Vitamin D supplementation before in vitro fertilisation in women with polycystic ovary syndrome: multicentre, double blind, placebo controlled, randomised clinical trial.
Hu KL, Liao T, Wu Q, Ma X, Cao Y, Tan J, Tian L, Wang J, Yin J, Liu Y, Zhao J, Zhao S, Li M, Cai L, Liu FT, Gan K, Xu Y, Wang Y, Cai J, Zheng B, Ma Y, Ma Q, Zheng J, Pu X, Zhang H, Hao C, Xie Q, Zhang C, Jiang L, Zhang S, Yan L, Meng Q, Li W, Mol BW, Li R, Wang R, Zhang D, VitD-PCOS trial group
- DOI
- 10.1136/bmj-2025-087438
- Record issued
- 2026-08-10
- Engine
- 7.29.0
- Exported
- 2026-09-21
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/7aad2513-510f-4d93-88b2-0bd76010fbc0 is authoritative.
How this rating was calculated
- IntegrityIntegrity concern−0.5★
- ReportingEthical approvals partially met−0.25★
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.
This is a well-designed, rigorously reported multicentre double-blind placebo-controlled RCT of vitamin D for live birth in PCOS-IVF, with computer-generated stratified randomisation, a priori power calculation, prospective registration, a CONSORT flowchart, and deidentified data plus code openly deposited at OSF. Both independent reviewers scored 96/100, with the only sub-threshold element being the absence of an explicit regulatory-framework compliance statement (e.g., Declaration of Helsinki / ICH-GCP). A minor internal inconsistency in the CONSORT footnote (7 enumerated exclusion reasons vs 11 excluded participants) warrants clarification.
Both reviewers reviewed the full text and converged on all eight dimension statuses (both scored 96); minor sub-criterion divergences (demographics race/ethnicity reporting; exact p_values applicability) were weighed and resolved without changing any status. The statistics component recomputed only 2 of the reported tests (both consistent) — the remaining analyses are unverified, not confirmed. No references were retracted or unresolved, and both data/code links are live.
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 2 tests: 2 consistent, 0 inconsistent; 2 via agent-written checks.
- CONSISTENTreported p = .670 · recomputed p = .656Reviewer 1Implied two-tailed p for the primary outcome adjusted risk ratio (the paper reports the estimate and CI but no p-value; the CI is consistent with a null result)
“226 (52.0%) live births occurred in the vitamin D group and 216 (50.2%) in the placebo group (adjusted risk ratio 1.03, 95% confidence interval 0.91 to 1.18)”
Taken as given: 1.03 is the adjusted risk ratio estimate; 0.91 and 1.18 are the two-sided 95% confidence bounds; log=1 because 1.03 is a risk ratioMethod: pCI recovers the standard-normal z from the log-RR and its 95% CI, then returns the two-tailed p.How we recomputed it: pCI(1.03, 0.91, 1.18, 1) - CONSISTENTreported p = .170 · recomputed p = .176Reviewer 1Implied two-tailed p for the 12-month cumulative live birth adjusted risk ratio (reported as estimate + CI, no p-value)
“12 months | 281 (64.6) | 260 (60.5) | 1.07 (0.96 to 1.18) | 1.07 (0.97 to 1.18)”
Taken as given: 1.07 is the adjusted risk ratio estimate; 0.97 and 1.18 are the two-sided 95% confidence bounds; log=1 because 1.07 is a risk ratioMethod: pCI recovers the standard-normal z from the log-RR and its 95% CI, then returns the two-tailed p.How we recomputed it: pCI(1.07, 0.97, 1.18, 1)
- lowinternal contradictionThe CONSORT footnote enumerates 7 exclusion reasons (3 in one group and 4 in the other), but 876 randomised minus 865 analysed leaves 11 participants excluded, so 4 exclusions are not accounted for by the printed footnote.
“876 participants were randomised and 865 were included in the primary intention-to-treat analysis”
Figure 1Find in source
Overstated conclusions
None foundConclusions 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.
