PLOS One retracts review on armed conflict and maternal-child care seeking in sub-Saharan Africa
PLOS One has retracted a systematic review examining how armed violent conflict affects healthcare-seeking behavior for maternal and child health in sub-Saharan Africa after editors concluded that unresolved methodological problems undermined the validity of the published results and conclusions. The retraction centers on core weaknesses in the review’s design, evidence handling, and certainty assessment, not a minor reporting defect.
A systematic review lives or dies on reproducible methods. When the search strategy, eligibility criteria, and synthesis plan are not aligned, the conclusions can drift far from the underlying literature even when the topic is clinically important.
Methodological gaps drove the editorial decision
The editorial notice points to several problems that were raised after publication. The supporting information file labeled S1 Checklist was a PRISMA-P checklist for review protocols rather than the standard PRISMA checklist for the completed systematic review, and it was not filled out with page numbers or section titles to guide readers to the relevant content in the manuscript. That omission matters because transparency in reporting is what allows clinicians, epidemiologists, and policy readers to judge whether a review can be trusted and reproduced.
The editors also identified an unexplained exclusion of articles published in 2023. In a rapidly evolving evidence base, selective exclusion without a clear rationale can alter the direction of a review, especially when the topic concerns conflict settings where publication patterns are uneven and data are sparse. On a topic such as maternal and child health in war-affected settings, even a small shift in the included literature can change the apparent balance of findings.
The stated aims, objectives, and research question were not internally consistent. At one point the review appeared to aim at describing the literature and identifying knowledge gaps; elsewhere it seemed intended to assess the impact of exposure to armed conflict on healthcare-seeking behavior in sub-Saharan Africa. Those are related but not identical tasks. A map of the literature is not the same as an analytic assessment of effect, and the difference changes how studies should be grouped, weighed, and interpreted.
Narrative synthesis was not judged reproducible
Editors said the methods used to synthesize the evidence were not adequately described. Instead of a systematic cross-study synthesis that compared outcomes, contextual factors, and study quality, the paper relied on a directional count of study outcomes and selective narrative descriptions organized study by study rather than outcome by outcome. In practice, that approach can produce a veneer of order while leaving the reader unable to determine why certain findings diverged, which evidence was stronger, or whether differences reflected setting, population, design, or bias.
This kind of synthesis problem is especially consequential in conflict research. Exposure to violence can reshape transport, displacement, trust in institutions, household decision-making, and access to clinics all at once. A review that does not compare these contextual drivers risks flattening a complex causal chain into a simple tally of favorable and unfavorable findings.
The editors remained concerned that the conclusions about the impact of conflict on healthcare-seeking behavior, the comparison between maternal and child health, and the proposed explanatory factors were not adequately supported by the Results section. In other words, the paper drew interpretive lines that the evidence presentation did not firmly support.
The data did not earn the conclusions.
GRADE was applied in ways that raised further concern
A separate problem involved the treatment of publication bias and GRADE certainty ratings. Publication bias was handled as if it were an individual-study risk-of-bias domain rather than an outcome-wide certainty issue under the GRADE framework. GRADE assessments were also applied to individual studies rather than across studies for each outcome.
Editors further objected to the way certainty was assigned. The review gave a “high” quality rating to nearly all observational and qualitative studies without documenting justification for upgrading. In evidence appraisal, that is a serious departure from standard practice. Observational and qualitative studies can be highly informative, especially in humanitarian and conflict settings where randomized evidence is rare, but high certainty ratings require explicit reasoning and careful accounting for limitations, not blanket upgrading by default.
For readers trying to use the paper in policy or program design, that misapplication could create a false impression that the evidence was stronger and more settled than it actually was. In maternal and child health research, overconfident certainty ratings can distort decisions about service delivery, outreach, and resource allocation in fragile settings where the consequences of error are immediate.
Corresponding author response did not resolve the core issues
After the concerns were raised, the corresponding author provided a completed PRISMA 2020 checklist with page numbers and stated that the review sought to answer what evidence has been reported regarding the relationship between armed conflict exposure and maternal and child health-care seeking behaviors in sub-Saharan Africa. The author also stated that data were synthesized narratively to identify recurring themes and that the evidence synthesis was presented in the Results section.
Editors were not persuaded that these responses resolved the central concern: the synthesis methods were still not described with enough precision to make the review reproducible, and the conclusions remained insufficiently supported by the results as presented. A narrative synthesis can be appropriate in heterogeneous literatures, but it still requires a clear and disciplined structure. Without that, narrative review can slide into selective emphasis rather than transparent evidence integration.
The retraction notice states that all authors did not agree with the retraction. The decision nonetheless stands, because the editors concluded that the unresolved methodological problems called the published results and conclusions into question.
Why this retraction matters for conflict and maternal-child health research
Research on healthcare seeking in conflict zones occupies a difficult evidence environment. Civilian displacement, insecurity, damaged infrastructure, and disrupted supply chains all interfere with access to antenatal care, facility delivery, immunization, and treatment for childhood illness. Those pathways are plausible, clinically important, and often ethically urgent to study. But the more consequential the policy implications, the higher the bar for transparent synthesis.
When a systematic review examines conflict exposure and health behavior, the review should separate maternal outcomes from child outcomes only when the underlying data justify that split, and it should explain the grounds for any comparison between them. It should also make explicit how study design, context, and certainty grading influenced the overall interpretation. Without that scaffolding, the final narrative can look decisive while resting on a fragile method.
The retraction does not erase the underlying public health problem. It does, however, remove a review that the journal now considers too methodologically weak to support its claims about how armed violence shapes healthcare-seeking behavior in sub-Saharan Africa.