Introduction:

It is advisable to validate and compare clinical prediction models because they can perform differently according to the ethnic, socio-demographic, genetic, cultural and idiosyncratic characteristics of populations.

Methodology:

A validation and comparison study of prognostic scales in a prospective cohort of patients over the age of 18 with a diagnosis of acute coronary syndrome with no ST-segment elevation.

Results:

five hundred and seven patients were included in the study. The mean age was 65±11, 54.4% were aged over 65, 55.6% were males, 43.8% had at least three coronary risk factors, and the type of coronary syndrome was unstable angina in 52.3% and myocardial infarction with no ST-segment elevation in 47.7%. Between the different risk groups there is a contrast in the distribution of hospital mortality (n=21, 41%) due to any cause, with significant differences for both scores (TIMI p=0.00001; GRACE p=0.0016); likewise for the outpatient GRACE (p=0.00001). The models were appropriately calibrated (Hosmer Lemeshow >0.05). Discrimination of isolated outcome of death was good for both scales in an inpatient setting (AUC-ROC TIMI 0.75 vs GRACE 0.79, p=0.37), with significant differences at 30 days (AUC-ROC TIMI 0.71 vs GRACE 0.85, p=0.0049) and at 6 months (AUC-ROC TIMI 0.75 vs GRACE 0.84, p=0.0194) in favour of the GRACE scale.

Discussion:

Both scales perform similarly in an inpatient setting. The GRACE model has greater predictive capability at one month and at 6 months.

        · abstract in Spanish     · text in Spanish     · Spanish ( pdf )