This systematic review was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 guidelines and the PRISMA-S extension for reporting literature searches in systematic reviews. This was aimed at ensuring transparency and reproducibility throughout our review process.

The search strategy was designed to capture all relevant studies on the effects of aerobic and anaerobic training in post-stroke recovery. Two independent reviewers conducted searches across several electronic databases, namely PubMed, Scopus, Web of Science, and PEDro, spanning from January 2014 to May 2024.

These searches aimed to identify trials evaluating the effects of aerobic and anaerobic training interventions on post-stroke rehabilitation and were performed in accordance with the PRISMA guidelines. Boolean operators and MeSH terms related to “aerobic training”, “anaerobic training”, and “stroke” were employed. The search terms were developed in consultation with a medical librarian and included a combination of MeSH terms and free text words related to “stroke”, “cerebrovascular accident”, “aerobic exercise”, and “anaerobic exercise”. The search strategy for PubMed was: (“aerobic exercise” [MeSH Terms] OR “aerobic training” [All Fields]) OR (“anaerobic exercise” [MeSH Terms] OR “anaerobic training” [All Fields]) AND “stroke” [MeSH Terms]. Initial screening focused on titles and abstracts for eligibility, followed by a full-text review of potential studies. Reference lists of relevant reviews and meta-analyses were also examined. Disagreements were resolved by consensus, with a third party available if needed. The complete search strategy, including all search terms and combinations used, is provided in Appendix A.

The risk of bias in each study was evaluated using the Cochrane Risk of Bias Tool. Our findings indicated a low risk of selection bias in 100% of the studies. Performance bias was high in 6% of the studies, and detection bias was high in 6% of the studies. Attrition bias was low in 97% of the studies, with only 3% showing a high risk. Detailed results of this assessment are provided in Appendix A.

Were included randomized controlled trials (RCTs) and quasi-experimental studies that compared the effects of aerobic training (TAE) or anaerobic training (TAN) on post-stroke recovery outcomes.

The study selection process is detailed in the PRISMA flow diagram ( Figure 1 ), which illustrates the number of records identified, included, and excluded at each stage of the review, along with reasons for exclusions at the full-text review stage.

This systematic review protocol was registered in the International Prospective Register of Systematic Reviews (PROSPERO) under the registration number CRD42024528759, ensuring transparency and reducing the risk of bias publication.

Studies meeting the following criteria were included: 1) human studies, 2) prospective design, 3) written in English, 4) confirmed stroke diagnosis, 5) inclusion of aerobic or anaerobic training, 6) pre- and post-intervention evaluation, 7) at least two experimental groups, and 8) a minimum of five randomized patients.

Following the evaluation of all the articles, the tests were compiled and categorized into the most used tests in the same scope. Five categories were selected based on the frequency of appearance of tests: Functionality, Walking Improvement, Strength, Balance and Cardiorespiratory Capacity. Subsequently, the most utilized tests for both TAE and TAN training were selected from each category.

To assess the different variables and their impact, various tests were employed, each specifically chosen for its relevance to the corresponding analysis. For the Functionality analysis, three tests were utilized: the 10-Meter Walk Test, the Timed Up and Go test, and the Barthel Activities of Daily Living Index (BarthelADL). These tests were selected because they provide valuable insights into an individual’s mobility, functional capacity, and ability to perform daily activities [25] [26] .

To evaluate Walking Improvement, the test employed was the 10-Meter Walk Test. This test was chosen due to its widely recognized and established measures of an individual’s walking ability and overall gait performance, providing valuable information on improvements in walking speed and quality [27] [28] .

For the variable of Strength, the Maximal Isometric Strength (MaxIS) test was employed. This test specifically focuses on measuring an individual’s maximal force production during static muscle contractions. By assessing strength levels, this test provides insights into an individual’s muscular capabilities, helping to evaluate changes and improvements in strength over time [29] .

The evaluation of Balance utilized the Berg Balance Scale, a well-known and widely used assessment tool [30] . This test evaluates an individual’s balance abilities, including both static and dynamic balance, by assessing their ability to perform various tasks while maintaining stability. It provides valuable information on balance control and the potential for improvement [31] .

