What proportion of symptomatic side-effects in patients taking statins are genuinely caused by the drug? A response to letters

Abstract

Dear Sirs We thank Professors Collins, Mancini and Golomb for their interest in our study.1 Professor Collins points out that although we did not find an excess of muscle-related adverse events in the statin regimes used in our analysis, the statin trials generally used low-intensity doses. Our further analysis1 comparing muscle side-effects between high- and low-intensity statin regimes did indeed show a statistically significant increase in myopathy symptoms with creatine kinase elevation >10 ×upper limit of normal, and of muscle aches, in the high-intensity statin regimes. In our paper1 we presented side-effect data separately in Tables 1 and 2 for the primary and secondary prevention population. However, we agree with Professor Collins’ reasoning that for effects other than on vascular events, incremental rates above placebo are likely to be similar between those who already have had a vascular event and those who have not. We agree that it is therefore useful to present the aggregate symptom rate data for the combined patient group (Table 1). Analysis of events reported in primary and secondary prevention randomized controlled trials. Analysis of events reported in primary and secondary prevention randomized controlled trials. We thank Professor Mancini for his gracious comments and for highlighting the many numerical calculations of clinical value. His NNTs and NNHs are the key for patients deciding whether to start therapy. The Proportion of Symptoms Non-pharmacological we describe is for patients, experiencing a symptom, who are deciding with their doctor whether to stop. Professor Mancini’s attention to placebo subtraction is wise and broadly applicable. We have done this for chronic beta blockade in heart failure,2 where the results were again eye opening. We welcome Professor Golomb’s insights. We are pleased that she agrees that, in the patients assessed in clinical trials and with the statin strengths used in those trials, there appears to be net neutrality on symptoms, along with a reduction in heart attacks, stroke and death, accompanied by an increase in diagnosis of diabetes. Certainly with any intervention, some individuals may gain and others lose. It is our role as physicians to give advice to help patients make the right choice for their individual preferences. Some will focus on net gain. Others focus exclusively on avoidance of harm. Others are extremely averse to taking regular medication.3 Yet others are suspicious of any medical recommendation for intervention, perhaps sensitized by past misadventures of which our profession should not be proud. Professor Golomb makes an excellent recommendation that future randomized blinded trials should meticulously collect information on symptoms. Fatigue is indeed a widespread symptom, affecting both patients and doctors. The pervasiveness of fatigue is evident: while in blinded trials 3.1% of the statin recipients reported it, 2.9% of the placebo recipients had it too. Thus while any reliable information on its origin and amelioration would be valuable for those taking statins, clearly it would be just as valuable for their counterparts who are not. Her hesitancy to immediately assume the glucose elevation from statins to be definite physiological harm – even though it increases diabetes diagnosis rates – is well thought through. There are many anti-diabetic drugs that lower glucose dramatically but do not deliver significant cardiovascular event reduction that would be expected if glucose level was the sole mediator of harm in diabetes. Therefore we need not assume that drugs raising glucose must definitely cause the cardiovascular harm that would be expected from diabetes. We continue to read her careful physiological studies with fascination. Professor Golomb’s suspicion that trials are designed to enrol patients most likely to benefit is well founded and perhaps not irrational for funders. We share her disappointment that many clinicians do not realize that very large sample sizes do nothing to remove bias.4 In fact, they increase the tendency for small biases to cause statistically significant effects. We are glad Professor Golomb saved the dying man. This is an important lesson to bear in mind for future patients with life-threatening muscle wasting. We must match this vigilance with continued preventive efforts against cardiovascular disease, which kills over 1000 people per day in our two countries,5 and is still the world’s largest killer.5