Are Racial and Ethnic Disparities in Mortality from Acute Leukemia Due to Socioeconomic Status Factors? Data from the Surveillance Epidemiology and End Results Database Linked to the National Longitudinal Mortality Study

Background: Previously, we demonstrated mortality disparities for minorities with Acute Leukemia (AL) despite favorable demographic and genetic prognostic factors. We also showed that differences in treatment by race/ethnicity explained a large component of this disparity. However, due to the limitations of the Surveillance Epidemiology and End Results (SEER) stand-alone database, we could not explore the association between AL mortality and socioeconomic status (SES) factors. The purpose of the current study is to determine how SES impacts racial/ethnic differences in AL mortality using an expanded SEER-SES linked dataset. Because SES may influence the receipt of high quality care, we hypothesize that SES factors will explain some proportion of AL mortality disparities.

Methods: Patients with acute lymphocytic (ALL) and acute myeloid (AML) leukemia were identified in SEER and linked to the National Longitudinal Mortality Study (NLMS). The NLMS contains patient-level SES factors collected by in-person and telephone interview surveys for a random sample in the United States Census Bureau Current Population Surveys (CPS) from the years 1979 to 2011. The SEER-NLMS linkage includes detailed cancer information (date of diagnosis, type of cancer, cause of death) and detailed SES factors (marital status, education, income, home ownership, occupation, insurance status) as well as individual demographic factors (age, gender, race/ethnicity). Cox proportional hazard models were built to estimate the hazard of mortality by race/ethnicity after consideration of individual, clinical, and SES factors.

Results: A total of 621 patients were diagnosed with ALL and AML during the study period. Thirty-six records were excluded due to missing data leaving 124 (21%) patients with ALL and 461 (79%) with AML in the analysis.

ALL: The majority of patients were non-Hispanic white (NHW) (n=73; 59%), followed by Hispanic (n=22; 18%), Asian Pacific Islander (API) (n=8; 6%), non-Hispanic black (NHB) (n=8; 6%) and unspecified/other race (n=13; 10%). Hispanic patients had a 3-fold increase in the hazard of death (HR 3.21; 95% CI (2.00-5.17)) when compared to NHW patients. Education and income decreased the hazard of death for Hispanic compared to NHW patients (HR 2.33 95% CI (1.21-4.48)). Insurance status further reduced this disparity (HR 2.02 95% CI (1.01-4.04)). Further adjustment for occupation and household ownership completely neutralized these disparities (HR 1.14; 95% CI (0.52-2.41)).

AML: The sample included 67% NHW (n=311) patients; 7% Hispanic (n =31), 5% NHB (n=22), 6% API (n=29) and 15% patients with unspecified/other race (n=68). Similar to ALL, there was a significant association between Hispanic ethnicity and increase hazard of mortality (HR 1.50; 95% CI (1.06-2.11)). There was a marked decreased hazard of mortality associated with API (HR 0.64 95% CI (0.44-0.91)) compared with NHW patients. The results in models adjusted for select SES factors demonstrated a persistent, unchanged mortality disparity for Hispanic patients.

Conclusions: SES factors explain some proportion of disparities in acute leukemia mortality, with the most impact in ALL; however, SES factors do not have as strong of an association in AML. Future interventions should be attentive to the underlying and specific factors that can reduce disparities in these diseases.