Rationale: It is well known that distinctly abnormal blood potassium values can cause serious complications such as cardiac arrhythmias. Although potassium regulation is generally considered important, hardly any research has been done about potassium regulation in intensive care patients. The investigators hypothesize that different potassium target-values, within the as normal accepted range, may have different effects in critically ill patients.

Study design: A prospective trial comparing two different potassium target-values. Potassium will be tightly regulated with the already fully operational GRIP-II computer program.

Study population: 1200 adult patients admitted at the thoracic intensive care unit of the University Medical Center Groningen.

Intervention: Comparison between two variations of standard therapy: potassium target-value of 4.0 mmol/L versus 4.5 mmol/L.

Main study parameters/endpoints: The primary endpoint is the incidence of atrial fibrillation or atrial flutter from ICU-admission to hospital discharge. Secondary endpoints are serum levels of potassium and the other main electrolytes, renal function and renal potassium excretion, the relation with insulin and glucose, the cumulative fluid balance, (ICU) length of stay and mortality.

Inclusion Criteria:

• All adult patients admitted to the thoracic and surgical ICU of the University Medical Center Groningen.
• Computerized potassium regulation with GRIP-II

Exclusion Criteria:

• Patients who are not potassium regulated with GRIP-II (patients who take their own meals).