Comparative effectiveness approaches not always the most effective

by Robert Temple, M.D., Deputy Center Director for Clinical Science

Robert Temple, M.D. caricatureComparative effectiveness is on everybody's mind, and that makes sense. After knowing that a drug works, usually the most important questions physicians, and patients have are:  how well does this drug work compared to the alternatives that I could use; is it better, worse, or the same.  

So the desire to have comparative data, broadly referred to as comparative effectiveness research, is not surprising. The main difficulty with doing comparative studies is that the effects of most drugs, while valuable, are not very large, so that even showing a difference between the drug and no treatment (a placebo treatment) is not easy. Showing a difference between two effective drugs, a difference much smaller than the difference between a drug and no treatment, is very challenging and will usually need a very large study. Comparative data do exist that show advantages for some members of a class over others, but they are not common.

Comparative effectiveness studies that were successful

Within a class of drugs like antidepressants or antipsychotics, there are relatively few cases where we have been able to say that one drug is better than another; however, there are a few.

The antipsychotic drug clozapine, is generally thought to be more effective than other drugs in its class, however, it has a toxicity that no other members of the class have. Clozapine causes a marked decrease in the number of certain white blood cells. But it was shown to work in people who did not respond to other drugs.  This showing was critical to its approval.

Some drugs that inhibit platelets in patients with coronary artery disease have been studied in trials that compared the new drugs with an older drug. In some cases, the newer drugs had better effects, reducing the rate of heart attacks, though sometimes causing more bleeding. It is fairly common for a new cancer drug to be more effective than an older therapy.

There is a popular class of anti-hypertensive drugs called angiotensin receptor blockers. With considerable effort, and quite large studies, two companies have shown that their drug had a larger effect on blood pressure than other members of the class.

Individualized response studies

There is some hope that instead of doing large, randomized trials to show differences, between treatments, observational studies, i.e., epidemiologic data, can be used to do this. There is considerable disagreement on whether this is possible. My belief is that any differences between effective drugs are likely to be very small and can be credibly detected only in randomized trials. Whether pooling the results of multiple controlled trials (meta-analyses) will work is an area for discussion, but to do this you need many trials with the same comparison, a rare occurrence.

A kind of study that I think has a lot of potential is one that is almost never done: take people who have not responded to one drug, and then randomize them back to the drug that didn't work and to another drug. This is particularly interesting when the two drugs are in the same pharmacologic class.

There is a widespread relief that people respond different to similar drugs. Ask most physicians and they will tell you that there are individualized responses to treatments; some people do better on one drug than another very similar drug  and you don't always know the reason. They would expect that to be the case for pain medication (e.g., NSAIDS) or anti-depressants. If it is really true that people who don't respond to one member of a drug class actually respond to another member of the class, we can design the perfect study to test this.

For example, you could take people who don't respond to a migraine drug and randomize them back to a new migraine drug, the old migraine drug the one they didn’t respond to, and a placebo. If it's true that there are individual differences that are important, then the new drug ought to be able to show its advantage over the drug that didn’t work.

I am aware of only a few studies that have ever tried to do this:

  • Clozapine, as I noted, is marketed because it was shown to work in people who had failed standard anti-psychotic therapy.
  • Years ago, CDER was very worried about the first angiotensin-converting-enzyme inhibitor, captopril, because it caused agranulocytosis at high doses. FDA would probably not have approved it except that there was a trial in people who had failed what was then considered the best available therapy, “triple therapy,” a combination of reserpine, hydralazine, and a diuretic. Patients who had failed to respond to this treatment or to randomized back to triple therapy or to captopril; captopril dramatically beat the triple therapy. Captopril was then approved for patients who had failed triple therapy.
  • More recently, Merck did a study in which they took people who had failed on Celebrex, a non-steroidal anti-inflammatory drug, and randomized them to Celebrex and to Vioxx. The company hoped, and I’m sure expected, that the people who failed Celebrex would respond better to Vioxx. The study, however, showed that there was no difference at all between the two drugs. Based on that study, it remains to be seen whether there are individual differences in response to anti-inflammatory drugs.

The question of whether a person who doesn’t do well on one member of a drug class might respond better to a closely related drug or drug of a different class is obviously an interesting and very important one. It still surprises me that this kind of study is rarely done.

Evaluating side effects

A closely related area of interest is whether, if a drug causes a side effect another member of the drug class, or a drug of a different class can be substituted and not cause the side effect. Again, this is a type of study that is very rarely done. But it is an interesting and therapeutically important question.

It has been shown that anti-depressants cause poor sexual function in women -- a common and significant effect. But do all anti-depressants have this effect or have it to the same extent? In one study, people who had sexual dysfunction while taking an anti-depressant were randomized to the anti-depressant that caused that dysfunction and to bupropion; bupropion, a pharmacologically different drug, didn't cause the sexual dysfunction. While the result was not really a surprise, the fact that the study was actually done was unusual.  There would seem to be a great many opportunities for such studies.

I would like to see more of these relatively small, easily-designed studies. They are not hard to conduct and the data have in some cases proved important.


Robert Temple, M.D., Deputy Director for Clinical Science

Robert Temple serves as CDER’s Deputy Center Director for Clinical Science and also Acting Deputy Director of the Office of Drug Evaluation I (ODE-I). He has served in this capacity since the office's establishment in 1995. 

Dr. Temple received his medical degree from the New York University School of Medicine in 1967. In 1972 he joined CDER as a review Medical Officer in the Division of Metabolic and Endocrine Drug Products. He later moved into the position of Director of the Division of Cardio-Renal Drug Products. 

In his current position, Dr. Temple oversees ODE-1 which is responsible for the regulation of cardio-renal, neuropharmacologic, and psychopharmacologic drug products. Dr. Temple has a long-standing interest in the design and conduct of clinical trials. He has written extensively on this subject, especially on choice of control group in clinical trials, evaluation of active control trials, trials to evaluate dose-response, and trials using “enrichment” designs.

Page Last Updated: 02/26/2016
Note: If you need help accessing information in different file formats, see Instructions for Downloading Viewers and Players.