Clinical Characteristics of Patients with Siadh Who Respond to Fluid Restriction in the Management of Hyponatremia from a Global Registry

Program: Abstracts - Orals, Poster Preview Presentations, and Posters
Session: SUN 0439-0452-Health Outcomes
Sunday, June 22, 2014: 1:00 PM-3:00 PM
Hall F (McCormick Place West Building)

Poster Board SUN-0448
Joseph G. Verbalis, MD1, Susan Boklage, MS, MPH2, Joseph Chiodo III, PharmD2 and Richard D Pollack, PhD3
1Georgetown University Medical Center, Washington, DC, 2Otsuka America Pharmceutical, Inc., Princeton, NJ, 3Advanced Analytic Solutions, Newtown, PA
INTRODUCTION: Hyponatremia (HN) is the most common electrolyte disorder and an independent predictor of increased mortality in hospitalized patients (pts). Fluid restriction (FR) is often the first line of therapy for HN; however, data assessing its effectiveness is limited.

HYPOTHESIS: Although a large number of demographic and clinical characteristics exist for pts with HN, a relatively small number of variables will discriminate pts responses to FR.

METHODS: Medical records of pts meeting the HN registry ( #NCT01240668) entry criteria for SIADH (age ≥18 yrs, euvolemic, serum sodium ([Na+]) ≤130 mEq/L) were abstracted. Pts treated with only FR on the first day a pt‘s [Na+] was ≤130 mEq/L (baseline [BL]), and who had a follow-up [Na+] within an 18-36 hour window of the first [Na+] were included. Response to FR was defined as a change from BL to follow-up [Na+] within 24 hours of at least: 2 mEq/L, 3 mEq/L, or 4 mEq/L. Exhaustive CHAID (CHi-squared Automatic Interaction Detection) was chosen as the segmentation method due to its ability to detect and evaluate both linear and non-linear relationships. Each change in response was modeled separately.

RESULTS:  365 pts were included in the final analysis. Pt demographics at admission and baseline lab results at the time of HN diagnosis were included in the model. By design, all variables included in each of the final segmentations were significant (p<0.05). Four variables were the most significant in each of the 3 models: serum [Na+], urine osmolality (UO), urine [Na+] and serum creatinine. The cutoff of UO (mOsm/kg H2O) for predicting response was <500 for pts with [Na+] 120-125 and <350 for [Na+] <120 (i.e., pts with UOs greater than these cutoffs had a higher probability of not responding).

CONCLUSIONS:  HN pt response to FR is associated to a relatively small number of parameters that differ slightly with the desired response in [Na+]. For pts with [Na+] >125, no variables predicted a response >50%. For pts with [Na+] ≤125, UO was the most critical factor in determining responses from 2-4 mEq/L; this is consistent with clinical experience that pts with lower free water clearances are less likely to respond to FR. Yet in this population, only 43% of physicians recorded UO at initial diagnosis. These results confirm the importance of easily measured laboratory parameters as predictors of response or lack of response to FR, and indicate a general deficiency of appreciation for these important clinical associations and how they can be utilized to enhance the management of HN pts.

Disclosure: JGV: Consultant, Cornerstone Therapetics, Consultant, Ferring Pharmaceuticals, Consultant, Otsuka, Principal Investigator, Otsuka, Speaker, Otsuka, Advisory Group Member, Otsuka. SB: Employee, Otsuka. JC III: Employee, Otsuka. RDP: Consultant, Otsuka.

*Please take note of The Endocrine Society's News Embargo Policy at