Design: Post-hoc secondary analysis of a randomised trial (SODa-BIC) with composite outcomes
Population: 488 critically ill adults with metabolic acidosis
Primary outcome: Win-ratio composite of death, renal replacement, renal dysfunction, length of stay
Effect: Win ratio 0.97 (95% CI 0.79–1.19); desirability-of-outcome probability 0.49

The SODa-BIC trial tested sodium bicarbonate for metabolic acidosis in critically ill patients and found no effect on major adverse kidney events at 30 days. This secondary analysis re-examined the data using win-ratio and desirability-of-outcome-ranking composites that weight death, dialysis, persistent renal dysfunction and length of stay.

The neutral result held: the win ratio was 0.97 (95% CI 0.79–1.19) and the desirability-of-outcome probability 0.49, essentially a coin toss, with most comparisons decided by mortality and length of stay. There was exploratory evidence that effect varied by acute kidney injury status and severity of acidaemia.

The takeaway is that reframing the endpoints did not rescue a benefit — routine sodium bicarbonate for ICU metabolic acidosis remains unsupported. The subgroup signals are hypothesis-generating only; they suggest where a future trial might look (severe acidaemia, established AKI), not a reason to give bicarbonate selectively now.

Re-analysing SODa-BIC with composite outcomes confirmed no overall benefit of bicarbonate.
The win ratio was 0.97 and the desirability-of-outcome probability near 0.5.
Effect appeared to vary by AKI status and acidaemia severity — exploratory only.
Routine bicarbonate for ICU metabolic acidosis remains unsupported.

Why it matters

It shows a neutral trial stays neutral under more favourable statistics, closing off a tempting reinterpretation.

The statistics, in plain English

A win ratio of 0.97 with an interval spanning 1.0, and a 49% desirability probability, both mean no advantage either way; the subgroup differences are exploratory and could easily be chance.