Exclusive breastfeeding and infant colic: a systematic review and meta-analysis

Abstract

Background

The benefits of exclusive breastfeeding for infant health outcomes have been widely reported; however, its association with infant colic remains unclear. This systematic review and meta-analysis aimed to examine whether exclusive breastfeeding is associated with lower odds of infant colic.

Methods

We searched MEDLINE via OVID, EMB Reviews, Embase, MIDIRS, and Global Health for relevant publications up to March 8, 2026. We included observational studies reporting infant colic among exclusively breastfed infants compared with infants receiving other types of feeding (mixed feeding, formula feeding, or non-exclusive breastfeeding) in children under 12 months of age. Two authors independently screened studies using Covidence, assessed risk of bias, and extracted data using Excel. Overall pooled estimates were obtained using multilevel random-effects meta-analysis, and subgroup analyses were conducted using multilevel meta-regression models, with additional stratified random-effects models for comparator-specific estimates.

Results

Of 13,349 articles identified, 15 studies involving 216,314 infants were included in the meta-analysis. Exclusively breastfed infants had 19% lower odds of colic than non-exclusively breastfed infants (pooled OR = 0.81, 95% CI: 0.67–0.97, p = 0.021). Substantial heterogeneity was observed across studies (I² = 92.6%). Additional analyses suggested that the pooled estimate was robust and unlikely to be substantially affected by publication bias or small-study effects. In subgroup analyses, exclusive breastfeeding was associated with a significantly lower likelihood of infant colic relative to mixed feeding (OR = 0.72, 95% CI: 0.57–0.93; p = 0.01). However, the association was weaker and not statistically significant when exclusive breastfeeding was compared with formula feeding (OR = 0.92, 95% CI: 0.71–1.17; p = 0.470).

Conclusions

Exclusive breastfeeding was associated with lower odds of infant colic. Further high-quality studies are needed to address limitations in the existing evidence and to better understand the mechanisms underlying this association.

Clinical trial number

PROSPERO 2025 CRD420251051099. https://www.crd.york.ac.uk/PROSPERO/view/CRD420251051099.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12887-026-07317-0.

Introduction

Infantile colic is a common yet challenging condition that affects approximately 17%–25% of otherwise healthy infants during the first six weeks of life [1]. In research, both Wessel and the Rome IV criteria, infants with colic are typically described as crying excessively without an apparent cause and remaining difficult to soothe despite caregiver efforts, in the absence of fever, failure to thrive, or signs of acute illness [2, 3].

Although infant colic is generally considered a self-limiting condition, it may have important implications for both infant and family well-being. Infants with colic are more likely to experience feeding difficulties [4, 5], and sleep disturbances [6, 7]. A recent study suggests that infantile colic may be associated with early-life biological vulnerabilities linked to later gastrointestinal, atopic, and neurodevelopmental disorders [8]. In addition to its effects on infants, excessive crying can impose a substantial emotional burden on caregivers, contributing to increased symptoms of stress, anxiety, and depression, as well as strain on family relationships [8, 9]. Infant colic is also a common reason for healthcare utilization, accounting for up to 20% of health-seeking consultations during early infancy [10].

The pathophysiology of infantile colic remains unclear; although several infant- and maternal-contributing factors have been proposed. Infant-related factors include gastrointestinal immaturity, gut microbiota dysbiosis, neurodevelopmental dysregulation, neonatal antibiotic exposure, birth by cesarean delivery and a family history of colic [8, 11]. Maternal characteristics and exposures have also been implicated, including prenatal anxiety or depression, alcohol consumption, smoking during pregnancy, atopic diseases, irritable bowel syndrome, and irregular work schedules [8, 12, 13].

Exclusive breastfeeding is a potentially modifiable early-life exposure that may influence the risk of infantile colic through several biological pathways, including gut microbiota development, gastrointestinal maturation, immune regulation, and the microbiota–gut–brain axis [14, 15]. Its benefits for infant health, compared with predominant, partial, or no breastfeeding, are well established [16–18]. Given the high prevalence of infantile colic and its impact on infant and family well-being, understanding whether exclusive breastfeeding influences colic risk has important implications for infant health promotion and parental counseling. However, the existing evidence remains inconclusive. While some studies have reported a protective association between exclusive breastfeeding and infantile colic, others have found no significant relationship [12, 19–21]. These inconsistencies may reflect methodological differences across studies, including variations in the definitions and measurement of exclusive breastfeeding and infantile colic, the timing of exposure and outcome assessment, and residual confounding arising from the multifactorial and incompletely understood etiology of infantile colic. We therefore conducted a systematic review and meta-analysis of epidemiological studies to evaluate the association between exclusive breastfeeding and infantile colic.

