We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
2015 - In adults without CV disease, most prediction scores overestimated risk for atherosclerotic CV events.
- Authors
Navar, Ann Marie; Pencina, Michael
- Abstract
Methods Design Prospective, population-based cohort study (Multi-Ethnic Study of Atherosclerosis [MESA]), with a median 10.2 years of follow-up. Setting 6 communities in the USA. Patients 4227 patients 50 to 74 years of age (mean age 62 y, 54% women) who self-identified as white, African-American, Hispanic, or Chinese and had no CVD or diabetes at baseline. Description of prediction guides American Heart Association and American College of Cardiology ASCVD (AHA-ACC-ASCVD) score, Framingham Risk Score for CVD (FRS-CVD), FRS for coronary heart disease (FRS-CHD), Adult Treatment Panel III (ATPIII) FRS-CHD, and Reynolds Risk Score (RRS). All scores were based on age, total and high-density lipoprotein cholesterol levels, blood pressure, smoking status, and sex; 4 included diabetes status (FRS-CHD, FRS-CVD, RRS, AHA-ACC-ASCVD); 3 included use of antihypertensive drugs (FRS-CVD, ATPIII-FRS-CHD, AHA-ACC-ASCVD); 1 included ethnicity (AHA-ACC-ASCVD); and 1 included family history, high-sensitivity C-reactive protein level, and hemoglobin A1c level (RRS). Outcome ASCVD composite outcome was score-specific and included myocardial infarction and CHD death (all risk scores); stroke (FRS-CVD, RRS, AHA-ACC-ASCVD); angina (FRS-CHD, FRS-CVD); coronary insufficiency (FRS-CHD); coronary revascularization (RRS); and transient ischemic attack, peripheral vascular disease, and heart failure (FRS-CVD). Events were identified by self-report during telephone interviews done every 9 to 12 months, confirmed by medical records, and adjudicated by physicians. Main results Observed and predicted ASCVD event rates for each risk score at 10 years are shown in the Table. In men, discordance between predicted and observed event rates ranged from 9% with RRS (11% vs 10%) to 154% with ATPIII-FRS-CHD (11% vs 4.4%). In women, discordance ranged from -21% with RRS (4.4% vs 5.6%) to 67% with AHA-ACC-ASCVD (6.8% vs 4.1%). C-statistics for the risk scores, representing ability to discriminate between patients with and those without an event, are shown in the Table. C-statistics were 0.69 to 0.71 in men and 0.60 to 0.72 in women. Conclusion In community-dwelling adults without cardiovascular disease, most prediction scores overestimated risk for atherosclerotic cardiovascular events; 1 score underestimated risk in women.
- Subjects
UNITED States; C-reactive protein; CARDIOVASCULAR diseases; CARDIOVASCULAR diseases risk factors; CORONARY disease; ETHNIC groups; GLYCOSYLATED hemoglobin; LIPIDS; LONGITUDINAL method; MEDICAL protocols; RISK assessment; INDEPENDENT living; RETROSPECTIVE studies; FAMILY history (Medicine); DESCRIPTIVE statistics
- Publication
ACP Journal Club, 2015, Vol 163, Issue 2, p12
- ISSN
1056-8751
- Publication type
Article