” OR “physical activity” OR sleep OR insomnia OR nutrition OR food OR diet OR vegetable OR fruit OR meal OR eating OR alcohol OR smoke*), which were combined using the Boolean operator “AND.” The search strategy was modified for each database (Supplementary Material). Reference lists of relevant systematic reviews were manually searched to identify additional eligible studies.

Selection of studies/criteria for inclusion or exclusion

Based on the PICO framework (Higgins et al., 2024), studies were eligible if they included (a) self-identified informal caregivers (aged 18 years and older) of older adults (aged 60 years and older) (Population); (b) an intervention that promoted and/or supported changes in physical activity, sleep, nutrition, smoking, or drinking behaviors and targeted improving caregivers’ lifestyle behaviors as the main intervention objective (Intervention); (c) a comparison of the intervention to active control or usual care (Comparison); (d) reported at least one behavioral, physical health, and/or mental health outcome (Outcome); (e) a randomized controlled trial (RCT) design; and (f) publication in a peer-reviewed journal in English. Interventions delivered by any mode (face-to-face, telephone, digital) and format (groups or individuals) were included. Studies were excluded if the intervention: (a) targeted domestic helpers; (b) included any pharmacological component other than those for managing an existing mental health condition; or (c) targeted improving physical and mental health without lifestyle components, e.g., Cognitive behavioral therapy (CBT) for carer burden or depression. The scope of lifestyle intervention components was guided by lifestyle medicine, in which the interventions should aim at promoting lifestyle change, involving changes in physical activity, nutrition, sleep, smoking, and alcohol intake. Since most studies did not specify care recipients’ age as an inclusion criterion or report their age range, they were included if the care recipients had a mean age of 60 years or older or had an age-related condition, such as dementia. This was done to enhance the number of included studies related to caregivers of older adults, while ensuring that care recipients in the studies were predominantly older adults. Each title and abstract was screened by two researchers independently, and discrepancies were resolved through discussion.

Data extraction

The extracted data included information on the study characteristics (e.g., author, year, country), caregiver characteristics (e.g., age, sex), intervention details (e.g., duration, type), attrition rate, and all reported outcomes. Outcomes were grouped into three domains: (a) health behavior, (b) mental health, and (c) physical health. Data for each study were initially extracted by one researcher. A second researcher then validated the process by verifying the extracted data for every study against the original articles. Discrepancies were resolved through discussion.

For synthesis, outcomes were included in both the narrative and quantitative analyses only if they were reported by at least four studies, to ensure that findings were based on more than minimal evidence. For studies included in the meta-analysis, their reported pre- and post-intervention quantitative data were recorded (i.e., sample sizes, means, standard deviations [SDs], or standard errors) for each group and for each outcome measure. When multiple measures were reported for the same outcome within a study, the following criteria were applied: (a) total scores were prioritized over subscale scores to provide a more comprehensive assessment (e.g., overall sleep quality over sleep duration, sleep disturbance, or use of sleeping medication), and (b) measures that were conceptually comparable across studies were selected to maximize consistency in synthesis, with preference given to the measurement approach most commonly used across included studies (e.g., subjective over objective caregiving burden).

Assessment of study quality and bias

The risk of bias for each study was evaluated using the Revised Cochrane Risk of Bias tool (RoB 2) (Sterne et al., 2019). Five domains of bias were evaluated in the assessments: randomization process, deviations from intended interventions, missing outcome data, measurement of the outcome, and selection of the reported result. Each study was initially evaluated by one researcher, then a second researcher reassessed each study to validate the results. When a discrepancy in the overall risk level was identified, the specific risk level for each domain was reexamined. All discrepancies were resolved through discussion.

Data analysis and synthesis

A meta-analysis was conducted using a random-effects model to compare lifestyle intervention with control conditions (active controls or usual care). The effect sizes for each study were summarized using Cohen’s d standardized mean differences (SMDs) with corresponding 95% CIs, with 0.2, 0.5, and 0.8 representing small, medium, and large effect sizes, respectively (Cohen, 2013). The degree of heterogeneity across studies was estimated using the I² statistic, and the p-value for the χ² statistic was used to measure the presence of heterogeneity (Deeks et al., 2024). The results were summarized using forest plots. For the outcomes without sufficient data for meta-analysis, narrative synthesis methods were used to report and to integrate the narrative and meta-analysis data. Meta-regression analyses were conducted to examine possible sources of heterogeneity with study characteristics, including publication year, type of control group, and risk of bias. Publication bias was evaluated using Egger’s test (Egger et al., 1997). Analyses were performed in RStudio using the meta package (R Foundation for Statistical Computing, Vienna, Austria) (Balduzzi et al., 2019).

