Comparative effectiveness research — investigations that determine which treatments are best — has attracted attention in the health care debate. Critics charge that these studies are designed to restrict choice. The Center for Medicine in the Public Interest released a report that suggested that they would stifle innovation. Often they are framed as studies to support efforts to keep useful but expensive therapies from patients.
But what if these studies, done well, revealed that some medications were better than others? What if they overturned conventional wisdom about understudied drugs, demonstrating that many patients were receiving ineffective treatments? What if they showed that some patient were actually being harmed? What if more knowledge about the benefit and risk of treatments in medicine, compared with their alternatives, is just what patients need?
Recently, we had an example of comparative effectiveness research done well. The National Institutes of Health (NIH) has produced a gem of a study with critical implications for patient decisions and lessons for future research. The study, called AIM-HIGH, was designed to evaluate whether adding Niaspan, a proprietary formulation of extended release niacin sold by Abbott Laboratories (Abbott Park, IL), to statin treatment in patients with heart and vascular disease would reduce their risk.
Niacin, also known as nicotinic acid or vitamin B3, is one of the body’s essential nutrients. In high doses niacin increases the good cholesterol (HDL) and decreases triglycerides. The drug causes several unpleasant side effects that limits its use in high dose, including an uncomfortable flushing sensation. The extended release niacin is better tolerated and often prescribed to patients. Whether the drug reduces a person’s risk of heart disease is not known.
The AIM-HIGH investigators enrolled 3,414 patients who had cardiovascular disease and were taking statin drugs to reduce their risk. These patients had low levels of the HDL cholesterol and high levels of triglycerides, markers of increased risk. The subjects were randomly assigned to Niaspan or placebo and observed for several years.
Although the study has yet to be published, enough information was released last week to give us a good picture of the results. The trial was stopped more than a year early because there was not a hint of benefit for patients to this point, making further evaluation unnecessary. There was even a suggestion of harm, with more strokes in the Niaspan group – but that may have been a chance occurrence.
The NIH study is an example of comparative effectiveness research that can be embraced by patients, clinicians, industry, payers and government because of their approach. Let’s take a close look at what they did right.
1. The study targeted an important area of uncertainty for patients. Niaspan is a popular drug that is known to raise the HDL cholesterol and decrease triglycerides. Sales of the drug last year were close to $1 billion. Unfortunately we do not know whether the drug reduces the risk of heart disease or stroke – or improves survival.
The approval of the drug was based on its effect on laboratory tests, not patient outcomes. This proprietary formulation that is touted widely in direct-to-consumer advertising and in colorful promotional pages in medical journals is not known to reduce patient risk when added to a statin drug. We have evidence from other studies that drugs that modify risk factors do not always affect patient risk in predictable ways. The only large study of niacin (not the extended release formulation) was conducted more than 30 years ago, before treatment with statins and the current era heart care.
Patients could make better decisions about whether to take this medication by knowing if it reduces their risk when added to a statin drug.
2. The study compared Niaspan with the a reasonable alternative strategy. Too often studies compare one drug with an inferior strategy. In this case, Niaspan was compared with good treatment with a statin and sometimes another drug to ensure that the bad cholesterol (LDL) was properly controlled. Statins are known to reduce risk and would be considered standard care for these high risk patients.
It was essential that Niaspan be tested in conditions that replicate the scenario in which patient need to decide with their clinicians about adding another medication. The study appropriately asks whether it is useful to add Niaspan to the regimen of patients who are taking statins.
3. The conclusions of the study did not go beyond the evidence. The press release focused on the absence of an effect of Niaspan, but did not generalize further to other drugs that may affect the levels of HDL cholesterol. Each drug must be tested in its own right because it has its own spectrum of effects on human biology and it is likely that there are many effects for each drug that are yet to be uncovered. Even within a class we have seen marked differences in the risk profiles of drugs – with Baycol and Zocor as examples – or Avandia and Actos.
Also, the investigators noted that the results should not be extrapolated to other patient populations. Whether there are specific groups who might benefit is unknown. What this study can say is that the only large study of this drug in a relatively high risk population has failed to detect an important benefit.
4. The NIH and Abbott were involved in the study. The NIH and the study investigators were wise to involve Abbot because it relieved taxpayers from paying the entire bill and engaged the company in a test of its product. Abbott Laboratories paid more than half of the $52 million study costs. Their involvement demonstrates a degree of consent to the study and commitment to it. Abbott should be commended for supporting the study and seeking truth about the effect of their drug on patient outcomes. No pharmaceutical company should want to sell a product that does not help patients. Abbott demonstrated their values by supporting a test of a billion dollar drug whose sales were based on an assumption of benefit. Moreover, Abbott did not criticize the study.
Having NIH oversee the study was also important. The NIH has impeccable research credentials and their policies and procedures protect the scientific integrity and transparency of the process. Unfortunately many industry trials have been tainted by concerns of misbehavior in the handling of data and the publication process. The credibility of this study was assured by its conduct under the auspices of the NIH.
5. The study had close oversight by an independent advisory group. The study had an independent group of experts protecting the interests of the patients. These experts were looking at the results as they were accumulating and entrusted with decisions about whether it was worthwhile to continue the study. They were also monitoring for any dangerous effects of the drugs that would mandate stopping the trial. They were the only ones who were allowed to see the results as the study progressed. It was their decision to stop the study based on the statistical certainty that it would not show Niaspan to be useful. Having such an independent group, which is not controlled by the sponsors and is accountable to the subjects, is essential in these studies.
The challenge is now to act on the information. The information that was released should be considered reliable enough to inform patients that adding Niaspan to statin treatment does not reduce patient risk even as it does change HDL cholesterol and triglyceride levels. Every patient on this drug – every patient for whom this drug is contemplated – should receive the information that was presented by the NIH and its investigators about AIM-HIGH.
We will learn more when the study is formally presented. The scientists need some time to perform all the analyses that were specified in their protocol. But the major finding is now known and dissemination of the information to patients should occur.
This is the type of comparative effectiveness study that we need. The focus is on describing how one strategy compares with another with respect to risks and benefits that patients experience. Patients are better informed about Niaspan than they were a week ago. The better information will presumably lead to better decisions. It is an outcome we should all embrace … and seek.
Harlan M. Krumholz is a professor of cardiology, epidemiology and public health at Yale University School of Medicine. He blogs at Forbes, where this article originally appeared.