Guest post by Jeremy Howick
Trials show that drugs called ‘interferon alpha’ extend life in people with advanced skin cancer (by a bit). If we invented a new drug to treat advanced skin cancer, most patients would want to know whether the new drug was better than interferon alpha. It would be less useful to know that the new drug was better than a sugar pill (a placebo). So in our study it surprised us to find that Eggermont and colleagues gave some advanced skin cancer patients placebos when testing their new drug called ipilimumab. Was this ethical?The answer depends on which version of the World Medical Association (WMA) ethical guidelines (codified in the ‘Declaration of Helsinki’) we look at. The original Declaration was published in 1964 as a response to the horrific Nazi experiments with humans in World War II. According to this original version, it was unethical to give people placebos if there was an established treatment. This makes sense to doctors and patients. If doctors could have provided a treatment they knew worked, they risk doing relative harm by prescribing placebos, and harming patients is unethical. And most patients want to know which treatment option is the best one, not whether a new treatment is better than a placebo.
It’s not that the original Declaration banned placebos in trials altogether—they were allowed if there was no alternative. For instance there is no cure for the common cold. So testing a new treatment for the common cold against a placebo would be okay. There were also two other kinds of cases where they could be okay.
- The first is in a ‘placebo add-on’ trial. In these trials, all patients in the trial receive the best existing treatment. Then, some receive the new treatment as an ‘add-on’ to the existing treatment, while others receive a placebo in addition to the existing therapy. This type of trial is useful in some cases. But when people talk about placebo controlled trials they usually don’t mean add-on trials.
- The second is placebo trials in developing countries, where people wouldn’t be able to afford the standard drug. If someone invented a new cheaper drug that they could afford in the developing country, then it might be okay to compare that new drug against a placebo. I don’t agree with this point because the new drug could generate large profits, so I think the manufacturers owe the people in the trial the chance of getting something that we already know works. It turns out that it doesn’t matter whether you agree with me, because in 2000 the Declaration of Helsinki was revised to allow placebo trials in developed countries too.
The revised Declaration states:
a placebo controlled trial may be ethically acceptable even if proven therapy is available … Where for compelling and scientifically sound methodological reasons its use is necessary to determine … efficacy or safety
A problem was that they didn’t explain what the ‘compelling and scientifically sound methodological reasons’ are. I searched for them and found two, and neither makes much sense. The first is that placebo controlled trials are supposedly better at telling us whether a new treatment works or not. The geek word for this is ‘assay sensitivity’. Promoters of placebo controlled trials use the example of selective serotonin reuptake inhibitors (SSRI) antidepressants to make this point. While SSRI antidepressants are quite effective for people with severe depression, they barely beat placebos for people with mild depression.
This means that (at least in trials) SSRI antidepressants are not always better than placebos for people with mild depression. So if we compared a new antidepressant drug against an SSRI in people with mild depression. If the new drug were just as good as the SSRI we wouldn’t know if the new one was better than a placebo or not.
But if the reason we need a placebo-controlled trial is because the old drugs aren’t that good, then why not demand that the new drug be better than the old one? Many promoters of placebo controlled trials think it is okay for a new drug to be as good as an old one, rather than better. You might believe this is okay if the new drug has fewer side effects. But the best way to prove that the new drug has fewer side effects than the old one is to do a trial in which we compare the new drug with the older one. (An option is to conduct a trial that has three groups: one group would get a placebo, the other would get the new drug, and the third would get the old drug.) Financial interests could also play a role here. If we compare the new drug with an older one, the new one might come out as a loser even if both are better than placebo. This could explain why Eggermont’s placebo controlled trial was funded by industry, whereas trial funded by the National Cancer Institute is currently underway that compares the new drug (ipilimumab) with the old one (interferon alpha).
The second ‘methodological and scientific’ reason to prefer placebo controlled trials is that they supposedly give us a measure of ‘absolute effect size’. Basically this means that if we compare a new drug with an old one, we don’t know how much better the new one is compared to a placebo. But why does this matter? Patients and doctors want to know how the new one compares to what they already have, not what it does compared with a placebo. There is another reason why this is wrong: placebos don’t provide a ‘baseline’ against which an ‘absolute’ measure can be obtained. For example in ulcer drug trials, the placebo effects ranged from 0% and 100% of the drug effect. Like with ‘real’ drugs, we don’t always know why placebo effects vary, but they do.
Besides not making much sense, our recent study suggests that real doctors or patients don’t agree with them. If real doctors and patients believed there were good ‘methodological’ or ‘scientific’ reasons, then we would expect to see lots of placebo controlled trials used where there was an established therapy. But we don’t. We found that just four (6% of our sample) of trials used placebo controls even though they could have used an established treatment. The Eggermont trial of ipilimumab described above was one of them.
I’ve tried contacting Eggermont to ask whether patients in the trial knew about the existence of interferon alpha but haven’t had a response. And our study shows that real doctors and patients don’t seem to agree with the WMA revision to the Declaration of Helsinki. The confusing arguments about the supposed ‘methodological’ advantages of placebo controls seem controversial at best. This means the WMA’s policy on the ethics of placebos in trials is unethical.