Using COVID-19 treatments before a clinical trial: a benefit-to-harm analysis
Updated: Apr 15, 2020
New treatments are dreamt up all the time. In both the minds of doctors, and in their test tubes. These ideas need to be translated into treatments proven to work. This is achieved by subjecting the treatments to clinical trials. However, there is often a lag between the conception of a treatment and the completion of a clinical trial. Clinical trials take time to recruit patients, time to complete, and are very expensive. Sometimes the lag is months, sometimes years, and sometimes decades. This is known as the translational time lag.
I would like to address in this email: why we need clinical trials to prove whether a COVID-19 treatment works, and whether it is smart to try these treatments before clinical trials are completed. I will perform a benefit-to-harm analysis using hydroxychloroquine as an example.
(I will not be addressing the ethical issue of whether a patient has the right to try these treatments. Legally, it has been resolved in the US. According to the Right To Try Act passed in 2018, patients have the right to try treatments that have not yet been proven to work in a clinical trial.)
Why we need clinical trials
Let's first take a dive into the history or Western Medicine, where doctors and patients constantly had to face life and death situations. Thinkers like Aristotle believed that answers to all questions facing humanity lay dormant in the brains of humans. One just needed to extract them using the process of deduction. This approach was called rationalism. There was no need for clinical trials. If one could present a logical argument of why a treatment would work, it was good enough.
Greek Medicine, aka Humoral Medicine, was one of the fruits of rationalism. It was created around 400 BCE by its main architect, Hippocrates of Kos. The premise: as the universe has 4 elements (earth, water, fire, air), so does the human body. The body has 4 fluids or "humors": yellow bile, black bile, phlegm, and blood. All disease is caused by an imbalance of these humors, and all disease can be cured by rebalancing these humors. Rebalancing is achieved by: purging stool (laxatives), vomiting (emetics), extracting fluid from the skin (blistering with skin irritants), and letting blood from the veins. Other rebalancing treatments included: exercise, diet, stress reduction, and baths. The legitimacy of Humoral Medicine was sealed when it was championed by Galen. He was the doctor to the Roman Emperors, and a trusted authority figure. Humoral medicine dominated European and Arabic medical practice for 2,100 years and was the main treatment for every devastating epidemic from smallpox, to malaria, to the plague.
Humoral Medicine culminated in the 1700's with a movement known as Heroic Medicine. As the name implies, Heroic Medicine was a more aggressive form of Humoral Medicine. Their motto, "If a little is good, more is better". It involved very aggressive bloodletting, induced diarrhea and vomiting, and the use of potent plant remedies and heavy metals like mercury. It’s greatest proponent was American doctor, Benjamin Rush. Rush was certainly a great American. He supported the Amercian Revolution, opposed slavery, and was one of the signers of the declaration of independence. He rose to almost saint status during the 1793 Yellow Fever epidemic in Philadelphia. Despite many doctors leaving town, Dr Rush stayed, risking his life to treat hundreds of sick patients. Heroic indeed. Good for Rush's reputation, but not so good for the patients. Instead of dying from Yellow Fever, patients died from anemia, diarrhea, vomiting, dehydration, skin infections, mercury poisoning, AND Yellow Fever.
It's been speculated that George Washington, the founder of our nation, was a victim of Heroic Medicine. Not that surprising. If anyone is going to be offered an aggressive new treatment - it's the rich and famous. In fact, doctors with big egos, big ideas, and big treatments, love the rich and famous. The rich and famous can fund their experiments, partake as Guinea Pigs in their experiments, and provide them the greatest gift a doctor can get - a celebrity anecdote and endorsement. At the age of 67, and otherwise in good health, General Washington developed a respiratory tract infection while at his home in Mount Vernon. It started with a sore throat and hoarseness, but quickly progressed to shortness of breath and fever. His doctors, who were big proponents of Heroic Medicine, bled over 80oz of blood from his body (40% of his total blood volume). He was given emetics to induce vomiting, laxatives to induce diarrhea, and Spanish fly to cause massive blistering of his skin. All of this was to “draw out the bad humors”. To this day, historians and doctors are unsure whether he died from the respiratory infection, or the “Heroic“ measures. I guess his doctors never got the celebrity endorsement they were looking for. No problem, I am sure they blamed his death on the disease.
It's not surprising that medical reformers like Oliver Wendall Holmes said in 1860,
“I firmly believe that if the whole materia medica as now used, could be sunk to the bottom of the sea, it would be all the better for mankind and all the worse for the fishes.”
