Authors: Holger J Schünemann, Andrew D Oxman, Gunn E Vist, Julian PT Higgins, Jonathan J Deeks, Paul Glasziou and Gordon H Guyatt on behalf of the Cochrane Applicability and Recommendations Methods Group.
Key points
The GRADE approach, adopted by The Cochrane Collaboration, specifies four levels of quality (high, moderate, low and very low) where the highest quality rating is for a body of evidence based on randomized trials. Review authors can downgrade randomized trial evidence depending on the presence of five factors and upgrade the quality of evidence of observational studies depending on three factors.
Quality ratings are made separately for each outcome.
Methods for computing, presenting and interpreting relative and absolute effects for dichotomous outcome data, including the number needed to treat (NNT), are described in this chapter.
For continuous outcome measures, review authors can present pooled results for studies using the same units, the standardized mean difference and effect sizes when studies use the same construct but different scales, and odds ratios after transformation of the standardized mean differences.
Review authors should not describe results as ‘not statistically significant’ or ‘non-significant’, but report the confidence interval together with the exact P value.
Review authors should not make recommendations, but they can – after describing the quality of evidence and the balance of benefits and harms – highlight different actions that might be consistent with particular patterns of values and preferences.
12.2 Assessing the quality of a body of evidence
12.4 Interpreting results of statistical analyses
12.5 Interpreting results from dichotomous outcomes (including numbers needed to treat)
12.6 Interpreting results from continuous outcomes (including standardized mean differences)
Box 12.8.a: The Cochrane Applicability and Recommendations Methods