Our president is going to get people killed. I don’t say this lightly. His actions today fall nothing short of reckless. Earlier today, President Trump tweeted:
— Donald J. Trump (@realDonaldTrump) March 21, 2020
Yes, calling for the FDA to act immediately to approve the drugs hydroxychloroquine and azithromycin for COVID-19 treatment with very little evidence. This is dangerous.
I think it’s appropriate to disclose here that I am not a virologist. I do have 20+ years experience with clinical trial design though, and I wondered the source of Trump’s enthusiasm. It took me about two minutes to find a link to a paper on Google Drive that seems to have inspired this latest madness…
It is very possible there is some promise to hydroxychloroquine and azithromycin, but this study is not enough to conclude that. It’s a basic, observational study and highly problematic, at best. Here’s why…
This team of investigators studied patients of three sorts:
- Patients from their center, who received hydroxychloroquine (H) (n=20)
- “Control” patients from another center who served as a control (n=16)
- Patients from their center who received H and the antibiotic azithromycin (A) (n=6, but really n=3)
The authors’ major conclusion is “that hydroxychloroquine treatment is
significantly associated with viral load reduction/disappearance in COVID-19 patients and its
effect is reinforced by azithromycin..” This is not the conclusion any reasonable scientist would reach and is inflammatory (pardon the pun) at best. The findings are titillating but not conclusive.
Azithromycin is an antibiotic used commonly to treat respiratory tract infections. Many of us have had our doctors issue us a Z-pack for bacterial infections. There has been a lot of debate this morning about the anti-viral properties of A. There’s interesting evidence that A has both anti-inflammatory and anti-viral properties that are secondary to its anti-bacterial properties. While those are interesting, they are not at all why patients in this study received A and I will limit my discusison only to the authors’ manuscript and conclusions.
Allow me to break down some facts and fictions related to the use of H & A in COVID patients as it related to the paper and Trump’s comments.
Fact: This was not a randomized, controlled clinical trial. The holy grail of medical research is the double-blind, randomized clinical trial where neither the investigator nor the patient knows the treatment being given. One could argue that viral titers might not be subjected to bias, but that’s not clear. Even without malicious intent, it is still possible for investigators to introduce unconscious bias when processing samples when they know the origin of those samples. They might process them slightly differently if they know the origin. That’s not to impune the integrity of the investigators. We’re all potentially at risk for bias and it’s why the most meaningful studies come from blinded research. It is not clear from the paper whether the investigators were blinded to the origin of the sample. In every human study I have ever participated in, if I had to review or process data, I was blinded to the origin of the sample. I make my students practice the same way. It helps keep everyone’s personal stakes out of the data.
The control group is also not a control group, at all. They’re from a different center and it is not clear why they were selected. In an ideal design, all of the patients would have been treated by the same team at the same location and would have been matched for age, sex, days of symptoms, and comorbidities.
Fact: Azithromycin was not given to patients with the intent to treat COVID. In both the H and H&A groups, participants were much more likely to have a lower respiratory tract infection or be symptomatic. This may mean that they were sicker longer. If that’s true, the virus may clear quicker because that’s the natural course of the disease. We don’t know if the H and H&A groups were treated differently in other ways because they were more likely to be symptomatic. At the very least, because the trial is not randomized, we cannot eliminate the possibility that chance is the biggest contributor to these findings.
Fact: Patients that were lost to the ICU or died were not included in an intent to treat analysis. In well-conducted clinical research, investigators perform an “intent(ion) to treat analysis.” That means that you analyze everyone enrolled in the study, regardless of whether they complete the trial or not. This allows the effects of death, non-compliance, cessation of treatment because the side effects are intolerable, etc, to be included and for the impacts of a treatment to be fully considered beyond a narrow group. It gives you more of the drug’s true effect, not simply the effect in the best, most compliant patients. In the H group, participants who stopped taking the drug, died, or went to the ICU were not included. This is hugely problematic because the criteria were not applied uniformly and the outcome in the H group may be the result of only including very robust participants. Patients that are dead or in the ICU and not monitored may have higher viral loads that would not be reflected in the data if these participants are eliminated.
These data should be viewed with absolute caution. These data are from a small sample of potentially handpicked patients with no clear criteria for why they were included in the analysis. Why these 16 controls? Why these 6 H&A patients? Be skeptical…
Fact: H&A is not FDA-approved for the treatment of coronaviruses. The president has made comments recently about these being FDA approved drugs. They are FDA approved, but for other indications. They are not approved for use in coronavirus. It is still not known if they are safe and effective here.
Fact: H&A is not universally safe. In cases where H or A are approved for use, part of the deliberation is whether the benefit of the drug outweighs the side effects. For current indications, this is true. It is not known whether it is true in coronavirus. Both of these drugs are associated with cardiovascular events, including something called long QT syndrome, in patients with existing cardiovascular risk. The electrocardiogram (ECG) was measured in patients receiving H&A, but no data are reported. Long QT is an arrhythmia of the heart that is commonly seen in the media because child and teenage athletes can die unexpectedly from it. It can be fatal. It is not known where the risk-benefit of using these drugs lies and whether it is safe to use them synergistically. Safety is at the core of the FDA’s mission. Other drugs that have been presumed to be innocuous, that we have used as a “last resort” because people were dying, have taught us that we can’t extrapolate the safety or efficacy of a drug for every indication.
Fact: By giving the public hope that there is a cure on the immediate horizon tomorrow, the president’s words may encourage the public to ignore public health measures. Why would you need to social distance or self-isolate if there is a cure? The president’s words may encourage the public to eschew common sense warnings from infectious disease experts. The current public health warnings are based heavily on quality data. Recommendations based on this study? Not so much…
Fact: Reckless dialogue from the president may cause a shortage of medications that have a lifesaving purpose for other indications. Can any of us get toilet paper right now? Fear has led people to stockpile what they believe to be “essential goods.” Thing is, toilet paper is not really essential. As I have reminded several people recently, as long as you can still wash your backside, toilet paper is not essential. A luxury in the end times, but not essential. H & A are actually essential for the treatment of life-threatening diseases like lupus. There are already reports of shortages of these drugs. This means that the people who may need them most cannot get them
Fact: Publishing this study may not be ethical. The authors state “For ethical reasons and because our first results are so significant and evident we decide to share our findings with the medical community, given the urgent need for an effective drug against SARS-CoV-2 in the current pandemic context.” Sharing research is always ethical, but not being transparent about the parts that are problematic is not. Not posting it to a platform where the scientific community can comment is not ethical.
If anything, this study reminds us of the importance of peer review and the danger of over-interpreting preliminary data. Are these data worth a follow-up? Sure. Are they enough to change our current treatment approach? Absolutely not.