The relationship between serum potassium, potassium variability and in-hospital mortality in critically ill patients and a before-after analysis on the impact of computer-assisted potassium control

Lara Hessels, Miriam Hoekstra, Lisa J Mijzen, Mathijs Vogelzang, Wim Dieperink, Annemieke Oude Lansink, Maarten W Nijsten

IntroductionThe relationship between potassium regulation and outcome is not known. Our first aim was to determine the relationship between potassium levels and variability in ICU stay and outcome. The second aim was to evaluate the impact of a computer-assisted potassium regulation protocol.MethodsWe performed a retrospective before-after study including all patients >15 years admitted to the ICU of our university teaching hospital for more than 24 hours between 2002 and 2011. Potassium control was fully integrated with computerized glucose control (GRIP-II). The potassium metrics that we determined included mean potassium, potassium variability (defined as the standard deviation of all potassium levels), percentage of ICU-time below and above the reference range (3.5 through 5.0 mmol/L). These metrics were determined for the first ICU day (early phase) and the subsequent ICU days (late phase, that is day 2 to day 7). We also compared potassium metrics and in-hospital mortality before and after GRIP-II was implemented in 2006.ResultsOf all 22,347 ICU admissions, 10,451 (47%) patients were included. A total of 206,987 potassium measurements were performed in these patients. 4,664 (45%) patients were regulated by GRIP-II. The overall in-hospital mortality was 22%. There was a U-shaped relationship between the potassium level and in-hospital mortality (P <0.001). Moreover, potassium variability was independently associated with outcome. After implementing GRIP-II, in the late phase the time below 3.5 mmol/l decreased from 9.2% to 3.9% and the time above 5.0 mmol/L decreased from 6.1% to 5.2% and potassium variability decreased from 0.31 to 0.26 mmol/L (all P <0.001). The overall decrease in hospital mortality from 23.3% before GRIP-II to 19.9% (P <0.001) after the introduction of GRIP-II was not related to a specific potassium subgroup.ConclusionsHypokalemia and hyperkalemia and potassium variability were independently associated with increased mortality. Computerized potassium control clearly resulted in improved potassium metrics.