Authors' reply Sir—In “intention-to-treat” analyses of the 4S trial, the 5-year average difference in plasma LDL cholesterol concentration was 1·7 mmol/L, and the reduction in the risk of “major coronary events” (defined as non-fatal myocardial infarction or coronary death) was one third.1Scandinavian Simvastatin Survival Study GroupRandomised trial of cholesterol lowering in 4444 patients with coronary heart disease: the Scandinavian simvastatin survival study (4S).Lancet. 1994; 344: 1383-1389Summary PubMed Scopus (11405) Google Scholar By contrast, in the intention-to-treat analyses of HPS2Heart Protection Study Collaborative GroupMRC/BHF Heart Protection Study of cholesterol-lowering with simvastatin in 20 536 high-risk individuals: a randomised placebo-controlled trial.Lancet. 2002; 360: 7-22Summary Full Text Full Text PDF PubMed Scopus (7749) Google Scholar (and of the other main trials of 5 years of statin versus placebo3Smith SC Lessons from cholesterol-lowering trials.Am J Med. 1998; 104: 28S-32SSummary Full Text Full Text PDF PubMed Google Scholar), the average LDL difference was about 1 mmol/L, and the risks of major coronary events and of “major vascular events” (defined as major coronary event, stroke, or revascularisation) were reduced by about one quarter. When comparing the intention-to-treat analyses of 4S and HPS, however, Scott Metcalfe and colleagues fail to take any account of these differences between the absolute LDL reductions achieved on average during each trial. Moreover, they seem not to have understood that the Conclusion of our report was intended to provide estimates (derived from the intention-to-treat analyses) of the benefits of actual use of 40 mg simvastatin daily among the types of high-risk patient studied in HPS (rather than comparisons with other trials), since such estimates should be of most direct relevance for high-risk patients who remain compliant, for the doctors who are treating them, and for health authorities that require appropriate cost-effectiveness analyses. Observational studies in different populations indicate that the proportional difference in coronary disease risk associated with a given absolute difference in usual LDL cholesterol concentration is similar throughout the LDL range that has been studied, without any “threshold” below which a lower concentration is not associated with lower risk.4Law MR Wald NJ Thompson SG By how much and how quickly does reduction in serum cholesterol concentration lower risk of ischaemic heart disease?.BMJ. 1994; 308: 367-373Crossref PubMed Scopus (1115) Google Scholar Despite these consistent observations, unduly selective emphasis in a previous trial on a retrospectively defined subgroup of patients with pretreatment LDL below 3·2 mmol/L, which involved few major coronary events (89 [21·7%] pravastatin vs 93 [21·1%] placebo),5Sacks FM Pfeffer MA Moye LA et al.for the Cholesterol and Recurrent Events Trial InvestigatorsThe effect of pravastatin on coronary events after myocardial infarction in patients with average cholesterol levels.N Engl J Med. 1996; 335: 1001-1009Crossref PubMed Scopus (7268) Google Scholar has been used to suggest that there might be a threshold at about this level below which lowering LDL would not reduce risk. By contrast, HPS prospectively planned to assess the effects of lowering LDL substantially among about 7000 participants who presented with LDL below 3·0 mmol/L, and it has shown unequivocally that lowering LDL from below 3 mmol/L to below 2 mmol/L reduces the risk of major vascular events (598 [17·6%] simvastatin-allocated vs 756 [22·2%] placebo-allocated, p<0·0001) by about one quarter, which is similar to the proportional risk reduction produced in HPS by a 1 mmol/L reduction among those who started at higher LDL concentrations.2Heart Protection Study Collaborative GroupMRC/BHF Heart Protection Study of cholesterol-lowering with simvastatin in 20 536 high-risk individuals: a randomised placebo-controlled trial.Lancet. 2002; 360: 7-22Summary Full Text Full Text PDF PubMed Scopus (7749) Google Scholar It is surprising, therefore, that Metcalfe and colleagues seem to give as much weight in their discussion to the small numbers of events in a retrospectively defined subgroup of the CARE trial (and, indeed, do not even consider the prospectively defined subgroup in that trial) as they do to the very much more robust findings from HPS or from the aggregate of all the randomised trials. Moreover, by contrast with their assertion, the absolute benefits produced by a 1 mmol/L reduction in LDL among the types of high-risk participants studied in HPS were similar, and substantial, both among those with pretreatment LDL below 3·0 mmol/L and among those with higher levels.2Heart Protection Study Collaborative GroupMRC/BHF Heart Protection Study of cholesterol-lowering with simvastatin in 20 536 high-risk individuals: a randomised placebo-controlled trial.Lancet. 