Statistical Reporting and Reproducibility Policy | European Gynecology & Obstetrics

CONTENTS

Scope and purpose
Definitions
General principles of statistical reporting
Minimum requirements for statistical reporting
Description of statistical methods
Statistical software disclosure
Sample size and power calculations
Management of missing data
Multiple testing, subgroup analyses, and sensitivity analyses
Pre-specified versus exploratory analyses and Statistical Analysis Plan
Reporting by study design: specific requirements
Reporting guidelines (EQUATOR): submission requirements and procedures
Sex and gender analysis in reporting
Statistical review in the peer review process
Computational reproducibility and analytical code
Presentation of statistical results in tables and figures
Violations and consequences
Post-publication statistical concerns
Annual review
Cross-references within the EGO Editorial Policy Framework
Normative references

SCOPE AND PURPOSE

This Statistical Reporting and Reproducibility Policy establishes the requirements for the transparent, accurate, and reproducible reporting of statistical methods and results in manuscripts submitted to and published in European Gynecology and Obstetrics (EGO). It applies to all authors of manuscripts submitted to the Journal, to peer reviewers in their assessment of statistical methods, and to the Editor-in-Chief and the editorial team in managing the editorial process.

The Policy operationalizes the commitments to research transparency established in Section 6 of the Editorial Policy Statement and the reporting standards requirements set out in Section 20 of the Research Ethics Policy. It complements the Data Availability and Sharing Policy regarding statistical code and computational reproducibility, and the Submission Guidelines regarding the mandatory use of EQUATOR reporting checklists. In the event of inconsistency on matters strictly related to statistical reporting and reproducibility, this Policy prevails; in all other areas, the Editorial Policy Statement remains the governing document. Requirements imposed by the Research Ethics Policy and the Submission Guidelines remain binding in their respective domains and are not overridden by this Policy.

Edikta S.r.l. issues the Policy on behalf of the European Society of Gynecology (ESG). It applies to all manuscript types that report primary or secondary data analysis, including original research articles, systematic reviews and meta-analyses, study protocols, and data notes. It applies to all manuscripts submitted from the date of its entry into force, subject to the study-design-specific provisions and justified exceptions explicitly set out in this Policy.

This Policy also encompasses operational aspects of peer review for statistical content (Section 14), the editorial consequences of statistical violations (Section 17), and post-publication investigation procedures involving statistical or data integrity concerns (Section 18). Where artificial intelligence or machine learning methods are used for inference or prediction, the applicable reporting and transparency requirements of this Policy apply in full.

DEFINITIONS

For this Policy, the following definitions apply:

- Statistical method: any procedure - quantitative or algorithmic - used to summarize, analyze, model, or draw inferences from data, including descriptive statistics, hypothesis tests, regression models, survival analyses, Bayesian methods, and machine learning algorithms used for inference or prediction.

- Effect size: a standardized, quantitative measure of the magnitude of a difference, association, or effect, independent of sample size. Effect size measures include mean difference, standardized mean difference (Cohen's d), odds ratio (OR), relative risk (RR), risk difference (RD), hazard ratio (HR), correlation coefficient (r), and partial eta-squared.

- Confidence interval: a range of values, derived from sample data, within which the true population parameter is estimated to lie with a specified level of probability (typically 95%). Confidence intervals quantify both the direction and precision of an estimate.

- p-value: the probability, under the assumption that the null hypothesis is true, of observing a result at least as extreme as the one obtained. A p-value does not measure the probability that the null hypothesis is true, the clinical significance of an effect, or the magnitude of a difference.

- Statistical significance: a conventional designation applied when a p-value falls below a pre-specified threshold (typically alpha = 0.05). Statistical significance does not imply clinical significance, nor does the absence of statistical significance imply the absence of a clinically meaningful effect.

- Power: the probability that a study will detect a true effect of a specified magnitude, given the sample size and significance threshold. Adequate power is conventionally set at 0.80 (80%) or above.

- Sample size calculation: a formal pre-study calculation determining the number of participants required to detect a hypothesized effect with specified power and significance level.

- Statistical Analysis Plan (SAP): a pre-specified, formally documented plan describing all primary and secondary analyses to be conducted, the statistical tests to be applied, the handling of missing data, and the criteria for subgroup and sensitivity analyses, prepared before unblinding of data or before data extraction in systematic reviews.

- Pre-specified analysis: any analysis defined in the study protocol, trial registration, or Statistical Analysis Plan before data collection is completed or before data extraction is initiated.

- Exploratory analysis: any analysis not pre-specified in the protocol, trial registration, or SAP, conducted after examination of the data, including data-driven subgroup analyses, post-hoc comparisons, and secondary analyses not listed in the original protocol.

- Multiple testing: the conduct of more than 1 statistical hypothesis test on the same dataset, which inflates the probability of at least 1 false-positive finding (type I error) beyond the nominal alpha level.

- Imputation: a statistical procedure for replacing missing data values with plausible estimates, including single imputation (e.g., mean substitution) and multiple imputation by chained equations (MICE).

- Reproducibility: the ability of an independent researcher to obtain the same results as the original authors by applying the same analytical methods to the same data. Computational reproducibility refers specifically to the ability to reproduce results by running the same code on the same data.

GENERAL PRINCIPLES OF STATISTICAL REPORTING

The Journal is committed to the highest standards of statistical transparency and reproducibility in published research. All statistical methods and results must be reported with sufficient clarity and completeness to allow independent assessment of the validity of the findings, replication of the analyses, and inclusion of the results in meta-analyses.

