Statins and primary prevention of cardiovascular events

BMJ 2008; 337 doi: http://dx.doi.org/10.1136/bmj.a2576 (Published 14 November 2008) Cite this as: BMJ 2008;337:a2576
  1. Norbert Donner-Banzhoff, professor1,
  2. Andreas Sönnichsen, professor2
  1. 1Department of General Practice, University of Marburg, D-35032 Marburg, Germany
  2. 2Institute of General Practice, Paracelsus Medical University, A-5020 Salzburg, Austria
  1. Norbert{at}med.uni-marburg.de

    No change in strategy is needed despite the hype surrounding the recent JUPITER study

    This week, the New England Journal of Medicine published the randomised controlled trial JUPITER,1 which compared rosuvastatin (20 mg daily) with placebo in 18 000 patients with no apparent vascular disease, low density lipoprotein cholesterol (LDL-C) of less than 3.4 mmol/l (130 mg/dl), and high sensitivity C reactive protein concentrations of 2.0 mg/l or higher. The combined primary end point was myocardial infarction, stroke, arterial revascularisation, hospital admission for unstable angina, or death from cardiovascular causes. The trial was stopped after a median of two years after a highly significant improvement in the primary end point with rosuvastatin (hazard ratio 0.56; 95% confidence interval 0.46 to 0.69; P<0.00001). It is hardly surprising that the JUPITER study is seen by many as opening the door to new avenues to prevention.

    What do the results mean? Do we really have to change our ways of targeting our preventive efforts—for example, measure high sensitivity C reactive protein on a regular basis? Although the relative risk reduction is impressive, the absolute risk puts the results into perspective. Of 100 people in the control arm, 1.36 experienced a primary outcome within one year, and this was reduced to 0.77 by the intervention. Although this was significant, its clinical relevance is doubtful. Participants were free of vascular disease, but half of them had a Framingham risk score >10%, and more than a third had the metabolic syndrome. A closer look at subgroup analyses (size of plots, exact numbers are not reported) indicates that most events occurred in high risk groups. Wouldn’t old fashioned risk estimation by traditional methods have produced similar results?

    The relative risk reduction (44%) was also much higher than in previous trials. Although the authors suggest that measuring C reactive protein allows the selection of a group of patients who benefit more than others, alternative explanations should be considered. Because participants in both the intervention and control groups had low concentrations of LDL-C, controls were presumably not taking non-study statins. This is in contrast to many previous trials, where up to 17% of patients in the control group were taking non-study statins, which diluted the positive effect.2 3 Moreover, a four week placebo run-in phase allowed the JUPITER investigators to select highly compliant patients. This also limits the external validity of the trial.

    Although the JUPITER investigators talk about a “strategy” based on measurement of LDL-C and C reactive protein, this is not what they studied. In fact, they present a conventional drug trial on a highly selected group of patients. Testing a strategy in a pragmatic trial would have to involve measuring high sensitivity C reactive protein in addition to detecting conventional risk factors to identify patients like the ones included in JUPITER. In a control arm only the usual risk factors would be used to target treatment at high risk patients. A trial of this kind conducted in relevant settings, such as primary care, is needed before a strategy that includes measuring high sensitivity C reactive protein can be routinely recommended. Only then can we know whether the additional complexity and cost (high sensitivity C reactive protein is measured differently from conventional C reactive protein) would translate into reduced disease outcomes.Furthermore, from a recent analysis of the Framingham study we learn that the additional prognostic value of laboratory measurements is negligible.4 Against the hype of new tests and drugs we should not forget that to work at a population level interventions have to be simple and easily reproducible.

    So what can we learn from the JUPITER trial? Firstly, that rosuvastatin is efficacious and safe. This is important because a previous trial with rosuvastatin 10 mg versus placebo was negative, with a non-significant relative risk reduction of only 8%.5 However, because rosuvastatin was not compared directly with simvastatin and other statins, we do not know whether it is really better. Until we know that, established (and cheaper) statins should be preferred. Secondly, statins work independently of LDL-C concentrations, but in low risk populations the effects on absolute risk are small. Thirdly, this study—in which almost half of the subjects were below the target level for LDL-C lowering in ischaemic heart disease—makes the whole concept of lipid lowering to a specific LDL-C target look more obsolete than before. Instead of lowering lipids we should talk about global cardiovascular risk assessment. Statins with their multiple actions are an effective means to improve the prognosis of high risk patients, as are aspirin and antihypertensive drugs. Possible effects should be discussed with patients,6 and if doctor and patient agree that a statin is indicated, a fixed dose should be given.7

    All this is hardly new. No change in practice is warranted on the basis of the JUPITER study. Future studies should evaluate alternative strategies of risk assessment and intervention in usual care settings that may include measuring high sensitivity C reactive protein. Current preventive practices should be reconsidered only if an effect can be shown in a pragmatic study.


    Cite this as: BMJ 2008;337:a2576


    • Competing interests: None declared.

    • Provenance and peer review: Commissioned; not externally peer reviewed.