While severe hyperkalemia is commonly encountered, its manifestation in hospitalized patients and related outcomes are unclear. We aimed to examine development of hyperkalemia in hospitalized patients and associated outcomes.

Data from a county hospital electronic health record were used to assess all inpatient admissions, 2012-2016, for non-dialysis-dependent patients with ≥1 K value for development of hyperkalemia. Unadjusted odds ratios (ORs) were calculated for associations of the maximum K value with in-hospital mortality and adjusted ORs were calculated for death associated with hyperkalemia.

In 47,018 individual patient hospitalizations, 1.3% had a maximum K ≥6.0 mEq/L and 4.2% <3.5 mEq/L. Fifth and 95th percentiles for maximum K values were 3.5 and 5.3 mEq/L. For high-K patients with a prior K value, the mean (SD) of the immediate pre-maximum K value was 5.0 ± 1.0 mEq/L, and the mean difference in K values (immediate pre-maximum to maximum) was 1.5 ± 1.1 mEq/L; mean duration between these two K tests was 10.7 ± 14.9 hours. Compared with maximum K values 3.5 to 4.0 mEq/L, ORs for death were 37.1 (95% confidence intervals, 25.8-53.3) for K 6.0 to <6.5, 88.6 (56.8-138.2) for K ≥7.0, and 18.9 (4.3-82.2) for K <3.0 mEq/L. In adjusted models, the OR for death for K ≥6.0 mEq/L was 4.9 (3.7-6.4).

Peak K values ≥6.0 mEq/L were associated with mortality. Values tended to increase rapidly, limiting opportunities for detection and treatment. Systems-based approaches to detect life-threatening hyperkalemia should be studied.