Methods

Study design

This systematic review and meta-analysis were designed according to the PICO framework. The population comprised healthy infants, the exposure was exclusive breastfeeding, the comparator was formula feeding and/or mixed feeding, and the outcome was infantile colic (Table 1). The review was conducted and reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [22].

Table 1.

Inclusion and exclusion criteria

Criteria	Inclusion	Exclusion
Population	• Infants under 12 months	• Studies exclusively among preterm, low-birthweight or any other health conditions
Exposure	• Exclusive breastfeeding: only breastmilk, no other liquids, or solids, except for vitamins, minerals, or medicines prescribed by a healthcare provider. • Any mechanism of feeding (e.g., feeding at the breast [direct breastfeeding] vs. feeding expressed human milk) • Measured point in time based on interview, questionnaire, or clinical record or via maternal or parental recall up to child aged 12 months • Infants exposed to exclusive breastfeeding from birth till the time of assessing colic or at least 3 months in retrospective studies among infants less than 12 months	• Any breastfeeding • Prominent breastfeeding • Studies that did not report or define time points of breast • feeding outcomes • Exclusive breastfeeding and colic either not collected at the same time or less than 3 months in retrospective studies among infants less than 12 months
Comparators	• Partial breastfeeding or mixed feeding or complementary feeding: breastmilk used along with any food or liquid, including non-human milk and formula or as defined by the author. • Formula feed is used only in formula as the source of nutrition for infants or any food or liquid. • Non-exclusive breastfeeding: no exclusive breastfeeding defined by authors or Partial breastfeeding or mixed feeding or complementary feeding or Formula feed	• All other comparisons; no comparison
Outcomes	Colic observed at any point under 6 months following: • Wessel criteria • the Rome IV consensus group • Identified by a physician • Mother/caregiver reported	• No data of colic prevalence reported by exposure and comparator or no adjusted or unadjusted results were reported
Study designs	• Observational studies comparing the occurrence of infantile colic between exclusively breastfed infants and one or more comparison groups	• Any kind of review • Studies with no comparison groups • Case study, case series
Publication language	No restriction during search and screen for tittle and abstract	Full text articles not in English

Eligibility criteria

Table 1 presents the inclusion and exclusion criteria used to guide the literature search and determine study eligibility for the systematic review and meta-analysis. We included studies of healthy infants younger than 12 months that defined colic according to Wessel’s criteria [2], the Rome IV criteria [3], physician diagnosis, or maternal report (yes/no). We included different definitions of infantile colic because it is a symptom-based condition without a definitive diagnostic test. Limiting inclusion to a single definition would have excluded much of the available evidence and reduced the review’s power and generalizability. Data were collected through diary records, health records, caregiver interviews, or self-administered questionnaires. We excluded studies focused on preterm or low-birth-weight infants, as well as infants with specific medical conditions.

This systematic review included observational studies, including prospective and retrospective cohort studies and cross-sectional studies, that evaluated the association between exclusive, partial, or non-exclusive breastfeeding and infant colic. No language restrictions were applied during the search process; however, studies published in languages other than English were excluded during full-text assessment.

The exposure of interest was exclusive breastfeeding, regardless of feeding method (e.g., direct breastfeeding or feeding expressed human milk). Exclusive breastfeeding was defined according to the World Health Organization definition as feeding only breast milk, with no other liquids or solids except vitamins, minerals, or prescribed medicines [23] since birth. Studies were eligible for inclusion if they explicitly defined the exposure as exclusive breastfeeding (including terms such as “breastfeeding only” or “breastmilk only”) and included a comparison group comprising mixed feeding, formula feeding, or non-exclusive breastfeeding. Studies that classified infants simply as breastfed without distinguishing exclusive breastfeeding from predominant, partial, or any breastfeeding were excluded.

The comparators are non- exclusive breastfeeding, mixed feeding, and/or formula feed. Partial breastfeeding or mixed feeding or complementary feeding, defined as breastmilk use along with any food or liquid, including non-human milk and formula or as defined by the author. Comparator groups included non-exclusive breastfeeding, mixed feeding, and/or formula feeding. Partial or mixed feeding was defined as the use of breast milk together with other foods or liquids, including formula or non-human milk, or according to the authors’ definitions. Formula feeding was defined as the exclusive use of commercial infant formula.

Feeding practices were assessed through interviews, questionnaires, clinical records, or maternal/parental recall up to 12 months postpartum. To ensure temporal relevance, exclusive breastfeeding had to be measured either concurrently with colic assessment or within the preceding three months in retrospective studies, as infantile colic is uncommon after nine weeks of age [1].

