Dynamic Variability of Sodium and Potassium Predicts Mortality in Critically Ill Patients With Ischemic Stroke: Evidence From Two Critical Care Cohorts.
Zhang J, Zhang W, et al. • Brain and behavior • 2026
Fluctuations in sodium and potassium levels were strongly linked to increased short-term and long-term mortality risk in ICU patients with ischemic stroke, with patients having high variability in both electrolytes showing the worst prognosis.
Key Findings
Results
Continuous sodium variability (CV) was significantly associated with increased mortality at 30-day, 90-day, and 360-day intervals in MIMIC-IV cohort.
Hazard ratios for continuous sodium variability were 1.22 (30-day), 1.18 (90-day), and 1.17 (360-day)
Patients in the highest quartile (Q4) of sodium variability showed significantly increased mortality risk with HRs of 1.78, 1.51, and 1.49 at 30-, 90-, and 360-day intervals respectively
Analysis was conducted using multivariable Cox proportional-hazards models
2593 patients from MIMIC-IV were included in this analysis
Results
In the NWICU validation cohort, Q4 sodium variability was an independent risk factor for 30-day mortality.
HR = 4.04 for Q4 sodium variability in NWICU cohort (p < 0.001)
673 patients from NWICU were included in this validation cohort
This finding validated the MIMIC-IV findings in an independent dataset
The substantially higher HR in NWICU (4.04) compared to MIMIC-IV (1.78) suggests a consistent but potentially stronger effect in this population
Results
Potassium variability was also significantly associated with mortality at 30-day, 90-day, and 360-day intervals.
Potassium variability (CV) was analyzed alongside sodium variability across both cohorts
Patients were stratified into quartiles based on CV values for potassium variability
Kaplan-Meier cumulative incidence estimates were used to visualize survival differences across quartiles
Restricted cubic spline (RCS) analyses were conducted to examine associations of potassium CV with primary outcomes
Results
Patients with high variability in both sodium and potassium had the worst prognosis among all groups.
Combined high variability in both electrolytes was associated with the worst outcomes
The study included a total of 3266 IS patients across both cohorts (2593 from MIMIC-IV and 673 from NWICU)
This finding was derived from a retrospective cohort design using the MIMIC-IV database and NWICU
Multivariable Cox models were used to control for confounders when assessing this association
Results
Sensitivity analyses confirmed that both abnormally high and low sodium and potassium levels were independently associated with increased mortality risks.
Sensitivity analyses examined the impact of both electrolyte levels and variability on mortality
Both hypernatremia/hyperkalemia and hyponatremia/hypokalemia were associated with increased mortality
These analyses were conducted to separate the effects of electrolyte level from electrolyte variability
Results confirmed that variability itself, beyond just absolute electrolyte levels, contributed to mortality risk
Methods
The study used coefficient of variation (CV) as the primary measure of electrolyte variability in critically ill ischemic stroke patients.
CV was calculated for both sodium and potassium measurements during ICU stays
Patients were stratified into quartiles (Q1–Q4) based on CV values
The study focused on 30-, 90-, and 360-day all-cause mortality as primary outcomes
This was a retrospective cohort study utilizing two databases: MIMIC-IV and NWICU
What This Means
This research suggests that how much sodium and potassium levels fluctuate during an ICU stay — not just whether they are too high or too low — is strongly linked to the risk of dying in patients who have had an ischemic stroke (a stroke caused by a blocked blood vessel). The researchers analyzed data from 3,266 critically ill stroke patients across two separate hospital databases and found that patients whose sodium or potassium levels varied the most over time had significantly higher rates of death at 30 days, 90 days, and one year after admission. Patients with the highest level of variability in both electrolytes at the same time had the worst survival outcomes overall.
The association was consistent across both datasets used in the study. In the main dataset (MIMIC-IV), patients in the top quartile of sodium variability had a 78% higher risk of dying within 30 days compared to those with the least variability. In the validation dataset (NWICU), the top quartile of sodium variability was associated with more than four times the risk of 30-day death. Sensitivity analyses also showed that having either too much or too little sodium and potassium — in addition to frequent fluctuations — was harmful, suggesting that both instability and extreme values independently contribute to poor outcomes.
This research suggests that monitoring and managing electrolyte stability — not just keeping levels within a normal range at any single point in time — may be an important consideration in the care of stroke patients in the ICU. The findings highlight that electrolyte variability could serve as a useful indicator for identifying patients at particularly high risk of death, potentially allowing for more targeted clinical monitoring and intervention strategies.
Zhang J, Zhang W, Zeng B, Xie L, Zhu J. (2026). Dynamic Variability of Sodium and Potassium Predicts Mortality in Critically Ill Patients With Ischemic Stroke: Evidence From Two Critical Care Cohorts.. Brain and behavior. https://doi.org/10.1002/brb3.71626