Research design and searching strategy

This umbrella review was conducted following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [17]. Additionally, to ensure the study’s rigor, we adhered to the PRIOR statement, a reporting guideline for overviews of reviews on healthcare interventions [18].

Articles were searched across four databases including PubMed, Scopus, Web of Science, and Google Scholar. Additionally, a complementary search was conducted in Google Scholar to capture grey literature sources. We performed a comprehensive systematic search for SRMs that assessed the association between prenatal folic acid and/or multivitamin supplementation and the risk of ASD on a global scale, using the PICO (Population, Intervention, Comparison, and Outcome) framework. The search strategy incorporated MeSH terms, keywords, and their combinations. Moreover, we manually reviewed references cited in relevant systematic reviews and conducted a snowball search to identify linked articles from these studies. The search terms included: (Preconception OR pre-conception OR prenatal) AND (Multivitamin OR vitamin OR mineral OR micronutrient OR antioxidant OR diet OR folic acid OR folic-acid) AND (autism OR autistic OR ASD OR Asperger OR pervasive developmental disorder OR PDD) AND (meta-analysis OR systematic review OR review).

Study selection/eligibility criteria

The retrieved SRMs were imported into EndNote 20 software. Screening followed a two-stage process: title and abstract screening, followed by full-text review. Two independent investigators (BB and MA) applied pre-specified inclusion criteria to identify relevant articles. Eligible studies met the following criteria: [i] had a defined literature search strategy, [ii] appraised the included studies using a relevant tool, and [iii] used a standard approach to pooling data and providing summary estimates. Studies were excluded if they lacked relevant outcome measures, were not in English, or were narrative reviews, editorials, correspondence, conference abstracts, or methodological studies. Disagreements in both title/abstract and full-text screening were addressed through structured discussions during consensus meetings, where team members collaboratively reviewed the criteria and reached a resolution based on predefined guidelines.

Criteria for considering studies for this umbrella review

Quality assessment

The methodological quality of all included reviews was assessed by two independent reviewers using the Assessment of Multiple Systematic Reviews (AMSTAR) tool [19,20]. The quality was scored out of 11, with scores < 3, 4–7, and 8–11 indicating low, moderate, and high qualities, respectively. The decision as to whether to include a review was made based on meeting a pre-determined proportion of all criteria, or on certain criteria being met. Decisions about a scoring system or any cut-off for exclusion were made in advance and agreed upon by all reviewers before critical appraisal commences.

Data extraction

Data from the included SRM studies were extracted using a standardized data abstraction form, developed in an Excel spreadsheet. For each SRM study, the following data were extracted: (a) identification data (first author’s last name and publication year), (b) Review aim and type, (c) measure of association (odds ratio or relative risk with 95% confidence intervals), (d) number of primary studies included within each SRM study and their respective design type, (e total number of sample size included, (f) publication bias assessment methods and scores, (g) quality assessment methods and scores, (h) and data synthesis methods (random or fixed-effects model) [Table 1].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Characteristics of included studies.

https://doi.org/10.1371/journal.pone.0334852.t001

Statistical analysis

STATA v17.0 software was used for statistical analyses. Both narrative and qualitative methods, including tables and forest plots, were employed to summarize the estimates from the included reviews. We conducted an inverse variance-weighted random-effects meta-analysis to estimate pooled effect sizes with 95% CIs [21]. We assessed heterogeneity across studies using Cochrane’s Q test (Chi-square), the I² statistic, tau-squared (τ²), and corresponding p-values [22]. Heterogeneity was interpreted according to conventional thresholds, with I² values of 0% indicating no heterogeneity and values of 25%, 50%, and 75% representing low, moderate, and high heterogeneity, respectively [20,23]. For analyses where significant heterogeneity was detected, we applied the DerSimonian–Laird random-effects model. Subgroup analyses were conducted according to supplement type (folic acid versus multivitamin). To evaluate the robustness of findings, sensitivity analyses were performed by sequentially excluding individual studies. Publication bias was assessed using both graphical and statistical approaches: visual inspection of funnel plot symmetry and Egger’s regression test. Funnel plot asymmetry and a p-value <0.05 in Egger’s test were considered indicative of potential publication bias [24].

Credibility/GRADE assessment of each pooled analysis

We used the GRADE system to assess the strength of evidence for each pooled analysis, classifying it as convincing, highly suggestive, suggestive, or weak. GRADE uses five criteria: risk of bias, inconsistency, indirectness, imprecision, and publication bias. Downgrades occur based on a high risk of bias, inconsistency (high I² values), indirectness (multiple control comparisons), imprecision (wide confidence intervals or small sample size), and publication bias (significant Egger test p-value). Convincing evidence (Class I) required a p-value < 10 ⁻ ⁶, over 1,000 participants, low-to-moderate heterogeneity (I² < 50%), a 95% prediction interval excluding the null, and no small-study bias. Highly suggestive associations (Class II) involved over 1,000 participants, a p-value < 10 ⁻ ⁶, and a 95% prediction interval excluding the null. Suggestive evidence (Class III) required over 1,000 participants and a p-value ≤ 0.001, while weak associations (Class IV) needed a p-value ≤ 0.05, with p > 0.05 considered non-significant [25,26].

