Effects of cardiac telerehabilitation versus center-based cardiac rehabilitation on cardiorespiratory fitness in patients with coronary artery disease: a meta-analysis

Abstract

Background

Cardiac telerehabilitation (CTR) has emerged as an alternative to center-based cardiac rehabilitation (CBCR) for patients with coronary artery disease (CAD), but its effectiveness in improving cardiorespiratory fitness remains uncertain. This meta-analysis aimed to compare the effects of CTR versus CBCR on maximal oxygen uptake (VO_2max) in patients with CAD.

Methods

A systematic search of PubMed, Embase, Cochrane Library, Web of Science, Wanfang, and CNKI was conducted for relevant randomized controlled trials (RCTs). Pooled mean differences (MDs) with 95% confidence intervals (CIs) were calculated using a random-effects model.

Results

Ten RCTs involving 896 patients were included. Overall, CTR was not associated with a significant difference in VO_2max compared with CBCR (MD: 0.02 mL/kg/min, 95% CI: −0.71 to 0.74; p = 0.97; I² = 46%). Sensitivity analyses confirmed the robustness of the findings. Subgroup analyses showed consistent results across most factors. However, greater improvements were observed in studies with higher baseline VO_2max (p for subgroup difference = 0.03) and longer follow-up (> 26 weeks) (MD: 0.90 mL/kg/min, 95% CI: 0.05 to 1.75; p for subgroup difference = 0.01). Meta-regression indicated that baseline VO_2max and follow-up duration were positively associated with treatment effects. The certainty of evidence was rated as moderate due to imprecision, reflected by a pooled estimate close to null and CI crossing the line of no effect.

Conclusions

CTR provides comparable improvements in VO_2max to CBCR in patients with CAD. Potential benefits may be greater in patients with higher baseline fitness and with longer follow-up, although further high-quality trials are warranted.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12872-026-06041-y.

Introduction

Coronary artery disease (CAD) remains a leading cause of morbidity and mortality worldwide, imposing a substantial clinical and economic burden [1]. Exercise-based cardiac rehabilitation (CR) is a cornerstone of secondary prevention and is strongly recommended for patients with CAD, because it improves functional capacity, reduces recurrent cardiovascular events, and lowers mortality [2]. Central to the benefits of CR is the improvement in cardiorespiratory fitness, commonly assessed by maximal oxygen uptake (VO_2max), which is considered the gold-standard indicator of aerobic capacity [3, 4]. VO_2max is typically measured using cardiopulmonary exercise testing (CPET) and reflects the integrative function of cardiovascular, pulmonary, and muscular systems [3, 4]. Importantly, even modest increases in VO_2max are associated with clinically meaningful reductions in all-cause and cardiovascular mortality, making it a critical therapeutic target in rehabilitation [5–7]. Traditional center-based cardiac rehabilitation (CBCR) provides structured, supervised exercise training along with multidisciplinary support [8, 9]. However, its implementation is limited by low participation and adherence rates due to barriers such as travel distance, time constraints, and healthcare resource availability [8, 9].

To overcome these limitations, cardiac telerehabilitation (CTR) has emerged as an alternative model that delivers rehabilitation remotely using telecommunication technologies, including mobile applications, wearable devices, and telemonitoring systems [10, 11]. CTR enables flexible, home-based participation while maintaining structured exercise prescriptions and remote supervision. This approach may improve accessibility and adherence [12]. Despite its potential advantages, the real-world implementation of CTR remains complex and heterogeneous. Variability in technological infrastructure, healthcare system organization, and patient digital literacy may substantially influence the accessibility and effectiveness of CTR programs in routine practice [13]. Recent implementation-focused research highlights the importance of contextual factors, resource availability, and stakeholder engagement when translating telerehabilitation from controlled trials into clinical settings [14]. In addition, patient safety is a key consideration, particularly as care shifts from supervised, center-based environments to home-based models supported by mobile health technologies. Evidence from real-world mHealth implementation suggests that issues such as liability, patient comprehension, data security, and the need for standardized workflows are critical to ensuring safe and sustainable delivery of remote care [15]. Furthermore, there is increasing recognition that cardiac rehabilitation should be tailored to individual patient characteristics, including baseline functional capacity, comorbidities, behavioral factors, and personal preferences. Advances in digital health technologies, such as wearable devices, real-time monitoring, and mobile applications, offer opportunities for personalized rehabilitation through adaptive exercise prescriptions and continuous feedback [16]. However, challenges related to digital literacy, data privacy, and equitable access remain and may influence the effectiveness and scalability of these approaches. Therefore, a focused synthesis of high-quality randomized evidence is needed to clarify the comparative effectiveness of CTR versus CBCR on VO_2max and to explore potential factors that may influence treatment effects.

Recent randomized trials and meta-analyses have compared CTR with CBCR and generally suggest comparable effects on exercise capacity and other clinical outcomes [17–26]. However, previous reviews often included heterogeneous comparators, such as usual care or mixed rehabilitation models, and evaluated multiple outcomes rather than specifically focusing on VO_2max, a clinically important and objective indicator of cardiorespiratory fitness. In addition, potential effect modifiers, such as baseline fitness level and follow-up duration, have not been comprehensively explored. Meanwhile, findings regarding VO_2max remain inconsistent, with some studies reporting significant improvements favoring CTR, particularly during longer follow-up periods [27], while others show no significant difference between the two approaches [28]. Therefore, the present meta-analysis aimed to provide an updated and focused synthesis of randomized evidence comparing CTR directly with CBCR in patients with CAD, with specific emphasis on VO_2max and factors potentially influencing treatment effects.

Methods

This systematic review and meta-analysis followed established methodological guidance from the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) reporting framework [29] and the Cochrane Handbook for Systematic Reviews of Interventions [30]. The review protocol was registered prospectively with the PROSPERO international prospective register of systematic reviews (CRD420261376803). Clinical trial number: not applicable.

Study inclusion and exclusion criteria

Eligible studies were selected based on predefined criteria structured according to the PICOS framework.

• Population (P): Adult patients (≥ 18 years) with CAD were included, encompassing individuals with acute coronary syndrome (ACS), myocardial infarction (MI), stable CAD, or those undergoing coronary revascularization procedures such as percutaneous coronary intervention (PCI) and coronary artery bypass grafting (CABG). Studies involving mixed cardiovascular populations were eligible if data specific to CAD patients could be extracted separately.

• Intervention (I): The intervention of interest was CTR, defined as rehabilitation programs delivered partially or entirely through telecommunication technologies, including telemonitoring systems, mobile health applications, web-based platforms, or telephone- and video-based supervision. Both fully remote and hybrid programs were eligible. To ensure clinical relevance, interventions were required to include an exercise-based cardiac rehabilitation component.

• Comparator (C): The comparator was restricted to center-based cardiac rehabilitation (CBCR), defined as a structured, supervised exercise program delivered in a hospital, clinic, or rehabilitation facility. Studies comparing telerehabilitation with usual care or non-rehabilitation control groups were excluded to minimize heterogeneity in the comparator.

