Publication bias is a fundamental threat to the validity of systematic reviews and meta-analyses in clinical medicine. Yet current practice often reduces its assessment to the mechanical application of funnel plots, asymmetry tests, or single adjustment procedures, with limited attention to the underlying assumptions, alternative explanations, or implications for evidence certainty. This narrative methodological article reframes publication bias assessment as an interpretive and editorial responsibility rather than a purely technical problem. We examine what commonly used methods can and cannot reliably support. Detection tools function as nonspecific stress tests that identify deviations from simplified models; they do not diagnose selective publication but highlight situations in which the underlying assumptions require closer inspection. Adjustment approaches, including trim-and-fill, selection models, and regression-based methods, generate hypothetical estimates under unverifiable assumptions**, and therefore provide** sensitivity analyses rather than corrections that recover the true underlying effect. Divergence across adjustment methods is particularly informative, signaling inferential fragility rather than analytical failure. We identify five recurring misinterpretations encountered in peer review: equating asymmetry with proof of publication bias; privileging bias-adjusted estimates as inherently more credible; relying on a single adjustment method without examining assumption dependence; ignoring the plausibility of adjustment direction and magnitude; and overlooking implications for certainty of evidence. Editors and reviewers should prioritize transparency of assumptions, seriously consider alternative explanations, and calibrate conclusions proportionately. Viewing publication bias assessment as an interpretive responsibility rather than a methodological checklist promotes more disciplined inference and strengthens trust in clinical evidence synthesis.

Background
Large language model (LLM)–based tools are increasingly used to automate risk-of-bias assessment of randomized controlled trials with the revised Cochrane tool (RoB 2.0). However, their reproducibility, their accuracy, and their fidelity to the deterministic algorithm mapping signalling-question responses to domain judgments remain unclear.
Methods
In a conference workshop, participants used an identical prompt and tool versions to assess one RCT with three configurations and entered each tool’s output verbatim (13, 15, and 10 runs). One experienced reviewer’s assessment served as the reference. For each run we computed run-to-run reproducibility, agreement with the expert, and conformance between the tool’s stated domain judgment and the judgment implied by applying the RoB 2.0 algorithm to that run’s own signalling answers.
Results
All configurations showed substantial run-to-run variability under identical conditions, greatest in the conditionally complex domain 2 (Gemini pairwise agreement 0.34). Expert agreement varied widely across configurations (mean 5–60%), and one configuration systematically under-rated risk. Even in domains with a fully specified algorithm, stated judgments frequently diverged from the value implied by the tool’s own signalling answers (domain 2, mean 42%). Skip-logic violations occurred in 69%, 80%, and 100% of runs. Recomputing domain judgments from the signalling answers improved expert agreement for the general-purpose configurations.
Conclusion
LLM-based RoB 2.0 assessments exhibited variability and errors. Whatever tool is adopted, its characteristics and variability must be recognized. Having the tool perform only atomic (single) judgments while delegating aggregation to the algorithm, together with human review, may improve accuracy.

Intravenous intralipid is widely used for women with recurrent reproductive failure —including recurrent pregnancy loss (RPL), recurrent implantation failure (RIF), and repeated IVF failure—despite limited evidence. This systematic review evaluated the effectiveness and safety of intralipid infusion in this population. We searched seven electronic databases from inception to April 14, 2025. RCTs and non-randomized comparative studies (NRCSs) were included for effectiveness analysis; case series and reports for safety. Two reviewers independently screened and extracted data. Risk of bias was assessed using RoB 2.0 and RoBANS 2.0. Meta-analyses used a random-effects model (risk ratios [RR] with 95% CI). Evidence certainty was assessed using GRADE. Seventeen studies were included (8 effectiveness, 11 safety). Pooled RCT data showed intralipid significantly improved clinical pregnancy rate vs. no treatment (RR 2.31, 95% CI 1.42–3.74; I² = 0%; GRADE certainty: low); this effect was not observed in the pooled NRS data for live birth rate or clinical pregnancy rate, and one large NRCS reported a significantly higher miscarriage rate in the intralipid group (RR 1.12, 95% CI 1.04–1.20). No significant differences were observed vs. IVIG or steroids. Serious adverse events were rare. GRADE certainty was very low to low across all outcomes. Current evidence does not support the routine use of intralipid in women with recurrent reproductive failure. While pooled RCT data suggest a potential benefit in clinical pregnancy rate, evidence certainty is low, and this benefit was not observed in the pooled NRS data. Adequately powered RCTs with standardized protocols and live birth rate as the primary endpoint are needed.