3.1 Alignment with international standards. The statistical reporting requirements of this Policy are aligned with:

- the Statistical Analyses and Methods in the Published Literature (SAMPL) Guidelines (Lang and Altman);

- the International Committee of Medical Journal Editors (ICMJE) Recommendations (Section IV.A.3 on statistical methods), updated January 2026;

- the American Statistical Association (ASA) Statement on Statistical Significance and p-values (Wasserstein and Lazar, 2016) and its 2019 elaboration;

- the EQUATOR Network reporting guidelines applicable to the relevant study design;

- CONSORT 2025 for randomized controlled trials, including its Open Science section;

- the Committee on Publication Ethics (COPE) Core Practices (2022).

3.2 Statistical significance and clinical significance. Authors must distinguish between statistical significance and clinical significance when interpreting results. A statistically significant result does not necessarily imply clinical importance; a result that does not achieve statistical significance does not necessarily imply the absence of a clinically meaningful effect. Authors must report and interpret effect sizes and confidence intervals to support judgments of clinical relevance, and must not rely on p-values alone to characterize the importance of findings. The use of the term "trend" to describe a result with a p-value above the significance threshold is not acceptable; where a nominally non-significant p-value is considered noteworthy, the effect size and confidence interval must be the primary basis for the discussion.

3.3 Transparency as a scientific obligation. All statistical analyses, including those yielding null or negative results, must be reported completely and without selective omission of outcomes based on the direction or significance of findings. The Journal actively supports publication of well-designed studies with null results and considers methodological rigor - not statistical significance - as the primary criterion for publication.

3.4 Pre-registration and protocol alignment. Where a study protocol or Statistical Analysis Plan was registered before data collection or unblinding, the reported analyses must be consistent with the pre-registered plan. Deviations from the pre-registered analysis must be declared and justified in accordance with Section 10.

MINIMUM REQUIREMENTS FOR STATISTICAL REPORTING

The following requirements apply to all manuscripts that report quantitative data and statistical analyses. These are minimum requirements; additional requirements specific to study design are set out in Section 11.

4.1 Descriptive statistics. Data must be summarized using appropriate descriptive statistics:

- Continuous data that are approximately normally distributed: report means and standard deviations (SD). Use the form: mean (SD). Do not use mean +/- SD.

- Continuous data that are not normally distributed or where normality cannot be assumed: report medians and interquartile ranges (IQR) or full ranges. State the upper and lower boundaries of interpercentile ranges.

- Categorical data: report absolute frequencies and percentages. Report both numerators and denominators for all percentages.

- Total sample sizes and group sizes must be reported for each analysis.

4.2 Inferential statistics. All inferential statistical results must include:

(a) The test statistic (e.g., t, F, chi-square, z), its degrees of freedom (where applicable), and the exact p-value. p-values must be reported to 2 decimal places when greater than 0.01 and to 3 decimal places when less than 0.01. The threshold p < 0.001 is acceptable as the lower limit of exact reporting; values below this threshold should be reported as p < 0.001. p-values must not be reported merely as "significant" or "non-significant", nor as p < 0.05 or p > 0.05.

(b) A point estimate of the effect size with the 95% confidence interval (CI). The CI must always be reported for primary outcomes, regardless of statistical significance. For comparisons involving more than 2 groups, pairwise effect estimates and CIs must be provided where clinically relevant.

4.3 Two-sided tests. Two-sided p-values must be reported as the default. One-sided tests may be used only where pre-specified in the study protocol and justified by a theoretical rationale. Any use of one-sided tests must be explicitly declared and justified in the Methods section.

4.4 Pre-specified significance threshold. Authors must state the significance threshold (typically alpha = 0.05) used to designate statistical significance. If a different threshold was used (e.g., alpha = 0.01 for multiple testing correction), this must be stated and justified. Once a threshold is declared, results must be characterized as statistically significant or not; intermediate characterizations such as "approaching significance" or "marginally significant" are not acceptable.

DESCRIPTION OF STATISTICAL METHODS

5.1 Methods section requirements. The Methods section of all manuscripts reporting quantitative analyses must contain a dedicated statistical analysis subsection. This subsection must include:

(d) all statistical methods used in primary and secondary analyses, described in sufficient detail to allow an informed reader to assess their appropriateness and, where feasible, to replicate the analyses;

(e) the specific test or model applied to each outcome (e.g., "the primary endpoint was compared between groups using the independent samples t-test"; "association between exposure and outcome was modeled using multivariable logistic regression");

(f) the assumptions underlying the statistical methods applied and the procedures used to verify those assumptions (e.g., Shapiro-Wilk test for normality, Bartlett's test for homogeneity of variance, assessment of the proportional hazards assumption in Cox regression);

(g) covariates included in multivariable models and the rationale for their selection (a priori specification is preferred over stepwise selection);

(h) the statistical software used (name, version, and manufacturer; see Section 6);

(i) for regression models: the approach to variable selection, the method for assessing model fit (e.g., Hosmer-Lemeshow goodness-of-fit test, pseudo-R2, AIC, BIC), and the handling of collinearity.

5.2 Regression analyses. For all regression analyses (linear, logistic, Cox, Poisson, mixed-effects, and others), the following elements must be reported:

- the type of regression model and its specification;

- all covariates included, with unadjusted and adjusted estimates reported separately;

- 95% confidence intervals for all regression coefficients and transformed estimates (odds ratios, hazard ratios, rate ratios);

- measures of model fit and discrimination where applicable (e.g., area under the ROC curve [AUC] for logistic regression, C-statistic for Cox models);

- assessment of the proportional hazards assumption for Cox regression, with the method used;

- assessment of influential observations and outliers.