• Outcomes (O): Cardiorespiratory fitness was assessed using directly measured VO_2max (or peak VO₂) obtained from CPET with respiratory gas analysis, which is considered the gold-standard measure of aerobic capacity [3]. The outcome was reported as the mean difference (MD) between groups at the end of follow-up. Studies were included if they provided sufficient data for extraction or calculation of VO_2max.

• Study Design (S): Only randomized controlled trials (RCTs) with parallel group were included to ensure a high level of evidence and minimize confounding. No restrictions were applied regarding publication year or geographic location.

Studies were excluded if they (1) did not include patients with CAD or failed to provide extractable data for this population; (2) did not evaluate a telerehabilitation-based intervention with an exercise component; (3) did not include a comparator of CBCR; (4) did not report VO_2max or provided insufficient data for analysis; (5) were non-original studies, including reviews, meta-analyses, editorials, or case reports; (6) involved pediatric populations; or (7) represented duplicate publications from the same study population, in which case the study with the longest follow-up duration was included.

Database search

PubMed, Cochrane Library, Embase, Web of Science, Wanfang, and China National Knowledge Infrastructure (CNKI) were systematically searched using the combined terms including: (1) “coronary artery disease” OR “myocardial infarction” OR “acute coronary syndrome” OR “percutaneous coronary intervention” OR “coronary artery bypass grafting”; (2) “telerehabilitation” OR “cardiac telerehabilitation” OR “home-based rehabilitation” OR “remote rehabilitation” OR “internet-based rehabilitation” OR “mobile health” OR “eHealth” OR “mHealth”; (3) “cardiac rehabilitation” OR “exercise rehabilitation” OR “rehabilitation training”; and (4) “randomized controlled trial” OR “randomized” OR “RCT”. We considered only peer-reviewed, English or Chinese language full-text studies conducted in human participants. In addition, the bibliographies of relevant reviews and eligible articles were manually screened to identify any additional records. The final search update was performed on March 05, 2026, and the detailed database-specific search strategies are provided in Supplemental File 1.

Study quality evaluation

Two reviewers independently performed the literature search, screened the retrieved records for eligibility, extracted relevant data, and evaluated study quality. Any disagreements were resolved through discussion with the third author until consensus was reached among the authors. The methodological quality of the included randomized controlled trials was evaluated using the Cochrane Risk of Bias tool (RoB 2.0), which examines potential bias across five domains: the randomization process, deviations from intended interventions, missing outcome data, measurement of outcomes, and selection of the reported results [30]. Each domain was rated as low risk, some concerns, or high risk of bias, and these judgments were subsequently integrated to generate an overall risk-of-bias assessment for each study.

Data extraction

Extracted data included general study information (first author, publication year, and country), study design (double-blind, single-blind, or open-label), participant characteristics (diagnosis, number of patients included, mean age, sex distribution, and mean VO_2max at baseline), details of CTR and CBCR, CTR delivery model (hybrid or fully remote), whether psychological support was included in CTR, and follow-up durations. Hybrid CTR refers to programs that combine in-person, center-based sessions with remote/home-based training supported by telemonitoring or communication technologies. In contrast, fully remote CTR is delivered entirely at home via digital or telecommunication tools, without ongoing on-site supervised sessions. Psychological support in CTR includes behavioral guidance, education, counseling, or structured communication (e.g., feedback, motivational interviewing) aimed at improving emotional well-being, adherence, and self-management; it may range from basic support (e.g., coaching/education) to formal interventions such as cognitive behavioral therapy.

Statistical analysis

The effect of CTR on VO_2max compared with CBCR in patients with CAD was pooled using MDs with corresponding 95% confidence intervals (CIs). Between-study heterogeneity was assessed using Cochran’s Q test and quantified with the I² statistic [31]. A p value < 0.10 for the Q test or an I² value > 50% was considered indicative of substantial heterogeneity [31]. In accordance with current Cochrane recommendations, heterogeneity was interpreted by considering not only statistical measures, but also the magnitude and direction of effect estimates as well as potential clinical and methodological diversity among studies [30]. Pooled estimates were generated using a random-effects model to account for anticipated clinical and methodological diversity among trials [30]. The stability of the findings was evaluated through leave-one-out analyses [30]. Prespecified subgroup analyses were performed according to mean age, proportion of men, baseline VO_2max, CTR delivery mode (hybrid or fully remote), inclusion of psychological support, intervention duration, and follow-up duration. Medians of continuous variables were used as cutoff values for subgroup analyses to ensure a balanced distribution of studies across subgroups. In addition, the univariate meta-regression analysis was also performed to explore the study characteristics on the results of the meta-analysis, such as the sample size, mean age of the patients, proportion of men, mean VO_2max at baseline, durations of intervention, and follow-up durations. Potential publication bias was explored using funnel plot inspection and Egger’s regression test [32]. Statistical significance was defined as a two-sided p value < 0.05. All analyses were carried out using RevMan (version 5.3, Cochrane Collaboration, Oxford, UK) and Stata (version 17.0, StataCorp, College Station, TX, USA) software.

Certainty of evidence

Two reviewers independently assessed the certainty of evidence using the GRADE (Grading of Recommendations, Assessment, Development and Evaluation) approach, which evaluates the level of confidence in the results based on several domains, including risk of bias, inconsistency, indirectness, imprecision, and potential publication bias [33]. The overall certainty for each outcome was classified as high, moderate, low, or very low. Any disagreements between reviewers were resolved through discussion until a consensus was reached.

Results

Literature search

Figure 1 illustrates the study selection process for this meta-analysis. The database search initially identified 354 records, of which 273 remained after removing duplicates. Screening of titles and abstracts led to the exclusion of 244 articles that were not relevant to the study objective. The full texts of the remaining 29 articles were then reviewed in detail, and 19 were excluded for the reasons summarized in Fig. 1. Ultimately, 10 RCTs [17–26] met the eligibility criteria and were included in the quantitative synthesis.

Fig. 1.

Flowchart for the literature search and study inclusion

Characteristics of the included studies

The main characteristics of the included RCTs are summarized in Table 1. This meta-analysis included 10 RCTs published between 2002 and 2025, conducted across multiple countries, including the United States, Canada, Denmark, the Netherlands, New Zealand, Belgium, Spain, the Czech Republic, Greece, and China [17–26]. A total of 896 patients with CAD were enrolled. The included populations were heterogeneous but generally comprised patients with prior MI, PCI, CABG, stable angina, or ACS. Sample sizes ranged from 30 to 222 participants per study. The mean age of participants varied from 45.6 to 74.6 years, and the proportion of men ranged from 60.0% to 91.5%. Baseline cardiorespiratory fitness, assessed by VO_2max, ranged approximately from 14.6 to 27.5 mL/kg/min across studies. CTR interventions differed in technological complexity but generally included home-based exercise programs supported by telecommunication tools such as telephone follow-up, smartphone applications, wearable heart rate monitors, and web-based platforms. Five studies employed fully remote CTR models [17, 21, 22, 25, 26], whereas the remaining five studies used hybrid approaches combining initial supervised sessions with subsequent remote monitoring [18–20, 23, 24]. Psychological support was incorporated in CTR in six studies [17, 18, 20–23], typically in the form of counseling, motivational interviewing, or structured feedback during remote contacts. All studies compared CTR with CBCR, which typically involved supervised exercise sessions conducted in hospital or outpatient settings, often with continuous monitoring and guidance from healthcare professionals. The prescribed exercise intensity and frequency were broadly comparable between groups, although delivery modes differed. The intervention durations ranged from 6 to 40 weeks, with the follow-up durations varying from 12 to 65 weeks.

