Hospital resource utilization in childhood cancer.

To describe the patterns and predictors of hospital resource utilization in a cohort of children with newly diagnosed cancer, a retrospective cohort study of 195 consecutively diagnosed children with cancer at a single large Midwestern children's hospital was conducted. Patients were diagnosed between November 1995 and March 1997. All hospital encounters for these patients starting from the time of diagnosis to 3 years from diagnosis were identified using hospital administrative data. The patients were categorized into four diagnostic groups: lymphoid malignancies (acute lymphoblastic leukemia and lymphoma), myeloid leukemias (acute myeloid leukemia and chronic myeloid leukemia), central nervous system tumors, and solid tumors. Hospital charges and length of stay for patients in each diagnostic category were described. Predictive models for total resource consumption (total hospital charges) and intensive care use were derived. One hundred sixty-five of the 195 were admitted to Riley Hospital for Children at least once during the 3-year period following diagnosis. Among these 165, mean age at diagnosis was 6.9 years (minimum newborn, maximum 18.7 years). The ratio of boys to girls was 99:66 (1.5:1). The distribution of 165 diagnoses was as follows: 65 (39%) with lymphoid malignancy, 13 (8%) with myeloid leukemia, 36 (22%) with central nervous system tumors, and 51 (31%) with solid tumors. Sixty-two patients (38%) used the pediatric intensive care unit (PICU) at least once; 22 patients (13%) underwent stem cell transplantation. Sixty-five patients (39%) entered clinical trials. One hundred thirty-nine patients (84%) were alive at the end of 3 years. Three-year cumulative hospital charges were USD 16 million--almost USD 100,000/child hospitalized. Half of these charges were incurred in the first 4.5 months after diagnosis. Half of all hospital charges accrued to only 12.7% of patients; these patients were more likely to have a diagnosis of myeloid leukemia, to have undergone stem cell transplantation, and to have used the PICU. There were three independent predictors of hospital charges (log transformed): stem cell transplantation, PICU utilization, and death within 3 years of diagnosis. PICU utilization was predicted by tumor type (myeloid leukemia and central nervous system tumors were positive predictors of PICU utilization; lymphoid malignancy and solid tumors were negative predictors), stem cell transplantation, and death within 3 years of diagnosis. The authors conclude that hospitalization for childhood cancer is common, costly in the short term, and to some extent predictable. These data suggest that failures of current treatment not only lead to death but also add significantly to hospital resource utilization.