Area-Based Socioeconomic Status and Pediatric Allogeneic Hematopoietic Stem Cell Transplantation Outcomes: A CIBMTR Analysis

Introduction: Outcome disparities related to race and area-based socioeconomic status (SES) following allogeneic hematopoietic stem cell transplantation (allo-HCT) have been identified in adult patients [Baker et al. 2009]. The relationship between SES and outcomes in pediatric allo-HCT has not been previously described. Among a large cohort of pediatric allo-HCT recipients we sought to determine the impact of area-based poverty on 5-year outcomes of overall survival (OS), acute and chronic graft-versus-host-disease (aGVHD, cGVHD); as well as the short-term outcome of infection through day 100 .

Methods: We utilized the Center for International Blood and Marrow Transplant Research (CIBMTR) database to examine the association of sociodemographic variables with outcomes in two cohorts of pediatric transplant recipients aged <=18 years who received allo-HCT at U.S. centers between 2006-2015. Cohort 1 (malignant) included 2053 children who received myeloablative conditioning for any malignancy. Cohort 2 (non-malignant) included 1696 children who received myeloablative or reduced-intensity conditioning for any non-malignant disease. Zip codes of child's residence were categorized as high-poverty area (>=20% households living below 100% federal poverty level (FPL)) versus low-poverty area (<20% households below 100% FPL) by linkage to U.S. Census data [Krieger et al. 2002]. Individual-level sociodemographic variables including insurance (Medicaid-only vs Other), race (Caucasian vs Black vs Other) and ethnicity (non-Hispanic vs Hispanic) were included as covariates. Cox regression was used to examine the effect of patient-related (age, performance status, insurance, race, and ethnicity), disease-related (disease type and status), and HCT-related (donor/graft type, CMV status, stem cell source, HLA match, donor age and gender, conditioning regimen and intensity, GVHD prophylaxis, year of HCT) variables on the outcomes of interest between the two area-based poverty groups.

Results: Fifteen percent (N=299) of children in Cohort 1 lived in a high-poverty area; 35% (N=711) were insured by Medicaid-only; 11% (N=227) were African-American and 20% (N=417) Hispanic. Median follow-up of survivors was 74 months. In multivariable analysis, there was no association between area-based poverty and OS; however, OS was inferior in children with Medicaid-only insurance compared to those with private insurance (HR 1.22 (95% CI 1.06-1.40), p=0.0037) and in Black children compared to Caucasian (HR 2.02 (95% CI 1.10-3.73), p=0.0234). For the secondary outcomes of aGVHD, cGVHD or infection through day 100, there were no associations between area-based poverty, insurance, race, or ethnicity in multivariable analysis. To further explore the independent association of insurance with OS, we performed an ad hoc univariate analysis that demonstrated that insurance-related differences in OS for malignant disease appear to be driven by differences in treatment-related mortality (TRM) (5-year TRM: Medicaid-only 25% (95% CI 22-28) versus 18% (95% CI 16-21) other).

Thirteen percent (N=228) of children in Cohort 2 lived in a high-poverty zip code; 35% were insured by Medicaid-only (N=597); 20% (N=332) were African-American and 20% (N=344) Hispanic. Median follow-up of survivors was 74 months. In multivariable analyses, there was no association between area-based poverty, insurance, race or ethnicity and any outcome.

Conclusion: Area-based poverty is not associated with disparate outcomes in pediatric allo-HCT for malignant or non-malignant disease. In the setting of malignant disease, insurance-a household-level measure of socioeconomic status-and Black race are independently associated with inferior OS. These results suggest that future prospective investigation of more refined measures of household-level socioeconomic status may identify risk-factors for treatment-related mortality in this population.