Assembly of a cohort of children treated for acute myeloid leukemia at free-standing children's hospitals in the United States using an administrative database.

Pediatric Health Information System data were used to establish a multi-center cohort of 1,686 children treated for newly diagnosed acute myeloid leukemia (AML). The cohort assembly process, which included myeloid leukemia ICD-9 discharge diagnosis codes and manual review of induction chemotherapy, was validated by chart review at a single institution. The use of ICD-9 codes alone resulted in a poor positive predictive value (PPV; 31%). Inclusion of the results from the chemotherapy review improved the PPV to 100% without compromising sensitivity (95.7%). This cohort provides a reliable source for future comparative effectiveness and clinical epidemiology studies in pediatric AML.