QRISK: New CVD calculator better reflects risk of UK population

July 11, 2007

Nottingham, UK - British scientists have developed a new cardiovascular disease risk score for the UK population that they say better reflects the true risk of people living there [1]. Most doctors in the UK still use the Framingham model, but the researchers say this massively overpredicts the risk and to some extent identifies the wrong people for treatment.

Dr Julia Hippisley-Cox (University of Nottingham, UK) and colleagues reveal their new risk calculator, QRISK, in a report published online July 5, 2007 in BMJ. Hippisley-Cox—a GP and professor of clinical epidemiology and clinical practice—told heart wire that QRISK "helps separate out those who are genuinely at risk," and it incorporates a number of key things that Framingham did not—measures of social deprivation, body-mass index (BMI), existing antihypertensive treatment, and a family history of heart disease. "I think this stacks up, because I've written it, but we are interested to see what kind of peer review we get," she added.

Framingham researcher Dr Dan Levy (National Institutes of Health, Bethesda, MD) said: "I applaud the QRISK team for working so hard to develop better tools that can help UK physicians deliver better healthcare." However, he points out a couple of shortcomings, including the fact that QRISK was not compared with the most contemporary Framingham model available.

Framingham overestimates risk in UK by 1.5 million

Hippisley-Cox and colleagues set out to derive and validate a new cardiovascular risk score for the UK and test its performance against the established Framingham model used in the UK and a new score used in Scotland (ASSIGN), which does include a measure of social deprivation. They used figures from a general practice research database called QRESEARCH, which tracked the progress of 1.28 million healthy men and women registered at 318 general practices over a period of 12 years to April 2007, recording first diagnosis of cardiovascular disease. All the participants were aged between 35 and 74 at the start of the study.

This is "the largest study to have ever been undertaken and the first time that routine data in a UK general practice population have been used rather than an observational study in a predefined cohort," the researchers note.

They calculate that Framingham overpredicted cardiovascular disease risk—QRISK predicted 9% of patients to be at high risk compared with 13% for the Framingham equation and 14% for ASSIGN. QRISK identified 3.2 million patients at high risk in 2005, compared with 4.7 million from Framingham and 5.1 million from ASSIGN, they say.

QRISK also suggests that the risk in women is underestimated by Framingham—with 34% of women being at high risk with QRISK compared with 24% using Framingham. In contrast, Framingham overestimates the risk in men, with 73% of men aged 64 to 75 at high risk according to QRISK compared with 86% using the US data.

Hippisley-Cox told heart wire that GPs in the UK are currently missing the right people for treatment and that if QRISK were used it would better direct therapy toward those who most need it—people in deprived areas and women.

But did they use the right comparator?

Levy says one of the findings by Hippisley-Cox et al—that the ratio of total/HDL cholesterol was not predictive of cardiovascular disease risk in QRISK—was "counterintuitive and requires further explanation." Hippisley-Cox explained that the effect of cholesterol is reduced by inclusion of BMI, deprivation, and family history. "No other equation has all these in together, so we can't say that it is unprecedented."

She told heart wire that the Framingham model used in their paper was the 1990 Anderson equation [2]. "We recognize that other Framingham models have been produced, but we wanted to compare [QRISK] with the one used in clinical practice in the UK."

Levy says, "They did not use what I would consider to be the best Framingham risk model or the most contemporary. They have to understand that the old Framingham score [that they compared QRISK with] overestimates risk because when it was published the risk was higher than it is today. The whole process of developing risk models has to be a steadily evolving one."

He said that the most recent Framingham equation he is referring to—the ATP III model [3]—also includes antihypertensive treatment in the mix, unlike the model that is in use in clinical practice in the UK. However, he conceded that it does not include BMI, family history of heart disease, or measures of social deprivation.

How important is it to include BMI, family history, and deprivation?

With regard to BMI, Levy says a conscious decision was taken not to include it in Framingham models. "We know that obesity is an important contributor to elevated blood pressure, dyslipidemia, diabetes, etc, but part of the relationship of obesity to risk is mediated by these other risk factors. Obesity acts to a large extent through these, and by including BMI you would dilute the apparent effect of these other risk factors, for which we have irrefutable evidence of benefit from their modification."

"We agree that the independent contribution of BMI is weaker once blood pressure and serum cholesterol are accounted for," concedes Hippisley-Cox. "However, this tool is intended to inform management decisions, and BMI is nevertheless a major risk factor that is tangible for both clinician and patient."

With regard to family history of heart disease, Levy says the story is similar, "with a concern that adding family risk would dilute the apparent relationship of known modifiable risk factors with outcome."

Nevertheless, he says, Framingham researchers "have published widely on family history as a component that improves upon risk prediction, and it clearly implies an increased risk." They have also produced Framingham equations that do include family history, although it was not included in the most contemporary one he refers to nor in the one used as a comparator by the QRISK team.

Hippisley-Cox says: "There is substantial evidence that family history of premature coronary heart disease remains an independent risk factor once age, sex, blood pressure, cholesterol, and smoking are accounted for. The Framingham investigators identified such an effect in both parents and siblings."

And finally, on the issue of social deprivation, Levy concedes that this is "something we have not included in our risk models. I'm not sure we have collected it in the same systematic manner [as has occurred in the UK]."

Validation required ahead of major change in national policy in UK

The QRISK researchers point out that their validation was performed in a population similar to the one from which the algorithm was derived, so it potentially has a "home advantage" and therefore needs to be further tested in other populations. Hippisley-Cox told heart wire that they are currently in the process of performing a second validation using GP practices that utilize a different clinical computer system. "This will be a more severe test of the performance of QRISK," she noted.

These moves come at a time of major change in national policy on the identification of patients at high risk of cardiovascular disease in the UK. The National Institute for Health and Clinical Excellence (NICE) is in the middle of drafting new guidance to recommend that adults with a 20% or greater risk of developing cardiovascular disease over the next 10 years be offered statins, and the hope is that general practitioners—most of whom currently use the 1990 Framingham model to predict risk—will eventually use QRISK to pinpoint patients who should be targeted for medication and advice on lifestyle changes.

No disclosures were declared by either Hippisley-Cox or Levy in terms of pharmaceutical funding. Competing interests with regard to UK NICE and QRESEARCH—a not-for-profit organization—are listed in the paper.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.