2.1 | Study design, population, and data resources
After obtaining Research Ethics Board approval from the Hospital for Sick Children and Sunnybrook Health Sciences Centre, we carried out a population-based retrospective cohort study using the IMPACT Cohort. The need for individual consent was waived. The creation of the cohort has been described in detal previously.5 It comprises all Ontario residents, aged 15 to 21 years at diagnosis of one of six common cancers (including bone tumors) treated at an adult center (a Regional Cancer Center or a community hospital) or a pediatric center (one of the five specialized, tertiary care centers) from 1992 to 2012.5 In Ontario, AYA aged less than 18 may receive care at either pediatric or adult centers, whereas older AYA (≥18) almost always are treated in adult centers.7,9 IMPACT was built using two data sources. The Pediatric Oncology Group of Ontario Networked Information System (POGONIS) database captures demographic, disease- and treatment-related data of children diagnosed with cancer before the age of 18 who received their care at a pediatric center.22,23 AYA diagnosed between ages 15 to 21 who received their cancer care at an adult center were identified through the Ontario Cancer Registry (OCR) and their clinical data collected by trained chart abstractors.5,7 These data were combined to form the IMPACT database. All AYA patients diagnosed with OGS or EWS, except those who lived or were diagnosed out of the province (Ontario), were included in this study. The IMPACT database was linked deterministically using encrypted personal identifiers to population-based health services databases (CIHI-DAD for hospitalization records, NACRS for emergency room visits, OHIP for outpatients visits and physician’s billing information; see Appendix 1 ) hosted at ICES (formerly, the Institute for Clinical Evaluative Sciences).