This all changed in the 1700's, with the enlightenment. Thinkers, like John Locke and David Hume, demanded more than deductive reasoning. They started to rely on inductive reasoning as the main source of knowledge; the careful observation of nature through experiments. The age of empiricism began. In medicine, the great tool of empiricism was the clinical trial. Give some patients a treatment, and other no treatment, and see who gets better. Unfortunately, it would take another 300 years until it was perfected. The clinical trial, as we know it today, was not developed until the 1940’s.
The clinical trial is the ultimate medical invention. It is the invention that can test all the other inventions. The clinical trial not only informs us whether a treatment works, but also whether it causes harm. Once we realized we needed clinical trials to prove a treatment is safe and effective, everything changed. Yet, many patients and doctors, to this day, are uncomfortable waiting for trials to be completed before using new treatments. I would like to review a few recent and dramatic examples:
ALS (Lou Gehrig's disease) is a universally fatal disease of people in their prime. There is no treatment and hundred of treatments have failed. Regeneron developed a drug call BDNF for ALS in the 1990's. Anecdotally, and in uncontrolled trials, patients treated with BDNF were doing exceptional. They were "throwing away their canes". In fact, the company was publicly chastised for not allowing ALS patients compassionate use of their drug before the large trials came out. Finally in 1999, the large controlled trials came out, and showed the drug did no better than placebo. We were fooled, once again, by anecdote and uncontrolled trials.
Bone marrow transplant was a promising treatment for metastatic breast cancer in the 1990's. It was a very aggressive procedure with many horrible adverse effects; however, it allowed the oncologist to give very high doses of chemotherapy to kill the cancer. The first controlled trial from South Africa showed that it was working. Patients picketed outside insurance company headquarters demanded they pay for this expensive and dangerous procedure. However, all the subsequent trials, performed at other locations, failed. It turned out, the data from the South African study was fraudulent. Thousands of women had to suffer a death worse than metastatic breast cancer, a death from both metastatic breast cancer and a harmful bone marrow transplant.
The reason many patients die after suffering a heart attack is a fatal arrhythmia. Flecainide is an anti-arrhythmic drug that can suppress arrhythmia. If given to a patient after a heart attack, it could potentially prevent death. Based on this physiologic reasoning, cardiologists started using the drug in the 1980's, before controlled trials could confirm this hypothesis. When the clinical trials were finally published in 1990, it turned out that Flecainide INCREASED the risk of death. In fact, it has been extrapolated that between 1980 and 1990, we killed over 100,000 patients. All of this because we couldn't wait for the clinical trial to be completed.
I would like to emphasize that the preceding examples are the rule, not the exception. We are mostly fooled by fancy physiology, preliminary results, and anecdotes; furthermore, some of these treatments lead to great harm. With a track record like this, its easy to see how patient safety experts like Brent James MD have said,
“For most of human history doctors have done more harm than good.”
Should I take COVID-19 treatments before clinical trials are completed?
The answer that comes immediately to mind is, "Yes. COVID-19 is deadly, What do I have to lose?". But, this is not case. I will attempt to offer a benefit-to-harm analysis of one of the untested medications, hydroxychloroquine. It is currently the most popular and available drug for COVID-19.
Hydroxychloroquine: Assessing benefit
Let's try to calculate the probability of benefiting from hydroxychloroquine p(H) - and by benefiting, I mean preventing death. We will need to calculate the following numbers:
p(P) - Probability of a positive trial - In a prior email, I estimated a 5% likelihood of hydroxychloroquine working based on a systematic review and a Bayesian analysis. To clarify, this is the likelihood of hydroxychloroquine working in a large randomized clinical trail.
To figure out how this impacts you as an individual, we need to calculate some additional numbers...
p(E) - Effect size - When a treatment works, it does not work in 100% of cases. No drug is that good. In fact, the average effect size of most drugs is about 20%. This means that hydroxychloroquine would reduce the risk of death by 20%. If five patients were going to die of COVID-19, hydroxychloroquine would save one life.
p(D) - Probability of death - Calculating your probablity of death helps us determine how much benefit you will gain from taking hydroxychloroquine. Most patients with COVID-19 will not die. If a hundred people get COVID-19, about one out the hundred will die. Assuming all of them took hydroxychloroquine, only one of them actually benefited from it. Even if hydroxychloroquine was 100% effective, we would have to treat 100 patients with hydroxychloroquine to save one life. This is the NNT - number needed to treat. The other 99 patents would be exposed to the adverse effects of hydroxychloroquine with no benefit. We can start to estimate your probability of death by looking at the COVID-19 case fatality rate; estimates range from 0.033% to 5%. Estimates can be personalized by taking into account: the stage of the disease, your age, and your pre-existing conditions. For example, A 20 year old that recently developed a fever has approximately a 0.1% chance of death. The NNT in this situation is 1000. We would have to treat 1000 patients like her to save one life. An 80 year old in hospital on a ventilator has a 50% chance of death. The NNT in that situation is 2.
