TimelyScholarFree Tools › GRADE Evidence Rating Tool
Rate the certainty of evidence for each outcome with GRADE and build a Summary of Findings table with absolute effects per 1000, ready to paste into Word.
FreeNo account neededYour data stays in your browser
| Outcome | Anticipated absolute effects (95% CI) | Relative effect (95% CI) | Participants (studies) | Certainty of the evidence (GRADE) | Comments | |
|---|---|---|---|---|---|---|
| Risk with comparator | Risk with intervention | |||||
| All-cause mortality | 120 per 1000 | 98 per 1000 (79 to 122) 22 fewer per 1000 (from 41 fewer to 2 more) | RR 0.82 (0.66 to 1.02) | 2450 (8 RCTs) | ⊕⊕⊕◯ Moderate | Downgraded one level for serious imprecision. Confidence interval includes both benefit and no effect. |
| Hospital readmission | 250 per 1000 | 189 per 1000 (155 to 229) 61 fewer per 1000 (from 95 fewer to 21 fewer) | OR 0.70 (0.55 to 0.89) | 1890 (6 RCTs) | ⊕⊕◯◯ Low | Downgraded one level for serious risk of bias; downgraded one level for serious inconsistency. Unblinded outcome assessment in most trials; I² = 62%. |
| Quality of life (0 to 100) | MD 4.20 higher (1.10 higher to 7.30 higher) | – | 940 (4 RCTs) | ⊕⊕◯◯ Low | Downgraded one level for serious risk of bias; downgraded one level for serious indirectness. Higher is better. | |
Certainty: high ⊕⊕⊕⊕, moderate ⊕⊕⊕◯, low ⊕⊕◯◯, very low ⊕◯◯◯. Absolute effects use the comparator risk you enter; for risk ratios, risk with intervention = comparator risk × RR.
Study design, number of participants and studies, and the pooled effect.
Plus any reasons to rate up.
Paste into Word or download a CSV.
GRADE rates the certainty of a body of evidence for each outcome as high, moderate, low or very low. Evidence from randomized trials starts as high certainty and evidence from non-randomised studies as low.
Certainty is rated down for risk of bias, inconsistency, indirectness, imprecision and publication bias, usually by one level for each serious concern and by two for a very serious one. It cannot fall below very low.
Non-randomised evidence can be rated up for a large effect, a dose-response gradient, or when plausible residual confounding would have reduced the observed effect.
When non-randomised studies are assessed with ROBINS-I, the evidence starts as high certainty but is usually rated down for risk of bias, often by two levels.
For each important outcome: the comparator and intervention risks, the relative effect, the number of participants and studies, the GRADE certainty and explanations for any downgrading.
Use a typical risk for the population the table is about. If baseline risk varies little between studies, the Cochrane Handbook suggests the median comparator-group risk across the included studies; add rows for low- and high-risk groups when risks differ.
No. GRADEpro GDT is the GRADE Working Group's official software. This free tool helps you draft ratings and a table quickly.