Results

The initial search yielded 1,511 articles, and 271 were removed as duplicates. A total of 1,173 articles were removed from title and abstract screening, and 67 were selected for full-text review. From these, 18 articles were removed because they did not target caregivers’ lifestyle as the intervention goal; 13 did not report sufficient data to examine intervention effectiveness; nine involved care recipients who were not older adults; five were conference abstracts or protocols; and two did not measure relevant outcomes. Consequently, 21 studies were included in the systematic review, and 17 studies with pre- and post-intervention data were included in the meta-analysis.

Characteristics of studies

The studies included were published between 1997 and 2024 and were conducted in the United States (n = 11), Spain (n = 3), Australia (n = 2), Turkey (n = 1), China (n = 1), Japan (n = 1), Hong Kong (n = 1), and Finland (n = 1). Sample size ranged from 17 to 495, totaling 2,099 participants (989 intervention, 993 control). Their mean age ranged from 54 to 75 years, with the majority (67% to 100%) of them being female. Fifteen studies targeted dementia caregivers, and the rest included caregivers of older adults with cancer (n = 2), stroke (n = 1), age-related diseases (n = 1), and those who needed assistance with daily activities (n = 2).

Intervention characteristics and conceptual frameworks

Across the 21 included studies, substantial heterogeneity was observed in the design, theoretical framework, and delivery format. Seventeen studies targeted a single lifestyle behavior: 15 on physical activity (Brewster et al., 2020; Castro et al., 2002; Chan et al., 2020; Connell & Janevic, 2009; Farran et al., 2016; King & Brassington, 1997; King et al., 2002; Loi et al., 2024; Madruga et al., 2021; Montero-Cuadrado et al., 2020; Winters-Stone et al., 2016), one on nutrition (Koponen et al., 2022), and one on sleep (Xu et al., 2022). Among them, three studies incorporated exercise into caregiving skills-building interventions (Brewster et al., 2020; Farran et al., 2016; Montero-Cuadrado et al., 2020). The remaining four studies employed multicomponent interventions: two combined physical activity and nutrition (Crane et al., 2021; Marsden et al., 2010), and two addressed physical activity, nutrition, and sleep simultaneously (Elliott et al., 2010; Lök & Bademli, 2017). No studies included targeted smoking or drinking cessation. Interventions targeting physical activity typically involved structured on-site exercise sessions (Brewster et al., 2020; Castro et al., 2002; Connell & Janevic, 2009; Crane et al., 2021; de Dios-Rodríguez et al., 2023; Elliott et al., 2010; Farran et al., 2016; Hirano et al., 2011; Hives et al., 2021; King & Brassington, 1997; King et al., 2002; Loi et al., 2024; Lök & Bademli, 2017) or provision of exercise guidance (Chan et al., 2020; Madruga et al., 2021; Marsden et al., 2010; Montero-Cuadrado et al., 2020; Puterman et al., 2018; Winters-Stone et al., 2016), whereas interventions for nutrition and sleep relied on guidance‑based approaches.

In terms of conceptual foundations, fewer than half of the studies (n = 8) explicitly adopted a theoretical framework for intervention development. Among those that did, social cognitive theory (Bandura, 2004) was the most commonly cited framework (n = 6), primarily to enhance physical activity level through mechanisms such as self-efficacy. Other frameworks were used less frequently, including behavioral activation (Kennedy et al., 2024) to improve sleep quality by encouraging engagement in pleasant daytime activities (Xu et al., 2022), and motivational interviewing techniques (Miller & Rollnick, 2013) to support behavioral changes, for example, through additional calls for participants who did not meet their exercise goals (Puterman et al., 2018). The extent to which theoretical frameworks were consistently implemented and evaluated varied across studies. While some studies reported intervention fidelity and assessed theory-consistent mechanisms, such as self-efficacy or engagement in pleasant activities (e.g., Connell & Janevic, 2009; Xu et al., 2022), others provided limited detail on how theoretical constructs were translated into intervention components or did not measure these mechanisms (e.g., Brewster et al., 2020; Farran et al., 2016). Such variability limits the ability to assess their contribution to intervention effectiveness.

