We searched all published literature in four databases, Pubmed, Cochrane CENTRAL, and Web of Science, from the date of April 2014 to April 2024. Taking Pubmed as an example, our search strategy is “#1 Parturition [MeSH Terms] OR (Parturitions OR Birth OR Births OR Childbirth OR Childbirths OR Deliver* OR antenatal OR prenatal OR postpartum OR puerperium OR gestation OR postnatal OR Pregnan*) AND (in the last 10 years)[Filter] AND (Randomized Controlled Trial)[Filter] #2 (Fear of childbirth) OR FOC OR fear OR (fear of birth) AND (in the last 10 years)[Filter] AND (Randomized Controlled Trial)[Filter] #3 #1 AND #2.” In addition, reference lists of the included studies, and published systematic reviews and meta-analyses were hand searched to improve coverage. The specific search strategy is shown in the Supplementary material 1.

We defined the target trials according to the PICOS (population, interventions, comparisons, outcomes, study design) selection criteria.

(1) Not RCT, (2) conference papers, abstracts, meta-analyses, or systematic reviews, (3) papers with incomplete data, (4) papers published not in English.

Endnote X9.3.3 was used to manage the literature. Duplicates were excluded using the duplicate literature filtering function of the software, and two authors (ZHOU & ZHU) independently screened the papers based on study titles and abstracts to obtain papers that met the criteria. We then independently extracted data from the included studies. Disagreements were resolved by discussion between the two reviewers, and a third reviewer (LI) was consulted if necessary. If a consensus could not be reached, predefined tiebreaker criteria, such as prioritizing the most recent or methodologically rigorous study, were applied to ensure an objective resolution.

During data extraction, the primary outcome measure was FOB scales and the secondary outcome measures were mode of delivery, depression, stress, anxiety and delivery self-efficacy scales. Details of the extracted data included RCT registration number, first author, year of publication, country, sample size, baseline characteristics of participants (mean age and gestational week), interventions, days of duration, outcome measure indicators. Outcome data included baseline and endpoint data on FOB, depression, stress, anxiety, and childbirth self-efficacy scales, including means and standard deviations. In addition, the number of people who chose different modes of delivery was extracted. If the outcome of a trial included several post-intervention and follow-up scores, only the score immediately after the end of the intervention phase was used.

Two reviewers (ZHOU & ZHU) independently examined the methodological quality of individual studies by using Cochrane risk of bias tool 2.0. The assessment domains included: bias arising from the randomization process, bias due to deviations from intended interventions, bias due to missing outcome data, bias in measurement of the outcome, bias in selection of the reported result, and overall bias. The overall risk of bias judgement would be “high risk,” “low risk,” or “unclear,” which could be obtained based on signaling questions and corresponding algorithms. Any discrepancy was resolved by discussion and made a consensus.

The network meta-analysis (NMA) was conducted using Stata (version 17) based on frequentist principles. For studies reporting outcomes as pre- and post-intervention measures, we applied methods from the Cochrane Handbook to calculate the mean change and its corresponding standard deviation (SD). Continuous outcome indicators were analyzed using the standardized mean difference (SMD) with a 95% confidence interval (95% CI) as the effect size. The incidence of cesarean section (CS) was expressed as an odds ratio (OR) with a 95% CI. Statistical significance was defined as a p-value of less than 0.05.

Global inconsistency tests and node-splitting tests were used to assess inconsistency. A consistency model was used only if neither of the two reported significant inconsistency (p > 0.05). If inconsistency was reported in any network, sensitivity analysis was performed to identify the source of inconsistency, and the study was excluded from the network. Network diagrams of the interventions were constructed to show the correlation between treatment options. Publication bias within each network was assessed using funnel plots. The relative effectiveness of interventions was ranked by the surface under the cumulative ranking curve (SUCRA). Funnel plots were used to detect potential publication bias and small-sample effects. To compare the effectiveness of non-pharmacological interventions for reducing cesarean section rates. Between-study heterogeneity was evaluated using the Cochrane Q test (with significance set at p < 0.1) and I² statistics (with values above 50% indicating substantial heterogeneity). If heterogeneity was not significant, a fixed-effects model was used for statistical analysis; otherwise, a random-effects model was applied. To assess the effectiveness of non-pharmacological interventions in reducing fear of childbirth among primiparous women and those with severe FOC (W-DEQ-A score ≥ 85), subgroup analysis was performed.