To assess Cardiorespiratory Capacity, the 6-Minute Walk Test was employed. This test measures an individual’s endurance and cardiovascular fitness by assessing their ability to walk as far as possible within a six-minute timeframe. It is a reliable and commonly used test to evaluate an individual’s aerobic capacity and to monitor changes in their functional status over time [32] .

By utilizing these specific tests for each analysis, a comprehensive evaluation of different variables, such as functionality, waking improvement, strength, balance, and cardiorespiratory capacity, can be conducted. These tests were carefully selected based on their proven validity and reliability in assessing the respective variables, enabling researchers to gather meaningful data and insights for their analysis. To provide clarity on the purpose of each test, the following chart ( Table 1 ) was created:

The following information from selected publications was extracted: 1) characteristics of included subjects (amount, age, sex, time since stroke, stroke etiology, stroke location), 2) study design used, methodological quality (parallel groups/ crossed, PEDro scale), 3) description of the intervention applied (amount and duration of intervention sessions applied, type and intensity of intervention) 4) outcomes (evaluations used, differences detected between groups). The methodological quality of included trials (such as random allocation, baseline comparability, blinding etc.) was assessed using the PEDro scale.

To synthesize the data and perform the meta-analysis, we employed comprehensive statistical methods. The effect sizes for the outcomes within each study were calculated using standardized mean differences (SMDs) for continuous outcomes and risk ratios (RRs) for dichotomous outcomes, along with their 95% confidence intervals (CIs). A random-effects model was applied to account for the expected heterogeneity among the included studies.

We assessed the heterogeneity of the included studies using the I² statistic, where values over 50% were considered indicative of high heterogeneity. To explore the potential sources of this heterogeneity, subgroup analyses were planned based on predefined study characteristics, such as study design, population, and type of intervention.

Sensitivity analyses were conducted to assess the robustness of the findings by excluding studies with a high risk of bias. This allowed us to determine the impact of individual studies on the overall meta-analysis results.

To address potential reporting biases, we employed funnel plots and Egger’s regression test, particularly when the meta-analysis included enough studies. These methods helped to assess the presence of publication bias and other types of reporting biases in the included studies [33] [34] .

The certainty of evidence for each outcome was evaluated using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach. This systematic and transparent methodology considers several factors, including the risk of bias, inconsistency, indirectness, imprecision, and publication bias.

Each outcome was assessed and assigned a GRADE rating ranging from “very low” to “high”, reflecting confidence in the effect estimates. The criteria for downgrading or upgrading the evidence were clearly defined, with the risk of bias assessed for each included study, inconsistency evaluated through the analysis of heterogeneity, and imprecision considered based on the width of confidence intervals and the effect sizes.

By incorporating these detailed methods into our meta-analysis and applying the GRADE approach, we aimed to provide a thorough and reliable synthesis of the evidence, facilitating informed conclusions about the effectiveness of aerobic and anaerobic training interventions in post-stroke rehabilitation [33] [34] .

Given that this study was based on published studies and did not involve collecting primary data, ethical approval was not required.

The analysis of 29 articles, encompassing a total of 1682 individuals, allowed for an examination of the effectiveness of different training interventions in relation to the 10 Meter Walk Test, the Timed Up and Go test, and the BarthelADL scale. The findings demonstrated varying outcomes across these assessments for both TAE and TAN training interventions.

The results from the 10MWT ( Figure 2 ) show that TAN was significantly more effective at improving walking speed than TAE (p < 0.0001). The Chi² value of 5.06 with a degree of freedom of 1 (p = 0.02) suggested a significant association, while the I² value of 80.2% indicated a moderate-to-high degree of heterogeneity among the studies. This suggests that anaerobic training is particularly beneficial for enhancing short-duration functional movements, which are crucial for stroke recovery.

These results highlight the potential of TAN for improving walking efficiency, a critical aspect of functional recovery, while TAE may be better suited for enhancing overall mobility, as reflected in the TUG test results ( Figure 3 ).