5.3 Survival analyses. For all time-to-event analyses, the following elements must be reported:

- the Kaplan-Meier survival function for each group, with 95% confidence intervals;

- the number at risk at specified time points, presented below the survival curve;

- the log-rank test statistic and p-value for comparisons of survival functions;

- hazard ratios with 95% CIs from Cox regression, with verification of the proportional hazards assumption;

- censoring rates and the reasons for censoring.

5.4 Meta-analyses. For systematic reviews and meta-analyses, the following statistical elements must be reported:

- the pooling method (fixed-effects or random-effects model) and the rationale for the choice;

- the measure of heterogeneity: I2 statistic with its 95% CI, Cochran's Q statistic and p-value, and tau2 for random-effects models;

- the pooled effect estimate and its 95% CI presented in a forest plot with individual study estimates;

- methods for assessing small-study effects or reporting bias, including funnel plots and statistical tests for funnel plot asymmetry (e.g., Egger's test, Begg's test), where appropriate; such tests should generally be used only when at least 10 studies are included in the meta-analysis, as tests have insufficient power with fewer studies; results must be interpreted alongside visual inspection and clinical or methodological heterogeneity;

- sensitivity analyses for influential studies;

- subgroup analyses, pre-specified or clearly designated as exploratory.

STATISTICAL SOFTWARE DISCLOSURE

6.1 Mandatory disclosure. All manuscripts reporting statistical analyses must include a declaration of the statistical software used. This declaration must appear in the statistical analysis subsection of the Methods section and must specify:

(j) the name of the software package (e.g., R, SAS, Stata, SPSS, Python, MATLAB);

(k) the version number of the software (e.g., R version 4.4.2, SAS version 9.4);

(l) the name of the software developer or distributor (e.g., R Core Team, SAS Institute, StataCorp);

(m) for R and Python analyses: the names and version numbers of all packages or libraries used for statistical analyses (e.g., survival, lme4, meta, metafor in R; scipy, statsmodels, lifelines in Python).

6.2 Justification for non-standard software. Where analyses are performed using custom-written code or non-validated software tools not widely recognized in the biomedical literature, authors must justify their choice and, where available, provide a reference to a published description or validation of the software.

6.3 Code availability. Analytical code - including scripts, syntax files, and annotated notebooks - that can be used to reproduce the statistical analyses reported in the manuscript must be made available at the time of publication unless legal, institutional, or ethical constraints prevent disclosure. Authors who are unable to share their analytical code must provide a detailed justification in the Code Availability Statement (see Data Availability and Sharing Policy, Sections 4 and 8 and Section 15.4).

SAMPLE SIZE AND POWER CALCULATIONS

7.1 Requirement for sample size justification. For all prospective studies - including randomized controlled trials, observational studies with a priori designed recruitment targets, diagnostic accuracy studies, and study protocols - authors must provide a sample size calculation or a formal justification for the sample size used.

7.2 Reporting elements. Sample size calculations must be reported in the Methods section with the following elements:

(n) the primary outcome and the specific measure of treatment effect or association on which the calculation is based (e.g., mean difference, OR, RR, hazard ratio, diagnostic sensitivity);

(o) the anticipated effect size used in the calculation, with its source (e.g., from a pilot study, prior publication, or clinical judgment), and the minimum clinically important difference (MCID) where applicable;

(p) the significance level (alpha) and the power (1 - beta) used in the calculation;

(q) the expected attrition, dropout, or loss-to-follow-up rate and the inflation applied to the sample size to account for it;

(r) the statistical software or formula used for the calculation (name and version; see Section 6), with a reference if a published formula was used;

(s) the total and group-specific sample sizes resulting from the calculation.

7.3 Deviations from planned sample size. Where the enrolled sample size deviates from the planned sample size, the reasons for the deviation must be declared in the Methods section, and the actual power of the study at the achieved sample size must be reported where feasible.

7.4 Retrospective studies. For retrospective studies and for studies with fixed-size datasets, a formal a priori power calculation may not be feasible. Authors must not report observed power (post-hoc power) as a justification for interpretability of results, as the observed p-value algebraically determines this measure and provides no independent information. Instead, authors must provide a precision-based justification: the observed or expected width of confidence intervals for primary outcomes, the minimum detectable effect at the achieved sample size for clinically meaningful thresholds, or a sensitivity analysis showing which effect sizes the study could reasonably have detected. Authors are also encouraged to contextualize null or indeterminate findings with respect to the clinical relevance of the observed effect size and its confidence interval. Where exploratory analyses are conducted, they are permitted only for hypothesis generation. Exploratory analyses must be clearly labeled as such. They must not be used to justify inferential claims that are unsupported by pre-specified analyses or to reclassify post hoc findings as confirmatory results.

MANAGEMENT OF MISSING DATA

8.1 Declaration of missing data. The Methods section must describe the extent, pattern, and plausible mechanism of missing data for all primary variables. Where mechanisms of missingness are invoked - such as Missing Completely at Random (MCAR), Missing at Random (MAR), or Missing Not at Random (MNAR) - these must be presented explicitly as assumptions, not as facts. Authors must support these assumptions with available evidence (e.g., Little's MCAR test, comparisons of observed characteristics between complete and incomplete cases) and accompany them with sensitivity analyses when the assumption may materially affect the conclusions. The choice of missing data handling method must be consistent with the declared assumption.