Table 1.

Characteristics of the included RCTs

Study	Country	Design	No. of patients	Diagnosis	Mean age (years)	Men (%)	Baseline mean VO2max (ml/kg/min)	Details of CTR	CTR delivery model	Psychological support in CTR	Details of CBCR	Follow-up duration (weeks)
Gordon 2002 [18]	USA	R, OL	94	CAD (prior MI, CABG, PCI, or stable angina); Low or moderate risk	60.4	72.0	21.6	Technology used: Telephone (telephone visits with nurse-case manager); Exercise component: Aerobic (individually prescribed home-based exercise plan); Frequency: Unsupervised home exercise encouraged ≥ 3 days/week; Telephone visits at weeks 2, 4, 8, 10; Office visits at weeks 0 and 6; Duration: 12 weeks	Hybrid	Yes — Counseling (one-on-one counseling for stress management as part of lifestyle modification)	Supervised: Yes (continuous ECG telemetry monitoring) Frequency: 3 sessions/week Duration: 12 weeks Setting: Hospital	12
Arthur 2002 [17]	Canada	R, OL	222	CAD (post-CABG surgery, all patients 35–49 days post-CABG); Low-risk	63.4	81.4	15.1	Technology used: Telephone (biweekly telephone calls from exercise specialist); Exercise component: Aerobic (self-paced walking, warm-up/cool-down stretching); Frequency: Advised 5 sessions/week; Actual: 6.5 ± 4.6 sessions/week; Duration: 42.8 min (first 3 months), 47.2 min (next 3 months) Duration: 26 weeks	Fully remote	Yes — Support/education provided during telephone calls. Also access to clinic psychologist (42% used, mean 2.6 visits)	Supervised: Yes (exercise specialists and kinesiologists) Frequency: Advised 3 sessions/week + additional home sessions; Actual: 2.3 ± 0.9 classes/week + 1.4 ± 1.2 additional sessions Duration: 26 weeks Setting: Hospital	26
Oerkild 2011 [19]	Denmark	R, OL	75	CAD (post-MI, post-PCI, or post-CABG); Age ≥ 65 years	74.6	60.0	14.6	Technology used: Telephone (one telephone call between two home visits) + in-person home visits (physiotherapist at weeks 0 and 6) Exercise component: Aerobic (self-paced brisk walking, stationary bicycling) Frequency: Prescribed 6 days/week, 30 min/day; Intensity: Borg 11-13 Duration: 6 weeks (intervention period);	Hybrid	No	Supervised: Yes (group-based supervised exercise) Frequency: 2 sessions/week (60 min each) + encouraged to exercise at home Duration: 6 weeks (intensive program) Setting: Hospital	52
Kraal 2017 [20]	The Netherlands	R, OL	78	CAD, (post ACS, PCI, or CABG); Low to moderate risk	59.2	88.9	24.2	Technology used: Heart rate monitor with chest strap + web application for data upload/review + weekly telephone coaching Exercise component: Aerobic (continuous training on preferred modality: cycling, walking, running) Frequency: Prescribed ≥ 2 sessions/week, 45–60 min/session at 70–85% HRmax Duration: 12 weeks (plus continued use of HR monitor and web app encouraged thereafter)	Hybrid	Yes — Motivational interviewing during weekly telephone calls + goal-setting theory	Supervised: Yes (by physical therapists and exercise specialists) Frequency: 2–3 sessions/week, 45–60 min/session at 70–85% HRmax Duration: 12 weeks Setting: Hospital outpatient clinic	52
Maddison 2019 [21]	New Zealand	R, OL	140	CAD (documented diagnosis within 6 months, AP, MI, coronary revascularization)	61.2	85.5	27.5	Technology used: Smartphone (loaner) + chest-worn wearable sensor + bespoke smartphone and web apps + custom middleware; real-time remote exercise monitoring and audio coaching Exercise component: Aerobic (walking preferred, alternatives allowed); 3 sessions/week for 12 weeks; 30–60 min/session at 40–65% HRR; individualized and progressive prescription Frequency: 3 sessions/week monitored in real-time; Duration: 12 weeks	Fully remote	Yes — Real-time audio coaching, feedback, social support; theory-based behavioural strategies	Supervised: Yes (by clinical exercise physiologists in cardiac rehabilitation clinics) Frequency: 3 sessions/week for 12 weeks Duration: 12 weeks Setting: Cardiac rehabilitation clinics	12
Avila 2020 [22]	Belgium	R, OL	55	CAD, (angiographically-documented or previous MI); low-to-moderate risk	61.2	88.9	26.2	Technology used: Heart rate monitor + web platform for data upload/review + weekly phone or email feedback Exercise component: Aerobic (≥ 150 min/week at 70–80% HRR, preferably 6–7 days/week) Frequency: 3 initial supervised sessions, then home-based; weekly contact via phone/email Duration: 12 weeks	Fully remote	Yes — Weekly contact used to discuss adherence, barriers, and provide feedback (supportive coaching)	Supervised: Yes (by physiotherapists) Frequency: 3 sessions/week, ~ 150 min/week total Exercise: Cycling, treadmill, arm ergometry, rowing, calisthenics Duration: 12 weeks Setting: Hospital outpatient clinic	52
Dalli Peydro 2022 [24]	Spain	R, OL	58	Low-risk ACS (UA, NSTEMI, STEMI); LVEF ≥ 50%	59.2	91.5	23.8	Technology used: Smartphone application + web platform + heart rate monitor + call center for technical assistance Exercise component: Aerobic (walking, strength exercises with resistance bands); target HR 60–80% HRR; Borg RPE 12-14 Frequency: 2 weeks with 4 supervised sessions initially, then home-based; daily exercise and data entry via app for 10 months Duration: 40 weeks	Hybrid	No	Supervised: Yes Frequency: 16 supervised exercise sessions over 8 weeks Duration: 40 weeks Setting: Hospital (University Hospital La Fe, Valencia) Exercise: Routine workouts and aerobic cycling training	40
Batalik 2021 [23]	Czech	R, OL	44	CAD (AP, MI within 6 months, post-revascularization: PCI or CABG); LVEF > 45%	56.6	81.3	23.4	Technology used: Wrist heart rate monitor + web application for data upload/review + weekly telephone feedback Exercise component: Aerobic (walking or cycling); 3 sessions/week, 60 min aerobic phase at 70–80% HRR; 10 min warm-up, 10 min cool-down Frequency: 2 initial supervised sessions, then home-based; weekly telephone calls for feedback Duration: 12 weeks	Hybrid	Yes — Weekly telephone calls provided feedback, advice, and training motivation	Supervised: Yes (by physiotherapist specializing in CR) Frequency: 3 sessions/week, 60 min aerobic phase at 70–80% HRR; 10 min warm-up, 10 min cool-down Duration: 12 weeks Setting: Outpatient clinic Exercise: Cycling on ergometers, walking on treadmill	65
Mitropoulos 2024 [25]	Greece	R, OL	30	CAD (recent MI < 4 weeks; stable clinical status; excluded LVEF < 30%)	53.8	73.3	27.0	Technology used: Smartwatch with ECG and HR readings + oximeter + automated BP monitor + electric body scale + Zoom web-based platform for real-time live video supervision Exercise component: Aerobic (music workout) + resistance training (resistance bands) + balance and flexibility training Frequency: 3 sessions/week for 6 months (26 weeks) Monitoring: Real-time transmission of hemodynamic responses (HR, ECG, SpO₂) via mobile app to health instructor and cardiologist during exercise; RPE monitored regularly Duration: 26 weeks	Fully remote	No	Supervised: Yes (physical education teacher experienced in cardiac rehabilitation) Frequency: 3 sessions/week for 6 months Duration: 26 weeks Setting: Community-based health club (gym) near participants’ residence Note: Gym-based group did not use wearable monitoring; RPE was used to monitor intensity	26
Li 2025 [26]	China	R, OL	110	CAD (post-PCI; Age ≤ 60 years)	45.6	77.0	17.1	Technology used: Heart rate monitor + mobile terminal app/WeChat for exercise recording + remote communication with therapist via phone/WeChat Exercise component: Aerobic + resistance + flexibility + balance training; 60–90 min/session, 3 sessions/week for 12 weeks; individualized prescription based on CPET Frequency: Weekly follow-up via phone or WeChat; patients recorded exercise data via mobile terminal Duration: 12 weeks	Fully remote	No	Supervised: Yes (by rehabilitation therapist) Frequency: 3 sessions/week, 60–90 min/session Duration: 12 weeks Setting: Hospital outpatient department Monitoring: Single-lead ECG monitoring during sessions	12

