Association of first-24-hour hypernatremia and hyperkalemia co-occurrence with 30-day mortality in critically ill adults: a retrospective cohort study

First-24-hour hypernatremia–hyperkalemia co-occurrence may mark severe illness, but its incremental prognostic value is uncertain. We analyzed 13,383 adult ICU admissions (11,942 patients) in Taiwan during 2018–2020. Peak sodium > 145 and potassium > 5.5 mmol/L defined four phenotypes; peaks need not be simultaneous. This prognostic association analysis used information accumulated during the first ICU day. Patient-clustered logistic regression compared total APACHE II adjustment with complementary adjustment excluding its sodium/potassium points. Twenty repeated patient-grouped five-fold cross-validations assessed discrimination. Thirty-day mortality was 12.2%, 33.7%, 32.7%, and 74.4% in the neither ( n = 12,097), hypernatremia-only ( n = 841), hyperkalemia-only ( n = 367), and co-occurrence ( n = 78) groups. With total APACHE II, the co-occurrence odds ratio was 1.90 (95% CI 1.07–3.39), global phenotype P = 0.094, and mean AUC change 0.00007. With component-excluded adjustment, the corresponding odds ratio was 2.70 (1.53–4.75), global P = 0.002, and mean AUC change 0.00037. Additional sensitivity analyses examined outcome availability, in-hospital mortality, sparse data, and alternative thresholds. Outcome missingness and unavailable measurement/death timing limit interpretation. Co-occurrence characterized a small high-risk subgroup, but incremental discrimination was negligible; causality and clinical predictive utility were not established.