Physicians ’ preference towards the non-evidence based hydroxychloroquine treatment for COVID-19 : the pandemic effect

| OBJECTIVE: To evaluate whether there is a "pandemic effect" promoting irrational medical reasoning. METHODS: Crosssectional study. State medical councils were asked to send all registered physicians two sequential email invitations to complete a Google Form questionnaire. Between April 15 and May 3, 2020, 370 doctors answered our questionnaire with questions about the prescription of hydroxychloroquine for COVID-19 and vitamin C for sepsis. The questionnaire had a five-point Likert scale (higher number, higher support) on the degree of propensity to prescribe the medication and a binary question (yes or no) to express the doctor's final decision to prescribe. These two questions, in the same order, were asked for scenarios of mild, moderate and severe cases. RESULTS: The propensity to prescribe hydroxychloroquine for COVID-19 increased with the severity of the clinical presentation: for mild cases, 37% (95% CI 32%-42%) of the physicians chose "yes", increasing to 68% (95% CI 63%-72%) and 89% (95% CI 85%-92%) for moderate or severe cases, respectively (Cochran's Q test: P<0.001). The medians and interquartile ranges of the Likert scales for hydroxychloroquine were 2 (1-4), 4 (2-4), 4 (4-5) in mild, moderate and severe cases of COVID-19 (Friedman test: P<0.001). CONCLUSIONS: The propensity of the Brazilian physician to prescribe hydroxychloroquine for COVID-19 is high and, according to the severity of the disease, ranged from 37 to 89%. On the contrary, the propensity to prescribe vitamin C for sepsis, a non-pandemic situation, was lower and not associated with clinical severity. Our data suggest a "pandemic effect" promoting irrationality in medical reasoning.

to express the doctor's final decision to prescribe. These two questions, in the same order, were asked for scenarios of mild, moderate and severe cases. RESULTS: The propensity to prescribe hydroxychloroquine for COVID-19 increased with the severity of the clinical presentation: for mild cases, 37% (95% CI 32%-42%) of the physicians chose "yes", increasing to 68% (95% CI 63%-72%) and 89% (95% CI 85%-92%) for moderate or severe cases, respectively

Introduction
Hydroxychloroquine has been widely prescribed as a treatment for COVID-19 patients since mid-March 1 . Prescription enthusiasm was triggered by a high risk of bias French study 2 , with an impressive Altimetric of 7.748 as of May 1st, which was more ideologically than scientifically endorsed.
Hydroxychloroquine has potential unintended consequences 3 and no proof of efficacy according to evidence-based principles 4 . It is not known whether physicians feel compelled to prescribe hydroxychloroquine because of the collective commotion or if prescription is mediated by their own preferences towards the treatment. We hypothesized that there is a "pandemic effect" promoting irrational medical reasoning.

Methods
Six state medical councils were asked to send all registered physicians two sequential e-mail invitations to fill a Google Form questionnaire. The Brazilian National Ethics Research Committee approved the protocol (CAAE 31015520.6.0000.0048).
The questionnaire had a header stating: "Hydroxychloroquine has been proposed as a treatment for COVID-19 but there is no scientific proof of efficacy so far". Then, it was followed by a five-point Likert Scale (higher number, higher support) on the degree of propensity to prescribe the drug and a binary question (yes or no) to express physician final decision to prescribe. These two questions, in the same order, were performed for scenarios of mild, moderate and severe cases.
As a control question, we used the same questions for vitamin C in sepsis, a non-evidence based treatment, but not related to SARS-Cov-2 pandemic.
The primary endpoint was predefined as the binary response. We planned a sample size of 385 individuals to provide a precision of ± 5% for the binary frequency of positive response regarding prescription, assuming a prevalence of 50% (95%CI).
Four of the six invited medical councils did not feel comfortable approaching the issue in the midst of the pandemic. The two councils of Alagoas and Goias states agreed to send the questionnaire to their physicians, comprising a total of 21.962 invitations. Between April 15 and May 3, 370 physicians responded, age 42 ± 11 yrs, 61% males, 37% surgeons.

Table 1. Prescription Propensity in a Likert Scale and Binary Response
The only variable associated with hydroxychloroquine propensity was years of professional experience (prescriptors versus non-prescriptors: 19±11 and 14±11; student's t test: P<0.001). Gender, academic activity and post-doctoral titles were not related to the outcome.
Brazilian physician's propensity to prescribe hydroxychloroquine for COVID-19 is high and, according to disease severity, varied from 37 to 89%. On the contrary, propensity to prescribe vitamin C for sepsis, a non-pandemic situation, was lower and not associated with clinical severity. Our data suggests a "pandemic effect" promoting irrationality on medical reasoning.
The key limitation of the present study includes a sample that was not representative of the entire population of physicians in the country, which might impact generalizability of our findings.
In times where rational clinical decision-making is of utmost importance, our preliminary data raises concern regarding the role of collective emotional stress guiding the prescription of non-evidence based therapies.

Author contributions
Correia LC, Barreto Filho JA were responsible for the literature review, conception and planning, study design and methods, data analysis and conclusion and writing of the manuscript. Lopes JR, Garcez FB, Le Campion E, Barcellos G were responsible for the literature review, study design and methods, data analysis and discussion of the results

Competing interests
No financial, legal or political competing interests with third parties (government, commercial, private foundation, etc.) were disclosed for any aspect of the submitted work (including but not limited to grants, data monitoring board, study design, manuscript preparation, statistical analysis, etc.).