Search strategy

We searched the following databases: (1) Medline via Ovid (1946 to June 27, 2025), (2) EBM Reviews (1991 to June 25, 2025), (3) Embase (1947 to June 27, 2025), (4) Maternity & Infant Care Database (MIDIRS, 1971 to June 17, 2025), and (5) Global Health (1910 to 2025 Week 26). The search terms included study population (infant OR newborn OR neonat* OR babies OR baby), outcome of infantile colic (Colic* OR ‘infantile colic’ OR ‘infant colic’ OR gastric* or gastro* OR cry OR crying OR ‘excessive crying’ OR fuss*) and breastfeeding type as exposure (breastfe* or ‘breast feeding’ OR breast-fed OR ‘human milk’ OR breastmilk OR breast-milk OR ‘breast milk’ OR feed* OR fed OR nutrition OR diet*). We conducted the initial search on June 27, 2025, and updated it using the same search terms on March 8, 2026, which identified one additional study. Details of the search strategies and results for each database are provided in Supplement 1. We also conducted a manual search of reference lists but identified no additional relevant studies.

Article screening and selection

We used EndNote to remove duplicate records and imported the remaining studies into Covidence for screening according to the PRISMA framework. Two reviewers independently screened titles and abstracts against the predefined inclusion criteria to identify potentially eligible studies. Disagreements were resolved by a third reviewer. Full texts of potentially relevant articles were then independently assessed by two reviewers, with reasons for exclusion documented. Any discrepancies at this stage were also resolved by a third independent reviewer. All stages of study selection were conducted in Covidence.

Data were extracted into standardized Excel forms. One reviewer performed the primary data extraction, and a second reviewer checked the data for accuracy and completeness. Extracted information included publication year, first author, study design, study location, sample size, definitions and measurement time points for exclusive breastfeeding and comparator feeding practices, definitions and timing of colic assessment, and outcome data. We extracted either raw tabulated data or effect estimates, including odds ratios (ORs) and corresponding 95% confidence intervals (CIs), from both univariate and multivariate analyses.

Risk of bias assessment

Two independent reviewers assessed the risk of bias using the adapted Newcastle–Ottawa Scale (details provided in Supplement 2), as all included studies were observational in design [24]. The first author compared the assessments and resolved any disagreements.

Statistical analysis

Model selection and structure

Statistical analyses were performed in R (version 4.4.1) using the metafor package [25]. To examine whether the protective effect of exclusive breastfeeding varied by the type of alternative feeding, we conducted a moderator analysis using feeding type as the comparator [25]. Because included studies contributed multiple effect sizes (e.g., comparisons of exclusive breastfeeding with both mixed feeding and formula feeding within the same population), we used a multilevel random-effects model to account for the hierarchical structure of the data.

To address the non-independence of effect sizes nested within studies, we specified a model with random effects for both the study level (between-study variance) and the individual effect size level (within-study variance).

Parameters were estimated using Restricted Maximum Likelihood (REML). Effect sizes were analyzed as Log Odds Ratios (log OR) and back-transformed to Odds Ratios (OR) with 95% confidence intervals (CI) for reporting and visualization.

Heterogeneity analysis

Heterogeneity was quantified using the I² statistic, which represents the percentage of total variation across studies attributable to heterogeneity rather than chance. Both overall and subgroup specific I² values were calculated. We considered I² values greater than 80% to indicate substantial heterogeneity, 60%–80% to indicate moderate heterogeneity, and values below 60% to indicate low or unimportant heterogeneity [26].

Subgroup analysis

To explore potential sources of heterogeneity, subgroup analyses were conducted using multilevel random-effects meta-regression models. A range of study-level characteristics were examined as categorical moderators, including year of data collection (before 1990 vs. 1990 or later), study setting (hospital-based vs. community-based), study size (< 500, 500–1,499, vs. >1,500 participants), study design (cohort vs. cross-sectional), adjustment for confounders (yes vs. no), country income level, colic definition (crying for at least 3 h per day on at least 3 days per week vs. no specified duration of crying), and age at outcome assessment (< 6 months vs. 6–12 months). Each moderator was analysed separately within the same multilevel framework to assess effect modification. The models accounted for the hierarchical structure of the data by including random effects for study and effect size levels, allowing for multiple comparisons contributed by the same study. Pooled odds ratios and 95% confidence intervals were obtained for each subgroup using model-based predictions. Differences between subgroup levels were assessed using omnibus Wald-type tests (QM tests).

To examine the association between exclusive breastfeeding and infant colic across different feeding comparisons, we conducted subgroup analyses by comparator type: mixed feeding and formula feeding. For these analyses, the dataset was filtered to ensure each study contributed only one independent effect size per comparison. Consequently, individual random-effects models (rma) [25] were employed for each subgroup to estimate pooled effects, 95% confidence intervals, and subgroup-specific heterogeneity (I²).