Background
Post-thyroidectomy hypocalcemia is the most frequent complication after total thyroidectomy. Preoperative vitamin D deficiency has been suggested as a potential risk factor, but inconsistencies exist in the literature, possibly related to methodological differences. To evaluate whether study design and risk of bias influence the association between preoperative vitamin D deficiency and postoperative hypocalcemia in patients undergoing total thyroidectomy.
Methods
This is a secondary analysis of a previously conducted systematic review. We included observational studies evaluating the association between preoperative vitamin D levels and postoperative hypocalcemia. Methodological quality was assessed using the QUIPS tool. Subgroup analyses were performed based on study design (prospective vs. retrospective) and overall risk of bias (high vs. low/moderate).
Results
Twenty-eight studies comprising 4994 patients were included. Nineteen studies had a prospective design. Both prospective and retrospective studies showed an association between vitamin D deficiency and hypocalcemia; however, the effect size was lower in prospective studies (OR 1.95; 95% CI 1.28-2.97) compared to retrospective ones (OR 2.18; 95% CI 1.02-4.7). Studies with high risk of bias showed a significant association (OR 2.55; 95% CI 1.4-3.6), while those with low/moderate risk did not (OR 1.71; 95% CI 0.96-3.06).
Conclusion
Study design and methodological quality influence the reported association between vitamin D deficiency and postoperative hypocalcemia. These findings suggest caution when recommending preoperative vitamin D supplementation based solely on observational data.

Systematic reviews and meta-analyses are pivotal for evidence-based decision-making but depend on the availability of precise statistical data. Researchers often encounter studies where essential statistics are missing or presented only in graphs, leading to potential data exclusion and selection bias. This study aims to provide specific methodologies for extracting or reconstructing the statistical parameters required for meta-analysis—specifically effect sizes (MD, OR, RR, HR) and their corresponding variance measures (SD, SE, variance)—from incomplete or graphically reported data. We describe calculation and extraction protocols for five specific scenarios encountered in medical literature: (1) continuous data missing standard deviations; (2) categorical data missing standard errors; (3) calculating risk estimates from frequency tables; (4) extracting continuous data presented solely in graphs; and (5) reconstructing hazard ratios from Kaplan-Meier survival curves. Valid meta-analysis requires both an effect size and a measure of variance. When these are not explicitly reported, they can often be derived from other available statistics or digital extraction from figures. While heterogeneity is inherent in meta-analysis, the methodology allows for error adjustment and robust synthesis. Therefore, preventing data loss via these extraction methods is preferable to excluding studies. Maximizing data inclusion enhances the comprehensive value and statistical power of the final analysis.

This review explores the current landscape of artificial intelligence (AI)-assisted semi-automation tools used in systematic reviews and guideline development. With the exponential growth of medical literature, these tools have emerged to improve efficiency and reduce the workload involved in evidence synthesis. Platforms such as Covidence, EPPI-Reviewer, DistillerSR, and Laser AI exemplify how machine learning and, more recently, large language models (LLMs) are being integrated into key stages of the systematic review process—ranging from literature screening to data extraction. Evidence suggests that these tools can save considerable time, with some achieving average reductions of over 180 hours per review. However, challenges remain in transparency, reproducibility, and validation of AI performance. In response, international initiatives such as the Responsible AI in Evidence Synthesis (RAISE) project and the Guideline International Network (GIN) have proposed frameworks to ensure the ethical, trustworthy, and effective use of AI in health research. These include principles like transparency, accountability, preplanning, and continuous evaluation. This review highlights both the opportunities and limitations of adopting AI in evidence synthesis and underscores the importance of human oversight and rigorous validation to ensure that such tools enhance, rather than compromise, the integrity of systematic reviews and guideline development.

Evidence-Based Medicine (EBM) demands systematic changes across the healthcare system, essential for enhancing patient safety and quality of medical care. To address the question, "Are we adopting scientific methods to optimize patient safety and enhance treatment efficacy?", assessing the level of EBM implementation is crucial. The adoption rate of evidence-based medical practices varies across countries and medical fields, often being lower in resource-limited settings. In South Korea, there have been several documented cases where the adoption of non-evidence-based practices, such as CARVAR surgical procedures not based on scientific evidence, has led to severe patient safety issues, thereby raising significant concerns about the quality of medical care provided. Conversely, the ABBA Study exemplifies successful application of EBM, demonstrating how scientific research assessed the risk of intracranial hemorrhage in patients with low-dose PPA in OTC cold medicines. This study not only confirmed the associated risks but also influenced health policy, resulting in the withdrawal for PPA-containing OTC cold medicines in Korea. This positive example highlights the imperative for governments, healthcare institutions, and medical schools to expedite the transition to evidence-based, patient-centered healthcare by fostering a robust commitment to systematic reviews and enhanced support for clinical research. The Korean Society of Evidence-Based Medicine (KSEBM) is expected to play a significant role in embedding these core strategies domestically