Causal Variance Decompositions for Evaluating Healthcare Provider Performance
Bibliographic record
Abstract
There has been a growing interest in striving for accountability and transparency in healthcare by quantifying and comparing the performance of providers. This requires measuring the quality of care provided to patients, which is often done using disease-specific quality indicators (QIs). Before considering intervening and improving quality gaps identified by using QIs, the chosen indicator has to demonstrate existing differences in the quality of care. Therefore, the overarching aim of this thesis is to develop metrics that assess the usefulness of QIs in identifying existing differences in the quality of care provided between providers. This can be achieved by decomposing and quantifying the overall variation between providers in a causal inference framework. The first objective is to develop a three-way causal variance decomposition to attribute observed variation in care received into that causally explained by hospitals’ performance, the patient case-mix itself, and unexplained variation. In the context of surgical cancer treatments, the performance variations can be due to hospital and/or surgeon level differences, creating a hierarchical clustering. The second objective is to generalize the causal variance decomposition for several exposure hierarchies. For this purpose, I derive a decomposition of the observed variation in care received into four parts, explained by patient case- mix, causally explained by hospitals’ performance, causally explained by surgeons’ performance, and unexplained variation. While the use of variance decompositions can demonstrate differences in quality of care, causal mediation analysis can be used to investigate the care pathways that lead to these differences in performance between hospitals. Combining these two approaches enables decomposing between-hospital variation in an outcome-type indicator into two parts: the part that is mediated through a specific process of care and the part that is due to all other pathways. The third objective is to derive a causal mediation analysis decomposition of the between-hospital variance into three parts, the natural indirect effects, the natural direct effects, and a covariance-type term. Model-based estimators are formulated for the variance decompositions, and their performance are investigated through simulation studies. All methods are demonstrated in real data applications in Ontario kidney cancer data.
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 | 0.061 | 0.176 |
| Meta-epidemiology (narrow) | 0.003 | 0.001 |
| Meta-epidemiology (broad) | 0.002 | 0.007 |
| Bibliometrics | 0.009 | 0.008 |
| Science and technology studies | 0.001 | 0.003 |
| Scholarly communication | 0.003 | 0.003 |
| Open science | 0.002 | 0.004 |
| Research integrity | 0.002 | 0.004 |
| Insufficient payload (model declined to judge) | 0.011 | 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; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
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".