8.2 Handling methods. The method used to handle missing data must be explicitly declared and justified:

- Complete case analysis (listwise deletion): acceptable only where the extent of missing data is small (typically fewer than 5% of cases) and a MCAR mechanism is plausible. If complete case analysis is used with a higher proportion of missing data, a sensitivity analysis is required.

- Single imputation (e.g., mean substitution, last observation carried forward): generally not acceptable for primary analyses due to underestimation of uncertainty; if used, a sensitivity analysis with complete case analysis is required.

- Multiple imputation: the preferred approach for MAR data. Where multiple imputation is used, authors must report the number of imputations performed, the variables included in the imputation model, and the pooling rule applied (e.g., Rubin's rules).

- Model-based methods (e.g., mixed-effects models for missing at random longitudinal data): acceptable with appropriate specification and justification.

8.3 Sensitivity analyses. Where the primary analysis relies on a missing-data handling method that involves assumptions, a sensitivity analysis using a different, plausible method must be conducted and reported to assess the robustness of the findings to the missing-data approach.

MULTIPLE TESTING, SUBGROUP ANALYSES, AND SENSITIVITY ANALYSES

9.1 Multiple testing. Where multiple statistical hypotheses are tested on the same dataset, the type I error rate is inflated beyond the nominal alpha level. The following requirements apply:

- Where multiple testing is conducted, the method for controlling the family-wise error rate or the false discovery rate must be stated (e.g., Bonferroni correction, Holm-Sidak method, Benjamini-Hochberg procedure).

- Alternatively, where no formal correction is applied, the authors must declare this explicitly, acknowledge the increased risk of false-positive findings, and present the results as descriptive or hypothesis-generating rather than confirmatory.

- Secondary and exploratory outcomes should be presented with effect estimates and confidence intervals rather than p-values as the primary metric, to mitigate the interpretative consequences of multiple testing.

9.2 Subgroup analyses. Subgroup analyses must comply with the following requirements:

(t) Subgroup analyses must be explicitly designated as pre-specified (defined in the protocol, trial registration, or SAP) or post-hoc (defined after data collection or examination of results). Pre-specified and post-hoc subgroup analyses must be presented separately.

(u) The total number of subgroup analyses conducted must be declared.

(v) Tests for interaction (heterogeneity of effect across subgroups) must be reported. Differences in subgroup results must not be interpreted as evidence of a treatment-subgroup interaction without a significant test of interaction.

(w) Subgroup analyses should be regarded as hypothesis-generating unless pre-specified with adequate statistical power.

9.3 Sensitivity analyses. Sensitivity analyses conducted to assess the robustness of primary findings to analytical assumptions must be clearly identified and their purpose stated. Sensitivity analyses do not require adjustment for multiple testing where they test the same primary hypothesis under different assumptions; however, they must be distinguished from primary analyses and presented as secondary or supportive evidence.

PRE-SPECIFIED VERSUS EXPLORATORY ANALYSES AND STATISTICAL ANALYSIS PLAN

10.1 Designation of analyses. All analyses reported in the manuscript must be designated as:

- Primary analysis: the analysis directly addressing the primary research question or primary outcome;

- Pre-specified secondary analysis: any analysis defined in the protocol, trial registration, or SAP as a secondary objective before data collection or unblinding;

- Exploratory or post-hoc analysis: any analysis not included in the pre-specified plan, including data-driven subgroup analyses, additional outcome measures examined after data collection, and analyses prompted by reviewer suggestions.

This designation must be stated in the Methods section and, where relevant, reflected in the presentation of results.

10.2 Statistical Analysis Plan. For randomized controlled trials and prospective observational studies with pre-specified hypotheses, authors are strongly encouraged to prepare and register a Statistical Analysis Plan before data collection or, for clinical trials, before unblinding of treatment allocation. Where a SAP was prepared and registered:

- the registration number or URL of the SAP must be stated in the Methods section;

- any deviations from the SAP must be explicitly declared in the Methods section, with justification;

- analyses conducted in addition to those pre-specified must be clearly identified as exploratory.

10.3 Registered Reports. The Journal recognizes and welcomes submissions under the Registered Report format, in which the study protocol and statistical analysis plan are peer-reviewed before data collection. Manuscripts submitted as Registered Reports must follow the additional requirements specified in the Peer Review Policy and on the Journal's website. Registered Reports that received Stage 1 in-principle acceptance are evaluated at Stage 2 based on their adherence to the pre-registered protocol, not based on the statistical significance or direction of results.

REPORTING BY STUDY DESIGN: SPECIFIC REQUIREMENTS

The following requirements supplement the minimum requirements in Section 4 for specific study designs commonly published in European Gynecology and Obstetrics (EGO).

11.1 Randomized controlled trials. For randomized controlled trials, the CONSORT 2025 checklist is mandatory. In addition:

- baseline characteristics must be presented descriptively without significance testing, in accordance with CONSORT 2025 Item 25; p-values for baseline comparisons between randomized groups are not appropriate and will be requested for removal during peer review;

- the analysis population must be declared (intention-to-treat [ITT], modified ITT, per-protocol) and justified;

- for superiority trials: the confidence interval for the primary outcome must be presented alongside the p-value; the clinical interpretation must address both the lower and upper bounds of the CI;

- for non-inferiority and equivalence trials: the pre-specified non-inferiority margin must be declared, with justification; the primary result must be expressed as a confidence interval in relation to the non-inferiority margin; a p-value alone is insufficient for non-inferiority inference.

