RESUMEN
BACKGROUND: Hypernatremia has been associated with mortality in neurocritically ill patients, with and without traumatic brain injury (TBI). These studies, however, lack concomitant adjustment for hyperchloremia as a physiologically co-occurring finding despite the associations with hyperchloremia and worse outcomes after trauma, sepsis, and intracerebral hemorrhage. The objective of our study was to examine the association of concomitant hypernatremia and hyperchloremia with in-hospital mortality in moderate-severe TBI (msTBI) patients. METHODS: We retrospectively analyzed prospectively collected data from the OPTIMISM-study and included all msTBI patients consecutively enrolled between 11/2009 and 1/2017. Time-weighted average (TWA) sodium and chloride values were calculated for all patients to examine the unadjusted mortality rates associated with the burden of hypernatremia and hyperchloremia over the entire duration of the intensive care unit stay. Multivariable logistic regression modeling predicting in-hospital mortality adjusted for validated confounders of msTBI mortality was applied to evaluate the concomitant effects of hypernatremia and hyperchloremia. Internal bootstrap validation was performed. RESULTS: Of the 458 patients included for analysis, 202 (44%) died during the index hospitalization. Fifty-five patients (12%) were excluded due to missing data. Unadjusted mortality rates were nearly linearly increasing for both TWA sodium and TWA chloride, and were highest for patients with a TWA sodium > 160 mmol/L (100% mortality) and TWA chloride > 125 mmol/L (94% mortality). When evaluated separately in the multivariable analysis, TWA sodium (per 10 mmol/L change: adjusted OR 4.0 [95% CI 2.1-7.5]) and TWA chloride (per 10 mmol/L change: adjusted OR 3.9 [95% CI 2.2-7.1]) independently predicted in-hospital mortality. When evaluated in combination, TWA chloride remained independently associated with in-hospital mortality (per 10 mmol/L change: adjusted OR 2.9 [95% CI 1.1-7.8]), while this association was no longer observed with TWA sodium values (per 10 mmol/L change: adjusted OR 1.5 [95% CI 0.51-4.4]). CONCLUSIONS: When concomitantly adjusting for the burden of hyperchloremia and hypernatremia, only hyperchloremia was independently associated with in-hospital mortality in our msTBI cohort. Pending validation, our findings may provide the rationale for future studies with targeted interventions to reduce hyperchloremia and improve outcomes in msTBI patients.