HOUSTON -- Hypochloremia proved to be a "robust predictor" of all-cause mortality among critically ill patients, although the reasons for the association remained unclear, a large cohort study showed.
Patients with serum chloride values ≤98 mEq/L had more than a twofold higher 30-day mortality compared with the reference value of 107 mEq/L to 108 mEq/L. Hyperchloremia (≥112 mEq/L) almost doubled the mortality risk.
Both hypo- and hyperchloremia predicted an increased mortality risk in a multivariable analysis that excluded sodium levels. After adjustment for sodium, only hypochloremia remained an independent predictor of mortality.
"Although this is associative data, there is no causality," Kenneth Christopher, MD, of Brigham and Women's Hospital in Boston, said here at the Society of Critical Care Medicine meeting. "It is hypothesis generating and it is topical because it is responding to a lot of literature that shows that hypochloremic states are associated with mortality in animal models."
Abnormal serum chloride values -- elevated or reduced -- are common among ICU patients. However, the mechanisms controlling chloride levels in critically ill patients have remained poorly understood.
Studies of the relationship between chloride and mortality have produced conflicting results, as some studies have shown an association, whereas others have identified hyperchloremia as a mortality risk in critically ill patients, according to the background of the poster presentation by Christopher and colleagues. In general, hypochloremia is not an accepted risk factor.
The investigators hypothesized that abnormal chloride after admission to an ICU would be associated with increased mortality. To examine the relationship, they analyzed data on 51,789 critically ill patients at Massachusetts General and Brigham and Women's Hospital in Boston from 1998 to 2009.
Patients were grouped into deciles of serum chloride values ranging from ≤98 mEq/L to ≥112 mEq/L. The primary endpoint was 30-day mortality, and all comparisons were versus the reference value of 107 to 108 mEq/L.
In the unadjusted analysis, chloride values of ≤98 mEq/L to 104 mEq/L were associated with significantly higher mortality (OR 2.56 to 1.15, P=0.02 to P<0.001) as was a value ≥112 mEq/L (P<0.001).
Investigators developed two models that adjusted for age, gender, race, medical versus surgical patient, Deyo-Charlson Index, sepsis, creatinine, hematocrit, anion gap, HCO3, and white blood count. One of the models also included sodium.
In the model that omitted sodium, serum chloride values ≤98 mEq/L to 103 mEq/L remained predictors of increased mortality (OR 1.15 to OR 1.53, P=0.011 to P<0.001), as did values ≥112 mEq/L (OR 1.33, P<0.001).
After inclusion of sodium, hyperchloremia no longer predicted 30-day mortality risk (OR 1.01). However, hypochloremia remained an independent predictor of increased mortality across the range of ≤98 mEq/L to 104 mEq/L (OR 1.16 to OR 1.66, P=0.02 to P<0.001).
"The present study illustrates a graded, independent relationship between chloride at critical-care initiation and 30-day all-cause mortality," Christopher and colleagues concluded their poster presentation.
In a discussion that followed the presentation, Christopher said the study had the statistical power to demonstrate the existence of a true relationship between chloride and mortality among critically ill patients.
"Interestingly, hypochloremia really hasn't been studied that much," said Christopher. "Hypochloremia is caused by excessive water, and that would include patients with SIADH (syndrome of inappropriate antiduretic hormone), CHF, excess chloride loss, and patients on diuretics."
Primary source: Society of Critical Care Medicine
Source reference:
McMahon G, et al "Chloride is associated with mortality in the critically ill" SCCM 2012; Abstract 356.
Source reference:
McMahon G, et al "Chloride is associated with mortality in the critically ill" SCCM 2012; Abstract 356.
No comments:
Post a Comment