December 23rd, 2013

Statins: Targeting Risk, but Risking Diabetes?


One of Richard Lehman’s recent blog posts lauding a BMJ editorial about targeting cardiovascular risk rather than cholesterol concentration from CardioExchange’s Editor, Harlan Krumholz, struck a chord with member Emma Morton-Eggleston. Emma, who is an endocrinologist at the Brigham and Women’s Hospital in Boston, wrote to Richard with her perspective, who responded, and they’ve agreed to share their debate with CardioExchange:

Dear Richard,

Hope all is well — and as always thanks for your weekly reviews. They are guaranteed to provoke a smile at your incisive (and instructive) skepticism. But I have to disagree with you on Dr. Krumholz’s editorial — which surprised me, as did your response to it, precisely because it was so optimistic.

While I completely agree with his primary points regarding the jettisoning of targets and the importance of grounding the translation of guidelines in shared decision-making with patients, the suggestion that we can let time and further studies fix the potential over-medicalization of millions of people (and if the global medical community follows suit, hundreds of millions) from a likely flawed risk calculator is puzzling to me. To my mind, the potential medicalization of low-risk populations for primary prevention is a fundamental problem that can’t be left to a promissory note on future work that accurately defines risk. Even if such work comes to pass, as noted in the JAMA Viewpoint you also referenced by John Ioannidis, risk prediction is an inherently tricky business even in the most thoughtful and rigorous of hands.

The diabetes experience made clear that once a guideline is in place, distracted following is the norm given the time pressures of modern medicine and the natural tendency towards inertia in critical thought that you have described so well. Why now is it statins and heart disease, not diabetes and anti-glycemic medications, do we do an about-face on decrying over-medicalization? And abandon prior attention to absolute risk of disease, and of benefit and harm of intervention, in the process?

You have howled (as have I across the Atlantic in agreement) about the potential creation of vast global pharma markets for quasi “diseases” including pre-diabetes. Why no howling now? The availability of generic statins may mean little profit is made for the time being, but it seems highly unlikely that opening the door to medical treatment for the prevention of arteriosclerotic cardiovascular disease (CVD) in low-risk populations will not herald targeting of these populations for new antilipidemic agents in the future.

The potential harms attributed to statins are no small thing and can impact quality of life in very real ways — I am not sure how this squares with patient-centered care in low-risk patients. Last — and you knew this one was coming as I am an eternal hammer when diabetes is the nail — the dismissal of potential diabetes risk with statin use in patients at low risk for CVD because of the absence of long-term data is baffling. It is an increased risk that two large studies — an RCT and a retrospective cohort analysis — have suggested may approximate 50% in women. The CV benefit of statins in high-risk groups is about as close as we can get to true evidence-based medicine, and benefit appears to far outweigh risk of diabetes in these patients. But it is not at all clear that small decrements in absolute risk of CVD outweigh harms in low-risk patients.

Muscle aches (at times profound), myopathy, and diabetes aside, overmedicalization is never a good thing. Okay, I’ll say rarely since nothing is never except immortality and nothing is always other than death, splendid gardens, friendship/family for the lucky, and good cheese with a ripe pear.

Looking forward to your thoughts,


Dear Emma,

Thank you so much for writing. I’m really glad you’ve raised these issues because they puzzle me. I am trying to draw up a simple option grid to help people discuss this with their doctors, since most people still need to see a doctor to get hold of a statin. This discussion might then proceed to a personal cardiovascular risk assessment using one of the current instruments — not the flawed one you mention, but either Victor Montori’s or James McCormack’s — to give people a ball-park idea of where they lie in the risk spectrum and what the absolute benefit might be from taking a statin.

I would leave guidelines aside here — this is not about health professionals but about a personal decision by a healthy person about how they want to play the odds of dying in a particular way. We can point people to fairly accurate information about that in relation to this class of drugs, though we could argue endlessly about the detail.

