A bibliographic search was performed on July 23, 2024, using the following databases: US National Library of Medicine (PubMed or MEDLINE), SCOPUS, Web of Science, and the Cochrane Database of Systematic Reviews. No temporal restrictions were applied for the inclusion of articles. The study question was framed using the Population, Intervention, Comparison, and Outcome (PICO) model [22] available in Table 1. Medical Subject Headings terms and keywords were used for the search string, in accordance with the specifications of each consulted database (Multimedia Appendix 2). To strengthen the research, reference lists from both the selected studies and pertinent literature reviews on the subject were examined as well. The primary goal was to investigate the effects of telerehabilitation treatment compared with other forms of treatment and with no treatment in terms of pain, upper limb functionality, and QoL. The secondary goal was to determine if there is a difference in the rate of complications among those who underwent telerehabilitation treatment, other forms of treatment, or no treatment.

This SR included (1) articles in English and Italian; (2) randomized controlled trials (RCTs); (3) studies that included a population of adult women (age >18 y) who underwent surgery for BC; (4) telerehabilitation treatment versus other types of treatment or no treatment; and (5) studies that measured the effects of the treatments in the domains of QoL, pain, function, complications incidence, and body weight composition.

All RCT studies that included a group undergoing telerehabilitation treatment compared with one or more groups undergoing other types of treatment or no treatment were included. Telerehabilitation treatment refers to all rehabilitation interventions delivered remotely, even without direct supervision by health care personnel. Telemonitoring interventions (eg, monitoring of symptom variations such as pain and exercise-related fatigue) were also included.

All publications that had one or more of these elements were excluded: (1) articles written in languages other than English or Italian; (2) non-RCT studies, observational studies, case reports, case series, narrative reviews and overviews, study protocols, SRs, meta-analyses, and gray literature; (3) pediatric population (age <18 y); and (4) conference papers.

Duplicates were initially identified and removed using the Rayyan web app for SRs [23]. Two reviewers (FS and BC) independently assessed the titles and abstracts to determine their eligibility. At the end of this phase, the 2 reviewers examined, always independently, the full text of the articles to verify their eligibility criteria. After selecting the studies to include, data were extracted. Any discrepancies or disagreements between the 2 reviewers in the previous phases were discussed and resolved with the help of a third reviewer (MB).

The Cochrane risk-of-bias tool for randomized trials version 2 (RoB 2) [24] was used to assess articles included in this SR. To consider all potential areas of bias, the 5 items of the tool (the randomization process, deviations from the intended interventions, missing outcome data, measurement of the outcome, and selection of the reported result) were evaluated as “low,” “some concerns,” or “high.” A proposed judgment about the risk of bias arising from each domain is generated by an algorithm based on answers to the signaling questions. Two authors (FS and BC) independently evaluated the RCTs included. Any disagreements that arose between the primary 2 reviewers were resolved through discussion or with a third reviewer (MB).

For continuous variables, the mean was calculated according to the guidelines of the Cochrane Handbook for Systematic Reviews of Interventions version 6.3 [25]. If the results were reported as median, IQR, mean value, or 95% CI, a conversion was applied. For studies with incomplete or not directly accessible data, an attempt was made to contact the corresponding author for feedback. In case of no response or inability to provide further data, the articles were excluded from the review. A random-effects model was used a priori due to anticipated clinical and methodological heterogeneity among included studies, including differences in intervention duration and outcome measures. This choice is consistent with recommendations that the use of random-effects models should be based not only on statistical heterogeneity, but also on conceptual considerations [26]. We prespecified subgroup analyses for follow-up time (end of treatment, approximately 12 weeks and 24 weeks) in outcomes reported at multiple timepoints. Where sufficient data were available (≥3 studies per subgroup), we also planned subgroups by outcome instrument family (eg, Functional Assessment of Cancer Therapy vs European Organisation for Research and Treatment of Cancer [EORTC] for QoL) to examine whether the choice of measurement tool influenced pooled effects. Additional subgroup or sensitivity analyses (eg, by comparator type—standard care vs no rehabilitation and by telerehabilitation delivery mode—synchronous, asynchronous, and hybrid) were considered exploratory and conducted only when ≥2 studies informed each category. All analyses were performed using Review Manager (RevMan) version 5.4 (Cochrane Collaboration, 2020), which was used for meta-analyses, forest plots, and chi-square tests for subgroup differences. GRADEpro (guideline development tool; Evidence Prime) was used to generate evidence profiles and a summary of findings.