When considering the BarthelADL index ( Figure 4 ), neither the TAE nor the TAN groups showed any significant improvements. This lack of significant changes suggests that both training modalities were equally ineffective in impacting the BarthelADL scores.

To provide a comprehensive analysis of the impact of TAE and TAN on various functional outcomes in post-stroke rehabilitation, the following tables offer an overview of the studies investigated. These tables summarize the findings from multiple research articles, highlighting the efficacy of TAE and TAN in improving

specific functional measures. Table 2 presents an overview of the studies investigating TAE and TAN in the 10MWT. Table 3 summarizes the studies investigating TAE and TAN in the TUG test. Table 4 provides an overview of the studies investigating TAE and TAN in the BarthelADL. These tables provide detailed insights into the methodologies, sample sizes, and key outcomes of the studies, allowing for a clearer understanding of the comparative effectiveness of TAE and TAN in post-stroke rehabilitation.

A total of 654 individuals were selected in a total of 14 articles. Figure 2 shows a significant improvement in the 10MWT for TAN participants, with a p-value of < 0.00001. This improvement reflects the ability of TAN to enhance walking speed and quality, which is vital for stroke survivors regaining mobility.

The heterogeneity (I² = 80.2%) indicates variability across studies, suggesting that TAN interventions may have different effects depending on the intensity and frequency of training.

Surprisingly, TAE showed better results than TAN in the MaxIS test, with a p-value of <0.00001. This may be due to TAE’s focus on sustained aerobic activity, which indirectly supports strength development by improving cardiovascular efficiency.

Although TAN is typically associated with strength improvements, this result suggests that TAE can also play a role in enhancing muscular capacity, especially in post-stroke individuals who may benefit from improved oxygen delivery to muscles ( Figure 5 ).

To thoroughly assess the impact of TAE and TAN on various outcomes in post-stroke rehabilitation, a compiled table was made that summarize the studies reviewed. This table consolidates findings from multiple research articles, showcasing the effectiveness of TAE and TAN in enhancing specific functional measures.

Table 5 presents an overview of the studies investigating TAE and TAN in MaxIS. This table provides detailed insights into the methodologies, sample sizes, and key outcomes of the studies, allowing for a clearer understanding of the comparative effectiveness of TAE and TAN in post-stroke rehabilitation.

As demonstrated in Figure 6 , TAN resulted in a significant improvement in BBS scores (p = 0.05), with a Chi² value of 4.75. This highlights the efficacy of anaerobic training in enhancing balance, likely due to its emphasis on strength and coordination.

These findings suggest that TAN interventions may be more effective in addressing balance deficiencies, which are common among stroke survivors and critical for preventing falls.

Table 6 provides an overview of the studies investigating TAE and TAN in the Berg Balance Scale (BBS). This table offers detailed insights into the methodologies, sample sizes, and key outcomes of the studies, enabling a clearer understanding of the comparative effectiveness of TAE and TAN in post-stroke rehabilitation.

Both TAE and TAN demonstrated significant improvements in the 6MWT (p < 0.00001), as seen in Figure 7 . This result suggests that both types of training are effective in enhancing cardiorespiratory fitness, which is essential for long-term recovery.

Given the equivalent benefits of TAE and TAN on cardiorespiratory capacity, rehabilitation programs can integrate both modalities to optimize endurance and overall fitness in stroke survivors.

To assess the impact of TAE and TAN on post-stroke rehabilitation, Table 7 summarizes studies on the 6MWT. It details methodologies, sample sizes, and key outcomes, providing insights into the comparative effectiveness of TAE and TAN.

The superiority of TAN over TAE training in improving performance on the 10MWT can be attributed to the specific physiological adaptations induced by TAN. Anaerobic exercise focuses on high-intensity, short-duration activities, targeting muscular strength, power, and anaerobic capacity [73] . These adaptations are particularly relevant for enhancing walking speed and efficiency, which are key components measured by the 10MWT [74] .

On the other hand, the effectiveness of TAE training in improving TUG test performance can be attributed to its emphasis on aerobic capacity, cardiovascular fitness, and functional mobility [75] . The TUG test primarily assesses the time taken to complete a series of functional tasks, such as standing up, walking, and turning [76] .