Checked — nothing surfaced.
8 major claims checked against the paper's own evidence: all adequately supported.
- supportedReviewer 1Vitamin D supplementation does not improve live birth rates in women with PCOS undergoing IVFThe primary analysis (adjusted RR 1.03, 95% CI 0.91 to 1.18) directly supports the null finding, with per-protocol and complete-case analyses consistent.Evidence: Adjusted risk ratio 1.03, 95% CI 0.91 to 1.18, for live birth; per-protocol and complete-case analyses consistent
“this does not translate to improved live birth rates after the first transfer for patients with polycystic ovary syndrome”
AbstractFind in source - supportedReviewer 1Vitamin D supplementation increases serum 25-OHD levelsSerum 25-OHD was significantly higher on trigger day and pregnancy test day in the vitamin D group, directly supporting the claim.Evidence: Trigger-day 25-OHD 32.3±11.2 vs 18.2±7.6 ng/mL, adjusted mean difference 13.6 (95% CI 10.9 to 16.3)
“On the day of triggering, the serum 25-OHD level was significantly higher in the vitamin D group than in the placebo group (32.3±11.2 v 18.2±7.6 ng/mL, adjusted mean difference 13.6, 95% confidence interval 10.9 to 16.3)”
AbstractFind in source - supportedReviewer 1No interaction was observed between baseline vitamin D levels and treatment effect on live birthThe subgroup interaction was tested with an interaction term and fractional polynomial approach, yielding P=0.731.Evidence: Interaction P=0.731; fractional polynomial linear term best fitting
“No interaction was observed between baseline vitamin D levels and treatment effect on live birth (interaction P=0.731).”
ResultsFind in source - supportedReviewer 1These findings challenge the routine use of vitamin D supplementation as an adjunctive treatment in women with PCOS undergoing IVFThe null primary result, while not ruling out a modest benefit, supports the conclusion that routine use is not supported by this evidence.Evidence: Null primary outcome with CI 0.91 to 1.18; no benefit in secondary outcomes
“These findings challenge the routine use of vitamin D supplementation as an adjunctive treatment in women with polycystic ovary syndrome undergoing IVF”
What this study addsFind in source - supportedReviewer 2Vitamin D supplementation did not improve live birth rates in women with PCOS undergoing IVF.The primary outcome shows no statistically significant difference between groups, with an adjusted risk ratio of 1.03 (95% CI 0.91 to 1.18).Evidence: Table 2: 226 (52.0%) live births in vitamin D group vs 216 (50.2%) in placebo group.
“226 (52.0%) live births occurred in the vitamin D group and 216 (50.2%) in the placebo group (adjusted risk ratio 1.03, 95% confidence interval 0.91 to 1.18).”
Table 2Find in source - supportedReviewer 2Vitamin D supplementation increases serum 25-OHD levels.The data show a significant increase in serum 25-OHD on trigger day, with a mean difference of 13.6 ng/mL (95% CI 10.9 to 16.3).Evidence: Supplementary table 4 and Figure 2: mean levels 32.3 vs 18.2 ng/mL.
Serum 25-OHD levels were significantly higher in the vitamin D group than in the placebo group: mean levels on trigger day were 32.3±11.2 v 18.2±7.6 ng/mL (adjusted mean difference 13.6, 95% CI 10.9 to 16.3).
Results ¶2reviewer’s wording - supportedReviewer 2No evidence of differences in participants with varying baseline serum 25-OHD levels.The interaction test between baseline vitamin D levels and treatment effect on live birth was not significant (P=0.731).Evidence: Results paragraph 7: No interaction was observed (P=0.731).
“No interaction was observed between baseline vitamin D levels and treatment effect on live birth (interaction P=0.731).”
Results ¶7Find in source - supportedReviewer 2The findings do not rule out a modest benefit of vitamin D supplementation.The 95% confidence interval for the primary outcome (0.91 to 1.18) includes a modest benefit, and the authors explicitly state this caveat.Evidence: Discussion: 'the 95% CI indicates that more modest effects cannot be excluded'.