Interventions also differed in their delivery format and intensity. Most were delivered in an individual format (n = 14), and the remaining were delivered in groups (n = 2) or dyads (n = 3). The length of the interventions ranged between 7 weeks and 12 months. Delivery modes included in-person sessions (n = 11), telephone-based contacts (n = 5), and hybrid formats (n = 5). Hybrid format refers to the combination of in-person sessions and technology assistance, including telephone calls, text messages, and wearable devices.

Considerable variability was also observed in control conditions. Nineteen studies adopted two-arm designs, with one arm assigned to the lifestyle intervention and the other to the control condition. The control groups comprised eight arms with no treatment, four with attention control, four with waitlist control, and three with caregiving skills interventions. Two studies included three arms. Loi et al. (2024) compared a physical activity program with attention control and no treatment. Brewster et al. (2020) compared a physical activity program with an attention control and caregiving skills intervention.

Risk of bias

Fifteen studies were considered to have a high risk of bias, three of some concern, and two low risks of bias. Risk of bias was concentrated in the measurement of outcomes and missing outcome data. Outcomes were largely self-reported without participant blinding, which may introduce response bias due to perceived improvement following intervention. However, self-reported measures are appropriate for subjective constructs such as psychological well-being, and excluding such studies would substantially reduce the evidence base and risk introducing selection bias. Most studies also provided insufficient information to determine whether missingness was related to true outcome values (e.g., baseline differences between completers and dropouts). This reflects a broader pattern of incomplete reporting in the field rather than an issue specific to individual studies. Excluding these studies on this basis would likewise reduce available evidence and potentially bias the review. Therefore, all studies were retained, and these limitations were considered when interpreting the findings.

Outcomes

The meta-analysis examined six outcome measures across three domains: health behavior (physical activity level and sleep quality), mental health (depressive symptoms, caregiving burden, and stress), and physical health (self-rated health). Results from meta-analyses, subgroup analyses, and meta-regression analyses are presented in Tables 1–3, respectively.

Table 1.

Effects of lifestyle interventions for carers of older adults compared with control conditions at post-assessment.

	k	SMD	CI	p	I 2	CI	PI
Physical activity	9	0.88	[−0.69; 2.44]	.232	90.8%	[84.8%; 94.4%]	[−3.78; 5.53]
− Outlier removeda	8	0.28	[−0.45; 1.01]	.396	78.7%	[58.2%; 89.1%]	[−1.60; 2.16]
Sleep quality	3	0.18	[−0.17; 0.53]	.154	6.8%	[0.0%; 90.3%]	[−0.16; 0.52]
Depression	13	−0.34	[−0.47; −0.20]	.000	0.0%	[0.0%; 56.6%]	[−0.51; −0.16]
− Trim-and-fill method (four studies added)	17	−0.25	[−0.40; −0.10]	.002	24.6%	[0.0%; 58.0%]	[−0.55; 0.05]
Burden	13	−0.25	[−0.65; 0.15]	.206	77.0%	[60.9%; 86.5%]	[−1.51; 1.02]
− Outlier removeda	12	−0.13	[−0.41; 0.15]	.317	62.4%	[29.6%; 79.9%]	[−0.90; 0.63]
Stress	4	−0.21	[−0.43; 0.01]	.056	0.0%	[0.0%; 84.7%]	[−0.56; 0.14]
Self-rated health	4	0.28	[−0.01; 0.58]	.056	26.9%	[0.0%; 72.5%]	[0.03; 0.54]

Note. k = numbers of comparisons; SMD = standardized mean difference; CI = confidence interval; PI = prediction interval.

^aRemoved Lök and Bademli (2017).

Table 2.

Subgroup analyses of lifestyle interventions for carers of older adults compared with control conditions at post-assessments.

