Covariate Adjusted Confidence Intervals for the Incremental Cost-Effectiveness Ratio in Non Randomized Trails, Comparing Fieller and Bootstrap Methods - An Example from the Sarah Study
Bibliographic record
Abstract
Background: Assessment of uncertainty in cost-effectiveness analysis is of high importance as it greatly impacts use of cost-effectiveness results in decision making and priority setting. While in recent years various methods have been suggested for the estimation of confidence intervals to the incremental cost-effectiveness ratio (ICER) when groups are randomly assigned to the treatment groups, relatively little work has been done in the case of non-randomized studies. Objective: Aim of this study is to describe a method of resampling based confidence intervals in the context of non-randomized trial requiring covariate adjustment and to compare it to a recently suggested parametric approach. Its suitability and first indication of relative performance are discussed using data from a non randomized trial comparing inpatient and outpatient rehabilitation in patients after an acute myocardial infarction (SARAH-Study). Methods: Covariate adjustment was performed using a seemingly unrelated regression (SURE) approach. Regression equations for costs and effects were jointly estimated allowing to incorporate the correlation between cost and effects. Point estimate of the ICER ware calculated by the quotient of the coefficient of the therapy-dummy in the cost and effect equation. Parametric confidence intervals for the ICER were calculated using the Fieller formula. Non parametric approach was based on bootstrapping the SURE regression model. Resampled point estimates of the ICER were transformed into angular deviations and standard methods for bootstrap confidence intervals were applied. Results: The 163 patients enrolled into the SARAH-trial preferred predominantly the inpatient to the outpatient rehabilitation (112 vs. 51). Based on pre-post rehabilitation data patients in the outpatient group incurred lower gains in quality of life (p=0,094) but caused also considerably lower cost (p0.12 since conclusions of the analytic term became complex with smaller levels of alpha, reflecting the insignificance of the effect estimate. Due to the non-parametric nature of the bootstrap method intervals could be calculated for any confidence level. They also tended to be wider for the same alpha levels compared to Fieller confidence intervals. Conclusion: Assessment of the uncertainty in the ICER is crucial also in a non randomized context. Different methods can be applied for this purpose. The Fieller method which is based on the normality assumptions tends to give exact solution if assumptions are met. In the example however it exhibited its known problem of yielding no solutions for alpha levels of interest if differences in effects and/or costs are too small. The approach of bootstrapping the regression seemed here suitable, although it tended to show wider confidence intervals. This non-parametric approach offered a valuable option in this non randomized trial setting, of which a detailed analysis of performance will be investigated in further simulation analyses.
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 | 0.192 | 0.390 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.002 | 0.004 |
| Bibliometrics | 0.002 | 0.002 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.001 | 0.002 |
| Open science | 0.002 | 0.001 |
| Research integrity | 0.002 | 0.002 |
| Insufficient payload (model declined to judge) | 0.002 | 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 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".