The certainty of evidence for each outcome was evaluated using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach [27]. In total, 5 domains were considered: risk of bias, inconsistency, indirectness, imprecision, and publication bias. Evidence profiles and summary of findings tables were generated with the GRADEpro guideline development tool.

A total of 4617 studies were found after the systematic search of online databases. From duplicate analysis conducted using Rayyan online software, 1414 articles were deleted.

After the analysis of title, abstract, and keywords, 3045 articles were excluded since they did not meet the inclusion criteria (background article: 1777, wrong outcome: 527, wrong publication type: 383, wrong study design: 253, wrong population: 133, wrong drug: 43, and foreign language: 11). Therefore, 158 articles were analyzed in detail through full text screening to verify their eligibility: 37 studies were excluded due to the publication type (ie, study protocol), 14 were excluded because the study design did not meet the inclusion criteria, 31 articles were excluded because their population was different from patients undergoing BC surgery, 33 articles were excluded because their intervention did not meet the PICO question, and 32 articles were excluded because the outcomes did not meet the PICO question and the inclusion criteria. At the end of the screening process, 11 RCTs [28-38] were selected as they met all the eligibility criteria.

The study selection process and the included trials are reported in Figure 1.

A total of 11 RCT studies [28-38] were included and summarized in this SR.

The articles were published between 2016 [30] and 2023 [36,37]. A total of 748 women who underwent BC surgery were enrolled in the control and experimental groups. Table 2 summarizes the sociodemographic and clinical characteristics of participants across the included studies.

The interventions analyzed were highly heterogeneous regarding the types of instruments used, training frequency, intensity, and duration. However, they can be classified according to the method of therapy delivery.

In total, 5 studies [28,29,32,34,38] based their interventions on mobile apps, albeit in different therapeutic approaches. For instance, Çınar et al [28] designed a mobile app that offered information, advice, and consultations from qualified nurses, alongside daily reminders. In contrast, Dong et al [29] used a step-tracking app for monitoring PA, complemented by remote video instructions for muscle training and supportive messages via social media, while de Aviz et al [38] performed an asynchronous treatment with the sending of exercises via instant messaging app, in order to induce physical improvements, in the QoL and reduce fatigue and pain in postoperative patients with BC.

Only the studies of Galiano-Castillo et al [30,31] focused on a web-based intervention, using the e-CUIDATE system, which provided a personalized exercise program accessible through a secure online interface, while Zhou et al [37] used a multimodal care program via WeChat (Tencent Holdings Limited), including relaxation training, expression of feelings, and social support.

The remaining studies [33,35] combined several rehabilitation interventions: Park et al [35] used the Xbox One Kinect (Microsoft), integrating it with virtual reality experiences; Loubani et al [33] blended clinical sessions with home-based telerehabilitation through the Kinect-based CogniMotion system (ReAbility Online, Gertner Institute, Sheba Medical Center, Ramat Gan, Israel) for both motor and cognitive exercises, while Swartz et al [36] combined virtual exergaming via SecureVideo with PA behavioral coaching.

It is important to highlight that nearly all the studies provided patient education within their interventions. This may include individualized information delivered through online coaching sessions [36], content available via mobile apps [28,29,32] or instant messaging apps [38], or combined methods using email and apps [34], as well as online consultations using platforms such as WeChat [37], exercise guidelines, or written recommendations [35]. Galiano-Castillo [30,31] reported having conveyed the significance of the exercise program in the context of post-BC rehabilitation to the study cohort.