		k	SMD	95% CI	I 2 (%)	95% CI (%)	p
Physical activity (outlier removed)
Dementia specific	Yes	4	0.71	[−0.46; 1.87]	69.5	[12.1; 89.4]	.089
	No	4	−0.19	[−1.40; 1.02]	78.6	[42.6; 92.0]
Format	Telephone	4	0.02	[−1.39; 1.43]	85.1	[63.2; 94.0]	.755
	In-person	3	0.16	[−0.02; 0.34]	0.0	[0.0; 89.6]
Theoretical framework	Social cognitive theory	4	0.02	[−1.39; 1.43]	85.1	[63.2; 94.0]	.755
	None	3	0.16	[−0.02; 0.34]	0.0	[0.0; 89.6]
Control group	Attention control	3	−0.21	[−2.70; 2.28]	88.7	[68.9; 95.9]
	No treatment	3	0.32	[−0.18; 0.79]	0.0	[0.0; 89.6]	.382
Depression
Dementia specific	Yes	10	−0.35	[−0.51; −0.18]	2.5	[0.0; 63.3]	.804
	No	3	−0.31	[−0.91; 0.29]	0.0	[0.0; 89.6]
Format	Telephone	4	−0.18	[−0.37; 0.02]	0.0	[0.0; 84.7]	.004
	In-person	4	−0.36	[−0.74; 0.02]	0.0	[0.0; 84.7]
	Hybrid	5	−0.52	[−0.75; −0.29]	0.0	[0.0; 79.2]
Intervention	Exercise + caregiving skills	4	−0.48	[−0.97; 0.01]	43.9	[0.0; 81.3]	.263
	Exercise	8	−0.29	[−0.45; −0.13]	0.0	[0.0; 67.6]
Theoretical framework	Social cognitive theory	6	−0.30	[−0.58; −0.03]	30.0	[0.0; 71.4]	.401
	None	6	−0.42	[−0.62; −0.21]	0.0	[0.0; 74.6]
Control	Attention control	5	−0.41	[−0.67; −0.16]	0.0	[0.0; 79.2]	.314
	Caregiving skills	3	−0.38	[−1.07; 0.31]	32.1	[0.0; 92.9]
	No treatment	4	−0.20	[−0.54; 0.13]	0.0	[0.0; 84.7]
Country	United States	6	−0.30	[−0.60; −0.00]	31.0	[0.0; 72.0]	.520
	Asia-Pacific	3	−0.22	[−0.44; 0.00]	0.0	[0.0; 90.0]
Burden (outlier removed)
Dementia specific	Yes	8	−0.23	[−0.51; 0.05]	50.8	[0.0; 78.0]	.239
	No	4	0.16	[−0.81; 1.12]	75.7	[33.0; 91.2]
Format	Telephone	3	−0.04	[−1.07; 0.98]	71.1	[1.8; 91.5]	.918
	In-person	5	−0.01	[−0.80; 0.79]	76.5	[42.7; 90.3]
Intervention	Exercise + caregiving skills	4	−0.30	[−0.78; 0.18]	42.2	[0.0; 80.6]	.538
	Exercise	4	−0.09	[−1.05; 0.86]	78.9	[43.6; 92.1]
Theoretical framework	Social cognitive theory	5	−0.31	[−0.62; −0.01]	23.8	[0.0; 68.9]	.401
	None	6	−0.02	[−0.60; 0.57]	70.7	[31.6; 87.4]
Control	Attention control	5	−0.23	[−0.58; 0.11]	37.6	[0.0; 76.8]	.625
	Caregiving skills	3	−0.30	[−1.19; 0.59]	60.1	[0.0; 88.6]
	No treatment	3	0.13	[−1.57; 1.82]	81.6	[42.9; 94.1]

Note. k = numbers of comparisons; SMD = standardized mean difference; CI = confidence interval; PI = prediction interval.

Table 3.

Meta-regression of lifestyle interventions for carers of older adults compared with control conditions at post-assessment.

	k	Estimate	SE	p	95% CI
Physical activity (outlier removed)
Year of publication	8	−0.06	0.03	.080	[−0.12; 0.01]
Mean age	7	0.02	0.09	.849	[−0.21; 0.24]
Percentage of women	5	0.02	0.05	.784	[−0.14; 0.17]
Intervention duration (weeks)	8	0.01	0.02	.671	[−0.05; 0.07]
Depression
Year of publication	12	0.00	0.01	.596	[−0.02; 0.01]
Mean age	11	0.03	0.01	.048	[0.00; 0.05]
Percentage of women	11	0.01	0.01	.460	[−0.01; 0.02]
Intervention duration (weeks)	13	0.00	0.00	.287	[−0.00; 0.01]
Burden (outlier removed)
Year of publication	11	0.02	0.02	.291	[−0.02; 0.06]
Mean age	10	0.03	0.02	.290	[−0.03; 0.09]
Percentage of women	10	−0.02	0.01	.175	[−0.06; 0.01]
Intervention duration (weeks)	12	−0.01	0.01	.223	[−0.03; 0.01]

Note. k = numbers of comparisons; SE = standard error; CI = confidence interval.