2002; 360: 7-22Summary Full Text Full Text PDF PubMed Scopus (7749) Google Scholar The prespecified data analysis plan for HPS states that comparisons of the effects of simvastatin allocation among participants in different subcategories would be based on the large number of major vascular events that were anticipated, supplemented by the results for the smaller number of major coronary events (see http://www.hpsinfo.org). Figures 7 and 8 in the main report indicate that, despite analysis of a large number of prespecified subcategories, the proportional reduction in major vascular events was about one quarter in each subcategory of participants studied.2Heart Protection Study Collaborative GroupMRC/BHF Heart Protection Study of cholesterol-lowering with simvastatin in 20 536 high-risk individuals: a randomised placebo-controlled trial.Lancet. 2002; 360: 7-22Summary Full Text Full Text PDF PubMed Scopus (7749) Google Scholar Metcalfe and Dawson have chosen instead to emphasise the observed results in particular subcategories for the statistically less robust outcome of major coronary events, and then mistakenly suggest that there was significant heterogeneity between the proportional risk reductions observed among those with or without prior coronary disease. By contrast, there were similar highly significant reductions in major coronary events (webfigure 1 with main report), as well as in major vascular events (figure 7), among people in each of these prior disease categories.2Heart Protection Study Collaborative GroupMRC/BHF Heart Protection Study of cholesterol-lowering with simvastatin in 20 536 high-risk individuals: a randomised placebo-controlled trial.Lancet. 2002; 360: 7-22Summary Full Text Full Text PDF PubMed Scopus (7749) Google Scholar With appropriate allowance for multiple comparisons (as prespecified), there was not significant heterogeneity between the effects observed on major coronary events among participants subdivided with respect to ACE inhibitor or β-blocker use at study entry (webfigure 2), and this lack of significant heterogeneity is supported by the more robust findings for major vascular events (figure 8). Hence, HPS does not provide any good evidence that the proportional effects of simvastatin on coronary or other major vascular events differ materially in the presence or absence of ACE inhibitors or β blockers (or of any other treatments being used) from the substantial risk reductions observed overall.2Heart Protection Study Collaborative GroupMRC/BHF Heart Protection Study of cholesterol-lowering with simvastatin in 20 536 high-risk individuals: a randomised placebo-controlled trial.Lancet. 2002; 360: 7-22Summary Full Text Full Text PDF PubMed Scopus (7749) Google Scholar We can confirm for Peter Wilmshurst that the colour of the vitamin and placebo capsules was changed from red to dark brown early in HPS in order to ensure that some slight discolouration during long-term storage would not lead to unblinding of the treatment allocation. Finally, we agree with Andrea Poli and Alberico Catapano that the results of HPS and the other major statin trials are entirely consistent with the observed reductions in the risk of vascular events being largely or wholly a consequence of the reduction in LDL cholesterol concentration. P Sleight has received honoraria from the pharmaceutical industry for participating in scientific meetings, but the other authors have not. Heart Protection StudyWe have concerns about how the Heart Protection Study (HPS; July 6, p 7)1 results were reported. Full-Text PDF Heart Protection StudyThe data from HPS1 contain information which could be relevant to the debate on the clinical relevance of the non-lipidic or pleiotropic effects of statins. Full-Text PDF Heart Protection StudyThe reports of the Heart Protection Study1,2 state that a 2×2 factorial design was used to allocate patients to active drug (simvastatin or vitamins) or matching placebos. Investigators who attended the meeting at which changes in drug formulation were announced informed me that medication and placebos were changed part way through the study because the appearance and smell of active drugs and placebos were found to differ over time. Irrespective of whether this change would have affected outcome, investigators have a responsibility to ensure that research methods are accurately reported. Full-Text PDF