Sensitivity analysis

To assess the robustness of the pooled estimates, leave-one-out sensitivity analyses were conducted for the overall model and all statistically significant subgroups [27]. In this iterative procedure, each study, including all associated effect sizes, was sequentially excluded to identify potential influential outliers and determine whether the findings were disproportionately driven by any single study.

Publication bias

Publication bias was assessed through visual inspection of a funnel plot, in which standard errors were plotted against log odds ratios. Funnel plot asymmetry was further evaluated using a multilevel adaptation of Egger’s regression test, with the standard error included as a moderator in the rma.mv model [28] A p-value < 0.05 was considered indicative of significant publication bias.

Results

Characteristics of included studies

We screened the titles and abstracts of 13,349 records and assessed 136 full-text articles for eligibility. Fifteen studies were ultimately included in the meta-analysis (Fig. 1).

Fig. 1.

Flow chart of study selection for the systematic review and meta-analysis of breastfeeding and infantile colic

The included evidence was primarily derived from prospective studies [29–38], followed by cross-sectional (4/15) [5, 39–41] and retrospective studies (1/15) [6]. Sample sizes varied substantially, ranging from 321 to 88,186 infants (Table 2). Most studies (11/15) were conducted in high-income Western countries [29–34, 37, 39–41] while four were conducted in Asian countries [5, 35, 36, 38].

Table 2.

Characteristics of included studies

No.	Author (Year)	Data Year	Country	Design	Target population	N	Data Collection Methods	Colic Definition	Age at Outcome	Exposure & Comparator	Adjusted for confounders	Quality Assessment
1	Traisman (1966) [29]	1958–1963	USA	Prospective cohort	Hospital-based	530	In-person interviews by health care providers	By mother, described as “severe crying”	1–6 months	EB since birth vs. Formula feeding only	No	Moderate
2	Rubin (1984) [30]	1979–1980	UK	Prospective cohort	Hospital-based	1,019	In-person interviews by health care providers	Listed colic as a health issue for infants for mothers to select	4 months	Breastfeeding vs. <4 months vs. Bottle feeding	No	Moderate
3	Thomas (1987) [31]	1982–1984	USA	Prospective cohort	Hospital-based	964	Phone interview	By mother, described as “recurrent episodes of unexplained crying and irritability”	≤ 17 weeks	EB vs. Formula feeding vs. Mixed feeding	No	Moderate
4	Howie (1990) [32]	1983–1986	UK	Prospective cohort	Community-based	674	In-person interviews by health care providers	By mother, described as “Intermittent attacks of abdominal pain when the baby screamed and drew up his or her legs but was well between episodes”	≤ 13 weeks	EB vs. Mixed or formula feeding	No	High
5	Crowcroft (1997) [39]	1975–1988	UK	Cross-sectional	Community-based	76,747	In-person interviews by health care providers	Asked parents whether their infant had been taken to a health care provider for colic.	24–37 days	EB vs. Mixed or formula feeding	Repair of housing, District or residence, Mother’s education, Father’s education, Parity, Mother’s age	High
6	Vander Wal (1998) [41]	1995–1996	Netherlands	Cross-sectional	Community-based	1,826	In-person interviews by health care providers	Colic defined excessive crying at least 3 h per day	2–3 months	EB vs. Mixed or formula feeding	Sex, parity, ethnic of origin, and education of mothers	High
7	Castro-Rodriguez (2001) [33]	1980	USA	Prospective cohort	Hospital-based	983	In-person interviews by health care providers	Identified by a physician	2 months	EB vs. Mixed or formula feeding	No	High
8	Canivet (2008) [34]	1997	Sweden	Prospective cohort	Hospital-based	1,620	Phone interview	Colic defined excessive crying at least 3 h per day, 3 days per week	4½–5½ weeks	EB vs. Mixed or formula feeding	Age	High
9	Talachian (2008) [35]	2013	Iran	Prospective cohort	Hospital-based	321	In-person interviews by health care providers	Colic defined excessive crying at least 3 h per day, 3 days per week	12 weeks	EB in 12 weeks (yes vs. no)	No	Moderate
10	Fazil M (2011) [36]	2008	Pakistan	Prospective cohort	Hospital-based	426	In-person interviews by health care providers	Colic defined excessive crying at least 3 h per day, 3 days per week	12 weeks	EB in 12 weeks (yes vs. no)	No	High
11	Helseth (2022) [6]	1999–2008	Norway	Retrospective cohort	Community-based	88,186	In-person self-administered questionnaire	Listed colic as a health issue for infants for mothers to select	6 months	EB 0–3 months (yes vs. no)	No	Low
12	Murphy (2023) [37]	2008–2009	Ireland	Prospective cohort	Community-based	41,185	In-person interviews by health care providers	Asked parents whether their infant had been taken to a health care provider for colic.	9 months	EB 90–180 days (yes vs. no)	No	Low
13	Bante (2024) [5]	2022	Ethiopia	Cross-sectional	Community-based	449	In-person interviews by health care providers	Colic defined excessive crying at least 3 h per day, 3 day per week within 3 weeks	6–12 months	EB up to 6 months (yes vs. no)	Educational status, Perceived Social Support, feeding problems, and maternal Depression, Anxiety, and Stress symptoms	High
14	Garnacho-Garnacho (2024) [40]	2024	Spain	Cross-sectional	Community-based	528	Online self-administered questionnaire	Infant Colic Severity Questionnaire (ICSQ); score ≥ 50 = colic	0–12 months	EB vs. Mixed or non-breastfeeding	No	Moderate
15	Doan et al. (2026) [38]	2020–2021	Vietnam	Prospective cohort	Hospital-based	856	Phone interview	Colic defined excessive crying at least 3 h per day, 3 days per week	3–21 weeks	EB since birth vs. Mixed feeding	Maternal age, maternal education, prepregnancy body mass index, parity, gestational age at birth, mode of delivery, birthweight, sex of infants, age of the child, and randomized group	High

