Routine Sodium Checks Before Surgery May Not Save Lives in the Short Term, Landmark Study Finds

Hyponatremia—abnormally low sodium in the blood—is the most common electrolyte disorder seen in patients heading into the operating room, with reported prevalence rates ranging from 7.8 percent to 23.8 percent in large surgical cohorts. For years, clinicians have known that patients who arrive for surgery with diluted blood fare worse: they die more often, land in the intensive care unit more frequently, and stay in the hospital longer. What has remained stubbornly unclear is whether the low sodium itself is a cause of these complications or merely a chemical fingerprint of deeper disease. Now, a study published in the Journal of General Internal Medicine offers the most direct answer yet, and its conclusion is likely to unsettle a practice that thousands of hospitals treat as gospel.

Researchers at Singapore General Hospital, a quaternary academic medical center with more than 2,000 beds and over 10,000 operations each year, took advantage of something rare in medicine: a uniform, institution-wide protocol that has been applied to every surgical patient since 2013. Under that protocol, any patient whose preoperative serum sodium falls below 130 mmol/L must have surgery deferred and receive a mandatory referral to an internal medicine perioperative team for evaluation and treatment. Patients with sodium between 125 and 129 mmol/L undergo a structured assessment for culprit medications and conditions such as heart failure, cirrhosis, chronic kidney disease, and the syndrome of inappropriate antidiuretic hormone secretion, followed by internist review. Because the rule switches on at a precise numerical threshold, it created a natural experiment that the investigators could exploit statistically.

The team, led by Gerald Gui Ren Sng of the Department of Endocrinology together with anesthesiologist Hairil Rizal Abdullah and colleagues, drew on the SingHealth Perioperative and Anesthesia Subject Area Registry, a comprehensive data mart capturing preoperative, intraoperative, and postoperative information. From 180,926 adults operated on between 2013 and 2020, they identified 24,846 patients with preoperative sodium values between 125 and 135 mmol/L. After excluding 43 patients with incomplete data, the final cohort of 24,803 split into 1,807 patients (7.3 percent) below the 130 mmol/L threshold and 22,996 (92.7 percent) at or above it. Adherence to the protocol appeared high: only 1.27 percent of all cases that proceeded to surgery ultimately went ahead with a sodium value below the cutoff.

The raw numbers told a familiar and grim story. Compared with patients whose sodium sat between 130 and 135 mmol/L, those below 130 mmol/L had a 30-day mortality of 3.32 percent versus 1.33 percent, a one-year mortality of 21.25 percent versus 11.03 percent, an ICU admission rate of 18.65 percent versus 11.50 percent, and a mean hospital stay of 20.64 days versus 13.42 days—all differences statistically significant at p less than 0.001. The low-sodium group was also older, more often male, and carried a heavier burden of chronic kidney disease, cardiac disease, hypertension, and insulin-treated diabetes. On its face, the case for aggressive preoperative intervention seemed overwhelming.

But raw comparisons between groups this different can mislead, which is where the study’s methodological ambition comes in. The researchers employed a regression discontinuity design, a quasi-experimental technique borrowed largely from economics that can approximate causal inference from observational data. The logic is elegant: patients whose sodium is 129.9 mmol/L and those at 130.1 mmol/L are, for all practical purposes, nearly identical in every measurable respect, yet only the former triggers the protocolized referral. If the intervention works, outcomes should jump—or discontinue—abruptly at the 130 mmol/L line. The team fitted three variants of the model: a sharp design assuming a constant slope on either side of the threshold, a kink design allowing different slopes that approximate an exponential relationship, and a triangular kernel design that weights observations closer to the cutoff more heavily. A bandwidth of 5 mmol/L on either side of the threshold was chosen and validated using established optimal-bandwidth methods.

The verdict on short-term outcomes was unambiguous. For the primary outcome of 30-day mortality, the best-fitting sharp model showed no significant discontinuity at the threshold (treatment effect estimate 1.22, p = 0.138). ICU admissions (treatment effect 0.014, p = 0.564) and length of stay (treatment effect 0.184, p = 0.943) likewise showed no jump at 130 mmol/L, and none of the subgroup analyses—patients with ischemic heart disease or congestive cardiac failure, those with chronic kidney disease defined by an estimated glomerular filtration rate below 60 mL/min/1.73 m², those at moderate or high operative risk, and those undergoing emergency surgery—revealed a threshold effect. In plain terms, the elaborate machinery of referral, deferral, and sodium correction did not measurably change how patients fared in the first month after surgery.

