The protocol has been registered with the International Registry of Prospective Systematic Reviews (PROSPEROID: CRD42023395515). A systematic literature search was conducted according to PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines (18).

Two researchers, W.Y.F. and H.L.L., independently conducted searches in four databases: PubMed, Embase, Web of Science, and the Cochrane Library. A systematic search was performed using the terms “aerobic exercise” AND “prediabetes.” (For a comprehensive list of search terms, please refer to Supplementary Table S1 ). Studies published from the inception of the databases up to January 7, 2023, were included. Additionally, we screened the references of relevant studies to supplement the search results from these databases. No language restrictions were applied in this study.

The inclusion criteria were determined based on PICOS (population, intervention, control, outcome, and study design) criteria (18). The study participants included individuals with prediabetes, including those with IGT and IFG, without any gender or age restrictions. The intervention must be supervised aerobic exercise training. For comparison, the control group should consist of participants without any interventions (referred to as “blank controls”). We included randomized controlled trials (RCTs) that compared aerobic exercise interventions with usual care (no exercise). If the control group implemented a healthy diet, the intervention group also needed to incorporate the same dietary control to ensure that aerobic exercise was the sole intervention. If multiple interventions were available, only data from the aerobic and control groups were extracted. At least one primary outcome was reported (BMI, FBG level, any change in 2hPG or HbA1c).

The criteria for diagnosing prediabetes are as follows: 1) FBG ranging from 100 mg/dL (5.6 mmol/L) to 125 mg/dL (6.9 mmol/L) (IFG). 2) Blood glucose concentration of 140 mg/dL (7.8mmol/L) to 199 mg/dL (11.0 mmol/L) two hours after intake of 75g oral glucose (IGT). 3) HbA1c ranging from 5.7% to 6.4% (4).

Participants who had diabetes, cardiovascular disease, were pregnant, had major musculoskeletal problems, or any other medical conditions that would substantially impact the exercise interventions were excluded from the study. Additionally, individuals who were taking hypoglycemic or lipid-lowering medications were also excluded.

Two investigators independently assessed search results and extracted data using a standardized data extraction format. The following data were extracted: first author, year of publication, study region, study design, age and sex of the included patients, proportions, baseline BMI, type, intensity, frequency, and duration of the aerobic interventions. Any disagreements were resolved through consensus. If a consensus could not be reached, a third reviewer (H.L.W) was consulted.

Two reviewers (Y.F.W. and H.L.L) independently rated the risk of bias of the RCTs using the revised Cochrane risk of bias, version 2 (RoB 2) tool. The risk of bias was categorized as high, low, or having some concerns. In cases where there was disagreement between the two researchers, a third party was involved to facilitate discussion and reach a consensus.

Stata15.1 software was used for analyzing the continuous outcome indicators of the included studies. The weighted mean difference (WMD) statistic was used to combine effect sizes, and a 95% confidence interval (CI) was provided for each effect size. Effect sizes for the primary outcomes were expressed as WMD. Cochran’s Q and I² were employed to assess study heterogeneity, with I² <25% indicating low heterogeneity, 25% to 50% indicating moderate heterogeneity, and > 50% indicating high heterogeneity. The fixed effect model was applied for low and moderate heterogeneity, while the random effect model was used for high heterogeneity. Subgroup analysis was conducted to examine the influence of different time periods on the exercise and its outcomes. To explore the potential publication bias, Egger’s test was employed. The significance level was set at 0.05.

A total of 4283 papers were identified through database searches and other sources. After removing duplicates, 3222 papers remained. Among these, 3117 were excluded based on their titles and abstracts, leaving 105 papers for full-text examination. Finally, 10 RCTs were included in the review and meta-analysis (14, 15, 19–26) ( Figure 1 ).

The Cochrane Risk of Bias assessment (ROB2) indicated that eight of the studies had a low risk of bias, while the remaining two studies were found to have some concerns regarding bias ( Supplementary Figure S1 ).

A total of 815 patients with prediabetes were included in the analysis, derived from the 10 articles that were ultimately selected for the study. Among the studies included in the analysis, the average (SD) age of the participants was 56.1 (5.1) years, with males comprising 39.2% of the subjects. The BMI of the participants was provided in nine research, and the type, intensity, frequency, and duration of aerobic exercise were reported in all ten investigations.

All aerobic training was performed under supervised conditions. Modes of training include aerobic dance (14, 19–21, 24, 25), treadmill running (15, 26), walking (15, 22), and mixed aerobic exercise (23). In nine studies, exercise intensity was reported as maximum heart rate (HRmax), maximum heart rate reserve (HRRmax), or heart rate reserve (HRR) with a median intensity and range of 65% (50%-75%). Another study (26) reported on high-intensity interval training (HIIT): 4 x 4-minute intervals at 90% of maximum heart rate ( Supplementary Table S2 ).