NR not reported, CAD coronary artery disease, MI myocardial infarction, CABG coronary artery bypass grafting, PCI percutaneous coronary intervention, ACS acute coronary syndrome, UA unstable angina, NSTEMI non-ST-elevation myocardial infarction, STEMI ST-elevation myocardial infarction, LVEF left ventricular ejection fraction, VO_2max maximal oxygen uptake, CPET cardiopulmonary exercise testing, CTR cardiac telerehabilitation, CBCR center-based cardiac rehabilitation, HR heart rate, HRmax maximum heart rate, HRR heart rate reserve, RPE rating of perceived exertion, ECG electrocardiogram, SpO₂ peripheral oxygen saturation

Study quality evaluation

The risk of bias of the included studies was assessed using the Cochrane RoB 2.0 tool, and the results are summarized in Table 2. Overall, the methodological quality of the included trials was acceptable, with no study judged to be at high risk of bias. The randomization process was considered low risk in nine studies [17, 19–26], which reported appropriate methods such as computer-generated randomization, block randomization, or random number tables. However, one early study did not provide sufficient details regarding allocation concealment, resulting in some concerns in this domain [18]. All trials were conducted using open-label designs, which is common in exercise-based rehabilitation research. Consequently, the domain of deviations from intended interventions was generally rated as having some concerns. Nevertheless, the potential impact of performance bias was likely limited because the primary outcome, VO_2max, is an objective physiological measure. The risk of bias due to missing outcome data was low across studies, as follow-up rates were generally high (typically above 85%), and reasons for dropout were reported and balanced between groups. Outcome measurement was also considered to be at low risk of bias in nine studies because VO_2max was assessed using standardized CPET protocols [17–21, 23–26]. In one trial [22], this domain was judged as having some concerns because the outcome assessors were not blinded to the intervention. For selective reporting, all the studies reported prespecified outcomes. Overall, all included studies were judged as having some concerns rather than low risk of bias, primarily due to the open-label design and limited reporting of methodological details in certain trials.

Table 2.

