- Design
- Retrospective registry, regression-discontinuity design
- Population
- 24,803 adults with preoperative sodium 125–135 mmol/L, Singapore
- Primary outcome
- 30-day mortality
- Effect
- No threshold effect at 130 mmol/L (p = 0.138); discontinuity for 1-year mortality (p = 0.010)
A Singapore institution required internist referral and treatment for any patient with preoperative sodium below 130 mmol/L. Using registry data from 2013 to 2020, investigators compared 24,803 patients with sodium between 125 and 135 mmol/L in a regression-discontinuity design, which tests whether outcomes jump at the 130 threshold where the protocol starts.
Patients below 130 did worse in raw terms: 30-day mortality 3.3% against 1.3%, one-year mortality 21% against 11%, and longer stays. But there was no discontinuity at the threshold for 30-day mortality, ICU admission or length of stay — the protocol did not move short-term outcomes. One-year mortality did show a discontinuity, mostly in patients with cardiac disease, CKD or emergency surgery.
The likely reading is that low sodium marks sick patients, and correcting the number before theatre does little by itself in the short term. The one-year signal may reflect better management of the underlying heart or kidney disease that the referral uncovered.
- Treat preoperative sodium below 130 mmol/L as a marker of underlying illness and look for the cause.
- Do not delay non-urgent surgery solely to normalise the sodium number.
- Use the referral to review heart failure, kidney disease, diuretics and other causative drugs.
- Watch for rapid postoperative sodium rise, which was more likely with lower starting values.
Why it matters
It questions protocols that delay or refer purely on a sodium threshold rather than on the patient's underlying illness.
Don't overread it
The one-year mortality benefit comes from a quasi-experimental design and subgroups, and may reflect care of comorbidities rather than sodium correction.
The statistics, in plain English
Regression discontinuity compares patients just above and just below a cut-off, who should be similar apart from the protocol. No jump at 130 for 30-day mortality means the intervention did not detectably change it. Subgroup effects are less reliable than the main result.
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 internal medicine, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free