Table 1 Agreement on important characteristics, at re-extraction of a 10% random sample of included studies
Characteristic | Unweighted Cohen’s Kappa (95% CI) |
|---|---|
Funding source | 0.75 (0.5–1) |
Study Design | 0.65 (0.02–1) |
Type of analysis (confirmatory/hypothesis-based versus exploratory) | 0.14 (0–0.5) |
Mediation analysis is primary aim of study | 0.52 (0.12–0.92) |
Causal diagram included | 1 (1–1) |
Sample size | 0.75 (0.55–0.96) |
Power/sample size calculation included | 0 (unable to estimate, too infrequent) |
Method of mediation analysis | 0.84 (0.83–1) |
Type of time-to-event model | 0.91 (0.75–1) |
Competing risks considered | 0 (unable to estimate, too infrequent) |
If clustering of data, was this addressed in the analysis? | 0.61 (0.15–1.0) |
Outcome frequency > =10% | 0.79 (0.54–1) |
Rare outcome limitation for Cox model mentioned | Unable to estimate, all false (100% agreement) |
Temporal separation clearly defined | 0.76 (0.47–1) |
No unmeasured confounding of exposure/outcome | 0.82 (0.49–1) |
No unmeasured confounding of mediator/outcome | 0.85 (0.57–1) |
No unmeasured confounding of exposure/mediator | 0.6 (0.13–1) |
No exposure-dependent confounding of mediator-outcome | 0.64 (0–1) |
Accurate measurement of mediator | 0.65 (0.32–0.99) |
Interaction between exposure and mediator considered/tested | 0.6 (0.19–1) |
Was a method used to address confounding of exposure or mediator? | N/A (100% used regression for both exposure and mediator, 100% agreement) |
Sensitivity analysis relating to mediation analysis | 0.44 (0.05–0.75) |
Measures reported-indirect effect | 0.76 (0.46–1) |
Measures reported- proportion mediated | 0.86 (0.61–1) |
Precision estimate for indirect effect | 1.0 (N/A) |
Precision estimate for proportion mediated | 0.77 (0.34–1) |