In search of the Holy Grail
Articles tended to grow in statistical complexity, becoming so
increasingly difficult to be understood by general practitioners for whom
they are intended.
Patients were classified into high and low risk, based on treatment
thresholds of 20%. Although predicted risk varies across a continuum, risk
groups were created according to clinical decisions requirements.
QRISK2 was slightly superior to NICE Framingham in estimating 10-year
risk of a cardiovascular event (table 3), but the differences between the
two do not seem to be clinically significant. AUROC statistics were
similar in both risk scores, in men and women; then the overemphasized
superior incidence rate of cardiovascular events (per 1000 person years)
in the high risk group in QRISK2 was counterbalanced by a lower capacity
of capture of cardiovascular events. It had been as, by increasing the
specificity, sensibility had to be sacrificed; a question of tradeoff
between opposing interests (of gains and losses).
But, as fig 2 stresses in a suggestive way, in both models the
discrimination performance remained modest.
From what I could perceive, it seems then that has not been
discovered yet this time the mother lode gold mine
Competing interests: No competing interests