Prevalence of rare anatomic variants - publication bias due to selective reporting in meta-analyses studies
Introduction: Meta-analyses of prevalence studies reporting rare anatomic variants are prone to selective reporting of non-null, confirmatory results, thus resulting in publication bias. Aim: We aim to numerically approach this bias and evaluate the most widely used methods for its assessm...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Pensoft Publishers
2024-12-01
|
| Series: | Folia Medica |
| Online Access: | https://foliamedica.bg/article/137881/download/pdf/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Introduction: Meta-analyses of prevalence studies reporting rare anatomic variants are prone to selective reporting of non-null, confirmatory results, thus resulting in publication bias. Aim: We aim to numerically approach this bias and evaluate the most widely used methods for its assessment. Materials and methods: We used probability theory over three sets of real-world and a single set of simulation data to assess the maximum publication bias due to selective reporting in meta-analysis of rare anatomic variants. For each individual study, we approximated the theoretical maximum of the neglected, “not published”, part of the truth, as revealed by the corresponding null, non-confirmatory result. Furthermore, we computed the relevant pooled estimate using the Freeman-Tuckey double arcsine transformation under the random effects model and the restricted maximum likelihood (REML) estimation in STATA 18. Finally, we comparatively applied Egger’s and Begg’s test, trim-and-fill analysis, and Doi plot / LFK index to assess publication bias before and after correction for maximum selective reporting. Results: Meta-analyses of prevalence studies reporting rare anatomic variants may exhibit significant publication bias due to selective reporting. This bias grows larger as the included studies report less confirmatory cases and may theoretically reach 50%. From all tools assessing publication bias, the LFK index was suggested to be the most informative. Conclusions: Selective reporting might result in inflated publication bias in meta-analyses of prevalence studies reporting rare anatomic variants. Although the accurate assessment of this kind of bias is highly challenging in both theory and practice, the use of the LFK index is proposed as the most appropriate tool for that purpose. |
|---|---|
| ISSN: | 1314-2143 |