“the 95% CI indicates that more modest effects cannot be excluded.”
Discussion ¶1Find in source
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 multiple studies on the high prevalence of vitamin D deficiency in PCOS, the expression of vitamin D receptors in reproductive tissues, and conflicting evidence from prior trials. The authors explicitly state that 'Existing randomised controlled trials are limited by small sample sizes, short intervention durations, single centre design, and insufficient follow-up time', thereby addressing the limitations of prior work. The hypothesis that vitamin D supplementation may improve live birth rates is logically derived from the cited evidence.
“Existing randomised controlled trials in general infertility populations are limited by small sample sizes, short intervention durations, single centre design, and insufficient follow-up time.”
“vitamin D receptors are expressed in ovarian tissue, endometrium, and placenta, suggesting potential involvement of vitamin D in folliculogenesis, endometrial receptivity, and implantation”
“Therefore, we conducted a multicentre, randomised, double blind, placebo controlled trial”
“Evidence on vitamin D supplementation in women with infertility is conflicting.”
“Existing randomised controlled trials in general infertility populations are limited by small sample sizes, short intervention durations, single centre design, and insufficient follow-up time.”
“Therefore, vitamin D supplementation has been proposed as an adjunctive treatment for women with infertility—especially PCOS—before IVF to improve fertility outcomes.”
Randomisation used a computer-generated list with variable block sizes (2, 4, 6) stratified by centre, prepared by an independent statistician; unit is the individual participant. Double blinding was achieved with identical placebo capsules and masked analysis. The sample size calculation states power 80%, type I error 5%, and an effect of 38% to 48%. Inclusion/exclusion criteria are given, and the analysis population (modified ITT, negative imputation, per-protocol) is pre-specified. The bench-science sub-criteria (replicate distinction, wet-lab controls, independent replication) are not applicable to a human RCT.
“Randomisation was conducted using a computer generated list, with variable block sizes of two, four, or six, stratified by study centre.”
“Participants and trial investigators were unaware of the treatment allocation”
“a total of 768 participants were required (power 80% and type I error rate 5%)”
“Randomisation was conducted using a computer generated list, with variable block sizes of two, four, or six, stratified by study centre.”
“To show or refute a 10% increase in the live birth rate in the vitamin D treatment group (48%), a total of 768 participants were required (power 80% and type I error rate 5%).”
“Exclusion criteria were participants who had experienced three or more failed IVF cycles, if they were scheduled for pre-implantation genetic testing, oocyte donation IVF cycles, or with a known allergy to vitamin D.”
Sex is reported (all women with PCOS); a separate sex justification is not applicable because the study concerns a female-only condition. Age (29.4±3.6), BMI (median 24.2), infertility duration, AMH, and PCOS phenotype are reported in Table 1 with missing-data footnotes. Race/ethnicity is not explicitly stated but is implied by the Chinese setting; comorbidities are not individually tabulated but the PCOS population is well characterised. Species/strain/housing sub-criteria do not apply to a human trial.
“Age (years), mean (SD)* | 29.4 (3.6) | 29.4 (3.5)”
“Type D | 259 (59.5) | 266 (61.9)”
“women aged 20-42 years diagnosed as having PCOS according to the Rotterdam criteria”
“women with polycystic ovary syndrome”
“Age (years), mean (SD)* | 29.4 (3.6) | 29.4 (3.5)”
The study was approved by a named body with a protocol number (Women's Hospital of Zhejiang University, IRB-20200035-R) plus local committee approvals, and informed consent was obtained. However, the manuscript nowhere states adherence to a named ethical/regulatory framework such as the Declaration of Helsinki or ICH-GCP, so regulatory_compliance is not reported. Per the scoring rule, any applicable criterion that is not_reported yields a warn (a fixable reporting gap, not misconduct).
“The study was approved by the ethics committee of the Women’s Hospital of Zhejiang University (IRB-20200035-R), followed by ethics committee approval from each centre.”