11.2 Observational studies. For cohort, case-control, and cross-sectional studies, the STROBE checklist is mandatory. In addition:

- effect measures must be presented as odds ratios, risk ratios, rate ratios, or mean differences, as appropriate to the study design, with 95% CIs;

- confounding must be addressed by declaring all variables included in multivariable adjustment and the rationale for their inclusion;

- potential for residual confounding must be acknowledged in the discussion.

11.3 Diagnostic accuracy studies. For diagnostic accuracy studies, the STARD 2015 checklist is mandatory. In addition:

- sensitivity, specificity, positive and negative predictive values, and likelihood ratios must be reported with 95% CIs;

- the area under the receiver operating characteristic curve (AUC-ROC) must be reported where applicable;

- the reference standard must be clearly defined, and any limitations of the reference standard acknowledged.

11.4 Systematic reviews and meta-analyses. For systematic reviews and meta-analyses, the PRISMA 2020 checklist is mandatory. Statistical reporting requirements are specified in Section 5.4 of this Policy.

11.5 Prediction model studies. For prediction model studies and AI-based prediction models, the TRIPOD or TRIPOD+AI (where an AI algorithm is used) checklist is strongly recommended. In addition:

- model performance must be reported with calibration and discrimination metrics (AUC-ROC or C-statistic, calibration plot, Brier score where applicable);

- internal validation (bootstrap or cross-validation) and external validation results must be clearly distinguished;

- for AI-based models, the requirements of the Artificial Intelligence Policy apply in addition to this section.

11.6 Studies reporting proportions and rates. For studies reporting simple proportions, rates, or prevalences: absolute numbers and percentages must always be reported together; denominators must be clearly specified for each percentage; rates must be accompanied by the time period or person-time denominator from which they are derived.

REPORTING GUIDELINES (EQUATOR): SUBMISSION REQUIREMENTS AND PROCEDURES

12.1 Mandatory submission of EQUATOR checklists. In accordance with Section 12.3 of the Submission Guidelines, authors must upload the applicable EQUATOR reporting checklist as a supplementary file at the time of submission. The applicable reporting guideline is determined by the primary study design, as specified in the table below.

Study Design	Reporting Guideline	Requirement
Randomized controlled trial	CONSORT 2025. For manuscripts submitted before the adoption date of this Policy, the Editorial Office may request alignment with CONSORT 2025 during revision	Mandatory
Observational study (cohort, case-control, cross-sectional)	STROBE	Mandatory
Systematic review/meta-analysis	PRISMA 2020	Mandatory
Case report	CARE	Mandatory
Diagnostic/prognostic accuracy study	STARD 2015	Mandatory
Animal research	ARRIVE 2.0	Mandatory
Clinical trial protocol	SPIRIT 2025	Mandatory
Systematic review protocol	PRISMA-P	Strongly recommended
Qualitative research	COREQ or SRQR	Strongly recommended
Sex/gender analysis	SAGER	Strongly recommended
Economic evaluation	CHEERS 2022	Strongly recommended
Prediction model / AI-based model	TRIPOD+AI	Strongly recommended
Scoping review	PRISMA-ScR	Strongly recommended

12.2 Checklist completion. The checklist must be completed in full. For each checklist item, the page number in the manuscript where the item is addressed must be indicated. Not applicable items must be marked as "N/A" with a brief explanation. Incomplete checklists, those that contain only "see text" notations without page references, or those that do not apply to the stated study design will be returned to the corresponding author for correction before the manuscript proceeds to peer review.

12.3 Studies without an applicable EQUATOR guideline. Where, in the judgment of the Editorial Office, no established EQUATOR reporting guideline exists for the study design, the Editorial Office shall communicate this determination to the corresponding author; the Editor-in-Chief may be consulted where necessary. In such cases, authors must state this explicitly in the Methods section, provide a brief rationale for the study design chosen, and describe the statistical methods used with particular attention to the reporting requirements of Sections 4 through 10 of this Policy.

12.4 Extensions and supplementary guidelines. Where an applicable extension to a core EQUATOR guideline exists for the study population or context (e.g., CONSORT extensions for cluster trials, pilot and feasibility trials, or N-of-1 trials; PRISMA-P for systematic review protocols; PRISMA-IPD for meta-analyses of individual participant data), authors are strongly encouraged also to complete the relevant extension checklist.

SEX AND GENDER ANALYSIS IN REPORTING

13.1 Sex and gender as analytical variables. In accordance with the SAGER (Sex and Gender Equity in Research) Guidelines, authors must address the role of sex (biological) and gender (social/identity construct) in the design and reporting of their study.

13.2 Reporting requirements. Authors must:

- declare whether the study analyzed sex and/or gender as variables, and if so, how they were defined and measured;

- report results stratified by sex and/or gender where sample sizes permit and where clinically relevant;

- where sex- or gender-stratified analyses were not conducted, provide a justification and acknowledge this as a limitation;

- avoid the interchangeable use of "sex" and "gender" without explicit definition of the terms as used in the study;

- where data on sex or gender were not collected (e.g., in studies using administrative databases), declare this as a limitation.

13.3 SAGER checklist. The SAGER guideline checklist is strongly recommended for all clinical and observational studies submitted to the Journal (see the table in Section 12.1). Authors are encouraged to refer to the SAGER Guidelines available at genderbasicresearch.org.

