Hydroxychloroquine and azithromycin for the treatment of COVID-19–Review of study by Didier Raoult

Hydroxychloroquine has received lots of attention lately mainly due to one trial coming out of Marseille, France, but is the hype really warranted? Lets take a closer look. President Trump has been partly responsible for creating the hype in this tweet. He said hydroxychloroquine and ASA Thor Meissen taken together have a real chance to be one of the biggest game changers in the history of Medicine. The FDA has moved mountains. Hopefully they will both be put in use immediately. People are dying, move fast. Now, that’s as you will see, a very unfounded and dangerous recommendation. So that’s the paper by Professor DDA volt and co-workers from Marseille, France, hydroxychloroquine and ASA Thomason as treatment of Kovach 19 results of an open-label non-randomized clinical trial. So here’s what they did. They recruited 26 patients into the intervention group. They were all recruited in their own Center in Marseille and 16 controlled patients who were recruited in Nice, Avignon, Seoul or ma se. They all tested positive for the virus at baseline. They then either received 200 milligrams of hydroxychloroquine. Three times daily, with or without a serum isin or no treatment, the study’s outcome was the clearance of the viral RNA from nasopharyngeal swabs that were tested with PCR on days. One through six PCR basically looks for viral RNA in the specimen. Now, ideally, we would like to test the drug’s efficacy using treatment randomization, where it’s decided by chance who gets the drug and who doesn’t also in the ideal world patients who don’t get the drug should receive a placebo and all participants. Patients and doctors should be unaware of who gets what that’s called blinding. Now we are in an emergency pandemic situation and we need insights fast, so it’s understandable that a formal, randomized, controlled double-blind at trial was not formed, but even if we put those requirements aside, the study has a couple of major flaws. What we would like to see in a trial like that, that’s done, quote-unquote, quick and dirty is to balance groups. We would like to see at least some type of matching of participants on important variables like age, gender, Center disease, severity and disease duration. However, when you look at the age distribution, you can clearly see that patients in the hydroxychloroquine arm were much older than controls. There was also a higher proportion of male patients in the hydroxychloroquine group, whereas there were more asymptomatic patients in the control group as compared to the intervention group. So it really looks like the intervention group is at a significant disadvantage. Now one could argue that if we see a positive treatment results when the intervention group starts out with more severe disease, then the actual effect of the drug is even more impressive. But that’s not the case here, as you will see here, are the patients taken from the papers supplementary Table one in order to give you a better overview, I marked the asymptomatic patients, green patients with upper respiratory tract infections, yellow and patients with lower respiratory tract infections, Red so from less severe to more severe disease, and you can see that the hydroxychloroquine group seems to be much sicker. This to me suggests that the hydroxychloroquine group might actually have more advanced disease than the control group, as we move on you’ll see why that is important. Also, there were a couple of very sick patients who were removed from the intervention group. Three patients had to be transferred to the intensive care unit. One died, one left the hospital and one left. The study due to nausea now in a proper analysis, called intention-to-treat analysis, which they could have done. These patients would still have to included in the overall endpoint efficacy calculation in this paper they were just simply removed. Now, as you can imagine, removing the very sick from the treatment group could make the final results look much better than they truly are. So this dropout of sick patients is problematic. So, to reiterate, the hydroxychloroquine group is older, has more male patients have less asymptomatic patients and might be further along the disease trajectory than the control group. So here’s how they assess their chosen outcome. The presence of viral RNA in nasopharyngeal swabs and days one to six after the start of the study seen over here in the six right most columns, they used cycle threshold to see if a test was positive or negative. The cycle threshold is the number of cycles to be run for the PCR test. To turn positive in these analyses, negative was defined as a CT value above 35, the lower the number, the more virus that is present and if the number is higher, the less viral RNA is present. Now this all looks a little messy, so I colored all positive tests where the virus is present in red. All negative results were there was no virus in green and all tests that were not done in gray. On the top. You see the control patients on the bottom, you can see the hydroxychloroquine patients and what becomes obvious immediately is the high number of incidents where no tests were done in the control group. What’s up here now, here’s what they actually compared in their final analyses the proportion of negative tests on the various days between the different groups. This proportion was calculated as number of negatives, divided by total number of patients in the