Abstract Rationale Potassium disturbances are common in critically ill patients and may influence outcomes such as mortality and cardiac arrhythmias. While both hypokalemia and hyperkalemia have been associated with adverse events, the impact of potassium levels and their variability on ICU outcomes in septic patients remains unclear. This study aimed to evaluate the association between ICU admission potassium and early potassium variability with ICU mortality and new-onset atrial fibrillation (NOAF) in septic ICU patients. Methods We conducted a retrospective cohort study using the Medical Information Mart for Intensive Care IV (MIMIC-IV) database. Adult patients meeting the Angus sepsis criteria during their first ICU admission were included. Serum potassium data were extracted from laboratory measurements labeled “Potassium” and “Potassium, Whole Blood,” excluding non-serum specimens. Potassium-related variables included ICU admission potassium, mean potassium, standard deviation (SD), coefficient of variation (CV), and mean absolute percentage error (MAPE) during the first 72 hours of ICU stay. NOAF was identified through nurse documentation in the electronic health record, a method previously validated as a reliable proxy for ECG-confirmed atrial fibrillation. Generalized additive models (GAM) were used to assess nonlinear and linear associations between potassium parameters, ICU mortality, and NOAF. Results Of 546,028 ICU admissions in MIMIC-IV, 58,871 met the Angus sepsis criteria. After applying inclusion and exclusion criteria, 22,800 septic ICU patients were analyzed; 54% were male. ICU admission potassium showed a significant nonlinear association with ICU mortality (p 0.001 for smooth term), with the lowest mortality observed at approximately 3.5 mmol/L. Both hypokalemia and hyperkalemia were associated with increased mortality risk. ICU admission potassium was also significantly associated with NOAF (p 0.001), showing a linear relationship in the entire cohort and a nonlinear pattern among patients without preexisting AF. None of the potassium variability metrics (SD, CV, MAPE, or mean potassium during the first 72 hours) demonstrated significant associations with ICU mortality or NOAF. Conclusions In septic ICU patients, ICU admission potassium exhibits a significant nonlinear association with ICU mortality and a variable relationship with new-onset atrial fibrillation. Potassium variability during the early ICU course was not predictive of adverse outcomes. These findings highlight the importance of maintaining optimal potassium levels at ICU admission while suggesting that short-term potassium fluctuations may be less clinically relevant. This abstract is funded by: None
Mai et al. (Fri,) studied this question.