A comparison of the statistical performance of different meta-analysis models for the synthesis of subgroup effects from randomized clinical trials
Bibliographic record
Abstract
BACKGROUND: When investigating subgroup effects in meta-analysis, it is unclear whether accounting in meta-regression for between-trial variation in treatment effects, but not between-trial variation in treatment interaction effects when such effects are present, leads to biased estimates, coverage problems, or wrong standard errors, and whether the use of aggregate data (AD) or individual-patient-data (IPD) influences this assessment. METHODS: Seven different models were compared in a simulation study. Models differed regarding the use of AD or IPD, whether they accounted for between-trial variation in interaction effects, and whether they minimized the risk of ecological fallacy. RESULTS: Models that used IPD and that allowed for between-trial variation of the interaction effect had less bias, better coverage, and more accurate standard errors than models that used AD or ignored this variation. The main factor influencing the performance of models was whether they used IPD or AD. The model that used AD had a considerably worse performance than all models that used IPD, especially when a low number of trials was included in the analysis. CONCLUSIONS: The results indicate that IPD models that allow for the between-trial variation in interaction effects should be given preference whenever investigating subgroup effects within a meta-analysis.
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 | 0.443 | 0.708 |
| Meta-epidemiology (narrow) | 0.005 | 0.004 |
| Meta-epidemiology (broad) | 0.015 | 0.056 |
| Bibliometrics | 0.011 | 0.009 |
| Science and technology studies | 0.001 | 0.003 |
| Scholarly communication | 0.008 | 0.008 |
| Open science | 0.007 | 0.005 |
| Research integrity | 0.006 | 0.007 |
| Insufficient payload (model declined to judge) | 0.004 | 0.001 |
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".