However, in the case of the Barthel ADL, both TAE and TAN did not display noteworthy enhancements. This lack of significance suggests that neither training intervention resulted in substantial changes in activities of daily living. The Barthel ADL assesses a broader range of functional abilities beyond walking, and it is possible that different types of interventions or longer training durations may be required to elicit notable improvements in this area.

The varying effectiveness of TAE and TAN training interventions across different outcome measures can be attributed to the specific physiological adaptations targeted by each intervention. The superior performance of TAN in the 10MWT likely stems from its focus on enhancing muscular strength, power, and anaerobic capacity, which are directly related to walking speed and efficiency. On the other hand, the lack of significant results in the Barthel ADL suggests a potential need for alternative interventions or longer training durations to achieve meaningful improvements. The positive effects of TAE training in the TUG test highlight the importance of aerobic fitness and functional mobility for tasks involving standing, walking, and turning. This highlights the fact that functionality is an open capacity, encompassing both aerobic and anaerobic activities.

Integrating both aerobic and anaerobic elements into a training program could potentially enhance the range of physiological adaptations and improve outcomes across various functional measures. This approach may be particularly beneficial for individuals looking to improve both walking speed and efficiency as well as overall functional mobility.

In conclusion, while TAN training demonstrates superior efficacy in the 10MWT due to its focus on anaerobic adaptations, and TAE training excels in the TUG test through its emphasis on aerobic fitness and functional mobility, a combined approach incorporating both training modalities may offer the most comprehensive benefits. This holistic strategy could address the limitations observed in the Barthel ADL and provide a more effective means of improving overall functional capacity. Functionality, therefore, is an open capacity, encompassing both aerobic and anaerobic activities, and optimizing training programs to include both elements may yield the best results for enhancing various aspects of physical performance and daily living activities.

The observed significance of TAN in improving walking performance, as measured by 10MWT, can be attributed to several factors. Firstly, TAN protocols typically emphasize high-intensity exercises, targeting the development of muscular strength and power [77] . These exercises primarily focus on the lower extremities, which are crucial for walking and gait mechanics. By engaging in TAN, individuals can enhance their muscular strength and power, leading to improvements in walking ability and performance as reflected in the 10MWT results [74] .

Secondly, TAN often involves exercises that promote neuromuscular coordination and motor control [78] . The precise and coordinated movements required during TAN can enhance the communication and synchronization between the muscles, nerves, and central nervous system [79] . As a result, individuals who undergo TAN may experience improved motor control and movement efficiency, leading to better performance in the 10MWT.

Furthermore, the anaerobic nature of the training stimulates adaptations in the anaerobic energy system, enabling individuals to sustain higher levels of effort during walking. The improved anaerobic capacity contributes to enhanced walking speed, endurance, and overall performance, all of which are assessed in the 10MWT. For this population, displacements occur over short distances, not requiring too much aerobic capacity. This means that the anaerobic adaptations induced by the training are particularly beneficial, as they are more relevant to the short bursts of effort typically required in their daily activities. Consequently, the emphasis on anaerobic training aligns well with the functional demands of their environment, leading to significant improvements in their walking performance and overall mobility.

The observed heterogeneity among the studies included in the analysis could be attributed to variations in the duration, intensity, and specific protocols of the TAN interventions. Differences in participant characteristics, such as age, fitness level, and baseline walking abilities, may also contribute to the heterogeneity. Additionally, variations in study design, measurement tools, and statistical approaches used across the included articles could influence the degree of heterogeneity observed.

In summary, the significant association between TAN and Walking Improvement, specifically demonstrated by the 10MWT results, can be attributed to the development of muscular strength and power, enhanced neuromuscular coordination, and improved anaerobic capacity.

The unexpected superiority of TAE in promoting strength increase, as measured by the MaxIS, can be attributed to several factors, especially in individuals recovering from a stroke. TAE interventions typically involve sustained aerobic activities that target cardiovascular fitness, endurance, and whole-body muscle engagement. These exercises, such as walking, cycling, or swimming, stimulate the cardiovascular system, leading to improved oxygen delivery and enhanced energy production within the muscles, especially in the lower limbs [80] .

