An episode-based approach for assessing the impact of dyskalaemia on critical care outcomes – results from a territory-wide cohort study

Background: Dyskalaemias have been reported as an independent prognostic factor for adverse outcomes based on plasma potassium levels upon intensive care unit (ICU) admission or mean potassium levels across ICU stay, but the granular effects of discrete episodes of dyskalaemia have not been extensively studied. Methods: This retrospective observational cohort study included all adult patients admitted to ICUs in public hospitals in Hong Kong between January 2010 and June 2023. Discrete episodes of dyskalaemia were defined if plasma potassium measurements were beyond the reference range of 3.5–5 mmol/l. Patients were classified into four groups based on the potassium levels throughout their ICU stay (normokalaemic, hyperkalaemic, hypokalaemic and mixed dyskalaemic). Patients with prolonged ICU length of stay beyond 14 days were excluded. The main study outcomes were ICU mortality and the incidence of arrhythmias, which was defined based on a pharmacological surrogate of requiring amiodarone between the onset of the dyskalaemic episode and ICU discharge. Results: A total of 167,449 patients were included. A total of 60,953 (36.4%) patients remained normokalaemic. There were 21,820 (13.0%), 79,312 (47.4%) and 5364 (3.2%) patients in the hyperkalaemic, hypokalaemic and mixed dyskalaemic groups respectively. Dyskalaemia was significantly associated with increased ICU mortality (hyperkalaemia: adjusted odds ratio (aOR) 1.95, 95% confidence interval (CI) 1.79 to 2.12, P < 0.0001 vs hypokalaemia: aOR 1.83, 95% CI 1.70 to 1.96, P < 0.0001 vs mixed dyskalaemia: aOR 2·87, 95% CI 2.57 to 3.20, P < 0.0001). The same adverse effects of dyskalaemia were also observed in incidence of arrhythmias. Conclusions: Having dyskalaemic episodes during the ICU stay was significantly associated with increased odds of ICU mortality and arrhythmias.