Study quality evaluation via the Cochrane RoB 2.0 Tool

Study	Randomization process	Deviations from intended interventions	Missing outcome data	Measurement of the outcome	Selection of the reported results	Overall
Gordon 2002 [18]	Some concerns (The study states patients were “randomly assigned” but does not describe the allocation sequence concealment method (e.g., sealed envelopes, central randomization). No baseline imbalances were statistically significant, which is reassuring. However, lack of detail on randomization method raises some concerns)	Some concerns (The trial was open-label (participants and staff knew group assignment), which is unavoidable for exercise-based rehabilitation trials. The primary outcome (VO₂max) is an objective physiological measure unlikely to be influenced by knowledge of group assignment. The analysis appears to follow intention-to-treat principles (91.6% completion, similar dropout rates across groups)	Low risk (142 of 155 patients (91.6%) completed follow-up testing. Reasons for dropout were provided and were similar across groups (personal reasons, medical problems unrelated to study). The proportion of missing outcome data is small (< 10%) and balanced)	Low risk (VO₂max was measured using a standardized, calibrated metabolic cart during Bruce protocol treadmill testing to volitional fatigue. Identical procedures were used at baseline and 12 weeks. Outcome measurement was objective and not influenced by investigator or participant knowledge of group assignment)	Low risk (The primary outcomes (VO₂max, lipid profile, blood pressure, weight) were prespecified in the Methods section. Results were reported for all measured outcomes. No evidence of selective reporting)	Some concerns
Arthur 2002 [17]	Low risk (The study used a concealed randomization process: a data analyst prepared a blocked randomization schedule; group assignments were sealed in opaque envelopes and opened in sequence after consent and baseline data were obtained. This is adequate allocation concealment. Baseline characteristics were largely similar, with only a few differences (weight, resting heart rate, social support) that were controlled for as covariates)	Some concerns (The trial was open-label (patients could not be blinded to group assignment), which is unavoidable for exercise rehabilitation trials. However, the primary outcome (VO₂peak) is an objective physiological measure unlikely to be influenced by knowledge of group assignment. The analysis followed intention-to-treat principles)	Low risk (242 patients randomized; 222 (91.7%) returned for final assessment at 6 months (Home: 94.2%; Hospital: 89.3%). Reasons for withdrawal were provided. Follow-up for vital status was 100%. Missing outcome data < 10% and balanced)	Low risk (VO₂peak was measured using a standardized, calibrated metabolic cart during a symptom-limited cycle ergometer test with breath-by-breath analysis. The physicians who evaluated the primary outcome were blind to patients’ group assignment)	Low risk (The primary outcome (VO₂peak was prespecified in the Methods. Results were reported for all measured outcomes at 3 and 6 months. No evidence of selective reporting)	Some concerns
Oerkild 2011 [19]	Low risk (The study used computer-generated randomly permuted blocks with alternate block sizes of 4 to 6. Allocation concealment is implied. Baseline characteristics showed no significant differences between groups)	Some concerns (The trial was open-label. This is unavoidable for exercise rehabilitation trials. The primary outcome (peak VO₂) is an objective physiological measurement unlikely to be influenced by knowledge of group assignment. Analysis was by intention-to-treat)	Low risk (75 randomized; 71 (94.7%) completed 3-month follow-up; 64 (85.3%) completed 12-month follow-up. Reasons for dropout clearly reported. Deaths are expected in elderly cardiac populations and were balanced. Missing outcome data are relatively low and balanced)	Low risk (Peak VO₂ was measured using a symptom-limited cycle ergometer test with breath-by-breath expired gas analysis according to guidelines. Peak VO₂ was defined as the highest mean value within the final minute of exercise. Objective, standardized measurement)	Low risk (Primary outcome (peak VO₂) and secondary outcomes were prespecified in Methods. Results reported for all outcomes at 3 and 12 months. No evidence of selective reporting)	Some concerns
Kraal 2017 [20]	Low risk (Random allocation with variable block size using dedicated computer software. Allocation concealment using numbered, sealed opaque envelopes opened between baseline test and start of training. Baseline characteristics were similar between groups)	Some concerns (The trial was open-label (unavoidable for exercise rehabilitation trials). The primary outcome (peak VO₂) is an objective physiological measure unlikely to be influenced by knowledge of group assignment. Analysis was per-protocol— this is a minor deviation from intention-to-treat but acceptable given the low dropout rate of 9%)	Low risk (90 randomized; 78 (86.7%) completed 1-year follow-up (41 center, 37 home). Loss to follow-up: 12 patients (8 home, 4 center). Reasons provided (withdrawal, comorbidities, death). Missing data balanced and < 15%)	Low risk (Peak VO₂ was measured by maximal exercise test with respiratory gas analysis. Standardized protocol. Objective measurement)	Low risk (Primary outcomes (peak VO₂, physical activity) and secondary outcomes (quality of life, training adherence) were prespecified in the protocol Results reported for all measured outcomes at 12 weeks. No evidence of selective reporting)	Some concerns
Maddison 2019 [21]	Low risk (Computer-generated sequence with variable blocking and stratification. Allocation concealed using sequentially numbered, sealed, opaque envelopes until after baseline assessment. Baseline characteristics were balanced between groups)	Some concerns (Participants could not be blinded to treatment allocation (unavoidable for exercise trials). Primary outcome (VO₂max) is objective. Analysis by intention-to-treat with multiple imputation for primary outcome)	Low risk (162 randomized; 140 (86.4%) completed 12-week primary outcome. Attrition: 13.6% at 12 weeks, 17.3% at 24 weeks. Reasons for dropout provided. Missing data addressed with multiple imputation for primary outcome)	Low risk (VO₂max measured by online metabolic cart during individualized treadmill cardiopulmonary exercise test conducted by blinded exercise physiologists. Objective, standardized measurement with blinded assessors)	Low risk (Primary and secondary outcomes prespecified in published protocol. Results reported for all outcomes. No evidence of selective reporting)	Some concerns
Avila 2020 [22]	Low risk (Web-based random number generator. Baseline characteristics were comparable between groups)	Some concerns (Open-label (unavoidable for exercise trials). Primary outcome (VO₂peak) is objective. Intention-to-treat analysis performed for primary outcome)	Low risk (90 randomized; 80 (88.9%) completed 1-year follow-up. Dropouts: 4 HB, 1 CB, 5 CG — reasons not detailed but balanced. Missing data < 15%)	Some concerns (VO₂peak measured by maximal graded cycle ergometer test with breath-by-breath respiratory gas analysis). Peak VO₂ defined as 30-second average at highest workload. However, test personnel were not blinded)	Low risk (Primary and secondary outcomes prespecified. Results reported for all outcomes. No evidence of selective reporting)	Some concerns
Dalli Peydro 2022 [24]	Low risk (Randomization 1:1 using sealed envelopes from 7 blocks of 10; independent statistician performed randomization. Baseline characteristics were similar between groups)	Some concerns (Open-label (unavoidable for exercise trials). Primary outcome (VO₂max) is objective. Analysis by intention-to-treat)	Low risk (67 randomized; 59 (88.1%) included in ITT analysis. Dropouts: 6 CBCR, 2 CTR (reasons provided: readmissions, refractory angina, declined follow-up). Missing data balanced and < 15%)	Low risk (VO₂max measured by symptom-limited CPET using Bruce protocol with ergospirometer; breath-by-breath respiratory gas analysis; maximal criteria specified (RER > 1.1 or HR > 85% predicted). Outcome assessors blinded)	Low risk (Primary and secondary outcomes prespecified. Results reported for all outcomes. No evidence of selective reporting)	Some concerns
Batalik 2021 [23]	Low risk (Computerized allocation system applying an algorithm in 1:1 ratio. Baseline characteristics were similar between groups with no significant differences)	Some concerns (Open-label (unavoidable for exercise rehabilitation trials). Primary outcome is objective. Analysis appears to be per-protocol (completers only, n = 51), but dropout rate was low (9%) and balanced)	Low risk (56 originally randomized; 44 (78.6%) completed 1-year follow-up. Dropouts: 5 CBCR, 2 HBCT (reasons: COVID-19 pandemic, joint replacement, malignant illness, refusal). Missing data relatively balanced)	Low risk (VO₂ measured by individual maximal CPET with respiratory gas analysis on a bicycle ergometer according to ESC/AHA guidelines. Objective, standardized measurement)	Low risk (Primary and secondary outcomes prespecified. Results reported for all outcomes. No evidence of selective reporting)	Some concerns
Mitropoulos 2024 [25]	Low risk (Stratified randomization by independent statistician blinded to study procedures using nQuery software; allocation concealed; stratification based on sex, comorbidities, LVEF, and CRF. Baseline characteristics were similar between groups)	Some concerns (Open-label (unavoidable for exercise trials). Primary outcome (VO₂peak) is objective. Analysis by intention-to-treat (implied; completers analyzed)	Low risk (36 initially randomized; 30 completed (6 withdrew — 3 from each group). Reasons for withdrawal not detailed but balanced. Missing data < 20%)	Low risk (VO₂peak measured by CPET on treadmill ergometer with breath-by-breath gas exchange analysis; RER ≥ 1.10 used to verify maximal effort. Assessors blinded to group allocation)	Low risk (Primary and secondary outcomes prespecified. Results reported for all outcomes. No evidence of selective reporting)	Some concerns
Li 2025 [26]	Low risk (Random number table method; baseline characteristics similar between groups)	Some concerns (Open-label (unavoidable). Primary outcome is objective)	Low risk (110 randomized; 100 (90.9%) completed. Dropouts: 6 home-based, 4 outpatient. Missing data < 10%)	Low risk (VO₂max measured by CPET with ramp protocol (10–20 W/min). Standardized protocol)	Low risk (Primary and secondary outcomes prespecified. Results reported for all outcomes)	Some concerns

