Framinghamis a town inMassachusetts, not far fromBoston. It’s predominantly white and middle class. It provided a refuge for families persecuted in theSalemwitches trials and it’s where the Battle Hymn of the Republic was first sung.
So what’s it got to do with how much English primary care trusts spend on statins?
The answer to that is simple: theFraminghamheart study. Between 1968 and 1975, doctors gathered data on heart disease in 5,573 people inFraminghamand framed a formula to calculate an individual’s risk of having a heart attack. It was universally accepted and became the gold standard.
That is until now. NICE has recommended it be replaced with QRISK. It is a decision that will have profound impact for PCTs – both for the bottom line and for the health inequalities agenda.
Questions about theFraminghamrisk score began to emerge around 2002 when Dr Peter Brindle, a GP in inner cityBristoland a that time a Wellcome Trust fellow, questioned whether it was valid for his multiethnic, deprived population.
First he challenged it by comparing it with the British equivalent – the British regional heart study and found that theFraminghamscore over estimated the risk of non-fatal coronary events by 57 per cent. More worrying, he also found that 84 per cent of British deaths from heart disease were in the 93 per cent of men classified as low risk byFramingham.
“Basically, it did not work that well depending on where you were using it,” he says. While it included risk factors such as age, gender, smoking, blood pressure and total cholesterol it had no measures of deprivation or ethnicity.
Dr Brindlle, who is also R&D lead forBristol,South Gloucesterand North Somerset PCTs, adds: “My ambition since then has been to find an alternative.”
That something is QRISK. It was developed with a team of researchers from theUniversityofNottingham, Bristol PCT and the Universities of Queen Mary andBristol.
They used data from QRESEARCH, a general practice database set up by clinical system provider EMIS and theUniversityofNottingham. It is the largest database of its kind in the world with every EMIS practice pumping anonymised patient data in on a daily basis.
Instead of tracking 5,500 patients in one town, they tracked the progress of 1.28 million healthy men and women aged 35 to 74 registered at 318 general practices over a period of 12 years to April 2007. They recorded the first diagnosis of cardiovascular disease and looked at the relationship between this diagnosis and various known risk factors for CVD.
The result is a complicated algorithm to predict who is at risk of CVD. In addition to the risk factors used byFramingham, they were able to factor in family history, body mass index, a measure for deprivation and current antihypertensive prescription.
Applying the algorithm back to the original data, they were able to look at how accurate it was – and compare it withFraminghamand another risk score developed inScotlandcalled ASSIGN.
Not only was QRISK more accurate but also it predicted fewer people overall would be at risk of CVD. In patients aged 35 to 74, QRISK over predicted risk at 10 years by 0.4 per cent.Framinghamover predicted by 35 per cent and ASSIGN by 36 per cent. QRISK predicted 9 per cent of patients aged 35 to 74 to be at high risk of CVD compared to 13 per cent forFraminghamand 14 per cent for ASSIGN.
There was more. Because QRISK built in several additional risk factors, it identified a different group of patients on the basis of age, sex and social deprivation. Broadly speaking, it identifies more women and more people in deprived areas as being at risk.
Writing in the BMJ in July 2007, the authors concluded: “It is therefore likely to be a more equitable tool to inform management decisions about and help ensure treatments are directed towards those most likely to benefit.”
QRISK is online now and easy to use, requiring patients and/or their doctors to click through a series of questions before coming up with a risk score – and an explanation of what that risk means.
“It’s not perfect,” says Dr Brindle. “But it is fairer. We are now looking at a second generation QRISK that builds in the ethnicity data we have on tens of thousands of patients.”
That’s the academic side of things. Using a new tool such as QRISK requires research papers but also approval, acceptance and a political context. All three are coming together in 2008.
QRISK came along just as NICE was working on new guidelines on CVD risk assessment, lipid modification and primary and secondary prevention. In October 2007 NICE delayed them take QRISK into account.
By February it had accepted that QRISK was indeed preferable toFramingham, a decision that has proved controversial as (a) Dr Brindle and another QRISK developed Dr John Robson sat on the guideline development group and (b) not everyone – doctors from the cholesterol charity HEARTUKincluded – is convinced it works.
NICE has defended the GPs saying they left the room during discussions about the choice of risk assessment tool and pointing out that the other panel members were in unanimous agreement about adopting it. It is widely thought that QRISK will replaceFraminghamwhen the guidelines are published in May 2008.
Next the political. In January 2008 Gordon Brown made a surprise announcement. The NHS was to venture into primary prevention of CVD. He promised vascular screening, to be introduced this year or early 2009, to include a series of blood, fat and sugar tests in GP surgeries, alongside questions on age, gender, postcode, family history, height and weight.
It’s a tick list that looks very much like the variables in QRISK andDr David Stables, clinical director of EMIS, looks at the numbers through this primary prevention prism. “If it is adopted nationally and used correctly, we estimate that QRISK has the potential to save more than 200,000 lives over the next ten years.”
Dr Brindle believes that primary prevention is on its way. “There is such a big ground swell of opinion that this is the right thing to do that it will happen,” he says.
Nevertheless, he isdeeply ambivalent about it. If primary prevention of CVD becomes the norm and QRISK the risk assessment tool by which doctors identify who is at risk – either with the patient sitting in front of them or by trawling through records and calling in the people it highlights for a consultation – several things flow out.
Certainly it will help PCTs get a handle on tacking deprivation, he says. “But from a managerial point of view, those practices in the deprived areas will have a lot more work and I am not sure if anyone has fully worked through the implications.”
While QRISK will potentially reduce the number of men in affluent areas identified as at risk and therefore prescribed statins and/or exercise, it will bump up the number of women and people in deprived areas.
“We are talking about potentially another five million people on preventive drugs, needing blood tests and counselling, follow up and drugs prescribed. I am not sure we know what the true costs are.” The Vascular Board (a group at the Department of Health) is working on this now.
QRISK is just one output of QRESEARCH – albeit one that has set a number of hares running. “This is just an example of what QRESEARCH is doing,” says Dr Stables. “It has the potential for a significant change in the way outcomes and interventions are measured.”
Already, the Health Protection Agency uses a service called QSURVEILLANCE using anonymised data from 4,500 practices to track flu epidemics and vaccine use as well as health status in disaster zones such as after the Bunsfield fire.
Then there are the questions that the data could answer. Up to now, he says, clinical trials have been painstaking processes, based on a few hundred or a few thousand patients. QRESEARCH can access literally millions of pieces of high quality information derived from electronic patient records and could radically change this.
In theory, you could ask QRESEARCH any question about the primary care data. Is there a link between MMR and diagnosis of autism, for example. I am a diabetic, should I be on HRT? Is it just Viox that is linked with heart attacks or do other NSAIDs show the same pattern of side effects?
It could be used to flag up anomalous patterns, for example between prescription of a drug and unexpected side effects or a cluster of ill health.
Right now because of the way data is coded and searched, QRESEARCH takes three or four months to answer these questions. Dr Stables' vision is for the process to be done in minutes.
“This was my vision when I set up EMIS20 years ago,” says Dr Stables. “I thought it would take five years but we are not there yet. I have now revised it to a 30 year project.”
It is made possible by the willingness of GPs to share their data and by a system in which family doctors have cradle to grave responsibility for the primary care needs of their patients. The really big question is whether this system will survive long enough for Dr Stables’ vision to reach fruition.
For more information visit http://www.qrisk.org/Default.aspx