Physical activity level

Among the 20 lifestyle interventions involving physical activity components, ten studies assessed caregivers’ physical activity level, and eight provided sufficient pre- and post-intervention data for inclusion in meta-analysis. Meta-analysis showed no statistically significant pooled effects of these interventions on physical activity level compared with control groups (SMD = 0.88, 95% CI −0.69 to 2.44), with substantial heterogeneity (I² = 90.8%, 95% CI 84.8% to 94.4%). Sensitivity analysis indicated that the pooled estimate was strongly influenced by a single small study with an unusually large effect size (Lök & Bademli, 2017). Removal of this outlier substantially reduced the pooled effect (SMD = 0.28, 95% CI −0.45 to 1.01) and heterogeneity (I² = 78.7%, 95% CI 58.2% to 89.1%). Of the studies not included in the meta-analysis, both reported no between-group differences (de Dios-Rodríguez et al., 2023; Hirano et al., 2011).

Subgroup and meta-regression analyses, conducted with and without the outlier study, did not identify significant moderators. Subgroup analyses revealed no statistically significant subgroup differences in care recipient’s condition, delivery format, presence of social cognitive theory, or types of control conditions. Meta-regression analyses revealed no significant findings in year of publication, mean age, percentage of women, or duration of intervention. Egger’s test was not performed because there were fewer than ten studies.

One study assessed long-term effects at six months after the intervention and reported no significant between-group difference (Connell & Janevic, 2009). Two studies incorporated pedometers as objective indicators of behavioral change. de Dios-Rodríguez et al. (2023) reported no significant changes after the intervention, whereas Farran et al. (2016) did not report pre-post physical activity data. Taken together, findings across studies were inconsistent, and no clear sources of heterogeneity were identified. However, these results should be interpreted with caution given the small number of studies available for these analyses.

Sleep quality

Four studies assessed sleep quality in caregivers, of which three were included in the meta-analysis. Findings across individual studies were inconsistent. Three smaller trials (N = 31–100) reported significant improvements in sleep quality following the intervention, while the largest trial (N = 642) found no between-group difference (Elliott et al., 2010). Meta-analysis showed no statistically significant pooled effect of lifestyle interventions on sleep quality compared with control conditions (SMD = 0.18, 95% CI −0.17 to 0.53), with low heterogeneity (I² = 6.8%, 95% CI 0.0% to 90.3%). Egger’s test was not performed because fewer than ten studies were included. The pooled estimate was potentially influenced by the largest trial, which contributed approximately 80% of the total sample. The study did not specify a theoretical framework for the intervention, whereas two of the smaller studies reporting positive effects were theory‑based.

Depressive symptoms

Twelve studies assessed depressive symptoms, of which 11 were included in the meta-analysis. Ten interventions targeted solely physical activity and one solely sleep. These interventions were associated with moderate effects in reducing depressive symptoms in caregivers of older adults compared to control groups (SMD = −0.34, 95% CI −0.47 to −0.20), with low heterogeneity (I² = 0.0%, 95% CI 0.0% to 68.2%). No outlier study was identified. Egger’s test for the intercept indicated funnel plot asymmetry (p = .047), indicating evidence of publication bias. When correcting for publication bias, the trim-and-fill method showed comparable effect sizes with the main model (four studies added, SMD = −0.25, 95% CI −0.40 to −0.10). However, the prediction interval widened to include the null effect (PI −0.55 to 0.05), indicating uncertainty in the magnitude and replicability of effects in future studies. One study reported long-term effects at six months after the intervention and found no statistically significant between-group difference (Connell & Janevic, 2009).

Subgroup analyses showed a significantly larger effect size in interventions involving exercise delivered using a hybrid format (SMD −0.52, 95% CI −0.75 to −0.29) compared with telephone (SMD −0.18, 95% CI −0.37 to 0.02) or in-person (SMD −0.36, 95% CI −0.74 to 0.02). No statistically significant subgroup differences were observed in care recipient condition, use of social cognitive theory, or study location. Meta-regression analyses revealed no significant associations with year of publication, participant age, sex distribution, and duration of intervention. Although a potential moderator was identified from subgroup analysis, these results should be interpreted with caution given the small number of studies available for these analyses.

