BeyondApples.Org

Doctors have known for some time that in people who have already had a major coronary event (e.g., heart attack) statins reduce the risk of a second event. Most studies estimate that statins reduce the risk of death in patients who have established coronary heart disease by about 30 percent. Statins are also recommended as first-line therapy for hyperlipidemia or “high cholesterol,” one of the major modifiable risk factors for cardiovascular disease. Given the prevalence of cardiovascular disease, some people have wondered, only half-jokingly, whether statins should be added to the water supply (if it works for fluoride and tooth decay, why not heart disease?).

A recent study published in the British Medical Journal (BMJ) asked an important question about the potential merits of such widespread statin use — is there any benefit to treating people without established cardiovascular disease but who have cardiovascular risk factors with statins? (1) The answer has sparked controversy not only about statin use for the primary prevention of cardiovascular disease but also about the way we report the results of clinical trials and the role of industry-sponsored trials in shaping clinical practice.

The BMJ study was a meta-analysis of 10 randomized clinical trials of about 70,000 people followed for an average of 4 years. In these trials, people with risk factors for cardiovascular disease but no history of existing disease were randomized to receiving statins or no treatment. After pooling these individual trials, the authors found that compared to people not randomized to receiving statin therapy those treated with statins had fewer deaths (odds ratio 0.88), fewer coronary events (0.70), and fewer cerebrovascular events (0.81).

In brief, in people at risk for cardiovascular disease, statins lowered the risk of death, heart attacks, and strokes. Sounds amazing, right? Think again.

Problem 1: How The Results Were Presented

First of all, what is an “odds ratio?” According to Wikipedia, an odds ratio is “a measure of effect size, describing the strength of association or non-independence between two binary data.” (2) Okay, that wasn’t helpful. The way I think about it is that an odds ratio is a statistic that compares the odds of an event (e.g., probability of heart attack over probability of no heart attack) in the intervention group to odds of an event in the control group. Okay, still doesn’t make much sense.

In addition to odds ratios, the authors report risk reductions. Risk reduction tells us, simply, how much taking statins versus not taking statins affects our risk of an event. In the table below, you can see that the people in the statin group had a 12% “relative risk” reduction in total mortality, which means that people who took statins had a 12% lower chance of death than those who didn’t take statins.

But even these statistics can be deceptive, because sometimes the risk of getting a disease is so low in the first place that a significant reduction in that risk leads to a negligible improvement in health. That’s why there’s a movement in the academic literature towards reporting absolute risk reduction. Absolute risk reduction tells us by what amount (not by what percentage) our risk of a disease changes with an intervention. It also allows us to calculate “number needed to treat,” which many people find to be a useful way of presenting complex statistics. Number needed to treat, or NNT, is the number of people that would need to receive an intervention to prevent one bad outcome. So using the chart below (total mortality), 167 people would need to be treated with statins for 4.1 years in order to prevent one death.

While the authors provided data on absolute risk reduction, they did not report NNTs, which is quickly becoming a literature standard. As a rule of thumb, to recommend a medication the NNT should be less than 50. Many observers have questioned whether the NNTs were intentionally excluded because they did not meet this criterion.

Problem 2: How the Results Were Obtained

Adding to this concern is the extensive industry ties of the report’s authors. Here are the authors’ self-reported disclosures, taken verbatim from the article:

“Competing interests: AMG is a consultant for Genentech, Kowa, Martek, Merck, and Merck/Schering-Plough, and serves on the board of directors for Aegerion and Arisaph. He is a member of DuPont’s health advisory board and serves on the data safety monitoring board for Novartis. JS carried out consultancy work and receives support for research from Bristol-Myers Squibb. RGJW receives support for research from Bristol-Myers Squibb. HN has received travel grants and speaking honorariums from Sankyo. RHK has received research fees and speaking honorariums from Pfizer. PR has received research grant support from the National Heart Lung and Blood Institute, the National Cancer Institute, the Donald W Reynolds Foundation, the Leducq Foundation, Astra-Zeneca, Novartis, Merck, Abbott, Roche, and Sanofi-Aventis; consulting fees and lecture fees from Astra-Zeneca, Novartis, Merck-Schering Plough, Sanofi-Aventis, ISIS, and Vascular Biogenics; and is listed as a co-inventor on patents held by the Brigham and Women’s Hospital that relate to the use of inflammatory biomarkers in cardiovascular disease. These patents have been licensed to Siemens and Astra-Zeneca.”

Even though this is a scientific study that uses many standardized methods there are many opportunities for the introduction of bias. One such area is study selection. There are dozens of clinical trials that the authors could have used in their meta-analysis but they choose 10. They cite a variety of reasons to include some trials and exclude others. For example, they excluded “three studies with design problems, fewer than 20 events overall, and insufficient followup.” What were these “design problems?” They don’t say but clearly these kinds of decisions, while necessary in any scientific study, are potential avenues for bias, intentional or not.

With the number of potential conflicts of interest, many doctors and patients wonder, rightfully so, how much stock to place in this study and whether the results are applicable to clinical practice. As stated by one physician, “As a general rule, I refuse to take any advice from anyone, even an ‘opinion leader’ who has taken even one cent from a drug company. I await an analysis from a group with no connection to ‘industry’ before I change my recommendations for prevention of [cardiovascular disease].” (3)

Discussion

Increasingly, people in the general public are engaged in primary medical literature. This trend is welcome and exciting. After all, we are all paying for this research through NIH funding, and it pertains to our health, so why not get it directly from the source? The trouble is that even for a practicing physician like myself interpreting these studies is wrought with challenges. If I just read the abstract, it would be easy for me to get excited about the results and want to get all my patients with risk factors for cardiovascular disease on statins. But as I hopefully demonstrated these studies are not so straightforward.

So what do I think of statins for the primary prevention of cardiovascular disease? It’s hard to say. Even if I took all the results of this study at face value, it’s hard to know what to do. Many of my colleagues would say that with a NNT of 77 to prevent one coronary event statins should not be recommended for people with risk factors for cardiovascular disease but no preexisting history. Statins are not without side effects, after all, and certainly are not free. At the same time, they would argue, there are other interventions like dieting and exercise that can be just as beneficial without the potential for harm. But I don’t think it’s that easy to dismiss statins. I know some patients who would jump at the chance to reduce their risk of a heart attack by 1.3%. In addition, statins are well tolerated by most people and are usually covered by insurance. At the same time, there is a large contingent of patients who, despite repeated attempts, just can’t seem to lower their risk of cardiovascular disease through lifestyle modifications and in whom statins may be a reasonable adjunctive therapy.

Bottom Line

Statins are no panacea, and their benefits have been overstated. In select patients, statins may be a good choice but not before an aggressive course of lifestyle interventions. Before we go putting it in the drinking water, we need to step back and take a closer look at the evidence.

- Shantanu Nundy, M.D.

(1) http://www.bmj.com/cgi/content/full/338/jun30_1/b2376

(2) http://en.wikipedia.org/wiki/Odds_ratio

(3) Taken from ProCor Global Dialogue list serv email.