Abbreviation: EB Exclusive breastfeeding

Eight studies recruited participants from hospitals [29–31, 33–36, 38] whereas the remaining studies were community-based [5, 6, 32, 37, 39–41]. Most studies (12/15) included infants younger than 6 months [6, 29–36, 38, 39, 41], two included infants aged 6–12 months, and one included infants across the full 0–12 month age range. Data collection methods varied: 10 studies used in-person interviews conducted by healthcare providers [5, 29, 30, 32, 33, 35–37, 39, 41], three used telephone interviews [31, 34, 38], and two used self-administered questionnaires, including one in person [6] and one online survey [40].

Infant colic was assessed using a variety of methods. One study used a diary-based assessment [35], one relied on physician diagnosis [33] and the remaining studies used maternal reports. Seven studies defined colic as crying for three or more hours per day on at least three days, although the reference periods varied (e.g., during the past week, over three weeks, or since birth) [5, 34–36, 38, 41]. Three studies assessed colic using maternal responses to questions describing typical symptoms, such as “severe crying,” “recurrent episodes of unexplained crying and irritability,” or “intermittent attacks of abdominal pain” [29, 31, 32]. In another three studies, colic was included as an infant health condition for mothers to identify from a list [6, 12, 30]. One study used the Infant Colic Severity Questionnaire (ICSQ) to diagnose colic [40], while another identified colic based on whether parents sought healthcare for their infant’s symptoms [37].

Data on exclusive breastfeeding were collected using either point-in-time assessments or maternal recall, except for one study that used serial measurements over time [38]. Definitions of exclusive breastfeeding were often omitted, and few studies specified the feeding method (e.g., direct breastfeeding versus bottle-feeding expressed breast milk). Eight studies compared exclusive breastfeeding with formula feeding [29–34, 39–42], nine compared with mixed feeding [30–34, 38–41], and six compared with non-exclusive breastfeeding [5, 6, 35–37].

Risk of bias assessment

Overall, eight studies were rated as high quality, five as moderate or unclear quality, and two as low quality (Table 2). Across the 15 included studies, methodological quality was consistently high for sample representativeness and sample size, with all studies judged to have representative samples and adequate sample sizes.

However, study comparability was limited, as 10 studies did not adequately adjust for potential confounders [6, 29–33, 35–37, 40] whereas five studies did [5, 34, 38, 39, 41]. Ascertainment of feeding exposure was generally robust, with 11 studies rated as high quality [29–36, 38, 39, 41] while four were rated as unclear because of insufficient methodological detail [5, 6, 37, 40]. Response rates were considered satisfactory in nine studies [5, 30–32, 34, 36, 38, 39, 41] six studies were rated as low quality in this domain [6, 29, 33, 35, 37, 40]. Outcome assessment showed greater variability: seven studies were rated as high quality [5, 33–36, 38, 40], five as moderate quality [29, 31, 32, 37, 41], and three as low quality [6, 30, 39]. Statistical methods were appropriately described and applied in nine studies [5, 6, 32–34, 38–41] while six studies had limitations in statistical reporting or analysis [29–31, 35–37].

Association between exclusive breastfeeding and infantile colic

A total of 15 studies, comprising 23 distinct comparisons including 216,314 infants, were included in the meta-analysis. Among these, eight studies contributed data to multiple comparison groups. As shown in Fig. 2, exclusively breastfed infants had 19% lower odds of colic than non-exclusively breastfed infants (pooled OR = 0.81, 95% CI: 0.67–0.97; p = 0.021).