Therefore, the studies proposed a variety of exercises as part of their rehabilitative interventions for women with BC. Among these, the main types include video game–based PA [36], specific exercises for the upper limbs [29,33,35,37,38] or lower limbs, cardiorespiratory training [29], high-intensity interval training [34], relaxation exercises and guided imagination [28], and both resistance and aerobic training [30,31]. Naturally, the choice of exercises depends on the specific objectives of the study, such as improving physical function, QoL, or ROM.

Table 3 provides a comprehensive summary of the RCTs included in this review, detailing study design, sample characteristics, type and duration of interventions, comparators, outcome measures, and main findings.

The characteristics of the included interventions were further synthesized according to the Frequency, Intensity, Time, and Type framework, as reported in Table 4. While frequency and time or duration were consistently described across studies, intensity was often poorly defined or heterogeneous. Notably, in the study by Park et al [35], intensity was not expressed through quantitative measures (eg, load, percentage of one-repetition maximum, or rating of perceived exertion [RPE]) but only described in terms of progression in ROM and the introduction of active movements with or without dumbbells.

Across the 11 RCTs, a total of 748 women were included (373 allocated to telerehabilitation and 375 to control arms). The mean age of participants spanned from the early 40s to the mid-60s, with most cohorts representing middle-aged women, although the US trial (Swartz et al [36]) recruited an older sample (mean 63.8 y). Cancer stage was predominantly early disease (stages 1-2), although several trials [36] also included patients at stage 3 or 3A. Surgical management varied: breast-conserving surgery predominated in Taiwan (76/112, 67.9%) [32], whereas Turkish and Israeli cohorts [28,33] reported mixed lumpectomy and mastectomy distributions, and immediate reconstruction was frequent in the Korean multicenter trial (78/100, 78%) [35]. Axillary lymph node dissection was reported in the intervention group in the study by Park et al [35]. The 2 Spanish trials [30,31] recruited BC survivors who had completed adjuvant therapy, with a balanced distribution of lumpectomy, quadrantectomy, and mastectomy. In other cohorts, details of surgical procedures were often sparse. Beyond age, sex, stage, and surgery, additional sociodemographic descriptors (eg, education, employment, marital status, ethnicity) were underreported. Where available, most participants were married (69.1%), had a basic educational level (45.7%), and were housewives or on medical leave [31]. Overall, baseline characteristics were generally well balanced between intervention and control groups within studies, supporting internal validity. A full study-level summary is provided in Table 2.

In total, 9 articles [28-30,32-35,37,38] assessed QoL as part of their research. The most frequently used assessment was the Functional Assessment of Cancer Therapy-Breast (FACT-B). However, the specific version of the scale varied across studies. For instance, Zhou et al [37] used version 4 of the FACT-B scale, while Loubani et al [33] and Park et al [35] used the original version. Meanwhile, Çınar et al [28] opted for a different approach, using the Functional Assessment of Cancer Therapy-Endocrine Subscale QoL scale, which is tailored for endocrine symptoms and de Aviz et al [38] administered Functional Assessment of Cancer Therapy-General (FACT-G).

The EORTC QoL questionnaires were used in 2 articles [30,32]. Galiano-Castillo et al [30] administered the EORTC Quality-of-Life Questionnaire Core 30 (EORTC QLQ-C30). Similarly, Hou et al [32] also used the EORTC QLQ-C30, but supplemented it with the EORTC Breast Cancer–Specific Quality-Of-Life Questionnaire, a module specifically designed for patients with BC.

The EQ-5D was another instrument used in 2 studies. Ochi et al [34] assessed participants using the standard EQ-5D total score, while Park et al [35] opted for the EQ-5D-5L version.

