Emulating randomized trials: treading carefully and pushing the limits
Bibliographic record
Abstract
In their commentary, Wing and Leyrat1 discuss the advantages and potential challenges of designing an observational comparative effectiveness study using the target trial framework based on an existing randomized controlled trial (RCT). This framework helps researchers avoid common pitfalls in the design of such studies, including the choice of time zero, the definition of exposure, and immortal time bias.2 Observational studies based on an existing published RCT are useful to benchmark the findings against those from the RCT and, subsequently, to extend the analyses to populations not represented or under-represented in the RCT.1 However, the issue of biases in the reference RCT merits attention. While using an existing published RCT is appealing as the basis to design an observational study, an underlying premise is that the reference RCT is necessarily well-designed and produces clinically relevant results. However, as recently illustrated in asthma and chronic obstructive pulmonary disease (COPD), RCTs are not immune to design issues and related biases.3 Several large recent RCTs in patients with COPD had several methodological shortcomings related to the randomization scheme, leading to biased results and inconsistent findings across trials.3,-5 The requirement of the abrupt discontinuation of existing treatments followed by randomization can introduce important biases to the trials.3,-5 This effect is more pronounced if the random allocation is immediate after discontinuation, but it can also be important if it occurs after the introduction of a run-in period, depending on the common treatment given during the run-in.6 These design requirements do not mimic any real-life clinical situation and would be difficult, if not impossible, to replicate using an observational design.7,8 Thus, using such RCTs as the target trial for replication would not be suitable, neither in terms of design nor for clinical relevance.
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 | 0.633 | 0.806 |
| Meta-epidemiology (narrow) | 0.005 | 0.004 |
| Meta-epidemiology (broad) | 0.015 | 0.011 |
| Bibliometrics | 0.009 | 0.007 |
| Science and technology studies | 0.006 | 0.071 |
| Scholarly communication | 0.026 | 0.065 |
| Open science | 0.023 | 0.017 |
| Research integrity | 0.071 | 0.107 |
| Insufficient payload (model declined to judge) | 0.006 | 0.004 |
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".