Fig. 2.

Forest plot of fixed-effect and random-effects meta-analyses comparing infant colic between exclusively breastfed and non-exclusively breastfed infants

Significant residual heterogeneity was observed (Q = 184.3, df = 22, p < 0.0001), supporting the use of a random-effects model. Variance partitioning in the multilevel analysis showed that 67.0% of the total variance was attributable to differences between-study, while 14.7% was due to variation among effect sizes within studies. The remaining 18.3% was attributed to sampling error. Overall heterogeneity was substantial, with a total I² of 92.6%.

Sensitivity analysis and publication bias

Visual inspection of the funnel plot suggested a symmetrical distribution of effect sizes (Fig. 3). This was supported by a multilevel adaptation of Egger’s regression test, which showed no significant association between study precision (standard error) and observed effect sizes (z = − 0.025, p = 0.974).

Fig. 3.

Funnel plot assessing publication bias in studies examining the association between exclusive breastfeeding and infant colic

Leave-one-out sensitivity analyses demonstrated the robustness of the findings (Fig. 4). In all 15 iterations, the association between exclusive breastfeeding and lower odds of infant colic remained statistically significant (p < 0.05). The pooled odds ratio varied only slightly, ranging from 0.77 to 0.84 around the overall estimate of approximately 0.81.

Fig. 4.

Leave-one-out sensitivity analyses of studies included in the meta-analysis comparing colic between exclusively breastfed and non-exclusively breastfed infants

Subgroup analyses (Table 3) showed that exclusive breastfeeding was associated with a significantly lower likelihood of infant colic among studies conducted in 1990 or later (pooled OR = 0.68, 95% CI: 0.60–0.78; p < 0.001), community-based studies (pooled OR = 0.76, 95% CI: 0.59–0.99; p = 0.042), studies including more than 1,500 participants (pooled OR = 0.73, 95% CI: 0.55–0.98; p = 0.039), and studies that adjusted for potential confounding factors (pooled OR = 0.67, 95% CI: 0.49–0.93; p = 0.015). Differences of pool effect between subgroup levels, however, were only found between the study conducted before or after 1990 (p of QM tests < 0.001), no other characteristics including study setting, study size, study design, control for cofounding, country income level, categorization of colic definition, or age at infantile colic assessment.

Table 3.

 Exclusive breastfeeding and the odds of having colic: random- effects meta- analyses of the odds ratio of colic by subgroup

Subgroup analysis	Number of studies	Pooled OR (95% CI)	p- value	QM test p-value
Data collection time
Before 1990	6	1.07 (0.93–1.22)	0.340	< 0.0001
After 1990	9	0.68 (0.60–0.78)	< 0.001
Study setting
Hospital based	8	0.85 (0.65–1.11)	0.240	0.575
Community based	7	0.76 (0.59–0.99)	0.042
Study size
< 500 participants	3	1.05 (0.63–1.75)	0.860	0.498
500– 1499 participants	7	0.81 (0.61–1.08)	0.147
≥ 1500 participants	5	0.73 (0.55–0.98)	0.039
Study design
Cohort	11	0.82 (0.66–1.03)	0.091	0.724
Cross- sectional	4	0.76 (0.53–1.09)	0.136
Control for confounding
No	10	0.88 (0.70–1.11)	0.276	0.174
Yes	5	0.67 (0.49–0.93)	0.015
Country income level
High- income country	10	0.83 (0.67–1.03)	0.097	0.539
Middle- /low- income country	5	0.73 (0.51–1.05)	0.092
Categorisation of colic definition*
Crying at least 3 h a day and 3 days a week	8	0.77 (0.57–1.04)	0.085	0.692
No specify duration of crying	6	0.83 (0.65–1.06)	0.133
Age at outcome assessment**
Less than 6 months	12	0.84 (0.68–1.04)	0.119	0.506
6–12 months	2	0.70 (0.42–1.17)	0.172

The total number of estimates does not add up to 16:

*1 study defined colic based on scale

**1 study provided estimates on the association of breastfeeding with colic at different age categories (0–12 months),

Effect of exclusive breastfeeding on infantile colic by comparison type

To explore sources of heterogeneity, subgroup analyses were conducted by comparator type (mixed feeding or formula feeding) using multilevel models (Fig. 5). Although the omnibus test for moderators was not statistically significant (Q_M = 3.98, df = 2, p = 0.137), the subgroup analyses showed differing patterns of association. Exclusive breastfeeding was associated with significantly lower odds of infant colic when compared with mixed feeding (OR = 0.72, 95% CI: 0.57–0.93; p = 0.01). In contrast, the association was weaker and no longer statistically significant when exclusive breastfeeding was compared with formula feeding (OR = 0.92, 95% CI: 0.71–1.17; p = 0.470).

