Septic Shock and Septic Shock Mortality in Relation to Social Disorganization Index Among Children Being Treated for De Novo Acute Myeloid Leukemia on COG Clinical Trials

Introduction: Acute myeloid leukemia (AML) requires intensive chemotherapy that increases the risk of life-threatening complications such as septic shock (SS). The relationship between social determinants of health (SDOH) and risk of developing SS in children with AML is not well understood. We previously identified associations between race-ethnicity and public insurance and risk of SS in children receiving treatment for acute leukemias (PBC 2019 10.1002/pbc.27713 #2024). To explore this further, we examined associations between the Social Disorganization Index (SDI), an area-based composite measure of SDOH domains, and time to SS and SS-associated mortality (SS-mortality). We hypothesized that higher social disorganization is associated with decreased SS-free survival and higher SS-mortality.

Methods: We used a previously assembled cohort of children with de novo AML treated on two Children’s Oncology Group (COG) clinical trials merged with daily resource utilization data from the Pediatric Health Information System database (PHIS; PLOS One 2015 10:e0143480). We used the SDI (Health & Place 2017 10.1016/j.healthplace.2016.12.004) to characterize patients’ relative exposure to adversity across SDOH domains. The SDI was calculated using the American Community Survey 5-year estimates and the US Census Bureau’s 5-digit national ZIP Code Tabulation Area-level data and linked to patients’ 5-digit ZIP code recorded on first admission. SDI is calculated as the sum of the following proportions: overall unemployment, households receiving public assistance, low-income households at <100% and at 100–149% poverty level, high school dropouts, female-headed households, renter-occupied houses, and moved households within the last year. Higher SDI values indicate a more disorganized neighborhood. To define their exposure to neighborhood disorganization, patients were grouped into one of 5 categories (from least to most disorganized) based on their nationally normed SDI. We used our previously developed and validated method to detect SS events in PHIS (PBC 2019 10.1002/pbc.27713 #2024) and used COG data for patient and disease characteristics and for treatment data. Our secondary outcome, SS-mortality, was defined as death within 2 weeks of antecedent SS event (Lancet Oncol. 2015 10.1016/S1470-2045(15)00197-7). Follow-up started 7 days from first chemotherapy until observed end of therapy to capture treatment-associated SS events. Patients were censored at relapse or death for the time to SS analysis. Time to first SS was summarized using Kaplan Meier estimates. Cox regression models were used to estimate unadjusted hazard ratios (HR) and associated 95% confidence intervals (CI) of SS outcome by SDI category.

Results: We identified 701 COG-PHIS patients with AML who had at least 7 days of protocol therapy. Most patients were from SDI category 5 (highest disorganization; Table 1). Patients in higher SDI categories were more likely to be Hispanic or non-Hispanic Black and have public insurance compared to those in the lowest categories. There were 201 (28.7%) incident SS events observed during protocol therapy, which occurred a median of 78 days (IQR 34-122) from start of therapy. We found no significant difference in time to incident SS between SDI categories (log rank p=0.67) or in unadjusted HR for SS events among the SDI categories (Figure 1). Among patients who had SS events, 9 (4.4%) experienced SS-mortality, of which 8 (88.9%) were in higher disorganization categories (SDI 4 or 5). Seven (77.7%) patients who experienced SS-mortality were non-Hispanic White, 6 (66.7%) were female, and 6 (66.7%) had private insurance. SS-mortality occurred a median of 42 days (IQR 15-167) from start of therapy. Five (56%) of the 9 SS-mortality events occurred in cycle 1, 1 (11%) occurred in cycle 2, 3 (33%) occurred in either cycle 4 or 5.

Conclusion: In our unadjusted model, neighborhood disorganization was not significantly associated with time to SS. Almost 90% of SS-mortality occurred among patients exposed to higher disorganization neighborhoods (SDI 4 or 5). Over half of SS-mortality occurred with the first cycle of therapy, suggesting a need to investigate associations between the SDI, severity at presentation, and AML induction mortality (Am J Hematol. 2017 10.1002/ajh.24605). We are designing multivariable-adjusted models to address potential confounding and competing risk analyses.