Validity of a long-term cardiovascular disease risk prediction equation for low-incidence populations: The CAMUNI-MATISS Cohorts Collaboration Study

    loading  Checking for direct PDF access through Ovid

Abstract

Background:

Before introducing long-term cardiovascular disease (CVD) risk models in clinical practice, their external validity should be investigated. We assessed the validity of the CArdiovascular Monitoring Unit in Northern Italy (CAMUNI) 20-year risk score, developed in Northern Italy, and published previously, when applied to a population with different risk factors distribution and event incidence.

Methods:

The validation sample consisted of 5307 35-69 year-old subjects (2418 men) enrolled in Central Italy during the 1980s (Malattia ATerosclerotica Istituto Superiore di Sanità (MATISS) study). Baseline risk factor assessment and follow-up procedures, including MONICA definition of acute events, followed a shared protocol with the derivation cohorts. We estimated model calibration and discrimination (area under the ROC curve, AUC) in the validation set; as well as the net benefit of using the CAMUNI risk score as second-level screening in subjects at different levels of short-term risk.

Results:

The 20-year risk of event was 14% in men and 7% in women. Model calibration was satisfactory, and the strength of the association between predictors and the endpoint was the same as in the derivation population. The AUC was 0.734 (men) and 0.802 (women). The net benefit of the CAMUNI score was 3.9 (95% confidence interval: 2.1-5.7) and 2.9 (1.7-4.3) in men and women at low 10-year risk, respectively. Among subjects at high short-term risk, a significant net benefit of 9.8 was observed in men only. A pooled CAMUNI-MATISS risk score is provided.

Conclusions:

In this low-incidence European population, long-term CVD prediction through the CAMUNI risk score is accurate and it has the potential to improve current primary prevention strategies based on short-term risk scores alone.

Related Topics

    loading  Loading Related Articles