Additionally, for strength-related activities, the body primarily uses ATP reserves rather than glycogen or fat. This is particularly relevant for stroke survivors, whose muscles may have reduced efficiency and strength. Aerobic exercises help improve the overall metabolic efficiency of the body, ensuring that ATP is produced and utilized more effectively during muscle contractions. This increased efficiency in ATP utilization can lead to significant strength gains, even though the primary focus of TAE is on aerobic conditioning.

In the context of stroke rehabilitation, enhanced oxygen delivery and improved energy production within the muscles due to TAE can help rebuild muscle strength and endurance. The emphasis on sustained aerobic activities also aids in overall cardiovascular health, which is crucial for stroke survivors. By improving the efficiency of the anaerobic energy system and ensuring that ATP reserves are readily available, TAE helps stroke survivors increase their muscle strength and perform daily activities with greater ease.

Therefore, while TAE primarily targets aerobic capacity, it also indirectly supports strength gains by optimizing the body’s energy systems. This dual benefit makes TAE a valuable component of stroke rehabilitation, contributing to improved functional outcomes and quality of life for stroke survivors.

It is important to consider the heterogeneity observed among the studies included, which may contribute to the variations in the strength increase outcomes between TAE and TAN interventions. Variations in the duration, intensity, and specific protocols of the exercise interventions, as well as differences in participant characteristics (e.g., age, time since stroke, initial strength levels), can influence the heterogeneity observed.

Furthermore, the dissimilarities in control group characteristics between TAE and TAN studies may have influenced the outcomes and contributed to the apparent inferiority of TAN. In TAE studies, comparisons with well-established rehabilitation methods, such as massage and physiotherapy, provided a robust baseline for evaluating the aerobic training’s efficacy. Additionally, the incorporation of passive mobilization in the control group allowed for a comprehensive assessment of the TAE’s unique benefits.

On the other hand, TAN studies utilized control groups with combined interventions, potentially leading to confounding factors in the evaluation of anaerobic training alone, such as task-related activities. Moreover, the comparison of TAN on lower limbs and upper limbs plus oriented tasks in the control group may have introduced biased results, as upper limb tasks differ significantly from lower limb-focused training in stroke rehabilitation [81] . Furthermore, the contrast between TAN in isokinetic settings and TAN performed at home introduced variations in training environment and supervision, possibly affecting the intervention’s effectiveness.

Hence, the diverse and less conventional control groups used in TAN studies might have contributed to the observed disparities in outcomes compared to TAE interventions. As such, careful consideration of control group selection and standardization is crucial for future research investigating the efficacy of TAN in stroke rehabilitation.

Another potential explanation for the superior strength increase observed in the TAE group could be the focus on functional movements and activities that closely mimic daily tasks. TAE interventions often involve exercises that simulate real-life movements, which are directly relevant to the functional needs of individuals who have had a stroke. By specifically targeting these functional movements, TAE training may result in more significant improvements in overall strength. and functional capacity.

In conclusion, the unexpected finding of TAE demonstrates a significant response and superior strength increase compared to TAN in individuals who had a stroke, as measured by the MaxIS test, may be attributed to one or more factors cited previously.

The lack of significant improvement in strength observed in the TAN can be attributed to several potential factors. Firstly, TAN typically emphasizes high-intensity, short-duration activities that target muscular strength and power [74] . However, in the context of individuals who have had a stroke, it is possible that the intensity of the TAN prescribed in the interventions was not sufficient to elicit substantial strength gains. Stroke survivors may have unique physiological considerations and limitations that could affect their ability to tolerate and benefit from high-intensity TAN [82] . Additionally, factors such as muscle weakness, reduced motor control, and altered muscle activation patterns resulting from the stroke may impact the response to TAN [83] .

The observed significant response for TAN in improving balance, as measured by the BBS, can be attributed to several factors. First, TAN interventions generally involve exercises that focus on enhancing muscular strength, power, and explosive movements [84] . These exercises often include resistance training or plyometric exercises. Such activities can directly target the muscles involved in maintaining balance and stability, leading to improved muscular control and coordination [85] .

