To ensure the reproducibility of the study, the literature search was conducted using the following databases: PubMed, ProQuest, Scopus, China National Knowledge Infrastructure (CNKI), VIP, and Wanfang. Based on the following core considerations: ① the impact of timeliness on the quality of research, ② the progress of methodology, ③ the timeliness of policy and practice, ④ information overload and feasibility, the search period ranged from January 1, 2020, to August 27, 2024, with a language restriction to English. The search strategy has been registered in the International Prospective Register of Systematic Reviews (PROSPERO, CRD42024584877). The keywords used included, but were not limited to: “prenatal exercise”, “exercise intervention during pregnancy”, “caesarean section rate”, “randomized controlled trial”, “pregnancy outcomes”, as well as their synonyms and related combinations, such as “antenatal physical activity”, “cesarean delivery outcomes”. In addition, Boolean operators (AND, OR) were used to optimize the search formula, with specific examples as follows:

- “prenatal exercise AND pregnancy outcomes”

- “exercise intervention OR physical activity AND caesarean section rate”

- “randomized controlled trial AND antenatal physical activity”

To ensure comprehensive retrieval, we also screened the reference lists of relevant literature to obtain additional studies.

To clearly present the criteria, the inclusion and exclusion criteria are listed as follows:

Inclusion Criteria

① The study subjects are healthy pregnant women, with single pregnancy, cephalic presentation, normal pelvic and fetal positions, and no separation of the symphysis pubis;

② The intervention measures are prenatal exercises, including physical activities (such as running, swimming, yoga) and prenatal labor training (such as birth ball exercises, pelvic floor muscle training, Lamaze breathing training);

③ At least one outcome indicator is assessed, such as the rate of CS;

④ Sufficient data are reported, including sample size, mean, and standard deviation;

⑤ The study type is a RCT.

Exclusion Criteria

① The study subjects are unable to exercise due to twin pregnancy, breech presentation, placenta previa, PTB, miscarriage, cervical relaxation, separation of the symphysis pubis, other severe medical comorbidities (such as heart, kidney, liver diseases), or have a history of cognitive impairment and mental illness;

② The intervention measures are not prenatal exercises, or are not the only intervention;

③ There are no outcome indicators to assess;

④ Insufficient data are available;

⑤ The study is not a RCT (such as case reports, reviews, animal experiments, comments);

⑥ The study has issues with insufficient data, language barriers, others.

Two reviewers independently screened the literature and extracted the following data: first author, publication date, country, sample size, participant characteristics (such as age, mean, and standard deviation), intervention measures, and main outcome indicators (such as the rate of CS). If there was a disagreement, a third reviewer made the decision.

The assessment of study bias risk was conducted using the Cochrane Collaboration Risk of Bias Tool described in the Cochrane Intervention Systematic Review Manual (version 5.1.0, https://handbook.cochrane.org/), including the following seven dimensions:

① Random sequence generation (selection bias);

② Allocation concealment (selection bias);

③ Blinding of participants and personnel (performance bias);

④ Blinding of outcome assessment (detection bias);

⑤ Incomplete outcome data (attrition bias);

⑥ Selective reporting (reporting bias);

⑦ Other biases (such as funding support and conflicts of interest of researchers).

Each risk of bias was assessed as “low risk”, “unclear”, or “high risk”.

Data analysis was performed using Review Manager 5.4 software (https://revman.cochrane.org/). The significance of the p-value is defined as if p $\le$ 0.05, the result is considered statistically significant. If p $>$ 0.05, the results are not considered statistically significant. Heterogeneity was assessed using the Q test and I² statistic:

The Q test was used to determine the significance of heterogeneity, with p 0.1 indicating the presence of heterogeneity;

The I² statistic quantitatively assessed the degree of heterogeneity:

I² 25%: low heterogeneity;

25% $\le$ I² $\le$ 50%: moderate heterogeneity;

I² $>$50%: high heterogeneity.