group on that specific day. But if you don’t test in a group on a given day, then the people who don’t get tested have no chance of showing up in the numerator but they’re showing up in the denominator. So here you see what they did. They took the number negative patience and divided them by the total number of patients in that group on that day and displayed these numbers here and here so 50 % negatives in the hydroxychloroquine group on day three versus six point: three percent negatives in the control group And here are the numbers for the other days, let’s take days three, for example, 50 versus 6.3 percent. Lets look at day three here! Well, if there are so many tests that are not performed, obviously, there’s no way to detect the correct proportion of negatives. That’s a big big problem of this study. Also, let’s look back at the patients who dropped out these two here tested negative when they left the study, but the three individuals going to the ICU were tested positive when they left, as did the patient with nausea. These four individuals, if they had been included in the analysis, as would be a standard thing to do in an intention-to-treat analysis, would actually make the intervention group look worse than it does right here. Furthermore, in our research we found that there were two different documents of the study available online, one on a pre-publication server and the other one on the institutional website of the authors. Here they are in the browser, one on the pre-publication server and the other one on the institutional website. When we compared the supplementary table, one of these two versions, we found significant differences in the reported data on the left. You see the version on the pre-publication server on the right. You can see the version on the institutional website. Lets compare the columns one by one. There are already differences that can be seen here. Difference is here here here also here here and here. How can this be? You can check for yourself just a note on the side. This is the status as of March 24. 20:21 things I haven’t mentioned yet is that these folks down here received a Sethe or Meissen on top of their hydroxychloroquine and according to the authors, fared even better, when looking at the proportion of negatives on base three to six, we can see that up to a Hundred percent have cleared the virus, I’m not exactly sure what the rationale was behind giving both drugs in combination, assuming that they would have added benefits, but it seems to me like a very risky endeavor. These two compounds are known to prolong the QT interval, which in turn can cause severe rhythm problems known as ventricular tachycardia seen here. So it’s not a great idea to recommend widespread use of both drugs in combination. Mr

President, so let’s recap the major problems of the study we’ve encountered so far, which groups not being comparable, incomplete testing in the control group, patient, dropouts, inconsistent reporting and a potentially dangerous combination of drugs. But we haven’t even touched upon the biggest problem of all, and that is an incomplete understanding of the disease, pathophysiology and course on the part of the investigators. So what was actually defined as the baseline of the study? Ideally, that should have been the very beginning of symptoms. However, the baseline in this trial was defined as the inclusion of patients into the study, so the disease and the participants had already started prior to day zero of the study. Now we said that the hydroxychloroquine group looks as if it were already further along the disease trajectory based on primarily disease severity in that group, so it looks like their disease might have started earlier than the control group on average. Now we know from biologic studies that far slowed in the nasal pharynx goes down in almost all patients in the first 10 days of symptom onset, even in those who later develop severe disease. As seen in this page. This is an example of a patient that developed pneumonia. As you can see, the viral load in the nasal pharynx goes down, whereas the viral node in the sputum remains high, because the sputum comes from the lungs and therefore better reflects the true disease process. So if our load goes down in the nasopharyngeal swabs in all patients, irrespective of disease severity than testing, one group at a later point than the other group will always show a certain difference in viral load. Irrespective of what drug you give them what’s happening in the nasal, pharynx has nothing to do with the severity of the disease. Testing for viral RNA in the sputum would have been much preferred and obviously, what would be even better is to have hard clinical outcomes like recovery rate duration from start to finish of symptoms or a regression of changes. Seen on the CT scan. These are the things we really want to impact with treatment, not the viral load in the nasal pharynx. So, in summary, I would say that this study adds very little to almost nothing to a current understanding of how to treat Kovach 19 patients and it’s pretty dangerous to have a president who endorses the widespread use of a drug based on evidence like this. We will need much more solid data than that. Please note that we’ve just recently opened up our entire library of courses to hospitals that are taking care of COBIT 19 patients. So if you want to learn mechanical ventilation and other important ICU skills, please make sure to setup your institution with a free group subscription