“After giving informed consent, participants were randomly assigned 1:1”
“The study was approved by the ethics committee of the Women’s Hospital of Zhejiang University (IRB-20200035-R), followed by ethics committee approval from each centre.”
“After giving informed consent, participants were randomly assigned 1:1 to receive vitamin D or placebo.”
The vitamin D and placebo capsules are described with manufacturer (Sinopharm Xingsha Pharmaceuticals, Xiamen, China), formulation (cholecalciferol, 800 IU per capsule), dose (4000 IU/day), and regimen. The statistical software is identified as Stata 18.0 (StataCorp, USA). No antibodies, cell lines, or other reagents are used. All applicable resources are adequately reported.
“each vitamin D capsule contained 720-960 IU calcium-free cholecalciferol”
“All analyses were performed using Stata 18.0 (StataCorp, USA).”
“automatic analyser (SCIEX Triple Quad 4500MD) using the vitamin D 200M Assay Kit (DISIGNS Diagnostics, China)”
“Vitamin D and placebo capsules were produced by a licensed pharmaceutical company (Sinopharm Xingsha Pharmaceuticals, Xiamen, China).”
“All analyses were performed using Stata 18.0 (StataCorp, USA).”
The primary analysis uses modified Poisson regression with robust variance and a random effects model for adjusted estimates. The sample size calculation is provided. Exact p-values are reported for the interaction test (P=0.731), while the primary outcome is reported with risk ratios and 95% CIs, which is acceptable for estimation-based reporting. Statistical software is identified. Data presentation includes a CONSORT flowchart, baseline table, and outcome tables with counts and percentages. The percentages match the reported counts (e.g., 226/435 = 52.0%).
“adjusted risk ratio 1.03, 95% confidence interval 0.91 to 1.18”
“modified Poisson regression with a robust variance estimate was used”
“No interaction was observed between baseline vitamin D levels and treatment effect on live birth (interaction P=0.731).”
“For unadjusted estimates, modified Poisson regression with a robust variance estimate was used. For adjusted estimates, a random effects model was used to account for centre effects.”
“226 (52.0%) live births occurred in the vitamin D group and 216 (50.2%) in the placebo group (adjusted risk ratio 1.03, 95% confidence interval 0.91 to 1.18).”
“No interaction was observed between baseline vitamin D levels and treatment effect on live birth (interaction P=0.731).”
The data availability statement names a concrete route with a persistent identifier: 'Deidentified individual participant data, and the code for statistical analyses are available at: https://doi.org/10.17605/OSF.IO/F9BC5'. This satisfies data_availability_statement, repository_deposit, and code_sharing. Accession numbers are not applicable (no sequencing data). Trial protocol and statistical analysis plan are available in the appendix.
“Deidentified individual participant data, and the code for statistical analyses are available at: https://doi.org/10.17605/OSF.IO/F9BC5”
“Trial documentation, including the study protocol and statistical analysis plan, are available in the appendix.”
“Deidentified individual participant data, and the code for statistical analyses are available at: https://doi.org/10.17605/OSF.IO/F9BC5”
The methods are comprehensive, covering trial design, participants, randomisation, IVF procedures, outcomes, and statistical analysis. The trial is registered at ClinicalTrials.gov (NCT04082650). A CONSORT flowchart is provided, indicating adherence to the guideline. All pre-specified outcomes are reported, and a post-hoc outcome is clearly labelled as exploratory. Limitations are discussed in a dedicated section. Conclusions are appropriately cautious ('does not improve', 'cannot rule out a modest benefit'). Funding sources and competing interests are declared.
“The trial was prospectively registered with ClinicalTrials.gov ( NCT04082650”
“Fig 1 CONSORT (consolidated standards of reporting trials) flowchart.”
“The observed live birth rate in the placebo group (50.2%) substantially exceeded the prespecified assumption of 38%.”
“Trial registration ClinicalTrials.gov NCT04082650 (https://clinicaltrials.gov/ct2/show/NCT04082650)”
“Fig 1 CONSORT (consolidated standards of reporting trials) flowchart.”