For studies with high heterogeneity, a random-effects model was used for analysis to reduce the impact of inter-study differences. For missing data, priority was given to using full data analysis; if the missing proportion was small (10%), methods such as the expectation-maximization algorithm were used to input data.

The search process followed a systematic approach, resulting in a total of 243 articles initially retrieved from 6 databases (PubMed, ProQuest, Scopus, CNKI, Wanfang, and Weipu). After removing 110 duplicates using Endnote (X9, Clarivate Analytics, Seattle, WA, USA), 133 unique articles remained. During the title and abstract screening, 113 articles were excluded for the following reasons: they were irrelevant to prenatal exercise or cesarean delivery outcomes, focused on unrelated topics (e.g., non-pregnant populations), or lacked primary research data.

The full-text of 20 remaining articles was reviewed in detail. At this stage, 9 articles were excluded based on the following criteria:

① Non-randomized studies: 5 articles did not use RCT designs.

② Inadequate intervention details: 2 studies did not provide sufficient information on the prenatal exercise protocols used.

③ Lack of relevant outcomes: 2 studies did not report CS rates or comparable delivery outcomes.

Ultimately, 11 RCTs meeting all inclusion criteria were included in the meta-analysis [7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17]. These studies involved a total of 2553 pregnant women, with 1152 in the exercise groups and 1401 in the control groups (Table 1, Ref. [7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17]).

This stepwise filtering process is illustrated in the updated Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) flow diagram (Fig. 1), providing a clear visual summary of the study selection stages.

The risk of bias tool was used to display detailed information on the quality assessment of the studies. Of all the selected studies, 5 described the way in which randomization was generated and were rated as low risk. Another 6 studies only mentioned the method of randomization without specifying the way in which randomization was generated, and were rated as unclear. A small number of studies had poor blinding of study participants and patients, and poor blinding of outcome measures. Overall, all included studies were considered to be at low risk of bias (Fig. 2).

The results of the study showed significant heterogeneity between studies (I² = 68%, p 0.05, Fig. 3). Because of the significant heterogeneity, a random effects model was used. The results of the analysis showed that performing prenatal exercise reduced the rate of CS compared to the group without prenatal exercise (relative risk (RR): 0.66, 95% confidence intervals (CI): 0.52–0.85, Z = 3.31, p 0.05, Fig. 3).

All studies were divided into two subgroups according to the type of exercise: physical activity during pregnancy (e.g., running, swimming, yoga, others) and antenatal labor training (e.g., birthing ball exercises, pelvic floor exercises, Lamaze breathing exercises, others), and examined whether a combination of physical activity and antenatal labor training still reduces the rate of CS.

The results showed that there was a statistically significant difference in the cesarean delivery rate between the exercise and control groups within the subgroup that performed physical activity [RR = 0.71, 95% CI: 0.54–0.92, p 0.05, with high heterogeneity (I² = 71%)]. There was also a statistically significant difference in cesarean delivery rates between the exercise and control groups within the subgroup that underwent antenatal labor training [RR = 0.55, 95% CI: 0.41–0.74, p 0.05, no heterogeneity between included studies (I² = 0%)]. At the same time, the difference in cesarean delivery rates between the exercise and control groups when combining physical activity and antenatal labor training was statistically significant [RR = 0.61, 95% CI: 0.42–0.88, p 0.05, no heterogeneity between included studies (I² = 0%)]. In the single type of exercise subgroup, the difference of CS rate between the exercise group and the control group was also statistically significant [RR = 0.67, 95% CI: 0.51–0.90, p 0.05, with high heterogeneity (I² = 74%)]. This suggests that physical activity during pregnancy combined with antenatal labor training remains effective in reducing CS rates (Figs. 4,5).

