Advances in oncologic and supportive care have led to improved prognosis and extension of survival time in critically ill patients with hematologic malignancies (HMs). Azoulay et al1 reported a large, prospective, multicenter study of intensive care unit (ICU) and post-ICU outcomes in patients with HMs who were admitted to the ICU without following a guideline for ICU admission. ICU admission occurred 4 days after hospital admission; 451 patients (44.6%) were admitted within 1 day, including 267 patients (26%) who were admitted directly to the ICU. The hospital, 90-day, and 1-year mortality rates were 39.3%, 47.5%, and 56.7%, respectively. As acknowledged by the authors, no rigid criteria were used across the different participating centers to make the decision to send patients to ICUs. The ICU triage policies were based on the experience of the clinicians with patients with HMs; however, the clinical judgment of physicians is inaccurate. The mortality rates in patients not admitted to the ICU because they were considered too well or too sick to benefit from intensive care were 11.3% and 74%, respectively. The mortality rate of the entire group (admitted and not admitted to the ICU), excluding patients who declined to participate in the study, was 46% (624 of 1,355 patients).

The principle of justice would dictate that ICUs use a valid and reliable mortality prediction model to determine admission.2 Scoring systems have been developed to measure the severity of illness and predict patient outcome.3 Chan et al4 reported that the Acute Physiology and Chronic Health Evaluation (APACHE) II scores determined at the time of admission to an ICU were predictive of in-hospital mortality in critically ill patients with cancer. Recently, Kopterides et al5 reported that the scoring systems for illness severity, including APACHE II, the Sequential Organ Failure Assessment (SOFA), and the Simplified Acute Physiology Score (SAPS) II, calculated on the first day of ICU stay, were fairly accurate predictors of ICU mortality, with good discrimination and acceptable calibration. Because clinical experience is not accurate, the authors should have used a scoring system for illness severity such as SAPS II or APACHE II, and should have reported the standardized mortality ratios for their patient groups.