This doesn’t need to be done by a doctor — in fact it is probably best done by a web-based program, with the doctor merely signing the prescription according to what the person chooses. I don’t even see this as a domain of “shared decision making” because doctors really have no business telling asymptomatic people what to do. We can give our personal advice, but in the end it is the individual’s choice, based on full and accurate information — i.e., the kind that doctors rarely have the time and knowledge to provide. And as these are lifelong choices, they should be made at leisure and revoked at will — by the individual. In this way “overmedicalization” is not an issue — we simply accept the free choices of autonomous individuals. Some people may want an absolute risk reduction of 0.01% — that’s their affair. Others may have an irrational fear of all tablets and simply not take them whatever their risk — and there is nothing we can do to change that. Given such a free choice, “compliance” or “adherence” become meaningless terms.

That’s the way I see this issue. If a statin makes your muscles ache, you simply stop taking it. A recent study showed that most patients can be successfully started back on a different statin, though that is not my experience. I think — though I do not know for certain — that this applies to every adverse effect attributed to statins, including the rise in blood sugar which takes some people over the arbitrary threshold into “type 2 diabetes.” I would like to know more about the reversibility of this: I hope it is similar to the “diabetes” caused by thiazides, which stops when you stop the thiazide and has no prognostic importance anyway, since it is just a short-term biochemical effect of the agent and does not indicate beta-cell damage. If you are interested in the herd, rather than the individual, then the benefit of the statin in this borderline group outweighs any conceivable harm from a small rise in fasting blood sugar. But again, it’s up to the individual whether to carry on taking the stuff or not. You are not going to feel any different whether your fasting sugar is 6.5 or 7 and the odds are that you will live longer taking the statin, though not by much.

Does this make sense? I am sorry to find that my views on this differ from those of many friends who seem to think there is a pharma-led conspiracy to overmedicate the world. There is no money in this for anyone, and I just hope the whole saga will be one more nail in the coffin of target-driven “care”, where the quality of practice is measured by how many people you persuade to take a particular action. We are not here to increase the life expectancy of the adult general population by imposing interventions, but to help people live and die in the way that they choose, according to the skills we may possess. At least, that is my philosophy, and I still owe Harlan a book about it, which I hope to call “Sharing Medicine.”

I’d really value your thoughts about my option grid if you have time. And I hope you are enjoying/surviving the snow and obtaining a good supply of ripe pears and cheese.

All the very best,


Dear Richard,

Beautifully said! The framework for shared decision making you lay out is what we should all be working to ensure — and your reminder that the choice is entirely the patient’s to make is critical in informing how one views making clinical recommendations. I find it very helpful. But I still worry that the act of making the recommendation to start a statin is a form of medicalization. Even within the fluid bounds of shared decision-making, to start that conversation you have to invoke the specter of risk and lay a diagnosis of sorts on the patient. You are changing how they (and their insurance company) view themselves. Perhaps this perspective just shows my own medical paternalism (maternalism?). It is hard to shake.

Regarding glycemic thresholds, your point that the diagnosis of diabetes is but a number on a glucose continuum is well taken and, as you suggest, in the case of statins there may be little correlation between meaningful clinical outcomes and observed glycemic rise. But the metabolic and pancreatic reality underlying glycemic thresholds is often more complicated, and variable, than that and we simply don’t yet understand whether the “diabetes” of statin use is metabolically the same as the inflamed environment/beta-cell dance that is “diabetes” as we know it. Nor do we know whether statin-potentiated diabetes progresses or carries the same (or lower), risk of complications and all-cause mortality in low-risk patients. We do know that the large majority of observational studies demonstrate a strong relationship between glycemia and all-cause mortality – and at A1Cs well less than 7%. And we increasingly acknowledge that we can’t effectively decrease that risk with treatment in many people.

As far as I know (but please correct me if I’m wrong) there is little good evidence that statin therapy decreases all-cause mortality in low-risk patients. So are we saying that in groups at low risk for  arteriosclerotic CVD we will recommend a drug for primary prevention that for most may modestly, but for some (women) may substantially, increase their risk of a disease that is itself a strong risk factor for mortality from any cause — not to mention a whole host of lesser, but no less relevant to patients, harms over time? And one that we can’t treat well once it develops? I’m open to the possibility, but will wait on the means to accurately identify risk — and the data demonstrating clear benefit that outweighs harms, before I do recommend it.