“However, this trial has several limitations.”
Registered (1 ID: ClinicalTrials.gov). Reporting guideline cited: CONSORT.
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.
Checked 35 references by DOI: 35 verified.
Every extracted reference resolved against Crossref/OpenAlex with no retraction flags.
2 data/code links checked; 2 live.
- dataOSFLIVEHTTP 200https://doi.org/10.17605/OSF.IO/F9BC5Resolves to OSF (data repository).
- datahttps://public.flourish.studio/visualisation/27443480/LIVEHTTP 200Resolves, but the content could not be matched to the paper.
Copyediting
1 minorWording, consistency and formatting errors that need correcting before submission.
No major wording or formatting errors. 1 minor suggestion below.
1 copyedit issue flagged: mostly consistency.
- MINORconsistencyFig 1 CONSORT footnote“†One participant did not meet Rotterdam criteria for polycystic ovary syndrome (PCOS), one had abnormal liver function, one was not scheduled for in vitro fertilisation (IVF) or intracytoplasmic sperm injection (ICSI), and one was diagnosed as having tuberculosis”→ Ensure the footnote (3 + 4 = 7 reasons) accounts for all 11 participants excluded between randomisation (876) and the analysis population (865).The enumerated reasons (7) do not sum to the 11 excluded, so the flowchart likely shows additional exclusions not described in the caption.
The published work is robust: a transparently reported, prospectively registered RCT with a null primary result, deposited data and code, and proportional conclusions. An informed reader should weigh the minor reporting gaps (no named Helsinki/ICH-GCP compliance statement, unexplicit race/ethnicity, and the CONSORT exclusion-count discrepancy) and recognise that only a fraction of the statistics were independently recomputed; none of these threaten the core conclusion, but the CONSORT reconciliation would warrant a small correction or clarification.
- 1.HIGHreportingReconcile the CONSORT flowchart exclusion counts: the Fig 1 footnote enumerates 7 exclusion reasons (3 in one group, 4 in the other) but 876 randomised minus 865 analysed leaves 11 excluded — publish a correction or add a note explaining the missing 4 exclusions so the flowchart totals match.This is an internal numerical contradiction in the published trial accounting that readers and editors will notice and that undermines confidence in the CONSORT flow.
- 2.MEDIUMethicsAdd an explicit statement of adherence to a recognised regulatory framework (Declaration of Helsinki or ICH-GCP) to the ethics statement, via a correction or a note in the data repository.Named committee approval and consent are reported, but the absence of a compliance-framework statement is the only sub-threshold element in this audit and is a standard expectation for a clinical trial report.
- 3.MEDIUMreportingReport participant race/ethnicity explicitly in the baseline demographics table and add a note on the generalisability limits of a homogeneous Chinese population to non-Chinese settings.Neither reviewer found race/ethnicity explicitly tabulated, and this is a relevant biological variable for a vitamin-D trial whose findings may not generalise across populations with differing baseline 25-OHD.
- 4.LOWreportingReference the completed CONSORT checklist explicitly alongside the CONSORT flowchart (Fig 1) rather than relying on the flowchart alone to signal guideline adherence.Both reviewers noted that CONSORT compliance is currently inferred from the flowchart rather than stated — an explicit checklist reference removes ambiguity.
- 5.LOWethicsSpecify in the Randomisation and masking section whether informed consent was written or verbal.The manuscript states only 'After giving informed consent' without specifying the mode, a minor completeness gap for a clinical-trial methods section.
- 6.LOWdata codeAdd a sentence to the Data availability statement specifying any conditions for data access (e.g., whether a data-sharing agreement is required).Clarifying access conditions makes the otherwise-excellent OSF deposit fully actionable for re-analysis.
- 7.LOWreportingDescribe the DSMC charter and terms of reference in the trial-oversight paragraph, since the Data Safety Monitoring Committee is acknowledged but its role is not detailed.A brief description of DSMC oversight strengthens the trial-oversight account for readers assessing governance.
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.