According to the recommendations of the Cochrane Handbook, if the funnel plot is used for publication bias evaluation, the number of studies included in the meta-analysis of this indicator should not be less than 10 papers. Otherwise, because the number of studies included in this indicator is too small, the testing ability of the funnel plot will be reduced. As such, the truth of asymmetry cannot be judged. A total of 11 studies were included in this meta-analysis, and all of them were analyzed by publication bias funnel plot. The results showed more symmetry (Fig. 6), suggesting that it is less likely to have bias.

The meta-analysis included 11 RCTs with a total of 2553 pregnant women, 1152 in the exercise group and 1401 in the control group. The results showed that prenatal exercises significantly reduced the rate of CSs compared to the control group (RR: 0.66, 95% CI: 0.52–0.85, p 0.05, Fig. 3). These findings indicate that prenatal exercise is associated with a 34% reduction in the likelihood of CS (Fig. 7). Heterogeneity analysis showed moderate heterogeneity (I² = 68%; p 0.05), which was further explored through subgroup analyses (Fig. 5).

Sensitivity analyses were conducted by sequentially removing individual studies to assess their impact on overall heterogeneity and pooled effect sizes. The results of meta-analysis after both eliminating Virginia Y. Watkins et al. [13] and Yogyata Wadhwa et al. [14] showed that the heterogeneity decreased to 0% (RR: 0.68, I² = 0%, p 0.05, Fig. 8). It suggests that these two papers may be the main source of heterogeneity, which may be related to the sample size, as well as statistical methods used. A forest plot of the sensitivity analysis is shown in Fig. 8, illustrating the robustness of the findings.

Each pregnancy is influenced by a variety of factors, including the woman’s physical health, reproductive age, lifestyle, living environment, and genetic predisposition. With the improvement of living standards and the relaxation of China’s two-child policy, the average age of childbearing has significantly increased. A study has shown [20] that older mothers, compared to younger ones, face a higher risk of pregnancy complications, such as gestational hypertension and gestational diabetes. These risks not only complicate the delivery process, but also increase the likelihood of adverse outcomes, including low birth weight (LBW) and macrosomia. Furthermore, changes in lifestyle, such as excessive caloric intake and reduced physical activity, exacerbate the problem of weight gain in pregnant women. Given the increasing complications associated with excessive weight, including dystocia and postpartum obesity [21], weight management during pregnancy has become a key concern for healthcare providers. Traditional Chinese beliefs generally discourage physical activity during pregnancy and instead encourage increased nutritional intake. This cultural mindset leads to uncontrolled weight gain, further increasing the risk of CSs and other complications [22]. Weight gain is closely related to pregnancy outcomes. Excessive weight can lead to meconium aspiration syndrome, macrosomia, and childhood obesity. The risks faced by mothers include gestational diabetes, hypertension, miscarriage, and persistent postpartum obesity. When combined with other conditions, such as immune diseases and viral hepatitis, these factors significantly increase the likelihood of CS. The prevalence of hypertension in China is estimated to be 5%–10% [23], while gestational diabetes affects up to 21.8% of pregnancies [24]. Addressing these factors is crucial for improving maternal and child health outcomes.

Prenatal exercise effectively alleviates many factors that lead to CSs, including excessive weight gain, macrosomia, and psychological stress. Excessive weight gain is significantly associated with cesarean deliveries [25]. By promoting controlled weight gain, prenatal exercise reduces the risk of fetal macrosomia, a major cause of dystocia. Activities such as walking, yoga, and swimming help improve cardiorespiratory function, enhance the immune system, and reduce pregnancy complications such as hypertension and diabetes, which are common risk factors for CSs. Targeted exercises for the abdominal and pelvic muscles enhance physical preparation for delivery. Stronger pelvic muscles contribute to a smoother childbirth and reduce the risk of complications associated with prolonged labor. In addition, prenatal exercise helps to alleviate stress and anxiety by regulating emotional health, giving women more confidence in vaginal delivery. This psychological resilience is crucial for overcoming fears and pressures associated with CSs. From a physiological perspective, prenatal exercise aids in cervical ripening and dilation, which is essential for promoting natural delivery. Certain positions, such as sitting cross-legged or squatting, promote optimal fetal positioning and reduce breech presentation [26]. Correct fetal positioning is a key determinant for successful vaginal delivery. Furthermore, studies have shown that prenatal exercise improves newborn Apgar scores, an important indicator of neonatal health, highlighting its comprehensive benefits for maternal and child outcomes.

