We have recently seen much ethical hand-wringing around use of the two new and experimental treatments for Ebola, ZMapp and TKM-Ebola. Unfortunately, people are worried about the wrong things.
The first two patients to be treated were Americans. Prior to this came the old (and historically justified) fear of testing potentially dangerous new treatments on vulnerable populations in developing countries rather than on privileged first world patients. The first ever treatment was not given to an African doctor because of this concern . After Westerners were treated, complaints rose about giving infected Westerners access to a new drug while infected Africans went without. These two concerns are obviously mutually exclusive. People are concerned, they are just not sure about what, sometimes even voicing both concerns in the same article .
An important worry for any drug is that it might have damaging side effects. But if there were ever a disease for which this is not a big deal, it is Ebola. It seems extraordinarily unlikely that the drug could make matters any worse for an Ebola patient than they are already. No drug, approved as well as unapproved, is ever completely “safe.” The question is whether patients given the drug tend to do better, worse, or the same as those not given the drug, both in the short term and the long term. In the case of Ebola, I think we can reasonably ignore any concerns about the “safety” of the treatments and focus on figuring out whether they are effective.
To learn about effectiveness, we need to test these treatments on Africans, in the midst of a terrifying epidemic. Science isn’t just something that happens in laboratories, to be brought to the clinic when ready. To find out whether a treatment will work on real patients in clinics, there is only one way to do it: we need to test it on real patients in clinics. Those patients and clinics are in Africa. Today’s circumstances, with an abundance of patients, are exactly those in which the trials need to happen. And given the terrifying spread of this particular outbreak, the sooner we start, the better. The question is not whether this is the best or most ethical way to do a trial. This is the only way to do a trial, so let’s get on with it.
The WHO has given an ethical green light  to the use of these experimental therapies, with the caveat that “there is a moral obligation to collect and share all data generated, including from treatments provided for ‘compassionate use’ (access to an unapproved drug outside of a clinical trial).” They mention a number of ethical issues, but leave out the one that troubles me. The unethical behavior here, which was just given a green light by the WHO, is not doing an experiment, but doing an experiment without using a control group. There should be no compassionate use exceptions. Everybody who wants these treatments should have to enter a randomized trial to have a chance of getting them.
Five patients have received the ZMapp serum so far. Two US missionaries recovered, a Spanish priest died, and reports are not yet in for two Liberian doctors. Unfortunately, the supply is now exhausted, and we must wait until more is manufactured. Another experimental treatment, TKM-Ebola, also now has a green light for use in patients . I don’t know how many people we can treat with the quantity of TKM-Ebola available today. I hope it is many more than the five we could treat with ZMapp.
At least five patients have received a potentially effective treatment, but nobody has yet been assigned to a control group. This is the ethical travesty, and it needs to stop.
With 1013/1848 = 55% of reported Ebola patients dying during this outbreak, you may think that we don’t need a control group. Surely we will notice if the death rate goes down from there . But this is not as easy as you might think. Some of those infected patients haven’t died yet, but will die later, so the true death rate is likely a little higher than 55%. And neither the typically more privileged patients who receive the therapies, nor the care they receive during their treatment are likely to be representative of the average Ebola patient suffering and perhaps dying at home in their village.
A control group is not only essential for good science backed by rigorous statistics, but it is also ethically sound. Many people are understandably concerned about deliberately withholding potentially life-saving treatment and giving a placebo instead. But this is a moot point; we have no choice about withholding treatment. There isn’t enough of it to go around. The fairest way to distribute it is by lottery. This fairest plan also happens to be the best scientific plan. The question is whether we capitalize on our inability in the service of science by implementing a well-designed lottery, or whether we squander it instead with a less systematic lottery of who gets treated and who misses out.
This situation has an important and telling precedent. The very first randomized clinical trial ever published  in 1948 investigated whether antibiotics are effective against tuberculosis with Ebola-like death rates. Just like today, there was a shortage of a new drug, streptomycin. In the aftermath of World War II, Britain received only a very limited supply from American manufacturers. The Medical Research Council controlled this shipment, and seized the opportunity to dictate to doctors that if they wanted access to the drug for their patients, the only way to get it was by taking their chances within a randomized trial. Thus began a golden era of effective medical science , led by a strikingly novel and extraordinarily effective gold standard. What is more, to everyone’s surprise, the streptomycin treatment had limited effectiveness against tuberculosis. Patients improved initially, but then the bacteria evolved resistance to the drug and patients deteriorated again. This phenomenon of drug resistance had never been seen before. The randomized trial allowed this failure to be quickly recognized, leading to the improved and effective treatment protocols we have today.
We have so few doses available, and it is unethical to waste them in unsystematic tests that lack a control group. The US should have led the way by tossing a coin to see which of the two American patients got ZMapp and which got a placebo. And then it should have given its second dose to Africa on the condition that it be randomly distributed to one out of two preselected patients in the same manner. This would have been the ethical thing to do.