Combining the above 3 probabilities yields the probability of benefiting from hydroxychloroquine p(H)
p(H) = p(P) x p(E) x p(D)
20 year old with fever p(H) = 5% x 20% x 0.1% = 0.001%
80 year old on a ventilator p(H) = 5% x 20% x 50% = 0.5%
Hydroxychloroquine: Assessing harm
Compassionate use vs off-label use - Compassionate use refers to drugs that are new and experimental, and have not been tested for basic safety in humans. Examples of compassionate use for COVID-19 include: remdesivir, vaccines, or synthetic antibodies. Off-label use refers to the use of drugs that already have applications in other disease, but not yet proven to work for COVID-19. Examples of off-label drugs for COVID-19 include: Avigan, Kevsera, Kaletra, Ivermectin, Zithromax, and Hydroxychloroquine. The advantage of off-label drugs over compassionate use drugs is that off-label drugs already have a record of relative safety in humans. As such, we know that Hydroxychloroquine is a relatively safe drug.
Reported adverse effects of hydroxychloroquine - These include: QT prolongation, torsades de pointes, bone marrow failure, liver failure, nausea, vomiting, diarrhea, hypoglycemia, headache, convulsion, and psychosis.
Probability of reported adverse effects - It is common to experience nausea, headache, and other non-serious adverse effects of hydroxychloroquine. The more serious adverse effects, like cardiac, liver, and bone marrow, are much less common. Yet, with such a low probability of benefit from hydroxychloroquine, a small risk of serious harm is important. Unfortunately, these rare but serious adverse effects are difficult to accurately quantify. For instance, we know that hydroxychloroquine can cause sudden cardiac arrest and death; this is documented in hundreds of case reports. But the true probability of this happening is unknown. Hydroxychloroquine is an old and infrequently prescribed medicine. There are no clinical trials with tens of thousands of patients, and trials of that magnitude would be necessary to detect a one in thousand, or one in ten thousand adverse reaction.
Patients at risk of death from COVID-19 are the same patients at risk for hydroxychloroquine adverse effects - Most patients that die with COVID-19 have other medical co-morbidities - these include: cardiac issues, liver issues, bone marrow issues, and advanced age. Ironically, these same patients are most likely to suffer from the adverse effects of hydroxychloroquine as hydroxychloroqiune can potentially aggravate these very issues. This group of patients is also known to take multiple medications, and many common medications interact with hydroxychloroquine.
Confusing death from COVID-19 vs death from hydroxychloroquine -
A common misinterpretation of hydroxychloroquine in uncontrolled trials is if the patient died, they died from the disease, if the patient survived, they survived because of the given drug. This is not true. For example, a common way a for a sick patient in an ICU to die is a cardiac arrhythmia. If a terminally ill COVID-19 patient on hydroxychloroquine dies from an arrhythmia, was it the final stages of the COVID-19, or was it the QT prolongation of hydroxychloroquine? It would be impossible to differentiate drug-related adverse effects from the disease manifestations in the absence of a control group.
Potential for worsening the disease - Hydroxychloroquine is an immune suppressant. Although it has been hypothesized that an overactive immune system is the real culprit of COVID-19 disease, it is possible this hypothesis is completely wrong. In fact, an immune suppressant could potentially worsen any infection. Hydroxychloroquine increased the viral load in HIV and Chikungunya. Other immune suppressants, like prednisone, actually increase mortality in influenza pneumonia.
I find it almost impossible to generate a number for the probability of harm by hydroxychloroquine. However, it is not negligible - that is for sure. Even if it is a very small number, its probably not smaller than my above calculations of benefit.
Ultimately, this exercise demonstrates an important principle of medicine: the benefit of an unproven treatment, with a small effect size, in a population with a low risk of death, is very low; it is easily outweighed by the small risk of a serious adverse effect. (The use of Aspirin in healthy 70 years olds to prevent heart attacks and strokes is another example of this principle and covered in this email). If you have a low risk of dying from COVID-19, you will not likely benefit from hydroxychloroquine, and the harms of the drug outweigh the potential benefit. I you have a high risk of death, the conversation is more nuanced. Before you take any treatment, please call me to discuss your specific situation. We can calculate a benefit-to-harm analysis as it pertains to you.