Only 1 study, conducted by Dong et al [29], used the Short Form Health Survey 36 to evaluate QoL.

As illustrated in Table 3, the majority of studies [28-30,32,33,37,38] reported significant improvements in QoL among participants in the intervention groups. However, not all findings were consistent. Of the total, 2 studies [34,35] did not observe statistically significant differences between the intervention and control groups, even though both groups experienced some degree of improvement.

All articles analyzed function and PA, albeit using different evaluation and assessment methods. In general, PA was measured using accelerometers, assessing mean daily steps and mean minutes of moderate to vigorous PA only by Swartz et al [36].

Additionally, Swartz et al [36] used various performance tests to evaluate the effect of their intervention on physical function, including the Short Physical Performance Battery, gait speed, Timed Up and Go test, and 2-minute step test.

Furthermore, 2 other articles used performance tests to measure PA, in particular, Dong et al [29] used the Stand-Up and Sit-Down Chair Test, while Galiano-Castillo et al [31] used the 6-Minute Walk Test (6MWT).

In addition, 3 studies [33,35,38] used questionnaires to assess upper limb function. Specifically, Loubani et al [33] and de Aviz et al [38] used the Disability of Arm, Shoulder, Hand (DASH) questionnaire, while Park et al [35] administered the Quick Version of the Disability of Arm, Shoulder, Hand (QuickDASH) questionnaire score.

Despite the methodologies, results indicated a moderate effect size for the Short Physical Performance Battery and 2-minute step test, along with small effect sizes for gait speed, Timed Up and Go test, and moderate to vigorous PA. Significant improvements were noted in the experimental groups as reported by Dong et al [29], Loubani et al [33], and Galiano-Castillo et al [31].

In contrast, Park et al [35] documented improvements over time without significant differences between the 2 groups.

In total, 5 articles analyzed muscle strength. Specifically, Dong et al [29] evaluated the upper limb strength using the arm lifting test, while De Aviz et al [38] used a digital dynamometer to measure shoulder joint flexion, extension, abduction, external rotation, and internal rotation. Additionally, Loubani et al [33], Ochi et al [34], and Galiano-Castillo et al [30] assessed handgrip strength using a dynamometer.

Strength assessment results indicated significant improvements in the experimental group as reported by Dong et al [29], while de Aviz et al [38] found enhancements in both groups pertaining to upper limb strength, without a statistical difference. Regarding handgrip strength, Galiano-Castillo et al [30] reported significant enhancement solely in the experimental group, which was maintained at the 6-month follow-up. Ochi et al [34] showed no significant differences in between-group analyses.

Of the total, 2 studies [29,31] examined the cardiorespiratory capability: Dong et al [29] used the maximal oxygen consumption, while Galiano-Castillo et al [31] used the 6-Minute Walk Test to evaluate the functional capability. The results were inconsistent; the study by Galiano-Castillo et al [31] showed significant improvements in the intervention group during both T1 and T2 evaluations compared with the control group, while Dong et al [29] found no differences between the groups at the end of the intervention.

ROM was assessed by Park et al [35], specifically focusing on shoulder abduction and flexion and by de Aviz et el [38] for shoulder joint flexion, extension, abduction, external rotation, and internal rotation.

Only Ochi et al [34] evaluated BMI and body mass in kilograms. Both studies reported no significant differences in between-group analyses.

Table 3 presents the results of functional and physical assessments.

In total, 4 articles [30,35,37,38] evaluated pain as an outcome measure. Zhou et al [37] and Park et al [35] used the numerical rating scale (NRS), while de Aviz et al [38] used the visual analog scale, and Galiano-Castillo et al [30] applied the brief pain inventory.

The findings across these studies were variable: 3 studies [30,37,38] reported significantly lower pain in the experimental group than the control group. It is important to underline that, although the results did not present a statistically significant difference between the groups, in the study by de Aviz et al [38], the patients in the experimental group did not present pain at the intermediate and final evaluation.

