SMART: Balanced Crystalloid vs Saline in the ICU

Balanced crystalloid beat saline on a composite of death, AKI, RRT.

Population

15,802 ICU adults at a single academic center across 5 ICUs (medical, surgical, trauma, neuro, cardiac).

Intervention

Balanced crystalloid (lactated Ringer's or Plasma-Lyte A).

Comparison

0.9% saline (NS).

Outcome

MAKE30 (Major Adverse Kidney Events at 30 days): composite of death, new RRT, persistent renal dysfunction (final Cr ≥ 200% of baseline).

Cluster-randomized, multiple-crossover trial. Each ICU randomized monthly to receive saline or balanced crystalloid as the default; physicians could deviate if clinically indicated. The trial design reduced contamination + selection bias vs patient-level randomization for a fluid trial.

• MAKE30: 14.3% balanced vs 15.4% saline (OR 0.91, 95% CI 0.84-0.99, p=0.04). Modest absolute reduction (1.1%) but clinically meaningful at scale.
• In-hospital mortality: 10.3% balanced vs 11.1% saline (p=0.06).
• New renal-replacement therapy: 2.5% vs 2.9% (p=0.08).
• Effect amplified in patients with sepsis (subgroup). NS-induced hyperchloremic acidosis is a plausible mechanism.

Default to balanced crystalloid (LR or Plasma-Lyte) for ICU + perioperative resuscitation. Reserve saline for hypochloremic alkalosis (vomiting, NG suction) or severe hyperkalemia where the small K in LR (4 mEq/L) is a concern. The downstream BaSICS trial (JAMA 2021) was statistically neutral but trended in the same direction; confidence in 'balanced > saline' has grown rather than weakened.

• Single-center, may not generalize to all ICU populations.
• Open-label cluster design; clinicians knew which fluid was 'default.'
• Modest absolute risk reduction (NNT ~94).
• Did not study OR-only fluid choice (though SMART-MED + SALT-ED extended findings to ED + ward).

1. Does your institution still default to NS for trauma + sepsis resuscitation? What would it take to change?
2. When would you deliberately choose NS over LR perioperatively?
3. How does the Plasma-Lyte K of 5 mEq/L change your decision in CKD/dialysis patients vs LR (K 4) vs NS (K 0)?