Applying the Narrow Forms of PubMed Methods-based and Topic-based Filters Increases Nephrologists’ Search Efficiency
Bibliographic record
Abstract
Objective – To determine whether the use of PubMed methods-based filters and topic-based filters, alone or in combination, improves physician searching. Design – Mixed methods, survey questionnaire, comparative. Setting – Canada. Subjects – Random sample of Canadian nephrologists (n=153), responses (n=115), excluded (n=15), total (n=100). Methods – The methods are described in detail in a previously published study protocol by a subset of the authors (Shariff et al., 2010). One hundred systematic reviews on renal therapy were identified using the EvidenceUpdates service (http://plus.mcmaster.ca/EvidenceUpdates) and a clinical question was derived from each review. Randomly-selected Canadian nephrologists were randomly assigned a unique clinical question derived from the reviews and asked, by survey, to provide the search query they would use to search PubMed. The survey was administered until one valid search query for each of the one hundred questions was received. The physician search was re-executed and compared to searches where either or both methods-based and topic-based filters were applied. Nine searches for each question were conducted: the original physician search, a broad and narrow form of the clinical queries therapy filter, a broad and narrow form of the nephrology topic filter and combinations of broad and narrow forms of both filters. Significance tests of comprehensiveness (proportion of relevant articles found) and efficiency (ratio of relevant to non-relevant articles) of the filtered and unfiltered searches were conducted. The primary studies included in the systematic reviews were set as the reference standard for relevant articles. As physicians indicated they did not scan beyond two pages of default PubMed results, primary analysis was also repeated on search results restricted to the first 40 records. The ability of the filters to retrieve highly-relevant or highly-cited articles was also tested, with an article being considered highly-relevant if referenced by UpToDate and highly-cited if its citation count was greater than the median citation count of all relevant articles for that question – there was an average of eight highly-cited articles per question. To reduce the risk of type I error, the conservative method of Bonferroni was applied so that tests with a p less than 0.003 were interpreted as statistically significant. Main Results – Response rate 75%. Physician-provided search terms retrieved 46% of relevant articles and a ratio of relevant to non-relevant articles of 1:16 (p less than 0.003). Applying the narrow forms of both the nephrology and clinical queries filters together produced the greatest overall improvement, with efficiency improving by 16% and comprehensiveness remaining unchanged. Applying a narrow form of the clinical queries filter increased efficiency by 17% (p less than 0.003) but decreased comprehensiveness by 8% (p less than 0.003). No combination of search filters produced improvements in both comprehensiveness and efficiency. When results were restricted to the first 40 citations, the use of the narrow form of the clinical queries filter alone improved overall search performance – comprehensiveness improved from 13% to 26 % and efficiency from 5.5% to 23%. For highly-cited or highly-relevant articles the combined use of the narrow forms of both filters produced the greatest overall improvement in efficiency but no significant change in comprehensiveness. Conclusion – The use of PubMed search filters improves the efficiency of physician searches and saves time and frustration. Applying clinical filters for quick clinical searches can significantly improve the efficiency of physician searching. Improved search performance has the potential to enhance the transfer of research into practice and improve patient care.
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 | 0.298 | 0.587 |
| Meta-epidemiology (narrow) | 0.002 | 0.002 |
| Meta-epidemiology (broad) | 0.007 | 0.006 |
| Bibliometrics | 0.019 | 0.019 |
| Science and technology studies | 0.002 | 0.001 |
| Scholarly communication | 0.007 | 0.011 |
| Open science | 0.004 | 0.007 |
| Research integrity | 0.004 | 0.002 |
| Insufficient payload (model declined to judge) | 0.019 | 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".