In contrast, Park et al [35] found no significant differences between the groups.

Fatigue was assessed by Zhou et al [37] using the NRS, Galiano-Castillo et al [30] using the revised Piper Fatigue Scale, and de Aviz et al [38] using the Functional Assessment of Cancer Therapy-Fatigue. Zhou et al [37] reported no significant differences between the groups, while Galiano-Castillo et al [30] and de Aviz et al [38] demonstrated a significant improvement in total fatigue perception in the intervention group compared with the control group, with this effect maintained at follow-up (T2).

Table 3 presents the results of the assessments for pain, fatigue, and kinesiophobia.

The risk of bias assessment with RoB2 [24] across the included studies revealed variability in methodological quality (Figure 2). Most studies, in the overall assessment, are at low risk of bias, while 1 study [33] raises some concerns, and 1 study [29] is at high risk of bias.

Most studies demonstrated a low risk of bias in terms of randomization and allocation concealment, suggesting robust methodological designs. However, 1 study [29] seems to have adopted a per-protocol analysis without a clear balancing of dropouts, raising doubts regarding domain D2, deviations from intended intervention, thus resulting in a high risk of bias. A major limitation across multiple studies was the lack of blinding for both participants and intervention providers, which is particularly relevant given the reliance on patient-reported outcomes in telerehabilitation interventions. This raises concerns about performance and detection bias, potentially influencing the reported effectiveness of interventions. Incomplete outcome data and selective reporting were generally well-managed, with most studies showing a low risk of bias in these domains, ensuring transparency and data integrity. Nonetheless, other biases, such as small sample sizes [36], lack of preregistered protocols in English [33,37], and possible confounding factors, were present in some studies, warranting caution in interpreting their results. Overall, while the included studies provide valuable insights into telerehabilitation for patients post BC surgery, future research should prioritize improved blinding strategies, objective outcome measures, and standardized reporting protocols to enhance methodological rigor and reliability of findings.

The GRADE summary of findings (Table 5) for the main outcomes at different time points showed that at the end of treatment moderate evidence suggests that telerehabilitation was associated with a statistically significant improvement in QoL compared with standard care (SMD 0.59, 95% CI 0.24-0.95; moderate certainty); similarly regarding to handgrip strength (MD 2.93 kg, 95% CI 0.82-5.04; low certainty). A moderate reduction in pain was also observed (SMD 0.50 lower, 95% CI 0.79-0.22 lower; low certainty). No relevant effects were found for upper limb function (SMD 0.86 lower, 95% CI –2.02 to 0.31; low certainty). During the 12-week follow-up, telerehabilitation maintained a beneficial effect on QoL (SMD 0.54 higher, 95% CI 0.14-0.93; low certainty) and was associated with a reduction in pain intensity (SMD 0.33 lower, 95% CI –0.60 to –0.06; low certainty). At 24 weeks follow-up, the improvement in QoL was further increased (SMD 0.75 higher, 95% CI 0.45-1.05; low certainty), while pain reduction remained modest (SMD 0.32 lower, 95% CI –0.60 to –0.03; low certainty). The certainty of evidence ranged from low to moderate, mainly due to limitations in study design and imprecision related to small sample sizes and wide CIs.

This SR with meta-analysis evaluated the clinical effectiveness of telerehabilitation in women undergoing BC surgery, focusing on 4 core outcomes: QoL, pain, upper limb function, and handgrip strength. A total of 11 RCTs were included in the qualitative synthesis, of which 5 were eligible for meta-analysis. The overall findings support the use of telerehabilitation as a valid, safe, and in some cases superior alternative to routine care, especially in improving QoL and muscular strength, while some functional outcomes require further investigation.

