ASSIGN, QRISK, and validation

We challenge the recent QRISK validation and editorial concluding that QRISK is the cardiovascular risk score for the United Kingdom.1 2

ASSIGN, QRISK’s precursor, was launched in Scotland before QRISK appeared.3 Predicting that scores omitting social deprivation (socioeconomic status) as a risk factor could exacerbate social gradients in disease, we developed ASSIGN to include it. ASSIGN was adopted without external validation because it correlated highly with the gold standard Framingham score. Discriminating rather better, even after adjustment for self-testing bias, it removed Framingham’s social inequity.

Subsequent to ASSIGN’s launch, QRISK authors told us that they were developing their own score. Our offer of collaborative comparison was not accepted, and QRISK coefficients were kept secret after its launch. The initial partisan publication, however, did show that ASSIGN discriminated better than Framingham in the QRESEARCH database where QRISK originated.4

We have not seen how QRISK deals with social deprivation in analyses similar to ours—possibly because we have full 10 year follow-up of our cohort. QRESEARCH and the validation THIN database do not. Both these databases are missing 70% of data on lipids, and probably more on family history of cardiovascular disease. There is a surprisingly flat social gradient in cardiovascular event rates3 4 5 and in men at high risk with the Framingham score in QRESEARCH (figure⇓).4

Social gradient by deprivation fifth in men for mortality under 75 from coronary heart disease in England and Scotland, and cardiovascular event rates in QRESEARCH and Scottish heart health extended cohort studies

• Download figure
• Open in new tab
• Download powerpoint

ASSIGN and QRISK scores may serve different priorities with different advantages. We welcome debate and collaborative comparisons, but question whether QRISK is the preferred score for the United Kingdom when ASSIGN is already adopted in Scotland (www.assign-score.com).