Ethics approval and consent to participate

This study is a systematic review of published articles and does not require ethical approval.

Quality assessment

Based on AMSTAR-2 criteria, the majority of the SRMs (6 out of 8) [10,12,13,15,16,27] were assessed as high quality, while 2 out of 8 were rated as low quality [11,14] (S1 Table). The most common reasons for a low-quality rating based on the AMSTAR-2 checklist are: failure to register the protocol in PROSPERO [10–13,15,27], absence of an evaluation of the risk of bias in the meta-analysis [10,11,13–15,27], insufficient explanation of heterogeneity [11,16], and failure to assess publication bias [10,11] (S1 Table).

Effect of maternal prenatal folic acid and multivitamin supplementation on offspring ASD

Among the eight studies included, six reviews reported an association between prenatal folic acid and multivitamin supplementation and a reduced risk of ASD [10–14], whereas the remaining two reviews, one for folic acid and one for multivitamin, found no such association [15,16]. The effect estimates from these included studies ranges from 0.57 (95% CI: 0.41, 0.78) [12] to 0.91 (95% CI: 0.73, 1.13) [10].The random-effects model analysis revealed a 30% reduced risk of ASD (RR = 0.70, 95% CI: 0.62–0.78) in offspring of mothers who took folic acid and/or multivitamin supplementation during prenatal care, compared to those who did not. Subgroup analysis by supplement type showed that maternal prenatal multivitamin supplementation reduced the risk of ASD by 34% (RR = 0.66, 95% CI: 0.55–0.80; GRADE: highly suggestive), while folic acid supplementation was associated with a 30% reduction in ASD risk (RR = 0.70, 95% CI: 0.60–0.83; GRADE: highly suggestive) (Fig 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Forest plot shows the pooled effect of pre-conception FA and multivitamins for the prevention of autism.

https://doi.org/10.1371/journal.pone.0334852.g002

Sensitivity analysis

The sensitivity analysis indicated that any of the single studies did not influence the pooled result. The pooled sensitivity estimate for the association between maternal prenatal folic acid and multivitamin supplementation and offspring ASD ranged from 0.67 (95% CI: 0.61–0.74) [15] to 0.72 (95% CI: 0.64–0.81) [13] (S1 Fig).

Publication bias assessment

The symmetrical funnel plots and Egger’s test provided no evidence of substantial publication bias among the included studies (P-value = 0.14).

Credibility according to GRADE evidence assessment

Overall, the strength of evidence assessed using the GRADE framework indicates highly suggestive (Class II) evidence for both the pooled analysis of the effects of maternal prenatal folic acid and multivitamin supplementation in preventing ASD. Both pooled analyses met most of the GRADE criteria. Specifically, both analyses demonstrated low to moderate heterogeneity, with I² values of 0.00% (P = 0.74) and 60.76% (P = 0.04), respectively. No publication bias was detected in either analysis (Egger’s test, P = 0.09). Furthermore, both analyses met the other three GRADE criteria: 1) no serious indirectness, 2) no serious imprecision, and 3) low risk of bias assessment using AMSTAR (A Measurement Tool to Assess Systematic Reviews), where 75% of questions were answered ‘yes’ and 25% were rated as ‘unclear’ or ‘no.’ As a result, the overall certainty of the evidence for both folic acid and multivitamin supplementation in preventing ASD highly suggests a 30% reduction in the risk of ASD for folic acid supplementation and a 34% reduction in the risk of ASD for multivitamin supplementation (S2 Table).

Implications for practice, policy, and future research

Practice: The findings emphasize the importance of prenatal folic acid and multivitamin supplementation as part of routine maternal care to reduce the risk of ASDs in offspring. Healthcare providers should encourage women to start supplementation during the pre-conception period and continue throughout pregnancy. This proactive approach can be integrated into preconception and prenatal care programs to promote maternal and fetal health.

Policy: Policymakers should prioritize the inclusion of folic acid and multivitamin supplementation in national maternal health guidelines and public health initiatives. Public awareness campaigns are needed to educate women of reproductive age about the potential benefits of these supplements in reducing ASD risk. Subsidizing or ensuring the affordability of high-quality folic acid and multivitamin supplements, particularly in low-resource settings, can further enhance accessibility and adherence to these recommendations.

Research: Future research should focus on conducting large-scale prospective cohort studies and randomized controlled trials (RCTs) to confirm and expand these findings. Key areas of investigation include exploring the molecular mechanisms behind the protective effects of folic acid and multivitamins against ASD, determining the optimal dosage, duration, and timing of maternal supplementation for maximum risk reduction, and assessing the roles of specific nutrients like vitamin D, zinc, and folate in mitigating ASD risk. Additionally, it is important to evaluate the impact of maternal supplementation across diverse populations and geographic regions to enhance generalizability and ensure culturally relevant interventions. Addressing these gaps will strengthen evidence to refine clinical guidelines and improve preventive strategies for ASD.