INTRODUCTION: In medical education research, the transfer of knowledge depends heavily on the visibility of scientific publications. Whether Open Access (OA) actually increases this visibility through a citation advantage (Open Access Citation Advantage, OACA) is still unclear for the field of medical education. This study aims to determine the existence and size of an OACA for medical education articles indexed in PubMed between 2010 and 2019.
METHOD: In a retrospective bibliometric cohort study, all entries classified as research articles by iCite® with the MeSH term 'Medical Education' (N = 43,275) were analysed. OA was defined by the presence of a PubMed Central identifier (PMCID). Primary endpoints were total citations, citations per year, and the Relative Citation Ratio (RCR). Group comparisons were made using Yuen‑Welch t‑tests (20 % trimmed means, α = 0.05). Sensitivity analyses included negative binomial regression with year and journal fixed effects, and quantile regression for RCR.
RESULTS: 21.2% of articles had a PMCID. PMCID-indexed publications showed higher trimmed means for total citations (10.31 vs. 7.00), annual citations (1.29 vs. 0.79), and RCR (0.79 vs. 0.49; all p < .001). Robust effect sizes (δˆRAKP, Algina-Keselman-Penfield robust standardized difference) ranged from 0.27 to 0.35 (small to medium). Negative binomial models with year and journal fixed effects confirmed these findings (IRR range: 1.53-1.67).
CONCLUSIONS: PMCID-indexed articles in the field of medical education are cited significantly more frequently and have higher field‑normalised impact values than non‑OA publications. Despite financial hurdles and methodological limitations including potential selection bias, the moderate OACA supports strategically expanding OA publication funds to maximise the reach and impact of medical education research. Prospective studies should consider different OA types, altmetrics, and potential confounders to pinpoint the impacts of OA.
PMCID-indexed articles in the field of medical education are cited significantly more frequently and have higher field-normalised impact values than non-OA publications.
large observational comparisons consistently support direction, but self-selection and prestige confounding limit attribution
Negative binomial models with year and journal fixed effects confirmed these findings (IRR range: 1.53-1.67).
adjusted models reproduce the association but do not control article-level selection into PMCID indexing
Derived from the full evaluation — not a separate score.
Strengths
The analysis covers 43,275 medical-education research articles published over a decade and evaluates total citations, annual citations, and RCR. This provides a substantial evidence base for the defined corpus.
↳ Methods, cohort definition and primary endpoints; Results, cohort flow
Year and journal fixed-effect models, quantile regression, and an immediate-PMC-release sentinel analysis consistently support higher citation outcomes among PMCID-indexed articles.
↳ Results, sensitivity analyses; Supplement Tables S3–S5
The paper discusses randomized null findings and reverse associations alongside supportive studies, then connects PMCID coverage, embargo timing, self-selection, and prestige confounding to interpretation.
↳ Introduction, OACA literature review; Discussion; Limitations
Limitations
Authors may preferentially choose OA for manuscripts expected to have high impact, while PMCID indexing is associated with funder and institutional resources. The reported models do not directly identify or bound this selection mechanism.
↳ Limitations, paragraphs 1 and 3
The exposure misses gold, hybrid, institutional-repository, and bronze OA pathways outside PMC, limiting what the results establish about OA as a whole.
↳ Methods, OA operationalisation; Limitations, paragraph 1
Recommendations to prioritise or expand OA publication funds imply a practical benefit, although the article explicitly states that its observational association does not establish causation.
↳ Abstract, Conclusions; Implications and Future Directions
The large cohort, field-normalised RCR endpoint, robust trimmed-mean comparisons, and several sensitivity analyses support a consistent association between PMCID indexing and citation outcomes. Methodological Rigour is limited by self-selection and funder-prestige mechanisms that year and journal fixed effects do not directly resolve. Positioning scores more strongly because the paper engages null and contrary studies and traces its principal limitations through interpretation. The practical recommendation to expand OA support should be narrowed because the analysis does not identify a causal funding return.
Nabu’s assessment, alongside the field’s view.
Are you an author of this paper?
Sound3.7
Confidence highThe study supplies large-scale, field-specific evidence on the association between PMCID indexing and citation impact while adding RCR as a field- and time-normalised endpoint. The advance is meaningful but incremental within an established OACA literature.
“To date, there has been no large-scale study of medical education—a gap this study attempts to fill.”
Robust group comparisons, year and journal fixed-effect models, quantile regression, and an immediate-release sentinel analysis provide multiple checks on the observed association. Self-selection by expected impact and funder prestige remain unaddressed at the article level, warranting moderate downward pricing.
“Authors may choose open access for manuscripts expected to have a high impact.”
The question, endpoints, analyses, and non-causal conclusion form a followable argument. Precision is reduced by switching between trimmed and untrimmed effect summaries and by describing the evidence for a benefit as clear.
“Although our study does not allow a causal conclusion to be drawn, the evidence for a benefit is clear.”
The paper presents supportive, null, and contrary OACA findings and traces major limitations to their likely interpretive consequences. Its policy recommendations are somewhat stronger than that contextual caution, and field-specific mechanisms receive less development than the broader literature review.
“Overall, the literature suggests that OACA may exist, but that its extent is field- and time-dependent.”
Lower confidence on Methodological Rigour, Positioning — domain match limited.
Caveats4 of 4 checks
The reported numerical results are broadly consistent, but effect magnitude and practical interpretation are framed more assertively in the Discussion and Conclusions than the primary analysis and acknowledged confounding warrant.
Ethics handling is appropriate for public bibliometric data, and code availability is claimed, but the availability wording and absent conflict disclosure are minor reporting watch-outs.
Flags: 3 declared / 5 total
27 of 27 checkable references verified
28 references in manuscript 1 are books, websites or datasets — counted, but not index-checkable
References not verified automatically are resolved manually within 24 hours.
No retraction or expression of concern found in Crossref (includes Retraction Watch).
Sources: Crossref (incl. Retraction Watch) ✓
Where this paper’s evidence sits on the path from initial observation to real-world use.
The findings derive from a decade of real-world bibliometric observations and can inform discussion immediately. Residual selection and prestige confounding prevent translation into a causal estimate of the return from choosing or funding OA.
“However, this observational association does not establish causation.”
AI-generated, human-governed. Something look off? Contact us to request a review.
Keep reading
Other papers Nabu has evaluated blind, starting with the same field.
Nicholas Fraser, Fakhri Momeni, Philipp Mayr, et al. · 2020 · Other social sciences
bioRxiv-deposited journal articles had sizably higher citation and altmetric counts compared to nondeposited articles.
SupportedMcGargill, Liu, Kuhns, et al. · 2026 · Other social sciences
Reviewers effectively evaluated Quality and Impact as separate dimensions, with Quality scores being more consistent across reviewers than Impact scores.
TentativeNoémie Aubert Bonn, Andrew J. Stewart, Lukas Hughes-Noehrer · 2026 · Other social sciences
Lack of time and resources was the most frequently reported barrier.
Supported