A total of 2,830 articles were yielded by the search by keywords. An additional 4 articles were identified by searching the reference lists of the included studies. After eliminating 699 duplicate articles and screening the titles and abstracts, we retained 84 articles for full-text reading and review. Finally, 32 (Abdollahi et al., 2020; Alivand et al., 2023; Andaroon et al., 2017; Baas et al., 2023; Boz et al., 2021; Calpbinici and Yücel, 2022; Çankaya and Şimşek, 2021; Dai et al., 2021; Duncan et al., 2017; Emadi et al., 2023; Ghasemi et al., 2018; Gökbulut et al., 2024; Gönenç and Dikmen, 2020; Güney et al., 2022; Güven Santur and Özşahin, 2024; Kuo et al., 2022; Lee et al., 2023; Lida and Fariborz, 2017; Mortazavi and Mehrabadi, 2021; Navaee and Abedian, 2015; Ozbek and Pinar, 2022; Rondung et al., 2018; Rouhe et al., 2015; Sun et al., 2021; Thijssen et al., 2023; Toohill et al., 2017; Uludağ et al., 2022; Vakilian et al., 2023; Veringa-Skiba et al., 2022; Wang et al., 2023; Wilczyńska et al., 2023; Yesildag and Golbasi, 2024) articles were included in this meta-analysis. The study selection flow chart is shown in Figure 1.

The trials included a total of 3,187 pregnant women receiving 17 non-pharmacological interventions. The specific interventions are as follows: (1) motivational interview (MI) (n = 3), (2) high intensity interval training (n = 1), (3) prenatal education program (n = 11), (4) emotional freedom technique (EFT) (n = 2), (5) mindfulness-based interventions (n = 6), (6) deep relaxation exercise (n = 1), (7) acceptance and commitment therapy (n = 1), (8) breathing awareness training (n = 1), (9) haptonomy (n = 2), (10) CBT (n = 3), (11) simulation-based childbirth education (n = 3), (12) eye movement desensitization and reprocessing therapy (n = 1), (13) counseling therapy (n = 3), (14) dance and music therapy (n = 1), (15) music therapy (n = 1), (16) placebo (n = 6), (17) care as usual (n = 19). The table of basic characteristics of included literature can be found in Supplementary material 2.

In general, the randomized controlled trials included in our network meta-analysis showed an acceptable and relatively low risk of biases. Almost all the included trials reported randomization but did not describe the allocation concealment in sufficient detail and were judged to have an unclear risk bias. Seven studies were judged to have a high risk of bias with regard to “blinding of participants and personnel” domain. Six studies had a high risk of bias concerned in the “blinding of outcome assessment” domain. Most studies described in detail the loss of follow-up and excluded data. Most studies had an unclear risk of selective reporting bias. Other risk of bias was not identified in the included studies. Supplementary material 3 presents the results of the methodological quality evaluation of the included studies.

Separate network meta-analyses were conducted to evaluate improvements in FOB both immediately post-intervention (during the gestational period) and after delivery. During the gestational period, FOB was measured using the W-DEQ-A, CAQ, FOBS, PRAQ-R2, and the Childbirth Self-Efficacy Questionnaire (fear subscale). In the postnatal period, the W-DEQ-B and FOBS were used as measurement instruments.

Inconsistency of the data was detected according to the fit of the inconsistency model and node-splitting model, with closed-loop p > 0.05, suggesting no significant inconsistency between the results of direct comparison and indirect comparison. The Forest plots of consistency analyses are shown in Figure 5.

As shown in Figure 6, the scatter of points on the funnel plot has shown visually symmetrical in shape around the mean estimated treatment effect, indicating low risk of publication bias.

Our study included 32 randomized controlled trials, identifying 17 non-pharmacological interventions to address FOC in women through a network meta-analysis. We also extracted data on related outcomes, including depression, stress, anxiety, self-efficacy, and mode of delivery. This review demonstrated that dance and music therapy, CBT, and EFT were the top three interventions for reducing FOC. For alleviating FOC in postnatal period, EFT, counseling therapy, and haptonomy were the most effective.

Cluster analyses supported the recommendation of EFT, MI, haptonomy, counseling therapy, and CBT as effective non-pharmacological interventions for mitigating FOC. Subgroup analyses revealed that, compared with usual care, EFT was most effective for primigravida, while haptonomy had the greatest effect in women with severe FOC. Counseling therapy was found to be statistically significantly effective in reducing depression, stress, and anxiety symptoms in pregnant women compared to usual care. No single intervention showed a significant advantage over others in improving childbirth self-efficacy. However, haptonomy and antenatal education programs significantly reduced the cesarean section rate.

The study indicates that EFT can significantly reduce FOC. EFT, which blends cognitive and exposure therapies with acupressure, uses finger tapping on acupoints and is commonly known as “tapping” (Church et al., 2022). Clinical EFT is an evidence-based treatment for PTSD, with meta-analyses supporting its efficacy for depression, anxiety, and phobias (Clond, 2016; Sebastian and Nelms, 2017; Stapleton et al., 2023). EFT is accessible, easy to train, and safe, with remote instruction proving as effective as group sessions. This scalability makes it a promising, cost-effective intervention (Church et al., 2018). However, EFT has limited research in maternal health, and further high-quality studies are warranted, particularly for women with traumatic birth histories, to confirm EFT’ s efficacy in managing childbirth-related fears.