Fig. 5.

Forest plot of fixed-effect and random-effects meta-analyses comparing colic between exclusively breastfed infants and those receiving mixed feeding or formula feeding

Consistent with these findings, leave-one-out sensitivity analyses showed that the association between exclusive breastfeeding and mixed feeding remained statistically significant (p < 0.05) across all nine iterations, with pooled ORs ranging from 0.65 to 0.78 (Fig. 6).

Fig. 6.

Leave-one-out sensitivity analyses of studies included in the meta-analysis comparing colic between exclusively breastfed and mixed-fed infants

Discussion

This multilevel meta-analysis, covering studies published between 1966 and 2026, found that exclusive breastfeeding was associated with significantly lower odds of infantile colic. The primary analysis yielded a pooled OR of 0.81 (95% CI: 0.67–0.97; p = 0.021), indicating an overall inverse association across the included studies. Subgroup analyses demonstrated a stronger association in studies that adjusted for potential confounding factors (pooled OR = 0.67, 95% CI: 0.49–0.93; p = 0.015), included at least 1,500 participants (pooled OR = 0.73, 95% CI: 0.55–0.98; p = 0.039), and were community-based (pooled OR = 0.76, 95% CI: 0.59–0.99; p = 0.042). A stronger association was also observed when exclusive breastfeeding was compared with mixed feeding (OR = 0.72, 95% CI: 0.57–0.93; p = 0.010). In contrast, no statistically significant association was observed when exclusive breastfeeding was compared with formula feeding (OR = 0.92, 95% CI: 0.71–1.17; p = 0.470).

Although overall heterogeneity was substantial (I² = 92.6%), the multilevel variance partitioning provided a structured explanation for this variability. We found that 67.0% of the total variance was attributable to between-study differences, likely reflecting the diverse geographical and clinical settings represented across six decades of research. An additional 14.7% of the variance arose from differences within studies, indicating that different feeding comparisons within the same population may produce varying results. The remaining 18.3% was attributable to random sampling error, suggesting that most variability reflected meaningful systematic differences rather than chance alone. This variance structure supports the use of a multilevel random-effects model and suggests that, although the magnitude of the association varied across contexts, the overall direction of effect remained consistently protective.

Several biological mechanisms may explain the observed association between exclusive breastfeeding and infantile colic. A growing body of evidence suggests that gut microbiota, inflammation, and the microbiota–gut–brain axis may contribute to the development of infantile colic [43–46]. Breast milk plays a center role in establishing the infant gut microbiota, and these effects appear to be dose dependent [47, 48]. The composition of breast milk microbiota changes throughout lactation and differs between mothers who exclusively breastfeed and those who do not [49]. Consequently, infant gut microbiota profiles vary according to feeding practices. Previous meta-analyses have shown that non-exclusively breastfed infants tend to exhibit greater bacterial diversity and higher levels of Bacteroidetes and Firmicutes, resulting in a gut microbiota profile that more closely resembles that of adults compared with exclusively breastfed infants [50]. Differences in gut microbiota between breastfed and formula-fed infants are also well established [51, 52]. Microbial composition can change rapidly during transitions between breastfeeding and formula feeding [53]. The stronger association observed for exclusive breastfeeding versus mixed feeding than versus formula feeding may therefore reflect microbial alterations associated with feeding transitions rather than formula feeding itself. However, the biological pathways underlying these associations remain incompletely understood and warrant further investigation.

Our findings also provide additional support for the promotion of exclusive breastfeeding, as it was associated with lower odds of infantile colic. Although the etiology of infantile colic remains uncertain and its management continues to be debated, breastfeeding is widely recognized for its substantial short- and long-term health benefits for both infants and mothers. Some excluded studies were generally consistent with our results, suggesting lower odds of colic among infants who were exclusively breastfed during the first week of life [12], before colic is typically reported, or during the first 4–8 weeks after birth, when colic prevalence usually peaks [54–56]. These findings suggest that breastfeeding support may be especially important during the early postpartum period, when feeding practices are being established before the typical onset of colic symptoms. In addition, maintaining or transitioning to exclusive breastfeeding after temporary early formula supplementation may still provide benefits compared with continued mixed feeding [38]. Dietary modification, particularly a low-allergen diet excluding cow’s milk, eggs, peanuts, tree nuts, wheat, soy, and fish, may further reduce infant crying duration among breastfeeding mothers [57–59]. Probiotic supplementation has also been associated with reduced crying duration, although its effects appear less effective among infants receiving mixed or formula feeding [60]. Together, these findings suggest that interventions supporting exclusive breastfeeding and optimizing early microbial development may help reduce the burden of infantile colic.

