Peer support intervention (ABA-feed) to improve breastfeeding: UK based, multicentre, parallel group, randomised controlled trial.
Jolly K, Clarke J, Crossland N, Dombrowski SU, Gkini E, Hoddinott P, Ingram J, Johnson D, MacArthur C, Mann M, McKell J, Moss N, Roberts T, Sanders J, Savory N, Sitch A, Taylor B, Tearne S, Thomson G, Williams E, Woolley R, ABA-feed study group
- DOI
- 10.1136/bmj-2025-086558
- Record issued
- 2026-08-10
- Engine
- 7.29.0
- Exported
- 2026-09-22
Prepared by Alpha1. This document is confidential: it is intended for the recipient it was shared with and must not be redistributed. The live record at alpha1science.com/verify/0d742703-9d4f-4e71-9753-e26706c241cb is authoritative.
How this rating was calculated
- IntegrityIntegrity concern−0.5★
- ReportingEthical approvals partially met−0.25★
- References were not verified against Crossref/OpenAlex.
- The numeric-impossibility checks (GRIM/GRIMMER/DEBIT/SPRITE) did not run on this paper: the pass that reads its reported means did not complete. No reported mean was checked for arithmetic impossibility.
This Kaimen Rigor review uses Kaimen Rigor reviewers trained on a curated corpus of high-fidelity and retracted papers, with expert supervision and curation. It can still make mistakes; verify each finding against the source before relying on it.
The paper is a well-conducted and transparently reported RCT with a clear premise, robust design, and comprehensive statistical analysis. The main rigor gaps are the absence of an explicit regulatory compliance statement (e.g., Declaration of Helsinki) and a reference to a reporting guideline, both minor omissions. The copyedit issues are minor.
Both independent reviewers agreed on all dimensions except for a minor sub-criterion disagreement on biological variables (age weight health), which was resolved by considering the study context. The evaluation is based on the full published paper.
Numerical inconsistencies
1 finding · worst lowValues that contradict each other or are impossible for the stated sample: recomputed p-values and test statistics, GRIM/GRIMMER checks on summary numbers, percentages against their own counts, totals against their parts, and estimates against their own confidence intervals.
- Internal contradictions in the reported numbersAssessed
Recomputed 7 tests: 7 consistent, 0 inconsistent; 7 via agent-written checks.
- CONSISTENTreported p = .770 · recomputed p = .694Reviewer 2Primary outcome risk ratio p-value from CI
“Any breastfeeding at 8 weeks after birth | 1013 (69.8) | 698 (68.8) | 1.01 (0.96 to 1.06; 0.77)”
Taken as given: The 1.01, 0.96, and 1.06 are the risk ratio and its two-sided 95% CI lower and upper bounds for the same outcome.; The CI is two-sided at 95% (normal approximation to the model-based estimate).; The log scale is used for a ratio estimate.Method: Two-tailed normal-approximation p derived from estimate and 95% CI on the log scale (pCI).How we recomputed it: pCI(1.01, 0.96, 1.06, 1) - CONSISTENTreported p = .690 · recomputed p = .575Reviewer 2Primary outcome risk difference p-value from CI
“0.01 (−0.03 to 0.04; 0.69)”
Taken as given: The 0.01, -0.03, and 0.04 are the risk difference and its two-sided 95% CI lower and upper bounds for the same outcome.; The CI is two-sided at 95% (normal approximation).; Linear (difference) scale is used.Method: Two-tailed normal-approximation p from estimate and 95% CI on the linear scale (pCI).How we recomputed it: pCI(0.01, -0.03, 0.04, 0) - CONSISTENTreported p = .480 · recomputed p = .494Reviewer 2Hazard ratio for time to cease any breastfeeding from CI
“Median (IQR) time to cease any breastfeeding <16 weeks (days) | 21 (2-49) | 16.5 (1-42) | 0.95‡ (0.82 to 1.10; 0.48)”
