last year a new cardiovascular risk algorithm QRISK was launched.¹ It had been developed using a UK population and therefore appeared to be more relevant to UK general practice than the traditional Framingham risk score. QRISK also had the advantage of incorporating ethnicity and social deprivation into the risk algorithm. However, NICE recommended the modified Framingham risk score in its guideline on lipid modification, published in May, saying that 'research is needed to validate QRISK in independent and clinical datasets and to assess the performance of the scores against each other.'²

A new version of the QRISK algorithm QRISK2 has now been developed and validated in a prospective open cohort study. The study used data from 531 practices in England and Wales contributing to the QRESEARCH database between January 1993 and March 2008.³ QRESEARCH routinely extracts collected data from the EMIS clinical system and includes census ethnicity and social deprivation data.

The study included 2.3 million patients, who were aged 35-74, did not have established cardiovascular or cerebrovascular disease and were not taking statins at baseline. Around 97% of the participants were white or did not have their ethnicity recorded, 1% South Asian, 1% black African or black Caribbean and 1% of other ethnicity.

Participating practices were randomly assigned to two cohorts: a derivation cohort, which consisted of 1.54 million patients registered at 355 practices, and a validation cohort, which consisted of 0.75 million patients from 176 practices.

The first occurrence of CVD, including CHD, stroke and TIA, was noted. There were 140,000 recorded cardiovascular events during the study.

The QRISK2 risk algorithm was developed using data on multiple risk factors, including ethnicity; age; sex; smoking status; systolic blood pressure; total cholesterol: HDL cholesterol ratio; BMI; family history of CHD in a first degree relative <60; Townsend deprivation score; treated hypertension; type 2 diabetes; renal disease; atrial fibrillation and rheumatoid arthritis.

More than 112,000 patients were classified as high risk (≥ 20% risk of CVD over 10 years) by the modified Framingham score. Using the QRISK2 algorithm, 41% of these patients were classified as low risk. The observed 10-year risk in these reclassified patients was 16.6%.

Around 78,000 patients were classified as high risk using QRISK2, of whom 15% were classified as low risk using the modified Framingham score. The observed 10-year risk in these reclassified patients was 23.3%.

Patients with a QRISK2 score of ≥ 20% had an observed annual incidence of 32.5 cardiovascular events per 1,000 person-years for men and 30.6 per 1,000 person-years for women. By contrast, a 10-year CVD risk ≥ 20% predicted by the modified Framingham score corresponded with 26.4 events per 1,000 person-years for men and 25.7 events per 1,000 person-years for women.

Therefore, QRISK2 appears to outperform the modified Framingham risk score in a typical UK practice population, and also has the advantage of including ethnicity and deprivation data.

Furthermore, QRISK2 has refined the original algorithm by incorporating self-reporting of ethnicity and additional risk factors such as treated hypertension, type 2 diabetes, renal disease, atrial fibrillation and rheumatoid arthritis.

QRISK2 can be embedded in clinical systems to provide an accurate tool to calculate CVD risk and populate risk registers.

NICE has accepted that CVD risk scores are a rapidly evolving area and it may be that QRISK2 finds its way into routine practice in the future. In practical terms my advice at present would be to stick with Framingham; most GPs will be familiar with this algorithm and it is robust and well tested as long as appropriate adjustments are made (eg for family history and ethnicity). Although QRISK2 appears to be superior, it has only been validated in a single cohort and until there is additional validation in other populations, particularly with different ethnic mixes, and has national backing (eg from NICE or JBS), I doubt that it will find common use.

However, whichever risk score you decide to use it is important to remember that these tools can only provide an approximate risk and that no risk score should replace clinical judgement when assessing patients for appropriate lifestyle or medical interventions.