SUCRA in Network Meta-Analysis: What It Means and How to Report It
On this page
If your review compares more than two interventions using network meta-analysis, you've likely encountered SUCRA -- the Surface Under the Cumulative Ranking curve. It's one of the most frequently reported, and most frequently misinterpreted, outputs in network meta-analysis.
Example forest plot — each line is one study, the diamond is the pooled estimate.
What SUCRA actually measures
SUCRA converts the probabilistic ranking of each treatment in your network into a single number between 0 and 100%. A treatment with a SUCRA of 100% would be ranked best with certainty across every simulation; a SUCRA of 0% would be ranked worst with certainty. In practice, most treatments land somewhere in between, reflecting genuine uncertainty about their relative ranking.
Under the hood, SUCRA is calculated from the cumulative probability that a treatment ranks first, first-or-second, first-through-third, and so on, across all treatments in your network -- typically derived from a Bayesian model with thousands of simulated iterations, though frequentist equivalents exist too.
What SUCRA does not tell you
This is where most misinterpretation happens.
**SUCRA is not the same as effect size.** A treatment can have a high SUCRA while still having a clinically trivial effect, if every other treatment in the network performs even more poorly. SUCRA ranks treatments relative to each other -- it says nothing on its own about whether the best-ranked treatment is actually clinically meaningful.
**SUCRA does not account for evidence quality.** A treatment estimated from two small, high-risk-of-bias trials can produce a high SUCRA purely from favorable point estimates, while a treatment backed by several large, well-conducted trials ranks lower. Reporting SUCRA alongside a GRADE assessment for each comparison is what keeps this from misleading readers.
**A high SUCRA does not mean statistical significance.** Two treatments with overlapping credible intervals can still show meaningfully different SUCRA scores, simply because SUCRA reflects the full ranking distribution, not a pairwise significance test.
How to report SUCRA correctly
- **Pair every SUCRA ranking with the actual effect estimates and their credible intervals.** A ranking table alone, without the underlying numbers, invites readers to over-interpret small differences between adjacent SUCRA scores. - **Report your network geometry.** How many trials connect each pair of treatments matters -- a treatment connected to the rest of the network through only one small trial produces a less trustworthy SUCRA than one supported by multiple well-connected comparisons. - **Check transitivity and consistency before you report SUCRA at all.** If your network shows meaningful inconsistency between direct and indirect evidence, ranking the treatments with SUCRA before resolving that inconsistency can produce a confidently wrong ranking. - **Use a rankogram or a cumulative ranking plot alongside the SUCRA table**, since a single percentage value flattens out the real uncertainty in the underlying ranking distribution.
Why this matters for peer review
Reviewers evaluating a network meta-analysis increasingly know to check whether SUCRA rankings are reported in isolation or contextualized with effect sizes, evidence quality, and network geometry. A submission that leads with "Treatment A had the highest SUCRA" without that context is one of the more common reasons network meta-analyses get sent back for revision.
If you're running a network meta-analysis and want a second opinion on your SUCRA interpretation, your network geometry, or how to present rankings alongside GRADE ratings, that falls squarely within our meta-analysis consulting.
**[Get a quote](/systematic-reviews#meta-analysis-support)** for network meta-analysis consulting.
Related articles