Seven out of the ten included studies reported BMI indicators (14, 15, 20–23, 25). The combined analysis revealed a significant difference in BMI between the aerobic exercise group and the control group (WMD=-1.44 kg/m², 95%CI -1.89, -0.98, I² = 44%, P<0.001). To explore the potential impact of exercise at different stages, subgroup analysis was performed based on various training durations. After combining and analyzing data from all four intervention periods (3, 6, 12, and 24 months), significant differences in BMI levels before and after the intervention were identified (3 months: WMD=-1.75 kg/m², 95%CI -2.79, -0.71, I² = 79.5%, P=0.027; 6 months: WMD=-0.88 kg/m², 95%CI -1.54, -0.23, I² = 0.0%, P=0.881; 12 months: WMD=-2.54 kg/m², 95%CI -3.95, -1.13, I² = 29.6%, P=0.233; 24 months: WMD=-1.86 kg/m², 95%CI -2.82, -0.90, I² = 0.0%, P=1.000) ( Figure 2 ). The Egger’s test showed no evidence of publication bias in this result (t=-0.20, P=0.850) ( Supplementary Figure S2 ).

Nine of the studies included in the analysis provided data on fasting blood glucose (FBG) levels (14, 15, 19–22, 24–26). The pooled results revealed a significant difference in FBG levels between the aerobic exercise group and the control group (WMD=-0.51mmol/L, 95%CI -0.70 -0.32, I² = 98.6%, P<0.001). Significant differences in FBG levels were found during subgroup analysis, specifically when data from three intervention periods (3 and 12 months) were combined and examined (3 months: WMD=-0.53mmol/L, 95% CI -0.72 -0.35, I² = 66.7%, P=0.049; 12 months: WMD=-0.60mmol/L, 95%CI -0.84 -0.36, I² = 64.0%, P=0.040). In contrast, there was no statistically significant difference in FBG between the aerobic exercise group and the control group following 6 and 24 months of intervention (6 months: WMD=-0.33mmol/L, 95%CI -0.72 0.07, I² = 99.3%, P=0.000; 24 months: WMD=-0.72mmol/L, 95%CI -1.49 0.06, I² = 96.7%, P=0.000) ( Figure 3 ). The results of the Egger’s test suggested no publication bias in this finding, as indicated by a non-significant test statistic (t=0.01, P=0.995). ( Supplementary Figure S3 ).

Six studies among those identified in the analysis reported outcome measures of 2-hour postprandial glucose (2hPG) levels (14, 19–21, 24, 25). The combined results demonstrated a significant difference in 2hPG levels between the aerobic exercise group and the control group (WMD=-0.76mmol/L, 95% CI -1.14 -0.38, I² = 97.8%, P<0.001). Statistically significant differences in 2hPG levels were found when analyzing the combined data from all four intervention periods (3, 6, and 12 months), indicating changes in 2hPG levels before and after the intervention (3 months: WMD=-1.02mmol/L, 95%CI -1.57 -0.47; 6 months: WMD=-0.63mmol/L, 95% CI -1.13 -0.13, I² = 97.8%, P=0.000; 12 months: WMD=-0.96mmol/L, 95%CI -1.66 -0.26, I² = 77.9%, P=0.004). Nonetheless, no statistically significant difference was observed in 2hPG levels between the aerobic exercise and control groups after 24 months of intervention (WMD=-0.50mmol/L, 95%CI -1.03 0.03, I² = 66.4%, P=0.051) ( Figure 4 ). The Egger’s test indicated no indication of publication bias in this result (t=0.29, P=0.779) ( Supplementary Figure S4 ).

Among the ten studies included in the analysis, seven studies reported outcome measures of HbA1c levels (14, 15, 19, 21, 23–25). The pooled results indicated a statistically significant difference in HbA1c levels between the aerobic exercise group and the control group (WMD=-0.34%, 95CI -0.45 -0.23, I ² = 96.5%, P<0.001). Significant statistical differences in HbA1c levels were detected when analyzing the combined data from all four intervention periods (3, 6, and 12 months) and comparing the measurements before and after the intervention (3 months: WMD=-0.52%, 95%CI -1.01 -0.04, I ² = 90.1%, P=0.000; 6months: WMD=-0.28%, 95%CI -0.52 -0.05, I ² = 85.8%, P=0.000; 12 months: WMD=-0.28%, 95%CI -0.52 -0.05, I ² = 85.8%, P=0.000). After 24 months of intervention, no statistically significant difference was found in HbA1c levels between the aerobic exercise group and the control group (WMD=-0.43%, 95%CI -0.90 0.05, I² = 90.2%, P=0.000) ( Figure 5 ). The result was not influenced by publication bias, as indicated by Egger’s test (t=1.00, P=0.337) ( Supplementary Figure S5 ).