The lack of a significant association between exclusive breastfeeding and infantile colic when compared with formula feeding should be interpreted cautiously. First, infantile colic is a multifactorial condition influenced by a complex interplay of biological, environmental, and psychosocial factors. Therefore, feeding methods alone may not be sufficient to substantially alter colic risk. Second, although breastfeeding and formula feeding are associated with distinct gut microbiota profiles, these differences may not necessarily translate into clinically meaningful differences in colic risk. In contrast, mixed feeding introduces frequent changes in feeding exposures and microbial colonization patterns, which may result in greater instability of the developing gut microbiota and microbiota–gut–brain axis. This hypothesis is consistent with our finding that exclusive breastfeeding was associated with lower odds of colic compared with mixed feeding but not with exclusive formula feeding. Third, residual confounding cannot be excluded, as parental characteristics, feeding difficulties, maternal diet, infant temperament may differ across feeding groups and influence both feeding practices. Finally, exposure and outcome misclassification may also have contributed to the null findings. Evidence from studies using the International Atomic Energy Agency (IAEA) deuterium oxide dose-to-mother technique indicates that maternal reports may overestimate the prevalence of exclusive breastfeeding compared with objectively measured feeding practices [61]. Infantile colic was commonly identified through caregiver reports rather than physician-confirmed diagnoses or diary-based assessments of crying duration based on Barr’s definition [62] or the Rome IV criteria [3]. Despite the absence of significant differences according to the definition of infant colic or the age at outcome assessment, such non-differential misclassification of both feeding status and colic outcomes may have diluted true associations, thereby biasing effect estimates towards the null and reducing the ability to detect significant differences between feeding groups.

Strength and limitation

A major strength of this study was the comprehensive search strategy, which aimed to identify all relevant studies and reduce reporting bias. We also excluded studies that did not clearly define appropriate time points for measuring exclusive breastfeeding exposure, thereby improving the validity of the evaluation. In addition, the non-significant Egger’s regression test suggested that the pooled estimates were unlikely to be affected by publication bias or small-study effects. Leave-one-out sensitivity analyses further demonstrated the robustness of the findings, as statistical significance was maintained across all iterations for both the overall model and the mixed-feeding subgroup. Subgroup analyses indicated that the pooled effect varied only by the year of data collection and remained consistent across all other study-level characteristics supports the robustness of the overall association.

However, several limitations should be considered when interpreting these findings. In addition to the potential exposure and outcome misclassification discussed above, the duration and assessment of exclusive breastfeeding varied substantially across studies. As a result, some infants classified as exclusively breastfed may have transitioned to mixed or formula feeding during the early weeks of life, but these changes in feeding practices could not be consistently verified. This may have also led to exposure misclassification and attenuated the observed associations. The association was significant among studies that adjusted for potential confounding factors such as infant age, mode of delivery, prematurity, maternal smoking, maternal education [1], suggesting that uncontrolled confounding is unlikely to fully explain the observed inverse association. However, because the primary meta-analysis combined crude and adjusted estimates, residual confounding cannot be ruled out. Finally, because all included studies were observational in design, residual confounding cannot be excluded, and a causal relationship between exclusive breastfeeding and infantile colic cannot be established.

Conclusion

The meta-analysis included 15 studies involving more than 216,000 infants, providing a comprehensive synthesis of the available evidence. Overall, the findings suggest that exclusive breastfeeding is associated with lower odds of infantile colic, particularly when compared with mixed feeding and among studies with adjustment for potential confounding, larger sample sizes, and community-based recruitment. These findings support the hypothesis that exclusive breastfeeding during early infancy may confer benefits for infant gastrointestinal function and may reduce the occurrence of infantile colic. Nevertheless, the results should be interpreted with caution because of substantial between-study heterogeneity.

Authors’ contributions

Dr Duong TT. Doan and Prof Colin Binns conceptualized and designed the study, screened articles, extracted data, coordinated and supervised articles screening and extraction, drafted the initial manuscript, and critically reviewed and revised the manuscript. Dr Thuy P. Nguyen searched literature, designed the data collection instruments, conducted the analyses, and critically reviewed and revised the manuscript. Dr Lam TK. Le, Ngoc NH. Tran, and Thien H. Dang, screened articles, designed the data collection instruments, extraction, drafted the initial manuscript, and critically reviewed and revised the manuscript. All authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.

Funding

No funding was secured for this study.

Data availability

All data generated or analyzed during this study are included in this published article and its supplementary information files.

Declarations

Ethics approval and consent to participate

Not applicable.

Competing interests

The authors declare no competing interests.

Consent for publication

Not applicable.