Taken as given: The 0.95, 0.82, and 1.10 are the hazard ratio and its two-sided 95% CI lower and upper bounds.; The CI is two-sided at 95% (normal approximation on log scale).Method: Two-tailed normal-approximation p from estimate and 95% CI on the log scale (pCI).How we recomputed it: pCI(0.95, 0.82, 1.10, 1) - CONSISTENTreported p = .004 · recomputed p = .004Reviewer 2Social support (Medical Outcomes Study) median difference p from CI
“90.6 (75.0-100) | 84.4 (71.9-100) | 6.25 (1.99 to 10.51; 0.004)”
Taken as given: The 6.25, 1.99, and 10.51 are the median difference and its two-sided 95% CI lower and upper bounds.; The CI is two-sided at 95% (normal approximation on the difference scale).Method: Two-tailed normal-approximation p from estimate and 95% CI on the linear scale (pCI).How we recomputed it: pCI(6.25, 1.99, 10.51, 0) - CONSISTENTreported p = .040 · recomputed p = .041Reviewer 2GAD-7 median difference p from CI
“median difference −1.0 (95% CI −1.96 to −0.04, P=0.04)”
Taken as given: The -1.0, -1.96, and -0.04 are the median difference and its two-sided 95% CI lower and upper bounds.; The CI is two-sided at 95% (normal approximation on the difference scale).Method: Two-tailed normal-approximation p from estimate and 95% CI on the linear scale (pCI).How we recomputed it: pCI(-1.0, -1.96, -0.04, 0) - CONSISTENTreported p = .002 · recomputed p = .002Reviewer 2Medical Outcomes Study mean difference p from CI
“8 weeks | 83.1 (19.6); 1257 | 80.4 (21.6); 855 | 2.53 (0.95 to 4.11; 0.002)”
Taken as given: The 2.53, 0.95, and 4.11 are the adjusted mean difference and its two-sided 95% CI lower and upper bounds.; The CI is two-sided at 95% (normal approximation on the difference scale).Method: Two-tailed normal-approximation p from estimate and 95% CI on the linear scale (pCI).How we recomputed it: pCI(2.53, 0.95, 4.11, 0) - CONSISTENTreported p = .940 · recomputed p = .928Reviewer 2Hospital admission under 16 weeks risk ratio p from CI
“165/1149 (14.4) | 116/801 (14.5) | 0.99 (0.80 to 1.24); 0.94”
Taken as given: The 0.99, 0.80, and 1.24 are the risk ratio and its two-sided 95% CI lower and upper bounds.; The CI is two-sided at 95% (normal approximation on log scale).Method: Two-tailed normal-approximation p from estimate and 95% CI on the log scale (pCI).How we recomputed it: pCI(0.99, 0.80, 1.24, 1)
- lowinternal contradictionThe breastfeeding-initiation risk ratio 95% CI lower bound is exactly 1.00, which under a normal approximation implies a two-sided p near 0.05, yet the reported p is 0.13; the accompanying risk-difference CI (−0.004 to 0.04) is consistent with p=0.13, suggesting a rounding artifact at the model boundary rather than an error.
“Breastfeeding initiation | 1239 (94.2) | 844 (92.5) | 1.02 (1.00 to 1.04; 0.13)”
Table 3Find in source
Overstated conclusions
1 finding · worst lowConclusions that reach past what the paper's own results support — including a significance claim that no longer holds when the statistic is recomputed, and efficacy resting on an unvalidated surrogate endpoint.
- Conclusions only partially backed by the presented evidenceAssessed
7 major claims checked against the paper's own evidence: all adequately supported.
- partialReviewer 2No randomised controlled trial of breastfeeding peer support in the UK has shown peer support to be effective.This is a claim about the prior literature, not demonstrated by this paper's own data; it is asserted without a systematic citation of all UK trials in this manuscript.Evidence: No direct evidence presented in this paper beyond the author's assertion; the paper's own null result is consistent with it.
“No randomised controlled trial of breastfeeding peer support in the UK has shown peer support to be effective.”