Furthermore, TAN commonly incorporates movements that challenge balance and proprioception [86] . By engaging in exercises that require dynamic stability, individuals can improve their ability to maintain equilibrium and control body movements [87] . The focus on rapid changes in direction, quick transitions, and sudden stops during TAN exercises can enhance proprioceptive feedback, body awareness, and postural control, which are critical elements for maintaining balance [88] .

Another possible explanation for the significant response favoring TAN is the principle of specificity in exercise training [89] . TAN interventions often involve movements and exercises that closely resemble activities performed in daily life or specific sports. This specificity of training may lead to better transfer of skills to balance-related tasks assessed by the Berg Balance Scale.

However, it is important to consider the heterogeneity observed among the included studies, which may influence the overall response and the interpretation of the results [90] . Variations in the types and intensity of exercises, duration of interventions, participant characteristics, and outcome measures used across the studies could contribute to the observed heterogeneity.

In conclusion, the significant response favoring TAN over TAE in the comparison of their effects on balance improvement, as measured by the BBS, can be attributed to the focus on muscular strength, power, explosive movements, dynamic stability, and the principle of specificity in training. The incorporation of exercises that challenge balance, proprioception, and postural control likely contributes to the observed improvements.

The absence of significant differences between TAN and TAE interventions in improving Cardiorespiratory Capacity, as measured by the 6MWT, can be attributed to several factors. Firstly, both TAN and TAE interventions typically involve exercises that target the cardiovascular and respiratory systems [90] . Aerobic exercise, regardless of the specific modality, increases heart and respiratory rate, also oxygen consumption, thereby improving Cardiorespiratory Capacity [90] .

Secondly, the 6MWT primarily assesses an individual’s endurance and functional capacity to perform activities involving walking. Both TAN and TAE interventions can enhance aerobic fitness, muscular endurance, and overall physical conditioning, which positively influence an individual’s performance in endurance-based tasks such as the 6MWT [91] . Therefore, the significant results observed in both TAN and TAE groups can be attributed to the benefits of aerobic exercise on Cardiorespiratory Capacity.

It is important to consider that the lack of significant differences between the two interventions may also be influenced by the low degree of heterogeneity observed among the studies included. The consistency in exercise protocols, intervention durations, participant characteristics, and outcome measures across the studies might contribute to the absence of significant variation between TAN and TAE groups.

In conclusion, the significant results observed in both TAN and TAE interventions without significant differences between the groups in improving Cardiorespiratory Capacity, as measured by the 6MWT, can be attributed to the shared focus on aerobic exercise targeting the cardiovascular and respiratory systems. The 6MWT serves as a suitable measure for assessing an individual’s endurance and functional capacity related to walking activities.

In conclusion, this study has provided valuable insights into the effectiveness of different training modalities across various functional categories. Five key points emerge from our findings:

Regarding the category of Functionality, it is easy to determine that none of the training modality is more efficient than the other. Our results demonstrated that for the 10MWT, TAN appeared to be more efficient, while for the TUG test, TAE exhibited greater efficacy. Interestingly, no significant differences were observed for either training modality in the BarthelADL. This highlights the importance of tailoring the choice of training to individual patient needs and therapeutic objectives [92] .

In the category of Walking Improvement, our findings suggest that the 10MWT was more effective when coupled with TAN. This underscores the potential benefits of anaerobic training in enhancing gait performance for individuals with the conditions in this investigation.

For the Strength category, our results indicated that MaxIS testing favored TAE. However, this outcome is accompanied by several justifications, contravening initial expectations that TAN would be more efficient for strength improvement. Thus, it is imperative not to definitively conclude that TAE is superior for enhancing strength, as the influence of training methodology and individual participant characteristics necessitates further investigation [93] .

Within the Balance category, the BBS emerged as the more effective tool when combined with TAN. This suggests that TAN may hold promise for improving balance in individuals with the studied conditions [94] .

