Considerations for Conducting Meta-analysis in Diagnostic Pathology
Bibliographic record
Abstract
We read with great interest the recently published article entitled “Evidence-Based Pathology: Systematic Literature Reviews as the Basis for Guidelines and Best Practices” by Marchevsky and Wick1 and here submit the following comments concerning systematic review (SR) and meta-analysis (MA). Our comments seem supplemental to this elegant review of SR and MA.First, specific guidelines for conducting SR and MA have been developed and used, because the methodologic quality may vary significantly among studies, resulting in varied reliability, validity, and clinical applicability.2 These guidelines ensure a full disclosure of the process, reproducibility of the analysis process and results, and the inclusion of only qualified data/studies. The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guideline is the most widely used.2 The recently published PRISMA-P was created for MA planning and protocol development.3 The Newcastle-Ottawa Scale, another system to evaluate study/data quality, may also be used in spite of being less popular than PRISMA.4 An international prospective registry of SR and MA has also been created and used, which may further reduce selection and other biases. Unfortunately, to our knowledge, the published SRs and MAs in diagnostic pathology seem to less frequently follow these guidelines (M.K., L.Z., unpublished data), which certainly deserves more awareness and utilization. All researchers (and our patients) would benefit from the consistent use of these tools in pathology.Second, data heterogeneity is a critical factor for data inclusion and analysis model selection. Therefore, before conducting an MA, heterogeneity of the included study results should be interrogated by using either the I2 test or the Cochran Q test; forest plots are a visual method to explore potential heterogeneity. If significant heterogeneity is present in the reported/included data, one may proceed with the following options: reporting a summary of reported/included data without further MA, subgroup and sensitivity analyses, and analyses using random-effects models (versus a fixed-effects model, which is appropriate for a homogeneous group of study results).Third, interested investigators may use commercial statistical software to conduct MAs, including SAS (SAS Institute Inc, Cary, North Carolina), Stata (Stata-Corp LP, College Station, Texas), and SPSS (IBM, Armonk, New York). Free software is also available with limited functionality. Review Manager or RevMan (The Cochrane Collaboration, The Nordic Cochrane Centre, Copenhagen, Denmark) is one of the leading choices. Open Meta-analysis (http://www.cebm.brown.edu/open_meta) was sponsored by the US Agency for Health Research and Quality and is very easy to use. MetaEasy (http://www.statanalysis.co.uk/meta-analysis.html), an add-on for Microsoft Excel (Microsoft, Redmond, Washington), may also be considered.In summary, we provide some additional information pertinent to the interesting article by Marchevsky and Wick.1 Our recent study shows that MA has been underutilized in diagnostic pathology despite its similar adjusted citation ratios to those of review articles.5 Our study also shows an increase in frequency of MA in diagnostic pathology in recent years, suggesting a growing interest in MA of pathology. Nonetheless, we agree that more efforts are needed to increase the awareness, training, and utilization of SR and MA in pathology.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.803 | 0.929 |
| Meta-epidemiology (narrow) | 0.006 | 0.007 |
| Meta-epidemiology (broad) | 0.018 | 0.023 |
| Bibliometrics | 0.015 | 0.020 |
| Science and technology studies | 0.004 | 0.017 |
| Scholarly communication | 0.017 | 0.028 |
| Open science | 0.015 | 0.010 |
| Research integrity | 0.022 | 0.033 |
| Insufficient payload (model declined to judge) | 0.010 | 0.005 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; the direct Gemma label and the distilled Codex classifier agree on what is shown here.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".