Patients with hematologic malignancies requiring intensive chemotherapy are at risk for life-threatening complications. Organ failure may appear rapidly and delay in initiating life-sustaining interventions may result in increased mortality. This encourages great alertness although not all patients require close monitoring. It is therefore critical to identify which patients are the most at risk for clinical deterioration to consider increased surveillance in these patients. The benefit of early intensive care unit (ICU) admission, as soon as the first signs of organ dysfunction appear, must also be clarified. Such an intervention could increase survival of patients by close monitoring and early initiation of organ-specific interventions but could also be responsible for anxiety and increased use of ICU resources. Many teams have analyzed the impact of early warning systems (EWS) including vital signs to detect organ dysfunction early on. It has been shown that these EWS could positively impact survival in many medical fields (pre-hospital, medicine or surgery departments). A few retrospective studies have explored the impact of EWS in hematology, with overall good prediction for ICU admission and mortality. Until now, it has however not been formally demonstrated that early ICU admission, as soon as the first signs of organ dysfunction appear, could benefit patients with hematologic malignancies. A randomized controlled trial studying the impact of early intervention would clarify the role of such a strategy. In this study, the investigators will prospectively evaluate the implementation of the National Early Warning Score (NEWS), with systematic referral to the ICU in high-score patients, to improve the survival of patients receiving intensive chemotherapy in ten academic centers. This score is one of the most performant and most frequently used to predict organ failure. Its calculation only requires vital signs such as respiratory rate, peripheral oxygen saturation, need for oxygen therapy, body temperature, arterial pressure, heart rate, and level of consciousness. The investigators will therefore study the impact of ICU admission in patients with high NEWS in a randomized, controlled trial. A cluster randomization is planned in which the centers will be randomized between usual care (control group) and interventional care with transfer to the ICU in the event of a NEWS score ≥7 (interventional group). Each parameter used to calculate the NEWS will be collected at least three times a day by the attending nurse.