Linking Hyponatremia with Adverse Outcomes
Is it Time to Consider a New Reference Range for Serum Sodium?
By Genna Rollins
Hyponatremia is the most common electrolyte disorder in hospitalized patients and its effects have been studied in various patient groups and settings. However, new research extends prior findings to a sizable population and explores the relationship between different hyponatremia presentations and mortality, length of stay, and patient disposition. The study is the subject of this issue of Strategies.
Hyponatremia is a common electrolyte disturbance that has been associated with adverse outcomes—particularly mortality—in relatively small study populations and specific disease states such as heart failure, liver disease, and pneumonia. However, there is little evidence about the relationship between hyponatremia and other outcomes such as length of stay or patient disposition following hospitalization. The authors of a new study explored these questions in >50,000 hospitalized patients and across the spectrum of presentations of the disorder, including community-acquired, hospital-aggravated, and hospital-acquired hyponatremia (Arch Intern Med 2010;170:294-302).
“The implications of hyponatremia in hospitalized patients has not been fully studied, and we felt it would be important to garner a better understanding of what relationship it has with clinical outcomes,” explained lead author, Ron Wald, MDCM, MPH, assistant professor of medicine at the University of Toronto and scientist at the Keenan Research Centre of the Li Ka Shing Knowledge Institute, also in Toronto. “Our hypothesis was that hyponatremia would be associated with all the adverse outcomes we studied, including in-hospital mortality, duration of stay, and the likelihood that the patient will go to a chronic care facility and not their home environment after leaving the hospital.”
Wald and his colleagues explored these issues by accessing discharge abstracts and the electronic laboratory database of all adult inpatients at St. Elizabeth's Medical Center in Boston between 2000 and 2007. Obstetrical patients and those without a documented serum sodium value on day of admission or during hospitalization were excluded, resulting in a final analytic data set of 53,236 patient records. Patients were defined as having community-acquired, hospital-aggravated, or hospital-acquired hyponatremia or normonatremia based on serum sodium levels on the day of admission and subsequent times during hospitalization. Community-acquired hyponatremia was defined as 138 mEq/L at the time of admission; hospital-aggravated hyponatremia was defined as a further decline in serum sodium concentration ≥2 MEq/L during the first 48 hours in patients with community-acquired hyponatremia, while hospital-acquired hyponatremia was designated as a nadir serum ≤138 mEq/L when the admission serum sodium had been within the reference range in admissions lasting longer than 1 day.
The primary outcomes examined included in-hospital mortality, length of stay (LOS) and patient disposition, meaning discharge home, to short- or long-term care facilities, or against medical advice. The researchers conducted multivariate analysis and adjusted for confounders to evaluate the relationship between hyponatremia and these outcomes. They found community-acquired hyponatremia in 37.9% of patients, hospital-aggravated hyponatremia in 5.7%, and hospital-acquired hyponatremia in 38.2% of the admissions lasting longer than 1 day.
Community-acquired hyponatremia was associated with adjusted odds ratios (OR) of 1.52 for in-hospital mortality and 1.12 for discharge to a short- or long-term care facility, as well as a 14% adjusted increase in LOS. In comparison to community-acquired hyponatremia that did not worsen during hospitalization, hospital-aggravated hyponatremia was associated with adjusted ORs of 1.46 and 1.10 for in-hospital mortality and discharge to a short- or long-term care facility, respectively, and a 12% adjusted increase in LOS. In the case of hospital-acquired hyponatremia, there were adjusted ORs of 1.66 and 1.64 for in-hospital mortality and discharge to a facility, respectively, and a 64% adjusted increase in LOS. “Our primary conclusion is that hyponatremia of all presentations is associated with adverse outcomes,” said Wald. “Generally speaking, increasing severity of hyponatremia is associated with an increasing likelihood of each of the adverse outcomes that we evaluated.”
These findings have potentially important ramifications for patient care, according to Liviu Klein, MD, MS, assistant professor of medicine and director of the heart failure device program at the Northwestern University Feinberg School of Medicine and Northwestern Memorial Hospital in Chicago. “The significance of the study is that they extended on a larger scale prior studies and over a variety of diseases. The results are telling us that regardless of the underlying pathology that leads to hyponatremia, hyponatremia is a bad sign, and it's an easy marker to identify.” Klein, who has studied hyponatremia in heart failure, was not involved with the study.
In addition to their specific findings around hyponatremia and adverse outcomes, the researchers believe their study should prompt further investigation about appropriate reference ranges for serum sodium. “We felt uncomfortable using a range of 135 to 145 molar equivalents per liter because even though it's roughly the normal range for serum sodium concentrations used in most clinical labs, we felt that this so-called normal range is somewhat arbitrary,” explained Wald. “Before undertaking the study we looked at the data in an unadjusted way and noticed that admission serum sodium concentration was quite predictive of outcome.”
The researchers developed a restrictive cubic spline depicting the unadjusted relationship between serum sodium levels and in-hospital mortality before conducting any further analysis. “Not surprisingly, we found that at a serum sodium concentration of 140 molar equivalents per liter, in-hospital mortality was lowest. But as we went down in increments of one millimole per liter we found that when sodium concentrations hit 137 molar equivalents per liter there was a small but significant increase in the likelihood of in-hospital mortality. On the other side, when serum sodium went above 142 molar equivalents per liter, we noticed a take-off towards increased mortality.” As a result of these observations, the researchers used a narrow reference range of 138–142 mEq/L for their complete analysis.
Klein agreed that this evidence, along with other supporting studies, could lead hospitals to revisit their reference ranges for serum sodium. “We had a somewhat similar story a few years ago with digoxin, where formerly up to two nanograms per milliliter was considered normal. Then a couple of studies showed that a level above one nanogram per milliliter was leading to increased mortality, and in the last several years in our hospital, and I suspect others, anything above one nanogram per milliliter is considered abnormal,” he noted.
Wald and Klein also concurred that both clinicians and laboratorians may not have a full appreciation of the implications of low, but not critical, serum sodium values. “The basic chemistry panel is one of the most common, but I think doctors sometimes look more at creatinine or BUN and kind of overlook sodium until it's less than 130 or so,” observed Klein. Similarly, labs may have fairly low critical values for sodium, according to Wald. “Clinical laboratorians may need to reconsider the serum sodium concentrations at which they would alert clinicians in a more urgent fashion to an abnormal reading. They may only send the physician an urgent alert if the sodium is really low—say less than 125—but I think that what we consider mild-moderate hyponatremia may actually be harmful and related to adverse outcomes.”