Fifty years ago

Despite exceptional mortality rates half a century ago, anyone in Britain asking a doctor at that time how to avoid having a coronary heart attack, or wanting their cholesterol tested, risked being labelled a hypochondriac. Cholesterol was considered an American fad. Unlike in America, medical treatment here was free—no need to worry therefore, the National Health Service would look after you. Prevention was the province of quacks and charlatans, outside the remit of orthodox medicine, the exception being fringe doctors (often regarded as failed clinicians) in public health, along with occupational physicians, obstetricians and some enthusiastic general practitioners. Consultants in hospitals were there to practise diagnosis and treatment, and that included cardiologists. Many were critical of prevention. It appeared simultaneously to be both threatening and unscientific.

Things have changed

In the 21st century things are different. Evidence based medicine challenged prejudice. Framingham and other cohort studies had introduced the concept of risk factors,1 while Stamler had pioneered preventive cardiology in the 1960s.2 Britain followed-up its 1940s invention of randomised controlled trials with its cardiologists supporting really large multicentre trials in the 1980s and 1990s,3 easing their way into preventive thinking through secondary prevention, and increased numeracy on risk. Medical advice is now no longer exclusively channelled through a personal physician: the media and the internet are awash with it. Government wants to save money by prevention. ‘Big Pharma’ have produced ever more effective drugs, and sponsored much postgraduate medical education. Guidelines and box ticking replace a large part of what was often idiosyncratic clinical judgement. Computerised patient databases provide potential for auditing clinical behaviour, implementation of guidelines, and their outcomes, both in hospitals and in the general population. We now claim to know what doctors and their teams ought to be doing, can tell them what to do (with financial inducements), and some years later can assess whether they did it. So what happens when a centrally dictated algorithm on cardiovascular disease prevention meets the ‘wisdom of the crowd’ of general practitioners?

The accompanying paper by Van Staa et al4 describes the ‘high risk’ strategy of primary prevention of cardiovascular disease through prescription of statins in the general population, when studied in a large UK general practice database, in disease-free men and women aged 35–74 years. Guidelines had recommended statin treatment, after multifactorial risk assessment, for those whose projected 10-year risk estimate for a cardiovascular event was 20% or more. Family history and total cholesterol/HDL ratio over 6 were additional qualifying criteria.5 ,6

Increase in initiation of statins

The authors did not know which cardiovascular risk score was being used, if any, in a particular case. When they applied three commonly used scores separately7–9 the risk score distributions in the populations were fairly similar, although the individuals so classified were not necessarily the same. There was a major increase in initiation of statin treatment from 2007 onwards. Uptake nonetheless was heterogeneous, with heavy usage in some general practices and very little in others. Initiation was much higher proportionately in individuals at high risk (score ≥20) than in the much larger number designated as ‘low’ risk (score ≤15), but the total numbers treated in each group were fairly similar. Towards 2011 more at ‘low’ risk were initiating treatment than high—the guideline was being applied only partially.4

What might influence the initiation of statins? Here we can see the outcome, but not who initiated the preceding consultation, nor the context, nor the interaction leading to starting or not starting long term medication. We do not know who refused statins, or about the worried well requesting them. Secondary prevention is excluded, but what about the patient with non-specific chest discomfort? We also do not know about patients obtaining ‘over the counter’ low dose statins direct from the pharmacy without involving the general practitioner.

The data show that initiation tended to follow multiple consultations in the previous year, that nearly half those concerned were being treated for hypertension, and that two thirds—where information was available—were overweight. Numbers of men and women being treated were more similar than might be expected given the higher age specific risk in men, but women were on average 2 years older. This suggests opportunistic rather than systematic screening. For these reasons, information on those starting statins was more complete than on others. Data on smoking were not recorded for 31% of the general population, obesity 45%, blood pressure 27%, and cholesterol 76%. For women, who consult more frequently, data were more available than for men. Similar missing data, by contrast, on statin initiators were 13%, 26%, 3%, and 9%. Data are not provided on social deprivation, a major determinant of risk.10 Those of Indian ethnicity, with enhanced susceptibility, did not appear more frequently among initiators than the general population. Were they offered statins? Did they refuse?