If ten patients had been rigorously randomized such that only five of them received the scarce ZMapp treatments and the other five got a placebo, and if ZMapp really were a miracle drug reducing the death rate from 60% to 0%, then there is a 69% probability that we would have been able to conclude in a “statistically significant” fashion, from that tiny but well designed trial, that the drug works. We think of randomized trials as huge affairs. And it is true that to detect subtle differences, we do need huge trials. But when the improvements are not subtle, trials can be quite small. Sixteen patients (eight per group) would be enough to have a 95% chance of detecting the effectiveness of a miracle drug. Even if the drug were slightly less miraculous, dropping the death rate from 60% to 30%, the 10-patient trial that we could and should have done with existing resources would have had a 42% chance of detecting the drug’s effectiveness. Our unsystematic trial of five patients will tell us much less, squandering this opportunity to make best use of the limited supply of ZMapp that we had.
Even if the treatment is only moderately effective, we still want to prove this and use the treatment with confidence that we are doing good. The more patients we can test, the better our chances of spotting moderately effective treatments. The limit should be our ability to manufacture the drug quickly enough, not our failure to throw together a randomization procedure. Randomization procedures are not difficult; all you need is a coin or dice. Let’s not make this harder than it needs to be.
What we should learn from the history of randomized trials is that the biggest danger with randomization comes from the urge to cheat. It is only natural for doctors to have special sympathy for particular patients, and to rig the game so that their favorite patients end up in the treatment group rather than the placebo. This is the real historical reason that trials were made “double blind” so that neither patient nor doctor knows who gets what. There is a lot of talk about the power of the placebo effect, and it is important for some conditions such as pain, but for most diseases the placebo effect is overrated. A placebo won’t stop you dying of Ebola just because you believe in it. But a double blind design is still an important precaution to make it impossible for well-meaning doctors to rig the allocations.
This overwhelming temptation is why we should enforce randomization at the source. We should ship treatments out in randomized kits, where each “treatment” is labeled only with a color-coded key. This should be the only way that treatment can be had; no compassionate use exemptions, instead you receive a kit at random. The kits can come in more than two colors, to make it harder for doctors to guess which is which. And if a patient dies before finishing all doses, or seems to need an extra dose, doses can be exchanged between kits of the same color. A condition for receiving treatment kits is that you report back, for each kit, whether the patient(s) who received it lived or died. That’s all you need to report; it’s not complicated. Modern clinical trials are complicated, but the subtleties don’t matter for a life-and-death condition like Ebola. Let’s not make things more complicated than they need to be.
One final concern is that many Africans are suspicious of Western doctors and experiments, and that their fears will keep them away. That’s fine, at least for now. That’s what the ethical principle of autonomy is about, crystallized in the notion of informed consent. And yes, African patients can be informed. The message we need to give them is simple:
“This drug might work. We have some good reasons to hope so, but we don’t know. We don’t think the dangers of the drug are anything like as bad as the disease you already have, and it might save your life. If you sign up, whether you get the drug is like tossing the coin, except you won’t be told which way the coin landed, and we your doctors won’t know either. In any case, we will do our best to take care of you, but this disease is hideous, and with or without treatment, you might die. If you sign up, you will also be helping us learn for sure whether the drug works, when we see whether people who get it do better. It’s up to you.”
A Western education is neither necessary nor sufficient for understanding this simple message, which encapsulates what a randomized trial of an experimental treatment means. We should cut through the excessive paperwork of the “informed consent ritual” that few Westerners understand , and stick to the essentials.
Plenty of infected Africans will probably refuse to take part in the trial. That’s okay, because we don’t have enough treatments to go around anyway. I suspect that the brave doctors and nurses in Africa who have risked their lives caring for Ebola patients will believe in the system enough to volunteer for a chance at life-saving treatment. Indeed, in recognition of their bravery, I would put health care providers in the front of the queue to sign up for the trial. But neither they nor anyone else should get access to our very limited stocks outside of a trial. Nobody should have more than a 50% chance of getting the treatment rather than a placebo.
Let’s hope that both treatments can be manufactured quickly, that both work, and that this outbreak is contained soon. And that during the next outbreak, we can move on to testing the two treatments against one another rather than against a placebo, to see which of the two works better.
Joanna Masel is an evolutionary biologist at the University of Arizona, where she teaches a class in Evidence-Based Medicine. Her research investigates the robustness and evolvability of biological systems and the nature of different types of competition in both biology and economics.
 Opting Against Ebola Drug for Ill African Doctor, by Andrew Pollack, The New York Times, 12 August 2014.
 Ebola, research ethics, and the ZMapp serum, by Laura Seay, Washington Post, 6 August 2014.
 Ethical considerations for use of unregistered interventions for Ebola virus disease (EVD), World Health Organization, 12 August 2014.
 FDA eases restrictions on experimental Ebola drug as CDC warns of ‘inevitable’ spread to US, RT.com, 8 August 2014.
 Ebola prizes revisited, by Eric Crampton, Offsetting Behavior, 5 August 2014.
 Streptomycin Treatment of Pulmonary Tuberculosis, British Medical Journal, 30 October 1948.
 Controlled trials: the 1948 watershed, British Medical Journal, 31 October 1998.
 Informed consent: how much and what do patients understand?, by M.E. Falagas et al., American Journal of Surgery, September 2009.