RoB 2.0 revised Cochrane risk of bias tool version 2, VO_2max maximal oxygen uptake, VO_2peak peak oxygen uptake, CPET cardiopulmonary exercise testing, ESC European Society of Cardiology, AHA American Heart Association, RER respiratory exchange ratio, HR heart rate, CBCR center-based cardiac rehabilitation, CTR cardiac telerehabilitation

Meta-analysis results

The pooled results of 10 RCTs [17–26] showed that CTR was not associated with a significant difference in VO_2max compared with CBCR in patients with CAD (MD: 0.02 mL/kg/min, 95% CI: -0.71 to 0.74, p = 0.97; Fig. 2A), with moderate heterogeneity (Cochrane Q test p = 0.06, I² = 46%). These findings indicate that CTR provides comparable improvements in VO_2max to CBCR. Sensitivity analyses, performed by sequentially excluding individual studies, yielded consistent results (MD range: -0.15 to 0.19 mL/kg/min; all p > 0.05).

Fig. 2.

Forest plots for the meta-analysis evaluating the effect of CTR versus CBCR on VO_2max in patients with CAD. A, overall meta-analysis; B, subgroup analysis according to the mean ages of the patients

Subgroup analyses are presented in Figs. 2B, 3, 4 and 5B. As shown in Fig. 2B, stratification by age (< 60 vs. ≥ 60 years) revealed no significant difference between CTR and CBCR in either subgroup (MD: 0.00 vs. 0.06 mL/kg/min; p for subgroup difference = 0.94). In Fig. 3A, analyses based on sex distribution (< 80% vs. ≥ 80% men) similarly showed no significant differences (MD: -0.50 vs. 0.51 mL/kg/min; p for subgroup difference = 0.14). Further subgroup analysis based on baseline VO_2max (< 23 vs. ≥ 23 mL/kg/min) demonstrated a significant subgroup effect (p for subgroup difference = 0.03), with a tendency toward greater improvement in studies with higher baseline VO_2max (MD: 0.71 vs. -0.66 mL/kg/min; Fig. 3B). Further subgroup analyses showed that both hybrid and fully remote CTR (MD: 0.30 vs. -0.32 mL/kg/min; p for subgroup difference = 0.43; Fig. 4A) and CTR with and without psychological support (MD: 0.10 vs. -0.08 mL/kg/min; p for subgroup difference = 0.82; Fig. 4B) yielded comparable effects to CBCR. Moreover, the subgroup analysis suggested that intervention duration (6–12 vs. 26–40 weeks) did not significantly influence the results (MD: 0.22 vs. -0.42 mL/kg/min; p for subgroup difference = 0.43; Fig. 5A). However, a significant subgroup difference was observed according to follow-up duration (≤ 26 vs. > 26 weeks; p for subgroup difference = 0.01; Fig. 5B), with CTR associated with a significant improvement in VO_2max at longer follow-up (MD: 0.90 [0.05 to 1.75] mL/kg/min), but not at shorter follow-up (-0.64 [-1.50 to 0.22] mL/kg/min).

Fig. 3.

Forest plots for the subgroup analyses evaluating the effect of CTR versus CBCR on VO_2max in patients with CAD. A, subgroup analysis according to the proportion of men; B, subgroup analysis according to the mean VO_2max at baseline

Fig. 4.

Forest plots for the subgroup analyses evaluating the effect of CTR versus CBCR on VO_2max in patients with CAD. A, subgroup analysis according to the delivery mode of CTR; B, subgroup analysis according to whether psychological support was included in CTR

Fig. 5.

Forest plots for the subgroup analyses evaluating the effect of CTR versus CBCR on VO_2max in patients with CAD. A, subgroup analysis according to the intervention durations; B, subgroup analysis according to the follow-up durations

Univariate meta-regression analyses were performed to explore potential sources of heterogeneity (Table 3). Baseline VO_2max (coefficient = 0.11, 95% CI: 0.01 to 0.21, p = 0.03) and follow-up duration (coefficient = 0.034, 95% CI: 0.001 to 0.067, p = 0.04) were both positively correlated with the effect of CTR on VO_2max, explaining a substantial proportion of between-study variance (adjusted R² = 65.3% and 100%, respectively). In contrast, sample size, mean age, proportion of men, and intervention duration were not significantly associated with the effect estimates (all p > 0.05). These findings suggest that baseline fitness level and follow-up duration may partly contribute to the observed heterogeneity.

Table 3.

Results of univariate meta-regression analysis

Variables	MD for the influence of CTR on the changes VO2max in patients with CAD compared to CBCR
	Coefficient	95% CI	p values	Adjusted R2
Sample size	-0.0083	-0.0228 to 0.0061	0.22	10.2%
Mean age (years)	0.062	-0.062 to 0.187	0.28	0%
Men (%)	0.028	-0.071 to 0.127	0.53	0%
Baseline mean VO2max (ml/kg/min)	0.11	0.01 to 0.21	0.03	65.3%
Intervention duration (weeks)	-0.012	-0.108 to 0.083	0.77	0%
Follow-up duration (weeks)	0.034	0.001 to 0.067	0.04	100%

CTR cardiac telerehabilitation, CAD coronary artery disease, CBCR center-based cardiac rehabilitation, VO_2max maximal oxygen uptake, MD mean difference, CI confidence interval

Publication bias

Funnel plots assessing publication bias for the meta-analysis of VO_2max are presented in Fig. 6. Visual inspection suggested an approximately symmetrical distribution of the included studies, indicating a low likelihood of publication bias. This was further supported by Egger’s regression test, which did not detect significant asymmetry (p = 0.78). However, these findings should be interpreted with caution, as the number of included studies was relatively small (k = 10), which may limit the statistical power to detect publication bias.

Fig. 6.

Funnel plots evaluating the publication bias underlying the meta-analysis evaluating the effect of CTR versus CBCR on VO_2max in patients with CAD

Certainty of evidence

The certainty of evidence for the primary outcome was assessed using the GRADE approach (Table 4). Overall, the certainty was rated as moderate, mainly due to imprecision, as the pooled estimate was close to null and the CI crossed no effect. Risk of bias, inconsistency, and indirectness were not considered serious, given the objective measurement of VO_2max and generally consistent findings across studies. Publication bias was not evident, although the limited number of studies warrants cautious interpretation. Overall, CTR appears to provide comparable improvements in VO_2max to CBCR, but further high-quality studies may influence this estimate.