STATISTICAL REVIEW IN THE PEER REVIEW PROCESS

14.1 Statistical assessment by peer reviewers. In accordance with the Peer Review Policy, peer reviewers are required to assess the statistical methods and reporting as part of their standard review. Specifically, reviewers must assess:

- whether the statistical methods are appropriate for the study design, data type, and research question;

- whether the reporting of statistical results meets the requirements of this Policy (effect sizes, confidence intervals, p-values, software disclosure, sample size justification);

- whether the conclusions drawn by the authors are supported by the statistical results reported;

- whether the assumptions of the statistical methods applied have been verified and reported.

14.2 Specialist statistical review. The Editor-in-Chief may commission a specialist statistical review for manuscripts that employ statistical methods of particular complexity or novelty, including but not limited to:

- Bayesian analyses;

- adaptive trial designs;

- network meta-analyses;

- complex survival models (competing risks, multi-state models, time-varying covariates);

- machine learning-based prediction models;

- complex longitudinal mixed-effects models;

- genetic association studies (GWAS) and genome-wide analyses.

The Editorial Office will notify authors whose manuscripts are subject to specialist statistical review. The statistical reviewer's report is transmitted to the authors with the decision letter.

14.3 Statistical assessment during pre-screening. During pre-screening (Submission Guidelines, Section 24.2), the Editorial Office verifies that the EQUATOR checklist has been uploaded and is appropriate for the stated study design, and that the manuscript contains a dedicated statistical analysis subsection in the Methods section. Manuscripts that do not include these elements will be returned to the corresponding author before entry into peer review.

COMPUTATIONAL REPRODUCIBILITY AND ANALYTICAL CODE

15.1 Scope. This section governs the availability of statistical code and analytical scripts that support the findings reported in a manuscript. It complements the Data Availability and Sharing Policy (Section 11), which establishes the general framework for code and software availability.

15.2 Code availability requirement. Authors must make all analytical code - including statistical scripts, data processing pipelines, and annotated notebooks - available in an accessible format at the time of publication, unless legal, institutional, or ethical constraints prevent disclosure. Where code cannot be shared, authors must provide a detailed justification in the Code Availability Statement (Section 15.4). Code availability enhances reproducibility and supports post-publication scrutiny.

15.3 Repositories and formats. Analytical code should be deposited in a recognized open-access repository (e.g., Zenodo, GitHub with a DOI minted via Zenodo, Open Science Framework, or an equivalent), and a persistent identifier should be provided. The preferred formats are annotated scripts (e.g., R Markdown, Jupyter Notebook, Quarto) that integrate code, output, and explanatory text. Raw, unannotated syntax files are acceptable, but authors are encouraged to include sufficient comments to support independent execution.

15.4 Code availability statement. A code availability statement must be included in the Data Availability Statement (Section 5). It must state whether the analytical code is publicly available, available upon reasonable request, or not available, with the reason in the last case.

15.5 Software environment documentation. Where code is deposited, the software environment must be documented (e.g., via a requirements.txt file in Python, a sessionInfo() or renv.lock file in R, or an equivalent environment capture method). This documentation must specify the version of all packages used to ensure forward compatibility.

PRESENTATION OF STATISTICAL RESULTS IN TABLES AND FIGURES

16.1 Tables. Tables presenting statistical results must comply with the following requirements:

- all numerical values must be accompanied by appropriate units of measurement in column or row headings, not in individual cells;

- statistical measures of variation and precision (SD, SEM, IQR, range, CI) must be clearly defined in column or row headings, not by footnote alone;

- odds ratios, hazard ratios, relative risks, mean differences, and other effect measures must be accompanied by their 95% CIs in the same cell or column;

- statistical test results (test statistic, degrees of freedom, p-value) may be presented in a dedicated column or as table footnotes;

- table footnotes must define all abbreviations used in the table;

- the use of asterisks or symbols to indicate significance thresholds (e.g., *p < 0.05) is discouraged; exact p-values are preferred;

- tables must clearly distinguish pre-specified primary analyses from secondary and exploratory analyses.

16.2 Figures presenting statistical data. For figures presenting statistical data (including box plots, bar graphs, scatter plots, forest plots, Kaplan-Meier curves, and receiver operating characteristic curves):

- the sample size must be stated in the figure or its legend;

- error bars must always be labeled and defined (SD, SEM, 95% CI); unlabelled error bars are not acceptable;

- for box plots: the central line, boxes, and whiskers must be defined (median, IQR, and range or 1.5xIQR) in the figure legend;

- for Kaplan-Meier curves: the number at risk at specified time points must be presented below the survival curve; the p-value from the log-rank test and the HR with 95% CI from Cox regression must be stated;

- for forest plots: the pooled estimate and its 95% CI, the measure of heterogeneity (I2 and Q), and individual study weights must be shown;

- color should be used to distinguish groups only when it conveys information not otherwise available; figures must be interpretable in greyscale for accessibility and printing.

16.3 Numerical precision. Statistical results must be presented with appropriate precision. The following conventions apply unless otherwise justified:

- means, SDs, and medians: report to 1 more decimal place than the raw data;

- percentages: report to 1 decimal place; for proportions based on a denominator of fewer than 100, report to the nearest whole number;

- p-values: see Section 4.2(a);

- odds ratios, hazard ratios, relative risks: report to 2 decimal places;

- confidence interval bounds: report to the same number of decimal places as the point estimate.