Changing scene of cardiovascular disease prevention

How has the cardiovascular disease prevention scene changed since these guidelines of the last decade? A revolution in attitudes to risk scoring had occurred previously following statin trials in the mid 1990s.11 Before that, risk scoring was used to encourage behavioural change and could emphasise relative risk, greater in younger age groups.12 After that, justification for medication was said to demand ‘absolute risk’ for which increasing age was the most powerful determinant.13 When incorporated into risk scores it means that 20% 10-year cardiovascular risk is rarely reached below the age of 50, but commonly so in old age. In the ASSIGN score8 it is reached by a male, with otherwise average current Scottish values of risk factors, at the age of 60 if he smokes, and if he doesn't, at 65; in women 5 years later. Therefore, targetting highest risk would logically begin in those over 70. But the over 70s have substantial risk of death from competing causes, not accounted for in conventional cardiovascular risk scoring. Starting at 70, disease at younger ages would not be prevented. The 10-year, absolute risk approach is now being challenged in favour of ‘lifetime risk’, implying early medication. That would be taking us back full circle towards relative risk, and is debatable in other ways. Lifetime risk is promised for the forthcoming Joint British Societies third update (JBS3)14 overdue at the time of writing. The mean age of initiating statins in the current data was 59 in men and 61 in women, suggesting general practitioners were pursuing a middle road between the various different strategies.

Cost and safety issues

Cost is another consideration. The high risk strategy is predicated on effectiveness and efficiency. Statins have been a major drain on the National Health Service (NHS) budget, and there is a tendency to deprecate ‘over prescribing’ in the ‘low risk’. Yet the annual cost per person to the NHS of treatment is now a fraction of what it was a decade ago. Leading statins are rapidly losing their patent rights. The cutpoint for ‘low risk’ in this report is not truly low. Someone whose current risk is 15% will reach 20% within 5 years. A recent Cochrane collaboration report emphasises that statins benefit lower risk patients by the same proportionate risk reduction as those at higher risk.15 There is no obvious lower threshold. Any cutpoint is arbitrary. The threshold for treatment could therefore be reduced with benefit to more people. Indeed it was once argued,16 and still debated,17 that statins should be used, incorporated in a polypill, in everyone over age 50 or 55—no need for risk scoring.

Safety and side effects then become the issue. The Cochrane report, based on published trial results, suggests usage at low dosage is remarkably free of risk from side effects, and that those that do rarely occur are reversible, often without stopping treatment.15 A large population cohort study (from which QRISK2 was derived) suggests that the reality may be rather less reassuring, and unintended effects are not negligible, although dose dependent.18 Unlike that on cost, this problem is not scheduled to decline.

An effective treatment creates as many problems as it solves. These findings illustrate what happens with the implementation of a high risk prevention strategy on a population-wide basis. Results may be relevant to the implementation of other prevention strategies, such as that recently announced for breast cancer in the UK, using tamoxifen and raloxifene. It would be wrong, however, to suggest that a high risk strategy can prevent more than a portion of cardiovascular disease. Half or more of disease occurs in those not labelled as high risk,13 and statins reduce risk by 30–40% over 5 years in those maintaining such therapy, suggesting a 20% population impact where users do not stop, but many do.15 The major decline in coronary heart disease in the UK over recent years can therefore be only partly driven by statins used in primary prevention,19 ,20 although secondary prevention would be a separate additional contribution.

General practice databases are proving their increasing worth for epidemiological purposes and for national planning, as their data improve in quality and completeness. Their designers need to anticipate future questions by including relevant items in the specification. Accurate recording of data is as important as complying with guidelines.