Table 4.

Summary of Findings and Certainty of Evidence (GRADE)

Outcome	No. of participants (studies)	Study design	Risk of bias	Inconsistency	Indirectness	Imprecision	Publication bias	Relative effect:	Certainty of evidence (GRADE)	Comments
Changes of VO2max in patients with CAD after CTR compared to CBCR	896 (10 RCTs)	RCTs	Not serious – Although most trials were open-label and judged as “some concerns” in RoB 2.0, the primary outcome (VO2max) was objectively measured using standardized CPET, making bias from lack of blinding unlikely	Not serious – Effect estimates were generally consistent across studies, with moderate heterogeneity (I² = 46%), which was explained in subgroup and meta-regression analyses	Not serious – Included studies directly addressed the target population (CAD), intervention (CTR), comparator (CBCR), and outcome (VO2max)	Serious – The pooled estimate was close to null and the 95% CI crossed no effect (MD 0.02, − 0.71 to 0.74), indicating uncertainty in the magnitude and direction of effect	Not serious – Funnel plot appeared approximately symmetrical, and no substantial publication bias was suggested	MD (95% CI): 0.02 (-0.71 to 0.74)	⨁⨁⨁◯ Moderate	CTR appears to provide similar improvements in VO2max compared with CBCR; however, the effect estimate is imprecise and further high-quality trials may influence the results

Specific reasons for each GRADE domain, including:

Risk of bias: Downgraded if a significant proportion of studies had unclear or high risk of bias in key domains (e.g., random sequence generation, allocation concealment, or selective reporting)

Inconsistency: Downgraded if substantial heterogeneity was observed (I² > 75%) and could not be explained by subgroup analyses or meta-regression

Indirectness: Evaluated but not downgraded, as all included studies directly assessed the population and outcomes of interest

Imprecision: Downgraded if confidence intervals were wide, or if the overall sample size was small

Publication bias: Assessed using funnel plots and Egger’s test; downgraded if significant asymmetry suggested potential bias

Abbreviations GRADE Grading of Recommendations Assessment, Development and Evaluation, RCTs, randomized controlled trials, CAD coronary artery disease, CTR cardiac telerehabilitation, CBCR center-based cardiac rehabilitation, VO_2max maximal oxygen uptake, CPET cardiopulmonary exercise testing, MD mean difference, CI confidence interval, RoB 2.0 revised Cochrane risk of bias tool version 2

Discussion

This meta-analysis synthesizes current randomized evidence comparing CTR with CBCR in patients with CAD, with a specific focus on cardiorespiratory fitness as assessed by VO_2max. Overall, the findings indicate that CTR provides comparable improvements in VO_2max relative to CBCR, with consistent results across sensitivity and most subgroup analyses and moderate certainty of evidence. Rather than suggesting superiority of one approach over the other, the results support the concept that CTR can serve as an effective alternative to conventional rehabilitation delivery models. Given the well-established prognostic importance of VO_2max, even maintaining equivalent improvements is clinically meaningful, particularly in the context of expanding access to rehabilitation services.

The present findings should be interpreted in the context of prior systematic reviews evaluating remote and technology-supported cardiac rehabilitation. Previous meta-analyses have demonstrated that home-based or telehealth cardiac rehabilitation, particularly when supported by wearable devices, can improve cardiorespiratory fitness and may serve as a feasible alternative to conventional programs [34]. In addition, recent evidence has highlighted the safety of remotely delivered cardiac rehabilitation, with very low rates of serious adverse events reported over longer follow-up periods [35]. Other syntheses, including network meta-analyses, have further suggested that multicomponent telehealth interventions combining wearable devices, mobile applications, and communication tools may optimize both efficacy and adherence [36]. However, these prior reviews differ from the present study in several important aspects. First, many earlier meta-analyses included heterogeneous comparators, such as usual care or mixed rehabilitation strategies, which limits direct comparison with standard center-based cardiac rehabilitation. In contrast, the present analysis focused exclusively on head-to-head comparisons between CTR and CBCR, providing a more clinically relevant assessment of whether CTR can substitute for conventional programs. Second, while previous studies often evaluated multiple outcomes, our analysis specifically emphasized VO_2max as a primary endpoint, given its objective measurement and well-established prognostic significance. Third, the current study incorporates more recent randomized controlled trials that reflect contemporary digital health technologies and implementation strategies, which were not fully captured in earlier reviews [36, 37]. Finally, by applying subgroup and meta-regression analyses, this study provides additional insights into potential effect modifiers, such as baseline fitness level and follow-up duration, which may partly explain variability in treatment effects and have not been consistently explored in prior syntheses. Collectively, these differences suggest that the present meta-analysis provides a more focused and updated evaluation of the comparative effectiveness of CTR versus CBCR, while also offering clinically relevant insights into factors that may influence patient response to telerehabilitation.

From a clinical and physiological perspective, the comparable effects of CTR and CBCR on VO_2max are plausible. Both strategies are grounded in structured, exercise-based rehabilitation programs that aim to improve cardiovascular efficiency, peripheral oxygen utilization, and overall functional capacity [38]. Improvements in VO_2max are driven primarily by exercise dose (frequency, intensity, duration, and adherence), rather than the physical setting in which the training is delivered [39]. CTR programs, when appropriately designed, can replicate key elements of CBCR, including individualized exercise prescriptions, progression protocols, and monitoring of physiological responses [40]. Furthermore, CTR may offer unique advantages such as greater flexibility, improved convenience, and integration into daily life, which can enhance adherence and promote sustained behavioral change [41]. On the other hand, CBCR provides direct supervision, immediate clinical support, and structured environments that may benefit certain patient populations, particularly those at higher risk or with lower baseline functional capacity [42]. Therefore, the comparable outcomes likely reflect a balance between the strengths and limitations of each approach, suggesting that both models can achieve similar physiological benefits when adequately implemented.

The subgroup and meta-regression analyses provide additional insights into factors that may influence the effectiveness of CTR. Notably, greater improvements after CTR in VO_2max were observed in studies enrolling patients with higher baseline VO_2max. This finding may reflect several underlying mechanisms. Patients with higher baseline fitness are often more capable of engaging in higher-intensity exercise training, which is a key determinant of improvements in aerobic capacity. They may also have better exercise tolerance, fewer comorbidities, and greater motivation or self-efficacy, all of which facilitate adherence to prescribed training regimens in a remote setting [43]. In contrast, patients with lower baseline VO_2max may require closer supervision, gradual progression, and more intensive support, which may be more effectively delivered in a center-based environment [44]. In addition, baseline VO_2max may act as a surrogate marker for overall health status, influencing responsiveness to rehabilitation interventions [45].