Throughout history, diverse cultures have developed unique understandings and practices of pregnancy care. In traditional medical systems, such as traditional Chinese medicine, it is emphasized that pregnant women should exercise moderately to reconcile qi and blood, as well as promote fetal development. These traditional ideas and practices, although in different forms, all reflect the concern and adjustment of the body state of pregnant women, and provide rich cultural soil and historical reference for modern prenatal movement. Over time, traditional wisdom has been preserved and scientifically validated and integrated under the framework of modern medicine. In today’s era of globalization and rapid science and technology advancement, prenatal exercise has undergone significant innovation. From yoga, pilates, water birth preparation exercises, to the use of smart wearable devices for personalized exercise monitoring, the integration of technology elements not only makes prenatal exercise safer and more efficient, but also meets the requirements of pregnant women for diverse and personalized experiences. This innovative practice not only reflects the in-depth exploration of modern life science, but also shows the flexibility and creativity of culture in adapting to the pace of modern life. The cultural perspective of prenatal movement is also reflected in the respect and care for the individual differences of pregnant women. Each pregnant woman’s physical condition, psychological state and cultural background are different, so the design and implementation of prenatal exercise should fully consider these differences and provide personalized guidance and support. This includes adjusting the intensity and type of exercise according to the physical conditions of pregnant women, as well as through group activities, psychological counseling and other ways to help pregnant women establish a positive attitude and relieve anxiety and stress during pregnancy. This individual-centered care concept not only promotes the physical health of pregnant women, but also provides a solid guarantee for their mental health.

This systematic review stands out for its comprehensive literature search, timeliness, and methodological rigor. The search covered multiple databases, including PubMed, ProQuest, Scopus, CNKI, VIP, and Wanfang, using both subject-specific and free-text terms to ensure the inclusion of the latest research. In addition, the studies included are representative, involving samples from different countries (e.g., China, United States, Europe, others), and various types of prenatal exercises (e.g., aerobic exercise, yoga, swimming, others), providing a broader background and practical experience. Compared to earlier meta-analyses that focused on observational indicators, this approach provides more actionable insights, especially in exploring how prenatal exercise affects pregnancy outcomes, offering strong support for clinical practice.

Firstly, there is significant variability in the quality of the studies included. Many lack clear details on randomization, blinding, and allocation concealment, which may introduce potential biases. Secondly, the meta-analysis shows high heterogeneity among studies, which may be due to differences in geographic regions, healthcare systems, and medical standards. For example, some areas may be more inclined to use traditional prenatal rest methods, while others are more proactive in promoting prenatal exercise. Additionally, although most studies involve common types of prenatal exercise, there are differences in the specific ways of intervention (e.g., frequency, intensity, duration of exercise, etc.), which may affect the generalizability of the conclusions. In terms of the robustness of the studies, this review screened high-quality RCTs from the past five years, and the studies included have good diversity and representativeness in terms of sample size, type of intervention, and assessment indicators. However, many studies were excluded due to not meeting the quality requirements or lacking key information. Especially in some low-quality studies, there was insufficient detail on randomization or blinding, which may have affected the accuracy and reliability of the final results. Lastly, although most studies have positively evaluated the impact of prenatal exercise on CS rates, a few studies have failed to find significant effects, which may be related to study design, sample selection, or other uncontrolled confounding factors. Therefore, further high-quality studies, especially multicenter, long-term follow-up studies, are still necessary to confirm the long-term effects of prenatal exercise on different pregnancy outcomes.