Bias in identification of the best treatment in a Bayesian network meta-analysis for binary outcome: a simulation study
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
Network meta-analysis (NMA) has emerged as a useful analytical tool allowing comparison of multiple treatments based on direct and indirect evidence. Commonly, a hierarchical Bayesian NMA model is used, which allows rank probabilities (the probability that each treatment is best, second best, and so on) to be calculated for decision making. However, the statistical properties of rank probabilities are not well understood. This study investigates how rank probabilities are affected by various factors such as unequal number of studies per comparison in the network, the sample size of individual studies, the network configuration, and effect sizes between treatments. In order to explore these factors, a simulation study of four treatments (three equally effective treatments and one less effective reference) was conducted. The simulation illustrated that estimates of rank probabilities are highly sensitive to both the number of studies per comparison and the overall network configuration. An unequal number of studies per comparison resulted in biased estimates of treatment rank probabilities for every network considered. The rank probability for the treatment that was included in the fewest number of studies was biased upward. Conversely, the rank of the treatment included in the most number of studies was consistently underestimated. When the simulation was altered to include three equally effective treatments and one superior treatment, the hierarchical Bayesian NMA model correctly identified the most effective treatment, regardless of all factors varied. The results of this study offer important insight into the ability of NMA models to rank treatments accurately under several scenarios. The authors recommend that health researchers use rank probabilities cautiously in making important decisions.
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.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.501 | 0.460 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.014 | 0.010 |
| Bibliometrics | 0.000 | 0.003 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.001 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
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 it