- Design
- Post-hoc win-ratio analysis of the SODa-BIC randomised trial
- Population
- 488 critically ill patients with metabolic acidosis
- Primary outcome
- Hierarchical composite: death, renal replacement therapy, persistent renal dysfunction, hospital stay
- Effect
- Win ratio 0.97 (95% CI 0.79 to 1.19); DOOR probability 0.49 (0.45 to 0.54)
SODa-BIC tested sodium bicarbonate for metabolic acidosis in ICU patients, with major adverse kidney events at 30 days as the primary outcome; it found no significant difference, though the components pointed in different directions. This post-hoc analysis re-examined the data with a hierarchical win-ratio outcome ordered as death, renal replacement therapy, persistent renal dysfunction and hospital stay.
In 488 patients forming 59,250 pairs, the win ratio was 0.97 (95% CI 0.79 to 1.19) and the Desirability of Outcome Ranking probability was 0.49 (0.45 to 0.54). Most pairs were decided by mortality and length of stay. There was evidence of effect modification across subgroups defined by acute kidney injury and severity of acidaemia.
A post-hoc subgroup signal should be treated as hypothesis generating. The analysis supports the original neutral result and shows that the choice of outcome did not change it.
- Different outcome hierarchies gave the same neutral answer for bicarbonate in ICU acidosis.
- Subgroup differences by AKI and acidaemia severity were found but are exploratory.
- Bicarbonate decisions should rest on the clinical picture and local protocol.
- Do not infer benefit or harm in any subgroup from this analysis.
Why it matters
It tests whether a neutral primary result hid a benefit in how outcomes were combined.
Don't overread it
Post-hoc analysis of one trial, with subgroup effect modification that was not prespecified as a treatment recommendation.
The statistics, in plain English
A win ratio of 0.97 means that in pairwise comparisons, bicarbonate patients did slightly worse than controls about as often as better, with an interval from 0.79 to 1.19 that includes no difference. A DOOR probability of 0.49 expresses the same thing as the chance that a random bicarbonate patient has the better outcome.
Read the rest in the app
You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.
Scan to keep reading on your phone. No account needed to start.
Tomorrow morning, before your first patient
One edition a day for nephrology, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free