Burden

Fourteen studies assessed caregivers’ burden, of which 11 were included in the meta-analysis. Three of the 14 studies included a caregiving skills-building component, but none described how intervention components were designed to target caregivers’ burden or stress appraisal processes. Findings across individual studies were inconsistent. Several small trials reported significant reductions in burden, whereas others found no between‑group differences. Meta-analysis showed no statistically significant pooled effects of lifestyle interventions on caregiving burden compared with control groups (SMD = −0.25, 95% CI −0.65 to 0.15), with substantial heterogeneity (I² = 77.0%, 95% CI 60.9% to 86.5%). Sensitivity analysis indicated that the pooled estimate was strongly influenced by a single small study with an unusually large effect size (Lök & Bademli, 2017). Removal of this outlier substantially reduced the pooled effect (SMD = −0.13, 95% CI −0.41 to 0.15) and heterogeneity (I² = 62.4%, 95% CI 29.6% to 79.9%). Egger’s test indicated no risk of publication bias (p = .450). Of the studies not included in the meta-analysis, one reported greater reduction in caregiving burden in the intervention group than the control group (Hirano et al., 2011), whereas another found no significant differences (de Dios-Rodríguez et al., 2023). The final study did not report interaction effects, but that reduced caregiving burden was associated with self-efficacy related to physical activity participation but not physical activity exertion or adherence (King & Brassington, 1997).

Subgroup and meta-regression analyses, conducted with and without the outlier study, did not indicate significant moderators. Subgroup analyses revealed no statistically significant subgroup differences in care recipient’s condition, delivery format, and presence of social cognitive theory. Meta-regression analyses revealed no significant findings in year of publication, mean age, percentage of women, or duration of intervention.

Stress

Five studies assessed caregivers’ perceived stress, of which four were included in the meta-analysis. The meta-analysis showed no statistically significant pooled effect of interventions on stress compared with control groups (SMD −0.21, 95% CI −0.43 to 0.01) with low heterogeneity (I² = 0.0%; 95% CI 0.0% to 84.7%). The study not included in the meta‑analysis reported a significant intervention effect in reducing stress compared with the control group (Puterman et al., 2018). One study reported long-term effects at six months after intervention and reported no significant between-group difference (Connell & Janevic, 2009). Egger’s test was not performed because there were fewer than ten studies. While studies assessing stress were all individual, exercise-based programs targeting dementia caregivers varied in intervention duration, delivery format, and use of theoretical frameworks, and no clear pattern of intervention characteristics was identified to explain the variability in stress‑related outcomes.

Self-rated health

Four studies assessed self-rated health, all of which were included in the meta-analysis. The meta-analysis showed no statistically significant pooled effect in improving self-rated health in caregivers of older adults compared with control groups (SMD 0.28, 95% CI −0.01 to 0.58), with low heterogeneity (I² = 26.9%, 95% CI 0.0% to 72.5%). Egger’s test was not performed because there were fewer than ten studies. The only study reporting a positive effect employed a multicomponent intervention targeting physical activity, nutrition, and sleep (Elliott et al., 2010), whereas the others focusing on a single lifestyle behavior found no between‑group differences. Across studies, interventions varied in target population, duration, delivery format, and use of theoretical frameworks, and no clear pattern of intervention characteristics was identified to explain variability in outcomes.

Discussion

This systematic review and meta-analysis evaluated the effectiveness of lifestyle interventions in promoting behavioral change, mental health, and physical health outcomes among informal caregivers of older adults, and further examined subgroup analyses to examine potential effects from intervention characteristics. Despite the intention to synthesize evidence across multiple lifestyle domains, our findings indicate that the current evidence base is heavily weighted toward physical activity interventions and outcomes. Only one intervention specifically targeted sleep, while the remaining studies included nutrition or sleep as part of multicomponent interventions. None focused on smoking or alcohol use. This imbalance highlights a narrow lifestyle focus within the existing literature and identifies critical gaps that may limit the potential of lifestyle interventions to address caregivers’ broader health risks.

Pooled estimates derived mainly from physical activity interventions suggested a small effect in alleviating depressive symptoms that reached statistical significance, while no consistent effects were observed for other outcomes. Our findings are broadly consistent with prior systematic reviews that reported positive effects of lifestyle interventions on mental health, particularly in reducing depressive symptoms (Baik et al., 2021; Won et al., 2025). However, evidence of publication bias, high risk of bias in many included studies, and an adjusted prediction interval that included the null raise uncertainty regarding the robustness of the findings. The observed effect may be overestimated, and future studies could plausibly observe little or no association. More rigorous and methodologically robust studies are required before firm conclusions about the effects are warranted. Interestingly, improvements in mental health outcomes were observed while no significant behavioral changes in physical activity or sleep quality were reported. By reporting both behavioral and mental health outcomes, our study highlights that mental health improvements were not consistently accompanied by measurable behavioral change, an aspect that has received limited attention in previous reviews. Future studies can examine the potential mechanisms underlying the mental health benefits observed in lifestyle interventions for caregivers.

