Introduction: Patients with hematologic malignancy (HM) may require intensive care unit (ICU) admission because of malignancy, treatment, or immunosuppression. Previous studies have reported various prognostic factors, with heterogeneity in study populations and mortality. A comprehensive understanding of prognosis in such patients is needed to better inform shared decision making. Herein, we aimed to summarize the prognosis of patients with HM admitted to ICUs and potential prognostic factors for mortality.
Methods: We searched Medline from inception to August 1, 2024 for observational studies of critically ill adult patients with any type of HM, in whom mortality was assessed. Our primary outcome was ICU mortality. We also evaluated prognostic factors for ICU mortality. We used a random-effects model to pool mortality estimates and adjusted odds ratios (ORs) of potential prognostic factors. We used adjusted ORs from the final multivariate models of the included studies that used an appropriate ratio of events per covariate.
Results: We included 76 observational studies published between 1988 and 2024 involving 44,462 participants. ICU mortality was 45% (95% confidence interval [CI], 42-49); in-hospital, 30-day, 3-month, 6-month, and 1-year mortality were 54% (95% CI, 48-60), 55% (95% CI, 48-62), 61% (95% CI, 48-73), 77% (95% CI, 71-82), and 73% (95% CI, 65-82), respectively. Meta-regression analysis suggested a decline in ICU mortality over time (beta= -0.006% per year; P= 0.023). Factors associated with increased ICU mortality included higher SOFA (OR 1.19; 95% CI, 1.07-1.31), mechanical ventilation (OR 5.89; 95% CI, 3.19-10.87), renal replacement therapy (OR 2.98; 95% CI, 1.78-5.01), and use of vasopressors (OR 3.57; 95% CI, 2.45-5.19). Age, sepsis, Charlson comorbidity index, bone marrow or hematopoietic stem cell transplantation, or neutropenia were not such factors.
Conclusions: Although decreasing over time, ICU mortality of patients with HM remained high, as did mortality at subsequent time points. The prognostic factors identified were related to severity of illness and had little to do with baseline patient characteristics. These factors should be used cautiously in shared decision making and advance care planning. Studies to identify modifiable prognostic factors and interventions are warranted.
