Population-based impact of rurality and neighborhood-level socioeconomic disadvantage on pediatric cancer mortality in Washington State.

Abstract

6537 Background: Emerging evidence suggests mortality among children with cancer differs based on socioeconomic factors. The effects of residential location in relation to these factors are not well-characterized. We examined associations of rurality and neighborhood-level socioeconomic (SE) disadvantage with mortality in pediatric patients with cancer. Methods: We conducted a retrospective cohort study using Washington State (WA) Cancer Registry data (1992-2013) linked to state birth (1974-2013) and death records (1992-2013) to identify all children born in WA diagnosed with cancer < 20 years. We defined rural residence as patient addresses within 2010 census tract level rural-urban commuting area (RUCA) codes of ≥4.0 at diagnosis. Neighborhood-level socioeconomic disadvantage was determined using 2010 census block-group level Area Deprivation Index (ADI) quintiles normalized to WA. Patient addresses within the highest ADI quintile were categorized as having SE disadvantage. children in four mutually exclusive groups was compared using Kaplan-Meier analysis, pairwise log rank testing, and Cox proportional hazard ratios (HR): non-rural with SE disadvantage, rural without SE disadvantage, rural with SE disadvantage, and non-rural without SE disadvantage (). Models were adjusted for sex, race and ethnicity, age at diagnosis, birth year, and cancer type. Results: We identified 4,417 children for analysis. Median length of follow up among survivors was 5.0 years (inter-quartile range: 1.0-11.5). SE disadvantaged and 13% were rural. Pairwise log rank tests showed mortality differences among children living in rural, SE disadvantaged, or rural and SE disadvantaged neighborhoods when compared with children without either factor (individual p-values all < 0.005); no other differences were noted. Relative to children in non-rural areas without SE disadvantage, the mortality HR for those in non-rural areas with SE disadvantage was 1.68 (95% confidence intervals [CI] 1.37-2.07). The HR for children in rural areas without SE disadvantage was 1.59 (95% CI 1.19-2.12). The HR for children in rural areas with SE disadvantage was 1.56 (95% CI 1.20-2.03). In sub-analyses, associations for rurality and SE disadvantage remained significant for leukemia mortality, but CNS and solid tumor mortality was only associated with SE disadvantage (but not rural) status. Acute lymphoblastic leukemia mortality was associated with rural (but not SE disadvantage) status. Conclusions: Children with cancer living in socioeconomically disadvantaged and/or rural neighborhoods at diagnosis had higher mortality relative to those in non-rural areas with lower neighborhood deprivation. Associations varied by disease type. The individual effects of SE disadvantage and rurality suggest that interventions should be designed to target both factors.