Our findings provided some indication that delivery mode may potentially influence intervention effectiveness among caregivers of older adults. In subgroup analyses, reductions in depressive symptoms reached statistical significance only among physical activity interventions delivered in hybrid formats that combined in-person sessions with technology-supported monitoring, but not telephone-based or in-person formats. This pattern appears to echo emerging evidence that digital technology-supported interventions may enhance accessibility and engagement for caregivers managing multiple responsibilities (Baik et al., 2021). These technologies also allow for more precise tracking of behavioral changes (e.g., physical activity) and provide objective data, enabling more personalized and responsive intervention strategies (Puterman et al., 2018). The combination of personal connection and ongoing monitoring may be valuable in supporting caregivers’ mental well-being. However, interpretation of these findings is exploratory and does not provide sufficient evidence to draw conclusions about the comparative effectiveness of different delivery formats, particularly given the small number of studies and limited statistical power within subgroups.

Despite physical activity being the most frequently addressed behavior in the review, our analysis found no significant improvements in physical activity levels or physical health outcomes among interventions involving physical activity components. This aligns with previous inconsistent evidence regarding the impact of physical activity interventions on physical health (Baik et al., 2021). The reliance on self-reported physical activity may have introduced bias. Prior research indicated discrepancies between self-reported and objective assessments on physical activity, with self-reported assessment underestimating sedentary activities and overestimating moderate and vigorous physical activity (Ogonowska-Slodownik et al., 2022). Furthermore, most studies measured only leisure physical activity, while other domains, such as occupational, household, and transportation (2018 Physical Activity Guidelines Advisory Committee, 2018), were rarely assessed. Caregivers likely differ in their baseline engagement across these domains and in their capacity to incorporate leisure activity into daily life. Interventions focused solely on leisure physical activity might overlook the broader context of caregivers’ physical activity and its potential impact on health.

Similarly, while sleep disturbances are commonly reported among caregivers, only three studies in our review assessed sleep quality and showed limited positive effects. Several factors may contribute to this lack of effect, including diversity in measurements and caregiver-specific barriers to engagement. Caregivers often face substantial time constraints and caregiving responsibilities, which may hinder their ability to engage in health-promoting lifestyle behaviors in daily life. For example, sleep deficiency can be caused by overnight care and a care recipient’s sleep disturbances (McCurry et al., 2015). Many caregivers prioritize the needs of care recipients over their own, potentially neglecting their own health (Waligora et al., 2019). These factors are unlikely to be resolved through general lifestyle interventions alone. Our findings have also shown that participating in lifestyle interventions is not associated with reduced caregiving burden and stress. Future research could explore whether incorporating elements, such as respite services and strategies that enhance caregivers’ awareness of their own needs, can strengthen their capacity and motivation to adopt health-promoting behaviors and reduce caregiving burden and stress.

By comparing theory‑based and non‑theory‑based lifestyle interventions among caregivers, this review provides new insight into the relevance of behavioral change theories in this population. Over half of the included interventions were grounded in behavioral change theories, such as social cognitive theory. Nonetheless, although theory-based interventions have shown promise in other populations, such as cancer survivors (Stacey et al., 2015), our findings suggest they did not outperform those without explicit theoretical frameworks. Only one sleep-focused intervention reported an improved outcome. This may reflect limitations in applying individual-focused behavioral models to caregiving contexts, where caregivers need to navigate competing priorities between their own health and the needs of care recipients. Traditional behavioral models emphasize personal beliefs and social norms as drivers of behavior, yet in caregiving contexts, social norms may prioritize care recipients’ well-being over caregivers’ own health. This complexity may influence the applicability of these models. Stacey et al. (Stacey et al., 2016) similarly found that social cognitive theory may not be the most suitable framework for promoting behavior change among cancer survivors and their caregivers. Given the contextual parallels, it is plausible that social cognitive theory may be insufficient for explaining behavioral change in caregiving populations.

