Predictors of outcome for leukemia patients in the intensive care unit
Purpose: To identify the variables associated with outcome for leukemia patients admitted to the Intensive Care Unit (ICU). Methods: Our study was an 18 month (1/1/93-6/30/94) retrospective review of a cohort of leukemia patients admitted to the ICU of a single, tertiary care, teaching hospital. Seventy-five patients between the ages of 19 and 75 were identified; the overall mortality rate was 64%. Patient data collected included: age and gender, class of leukemia, date of diagnosis, bone marrow transplant history, anthracycline drug history, mechanical ventilation, duration of hospital stay prior to ICU admission, ICU length of stay, and hospital mortality. Results: Variables entered into a multiple logistical regression resulted in the following mortality model: Odds Parameter p Ratio Variable Estimates Value [95%CI] No. of Hosp. Days 0.05 0.09 1.05 Prior to ICU(x1) [0.99 to 1.11] BMT (x2) 0.99 0.14 2.70 [0.72 to 10.08] Ventilation (x3) 2.02 0.0017 7.56 [2.14 to 26.78] The logit equation resulting from the above data is: G(x)=-1.65 + (0.05)x1 + (0.99) x2+(2.02)x3 Conclusions: Our data support previous studies that mechanical ventilation is the strongest predictor of poor outcome for the leukemia patient in the ICU. In addition, a history of bone marrow transplantation and a longer duration of stay prior to ICU admission contribute to an increased probability of mortality. Our findings also indicate that outcome of the leukemia patient in the ICU is independent of age, class or duration of leukemia, and anthracycline history. Clinical Implications: A unique patient population, such as that of leukemia patients, warrants the construction of an alternative predictor of outcome model for the ICU. Further analysis of prospective data will be aimed at validating this model along with assessing the long term quality of life and duration of survival.