Previous studies suggest that red cell distribution width, a measure of erythrocyte size variability, may predict long-term mortality, particularly in cardiovascular disease. Less research has focused on the prognostic utility of red cell distribution width in an acutely hospitalized population.

We performed a secondary analysis of prospectively collected data on 74,784 consecutive hospitalized adults with red cell distribution width measured on admission. The primary outcome of interest was in-hospital mortality; a secondary outcome was unplanned transfer to the intensive care unit. We calculated multivariable logistic models adjusted for age, gender, race, and comorbid conditions.

The overall in-hospital mortality rate was 1.3% (95% confidence interval [CI], 1.2-1.4). As red cell distribution width increased, so did mortality, from 0.2% (lowest red cell distribution width decile) to 4.4% (highest red cell distribution width decile). Unadjusted red cell distribution width significantly discriminated between hospital survivors and nonsurvivors (area under the curve 0.74). In multivariate analyses, for every 1% increment in red cell distribution width at the time of admission, the odds for in-hospital mortality increased by 24% (odds ratio 1.24; 95% CI, 1.20-1.27); findings were robust across comorbidity subgroups. The rate of unplanned intensive care unit transfer was 7.0% (95% CI, 6.8-7.2) and in unadjusted analyses increased more than 2-fold from 4.5% in the lowest to 11.6% in the highest red cell distribution width decile. This relationship was significantly confounded but remained significant in multivariate analysis (odds ratio 1.04 per 1% red cell distribution width increment; 95% CI, 1.03-1.06).

Red cell distribution width strongly and independently predicted in-hospital mortality in this large cohort of hospitalized patients. It also was associated with acute decompensation among patients on the general ward, but to a lesser degree. The mechanisms underlying these findings are unknown.