4 Responses to “Statins: Targeting Risk, but Risking Diabetes?”

  1. Seth Martin, MD says:

    Dear Richard and Emma,

    Thank you for sharing your thoughtful debate on CardioExchange. Please allow me to offer some additional thoughts, though I can’t promise that they will be as eloquent as yours.

    Getting back to the new ACC/AHA cholesterol guidelines for a moment, I must point out that these guidelines capture the shared decision making framework. The guideline coins it a “risk discussion.” This is a patient-provider discussion of potential benefits and harms, drug-drug interactions, and patient preferences. So, rather than imposing millions of statin prescriptions, or a billion of them, the guideline encourages patient-provider dialogue about ASCVD risk and therapeutic options. The risk discussion is a key virtue of the guideline, in my opinion. It values the doctor-patient relationship, the art of medicine, and the personalization of care.

    A risk discussion is recommended when starting a statin for primary prevention in adults with LDL-C 70-189 mg/dL and >=5% estimated 10-year risk of MI/CVA. Yes, Emma, there is good evidence that statin therapy decreases all-cause mortality and ASCVD events in this “low risk” setting. Please see:

    Mihaylova B et al. The effects of lowering LDL cholesterol with statin therapy in people at low risk of vascular disease: meta-analysis of individual data from 27 randomised trials. Lancet. 2012;380:581-90.

    Even with this strong evidence, I can understand the concern for “overmedicalization.” While the population-level benefit is there, many individuals will not go on to have an event, regardless of statin therapy. So, it is logical to ask if there is a way that we could even better identify individuals who are most likely to go on to events…

    I wonder whether either of you have had a chance to carefully examine the recent literature on atherosclerosis imaging, particularly coronary artery calcium scoring. The 2013 ACC/AHA cholesterol treatment guideline incorporates this as an option when the risk assessment is considered uncertain. In the Multi-Ethnic Study of Atherosclerosis, persons with a coronary artery calcium score of 100 or more carry a risk of a future CVD event similar to persons who have already suffered a CVD event.

    In this way, the distinction between primary and secondary prevention becomes artificial. The disease is the same: atherosclerosis. The risk is the same. Does it matter that the patient is asymptomatic? Asymptomatic does not equal healthy in my book. Not in a disease that progresses silently over decades until the day it strikes.

    So if a “primary prevention” patient has visualized atherosclerosis and carries the same probability of a future CVD event as a secondary prevention patient, would you then agree that it is no longer appropriate to label that person “healthy” or as having a “quasi disease”? And that it is not “overmedicalization” to discuss this risk and offer proven options to reduce that risk?

    Getting back to the risk discussion, I agree that the focus should be on empowering individuals to make fully educated decisions. I also agree that high-quality web-based programs could help in this process. However, I must admit that I’m skeptical of the idea that an internet tool is all that patients need. Maybe I’m too young and idealistic, but I still think physicians have a large role to play in guiding patients through major preventive health decisions such as whether or not to take a statin.

    In any case, we will hopefully learn a great deal more on this issue as the 2013 ACC/AHA cholesterol treatment guideline implementation phase proceeds. To that end, Richard, you might want to share your option grid with the expert panel, as it could be helpful to risk discussions and guideline implementation.

    I’d finally like to say that your discussion on statin associated hyperglycemia/diabetes is very insightful. Indeed, the cholesterol treatment guideline addressed this topic and weighed the risk in its recommendations. I’d also like to highlight prior discussions on Cardioexchange with perspectives from my colleagues, Mike Minder ( and Roger Blumenthal (

    Thanks again for sharing your thoughts. I hope that you find my additional comments constructive.