DiscussionFind in source - supportedReviewers 1, 2The ABA-feed peer support intervention did not improve breastfeeding rates compared with usual breastfeeding support in a UK context.The primary outcome result (risk difference 0.01, 95% CI -0.03 to 0.04, P=0.69) directly supports this claim.Evidence: Primary outcome: any breastfeeding at 8 weeks: 69.8% intervention vs 68.8% usual care; risk difference 0.01, CI -0.03 to 0.04, P=0.69.
“Rates of any breastfeeding at eight weeks did not differ between the intervention group (1013/1452; 69.8%) and usual care group (698/1015; 68.8%); adjusted risk difference 0.01, 95% confidence interval −0.03 to 0.04.”
AbstractFind in source - supportedReviewers 1, 2At eight weeks the intervention group reported higher social support, but this was not sustained at 16 weeks.The paper reports a significant difference in Medical Outcomes Study score at 8 weeks (median difference 6.25, 95% CI 1.99 to 10.51, P=0.004) but not at 16 weeks (P=0.18).Evidence: Table 4: Medical Outcomes Study score at 8 weeks: median difference 6.25 (95% CI 1.99 to 10.51, P=0.004); at 16 weeks: median difference 3.13 (95% CI -1.45 to 7.70, P=0.18).
“Social support using the Medical Outcomes Study score was higher in the intervention group (90.6 (75.0-100) v 84.4 (71.9-100); also assessed using the bootstrapping method owing to the highly skewed data; median difference 6.25 (95% CI 1.99 to 10.51, P=0.004). These significant differences were not maintained at the 16 week postnatal time point.”
ResultsFind in source - supportedReviewers 1, 2The findings would not support the commissioning of universal high intensity one-to-one peer support interventions.The null primary result and description of the intervention's intensity (median 19 contacts) support the conclusion that such a universal intervention is not effective.Evidence: Primary outcome result and discussion of implications.
“The implications are that universal intensive peer support for infant feeding would not be recommended in the UK context.”
DiscussionFind in source - supportedReviewer 1Subgroup analyses suggested possible benefit in women with lower education and higher deprivation, but these were not significant.The paper reports absolute differences of 6-9% in these subgroups, but interaction tests were not statistically significant.Evidence: Results: 'absolute differences of 6-9% were seen in the primary outcome between the intervention and usual care groups in subgroups of women with GCSE or A level as highest education level, living in areas in the most deprived fifth... There were no statistically significant interactions.'
Subgroup analyses were performed and absolute differences of 6-9% were seen in the primary outcome between the intervention and usual care groups in subgroups of women with GCSE or A level as highest education level, living in areas in the most deprived fifth of index of multiple deprivation, and who planned to mixed feed... There were no statistically significant interactions between the intervention and subgroups.
Resultsreviewer’s wording - supportedReviewer 1The intervention did not improve formula feeding practices.The paper reports that adherence to formula preparation recommendations was low and similar between groups.Evidence: Results: 'When formula preparation was relevant in the intervention and usual care groups, adherence to recommendations was low for correct water temperature (45.4% v 49.3%) and making up formula to use away from the home (34.6% v 32.8%).'
“When formula preparation was relevant in the intervention and usual care groups, adherence to recommendations was low for correct water temperature (45.4% v 49.3%) and making up formula to use away from the home (34.6% v 32.8%).”
ResultsFind in source - supportedReviewer 2Anxiety scores were lower in the intervention group at eight weeks but not sustained at 16 weeks.The 8-week median difference (−1.0, 95% CI −1.96 to −0.04, P=0.04) is significant and the 16-week comparison is null, consistent with the claim.Evidence: GAD-7 median 3.0 vs 4.0 at 8 weeks; at 16 weeks medians equal (3.0 vs 3.0, median difference 0.0).
Anxiety (GAD-7 score at eight weeks) was significantly lower in the intervention group (median 3.0 (IQR 1.0-6.0)) compared with the usual care group (4.0 (1.0-7.0))... median difference −1.0 (95% CI −1.96 to −0.04, P=0.04).
Resultsreviewer’s wording
Data authenticity concerns
None foundAn adversarial read for patterns associated with data that may not be genuine: results that look too clean, implausibly large effects, duplicated data or images, and methods that do not match the results reported.
