2.2 | Outcome and variables
The primary outcome was event-free survival (EFS), defined as the time between primary diagnosis and the occurrence of a first event (progression, relapse, subsequent malignant neoplasm [SMN], or death) or the end of follow-up (December 31, 2018). Most events were captured through chart abstraction. To avoid missing events that occurred at centers different from the primary treatment center, validated health administrative data-based algorithms using billings for chemotherapy, radiation, or palliative care with service dates after the end of the initial therapy were utilized.24
The main exposure variable was a 3-level categorical variable combining age and LOC. Pediatric LOC included patients aged 15 to 17.9 years old who were treated at a pediatric center. Adult ≥18 LOCcomprised 18 to 21 years old patients who were treated at an adult center. Patients in pediatric age group (<18 years) who were treated at an adult facility or diagnosed at a pediatric center but treated at an adult center, were categorized as adult<18 LOC . A patient’s main treating institution was categorized based on delivery of the majority of chemotherapy and/or radiotherapy during first three months after diagnosis. Patients who received upfront chemotherapy at a pediatric center and radiation at an adult center, were considered as pediatric-center patients.
Demographic and patient-related variables included sex and socioeconomic variables (neighborhood median income quintile, rurality, immigrant status). Neighborhood median income quintile was determined utilizing the Canadian census and respective postal code at the date closest to the date of cancer diagnosis. Patients were classified as having urban or rural residence according to their residential postal code at diagnosis and the Rurality Index for Ontario (RIO)-2004.25 The immigration status of each patient was determined by utilizing the Citizenship and Immigration Canada (CIC) database and recorded as a binary (yes/no) variable. The main disease-level variables were histology (OGS or EWS), DIL, metastatic status at diagnosis (localized versus metastatic), and site of the primary tumor (limb versus axial).
To calculate DIL, all healthcare claims and visit records in the year preceding cancer diagnosis were organized as discrete episodes and each episode was assigned a single ICD code representing the main purpose (symptom or diagnosis) of that episode. Two pediatric oncologists (PN and SG ) defined the diagnostic codes that were consistent with an underlying OGS or EWS diagnosis, based on their clinical expertise. The DIL for each patient was calculated as the time interval between the date of the first relevant physician encounter with a definite OGS- or EWS-related symptom and the date of cancer diagnosis. To evaluate the association between DIL and survival, categorical variables with two levels according to the calculated median DIL (i.e., short delay ≤ median and prolonged delay > median) were created separately for OGS and EWS.
Treatment variables other than LOC included clinical trial enrollment (enrolled vs non-enrolled), time from diagnosis to local therapy, and chemotherapy cumulative doses (mg/m2). Time to local therapy was defined as the time from the date of diagnosis to the date of first definitive local treatment (surgery and/or radiation) and was categorized as early (≤ 3 months) versus late (>3 months). The cumulative doses of chemotherapeutic agents were identified and recorded. The Cyclophosphamide Equivalent Dose (CED) was calculated for alkylating agents, using the following equation:26