Background/Purpose Cardiovascular disease (CVD) is a leading cause of systemic lupus erythematosus (SLE) morbidity and mortality. However, the increased CVD risk is not completely attributable to disease activity and traditional risk factors. Poverty in SLE has been associated with increased mortality, irrespective of race. We examined how financial strain, stress, and depression in relation to more recognized CVD risk factors impact the development of CVD in a large cohort with significant numbers of African-American (AA) people with SLE.

Methods Georgians Organized Against Lupus (GOAL) is a population-based cohort of validated SLE patients in Atlanta, Georgia supported by the Centers for Disease Control and Prevention. Sociodemographic information, disease factors, CVD risk factors and social determinants of health measures were collected at baseline in 2016. Potential CVD events were identified by participant report and by matching with the Georgia Hospital Discharge Database for CVD-related codes. Associated medical records were reviewed by study physicians and adjudicated for CVD events (myocardial infarction, angina, transient ischemic attack, thrombotic stroke, and/or peripheral vascular disease) using validated algorithms. After participants with prevalent CVD events through 2016 were identified, CVD-naïve participants were surveilled for incident CVD events from 2017–2021. We analyzed a combined measure of financial strain (FS) by Conger et al. Univariate and multivariate Cox regression analyses were used to evaluate the ability of FS to predict incident CVD. Bootstrap bagging was utilized to identify the relative stability and reliability of predictors of CVD. Time-dependent receiver operating characteristic curves were used to access the discriminative performance of the final prediction model.

Results Out of 780 participants, 179 (23%) were adjudicated as having had prevalent CVD events through 2016. Two individuals died before reaching 2017. The majority of the remaining 599 CVD-naïve participants were AA (472, 78.8%) or White (115, 19.2%). In 2017–2021, 113 (18.9%) participants were adjudicated as having had an incident CVD event, with a mean time to event of 27.6 months (SD 16.9, range 0.7–60).

There were 33 deaths: 20 after incident CVD, 13 without CVD (excluded in the Cox regression). Those with incident CVD were older, had longer duration of SLE, were less employed, and had more Medicare/Medicaid insurance than those without CVD. They also had more SLE activity and organ damage and more traditional CVD risk factors. (table 1) Univariable Cox regression analysis informed the variable set for the multivariable Cox regression (tables 2 and 3). Predictive accuracy of the final model was good: time-dependent area under the curve values ~0.80. The most reliable predictors of incident CVD were stress, obesity, age, disease duration, and hypertension, all with reliability over 93% (table 4).

Conclusions The burden of CVD remains very high in this SLE cohort. As expected, age, disease activity, organ damage, disease duration, hypertension, obesity, and physical inactivity independently predicted incident CVD in SLE. However, educational attainment, stress, and FS also predicted incident CVD. In terms of relative reliability of risk factors to predict incident CVD, stress had the highest reliability (99.7%), followed by obesity. Of risk factors with a reliability of >50%, half (7/14) can be considered non-traditional CVD risk factors and more than half (8/14) were significantly more common in AA compared to White participants. Risk factors with the lowest reliability include hypercholesterolemia, hydroxychloroquine use, smoking, and depression. The disproportionate burden of negative social determinants of health in AA communities may be a significant driver of CVD and other disparities described in SLE. Further research into related causal pathways, mitigating factors, and biologic mechanisms is needed.

Lay Summary The burden of cardiovascular disease (CVD) remains very high in this systemic lupus erythematosus (SLE) cohort. As expected, age, disease activity, organ damage, disease duration, hypertension, obesity, and physical inactivity independently predicted incident CVD in SLE. However, educational attainment, stress, and FS also predicted incident CVD. In terms of relative reliability of risk factors to predict incident CVD, stress had the highest reliability (99.7%), followed by obesity. At least half of the most reliable predictors of CVD were considered non-traditional risk factors and were more common in AA participants, which could explain a main source of health disparities in SLE. Further research into related causal pathways, mitigating factors, and biologic mechanisms is needed.