Haptonomy is an emerging intervention that fosters a positive emotional connection between mother and fetus through touch, potentially improving maternal mental health, preventing psychiatric disorders, and promoting strong mother-infant bonding (Fernandez, 2019; Ksycinski et al., 2010). Our findings suggest that haptonomy may help alleviate severe FOC and reduce cesarean section rates. However, it remains uncertain if the reduction in cesarean rates is clinically significant, given the limited studies and the influence of factors such as medical indications, partner preferences, and socio-economic and cultural considerations on delivery choice (Sebastião et al., 2016). Additionally, current trials focused only on primiparous women, and future research should include a wider population to confirm the effectiveness of haptonomy.

MI originated in addiction treatment and fosters a partnership between clinician and patient through an encouraging and guiding conversation style. It aims to evoke motivation for change by reinforcing and exploring the patient’s reasons for change, emphasizing patient autonomy, and prioritizing their needs (Miller and Rollnick, 2015). MI has proven to be an evidence-based approach for treating substance addictions, including alcohol, marijuana, and tobacco (Bischof et al., 2021). More recently, its use has expanded to mental health areas like anxiety and eating disorders (Romano and Peters, 2015), where MI has shown greater effectiveness as an adjunct to cognitive-behavioral therapy (CBT) compared to CBT alone (Marker and Norton, 2018). In the maternal health field, MI has been less extensively applied, this study suggests that MI holds potential as a primary or adjunctive method to reduce FOC.

This study found that both CBT and counseling therapy significantly reduce FOC. These interventions are commonly used to address psychological symptoms in pregnant women. While substantial evidence supports CBT’s effectiveness in improving sleep, anxiety, and depression during pregnancy (Shortis et al., 2020; Zheng et al., 2023), its impact on FOC is less studied or yields inconclusive results (Aguilera-Martín et al., 2021). CBT includes various formats. Some studies suggest face-to-face CBT is more effective than digital CBT due to adherence challenges (Larsson et al., 2017; Rondung et al., 2018), while others find no significant differences across formats or the opposite (Yunus et al., 2022). Our study did not distinguish between CBT types, suggesting future research should explore their specific effectiveness in reducing FOC. Counseling therapy, typically led by midwives, obstetricians, or psychologists, is a traditional psychological intervention (Souto et al., 2022). Although its effectiveness in reducing FOC is well-documented, some studies report limited impacts on FOC, childbirth experiences, and cesarean rates (Larsson et al., 2015).

Each intervention has different strengths and limitations in reducing fear of childbirth. EFT significantly reduces FOC and is easy to learn, safe, and cost-effective. However, there is limited research in maternal health, particularly for women with traumatic birth histories, and more high-quality studies are needed to confirm its effectiveness. Haptonomy may improve maternal mental health and reduce CS rates. However, current studies focus only on primiparous women, and the clinical significance of reduced cesarean rates remains uncertain. MI shows potential as an effective intervention for FOC, and further research is needed to explore its role in this context. CBT and counseling therapies are commonly used to reduce FOC. However, differences between CBT formats need further exploration. Counseling therapy is also effective in reducing FOC, though its impact on CS rates and childbirth experiences may be limited, suggesting a need for more effective alternatives. Further large-scale studies are needed to compare their efficacy and guide healthcare professionals in developing more effective strategies. Future research should compare the long-term effects of these interventions in larger samples and develop clinically practical, highly sensitive FOC assessment tools for rapid screening.

This study has several limitations requiring attention in future research. First, we did not conduct stratified analyses on sociodemographic factors that may influence FOC, such as age, gestational age, educational level, and income, nor did we account for variables such as treatment initiation and duration, which could introduce heterogeneity into the study. Second, some interventions, like music and dance therapy, were evaluated in only one study with small sample sizes, limiting result applicability and reliability. Third, the use of varying measurement scales with different structures, scoring criteria, and sensitivities may have introduced inconsistencies. Fourth, although all studies reported FOC outcomes, most did not assess related indicators such as depression, anxiety, stress, childbirth self-efficacy, or delivery methods. Fifth, many studies lacked adequate allocation concealment and blinding, risking bias and potential overestimation of intervention effects. Sixth, the inclusion of only English-language published literature introduces a potential risk of bias, as it may exclude relevant studies published in other languages.

Future research should compare the long-term effects of these interventions in larger samples and develop clinically practical, highly sensitive FOC assessment tools for rapid screening.