Purpose: The SCORE risk estimation system is widely used across Europe to estimate risk of fatal cardiovascular disease
(CVD) events in primary prevention efforts. However, there remains a need to examine
a broader endpoint, to make comparisons with other systems. This study
describes the derivation and validation of a new SCORE risk estimation system
using a broader endpoint of combined fatal and non fatal CVD events.
Methods: Data were accessed from the FINRISK cohort study
(recruitment years 1982 and 1987). Persons with complete risk factor data were
retained for analysis (n=13 507). The primary analysis method was a Cox
proportional hazards regression, with age as the time variable; sex, smoking, systolic
blood pressure and cholesterol as the independent variables; and first
occurrence of any CVD event (fatal or non fatal) as the dependent variable.
Validation studies examined discrimination, calibration and measures of global
fit, in both FINRISK (internal validation) and in an independent cohort, the
British Regional Heart Study (BRHS).
Results: SCORE-type charts with estimates of 10-year risk of fatal
and non fatal CVD events by risk factor status were derived. Internal
validation showed good discrimination: area under the receiver operator
characteristic curve (ROC) was 0.78 for the cholesterol based model. Whilst the
Hosmer Lemeshow test suggested a lack of fit, calibration plots showed
appropriate callibration. External validation showed appropriate
discrimination, and external calibration improved after a recalibration to
external cohort event rates procedure.
Conclusions: A risk function for estimation of the risk of combined fatal and non
fatal CVD events was derived and validated. Strengths and limitations compared
with the original SCORE charts will be discussed.
European Heart Journal ( 2010 ) 31 ( Abstract Supplement ), 805