On Chloroquine

This article is written under rather peculiar circumstances, in fact quite dramatic ones to say the least: after a first case declared on November 17^th, 2019¹Josephine Ma, March 13^th, 2020. ‘Coronavirus: China’s first confirmed Covid-19 case traced back to November 17’, South China Morning Post. Available on-line., a COVID-19 pandemic²World Health Organization (WHO), March 11^th, 2020. WHO Director-General’s opening remarks at the media briefing on COVID-19, WHO. Available on-line. has spread throughout the world. Among several consequences, this pandemic caused the confinement of the French population from March 17^th, 2020 at noon. I wrote the following article in France during this confinement. Actually, the seriousness of the pandemic led a group of scientific mediators do launch a coordinate prevention message on March 14^th, 2020 at 6 p.m. I associated myself to this message.

Prior to confinement, on February 25^th, 2020, the IHU Méditerranée-infection released a video stating that a molecule derived from quinine, being chloroquine, is an effective treatment against COVID-19:

This was the starting point for a major media coverage of chloroquine and another derivative from quinine, being hydroxychloroquine. Since this media coverage is extremely problematic, I would like to try to give you some clearer insights. The first goal of this article will be to give you some reliable sources, as quite a lot of uninteresting opinions were given, and too few facts. Yet the facts are there. This article will therefore not contain any opinion.

Organisation of this document

As usual in this journal, following an academic practice, references to printed publications are given in footnotes. For this article, I have been able to provide links to online versions of all these references and I strongly encourage you to consult them. Overall, the footnotes serve to structure the information, but in no way constitute secondary information. As is customary on the Internet, sources published on websites are presented as hyperlinks in the text. These sources are of equal importance. As some events reported here originate in France, some references are in French.

A PDF version of this article is available. An audio version (with quite a robotic voice) is also available.

Understanding clinical trial methodology

Though serious COVID-19 cases require heavy clinical cares, in fact this is these cases that put hospitals under strong pressure, the infection is generally a benign one: more than 80 % of patients spontaneously regain good health. Therefore, a vital question is how to determine if a drug is effective against some illness. For example, if a patient is recovering from the disease, and imagining that, while he or she was ill, he or she ate salsify with butter every day, how can we determine if this dish had any role in the patient’s recovery?

By the way, I think you have guessed it: the reason I chose this example is that there is absolutely no evidence that salsify could be a cure for the disease. It goes without saying, but it is better to say it: there have already been too many premature announcements about this pandemic, let’s not add to it.

How to go beyond the anecdotal stage to determine that a treatment has been identified? Furthermore, there is an important point to consider.

A long-identified problem

In volume one, first book, in chapter XXI of his Essays, Michel de Montaigne reports the case of a merchant from French city Toulouse who suffered from renal colic, and to whom an apothecary regularly administered clysters. However, the apothecary did not actually inject him with anything at all³Michel de Montaigne, 1580. Essais, Simon Millanges editor, Bordeaux, French Kingdom. Available on-line. An English translation by Charles Cotton: Michel de Montaigne, 1877. Essays, William Carew Hazlitt editor, London, United Kingdom. Available on-line.. However, these pseudo-enema did relieve the patient, so that an effect did occur.

These are called “contextual effects”: a treatment is characterized not only by its active ingredient, but also by its entire context, including the expectations of the patient as well as those of the therapist. These contextual effects are variable, but can have a rather dramatic impact and are neither a form of illusion, nor a mental problem: the context clearly has a part in a treatment, so much so that in some cases the active ingredient itself has no additional effect.

Addressing this issue

Therefore, testing a treatment by simply administering it to a patient, telling him there is good hope to cure him or her with it, does not allow determining whether the active ingredient actually brings something more than contextual effects.

To solve this, patients treated with the new treatment are compared to a group of patients that are not, which group is called the control group. However, the context may differ, since in one group patients receive a drug, while they do not in the control group. As a consequence, the contextual effects will vary. To correct this problem, the control group is given a treatment that is identical in all respects, such as shape, colour and odour, to the one tested, but for which there is no active ingredient. This fake drug is known as a placebo, and such a test is a placebo-controlled test.

