Author + information
- Gmerice Hammond,
- Kenton Johnston and
- Karen E. Joynt Maddox
Social determinants of health (SDOH) are powerful predictors of cardiovascular (CV) outcomes, but are absent from guideline-recommended CV risk prediction models. It is unknown if including SDOH in clinical risk prediction models improves model accuracy.
We used the 2016 Medicare Current Beneficiary Survey to calculate observed to expected (O:E) ratios for CV hospitalization, mortality, and annual cost of care under three risk models: 1) age + gender 2) Model 1 + clinical comorbidities (similar to current risk prediction models), and 3) Model 2 + a robust panel of SDOH, including income, education, housing, and food insecurity.
There were 8,220 (weighted N=28,661,897) patients in our sample. Poor beneficiaries had higher CV hospitalization (9.2 vs 6.1 per 100, p=0.024), mortality (6.1% vs 3.4%, p%<0.001), and cost ($12,791 vs $8,412, p%<0.001), than non-poor beneficiaries. Blacks/Hispanics had higher CV hospitalization (12.0 vs 6.0 per 100, p%<0.001) and cost ($11,047 vs. $9,090, p=0.03), but similar mortality (3.4% vs 4.0%, p=0.36), compared to whites. Models including age, gender, and comorbidities under-predicted mortality and costs for poor beneficiaries, and CV hospitalization for minorities; adding SDOH improved prediction (Table).
Adjustment for SDOH improves the accuracy of risk models among poor and racial/ethnic minorities, and could improve risk prediction among vulnerable populations and guide more optimal use of prevention strategies.
Moderated Poster Contributions
Prevention Moderated Poster Theater, Posters, Hall A
Saturday, March 28, 2020, 10:15 a.m.-10:25 a.m.
Session Title: How Can a “Learning Health Care System” Inform Cardiovascular Prevention Community?
Abstract Category: 32. Prevention: Clinical
Presentation Number: 1006-05
- 2020 American College of Cardiology Foundation