The most consistent benefit of telerehabilitation across included studies was its effect on QoL. The meta-analysis showed a moderate, statistically significant improvement (SMD 0.59) with moderate certainty of evidence. This effect was maintained across multiple follow-up points and QoL instruments, including FACT-B, FACT-G, and EORTC QLQ-C30. These findings align with those reported by Wen et al [16], who observed that eHealth-based interventions significantly improved general and subdomain QoL metrics in women with BC. Similarly, Peng et al [15] noted improvements in fatigue and functional domains in patients with BC who engaged in structured tele-exercise programs.

Specific interventions, such as the e-CUIDATE system studied by Galiano-Castillo et al [30], showed both QoL gains and durability of effects over a 6-month follow-up, supporting the long-term potential of well-structured telerehabilitation protocols.

In contrast, Scaturro et al [12] found that synchronous telerehabilitation, delivered through a web-meeting platform, was effective, but face-to-face rehabilitation yielded greater QoL improvements, potentially due to individualized adjustment and closer supervision. This observation underscores the role of patient-physiotherapist interaction and feedback loops, which are harder to replicate in remote formats.

Pain control emerged as another area where telerehabilitation was effective. The pooled analysis showed a significant reduction in pain (SMD –0.50). In literature, the beneficial effects of rehabilitation and physical exercise on pain reduction in patients undergoing BC surgery are well-established [39]. However, to date, no SRs have comprehensively evaluated this specific outcome in the context of telerehabilitation in this specific population. To the best of our knowledge, this is the first SR with meta-analysis to demonstrate the reduction of pain in patients receiving telerehabilitation following BC surgery.

Generally, the efficacy of telerehabilitation in alleviating pain is documented for both acute [40,41] and chronic pain, short-term and long-term [41], in patients following cancer surgery. The findings of our SR are thus consistent with existing literature on this subject.

Handgrip strength improved significantly (MD 2.93 kg); nevertheless, given the low certainty of evidence and reference values for minimal clinically important differences reported in the literature, this change is unlikely to be clinically meaningful [42,43]. Grip strength is an important functional indicator post mastectomy or quadrantectomy and is correlated with patient independence and upper limb integrity [44].

Upper limb function did not significantly improve in the meta-analysis (SMD –0.86, 95% CI –2.02 to 0.31). This may be due to substantial variability in the tools used (DASH and Quick-DASH), the intensity of training, or baseline impairments. Scaturro et al [12] found outpatient rehabilitation to be more effective than telerehabilitation in improving upper limb functional scores, possibly due to the capacity for individualized correction and targeted kinematic feedback.

Park et al [35] attempted to overcome this by using virtual reality–based telerehabilitation with the Xbox Kinect, achieving encouraging improvements in shoulder function and patient engagement.

The clinical implications of these findings are substantial. Telerehabilitation overcomes several barriers that hamper access to rehabilitation services, including geographic isolation, transportation challenges, financial burdens, and caregiver constraints. Studies such as those by Keikha et al [17] and Anik et al [45] emphasize that telerehabilitation can significantly enhance service reach and continuity of care, particularly for underserved populations.

Furthermore, its cost-effectiveness is becoming increasingly recognized [46-48]. Remote protocols may reduce direct and indirect costs, a hypothesis supported by implementation studies showing decreased health care usage when telerehabilitation is adopted as part of routine cancer recovery [12].

Several trials also incorporated patient education and behavioral coaching components [28,34,36] via messaging apps or interactive video calls (eg, WeChat-based multimodal nursing by Zhou et al [37]), highlighting how multidomain interventions can reinforce adherence and long-term engagement.

The sociodemographic and clinical profiles of participants in the included trials provide important context for interpreting the findings of this SR. Most samples consisted of middle-aged women, with mean ages ranging from the early 40s to the mid-60s. Cancer stage was predominantly early (1-2), although several studies included stage 3 cases, while surgical management varied between breast-conserving surgery, mastectomy, and immediate reconstruction. These characteristics reflect the typical population of BC survivors eligible for rehabilitation, but also highlight potential limitations in generalizability.

