Association of Neighbourhood-Level Socioeconomic Deprivation with Mortality Among Patients with Sickle Cell Disease

Introduction: While treatment advances for Sickle Cell Disease (SCD) have improved outcomes for most patients, some studies (but not all) have demonstrated disparities related to socioeconomic conditions. Few studies, however, have examined associations between sociodemographic factors and clinical outcomes. The objective of this analysis was to assess the association of neighborhood deprivation index (NDI) with mortality among SCD patients using electronic health record (EHR) data from a diverse health system.

Methods: Our study population included SCD patients (identified by diagnosis codes) aged 18 years and older enrolled at Kaiser Permanente (KP) Mid-Atlantic States between 01/01/2004 – 12/31/2023. The index date was defined as the date of earliest enrollment during the study period. NDI (derived from geocoded addresses and multiple census tract level socioeconomic variables) was extracted at index date or the earliest available after index. We followed patients from their index date until death, disenrollment, or the end of the study period. Covariates included age, sex, insurance, comorbidities, and medication use. Kaplan-Meier curves described survival by quartiles of NDI. Extended Cox proportional hazard models assessed hazard ratios (HR) for the adjusted mortality risk (using time-varying covariates) associated with the highest vs. lowest quartile of NDI.

Results: Among 1,083 SCD patients, 685 (63.3%) were female, 894 (82.6%) were commercially insured, 437 (40.4%) had hypertension, 156 (14.4%), had a venous thromboembolism, 223 (20.6%) had heart disease, 137 (12.7%) had acute chest syndrome, 105 (9.7%) had a hemorrhagic cerebrovascular event, 271 (25.0%) had chronic kidney disease, and 238 (22.0%) were taking hydroxyurea. Covariates did not significantly differ across quartiles of NDI, however, the time to development of comorbidities was shorter among those living in the highest (most deprived) vs lowest quartile of NDI for virtually all conditions. The mean age at death was 60.7 years and 68.2 years for the highest and lowest quartiles of NDI, respectively. Survival was worst among those living in the most deprived neighborhoods with survival curves showing separation after 8 years of follow-up. (Confidence intervals for survival curves overlapped). In the age- and sex-adjusted model, patients with the highest NDI had a 56% increased hazard of death compared to those in the lowest NDI (HR: 1.56; 95%CI: 1.01, 2.42). In the fully adjusted model, the HR estimate was reduced to 1.27 (95%CI: 0.80, 2.01) and was no longer significant. Both models demonstrated trends of higher magnitudes of HRs for death associated with increasing quartiles of NDI.

Conclusions: In this study of adult SCD patients, we demonstrated that patients living in the most deprived neighborhoods are dying earlier than those in less deprived areas. A significantly higher risk of death was associated with neighborhood deprivation in age- and sex-adjusted models. In full models, the risk estimate moved towards the null, however, since comorbid conditions may be mediators, their adjustment would not be appropriate in the causal estimation of the total effect of NDI on mortality. Our study included a robust measure of socioeconomic status, extensive longitudinal EHR data, and was conducted within a health system where patients had equal access to care. Future studies should explore the impact of social conditions on earlier clinical endpoints to address inequities in care that could potentially extend life among persons with SCD.