Sensitivity analysis was conducted on a case-by-case basis, and the results remained stable and reliable after the analysis ( Supplementary Figures S6 - S9 ).

Our systematic review and meta-analysis provide an overview of the effectiveness of aerobic exercise therapy in patients with prediabetes. Although the studies included in this research have some limitations, the low-risk diversity displayed in the outcomes implies that the findings are reliable. In the included studies, the type of aerobic exercise chosen was suitable for the general population, promoting long-term adherence and providing additional benefits. Seven of the studies had training intensities ranging from 60% to 70%, and the training was supervised by professionally trained personnel, ensuring high compliance. Aerobic dance was the most frequently used exercise modality and gained high popularity among elderly Chinese women. In recent years, exercise therapy has been extensively demonstrated as a cost-effective, user-friendly, safe, and devoid-of-side-effects intervention for preventing or treating diabetes, distinguishing it from other approaches (35). Additionally, our study encompassed more outcome indicators in comparison to the meta-analysis conducted by Hrubeniuk et al. (17). A subgroup analysis was conducted, capturing the effects of various intervention durations on outcome indicators. We also included a larger number of subjects compared to the meta-analysis conducted by Huang et al. (27) to comprehensively assess the true impact of different durations of continuous aerobic exercise on patients with prediabetes. Yang (32) et al.’s meta-analysis observed that aerobic training was more effective in decreasing BMI in diabetic patients than resistance training. This finding provides valuable insights that can be applied to designing training programs for prediabetic patients. In the future, we also aim to explore the effect of combined aerobic and resistance training on prediabetic patients.

While our findings indicate that aerobic exercise training may lead to significant improvements in indicators among individuals with prediabetes, the results are subject to some degree of uncertainty due to the presence of heterogeneity, inaccuracies, inconsistencies, and the inability to implement blinded exercise interventions across various studies. In the conducted meta-analysis, aerobic exercise interventions were grouped together. It is important to acknowledge that the observed heterogeneity across the studies may be attributed to variations in the nature, duration, and intensity of the exercise programs. Our research did not differentiate specifically between patients with IGT and IFG. The number of primary literature sources included in our study was limited, and the conclusions drawn from the meta-analysis may be somewhat questionable. In the future, more high-quality studies are needed to obtain more precise results. Additionally, certain indicators such as HS-CRP, which are associated with the incidence of diabetes, could not be included in the meta-analysis due to a lack of available primary studies for analysis. Our study only considered aerobic training studies and excluded those that incorporated resistance training in the control group. However, Grøntved et al. (36) conducted a prospective cohort study that revealed a reduction of approximately 35% and 50% in the risk for men who performed resistance training or aerobic exercise for at least 150 minutes per week, respectively. Incorporating resistance training with aerobic exercise seems to be beneficial in averting T2D. For better generalization, it is recommended to select exercises that have a broader appeal when incorporating resistance training.

FBG, 2hPG, and HbA1c are three distinct indicators related to glucose levels, each reflecting different time periods and factors that influence them. The correlation between these indicators can lead to similarities or differences in the analysis results. Therefore, when assessing long-term glycemic control in diabetic patients, it is essential to measure these indicators simultaneously to obtain more comprehensive and accurate information. Despite the limitations of the studies included, this systematic review and meta-analysis offer an informative overview of the effectiveness of aerobic exercise training at various durations in patients with prediabetes. This summary can serve as a valuable resource for clinical discussions and decision-making processes.

For patients with prediabetes, BMI and HbA1c are crucial indicators in assessing the effects of exercise. It is essential for healthcare professionals to determine the ideal BMI range for older adults, considering that overweight and obese patients may require significant weight loss to decrease their risk of developing diabetes and reverse prediabetes symptoms. Previous prospective cohort studies have suggested that lifestyle modifications, such as exercise and dietary interventions, can effectively prevent and lower the occurrence of T2D in individuals with prediabetes. Furthermore, interventions solely focused on aerobic exercise have been found to reduce the risk of T2D by 46% (37). Based on the findings of this study, exercise training for approximately one year may be recommended for prediabetic patients. However, if there is no significant improvement in the indicators after one year, healthcare professionals should consider additional interventions, such as medication or other treatments.

Our study included middle-aged and elderly participants with a mean age of 56.1 years. Considering the rising prevalence of diabetes in younger populations, future research should explore the impact of exercise interventions on children and adolescents with diabetes and prediabetes (38). This necessitates additional high-quality RCTs to evaluate the impact of various training durations on diabetes incidence, determine the optimal exercise intensity, duration, and frequency, explore the potential benefits of combining aerobic and resistance training, and compare the advantages of different training types for individuals with prediabetes. Implementing exercise interventions for prediabetic patients to improve their indicators and prevent the progression of diabetes can significantly alleviate the global burden of this disease on public health.