David Brill
The long-established rules for cardiovascular risk assessment could be overhauled in the UK, following calls for the Framingham risk score to be abandoned in favor of a newer model.
An independent analysis found the QRISK score to be superior to Framingham on all measures – identifying a higher proportion of people who went on to develop cardiovascular disease.
The findings prompted a resounding endorsement of QRISK in a British Medical Journal editorial, recommending its routine adoption. Reports have since emerged that QRISK is under review at the UK National Institute for Health and Clinical Excellence (NICE), and is set to be integrated into software packages used by GPs.
University of Oxford statisticians validated QRISK using an independent cohort of 1.07 million patients from 274 UK general practices. There were 43,990 cardiovascular events over a median of 4.9 years’ follow-up. [BMJ 2009 Jul 7;339:b2584]
QRISK under-predicted cardiovascular risk by 12 percent (13 percent for men and 10 percent for women) but was considerably more accurate than the Anderson Framingham algorithm, which overestimated risk by 23 percent (32 percent for men and 10 percent for women).
“We believe this formula has the potential to save many thousands of lives, by helping clinicians to more accurately predict those at risk of developing cardiovascular disease – the nation’s biggest killer,” said Professor Julia Hippisley-Cox of the University of Nottingham, who led the team which developed the QRISK score.
“It will arm doctors with all the information they need to decide how best to target patients with preventative measures such as lifestyle advice and cholesterol-lowering treatments. We are delighted to receive another strong endorsement of the value of QRISK in assessing the risk of heart disease in the UK population,” she said.
QRISK includes most of the traditional risk factors seen in the Framingham equations but also includes family history, social deprivation, body mass index and use of antihypertensive treatments. It was first developed and validated in 2007, using data from 1.28 million patients from 318 UK general practices. [BMJ 2007 Jul 21;335(7611):136]
Further refinements to the model – QRISK2 – were published last year to include ethnicity and other conditions such as type 2 diabetes, hypertension, atrial fibrillation, renal disease and rheumatoid arthritis. The updated version, however, has yet to receive independent validation. [BMJ 2008 Jun 28;336(7659):1475-82]
The external validation of the original QRISK was commissioned by the UK Department of Health to compare the model against the Anderson Framingham algorithm, which is presently recommended by NICE as the basis for deciding on whether to prescribe statins.
The incidence rate of cardiovascular events was 30.5 per 1,000 person years among men who were classed as high risk by QRISK, compared to 23.7 per 1,000 person years among men classed as high risk by the Anderson Framingham score. In high-risk women, the incidence rates were 26.7 per 1,000 person years for those identified with QRISK, and 22.2 per 1,000 person years for those identified by Anderson Framingham.
Despite the improvements with QRISK, the data provide “a sobering message about the current state of cardiovascular risk prediction,” according to the authors of the accompanying editorial. Increased usage, improved data collection and further refinements to the system could however increase the accuracy of risk prediction in future, wrote Professor Rod Jackson of the University of Auckland, New Zealand, and colleagues. [BMJ 2009 Jul 7;339:b2673]
The findings prompted a resounding endorsement of QRISK in a British Medical Journal editorial, recommending its routine adoption. Reports have since emerged that QRISK is under review at the UK National Institute for Health and Clinical Excellence (NICE), and is set to be integrated into software packages used by GPs.
University of Oxford statisticians validated QRISK using an independent cohort of 1.07 million patients from 274 UK general practices. There were 43,990 cardiovascular events over a median of 4.9 years’ follow-up. [BMJ 2009 Jul 7;339:b2584]
QRISK under-predicted cardiovascular risk by 12 percent (13 percent for men and 10 percent for women) but was considerably more accurate than the Anderson Framingham algorithm, which overestimated risk by 23 percent (32 percent for men and 10 percent for women).
“We believe this formula has the potential to save many thousands of lives, by helping clinicians to more accurately predict those at risk of developing cardiovascular disease – the nation’s biggest killer,” said Professor Julia Hippisley-Cox of the University of Nottingham, who led the team which developed the QRISK score.
“It will arm doctors with all the information they need to decide how best to target patients with preventative measures such as lifestyle advice and cholesterol-lowering treatments. We are delighted to receive another strong endorsement of the value of QRISK in assessing the risk of heart disease in the UK population,” she said.
QRISK includes most of the traditional risk factors seen in the Framingham equations but also includes family history, social deprivation, body mass index and use of antihypertensive treatments. It was first developed and validated in 2007, using data from 1.28 million patients from 318 UK general practices. [BMJ 2007 Jul 21;335(7611):136]
Further refinements to the model – QRISK2 – were published last year to include ethnicity and other conditions such as type 2 diabetes, hypertension, atrial fibrillation, renal disease and rheumatoid arthritis. The updated version, however, has yet to receive independent validation. [BMJ 2008 Jun 28;336(7659):1475-82]
The external validation of the original QRISK was commissioned by the UK Department of Health to compare the model against the Anderson Framingham algorithm, which is presently recommended by NICE as the basis for deciding on whether to prescribe statins.
The incidence rate of cardiovascular events was 30.5 per 1,000 person years among men who were classed as high risk by QRISK, compared to 23.7 per 1,000 person years among men classed as high risk by the Anderson Framingham score. In high-risk women, the incidence rates were 26.7 per 1,000 person years for those identified with QRISK, and 22.2 per 1,000 person years for those identified by Anderson Framingham.
Despite the improvements with QRISK, the data provide “a sobering message about the current state of cardiovascular risk prediction,” according to the authors of the accompanying editorial. Increased usage, improved data collection and further refinements to the system could however increase the accuracy of risk prediction in future, wrote Professor Rod Jackson of the University of Auckland, New Zealand, and colleagues. [BMJ 2009 Jul 7;339:b2673]
No comments:
Post a Comment