Effectiveness of adverse effects search filters: drugs versus medical devices
Bibliographic record
Abstract
OBJECTIVE: The study tested the performance of adverse effects search filters when searching for safety information on medical devices, procedures, and diagnostic tests in MEDLINE and Embase. METHODS: The sensitivity of 3 filters was determined using a sample of 631 references from 131 rapid reviews related to the safety of health technologies. The references were divided into 2 sets by type of intervention: drugs and nondrug health technologies. Keyword and indexing analysis were performed on references from the nondrug testing set that 1 or more of the filters did not retrieve. RESULTS: For all 3 filters, sensitivity was lower for nondrug health technologies (ranging from 53%-87%) than for drugs (88%-93%) in both databases. When tested on the nondrug health technologies set, sensitivity was lower in Embase (ranging from 53%-81%) than in MEDLINE (67%-87%) for all filters. Of the nondrug records that 1 or more of the filters missed, 39% of the missed MEDLINE records and 18% of the missed Embase records did not contain any indexing terms related to adverse events. Analyzing the titles and abstracts of nondrug records that were missed by any 1 filter, the most commonly used keywords related to adverse effects were: risk, complications, mortality, contamination, hemorrhage, and failure. CONCLUSIONS: In this study, adverse effects filters were less effective at finding information about the safety of medical devices, procedures, and tests compared to information about the safety of drugs.
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 | 0.141 | 0.547 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.004 | 0.005 |
| Bibliometrics | 0.015 | 0.008 |
| Science and technology studies | 0.001 | 0.002 |
| Scholarly communication | 0.005 | 0.008 |
| Open science | 0.002 | 0.002 |
| Research integrity | 0.004 | 0.001 |
| Insufficient payload (model declined to judge) | 0.002 | 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".