However, depending on whether patients are told that they are receiving the tested drug or a placebo, their expectations will not be the same, so again the context is different. The patients tested are therefore not informed whether they are given the active ingredient or the placebo: this is a simply-blinded test.

But the context is still not the same: if the person administering the treatment knows whether it is a placebo or not, his or her expectations are not the same and his or her attitude is different, even unintentionally. As a result, the caregiver should also not know whether he or she is administering the active ingredient or the placebo: this is a double-blinding test.

While the context is now the same in all cases, there is one last problem to be solved. Indeed, one can easily understand that if, for example, one group consists mainly of Olympic champions while the other is made up of seniors, the tests will be biased. In order to avoid this problem, many subjects must be available and distributed in such a way that the two groups are comparable. This is achieved by randomly assign the subjects to one group or the other, while ensuring that the characteristics of the two groups are statistically equivalent. This is referred to as a randomized trial.

The need for caution

To properly discern the specific effects of an active ingredient, it is therefore necessary to conduct randomized, double-blind, placebo-controlled or reference treatment-controlled trials. Of course, there is a procedure in case of worsening of the health of a patient participating in the trials, but the overall principle is the one explained.

Contrary to what can be stated here and there, an emergency situation can never justify not following a rigorous methodology: an active ingredient, even after encouraging in vitro tests, may perfectly fail to treat and can even worsen patient’s health. This was already the case for chloroquine when it was tested against chikungunya⁴Franck Touret and Xavier de Lamballerie, 2020. ‘Of Chloroquine and COVID-19’, Antiviral Research 177. Doi: 10.1016/j.antiviral.2020.104762. Beyond chikungunya, after encouraging in vitro tests, chloroquine has already proved unusable in vivo in the case of dengue fever, Ebola, SRAS-3 and HIV, among others.

Chloroquine can cause heart problems, sometimes fatal, and other adverse effects. It is not a substance to be used lightly. In particular, it leaves very little room for manoeuvre, as the therapeutic dose is 20 mg/kg of body mass on average, for a toxic dose at 30 mg/kg and lethal at 40 mg/kg. Also, its half-life is very long, which means that it will take a long time to completely disappear from the body.

Before launching a treatment, it is therefore necessary to determine the benefits it really brings against the risks it poses: this is called the benefit-risk balance, which can only be assessed using a rigorous methodology. It is far too dangerous to administer active substances without having properly determined what the consequences may be.

An inconclusive first study

Shortly after its announcement, the IHU Méditerranée-infection published a study on a trial of hydroxychloroquine as a treatment for COVID-19⁵Philippe Gautret, Jean-Christophe Lagier, Philippe Parola, Van Thuang Hoang, Line Meddeb, Morgane Mailhe, Barbara Doudier, Johan Courjon, Valérie Giordanengo, Vera Esteves Vieira, Hervé Tissot Dupont, Stéphane Honoré, Philippe Colson, Eric Chabrière, Bernard La Scola, Jean-Marc Rolain, Philippe Brouqui, and Didier Raoult, 2020. ‘Hydroxychloroquine and azithromycin as a treatment of COVID-19: results of an open-label non-randomized clinical trial’, International Journal of Antimicrobial Agents. Doi: 10.1016/j.ijantimicag.2020.105949, using it on patients with mild symptoms. I invite all readers to consult it, as doing so at least allows realizing what is and what is not in it.

The standard in science is to discuss each other’s work. On the one hand, this allows on the one hand the dissemination of new results, but also, on the other hand, it facilitates the search for errors. The advent of the Internet has made this commentary process even more effective. Therefore, specialists in the field immediately commented on this article, in particular by using the specialized platform Pubpeer. Once again, I invite you to consult this discussion. Among other things, you will see that, contrary to what we hear, it does not discuss Didier Raoult’s physical appearance, personality, or whether he is inside or outside the system. It only discusses the methodology and statistical quality of this publication.

Incidentally, it is contradictory to say, on the one hand, that Didier Raoult is a unit director and a great professor admired by his peers and, on the other hand, to claim that he is not part of the system. If Didier Raoult is indeed the internationally renowned specialist described by Didier Raoult⁶In particular, one should look carefully at the name of the author of the article about Didier Raoult published on January 3^rd, 2019 on the Club Mediapart blog platform., then he is an integral part of the system. Anyway, his person has no interest in the case. All that matters are the results. Therefore, this article will not discuss him any further.

