Objective To explore the relation between the minimum serum potassium ion concentration at admission and short⁃term prognosis of ICU inpatients. Methods A total of 59 843 adult patients firstly admitted to ICU were screened as the research subjects from the MIMIC⁃Ⅳ database. Information such as general demography, vital signs, disease type, complications, laboratory examination indices, and length of hospital stay was collected. Patients were assigned to low potassium group (n=10 675), normal group (n=48 758), or high potassium group (n=410) according to the minimum serum potassium ion concentration at admission within 24 hours in ICU. Methods such as univariate analysis, multivariate COX regression model, restricted cubic spline model, threshold analysis, Kaplan⁃Meier survival curve, and subgroup analysis were adopted to analyze the effect of the minimum serum potassium ion concentration at admission on the 30⁃day mortality of ICU inpatients. Results Among the 59 843 ICU inpatients enrolled, 6831 died within 30 days after admission, therein 1488 dead cases were in the low potassium group, 5225 in the normal group, and 118 in the high potassium group, accounting for 21.78%, 76.49%, and 1.73%, respectively. The results of restricted cubic spline model analysis revealed that when covariates were not corrected, the minimum serum potassium ion concentration at admission was in a U⁃shaped relation with the 30⁃day mortality of ICU inpatients, with the minimum risk value being 4.07 mmol/L (P<0.001). After adjusting for all covariates, the minimum serum potassium ion concentration at admission was in a J⁃shaped relation with the 30⁃day mortality of ICU inpatients (nonlinear P<0.001). Threshold analysis results indicated that the minimum serum potassium ion concentration at admission <4.07 mmol/L was a protective factor for the 30⁃day mortality of ICU inpatients (HR=0.775, P<0.001); the minimum serum potassium ion concentration at admission ≥4.07 mmol/L was a risk factor for the 30⁃day mortality of ICU inpatients (HR=1.218, P<0.05). The results of the multivariate COX regression analysis interpreted that before adjusting for all covariates, the 30⁃day mortality of ICU inpatients in the high potassium group and the low potassium group was higher than that in the normal group (P<0.05); furthermore, after adjusting for all covariates, the 30⁃day mortality of ICU inpatients in the low potassium group was higher than that in the normal group (P<0.05), but there was no statistically significant difference in the 30⁃day mortality between the high potassium group and the normal group of ICU inpatients (P>0.05). The results of Kaplan⁃Meier survival curve described that the 30⁃day survival rate of ICU inpatients in the normal group was higher than that in the high and low potassium groups, and the 30⁃day survival rate of ICU inpatients in the low potassium group was higher than that in the high potassium group (P<0.05). Subgroup analysis results exhibited that an interaction was detected in the subgroup of patients with chronic obstructive pulmonary disease and malignant tumors (P<0.05). Conclusion The minimum serum potassium ion concentration at admission is the influencing factor for the 30⁃day mortality of ICU inpatients, maintaining the minimum serum potassium ion concentration at 4.07 mmol/L at admission may be the most beneficial for improving prognosis of ICU inpatients. The 30⁃day mortality risk of ICU inpatients with high hyperkalemia is higher than that with hypokalemia.

无