Referring surgical patients with sodium below 130 did not change 30-day outcomes

Design

Retrospective registry study, regression discontinuity design

Population

24,803 adults with preoperative sodium 125–135 mmol/L at one Singapore institution

Primary outcome

30-day mortality

Effect

No threshold effect for 30-day mortality (p = 0.138); possible effect on 1-year mortality (p = 0.010)

A Singapore registry study looked at 24,803 adults with preoperative sodium between 125 and 135 mmol/L, out of 180,926 surgical patients. The hospital required physician referral and treatment for anyone below 130 mmol/L. A regression discontinuity design compared patients just below and just above that line, which approximates a natural experiment.

Patients below 130 did worse overall: 30-day mortality 3.32% vs 1.33% and one-year mortality 21.25% vs 11.03%. But at the threshold itself, the protocol made no detectable difference to 30-day mortality, ICU admission or length of stay. There was a signal of lower one-year mortality, strongest in patients with heart or kidney disease. Lower starting sodium also predicted rapid correction after surgery.

The authors suggest the long-term effect may come from physicians finding and treating the underlying disease, not from fixing the sodium number. Mild hyponatraemia before surgery is a marker worth investigating, not a level to chase in the days before theatre.

• Look for the cause of preoperative hyponatraemia: drugs, heart failure, cirrhosis, kidney disease, hypothyroidism
• Review thiazides, SSRIs and other sodium-lowering drugs before elective surgery
• Do not expect correcting sodium alone to reduce short-term surgical risk
• Watch sodium closely after surgery, since lower starting values predicted rapid correction

Why it matters

It challenges the reflex to delay surgery or push sodium up without asking why it is low.

Don't overread it

This is observational, and the one-year benefit was a secondary outcome that may reflect treatment of underlying disease.

The statistics, in plain English

Regression discontinuity compares people either side of an arbitrary cut-off, who should be very similar except for the protocol. No jump in 30-day mortality at 130 mmol/L means the referral did not detectably change short-term risk. The subgroup findings are weaker, because several were tested.

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.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.