Introduction
Autism spectrum disorders (ASD) among children are characterized by impaired social interactions, communication deviance, and the presence of restrictive and repetitive behavior patterns [1]. Previous reviews reported that ASD is predominantly genetic. Genetic factors alone account for 20%–30% of ASD cases, whereas the remaining 70%–80% are due to a complex interaction between environmental risk factors and genetics [2]. Recent meta-analyses have identified the role of advanced parent age, gestational infections, low birth weight, fetal distress, intrauterine growth retardation, small for gestational age, neonatal icterus, preterm labor, cesarean section, preeclampsia, and labor complications in the pathogenesis of ASD [3-5].
Some researchers have suggested that breastfeeding may protect against ASD [6-8], while another study did not show this association [9]. A meta-analysis by Tseng et al. [10] found that children with ASD were significantly less likely to be breastfed than those without ASD. However, they assessed the prevalence of breastfeeding (less versus more breastfeeding) and different breastfeeding durations in children with versus without ASD. Therefore, the present meta-analysis evaluated not breastfeeding versus breastfeeding and the risk of ASD among children based on epidemiological studies.
Methods
We performed the meta-analysis according to the PRISMA (Preferred Reporting Items for Systematic Reviews) criteria.
1. Eligibility criteria
The exposure variable was not breastfeeding, while the outcome of interest was ASD in children aged ≤18 years. Observational studies included cohort, cross-sectional, and case-control studies irrespective of language; publication date; and participant nationality, race, and age. The exclusion criteria were systematic reviews and meta-analyses, case reports and series, controlled trials, letters to the editor, and studies lacking complete data.
2. Information sources and search
We systematically searched the PubMed, Web of Science, and Scopus databases for articles published through December 2021. The search terms were applied in combination: (not breastfeeding, not breastfed, no breast milk, lactation, breastfeeding, breastfed, artificial milk-fed) and (ASD, autism spectrum disorders, or autism). The reference lists of the retrieved articles were manually searched to identify any further studies. We contacted the authors of the included studies for additional data.
3. Study selection
We used EndNote reference management software (Chicago Manual of Style, USA) to merge the results from different databases. Duplicate studies were subsequently excluded. In addition, 2 researchers independently extracted all data, and disagreements between the 2 researchers were resolved by discussion.
4. Data extraction
Data from the included studies were extracted using a data sheet in Stata ver. 13 (StataCorp LLC, College Station, TX, USA). The following data were included: first author, year of publication, country, study design, sample size, control for confounding variables (adjusted, unadjusted), age range, and ASD diagnostic method.
5. Methodological quality
We evaluated study quality using the Newcastle Ottawa Scale (NOS) [11]. Any study could achieve a maximum of 9 NOS stars: 4 for selection quality, 2 for comparability, and 3 for exposure quality. A score ≥7 indicated high quality.
6. Heterogeneity and reporting biases
Interstudy heterogeneity was assessed using the chi-square test [12] and the I2 statistic [13], while publication bias was assessed using Begg [14] line regression test.
7. Summary measures
The association between breastfeeding and the risk of ASD was assessed using odds ratios (ORs) and 95% confidence intervals (CIs). We used fully adjusted ORs to control for potentially confounding factors. A random-effects model was used to analyze the data [15]. Stata ver. 13 was used for the analysis, with a significance level of 0.05.
Results
1. Description of studies
According to our search strategy, 766 references were identified until December 20, 2021. Of them, 189 were duplicates and 566 were excluded by the title and abstract screening. Eleven studies were subjected to full-text review, which eliminated 4 full papers. Therefore, 7 studies were ultimately included in the present meta-analysis (Fig. 1A): 5 with a case-control design [6-9,16] and 2 with a cross-sectional design [17,18]. The total study population was 3,270 individuals. All studies were published in English (Table 1).
2. Effects of exposure
Fig. 1B assessed the association between not breastfeeding and the risk of ASD. According to the random-effects model, the estimated OR of the risk of ASD associated with not breastfeeding was 1.81 (95% CI, 1.35–2.27; I2=0%) (Fig. 1B). The results of the included studies were homogenous.
3. Publication bias
Publication bias was assessed using Begg test. The P value for Begg regression among the children with ASD was 0.176. No evidence of publication bias was found among studies reporting an association between breastfeeding and the risk of ASD.
4. Subgroup analysis
We performed a subgroup analysis of the quality of the studies and designs. The pooled results of the cross-sectional and casecontrol designs were OR=1.62 (95% CI, 1.12–2.13) and 2.65 (95% CI, 1.57–3.73), respectively. There was no significant association in studies with a low-quality OR (2.98; 95% CI, 0.91–5.05; Table 2).
5. Study quality
In the evaluation of study quality based on the NOS scale, 6 studies were of high quality and one was of low quality (Table 1).
Discussion
To our knowledge, this is the first meta-analysis based on epidemiologic studies of the association between not breastfeeding versus breastfeeding and the risk of ASD. Our results showed that not breastfeeding was a risk factor for ASD. No heterogeneity was detected among the studies that reported an association between breastfeeding and the risk of ASD. The mechanisms underlying the association between breastfeeding and ASD risk are unclear and warrant further research. Different findings have been reported regarding ASD with different biological factors.
An inadequate intake of “beneficial” omega-3 and omega-6 polyunsaturated fatty acids (PUFA) derived from the mother during pregnancy or lactation may also play an important role in ASD. However, omega-3 and omega-6 PUFA deficiencies have been identified in children with ASD, while the relationship between poor breastfeeding and the risk of ASD has been investigated. Fatty acids are abundant in the colostrum secreted during the first 2–3 days after birth. Breast milk is a rich source of longchain polyunsaturated omega-3 and omega-6 fatty acids considered important in the development of the cognitive, social, and language abilities of children aged 6 months to 3.5 years [19,20]. Other studies reported that the elevated oxytocin levels experienced during infant sucking may protect against the development of ASD [21,22].
A meta-analysis by Ghozy et al. [23] in 2018 showed that breastfeeding decreased the risk of ASD by 58%, while exclusive breastfeeding decreased the risk by 76%. Breastfeeding for 12–24 months resulted in the most significant reduction in the risk of ASD.
Another meta-analysis by Tseng et al. [10] reported that children with ASD (clinical diagnosis or self-reported) were significantly less likely to be breastfed than those without ASD (OR, 0.61; 95% CI, 0.45–0.83; P=0.002). However, they assessed the prevalence (less versus more) and durations in children with ASD versus without ASD. The present meta-analysis evaluated not breastfeeding versus breastfeeding as a risk factor for ASD in children.
This study has some limitations. First, to control for known risk factors for ASD, we used an adjusted form. However, other studies have reported only the unadjusted OR. This may have introduced bias into our study. Second, the sample sizes of the included studies were small. Despite these limitations, our results suggested that not breastfeeding is a risk factor for ASD. Therefore, breastfeeding is necessary to decrease the risk of ASD in children.
In conclusion, our results showed that not breastfeeding was a risk factor for ASD. Therefore, breastfeeding is necessary to reduce the risk of ASD in children. Our findings suggest that the possible association between ASD and not breastfeeding should be added to the list of reasons to provide breastfeeding support.