GRADE criteria | Examples | Relevance to included studies | Outcome |
---|---|---|---|
Imprecision | Insufficient statistical power Wide confidence intervals Small no. of studies | Wide confidence intervals Small sample sizes No sample size calculations | Low |
Inconsistency | Results not similar across studies Assessed using I2 | I2 indicated considerable heterogeneity Heterogeneity unexplained Some contrasting findings | Low |
Risk of bias | Limitations in study design or conduct Quality of evidence | Use of methods with controls Poor reporting of randomisation or attrition Lack of information Mixed quality across studies | Low |
Indirectness | Evidence differs from the study’s intended population or outcomes | Populations of rats and mice only Outcomes reported using desired outcome measures Majority of studies addressed PICO question | Moderate |
Publication bias | Lack of studies Selective publication | Most studies with small sample sizes and mixed outcomes Conflicts of interest well reported | Moderate |