Precisely, the analysis of the publication shows that the study of the IHU Méditerranée-infection is not significant. Here are the most essential elements:

• This is not a randomized double-blind placebo-controlled study: a group of patients with COVID-19 was treated with a mixture of hydoxychloroquine and azithromycin (an antibiotic) at IHU Méditerranée-infection, and the data obtained were compared with those obtained from patients treated at another institution who received neither hydroxychloroquine nor azithromycin, the method of selection of these patients being unclear.
• The study involves but a few subjects, namely 26. Among them, 6 (or 23 % of the original population) were excluded from the study. Of these, two who stopped treatment or left hospital before the end of follow-up, which changes the statistical characteristics of the group, especially since the cohort originally contained few patients (this is known as attrition bias). Three were transferred to an intensive care unit. Finally, one died. The results seem all the more positive as the disappointing cases, which in this case all belonged to the group treated by the IHU Méditerranée-infection, were removed from the study.
• The used test declares several patients positive to COVID-19 one day, negative a few days later and then a little later positive again. This is one of the specificities of COVID-19 disease: the viral load in the nasopharynx changes over time and it is not uncommon for it to be undetectable in the worsening phase of the disease. Many people are negative in the days preceding their death, which was the case of the person treated at the IHU Méditerranée-infection who died during this study. In any case, referring only to the nasopharyngeal viral load to conclude to a remission is widely criticized.
• The group treated with chloroquine is compared to a group of 16 people followed in another facility, with no guarantee that the protocols for assessing viral load are the same or conducted in the same way.

Clearly, this study does not allow any conclusions as to the effectiveness of the treatment.

A no more conclusive second study

On March 27^th, 2020, the IHU Méditerranée-infection released a preparatory version of an article detailing not clinical trials, but an observation of 80 cases of COVID-19 patients, once again presenting rather mild symptoms, who were administrating a mixture of hydroxychloroquine and azithromycin⁷Philippe Gautret, Jean-Christophe Lagier, Philippe Parola, Van Thuan Hoang, Line Meddeb, Jacques Sevestre, Morgane Maihle, Barbara Doudier, Camille Aubry, Sophie Amrane, Piseth Seng, Marie Hocquart, Julie Finance, Vera Esteves Vieira, Hervé Tissot Dupont, Stéphane Honoré, Andreas Stein, Matthieu Million, Philippe Colson, Bernard La Scola, Véronique Veit, Alexis Jacquier, Jean-Claude Deharo, Michel Drancourt, Pierre Edouard Fournier, Jean-Marc Rolain, Philippe Brouqui, and Didier Raoult, in preparation. ‘Hydroxychloroquine-Azithromycin and COVID-19’. Available on-line.. To my knowledge, as of April 7^th, 2020, this article has not yet been submitted to a journal or a prepublication website. There was also no control group for this study.

Among these cases, 83 % were negative on the COVID-19 test after seven days of treatment, 93 % were negative after eight days.

However, the article does not give every detail of its statistical methodology. In addition, the method for reporting whether patients are positive or negative for the disease changes over the course of the study. Finally, given the population tested and the absence of a control group, this case study still does not allow to draw any conclusion about the effectiveness of the treatment.

A counter-study no more significant

In response to these two studies, another was also made available on March 30^th, 2020, describing the case of 11 patients severely affected by COVID-19⁸Jean-Michel Molina, Constance Delaugerre, Jerome Le Goff, Breno Mela-Lima, Diane Ponscarme, Lauriane Goldwirt, and Nathalie de Castro, 2020. ‘No Evidence of rapid antiviral clearance or clinical benefit with the combination of hydroxychloroquine and azithromycin in patients with severe COVID-19 infection’, Médecine et Maladies Infectieuses. Doi: 10.1016/j.medmal.2020.03.006. Once again, this was not a clinical trial and the study was only conducted over a short period (5 days), the number of cases being too small to be significant anyway. However, the study concludes that treatment with hydroxychloroquine does not provide any benefit to severely affected patients.