Limitations

This systematic review and meta-analysis has several limitations. First, the evidence base was uneven across lifestyle domains. Most studies focused on physical activity, with limited evidence on sleep and nutrition and none on smoking or alcohol use. The small number of studies also limited the robustness of subgroup analyses for several outcomes. These constraints reduce the interpretability and generalizability of the findings. Further research is needed to examine how intervention characteristics, such as delivery format and theoretical foundations, influence outcomes in caregiver populations. Second, the high degree of heterogeneity observed across studies should be considered when interpreting the pooled estimates. Differences in intervention design, delivery modes, outcome measures, and caregiver populations suggest that the pooled effects reflect an average across diverse approaches rather than a uniform intervention effect. For instance, physical activity interventions varied widely, ranging from two sessions per week over six months to four sessions per week over twelve months. This variability limits direct comparability and highlights the need for greater standardization in intervention development and outcome reporting in future research.

Third, most included studies were at high risk of bias. Unclear missing outcome data may introduce attrition bias, while reliance on non-blinded, self-reported measures may inflate perceived effects. Consequently, the pooled estimates should be interpreted with caution, as they may reflect both true effects and systematic bias. Moreover, the pooled estimates for physical activity and caregiving burden were disproportionately influenced by a small trial with large effect sizes (Lök & Bademli, 2017). In this study, the same researcher delivered the intervention and collected outcome data during face‑to‑face home visits, raising concerns about detection bias and social desirability bias. These findings highlight the need for larger, more rigorously blinded randomized controlled trials in future caregiver intervention research. Fourth, evidence has only shown short-term effects of lifestyle interventions. The limited number of studies examining long-term outcomes prevented us from conducting a meaningful analysis of sustained effects. Furthermore, the search strategy has several limitations that may have affected the comprehensiveness of the review. The search was restricted to three databases and to English‑language publications, which may have resulted in the omission of relevant studies indexed in additional databases or published in other languages. In addition, the restriction to peer-reviewed studies and exclusion of grey literature may have introduced publication bias, as studies with non-significant or negative findings are less likely to be published. Finally, fewer studies were available for primary analyses of outcomes such as sleep and stress and for subgroup analyses related to physical activity, depression, and caregiving burden. While the findings should be interpreted with caution, they highlight the potential value of tailoring intervention characteristics to the caregiving context, including delivery format and theoretical foundations.

Conclusion

This systematic review and meta-analysis examined the effectiveness of lifestyle interventions for informal caregivers of older adults in improving their behavioral change, physical health, and mental health. Most interventions focused on physical activity, with fewer on nutrition or sleep, and none on smoking or alcohol use. Pooled estimates derived mainly from physical activity interventions suggested a small reduction in depressive symptoms that reached statistical significance. However, no consistent effects were observed in physical activity, sleep quality, caregiving burden, stress, or self-rated health, regardless of intervention characteristics. Given the high risk of bias in many included studies and evidence of publication bias, these findings should be considered preliminary and interpreted with caution. Further confirmation in larger, methodologically robust randomized controlled trials is needed before firm conclusions can be drawn. Future research may also consider how lifestyle interventions can be designed to better engage caregivers and promote behavioral changes, as well as broadening the scope to include nutrition, sleep, and avoidance of risky substance use.

Supplementary material

Supplementary material is available at The Gerontologist online.

Funding

None declared.

Conflicts of interest

None declared.

Data availability

1. Study registration: The study was preregistered at PROSPERO (https://www.crd.york.ac.uk/PROSPERO/view/CRD42023439797).

2. Analytic plan preregistration: The analysis plan was not formally preregistered.

3. Analytic code availability: Analyses were performed in RStudio using the meta package.

4. Materials availability: The included studies in this review are publicly available via PubMed.

Author contributions

Dara K.Y. Leung (Conceptualization [equal], Data curation [equal], Formal analysis [equal], Investigation [equal], Methodology [equal], Project administration [equal], Writing—original draft [equal], Writing—review & editing [equal]), Lucy Xi Lu (Investigation [equal], Writing—original draft [equal], Writing—review & editing [equal]), Yun Han Wang (Data curation [equal], Investigation [equal], Writing—original draft [equal], Writing—review & editing [equal]), Yun Lin Wang (Data curation [equal], Investigation [equal], Writing—original draft [equal], Writing—review & editing [equal]), Artemis Sze Kok Fung (Data curation [equal], Investigation [equal], Writing—original draft [equal], Writing—review & editing [equal]), and Tingan Jin (Formal analysis [equal], Investigation [equal], Writing—original draft [equal], Writing—review & editing [equal])