    Happy holidays,

  2. Enrique Guadiana, Cardiology says:

    The distinction between primary and secondary prevention is not artificial. In primary prevention if you have a 5% in 10 years risk to develop CVD well in the next 10 years, 100 individuals with the same characteristics as you, 5 will develop CVD. In secondary prevention of CVD, 100% has CVD, they are not at risk they are there, now what is in question is the prognosis. Atherosclerosis is not synonymous of ischemic heart disease, stroke, PAD, etc. the distinction is not artificial either, is important, you can have atherosclerosis, live a long live and an never develop CVD sintomatology or visceversa; so the treatment, severity and prognosis are different.

    In medicine as in life there are no absolutes. Expert panels, papers. Meta Analisys are great and very helpful but not perfect, they are human tools and we are far from perfect. The best way to approach medicine is with an open mind.

  3. Thank you Emma and Richard for the refreshing discussion. Despite all the hoopla surrounding the guidelines, it was not until your exchange that I sought to understand why I (uncomfortably) support the guidelines. For those of us who believe that we should focus interventions based on risk, rather than targets, the guidelines are a welcomed nod of support; for high risk patients, we will no longer be scrutinized for not driving an LDL to < 70 when statin therapy is maximized, and no longer need to sit on our hands with an LDL of 65 on pravastatin 20mg (as I often see in our CCU). However, putting the burden of goal-setting for low-risk patients on shared-decision making feels uncomfortable. Mostly because we haven't figured out how to extend effective shared decision-making for the millions of people who fall into this category.

    Shared-decision making, in my mind, is the pinnacle of good medical care. In full disclosure, I get a professional high when I am able to effectively engage patients, understand their values, and find a path forward that leaves the patient walking out feeling at-peace, even elated, with their decision. It makes all the difference in the world to embark on a path with conviction and promise. Unfortunately, this happens way too infrequently. My grandfather, now 95, always shopped at Sims. Why; he likes their motto, "an educated consumer is our best customer." Isn't that true. So we come up with decision-aids or option grids to level the playing field; educate our patients about the real risks and benefits; and invite them into our world. But too often, they are not prepared to deal with data as they are overcome by the larger meaning of disease in their life. For the baby boomer generation, it is often the, "well, I knew I would get to this place at some point." A resignation that they are destined to have disease and take medication. For others who already have been 'medicalized,' the incisive discussion around risks and benefits falls flat.

    So where does this leave us? Is quantifying risk and offering a statin to a low-risk person another form of medicalization? I think not. But our current construct for shared decision making does not teach us how to illicit long-held values or to test the human capacity to take on challenges and new approaches to life. Our teaching on medical interviewing has evolved to circumvent the, "doc, I forgot to mention I am having chest pain" declaration as the patient walks out the door. But have we considered the conversation that ensues with a spouse or a friend after the visit, when the person has had time to process. Maybe that should guide our shared decision making process. Responding to the emotional context of what it means to be of low-risk but have a new diagnosis of high cholesterol may charge our patients in ways unexpected. Ways that cannot be captured in an option grid. Indeed, my former mentor, and now colleague, Dr. Rob Ostfeld, a cardiologist at Montefiore started a wellness program for people in the Bronx; he is succeeding in getting many patients in the Bronx to consider and even adopt a whole-foods, plant-based diet. Kudos to Rob! In the end, I think we must acknowledge that diets, exercise, and statins are proven strategies for reducing risk, some to greater degrees than others. We should embrace the new guidelines as a challenge to be better physicians; not to further burden our patients with numbers that too often fall flat, failing to address how they really feel when they kick off their shoes at night and wrestle with their new diagnosis.

  4. Thanks Emma, Richard and other coleagues for the interesting discussion.

    I would like to put ” true life” in perspective:

    …Unfortunately, studies have shown that not only do physicians not understand the issue of health literacy or the level at which information needs to be given but they are also very poor at judging the health literacy or the numeracy ability of their patients.This lack of awareness and therefore lack of addressing the issue adds an extra layer of complexity to the reasons surrounding patients’ poor perception of their CVD risk…. (Webster R te al Risk Management and Healthcare Policy 2010:3 49–60)