Association of admission hematocrit with 6-month and 1-year mortality in intensive care unit patients
This study examined the association of hematocrit (Hct) levels measured upon intensive care unit (ICU) admission and red blood cell transfusions to long-term (1-year or 180-day) mortality for both surgical and medical patients. Administrative and laboratory data were collected retrospectively on 2393 consecutive medical and surgical male patients admitted to the ICU between 2003 and 2009. We stratified patients based on their median Hct level during the first 24 hours of their ICU stay (Hct < 25.0%, 25% ≤ Hct < 30%, 30% ≤ Hct < 39%, and 39.0% and higher). An extended Cox regression analysis was conducted to identify the time period after ICU admission (0 to <180, 180 to 365 days) when low Hct (<25.0) was most strongly associated with mortality. The unadjusted and adjusted relationship between admission Hct level, receipt of a transfusion, and 180-day mortality was assessed using Cox proportional hazards regression modeling. Patients with an Hct level of less than 25% who were not transfused had the worst mortality risk overall (hazard ratio [HR], 6.26; 95% confidence interval [CI], 3.05-12.85; p < 0.001) during the 6 months after ICU admission than patients with a Hct level of 39.0% or more who were not transfused. Within the subgroup of patients with a Hct level of less than 25% only, receipt of a transfusion was associated with a significant reduction in the risk of mortality (HR, 0.40; 95% CI, 0.19-0.85; p = 0.017). Anemia of a Hct level of less than 25% upon admission to the ICU, in the absence of a transfusion, is associated with long-term mortality. Our study suggests that there may be Hct levels below which the transfusion risk-to-benefit imbalance reverses.