With the publication of the new Framingham Risk Score to predict 10-year absolute risk of a coronary heart disease (CHD) event in 1998,1 this practice – which already used Framingham scores extensively to discuss lifestyle risks with patients – re-evaluated the scores using the new guidelines. The risk assessment applied only to assessment for primary prevention of CHD, not to those who already had established CHD.
It is worth noting that the small print of the Framingham scoring table states that people with a family history of premature vascular disease have a higher risk than the predicted score given and that some European nations may have a lower risk in relation to standard risk factors.
This caused intense debate in the practice once the Framingham scores were calculated and discussed.
The practice searched on the BNF code 2.12 to identify patients who were on statins for primary prevention; from the practice of 12,000 patients, 178 were currently receiving statin treatment. Of these, 55 had a diagnosis of angina, ischaemic heart disease or myocardial infarction also coded on the computer, and were thus not in the primary prevention category. It was therefore assumed that all remaining 123 patients being treated with statins were receiving them for primary prevention.
The notes were pulled for these patients and the Framingham risk scores predicting 10-year absolute risk were calculated using the new scoring regimen (table 1, below), but before treatment.
|Table 1: Framingham Risk Score|
Those treated since 1992 would therefore have had a Framingham risk score calculated on their pretreatment level of 1992 with the new scoring system. Sadly the practice discovered large holes in the flagging of patients.
On looking at the summary cards, many of the 123 patients did indeed have evidence of ischaemic heart disease, myocardial infarction, bypass surgery or angina. Framingham scores were calculated for all the patients without any of these diagnoses and new scores were entered on the computer.
The results showed that 19 patients had a Framingham score between 0 and 10%, 34 patients had scores between 11 and 20% and 18 patients had scores of greater than 21%. However, these percentages are only absolute risk of CHD events and not relative risks.
This latter point is important as the relative risk will rise absolutely as patients age, and blanketing statements such as 'we should treat all patients with an absolute risk of greater than 20%' may be inappropriate if the patient is 65.
The audit taught the practice many things. One was that the vast majority of patients who were on statins and did not have computer entries showing that they were being treated for secondary prevention were new patients. The diagnosis appeared to be entered on the computer by the GPs, or as the result of new hospital letters, but there was a breakdown in the flagging of notes of new patients joining the practice, identified by the practice nurse, and being coded correctly on the computer.
Originally the practice was chastened to see that 1% of patients were on primary prevention treatment, but are somewhat heartened to see that now only 71 patients are on primary prevention, which is only just over 0.5% of the population – and of these only 19 have scores below 10%.
These 19 patients have generated intense debate, not least in light of the recent debate alluding to evidence-based medicine and the patients' wishes being paramount in treatment. This debate merits discussion in the practice.
First, there is a problem as to whether patients who take part in trials are similar to those seen in general practice. It is constantly suggested that patients in trials are fitter and have less morbidity than patients seen in general practice. The argument runs, therefore, that patients seen in general practice are more sick and therefore will not benefit as much as patients in the trials.
The second argument is that individual risk is not the same as population risk, and that population studies do not apply to an individual patient. Personally, I find this latter argument nihilistic and believe that the only way to improve a population's health is to improve every single individual's health, albeit in a small way.
Similarly, some of the patients in this apparent low risk group have a family history of very premature CHD, so that these patients, despite their low score of relative risk, should be treated. Conversely, patients with a high relative risk, but a low absolute risk, should probably not be treated.
By studying our prescribing of statins we have become aware that not only was our flagging of patients with ischaemic heart disease poor, but also the practice has begun to understand the difference between relative risk and absolute risk. The practice has also identified a subgroup of patients who have a low relative risk but should be treated.
The partnership feels that it has a much better understanding of the use of statins, those at risk who should be treated, both with absolute or relative scores according to the Framingham table, despite low scores, and those who should not be treated at all. Perhaps this is the key – providing treatment where it will be most effective and withdrawing it from patients in whom it will be less effective.
Using the Framingham risk score to predict any absolute risk of CHD event has changed the practice's awareness of the use of statins and generated lively and educated debate.
- Wilson PWF et al. Prediction of coronary heart disease using risk factor categories. Circulation 1998; 97: 1837-47.