Checked — nothing surfaced.
Reporting gaps
1 finding · worst mediumRequired detail the manuscript never states — study design, biological variables, ethics approval and consent, key resources, statistical reporting, data and code availability, and overall transparency.
- Ethics/consent reporting incompleteAssessed
The introduction cites the 2022 Cochrane review on breastfeeding support, notes that previous UK trials of peer support showed no benefit, and identifies possible reasons (low intensity, lack of proactive support). The ABA-feed intervention was developed to address these issues. Strengths and weaknesses of prior work are acknowledged, and the rationale for the study is logically linked to the evidence.
“Despite international evidence for the effectiveness of peer support on any and exclusive breastfeeding rates, several large randomised controlled trials of peer support in the UK did not show a beneficial effect.”
“Several large randomised controlled trials of peer support in the UK did not show a beneficial effect. Possible reasons include low intensity interventions, lack of timely support after the birth, and the need for a new parent to initiate contact.”
“The ABA-feed (Assets based feeding help Before and After birth-feed) intervention of infant feeding peer support was developed to deal with these issues”
Randomisation was via a secure central web-based system with a 1.43:1 ratio and minimisation for site and age. The trial is unblinded, which is appropriate for a pragmatic behavioural intervention. A detailed sample size calculation is provided (90% power, 5% significance, ICC 0.039). Inclusion/exclusion criteria are clearly defined. Handling of missing data (imputed as formula feeding) and outliers (sensitivity analyses, bootstrapping) are described.
“To ensure concealment of treatment allocation, the Birmingham Clinical Trials Unit used a secure, central, web based randomisation system to allocate women in a 1.43:1 ratio to receive intervention or usual care.”
“Assuming 90% power and a two sided 5% significance level, with a control group rate of 44% for the primary outcome (95% CI 30.0% to 58.7%; from ABA feasibility data), a sample size of 2136 women (1068 in each group) was required to detect a risk ratio of 1.16 (ie, an increase of 7%).”
“Regardless of their feeding intention, women were eligible for recruitment if they were nulliparous, between 20 weeks and 35 weeks and six days gestation with a singleton pregnancy, aged 16 years or older, and lived in the study area.”
“the Birmingham Clinical Trials Unit used a secure, central, web based randomisation system to allocate women in a 1.43:1 ratio to receive intervention or usual care”
“Missing data were imputed as formula feeding for the primary outcome, excluding pregnancy losses, still births, infant deaths, or maternal deaths.”
All participants were nulliparous women (sex inherent to the breastfeeding population), with mean age 30.6 years reported in both arms. Demographics are extensive (ethnicity, education, employment, family status, deprivation index, feeding intentions). Health/gestational status is defined by eligibility (20-35 weeks gestation, singleton pregnancy). Sex justification is inherent to a breastfeeding trial, so marked not applicable. Species/housing criteria do not apply to human participants.
“Mean age of the participants was 30 years, and most described their ethnicity as white (2132; 86.5%)”
“women were eligible for recruitment if they were nulliparous, between 20 weeks and 35 weeks and six days gestation with a singleton pregnancy, aged 16 years or older”
The trial was approved by a named ethics committee (East of Scotland Research Ethics Committee, reference 21/ES/0045). Informed consent procedures are described in detail. However, no statement of adherence to the Declaration of Helsinki or other regulatory framework is provided, which is a minor omission for a clinical trial.
“This trial was approved by the East of Scotland Research Ethics Committee (21/ES/0045).”
“Informed consent was obtained from each participant in person or by telephone or video call.”
“This trial was approved by the East of Scotland Research Ethics Committee (21/ES/0045).”
“Informed consent was obtained from each participant in person or by telephone or video call.”
The intervention is peer support (behavioural). No investigational medicinal product, device, or biological reagent is used. The peer supporters are volunteers, not a resource to be catalogued. Software tools (SAS, Stata) are identified in the statistical analysis section, but this is a software tool for analysis, not a key resource for the intervention. The trial meets the criteria for not_applicable.
