Over the past few weeks I have had several people I know tell me about a “cure” for COVID-19. They have sent me links of YouTube videos and Facebook posts of “America’s Frontline Doctors” (they’ve eared the quotations) discussing hydroxychloroquine, azithromycin and zinc as the magic cocktail to cure COVID. I have yet to discuss the studies of hydroxychloroquine and azithromycin, which I plan to do in future posts. They claim that if it were more widely utilized, at the “right time” and in the “right patient” we could cure COVID.
I always thought YouTube was designed mostly for entertainment. You know, watching people fall off a skateboard, or a group of silly musicians dressing up like Disney characters to perform Welcome to the jungle. Facebook was designed for looking at pictures of your best friend’s new puppy. These two platforms sure have come a long way. Content creators spread more invalid information than ever before. Possibly inducing a significant amount of anxiety, turmoil and degrading our intelligence. As much as I enjoy seeing memes of the crazy woman yelling at smudge the cat, I could go without 99% content that shows up in my news feed.
Anyways, onto the study.
This was a retrospective cohort performed at NYU Langone Health system. Retrospective means they looked back over data. Cohort indicates they created two groups based on exposure to the independent variable, in this case zinc. Then they looked for different outcomes such as death and need to be placed on a ventilator and compared the occurrence of these events between the two groups. Patients were included if they were admitted to the hospital and received hydroxychloroquine and azithromycin. To be clear everyone analyzed received these two drugs. They were excluded if their discharge disposition was unknown, or they received another investigational drug.
Review of Methods
If you’ve read my previous posts, you may have a general understanding of scientific appraisal. Retrospective cohorts are a favorite of mine because they often demonstrate the pitfalls of analyzing data without controlling for variables that impact results. Scientific experiments are really all about control of the unknown.
For the results of a retrospective cohort to be valid the risk and circumstances for exposure need to be equal in both groups, the methods and circumstances for detecting the outcome needs to be equal in both groups, and an assessment for confounding needs to be performed.
In this design, the circumstances to be exposed to zinc are unknown. Based on the fact that the study exists and there are two groups, some patients received zinc, and others did not. The authors did not discuss any protocol established by the health system to administer zinc. Nor do they provide any explanation for the use of zinc in COVID patients. Doctors gave some patients zinc, and others no zinc. Based on this alone all control of the study is lost. Bias is introduced potentially confounding the results. Doctors may have given patients doing “ok” zinc in the thought that they could prevent them from deteriorating, while those that weren’t looking good they avoided zinc, or vice versa.
This is why randomization and blinding are so important. It levels the playing field. It equalizes the prognosis of the groups from the beginning to control investigator bias. The fact that there is no randomization, no protocol, or even a date range for the use of zinc basically invalidates these results. Also, we don’t know if the patients received zinc upon admission to the hospital or was it much later. There is more to learn, so I’ll continue.
Next, were the circumstances and methods for detecting the outcome equal? The methods for detecting the outcome was equal in this case. They simply extracted the outcome from the electronic record. The circumstances in which some of the outcomes occurred may have changed. For example, need for intensive care admission. This is a subjective outcome without much control. “Need” varies from physician to physician. Any resident that has worked the ICU at a major tertiary care center has had an argument with some attending physician about the “need” for ICU care. ICU admission in retrospective studies is not a great endpoint. As our ability to treat COVID has improved there has been less utilization of ICU beds and more admission to general medical floors. Also, discharge to hospice is an endpoint with quite a bit of variability. Patients go into hospice for a number of reasons. The circumstances may be unrelated to the severity of COVID. Also, if a patient or a patient’s family is considering hospice as a disposition the doctor probably isn’t going to throw some zinc at them before they go.
The authors did perform an assessment for confounding, which swung much of their results in and out of statistical significance. I will try to elaborate more on this in the discussion.
Like most retrospective cohort studies, the control for bias and confounding variables is lost. The ability to draw definitive conclusions from the study is nil. It is the bane of retrospective analysis. That does not mean it isn’t thought provoking, or worthless. Let’s move onto the results.
Results
They compared 411 hospitalized patients that received zinc to 521 that did not. All patients received hydroxychloroquine and azithromycin. Zinc was associated with a decrease in mortality or transition to hospice with a confidence interval of (0.36 – 0.73), and reduction in need for mechanical ventilation with a confidence interval of (0.35 – 0.89). This is equivalent to roughly a 50% reduction in the odds that the event would occur. Statistical significance was lost after excluding non-critically ill patients. There was no benefit in critically ill patients.
March 25th was the day zinc therapy was started within the health system. After adjusting for this date, the statistical significance remained for discharge home and decreased mortality/discharge to hospice, however statistical significance was lost for other outcomes. Again, this result was driven by non-critically ill patients.
This is not stated in the results so I’ll state it here. The proportion of patients in the no zinc group in the ICU was 16% compared to 9% in the zinc arm. Bias?