Similarly, the observation that CTR was associated with greater improvements in VO_2max at longer follow-up durations suggests that the benefits of remote rehabilitation may accrue over time. CTR allows patients to integrate exercise into their daily routines in a sustainable manner, potentially leading to better long-term adherence and compliance compared with time-limited, facility-based programs [46]. The flexibility and convenience of home-based rehabilitation may improve sustained participation, particularly for patients who face barriers to attending center-based sessions, such as travel distance, work commitments, or limited accessibility. Over extended follow-up periods, this sustained engagement may translate into incremental gains in aerobic capacity [27]. In contrast, adherence to CBCR may decline after supervised sessions end. However, adherence and compliance outcomes were inconsistently reported across studies, and factors such as patient motivation, digital literacy, self-efficacy, and access to technology may substantially influence long-term participation in CTR programs. The positive association between follow-up duration and treatment effect identified in meta-regression further supports the importance of sustained engagement in achieving meaningful improvements in VO_2max.

In addition to effectiveness, safety and long-term sustainability are critical considerations when evaluating CTR as an alternative to CBCR. Emerging evidence suggests that remotely delivered cardiac rehabilitation is generally safe, with very low rates of exercise-related serious adverse events and no reported exercise-related mortality in recent systematic reviews of randomized trials [35]. These findings support the feasibility of CTR when appropriate patient selection, risk stratification, and remote monitoring strategies are implemented [35]. Advances in wearable devices and telemonitoring systems enable real-time assessment of physiological responses during exercise, which may help mitigate safety concerns in home-based settings [47]. Nevertheless, issues such as delayed clinical response, patient adherence to monitoring protocols, and variability in technological infrastructure should be considered when interpreting safety in real-world practice. From a long-term perspective, CTR may offer advantages in maintaining rehabilitation benefits beyond the supervised intervention period. As suggested by the present subgroup and meta-regression analyses, longer follow-up duration was associated with greater improvements in VO_2max, which may reflect sustained engagement with exercise facilitated by flexible, home-based programs. In contrast, adherence to CBCR often declines after the completion of structured, facility-based programs. However, evidence regarding long-term clinical outcomes, such as mortality, recurrent cardiovascular events, and rehospitalization, remains limited and heterogeneous across studies. Furthermore, sustainability of CTR may be influenced by factors such as digital literacy, patient motivation, and access to technology, highlighting the need for individualized approaches and ongoing support. Taken together, while current evidence supports CTR as a safe and potentially sustainable alternative to CBCR, particularly for low-to-moderate risk patients, further large-scale studies with extended follow-up are required to better define its long-term clinical benefits and optimal implementation strategies.

This meta-analysis has several notable strengths compared to previous meta-analyses [27, 28]. First, it includes an up-to-date and comprehensive literature search, incorporating recent RCTs [25, 26] that were not included in earlier meta-analyses. Second, only RCTs were included, enhancing the internal validity and minimizing confounding. Third, the analysis employed strict inclusion criteria, focusing specifically on patients with CAD and directly comparing CTR with CBCR, thereby reducing clinical heterogeneity related to mixed populations or heterogeneous comparators such as usual care. Fourth, the study concentrated on a clinically relevant and objective primary outcome (VO_2max), avoiding variability associated with surrogate or subjective endpoints. Finally, extensive subgroup and meta-regression analyses were conducted to explore potential sources of heterogeneity, providing additional insights into factors influencing treatment effects.

However, several limitations should be acknowledged. First, although the overall heterogeneity was moderate, there was variability in CAD subtypes, baseline characteristics, and prior treatments (e.g., PCI, CABG), which may influence responsiveness to rehabilitation. Second, differences in the content and delivery of rehabilitation programs, including exercise intensity, frequency, duration, and the inclusion of behavioral or psychological support, may contribute to between-study variability. Third, the analyses were based on study-level data, and the lack of individual participant data limited the ability to perform more detailed subgroup analyses or adjust for patient-level confounders, such as age, sex, and comorbidities. Fourth, blinding of participants and personnel was generally not feasible in rehabilitation trials, which may introduce performance bias. However, the primary outcome (VO_2max) was objectively measured, reducing the risk of measurement bias. Fifth, the intervention and follow-up durations were relatively short in several studies, which may underestimate the long-term effects of rehabilitation strategies. Finally, the number of included studies was modest, which not only limited the ability to detect publication bias reliably, but also reduced the statistical power and robustness of the subgroup and meta-regression analyses. Therefore, these exploratory findings regarding potential effect modifiers should be interpreted cautiously.

From a clinical perspective, these findings support the use of CTR as a viable alternative to CBCR, particularly for patients who face barriers to accessing traditional programs. CTR may play an important role in expanding the reach of cardiac rehabilitation, improving participation rates, and facilitating long-term adherence. In real-world practice, safe implementation of CTR requires careful patient selection, risk stratification, and the use of remote monitoring tools (e.g., wearable devices and telemonitoring systems) to ensure timely detection of abnormal physiological responses during exercise. However, patient selection remains important, and individuals with lower baseline fitness or higher clinical risk may still benefit from supervised, center-based programs.

In addition, successful implementation of CTR depends on effective stakeholder engagement, including patients, clinicians, and healthcare systems, to address barriers such as digital literacy, access to technology, and integration into existing clinical workflows. CTR also provides opportunities for personalized rehabilitation through flexible, home-based programs that can be adapted to individual functional capacity, comorbidities, preferences, and adherence patterns. In view of the limitations, the results should be interpreted cautiously, and CTR should be considered as part of a personalized rehabilitation strategy tailored to patient characteristics and healthcare resources.

Future research should focus on well-designed, adequately powered RCTs with standardized intervention protocols and longer follow-up durations to better evaluate the long-term effectiveness and safety of CTR. Studies incorporating individual participant data would enable more precise assessment of effect modifiers, including baseline fitness, comorbidities, and adherence patterns. In addition, future trials should aim to standardize the definition and reporting of rehabilitation components, including exercise dose and behavioral interventions, to improve comparability across studies. Evaluating cost-effectiveness, patient-centered outcomes, and implementation strategies will also be important to inform large-scale adoption of CTR in clinical practice.

Conclusions

In conclusion, this meta-analysis demonstrates that CTR provides comparable improvements in VO_2max to CBCR in patients with CAD. The benefits of CTR may be more pronounced in patients with higher baseline fitness and with longer follow-up durations. While CTR represents a promising and accessible alternative to traditional rehabilitation, further high-quality evidence is needed to optimize its implementation and to better define the patient populations most likely to benefit.

Authors’ contributions

Xingjie Zhong and Jiangqin Wu conceived the study. Xingjie Zhong and Liping Li performed database search and study identification. Xingjie Zhong and Jiangqin Wu performed study quality evaluation and data extraction. Xingjie Zhong, Jiangqin Wu, and Peiyun Feng performed statistical analysis and interpreted the results. Xingjie Zhong and Jiangqin Wu drafted the manuscript. All authors revised the manuscript and approved the submission.

Funding

The study did not receive any specific funding from funding agencies in the public, commercial or non-profit sectors.

Data availability

The original contributions presented in the study are included in the article; further inquiries can be directed to the corresponding author.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.