What the Chinese are up to?

Chinese studies exploring chloroquine versus COVID-19 are also discussed. What about them?

First mentions of the chloroquine trail

There are indeed two publications indicating that the chloroquine trail is encouraging. One is a letter⁹Jianjun Gao, Zhenxue Tian, and Xu Yang, 2020. ‘Breakthrough: Chloroquine phosphate has shown apparent efficacy in treatment of COVID-19 associated pneumonia in clinical studies’, BioScience Trends 14 (1), pp. 72 – 73. Doi: 10.5582/bst.2020.01047 that provides no data and is therefore unusable, the other is an article¹⁰Manli Wang, Ruiyuan Cao, Leike Zhang, Xinglou Yang, Jia Liu, Mingyue Xu, Zhengli Shi, Zhihong Hu, Wu Zhong, and Gengfu Xiao, 2020. ‘Remdesivir and chloroquine effectively inhibit the recently emerged novel coronavirus (2019-nCoV) in vitro’, Cell Research 30, pp. 269 – 271. Doi: 10.1038/s41422-020-0282-0 reporting on in vitro tests. However, as we have seen, encouraging in vitro results do not necessarily imply that the treatment will be usable in vivo. This is far too weak to say that a treatment has been found.

Case studies

Still, there are larger Chinese studies, two published in The Lancet¹¹Chaolin Huang, Yeming Wang, Xingwang Li, Lili Ren, Jianping Zhao, Yi Hu, Li Zhang, Guohui Fan, Jiuyang Xu, Xiaoying Gu, Zhenshun Cheng, Ting Yu, Jiaan Xia, Yuan Wei, Wenjuan Wu, Xuelei Xie, Wen Yin, Hui Li, Min Liu, Yan Xiao, Hong Gao, Li Guo, Jungang Xie, Guangfa Wang, Rongmeng Jiang, Zhancheng Gao, Qi Jin, Jianwei Wang, and Bin Cao, 2020. ‘Clinical features of patients infected with 2019 novel coronavirus in Wuhan, China’, The Lancet 395 (10,223), pp. 497 – 506. Doi: 10.1016/S0140-6736(20)30183-5 ¹²Nanshan Chen, Min Zhou, Xuan Dong, Jieming Qu, Fengyun Gong, Yang Han, Yang Qiu, Jingi Wang, Ying Liu, Yuan Wei, Jia’an Xia, Ting Yu, Xinxin Zhang, and Li Zhang, 2020. ‘Epidemiological and clinical characteristics of 99 cases of 2019 novel coronavirus pneumonia in Wuhan, China: a descriptive study’, The Lancet 395 (10,223), pp. 507 – 513. Doi: 10.1016/S0140-6736(20)30211-7 and one in The New England Journal of Medicine¹³Wei-jei Guan, Zheng-yi Ni, Yu Hu, Wen-hua Liang, Chun-quan Ou, Jian-xing He, Lei Liu, Hong Shan, Chun-Liang Lei, David S. C. Hui, Bin Du, Lan-juan Li, Guang Zeng, Kwok-Yung Yuen, Ru-chong Chen, Chun-li Tang, Tao Wang, Ping-yan Chen, Jie Xiang, Shi-yue Li, Jin-lin Wang, Zi-jing Liang, Yi-xiang Peng, Li Wei, Yong Liu, Ya-hua Hu, Peng Peng, Jian-ming Wang, Ji-yang Liu, Zhong Chen, Gang Li, Zhi-jian Zheng, Shao-qin Qiu, Jie Luo, Chang-jiang Ye, Shao-yong Zhu, and Nan-shan Zhong, 2020. ‘Clinical Characteristics of Coronavirus Disease 2019 in China’, New England Journal of Medicine. Doi: 10.1056/NEJMoa2002032. These articles describe the medical cares given to many patients with COVID-19. These studies are called observational studies: unlike clinical trials, the aim is not to test a treatment, but to describe the symptoms and course of the disease. This allows for gathering valuable information to health care personnel. However, such a study is insufficient to inform a specific treatment and in any case in the studies in question the use of chloroquine is punctual.