Finally, in the category of Cardiorespiratory Capacity, both TAE and TAN modalities demonstrated efficacy. This indicates that both approaches can be considered viable options for enhancing Cardiorespiratory Capacity in patients with the conditions under examination [95] .

In summary, this research has shed light on the nuanced effects of different training modalities across various functional domains. It is imperative to acknowledge the complexity of these relationships and to consider individual patient characteristics when selecting the most appropriate training approach. Further research is warranted to deepen our understanding of these findings and to identify the most effective therapeutic strategies for specific clinical contexts.

Given the overall findings, it is noteworthy that TAN consistently showed significant advantages in key performance metrics such as the 10MWT for walking improvement and the BBS for balance. These outcomes suggest that, while both TAE and TAN have their merits, TAN may offer superior benefits in enhancing functional performance, particularly in areas critical to daily living and mobility. This indirect indication of TAN’s superiority underscores the potential for anaerobic training to play a pivotal role in the rehabilitation and improvement of functional capacities in individuals with the conditions studied.

The complete search strategy, including all search terms and combinations used, is detailed below.

We conducted a comprehensive literature search across four databases: PubMed, Scopus, Web of Science, and PEDro. The search was restricted to articles published from January 2014 to May 2024. For each database, the search terms and combinations were developed in consultation with a medical librarian, employing a combination of MeSH terms and free-text words related to “stroke,” “aerobic training,” and “anaerobic training.” Boolean operators were utilized to refine the search strategy.

In PubMed, we used the following search terms: (“aerobic exercise” [MeSH Terms] OR “aerobic training” [All Fields]) AND (“anaerobic exercise” [MeSH Terms] OR “anaerobic training” [All Fields]) AND “stroke” [MeSH Terms].

In Scopus, the search was conducted using: TITLE-ABS-KEY (“aerobic training” OR “aerobic exercise”) AND (“anaerobic training” OR “anaerobic exercise”) AND (“stroke” OR “cerebrovascular accident”).

In Web of Science, the search terms were: TS = (“aerobic training” OR “aerobic exercise”) AND TS = (“anaerobic training” OR “anaerobic exercise”) AND TS = (“stroke” OR “cerebrovascular accident”).

In PEDro, we searched using: Abstract/Title: “aerobic training” OR “aerobic exercise” AND “anaerobic training” OR “anaerobic exercise” AND “stroke”

The inclusion criteria for the studies were as follows:

The study selection process involved an initial screening of titles and abstracts to determine eligibility, followed by a full-text review of potential studies. We also examined the reference lists of relevant reviews and meta-analyses. Any disagreements were resolved by consensus, with a third party available if needed.

Data extraction included details on participant characteristics (number, age, sex, time since stroke, stroke etiology, stroke location), study design and methodological quality (parallel groups/crossed, PEDro scale), intervention details (number and duration of sessions, type and intensity of intervention), and outcomes (evaluations used, differences detected between groups).

The risk of bias in each study was evaluated using the Cochrane Risk of Bias Tool. Our findings indicated a low risk of selection bias in 100% of the studies. Performance bias was high in 6% of the studies, and detection bias was high in 6% of the studies. Attrition bias was low in 97% of the studies, with only 3% showing a high risk.

Meta-analysis and statistical analysis were conducted with effect sizes for outcomes within each study calculated using standardized mean differences (SMDs) for continuous outcomes and risk ratios (RRs) for dichotomous outcomes, with 95% confidence intervals (CIs). A random-effects model was applied to account for expected heterogeneity. Heterogeneity among studies was assessed and addressed using the I² statistic, with values over 50% considered indicative of high heterogeneity. Sensitivity analyses were conducted by excluding studies with a high risk of bias. Funnel plots and Egger’s regression test were used to assess reporting biases.

The statistical analysis was conducted using the Revman 5 software.

The certainty of evidence for each outcome was evaluated using the GRADE approach, with outcomes assessed and assigned a GRADE rating from “very low” to “high”.

This systematic review was conducted in accordance with the PRISMA 2020 guidelines and the PRISMA-S extension for reporting literature searches in systematic reviews. The review protocol was registered in the International Prospective Register of Systematic Reviews (PROSPERO) under the registration number CRD42024528759.