ASCVD (atherosclerotic cardiovascular disease) 2013 risk calculator

Clinical background

The decision on statin treatment in primary prevention rests not on the LDL cholesterol alone, but on absolute cardiovascular risk. The ASCVD calculator (the Pooled Cohort Equations, PCE) was introduced in the 2013 ACC/AHA guidelines precisely in order to quantify the ten-year risk of a first atherosclerotic cardiovascular event, defined as non-fatal myocardial infarction, cardiovascular death or non-fatal stroke [1]. The tool addresses the clinical question in a patient without known ASCVD and with an LDL cholesterol of 70 to 189 mg/dL, in whom the degree of absolute risk determines whether pharmacological prevention is warranted beyond lifestyle measures.

Calculating the ASCVD risk

The PCE consists of sex- and ethnicity-specific Cox proportional hazards models derived from four large American population-based cohorts collected during the 1990s [1]. The calculator covers people aged 40 to 79 years and uses the following variables: sex, ethnicity (White/other and Black/African American), age, total cholesterol, HDL cholesterol, systolic blood pressure, treatment for hypertension (yes/no), smoking (yes/no) and diabetes mellitus (yes/no).

The risk formula can be summarised as:

$$\text{Risk}_{10\text{ years}} = 1 - S_0(10)^{\exp\left(\sum \beta_i \cdot X_i - \mu\right)}$$

where $S_0(10)$ is the sex- and ethnicity-specific baseline survival at ten years, $\beta_i$ are the Cox regression coefficients for each predictor, $X_i$ are the individual's values (with natural logarithms for age, total cholesterol, HDL cholesterol and systolic blood pressure, where the blood pressure coefficient varies with hypertension treatment status), and $\mu$ is the corresponding group-specific mean of the linear predictor. The model thus estimates the ten-year risk of a first ASCVD event in a person without previous cardiovascular disease.

Interpretation in practice

Under the 2013 ACC/AHA guidelines, the estimate is operationalised through risk categories that govern the indication for a statin [1]:

The threshold at 7.5% is the central one: it defines the transition from observation to active pharmacological prevention. For patients with diabetes mellitus aged 40 to 75 years, or with an LDL cholesterol ≥ 190 mg/dL, statin treatment is indicated whatever the estimated risk, so the calculator should not be used as the basis for decision in these groups. For patients at borderline risk, additional factors such as familial hypercholesterolaemia, a raised Lp(a), coronary artery calcium scoring or chronic kidney disease may tip the assessment towards statin treatment.

Validation and performance

Since its publication in 2013, the PCE has been validated in a large number of external cohorts. A systematic review and meta-analysis of 38 studies with a total of 112 external validations found that the PCE had been validated 61 times (30 times in men, 31 in women) [2]. Overall, all three models studied (Framingham Wilson, Framingham ATP III and the PCE) overestimated the ten-year risk, with observed-to-expected (OE) ratios below 1.0. The overestimation was most marked in high-risk individuals and in European populations [2].

In an external validation using data from the SPRINT trial (n = 4,057, mean age 64.5 years), the PCE overestimated the number of ASCVD events by 128.6%: against 133 observed events over a median 3.3 years of follow-up, 304 events were predicted [3]. Calibration was poor (Hosmer–Lemeshow, p < 0.001) and discrimination moderate, with an AUC of 0.65 (95% CI 0.60 to 0.69) [3]. The overestimation was more marked in the higher risk groups, in which the observed risk did not match the predicted risk. In the meta-analysis, substantial heterogeneity in the c-statistic between studies was noted, probably reflecting differences in population characteristics [2]. Discrimination was generally better in women than in men [2].

The overestimation is attributed mainly to temporal effects: the PCE was derived from cohorts from the 1990s and reflects a higher baseline risk than is seen in modern populations with wider statin use, better revascularisation and improved risk factor control [3].

In 2024 the AHA presented the PREVENT equations as a successor to the PCE [4]. PREVENT removed ethnicity as a variable, included renal and metabolic markers, and improved calibration in contemporary populations. In a nationally representative American study (n = 7,765), PREVENT was calculated to reclassify approximately 53% of adults into a lower risk category, while only 0.4% moved upwards [4]. This would correspond to approximately 14.3 million fewer adults meeting the criteria for statin treatment, with an estimated 107,000 additional myocardial infarctions or strokes over ten years as a consequence of reduced treatment [4].

Limitations

The PCE applies only to the age group 40 to 79 years. In people under 40 or over 79 there is no validation evidence and the calculator should not be applied. With known ASCVD, an LDL cholesterol ≥ 190 mg/dL, or diabetes at ages 40 to 75, statin treatment is already indicated under guidelines, so the PCE should not be used as the basis for decision.

The ethnicity variable is a particular weakness. The coefficients for Black/African American and for White/other were derived from American cohorts and reflect specific population risk patterns. In other populations, including European and Asian ones, there are no corresponding ethnicity-specific coefficients and calibration becomes uncertain. The meta-analysis found precisely that the overestimation was particularly marked in European populations [2]. That the calculator does not include Lp(a), familial hypercholesterolaemia or coronary artery calcium scoring as variables is a further limitation, particularly at borderline risk, where these markers may settle the decision.

The systematic overestimation of risk is the most concrete pitfall in clinical use: patients in the intermediate risk category may in reality be closer to a lower risk, which can lead to statins being started in patients in whom the absolute benefit is smaller than the estimate suggests [2, 3]. Against this background the PREVENT equations were developed, but the PCE is still used in clinical practice and in many guidelines that remain in force [4].

References

1. Goff DC Jr, Lloyd-Jones DM, Bennett G et al. 2013 ACC/AHA Guideline on the Assessment of Cardiovascular Risk. Circulation. 2014. PMID: 24222018.
2. Damen JA, Pajouheshnia R, Heus P et al. Performance of the Framingham risk models and pooled cohort equations for predicting 10-year risk of cardiovascular disease: a systematic review and meta-analysis. BMC Medicine. 2019. PMID: 31189462.
3. Kuragaichi T, Kataoka Y, Miyakoshi C et al. External validation of pooled cohort equations using systolic blood pressure intervention trial data. BMC Research Notes. 2019. PMID: 31088530.
4. Diao JA, Shi I, Murthy VL et al. Projected Changes in Statin and Antihypertensive Therapy Eligibility With the AHA PREVENT Cardiovascular Risk Equations. JAMA. 2024. PMID: 39073797.