VIOLATIONS AND CONSEQUENCES

17.1 Pre-publication violations. Non-compliance with the requirements of this Policy at the submission stage may result in:

(x) return of the manuscript to the corresponding author before entry into peer review, where the violation concerns the absence of required elements (e.g., no EQUATOR checklist, no statistical analysis subsection, no sample size justification);

(y) a major or minor revision request where the violation concerns inadequate reporting that can be corrected without re-analysis;

(z) rejection of the manuscript where the statistical methods are found to be fundamentally inappropriate for the research question or data type, or where results have been selectively reported in a manner that undermines the validity of the conclusions.

17.2 Post-publication violations. Post-publication statistical concerns - including selective reporting, p-hacking, outcome switching (reporting outcomes other than those pre-specified without disclosure), and data fabrication or manipulation affecting statistical results - are governed by the Corrections and Retractions Policy and the Complaints, Appeals and Whistleblowing Policy. Where a post-publication investigation identifies a statistical violation of material significance to the conclusions of the published article, the appropriate corrective action (Correction, Expression of Concern, or Retraction) will be applied in accordance with the Corrections and Retractions Policy and COPE flowcharts.

POST-PUBLICATION STATISTICAL CONCERNS

18.1 Reporting statistical concerns. Readers, reviewers, and third parties who identify potential statistical errors or misconduct in published articles may report their concerns to the Editorial Office using the procedures established in the Complaints, Appeals and Whistleblowing Policy.

18.2 Investigation process. Statistical concerns reported after publication are assessed initially by the Editor-in-Chief, who determines whether the concern is sufficiently substantiated to warrant formal investigation. The Editor-in-Chief may commission an independent statistical review to assess the validity of the concern. Authors are notified of the investigation and are invited to respond to the concerns raised.

18.3 Preservation of pre-publication materials. In cases where a post-publication investigation is opened, the Editorial Office may request from the authors:

- the original raw dataset underlying the published analyses;

- the statistical code or syntax used to generate the reported results;

- the Statistical Analysis Plan and any pre-registration documentation;

- correspondence with co-authors relevant to the statistical analysis.

Please provide these materials promptly; failure to do so may be considered a factor in determining the investigation outcome.

18.4 GDPR and data protection constraints. Any request for raw data in connection with a post-publication investigation shall be subject to applicable legal, ethical, privacy, data protection, and consent constraints. Where individual-level clinical or personal data are involved, authors may be required to provide de-identified or anonymized datasets, data dictionaries, analysis logs, or independent verification by an authorized third party, as appropriate. Pseudonymized or partially de-identified data that may permit re-identification remain personal data under Regulation (EU) 2016/679 (GDPR) and may only be shared through a controlled-access mechanism consistent with the data protection framework established in the Data Availability and Sharing Policy (Section 19) and the Privacy and Data Protection Policy. Authors are not required to share data that cannot be shared without breaching applicable law, ethical approvals, or participant consent, provided that this limitation is documented and communicated to the Editorial Office promptly. Where data cannot be shared due to legal, ethical, or regulatory constraints, authors are expected to propose and, where feasible, implement alternative mechanisms to support reproducibility to the greatest extent possible. Such mechanisms may include controlled-access data repositories managed by the hosting institution, independent third-party verification of results, or the provision of synthetic datasets that reproduce the original data's statistical properties without exposing personal information.

ANNUAL REVIEW

This Policy is reviewed annually by the Editor-in-Chief and Edikta S.r.l. The annual review assesses:

(aa) updates to the SAMPL Guidelines, EQUATOR reporting guidelines (including new CONSORT extensions, STROBE extensions, PRISMA extensions, TRIPOD updates), and ASA guidance on statistical inference;

(bb) the Journal's experience in applying this Policy, including recurring patterns of statistical reporting deficiency identified during pre-screening and peer review;

(cc) developments in open science and computational reproducibility relevant to statistical reporting in biomedical journals;

(dd) updates to Scopus and Web of Science source evaluation criteria relevant to statistical and methodological quality;

(ee) new statistical methods emerging in gynecological and obstetric research that may require policy-specific guidance.

Material revisions to this Policy are communicated to authors, reviewers, and editorial board members through the Journal's official communication channels. A versioned record of all editions of this Policy is maintained and publicly accessible on the Journal's website.

CROSS-REFERENCES WITHIN THE EGO EDITORIAL POLICY FRAMEWORK

This Policy is part of the integrated editorial policy framework of European Gynecology and Obstetrics (EGO). The following cross-references apply:

- Editorial Policy Statement: establishes the foundational commitment to research transparency and methodological rigor that this Policy operationalizes; declares statistical transparency and reproducibility as standing obligations of all published research.

- Research Ethics Policy, Section 20: identifies this Policy as the governing document for detailed statistical reporting requirements and EQUATOR compliance; declares the applicable reporting standard for each article type.

- Submission Guidelines, Section 12.3: establishes the table of mandatory and strongly recommended EQUATOR reporting guidelines, reproduced and expanded in Section 12.1 of this Policy; governs the submission of EQUATOR checklists as required supplementary files. Section 27 cross-references this Policy.

- Data Availability and Sharing Policy, Section 11: establishes the framework for code and software availability that this Policy implements for analytical code; governs the Code Availability Statement required under Section 15.4 of this Policy.

- Peer Review Policy: governs the peer review process within which statistical assessment takes place; this Policy defines the statistical assessment criteria applicable to peer reviewers (Section 14.1) and the conditions under which specialist statistical review is commissioned (Section 14.2).

- Artificial Intelligence Policy: governs AI-based methods used in research; where AI models are used for analysis or prediction, the statistical reporting requirements of Section 11.5 of this Policy apply alongside the AIP requirements.

