Abstract
AbstractBackgroundChildhood, teenage and young adult (CTYA, 0-24 years) cancers are rare and diverse, making timely diagnosis challenging. Studies based on adult cancers suggest that the development and integration of clinical decision tools in primary care aid earlier cancer detection, yet, these have not been explored for CTYA cancers.AimTo develop and validate a primary care-based risk prediction tool to identify CTYA who are at increased risk of cancer.Methods and analysisUsing the QResearch Database, a nationally representative primary care database, we will generate an open cohort of children, teenagers and young adults (0-24 years) who were registered with a GP between 1stJanuary 1998 and 31stDecember 2019. CTYA will be followed up from the date at which the first cancer-relevant symptom is recorded in the records (index date) until the date of cancer diagnosis/6-months, whichever comes first. Candidate variables will include symptoms, signs, blood test results and demographic factors. Model derivation will include two approaches, Cox regression and logistic regression. Apparent performance of the derived model will be explored and subsequently internally-externally cross-validated to investigate performance heterogeneity and geographical transportability.
Publisher
Cold Spring Harbor Laboratory