Discussion
The reported results in this case are striking. They report significant reduction in death, need for mechanical ventilation, and likelihood to be discharged home. There’s an old adage that goes: “If it sounds too good to be true, then in probably is.” Sad to say this is most likely the case. The absence of randomization, and the change in statistical significance after assessment for confounders, and the larger proportion of critically ill patients in the control group invalidates the results. Zinc may be helpful, but yet again it was not proven in this study. Its very hard to prove anything in an observational study. The authors even say this! They admit confounders played a significant roll and further investigation is required. They stated it shouldn’t be considered the standard in COVID treatment! Yet, doctors are doing it. I'm ashamed.
Randomization is so important. Its principles need to be considered even in a retrospective cohort. Randomization is the random assignment of participants to either the control of experimental group. The sequence prevents investigators from stacking the deck by placing sicker patients in one group compared to the other. This makes the groups equal in prognosis. In a clinical trial the sequence of assignment is controlled, the timing of the therapy is controlled, other protocols or interventions are controlled, so the investigators do not need to account for so many variables that may impact the results. In a retrospective cohort the investigators need to imitate this process. They must correct for the assignment of patients, the timing of the medication, and inherent bias by providers, and other protocols in care that may have changed. In this study none of this occurred. As you see in the results time to start using zinc, critical illness, and changes in COVID care had a significant impact on the results. We do not know if there was bias in terms of those given zinc vs. those not, but it’s likely given the absence of a protocol.
The other issue is the issue with confounding. The authors deserve credit for running some statistics on different variables to see if they impacted their results. After the analysis some of their endpoints lost statistical significance. This does not only demonstrate that confounding was present, but it proves the results were due to a confounder. In order to disprove this new evidence a prospective trial would need to be run controlling for this effect before the authors can claim their zinc discovery. This is likely the reason the article ended up in an obscure journal. The peer review process actually worked this time. Unfortunately, the cocktail made its way to social media.
On a scientific level, the hypothesis makes no sense. There is almost no zinc deficiency in the US. If you have access to food, you have plenty of zinc. It’s in almost everything. Unless you’re a pirate out to sea for 6 months on a wooden frigate or a hobbit traveling through middle earth eating biscuits, you don’t have zinc deficiency. Your body keeps and stores what it needs and excretes the excess. When zinc is used by a cell to perform some function the zinc does not all the sudden decay into some other element when that function is done, it gets reused. Your amazing body has spent many millennia evolving this system to work to perfection so you don’t have to worry about running out of some essential vitamin or mineral each day. It has checks and balances in the form of proteins designed to store exactly what you need. If you google zinc and immune system, you’ll find one article touting its ability to up-regulate the immune system, and the next article states it slows down the immune response. Your immune system is way too complex to think, “zinc that’s the answer.” It’s never that easy. It most likely gets excreted from your body a few hours after you take it.
You may be thinking…so what’s the harm? If it gets excreted, and doesn’t harm you, who cares if you try it? In my opinion the biggest harm is the persistent erosion of trust between patients and doctors. Stories like this discredit all of us, because the letters in front of our name lend us credibility. If zinc turns out to be another dud, which is likely, patients lose trust in advice from their doctors. Which is the reason why some of my patients take turmeric to treat their cancer, or use grape-seed extract for heart disease. The erosion of trust is causes indirect consequences. This leads to poor, misguided decisions, and then unnecessary pain and suffering.
The big difference between sitting in an office with me discussing your health and standing on the steps of the supreme court is that I have “skin in the game.” I am accountable directly to you. I am accountable to a medical licensing board. There are attorneys waiting in the wings for me to make a mistake that I am accountable to as well. I have a lot to lose when I make a recommendation to you. We’re in this together. That’s why I wouldn’t recommend something to you that I wouldn’t do myself if I were in your situation.
What accountability do these doctors have standing on the steps of the supreme court? None! They are recommending an unproven therapy for a disease we know little about. If you take zinc when you get COVID instead of going to the hospital and die, these individuals are not accountable. They did not make a direct recommendation to you. They would make some excuse along the lines of you not seeking appropriate care. No one could prosecute them, because there is no established standard of care. A licensing board can only investigate them on their actions, not their misinformed words.
They stand on the steps of the supreme court because its easy. It’s easy to stand up, wave your degree around in a white coat, and preach your “knowledge” with strong conviction. But they didn’t do anything to earn that knowledge. So, it’s cheap to them, and they give you cheap advice in dramatic fashion. It’s hard to prove what you say is the truth. If they had gone through the process of proving without doubt what they preach, there would be no need for any speech. The data would speak for itself. They could all go out for a nice meal at Charlie Palmer Steak during their trip to Washington.
In closing, I would like to throw down a challenge to these “doctors.” If you believe so much in your conviction then do what you are supposed to do. Collaborate your practices. Design a randomized controlled trial. Execute it properly. Then present your results. Several of the “doctors” in the video were willing to put their licenses on the line in order to prescribe this cocktail. If zinc doesn’t help anyone, then forfeiting your license is reasonable retribution for the people you misled. That’s accountability. That’s “Skin in the Game.”
Comments