The paper names all statistical tests (mixed effects log binomial, Cox proportional hazards, mixed effects linear regression, bootstrapping). Exact p-values are reported (e.g., P=0.69, P=0.13). Effect sizes are reported with 95% confidence intervals. Software (SAS 9.4, Stata 18) is identified. Data presentation includes per-group n, percentages, means, medians, IQRs, and CIs. No mathematical implausibilities were detected. Assumptions are handled through model choice and sensitivity analyses.
“Any breastfeeding at 8 weeks after birth | 1013 (69.8) | 698 (68.8) | 1.01 (0.96 to 1.06; 0.77) | 0.01 (−0.03 to 0.04; 0.69)”
“Statistical analysis was undertaken using the statistical software packages SAS version 9.4 and Stata version 18.”
“Statistical analysis was undertaken using the statistical software packages SAS version 9.4 and Stata version 18.”
“69.8% v 68.8%; risk difference 0.01, 95% CI −0.03 to 0.04, P=0.69”
“All percentages presented do not include missing data.”
The paper states that deidentified individual participant data are stored at a university repository (https://edata.bham.ac.uk/1555/) and are accessible for ethically approved research after registration. This is a concrete route. The statistical analysis code is included in the supplementary file. No accession numbers are needed for this type of data.
“Deidentified individual participant data stored at https://edata.bham.ac.uk/1555/ and are fully accessible for ethically approved research after registration with UBIRA eData.”
“The SAS version 9.4 and Stata version 18 statistical analysis code is included in the supplementary file.”
“Deidentified individual participant data stored at https://edata.bham.ac.uk/1555/ and are fully accessible for ethically approved research after registration with UBIRA eData.”
“The SAS version 9.4 and Stata version 18 statistical analysis code is included in the supplementary file.”
The trial is registered (ISRCTN17395671). Methods are detailed enough for replication. All pre-specified outcomes are reported, including non-significant results. Limitations and strengths are discussed in a dedicated section. Conclusions are proportional to the evidence. The paper does not explicitly state adherence to a reporting guideline, which is a minor omission.
“Trial registration ISRCTN Registry ISRCTN17395671.”
“A limitation was that we did not fully achieve recruitment numbers to reach 90% power but did exceed the target for 80% power.”
“Trial registration ISRCTN Registry ISRCTN17395671.”
“A limitation was that we did not fully achieve recruitment numbers to reach 90% power but did exceed the target for 80% power.”
Registered (1 ID: ISRCTN). No reporting guideline cited.
Broken references and links
None foundReferences checked against Crossref, OpenAlex and Retraction Watch for retractions and resolvability, plus declared data and code links probed for whether they resolve to content matching the paper.
Checked — nothing surfaced.
1 data/code link checked; 1 live.
- datahttps://edata.bham.ac.uk/1555/LIVEHTTP 200Resolves, but the content could not be matched to the paper.
Copyediting
7 minorWording, consistency and formatting errors that need correcting before submission.
No major wording or formatting errors. 7 minor suggestions below.
7 copyedit issues flagged: mostly punctuation, typo, grammar.
- MINORgrammarMethods, ABA-feed intervention, paragraph 1“In additional to usual care”→ In addition to usual carePreposition error.
- MINORclarityMethods, ABA-feed intervention, paragraph 1“SMS text, WhatsApp, and telephone calls”→ SMS, WhatsApp, and telephone calls'SMS text' is redundant; SMS already implies text.
- MINORpunctuationTable 3, risk ratio column“0.96 (0.89 to 1.03 0.29)”→ 0.96 (0.89 to 1.03; P=0.29)Missing semicolon before P value.
- MINORtypoMethods, ABA-feed intervention“In additional to usual care, participants allocated to the ABA-feed intervention were offered proactive feeding support”→ Change 'In additional to' to 'In addition to'.Grammatical slip.
- MINORtypoTable 3 footnote“Value >0 favour favours planned ABA-feed intervention.”→ Remove duplicated word: 'Value >0 favours planned ABA-feed intervention.'Duplicated word.
