Identifying subgroups at risk for childhood cancer death: A latent class analysis
Mentor: Dr. Kimberly Johnson
In 2018, an estimated 10,590 new cancer diagnoses and 1,180 cancer deaths are expected to have occurred in U.S. children. Although the overall five-year survival rate for childhood cancers has increased to approximately 80%, cancer remains the leading cause of death from disease among children, accounting for approximately 15% of all childhood deaths under 15 years of age. In addition to cancer type, prior studies have established various sociodemographic, economic and clinical factors impacting risk of childhood cancer death, including parents’ socioeconomic status, child’s race/ethnicity, health insurance coverage, and cancer stage at diagnosis. Almost all prior studies have focused exclusively on bivariate relationships between these factors and cancer death, treating other risk factors as cofounders or not considering them at all. Ignoring potential interrelationships among risk factors only tells a partial story of individual risk for cancer death. To address this problem, this project aims to: 1) identify patterns in demographic, socioeconomic, and clinical characteristics that are associated with childhood cancer death in the United States, and 2) determine subgroups of children who are at particularly high risk for cancer death for all cancers combined and major cancer types in children. To accomplish these goals, multiple demographic, socioeconomic, and clinical risk factors will be taken into consideration at one time through a latent class analysis. In latent class analyses, individuals can be classified into mutually exclusive and exhaustive clusters based on the pattern of their demographic, socioeconomic, and clinical characteristics that are associated with cancer death. We will use data from the Surveillance, Epidemiology, and End Results (SEER) 18 program that includes ~30% of U.S. cancer diagnoses. At the conclusion of this project, it is expected that we will have obtained a more complete picture of the interrelationships of demographic, socioeconomic, and clinical characteristics that impact risk for cancer death in U.S. children. A poster and/or manuscript of the results from this study will be prepared and submitted for presentation and/or publication. It is anticipated that the results from this project could lead to better identification of children who are at particularly high risk of cancer death due to combinations of demographic, socioeconomic, and clinical factors. This will allow for the development of more tailored interventions to reduce cancer death in U.S. children.
