(129) Short- and Long-Term Associations of Hyperkalemia-related Events on Serum Potassium Time Courses in Patients with Chronic Kidney Disease: A Retrospective Cohort Study Using Electronic Health Records Data from the United States

Background: Hyperkalemia (HK) is defined as an elevated serum potassium (sK+) level and can result in serious conditions including arrhythmias or cardiac arrest.

Objectives: To assess the short- and long-term associations of events (i.e., treatments, diagnoses, medical procedures) on the sK+ time course in patients with chronic kidney disease (CKD) from routine clinical care.

Methods: Using US Optum® electronic health records data from 01/2010-12/2020, we selected patients with CKD, defined as ≥2 eGFR values of 15–60 ml/min/1.73m2 and/or ≥2 UACR values ≥30 mg/g at least 90-365 days apart (second qualifying value = index date; prior 365 days = baseline). We identified an episode of HK using two elevated sK+ values (≥5.5 mmol/l) or one elevated sK+ plus either the initiation of K+ binders or a HK diagnosis, not longer than 7 days apart. A bootstrapping multi-variable Cox regression analysis suggested 300 variables influencing sK+. Population-averaged time series was used to align incident events of these variables during follow-up and to assess the short-term (days 0-3) and long-term associations (days 20-100) on sK+ levels adjacent to the respective event.

Results: Among the 1,771,900 patients meeting the selection criteria, median age was 75 years (interquartile range [IQR] 66–80), 57.7% were women. Most dominant baseline variables associated with elevated HK risk included diagnoses of HK (Median Hazard Ratio [HR], 2.85; IQR, 2.68-3.04) and liver cirrhosis (HR, 2.1; 1.94-2.36), prescriptions of sodium polystyrene sulfonate (HR, 2.71; 2.62-3.04) and ion exchange resins (HR, 2.7; 2.49-2.96); variables associated with a lower HK risk were atenolol/chlorthalidone (HR: 0.57; 0.55-0.58) and tretinoin (HR: 0.58; 0.57-0.59) prescriptions. Population averaged time courses show different characteristic shapes with short- and long-term associations on serum potassium levels during follow-up. Highest short-term associations were found for sodium polystyrene sulfonate prescriptions (-0.485 mmol/L), hemodialysis (-0.36 mmol/L), diagnosis of hyperkalemia (-0.34 mmol/L), prescription for phosphate binders (-0.29 mmol/L) and sevelamer (-0.28 mmol/L). Major long-term associations were found for spironolactone (+0.19 mmol/L) and valganciclovir prescriptions (+0.18 mmol/L), partial nephrectomy (+0.13 mmol/L), and prescriptions for metolazone (-0.11 mmol/L) and hydroxyurea (-0.1 mmol/L).

Conclusions: Our novel results identified and quantified clinical covariates associated with short- and long-term associations on sK+ levels in a large real-world cohort of patients with CKD. These findings highlight clinically relevant risk factors associated with sK+ variability which may inform optimal management of patients with CKD.