Exploring the chloroquine trail

On the other hand, there is a Chinese study indicating that there is no significant difference between patients with COVID-19 treated with hydroxychloroquine compared to patients not receiving this treatment¹⁴Chen Jun, Liu Danping, Liu Li, Liu Ping, Xu Qingnian, Xia Lu, Ling Yun, Huang Dan, Song Shuli, Zhang Dandan, Qian Zhiping, Li Tao, Shen Yinzhong, and Lu Hongzhou, 2020. ‘A pilot study of hydroxychloroquine in treatment of patients with common coronavirus disease-19 (COVID-19)’, Journal of ZheJiang University (Medical Sciences) 49 (1). Available on-line..

Finally, results of a clinical trial of hydroxychloroquine were pre-published on March 30^th, 2020¹⁵Zhaowei Chen, Jijia Hu, Zongwei Zhang, Shan Jiang, Shoumeng Han, Dandan Yan, Ruhong Zuhang, Ben Hu, and Zhan Zhang, in preparation. ‘Efficacy of hydroxychloroquine in patients with COVID-19: results of a randomized clinical trial’. Medrxiv doi: 10.1101/2020.03.22.20040758. Not expertized.. The protocol consists of a 6-day monitoring—measuring the condition of patients before treatment, then after 5 days of treatment—on the impact of administrating 400 mg daily of hydroxychloroquine to patients with mild pneumonia (with cough and fever). The purpose was to estimate the effectiveness of hydroxychloroquine treatment on the improvement of mild symptoms, hopefully avoiding their worsening and thus the need of intensive care.

Although this study claims to be a double-blind, placebo-controlled study, this is not the case: both groups received the same care, except that one group received in addition chloroquine, while the control group did not receive placebo. This is therefore an open-label study. Furthermore, the sample size is quite small, with 31 subjects in the control group and 31 in the hydroxychloroquine group. Moreover, its duration is rather short, 6 days, whereas it is now shown that the evolution towards a serious form is frequent after 6 to 10 days and this despite a frequent transient clinical improvement before aggravation. Furthermore, the most serious cases—that is, those most in need of treatment since, as we have seen, mild cases tend to heal spontaneously—were excluded from the study: during the period of inclusion in the study, 142 patients were confirmed as suffering from COVID-19 and the reason the cohort was reduced to 62 subjects is not clear. On the other hand, side effects were reported: two mild cases of migraine and rash.

The alleged results are the following:

• a one-day reduction in the average duration of fever—however, in the context of COVID-19, this criterion is not necessarily clinically significant;
• a one-day reduction in the average duration of cough;
• slight improvement in the progression of pneumonia.

In addition to the methodological shortcomings already noted, the evolution of pneumonia was followed only radiographically and other possible criteria (such as arterial blood gas¹⁶Arterial blood gas test is a test widely used in pneumology and reanimation. It consists of detecting gases dissolved in the blood and makes it possible to evaluate a patient’s respiratory function.) were not taken into account. Moreover, the methodology does not allow to draw any conclusion on the impact of hydroxychloroquine concerning the contagiousness of patients after treatment. In addition, existing risk factors, such as smoking or pregnancy, are not indicated although they may impact the results. Therefore, it is not possible to fully ascertain the homogeneity of the two groups. Moreover, the study only included relatively young patients with mild symptoms, whereas it would have been interesting to compare the results with older and more severely affected patients, that is to say, once again, those most in need of management. All in all, it is highly likely that the study sample is not representative. In any case, the small number of patients makes it difficult to extrapolate the results to the whole population.

Although this study concludes that there is a slight improvement in symptoms, it is anecdotal and the study is methodologically and statistically weak. Moreover, the study itself notes that the balance between benefit (gain of one day on certain symptoms) and risk (variable side effects) remains unclear.

All in all, Chinese studies do not lead to be really optimistic regarding chloroquine as a treatment for COVID-19.

Be rational: be cautious

A problem with methodology

As you have seen, most of the studies cited in this article are inconclusive. There are two major reasons for this.