- Corrections and Retractions Policy: governs post-publication corrective measures applicable when statistical violations are confirmed, including selective reporting, outcome switching, and data manipulation affecting published results.

- Complaints, Appeals and Whistleblowing Policy: governs the reporting and investigation of post-publication statistical concerns, including suspected selective reporting, p-hacking, and manipulation of statistical results.

- Privacy and Data Protection Policy: governs the processing and protection of personal data by the Journal and the Publisher; Section 18.4 of this Policy implements the data protection constraints applicable to post-publication investigations involving raw data.

- Publication Ethics Policy: establishes the ethical framework within which statistical misconduct (including data fabrication, falsification, and selective reporting) is classified as publication misconduct.

21. NORMATIVE REFERENCES

Statistical and reporting standards

- Lang TA, Altman DG. Basic statistical reporting for articles published in clinical medical journals: the SAMPL Guidelines. In: Smart P, Maisonneuve H, Polderman A (eds). Science Editors' Handbook. European Association of Science Editors, 2013.

- Wasserstein RL, Lazar NA. The ASA Statement on p-Values: Context, Process, and Purpose. The American Statistician. 2016;70(2):129-133. https://doi.org/10.1080/00031305.2016.1154108

- Wasserstein RL, Schirm AL, Lazar NA. Moving to a World Beyond "p < 0.05". The American Statistician. 2019;73(sup1):1-19. https://doi.org/10.1080/00031305.2019.1583913

- Hopewell S, Chan AW, Collins GS, et al. CONSORT 2025 statement: updated guideline for reporting randomized trials. PLOS Medicine. 2025;22(4):e1004587. https://doi.org/10.1371/journal.pmed.1004587. Published simultaneously in The Lancet, BMJ, and JAMA (2025).

- von Elm E, Altman DG, Egger M, et al. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. PLoS Medicine. 2007;4(10):e296. https://doi.org/10.1371/journal.pmed.0040296

- Page MJ, McKenzie JE, Bossuyt PM, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ. 2021;372:n71. https://doi.org/10.1136/bmj.n71

- Bossuyt PM, Reitsma JB, Bruns DE, et al. STARD 2015: an updated list of essential items for reporting diagnostic accuracy studies. BMJ. 2015;351:h5527. https://doi.org/10.1136/bmj.h5527

- Gagnier JJ, Kienle G, Altman DG, et al. The CARE guidelines: consensus-based clinical case reporting guideline development. Glob Adv Health Med. 2013;2(5):38-43. https://doi.org/10.7453/gahmj.2013.008

- Riley RD, Collins GS, Van Calster B, et al. TRIPOD+AI statement: updated guidance for reporting clinical prediction models that use regression or machine learning methods. BMJ. 2024;385:e078378. https://doi.org/10.1136/bmj-2023-078378

- Percie du Sert N, Hurst V, Ahluwalia A, et al. The ARRIVE guidelines 2.0: updated guidelines for reporting animal research. PLOS Biology. 2020;18(7):e3000411. https://doi.org/10.1371/journal.pbio.3000411

- Chan AW, Tetzlaff JM, Altman DG, et al. SPIRIT 2025 statement: updated guideline for protocols of randomized trials. The Lancet. 2025. https://doi.org/10.1016/S0140-6736(25)00770-6. Published simultaneously in Nature Medicine, JAMA, and PLOS Medicine (2025).

- Tong A, Sainsbury P, Craig J. Consolidated criteria for reporting qualitative research (COREQ): a 32-item checklist for interviews and focus groups. Int J Qual Health Care. 2007;19(6):349-357. https://doi.org/10.1093/intqhc/mzm042

- Moher D, Stewart L, Shekelle P. Implementing PRISMA-P: recommendations for prospective authors. Syst Rev. 2015;4:1. https://doi.org/10.1186/2046-4053-4-1

- Tricco AC, Lillie E, Zarin W, et al. PRISMA Extension for Scoping Reviews (PRISMA-ScR): checklist and explanation. Ann Intern Med. 2018;169:467-473. https://doi.org/10.7326/M18-0850

- Heidari S, Babor TF, De Castro P, Tort S, Curno M. Sex and Gender Equity in Research: rationale for the SAGER guidelines and recommended use. Research Integrity and Peer Review. 2016;1:2. https://doi.org/10.1186/s41073-016-0007-6

- Higgins JPT, Thomas J, Chandler J, et al. (eds). Cochrane Handbook for Systematic Reviews of Interventions, version 6.4. Cochrane, 2023. Available at: training.cochrane.org/handbook.

- International Committee of Medical Journal Editors (ICMJE). Recommendations for the Conduct, Reporting, Editing, and Publication of Scholarly Work in Medical Journals. Updated January 2026. Available at: icmje.org.

- Committee on Publication Ethics (COPE). Core Practices. Current version. Available at: publicationethics.org.

- EQUATOR Network. Reporting guidelines for health research. Available at: equator-network.org.

Editorial, indexing, and comparative policy references

- Elsevier / Scopus. Source Evaluation Criteria. Content Coverage Guide. Updated 2024. Available at: scopus.com.

- Wiley. Best Practice Guidelines on Publishing Ethics. Current version. Available at: authorservices.wiley.com.

- BJOG: An International Journal of Obstetrics and Gynecology. Editorial policies for authors. Available at: obgyn.onlinelibrary.wiley.com.

- Human Reproduction Update. Guidelines for authors. Available at: academic.oup.com/humupd.