GRADE assessment and Summary of Findings tables

The single idea that trips up most authors is that GRADE is applied per outcome, not per study. A review of the same trials can be high certainty for mortality and low certainty for quality of life, because each outcome carries its own risk of bias, its own consistency across studies, and its own precision. Rating the studies once and stamping a single grade across the whole review misreads the method and is the error reviewers catch first. We build the assessment around your prioritised outcomes so each certainty rating reflects the evidence that actually informs it.

The rating also starts from the design. Randomised trials begin at high certainty and observational studies begin at low, and from there the certainty can move down through five domains or up through three. This is where GRADE goes beyond a quality assessment of individual studies, and why it feeds on, but is not the same as, your risk of bias appraisal.

The fifth downgrading domain is publication bias. Where the smaller studies point one way and the picture from a funnel plot or a small-study-effects test suggests missing negative results, we document the concern and downgrade honestly rather than assuming the published record is complete. Each domain can cost one or two levels, so an outcome can fall from high to very low certainty when several concerns stack up.

Downgrading is only half of GRADE. For observational evidence that has not already been marked down, three criteria can raise the certainty rating. A large effect that is unlikely to be explained by bias, a clear dose-response gradient where more exposure tracks more effect, and plausible residual confounding that would only have worked against the observed effect all argue that the true effect is at least as strong as the estimate. We apply these criteria explicitly and record the reasoning, so an upgrade is defensible rather than optimistic.

1. 1

   Prioritise the outcomes

   We agree the critical and important outcomes to rate, ideally from your protocol, so the assessment stays focused on the outcomes that drive the decision.

2. 2

   Rate every domain per outcome

   For each outcome we assess risk of bias, inconsistency, indirectness, imprecision, and publication bias, then apply the upgrading criteria where they fit, with a footnote justifying every move.

3. 3

   Build the Summary of Findings table

   We assemble the table in GRADEpro GDT with participant and study counts, absolute and relative effects, the certainty symbols, and explanatory footnotes linked to each rating.

4. 4

   Draft the certainty narrative

   We write the certainty statements for your results and the plain-language summary for your abstract, phrased in the standard GRADE wording readers expect.

You receive a completed certainty rating for every prioritised outcome, a Summary of Findings table built in GRADEpro with linked footnotes, and the drafted narrative for your results and abstract. The assessment reads directly off your pooled estimates and heterogeneity statistics and off your risk of bias judgements, so the certainty ratings and the GRADE certainty of evidence table tell one consistent story with the rest of the review. Where an outcome cannot be pooled, we carry the certainty logic into the narrative synthesis instead of forcing a rating the data will not support.

• A certainty rating for each prioritised outcome, from high to very low
• A Summary of Findings table built in GRADEpro GDT ready for your manuscript
• An explanatory footnote justifying every downgrade and every upgrade
• Absolute and relative effects reported alongside the certainty for each outcome
• The certainty narrative drafted for your results section
• A plain-language evidence statement for your abstract in standard GRADE wording

We quote a fixed fee shaped by the number of outcomes you need rated, whether the evidence is randomised, non-randomised, or a mix that changes the starting certainty, and whether you want the Summary of Findings table and the certainty narrative drafted as well as the ratings. Send your outcomes, your pooled results, and your risk of bias appraisal, and we will return a fixed quote for a fully documented assessment.