Validity and Inter-Rater Reliability Testing of Quality Assessment Instruments [Internet]
Bibliographic record
Abstract
Background Numerous tools exist to assess methodological quality, or risk of bias in systematic reviews; however, few have undergone extensive reliability or validity testing. Objectives (1) assess the reliability of the Cochrane Risk of Bias (ROB) tool for randomized controlled trials (RCTs) and the Newcastle-Ottawa Scale (NOS) for cohort studies between individual raters, and between consensus agreements of individual raters for the ROB tool; (2) assess the validity of the Cochrane ROB tool and NOS by examining the association between study quality and treatment effect size (ES); (3) examine the impact of study-level factors on reliability and validity. Methods Two reviewers independently assessed risk of bias for 154 RCTs. For a subset of 30 RCTs, two reviewers from each of four Evidence-based Practice Centers assessed risk of bias and reached consensus. Inter-rater agreement was assessed using kappa statistics. We assessed the association between ES and risk of bias using meta-regression. We examined the impact of study-level factors on the association between risk of bias and ES using subgroup analyses. Two reviewers independently applied the NOS to 131 cohort studies from 8 meta-analyses. Inter-rater agreement was calculated using kappa statistics. Within each meta-analysis, we generated a ratio of pooled estimates for each quality domain. The ratios were combined to give an overall estimate of differences in effect estimates with inverse-variance weighting and a random effects model. Results Inter-rater reliability between two reviewers was considered fair for most domains (κ ranging from 0.24 to 0.37), except for sequence generation (κ=0.79, substantial). Inter-rater reliability of consensus assessments across four reviewer pairs was moderate for sequence generation (κ=0.60), fair for allocation concealment and “other sources of bias” (κ=0.37, 0.27), and slight for the remaining domains (κ ranging from 0.05 to 0.09). Inter-rater variability was influenced by study-level factors including nature of outcome, nature of intervention, study design, trial hypothesis, and funding source. Inter-rater variability resulted more often from different interpretation of the tool rather than different information identified in the study reports. No statistically significant differences were found in ES when comparing studies categorized as high, unclear or low risk of bias. Inter-rater reliability of the NOS varied from substantial for length of followup to poor for selection of non-exposed cohort and demonstration that the outcome was not present at outset of study. We found no association between individual NOS items or overall NOS score and effect estimates. Conclusion More specific guidance is needed to apply risk of bias/quality tools. Study-level factors that were shown to influence agreement provide direction for detailed guidance. Low agreement across pairs of reviewers has implications for incorporation of risk of bias into results and grading the strength of evidence. Variable agreement for the NOS, and lack of evidence that it discriminates studies that may provide biased results, underscores the need for more detailed guidance to apply the tool in systematic reviews.
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 | 0.586 | 0.784 |
| Meta-epidemiology (narrow) | 0.002 | 0.002 |
| Meta-epidemiology (broad) | 0.005 | 0.009 |
| Bibliometrics | 0.016 | 0.015 |
| Science and technology studies | 0.003 | 0.005 |
| Scholarly communication | 0.006 | 0.005 |
| Open science | 0.004 | 0.006 |
| Research integrity | 0.002 | 0.003 |
| Insufficient payload (model declined to judge) | 0.005 | 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; 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".