Then came the surprise. One-year mortality did show a statistically significant discontinuity at the threshold, with a treatment effect estimate of 0.081 (p = 0.010) in the best-fitting triangular kernel model, along with a significant change in slope. The effect was concentrated in exactly the groups clinicians worry about most: patients with cardiac disease showed a treatment effect of 0.162 (p < 0.001), those with chronic kidney disease 0.071 (p = 0.012), those at moderate operative risk 0.779 (p = 0.013), and those undergoing emergency operations 0.100 (p = 0.001). The authors offer a provocative interpretation: rather than sodium normalization itself extending lives, the internist referral may have uncovered and set in motion long-term management of the underlying conditions—failing hearts, failing kidneys, occult malignancy—that caused the hyponatremia in the first place. In this reading, the low sodium value is less a poison to be neutralized than a smoke alarm pointing to a fire elsewhere.

The study also quantified a safety concern that has haunted perioperative medicine. Rapid shifts in serum sodium are dangerous in both directions: worsening hyponatremia risks cerebral edema, while overly swift correction of chronic hyponatremia risks osmotic demyelination syndrome, a devastating and often irreversible neurologic injury. In this cohort, lower preoperative sodium strongly predicted rapid postoperative correction, defined conservatively as a rise of more than 8 mmol/L within 48 hours of surgery. For every 1 mmol/L decrease in preoperative sodium below 130 mmol/L, the odds of rapid correction increased by 19 percent (odds ratio 1.190, p < 0.001). The authors note, however, that osmotic demyelination is reported in as few as 0.05 percent of patients even with rapid correction, making it unlikely to explain the one-year mortality pattern.

The investigators are candid about the limits of their work. The regression discontinuity design can evaluate only the specific cutoff chosen; it cannot reveal whether a different threshold—135 mmol/L, say, or 137 mmol/L, both used in other studies—might yield benefit, nor whether a different intervention, such as endocrinology referral, would perform better. The retrospective registry design precluded direct chart review of protocol adherence or of what treatments internists actually delivered, the etiology of each patient’s hyponatremia, the duration of any surgical delay, or the costs incurred. Hyponatremia itself is a notoriously heterogeneous condition—depletional states call for saline, the syndrome of inappropriate antidiuretic hormone secretion calls for fluid restriction, hypervolemic states call for diuretics, and multiple causes often coexist in a single patient—making a conventional randomized trial of specific treatments genuinely difficult to design. The small bandwidth of 11 data bins also limited the ability to adjust for comorbidities within the models, leaving the comorbidity-management hypothesis just that: a hypothesis.

Still, the implications ripple outward. For hospitals, the finding suggests that routine deferral of surgery to chase a sodium number may impose delays, added utilization, and cost without improving the outcomes that matter most in the immediate perioperative window—while possibly delivering unanticipated long-term benefit through the generalist attention it forces. For the research community, the study is a showcase for regression discontinuity design, a method the authors describe as underutilized in medicine and particularly suited to perioperative research, where patient heterogeneity and entrenched practice make randomized experiments impractical. As institutions worldwide review their own hyponatremia thresholds, this analysis of nearly a quarter-million surgical encounters provides both a cautionary tale about correlation versus causation and a template for answering similar questions hidden inside everyday clinical rules.

Subject of Research: Effect of protocolized internist referral for preoperative hyponatremia on postoperative surgical outcomes

Article Title: Assessing the Impact of Routine Intervention for Perioperative Hyponatremia on Postoperative Outcomes

Article References: Sng, G. G. R., Lim, D. Y. Z., Le, Q., Xie, Q., Bee, Y. M., & Abdullah, H. R. (2026). Assessing the Impact of Routine Intervention for Perioperative Hyponatremia on Postoperative Outcomes. Journal of General Internal Medicine. https://doi.org/10.1007/s11606-026-10892-8

Image Credits: AI Generated

DOI: 10.1007/s11606-026-10892-8

Keywords: hyponatremia, perioperative medicine, surgery, regression discontinuity design, serum sodium, 30-day mortality, ICU admission, chronic kidney disease, cardiac disease, osmotic demyelination syndrome, internist referral, outcomes research