A relevant aspect emerging from the FIIT analysis concerns exercise intensity. Unlike frequency and duration, which were usually well defined, intensity was rarely standardized across studies. Some trials referred to perceived exertion scales (eg, Borg and RPE) or to international recommendations (eg, American College of Sport Medicine), while others, such as Park et al [35], described progression exclusively in terms of ROM and the transition from passive to active movements with or without dumbbells, without specifying external load. This heterogeneity, and in some cases the total absence of information, hinders direct comparison between protocols and limits clinical reproducibility. From a methodological perspective, the lack of clear criteria for load progression makes it difficult to determine whether observed improvements were due to training frequency, total volume, or actual modulation of intensity. Future telerehabilitation trials should therefore adopt shared and explicit parameters for defining exercise intensity (eg, percentage of one-repetition maximum, RPE, or progressive load adjustments), in order to ensure greater standardization and clinical transferability.

Despite promising results, this review presents some limitations. Only 5 studies were included in the meta-analysis due to heterogeneity in intervention types, duration, and outcome measures. Sample sizes were often small, and blinding was generally not feasible, increasing the risk of performance and detection bias. Furthermore, most studies lacked long-term follow-up data, making it difficult to assess the sustainability of effects beyond 3-6 months. The GRADE approach revealed moderate certainty for QoL but low certainty for other outcomes, primarily due to methodological limitations and imprecision.

One relevant limitation of this meta-analysis is the heterogeneity among the included telerehabilitation interventions in terms of delivery mode (eg, mobile app, videoconferencing, and SMS text messaging), session frequency, intensity, and duration. Additionally, the standard care comparator varied across studies, encompassing different rehabilitation strategies and intensities. This clinical and methodological variability may have influenced the estimated effect sizes. The use of a random-effects model was a deliberate methodological choice to account for and mitigate such heterogeneity. Future trials should aim to harmonize both intervention and comparator protocols to enhance comparability and validity. Furthermore, the underreporting of sociodemographic variables such as education, employment, or ethnicity in many trials limits our ability to fully explore how social context may interact with the effectiveness of telerehabilitation. Future research should systematically report these descriptors to improve external validity and allow stratified analyses.

From a clinical perspective, these findings support the integration of telerehabilitation into standard postoperative care for women after BC surgery. In particular, telerehabilitation appears to be a valuable complement to in-person rehabilitation, especially for patients who face barriers such as distance, limited mobility, or time constraints. To maximize its effectiveness, clinical programs should be individualized, taking into account disease stage, type of surgery, functional status, and patients’ access to and familiarity with digital technologies. From a research perspective, future trials should prioritize the use of standardized and validated outcome measures to ensure comparability across studies and to strengthen the evidence base for meta-analyses. Furthermore, consistent and detailed reporting of sociodemographic variables is needed, as these factors may significantly influence adherence to telerehabilitation and treatment effectiveness, ultimately affecting the external validity of the findings. Addressing these aspects will enhance both the robustness of future evidence and its applicability in clinical practice. Finally, while this review did not include formal cost-effectiveness analyses, the potential economic implications of telerehabilitation deserve attention. Telerehabilitation may reduce indirect costs related to travel, time, and health care resource usage, making it an attractive option for both patients and health care systems. However, these assumptions require confirmation through well-designed economic evaluations alongside clinical trials. Future studies should therefore integrate cost-effectiveness analyses to determine whether the clinical benefits of telerehabilitation are accompanied by sustainable economic advantages.

This SR and meta-analysis confirm that telerehabilitation is an effective and viable approach for postoperative care in women undergoing BC surgery. It significantly improves QoL, reduces pain, and enhances grip strength compared with standard care, with moderate to low certainty of evidence. No significant improvement was observed in the upper limb function, although the direction of effect favors telerehabilitation. The variability across interventions and outcome measures suggests a need for more standardized, high-quality trials. Further research should focus on long-term outcomes, technological enhancements, and cost-effectiveness to support broader implementation.