Identifying early predictors of mortality in pediatric patients with acute leukemia and pneumonia

Study objective: To identify clinical variables of pneumonia in children with acute leukemia that predicted respiratory failure and mortality. Design: A retrospective chart review of children with acute leukemia admitted to the hospital with the diagnosis of pneumonia or ARDS from March 1991 to April 1994. Setting: Lucile Salter Packard Children's Hospital at Stanford, a 168-bed teaching hospital and regional tertiary referral center for children in northern California. Patients: During this study period, 20% of the 174 admissions of children with acute leukemia had pneumonia at the time of admission or during the course of the hospitalization for a total of 36 admissions. The mean age of these children was 9.2 +/- 1.1 years. Results: Eleven percent of the children with pulmonary infiltrates in one quadrant on the chest x-ray film at the onset of pneumonia and 53% of the children with pulmonary infiltrates in more than one quadrant at the onset of pneumonia died. Fifteen percent of the children without sepsis at the onset of pneumonia and 70% of the children with sepsis at onset died. Eighteen percent of the children without shock at the onset of pneumonia and 75% of the children with shock at the onset died. None of the children died who required less than or equal to 3L/min of O-2 to maintain SO2 greater than or equal to 95%, but 79% of the children who required >3L/min O-2 died. Using the criteria ''>3 L/min O-2 by nasal cannula to maintain SO2 greater than or equal to 95%'' to identify the nonsurvivors had a sensitivity of 100% and specificity of 88%. This specificity was not increased by combining the criteria ''O-2 requirements at any time'' and ''the extent of pulmonary infiltrates at the onset of pneumonia.'' All children who required mechanical ventilatory support for respiratory failure had previously received >3L/min O-2 by nasal cannula to maintain SO2 greater than or equal to 95% for 37.8 +/- 12.9 h (range 3 to 96 h). Nine of the 10 children in our study who received mechanical ventilation died. Conclusion: In children with acute leukemia and pneumonia, the amount of O-2 required to maintain SO2 greater than or equal to 95% may identify those who are likely to develop respiratory failure hours before mechanical ventilatory support is needed. The ability to identify children at risk for respiratory failure is not increased by combining the risk factors ''oxygen requirements'' and ''extent of pulmonary infiltrates at the onset of pneumonia.'' Finally, only 10% of the children who required mechanical ventilatory support survived.