Flawed Coronavirus Studies

Flawed coronavirus treatment studies are being promoted by the mainstream media as a way to disparage the effective treatment of hydroxychloroquine.

Among the flaws in these studies are:

• Preferentially using hydroxychloroquine to treat the sickest patients (when compared to the control group)
• Faked Data
• Giving hydroxychloroquine in toxic doses
• Delayed initiation of treatment
• Varied initiation of treatment
• A lack of testing for SARS-CoV-2 in outpatients
• Not having a large enough sample size to achieve statistical significance.

Depending on which study the media cites to discourage the use of hydroxychloroquine, the studies have one or more of these flaws.

The VA study[edit]

Outcomes of hydroxychloroquine usage in United States veterans hospitalized with Covid-19, also known as the VA study, was one of the first studies to suggest that the use of hydroxychloroquine could be harmful. The paper's results said there were 27 deaths (27.8%) in the HCQ group, 25 deaths (22.1%) in the HCQ + azithromycin group, and 18 deaths (11.4%) in the no HCQ group.^[1] The first flaw was that it was a small study. It only had 368 patients and small studies are more prone to errors than larger ones. The second flaw was that it delayed treatment with hydroxychloroquine... the drug works best when it is given early in the disease progression. The third and most egregious flaw was that it was giving hydroxychloroquine preferentially to the sickest patients.

If one gives hydroxychloroquine preferentially to the sickest patients when compared to the control group, then the fact that more patients died when they were given hydroxychloroquine should be unsurprising.

Lancet study[edit]

Hydroxychloroquine or chloroquine with or without a macrolide for treatment of COVID-19: a multinational registry analysis^[2], also known as Surgisphere, was one of the next studies to suggest that hydroxychloroquine was harmful. According to its authors, hydroxychloroquine had a 18% mortality rate when used to treat COVID-19 and the mortality rate raised to 23.8% when it was combined with an antibiotic, whereas the control group only had a 9.3% mortality rate. The main problem with this study was that it was a complete fraud. The following is a partial list of its flaws.

• No reproducibility^[3][4]
• No traceability of the data^[3][4]
• No transparency of the review^[3][4]
• Data was too "clean" to have occurred naturally^[5][6]
• Impossible to collect the data in the way the researchers claimed without violating privacy laws^[7][6]
• Not enough authors to conduct a study of its size^[7]
• Too many Australian deaths... it's data claimed to have recorded 73 Australian deaths by April 21 but the number of COVID-19 deaths in Australia didn't reach 73 until April 23.^[7]
• Not adhering to standard practices in the machine learning and statistics community^[4]
• Implausible ratios of chloroquine to hydroxychloroquine use^[4]
• Mean daily doses of hydroxychloroquine that are 100 mg higher than FDA recommendations.^[4]

The RECOVERY Trial[edit]

One of the most commonly cited studies by the media is the UK's RECOVERY trial. This trial gave the drug in hospitalized patients which is a bit late to be giving the drug, although studies such as the ones conducted at Mount Sanai hospital and the Henry Ford hospital system have shown that hydroxychloroquine can reduce the number of deaths in those patients by about 50%. The RECOVERY trial's biggest flaw was that it used toxic levels of the drug... the maximum recommended dose for hydroxychloroquine is 800 mg per day^[8] and the RECOVERY trial used a daily dose of 2400 mg^[9], a dose that is three times higher!

It goes without saying that when you give the wrong dose of a drug, bad things happen. Perhaps the only surprising thing about this trial is that the hydroxychloroquine deaths weren't much higher than the people who weren't given the drug.

Observational Study of Hydroxychloroquine in Hospitalized Patients with Covid-19[edit]

This study authored by Joshua Geleris et. al. is a retrospective study commonly cited in the media that analyzed patient data from the New York–Presbyterian Hospital (NYP)–Columbia University Irving Medical Center (CUIMC). The following are inherent problems in this study.

• Hydroxychloroquine was given late.^[10]
• The treatment with hydroxychloroquine varied depending on which doctor was seen.^[10]
• It was given to sicker patients preferentially (when compared to the control group).

Association of Treatment With Hydroxychloroquine or Azithromycin With In-Hospital Mortality in Patients With COVID-19 in New York State[edit]

This study authored by Eli Rosenberg et. al. is another retrospective study commonly cited by the media. It analyzed the data from 25 different hospitals in the New York area. The following were problems with this study.

• The treatment with hydroxychloroquine varied depending on which doctor was seen.^[11]
• It was given to sicker patients preferentially (when compared to the control group).

A Randomized Trial of Hydroxychloroquine as Postexposure Prophylaxis for Covid-19[edit]

This is one of two studies that come from the University of Minnesota that are commonly cited by the media to suggest that hydroxychloroquine does not have a benefit when it is used early. This are several reasons to doubt that conclusion, one of which is that the study was approaching statistical significance when one looks at the patients who received the drugs within two days of their exposure. The other reasons to doubt this study are as follows:

• Only 3% of patients tested positive for COVID-19.^[12][13][14]
• The data was self-reported (as opposed to being recorded by a researcher).^[13][14]
• The treatment was delayed as it was being delivered through the mail.^[13][14]
• The sample size might not have been large enough to achieve statistical significance (see the next study for a more thorough explanation).

Hydroxychloroquine in Nonhospitalized Adults With Early COVID-19: A Randomized Trial[edit]

This is the second of the two studies that come from the University of Minnesota. The following are its flaws:

• Only 58% of patients tested positive for COVID-19.^[15]
• The data was self-reported.^[15]
• The treatment was delayed as it was being delivered through the mail.^[15]
• The sample size was not large enough to achieve statistical significance.^[15]
• The assessment "overall symptom severity" was subjective and not rigorous.^[15]

During the course of this study, the researchers discovered that the low event rate of hospitalization and deaths among the participants meant that the trial would have required 6000 participants in order to achieve meaningful results and the study only had 491. Consider the following example in order to understand the nature of the problem. Lets say that 1 out of 100 people infected with SARS CoV-2 will need to be hospitalized, the study has 400 participants evenly divided between the control and placebo groups, and hydroxychloroquine is known to be 50% effective. The study is executed and 2 people in the control group were hospitalized whereas only 1 person in the treatment group was hospitalized. One could argue that this confirms the hypothesis, but anyone familiar with statistics would tell you that that is not correct. The difference between 1 and 2 people being hospitalized could have easily happened due to chance and thus the difference is not statistically significant. The only way to achieve statistical significance would be to increase the sample size, in this example to 6000 people, which would result in 30 people being hospitalized in the control group and 15 being hospitalized in patients treated with hydroxychloroquine. That wasn't done and so the results are meaningless. What was done was analyzing whether or not one self-reported having a cough after 14 days in both the placebo and treated groups and then trying to use that to determine if the treatment was successful or not.^[15] That, to be perfectly frank, is not science (unless one was trying to evaluate the effectiveness of cough medicine... even then one would need a larger sample size).

Hydroxychloroquine for Early Treatment of Adults with Mild Covid-19: A Randomized-Controlled Trial[edit]

This is a study conducted in Spain and authored by Oriol Mitjà et. al. It only had 293 participants which is far too small of a sample size to achieve statistical significance.^[16]

References[edit]