The first is that scientific publications are a way for different teams of researchers to communicate with each other. For these teams, it is very useful to know what trails are being explored and whether these trails seem likely to prove successful. When a team finds that a lead is not very successful, it is useful information for other teams working on the same subject. However, great care and caution should be exercised when communicating about such studies to non-specialists, to avoid the risk of confusion. Generally, in such a context, it is even preferable not to talk about these non-significant studies at all, because for someone not working on the subject, they simply do not provide any information. In times of crisis, when bad information can have very bad consequences, it is advisable to be even more careful.

In the communication about chloroquine, however, some teams have departed from caution. The generalist media have be attacked and blamed for focusing on studies that did not deserve such focus, but the fact is that the communication of some specialists was simply incomprehensible. Logically, general media did not understand anything, which led to a great deal of confusion.

The researcher has a responsibility when communication to the public. Some have freed themselves from this responsibility in order to highlight their work.

The second reason is that some teams tend to free themselves at least partly from the rigour necessary for a scientific study. As a result, poor quality studies are regularly proposed. Generally, they do not pass the barrier of peer review. Some, however, are published all the same. The mechanism of continuous reviews and re-evaluations allows those involved in the discipline to identify these problematic cases. However, in the case of COVID-19, such studies were discussed without waiting for this process to operate and these flawed studies were highlighted, adding to the confusion.

In recent times a view seems to grow in popularity, stating that there are circumstances that would justify a departure from the scientific method, or even that this method would be a hindrance. However, the current situation does not support this view at all.

One element on which the scientific method was built on the systematic minimization of error, as everyone makes mistakes. You have to be aware of them and arm yourself to be able to identify them. Furthermore, this article tries to show that even in times of crisis, accurate and prudential communication can be achieved. Also, at the time of writing, there are teams working on a treatment against COVID-19 using a strong methodology.

Chloroquine and consequences

At the time of writing, it cannot be said that chloroquine is an effective treatment for COVID-19, nor can it be said that it is not. We must wait a least for the results of the Discovery clinical trial, large-scale tests that have been launched throughout Europe, to have reliable information on the effectiveness of the treatments being considered. Even with such a large study, uncertainties are likely to persist. Let’s take this opportunity to recall that the IHU Méditérannée-infection team is far from being the only one working on this issue and that hydroxychloroquine is only one of the leads for treating the disease.

However, the effect of the communication on chloroquine is to add even more pressure on health care teams and provoking irrational acts of taking substances just because it says “chloroquine” on it. Also, some testimonials (which should not be overgeneralized) from teams participating in the Discovery study indicate that the media coverage of chloroquine led some volunteers to want to be part of the group treated with hydroxychloroquine to the exclusion of all other, in defiance with the methodology. An opportunity in our misfortune: the number of patients is unfortunately high and consequently there are enough volunteers to simply dismiss those who come with such demands. Moreover, this catastrophic communication has been the starting point for many conspiracy theories.

Please keep in mind that as of April 7^th, 2020 there is no evidence to determine whether chloroquine is useful in the treatment of COVID-19 and whether it has a positive benefit/risk balance. For this reason, among others, the recommendation to general practitioners is, according to the current state of knowledge, not to prescribe chloroquine in case of symptoms of COVID-19.

Even if chloroquine were to prove effective against COVID-19, it must be borne in mind that there are contraindications to its use, so that the prescription cannot in any way be automated. In the event of symptoms of COVID-19, do not try to force physicians in prescribing any variant of chloroquine.

In any case, as of April 7^th, 2020, there is no proven treatment for COVID-19 and the only method we have for limiting hospital overcrowding is to adhere to containment and barrier gestures.

Acknowledgements

I would like to thank my fellows from Café des sciences for reviewing this article:

• Thibault Fiolet, specialist in epidemiology and food risk assessment, Quoi dans mon assiette ?;
• Jean, specialist in cryosphere, I Need space;
• Cyril Vidal, dental surgeon, Dans les dents;
• Corentin Lacroix, physician, Why Doc;
• Erik Gustafsson, psychologist, Science pour parents;
• Nicolas Minier, specialized in bioengineering, BioRev³;
• Marc Robinson-Rechavi, specialist in bioinformatics, Tout se passe comme si.