Background/Purpose

SLE patients are at increased stroke risk, but racial/ethnic variation in risk has not been examined in a population-based study. We examined risks by race/ethnicity among SLE patients in Medicaid, the US medical insurance program for the poor. We investigated whether differential loss to follow-up and variation in mortality between racial/ethnic groups influenced Cox regression model estimates.

Methods

From the Medicaid Analytic eXtract (MAX) 2000-2006, containing all billing claims for patients from 47 U.S. states and Washington D.C., we identified patients 18-65 with prevalent SLE (≥3 SLE ICD-9 codes of 710.0, >30 days apart) and/or lupus nephritis (additional >2 codes for nephritis, renal insufficiency or failure). The index date was the date when SLE or lupus nephritis definition was met. We extracted age, sex, US region, calendar year, zip code area-based socioeconomic status (SES). Baseline comorbidities and SLE-specific risk index (Ward M, J Rheum, 2000) were from ICD-9 and CPT codes until index date. Within inpatient claims, ICD-9 codes identified fatal and non-fatal, ischemic and hemorrhagic strokes (PPV 83%, Andrade SE, Pharmacoepi Drug Saf, 2012). Stroke incidence rates (IR) per 1,000 person-years with 95% CIs were calculated for each racial/ethnic group. Multivariable-adjusted Cox regression models calculated cause-specific hazard ratios (HRcs) for stroke from index date through end of follow-up, censoring for death or loss to Medicaid follow-up, adjusting for covariates (Table). We also used Fine and Gray proportional hazards models to calculate subdistribution HRs (HRsd), accounting for competing risks of death and loss to follow-up, adjusting for the same covariates.

Results

Of 42,221 SLE patients, 39,320 (93%) were female and 6,467 (15%) had lupus nephritis. Mean age at baseline was 38.13 (SD 12.29); 38% lived in the South, 23% in the West, 20% in the Northeast and 20% in the Midwest. Blacks represented 40%, Whites 38%, Hispanics 15%, Asian 5%, and Native Americans 2%. IRs were 10.02 (95% CI 9.44-10.64) per 1,000 person-years for all SLE patients, and 17.03 (95%CI 15.11-19.20) per 1,000 person-years for all lupus nephritis patients. After multivariable adjustment, Blacks had higher stroke risks (HRcs 1.31) than Whites. (Table) This risk remained similarly elevated in competing risks models (multivariable HRsd 1.36). Among lupus nephritis patients, stroke risks among Blacks vs. Whites were also high (multivariable HRsd 1.57). Stroke risks among other racial/ethnic groups did not significantly differ from those in White patients. 

Conclusion

Among US Medicaid SLE and lupus nephritis patients, stroke IRs were high. After adjusting for sociodemographic and clinical factors, Blacks compared to Whites with SLE had 36% increased risks and those with lupus nephritis had 57% increased risks. Accounting for competing risks did not substantially affect these estimates.