Moderate-to-Large Increases in Perioperative Serum Sodium Concentration Associated With Adverse Neurologic Events After Continuous Flow Left Ventricular Assist Device Implantation Michael Mazzeffi, MD, MPH,* Christopher Paciullo, PharmD,† J. David Vega, MD,‡ Duc Nguyen, MD,‡ and Michael Connor, MD§ Objective: It was hypothesized that preoperative hypona- tremia is associated with increased 30-day mortality after left ventricular assist device placement, and that large increases in sodium concentration are associated with adverse neurologic events and 30-day mortality. Design: Data were collected retrospectively on all patients having continuous flow left ventricular assist device implan- tation between January 1, 2009 and March 31, 2013. Pre- operative variables, operative variables, and perioperative sodium concentrations were recorded. Both 30-day mortal- ity and 72-hour adverse neurologic events (stroke or seizure) were recorded as primary outcome variables. Preoperative sodium and Δ sodium (postoperative sodium-preoperative sodium) were analyzed as tests for 30-day mortality and adverse neurologic events using receiver operating charac- teristic curves. Both crude and adjusted logistic regression analyses were used to estimate odds ratios for the outcome variables. Setting: Tertiary care academic medical center. Participants: Patients having durable continuous flow left ventricular assist device placement. Interventions: None. Measurements and Main Results: Among 88 patients, 30- day mortality was 14% (12 of 88) and the rate of perioper- ative stroke or seizure was 9% (8 of 88). There were 3 strokes and 5 tonic-clonic seizures. Preoperative sodium was a poor discriminative test for 30-day mortality and stroke or seizure (AUC ¼ 0.47 and 0.57, respectively). Δ sodium was a poor discriminative test for 30-day mortality, but a fair discriminative test for stroke or seizure (AUC ¼ 0.55 and 0.78, respectively). Δ sodium was a good discriminative test for seizure alone (AUC ¼ 0.82) and a fair discriminative test for stroke alone (AUC ¼ 0.70). It also increased the odds of stroke or seizure significantly, even when adjusting for possible confounders. Conclusions: Moderate-to-large increases in sodium con- centration during left ventricular assist device placement appear to be associated with adverse postoperative neurologic events. Preoperative hyponatremia has no relationship with 30-day mortality or adverse perioperative neurologic events. & 2015 Elsevier Inc. All rights reserved. KEY WORDS: hyponatremia, stroke, seizure, ventricular assist device, mortality M ORE THAN 5 MILLION adults in the United States carry a diagnosis of heart failure. 1 Heart transplantation remains the only definitive therapy for this chronic condition despite advances in medical management. Unfortunately, the number of patients on the heart transplant waiting list far exceeds the number of donor hearts. Another therapeutic option for end-stage heart failure patients is mechanical circulatory support with durable, implantable devices that provide suffi- cient cardiac output to meet the patients’ metabolic demands as both a bridge to transplantation or for destination therapy. Improving technology and insertion technique with increased management experience have helped left ventricular assist devices (LVADs) become an important option for these patients. In 2011, more than 1,600 LVADs were implanted in the United States. 2 The number of LVADs implanted is expected to rise with the increasing prevalence of heart failure, which is projected to increase by 25% by the year 2030, 1 and improving outcomes from destination therapy. For these reasons, a solid understanding of LVAD-related complications increasingly is necessary for healthcare providers. Patients with LVADs are susceptible to adverse events, with up to 90% experiencing an event within 60 days of implanta- tion. 3 Complications, including right ventricular failure, anemia, and coagulopathy, are common during the postoperative period. 4 Electrolyte disturbances in LVAD patients and their impact on outcomes previously have not been addressed in much detail. In hospitalized patients with cirrhosis, kidney disease, heart failure, and recent surgery, hyponatremia is common and predicts a worse outcome, including higher mortality. 5–10 Hyponatremia and fluctuations in serum sodium concentration also have been associated with increased intensive care unit (ICU) mortality. 11 Specifically, changes in serum sodium 412 mmol/L are associated with increased ICU mortal- ity, with a proportional increase for greater fluctuation. 9 Similar relationships have been found in cardiac surgery patients, with preoperative, postoperative, and ICU-acquired sodium distur- bances all being associated with higher mortality. 12–14 To the authors’ knowledge, there are no published studies of hyponatremia and outcomes in patients after LVAD implanta- tion. The objective of this study was to examine the effect of preoperative hyponatremia and perioperative sodium changes on 30-day mortality and adverse postoperative neurologic events in patients undergoing durable LVAD implantation. It was hypothesized that preoperative hyponatremia would be associated with increased 30-day mortality and that large increases in serum sodium concentration during the perioper- ative period would be associated with 30-day mortality and adverse postoperative neurologic events. From the *Department of Anesthesiology, University of Maryland, Baltimore, MD; †Department of Pharmacy, Emory University, Atlanta, GA; ‡Department of Surgery-Division of Cardiothoracic Surgery, Emory University, Atlanta, GA; §Emory Center for Critical Care & Division of Pulmonary, Allergy, and Critical Care Medicine, Emory University, Atlanta, GA. Address reprint requests to Michael Mazzeffi, MD, MPH, Depart- ment of Anesthesiology, University of Maryland School of Medicine, 22 South Greene Street, Room S11C00, Baltimore, MD 21201. E-mail: mmazzeff@hotmail.com © 2015 Elsevier Inc. All rights reserved. 1053-0770/2601-0001$36.00/0 http://dx.doi.org/10.1053/j.jvca.2014.07.029 360 Journal of Cardiothoracic and Vascular Anesthesia, Vol 29, No 2 (April), 2015: pp 360–366