Clinical Investigation
Outcomes, Health Policy, and Managed Care
Adding socioeconomic status to Framingham scoring to reduce disparities in coronary risk assessment

https://doi.org/10.1016/j.ahj.2009.03.019Get rights and content

Background

The purpose of the study was to examine the potential of adding socioeconomic status (SES) to Framingham Risk Scoring (FRS) to improve coronary heart disease (CHD) prediction by SES.

Methods

We assessed the effect of measures of SES (<12 years of education or low income) on model discrimination and calibration when added to FRS in a prospective cohort, Atherosclerosis Risk in Communities. We validated use of this model in a second cohort, the National Health and Nutritional Examination Survey linked to the National Death Index.

Results

Based on FRS alone, persons of higher and lower SES had a predicted CHD risk of 3.7% and 3.9%, respectively, compared with observed risks of 3.2% and 5.6%. Adding SES to a model with FRS improved calibration with predicted risk estimates of 3.1% and 5.2% for those with higher and lower SES, mitigating the discrepancy between predicted and observed CHD events for low-SES persons. Model discrimination (area under the receiver operator curve) was not significantly affected, and consistent findings were observed in the validation sample. Inclusion of SES in the model resulted in upgrading of risk classification for 15.1% of low-SES participants (95% CI 13.9-29.4%).

Conclusions

Standard FRS underestimates CHD risk for those at low SES; treatment decisions ignoring SES may exacerbate SES disparities. Adding SES to CHD risk assessment reduces this bias.

Section snippets

Samples

ARIC, our derivation sample, is an ongoing, prospective epidemiologic study conducted in 4 US communities. It was designed to examine the etiology and natural history of atherosclerosis and related diseases by race, sex, and location.8 The ARIC cohort component was initiated in 1987. Each ARIC field center randomly selected and recruited a sample of approximately 4,000 individuals aged 45 to 64 years from a defined population in 1 of 4 communities. Participants received extensive examination,

Results

There were 15,732 men and women included in ARIC. We excluded 2,324 (15%) subjects with prevalent CHD or diabetes (considered a CHD equivalent with >20% risk of CHD4) at baseline, leaving 13,408 persons. Next, we excluded 103 (<1%) subjects who were missing information needed to calculate the FRS and 743 (5%) subjects who were missing SES information (primarily income), leaving 12,562 subjects available for analysis. Based on the categories used, 26% of the eligible sample was in the lower-SES

Discussion

These findings demonstrate that standard FRS underestimates CHD risk for lower-SES US adults whether assessed by CHD events or CHD deaths. National Cholesterol Education Program guidelines for cholesterol treatment recommend FRS for those at intermediate risk.4 Application of these guidelines likely results in undertreatment, including use of behavioral and pharmacologic interventions, for those at lower SES due to poor calibration of FRS for this population.

Notably, we found that bias in CHD

Acknowledgements

Sean Meldrum, MS, who died unexpectedly before submission, conducted many of the analysis. There were no other contributors besides the named authors.

References (39)

  • FiscellaK. et al.

    Socioeconomic status and coronary heart disease risk prediction

    JAMA

    (2008)
  • BrindleP.M. et al.

    The accuracy of the Framingham risk-score in different socioeconomic groups: a prospective study

    Br J Gen Pract

    (2005)
  • The ARIC investigators

    The Atherosclerosis Risk in Communities (ARIC) Study: design and objectives

    Am J Epidemiol

    (1989)
  • EzzatiT.M. et al.

    Sample design: Third National Health and Nutrition Examination Survey

    Vital Health Stat

    (1992)
  • Centers for Disease Control and Prevention

    The Third National Health and Nutrition Examination Survey (NHANES III 1988-94) reference manuals and reports [CD-rom]

    (2005)
  • U.S.Census Bureau

    Poverty thresholds

  • StampferM.J. et al.

    Test of the National Death Index

    Am J Epidemiol

    (1984)
  • WentworthD.N. et al.

    An evaluation of the Social Security Administration master beneficiary record file and the National Death Index in the ascertainment of vital status

    Am J Public Health

    (1983)
  • WilliamsB.C. et al.

    The accuracy of the National Death Index when personal identifiers other than Social Security number are used

    Am J Public Health

    (1992)
  • Cited by (85)

    View all citing articles on Scopus
    View full text