Racial/Ethnic and Geographic Variations in Long-Term Survival Among Medicare Beneficiaries After Acute Ischemic Stroke

Xin Tong, MPH; Linda Schieb, MPH; Mary G. George, MD, MSPH; Cathleen Gillespie, MS; Robert K. Merritt, MA; Quanhe Yang, PhD

Disclosures

Prev Chronic Dis. 2021;18(2):e15 

In This Article

Methods

Data Sources and Study Sample

We used Medicare's enrollment databases to generate our study cohort among Medicare fee-for-service (FFS) beneficiaries and Medicare Provider Analysis and Review (MEDPAR) data to assess overall survival among beneficiaries hospitalized with AIS from 2008 through 2012. To select the final analytical cohort we 1) identified all Medicare FFS beneficiaries aged 65 or older with 12 months continuous enrolment in Medicare parts A and B during 2007–2012; 2) identified all hospitalizations with AIS as the primary diagnosis among FFS beneficiaries from 2007 through 2012, including multiple admissions; and 3) used a 12-month or longer lookback period to identify the first AIS hospitalization. The length of lookback time varied by the years of Medicare enrollment; for example, 12 months for beneficiaries aged 66 (Medicare eligible at age 65 years), 24 months for those aged 67, and so on. Because of the 12-month or longer lookback period, our final cohort included FFS beneficiaries aged 66 or older with AIS hospitalizations from 2008 through 2012 (2007 served as lookback time). We used MEDPAR files to identify AIS, our outcome of interest. The MEDPAR files contained records for inpatient hospital stays and skilled nursing facility stays for all Medicare beneficiaries, and we used the primary diagnosis codes (International Classification of Diseases, 9th revision [ICD-9-CM][4] codes 433.01, 433.11, 433.21, 433.31, 433.81, 433.91, 434.01, 434.11, and 434.91) to identify beneficiaries with AIS. We excluded all institutional long-term stay hospitalizations. We identified 1,019,267 FFS beneficiaries aged 66 or older in our study period who had AIS. Socioeconomic status (SES) in the community, defined by the percentage below the poverty level in the county of beneficiary residence in 2008, was linked to Medicare data from the Health Resources and Services Administration Area Health Resources Files (https://data.hrsa.gov/data/download).

Statistical Methods

We examined differences in the distribution of demographic features by χ2 test for categorical variables, and t test for continuous variables. The 5-year survival was defined as the time from the date of AIS to the date of death, or the date of end of follow-up (December 31, 2017), whichever came first. We used the National Death Index linked to Medicare data available through the Centers for Medicare and Medicaid Services (CMS) to determine the date of death. We performed 5-year survival analyses and subgroup analyses by age groups (66–74, 75–84, and ≥85), sex, race and Hispanic ethnicity (non-Hispanic White, non-Hispanic Black, Hispanic, and other non-Hispanic races), and SES at the county level (quartile distribution; higher quartiles indicate higher level of poverty). We identified Charlson Comorbidity Index (CCI) conditions[5] by using secondary diagnosis codes. We examined the variations in AIS survival across the states for all beneficiaries and by race and Hispanic ethnicity. Univariate and multivariate survival analyses of 5-year survival after AIS were carried out using the Kaplan–Meier life table, and Cox proportional hazards regression analyses adjusting for age, sex, race and Hispanic ethnicity, SES, state (Model 1); and for CCI (0, 1, 2, 3, and ≥4) (Model 2). For subgroup analyses, we defined insufficient data if the total events (deaths) per analytic group were fewer than 15 during follow-up. We used SAS, version 9.4 (SAS Institute) for analyses and considered a 2-sided P value of <.05 significant. Medicare data are available from CMS, US Department of Health and Human Services, for any qualified investigator.

processing....