- MINORpunctuationTable 3, Any breastfeeding at 24 weeks row“0.96 (0.89 to 1.03 0.29)”→ Insert a semicolon before the p-value: '0.96 (0.89 to 1.03; 0.29)'.Missing semicolon separating CI from p-value.
- MINORconsistencyTable 3, Breastfeeding initiation row“1.02 (1.00 to 1.04; 0.13)”→ Review the risk-ratio CI lower bound (1.00) against the reported p=0.13; the CI touching 1.00 implies a p near 0.05 under normal approximation.Possible rounding artifact at the model boundary; risk-difference CI is consistent with p=0.13.
The published study is robust and trustworthy. Readers should note the minor omissions: no explicit statement of adherence to the Declaration of Helsinki, and no reference to the CONSORT checklist. The statistical analyses are consistent and the data/code are accessible. No validity-threatening errors were found; an erratum adding the regulatory compliance statement would be a minor improvement.
- 1.HIGHethicsAdd a statement of adherence to the Declaration of Helsinki (or equivalent) in the Ethics section; this could be submitted as a correction to the journal.Standard requirement for clinical trials; missing statement is a reporting gap that readers and journal editors may note.
- 2.HIGHreportingReference the CONSORT 2010 checklist in the Methods section to explicitly state adherence; this could be added via a correction.Improves reporting transparency and meets journal requirements for randomised trials.
- 3.MEDIUMcopyeditCorrect 'In additional to usual care' to 'In addition to usual care' in the Methods section (ABA-feed intervention).Corrects a grammatical error that could distract readers.
- 4.MEDIUMcopyeditIn Table 3, insert a semicolon before the P-value in the rows where it is missing (e.g., '0.96 (0.89 to 1.03 0.29)' should be '0.96 (0.89 to 1.03; P=0.29)').Standard formatting for clarity and consistency.
- 5.MEDIUMcopyeditRemove the redundant 'text' in 'SMS text' and correct to 'SMS' in the Methods section.Improves clarity and avoids redundancy.
- 6.MEDIUMcopyeditFix the duplicated word in Table 3 footnote: 'Value >0 favour favours planned ABA-feed intervention' should be 'Value >0 favours planned ABA-feed intervention'.Corrects a typo that could cause confusion.
- 7.LOWstatisticsIn the Abstract or Results, note the minor inconsistency between the risk ratio CI lower bound of 1.00 and the reported p=0.13 for breastfeeding initiation; this is likely a rounding artifact from the model boundary and could be clarified in a supplementary note.Transparency for readers who may notice the discrepancy.
The star rating is the report’s one-glance summary. Every paper starts at 5★ and loses stars for the concrete problems the review finds — so a rating is never a vague average, it’s a running total you can read line by line under “How this rating was calculated.”
- Reporting — 8 dimensionseach dimension that fully fails−½★
- each dimension partially met−¼★
- Statistics · Integrity · Claimseach serious problem−1★
- each medium problem−½★
- Citationseach retracted or unverifiable reference−¼★
- Copyeditonly when the manuscript needs a full edit−½★
The rating never drops below 1★, and a demonstrable critical failure (an impossible statistic, a proven ethics violation) caps it at 1★ on its own — so the stars can never look healthy when the verdict is CRITICAL.
The rating draws on a panel of agents. Three independent Kaimen Rigor reviewers grade the eight dimensions below across several independent passes (the shown verdict is their majority vote — steadier than any single run), isolate the paper’s major claims and check its own evidence backs them, and flag integrity concerns. Alongside them, a citation agent resolves every reference against Crossref, OpenAlex, and Retraction Watch; a statistics agent recomputes reported tests; and rule-based checks verify that declared data/code links actually resolve. Full text is required — an abstract-only submission is not analyzed.
Graded against NIH, MDAR, ARRIVE 2.0, CONSORT, EQUATOR, and RRID guidelines. A dimension that doesn’t apply to the study type is skipped, never penalized.
This Kaimen Rigor review is model-assisted and is not a substitute for formal expert review. It complements human evaluation by surfacing potential methodological concerns — verify each finding against the source.