key: cord-303819-w1785lap
authors: Cortegiani, Andrea; Ippolito, Mariachiara; Ingoglia, Giulia; Iozzo, Pasquale; Giarratano, Antonino; Einav, Sharon
title: Update I. A systematic review on the efficacy and safety of chloroquine/hydroxychloroquine for COVID-19
date: 2020-07-11
journal: J Crit Care
DOI: 10.1016/j.jcrc.2020.06.019
sha: 
doc_id: 303819
cord_uid: w1785lap

PURPOSE: To assess efficacy and safety of chloroquine (CQ)/hydroxychloroquine (HCQ) for treatment or prophylaxis of COVID-19 in adult humans. MATERIALS AND METHODS: MEDLINE, PubMed, EMBASE and two pre-print repositories (bioRxiv, medRxiv) were searched from inception to 8th June 2020 for RCTs and nonrandomized studies (retrospective and prospective, including single-arm, studies) addressing the use of CQ/HCQ in any dose or combination for COVID-19. RESULTS: Thirty-two studies were included (6 RCTs, 26 nonrandomized, 29,192 participants). Two RCTs had high risk, two ‘some concerns’ and two low risk of bias (Rob2). Among nonrandomized studies with comparators, nine had high risk and five moderate risk of bias (ROBINS-I). Data synthesis was not possible. Low and moderate risk of bias studies suggest that treatment of hospitalized COVID-19 with CQ/HCQ may not reduce risk of death, compared to standard care. High dose regimens or combination with macrolides may be associated with harm. Postexposure prophylaxis may not reduce the rate of infection but the quality of the evidence is low. CONCLUSIONS: Patients with COVID-19 should be treated with CQ/HCQ only if monitored and within the context of high quality RCTs. High quality data about efficacy/safety are urgently needed.

The spread of the coronavirus disease 2019 caused by the novel SARS-CoV-2 has reached pandemic dimensions. Many drugs, both repurposed and new, have and are being investigated for preventing or treating the disease [1, 2] .

Chloroquine (CQ) and its related formulations (e.g. hydroxychloroquine -HCQ) were introduced at a very early stage of the pandemic as a potential treatment for COVID- 19 . At the time, only pre-clinical rationale, in vitro findings and meager animal model data were available [3] . The desperate need for an effective treatment has led to widespread use of the drug nonetheless. Dependent on location, CQ/HCQ are being used in the context of clinical trials or as standard care.

As the pandemic evolves, the amount of evidence accumulated regarding various treatment options is growing rapidly. However, the efficacy and safety of CQ/HCQ remains unclear [4] . We therefore aimed to systematically search, assess and summarize the existing literature related to the efficacy and safety of these drugs in the clinical context of treatment and prophylaxis of COVID-19. We also set out to pool and meta-analyze the most updated data, if possible, in order to ascertain whether any conclusions can be reached at this time regarding the association between CQ/HCQ and hospital mortality in patients with COVID-19 or disease prevention in those exposed.

We prospectively registered the protocol of this review on Open Science Framework (OSF) on 12 th May 2020 (osf.io/3yka8).

Journal Pre-proof PICO question: The current review covers studies comparing adult patients with or at risk of COVID-19 (P) who had been administered CQ or related formulations, alone or in combination with other drugs (I) to those given standard care or other regimens or drugs (C). The outcomes of interest were both efficacy (i.e. mortality, viral clearance, infection rate) and safety (i.e. adverse events, focusing on cardiac events)

Search strategy: The search strategy is presented in full in Supplementary Material 1. In brief, we performed a comprehensive search of MEDLINE, PubMed and EMBASE from inception to 8 th June 2020 for both randomized and nonrandomized studies, both retrospective and prospective [5] addressing the PICO question. We did not apply any language or quality restrictions. Following full-text download (see below) the reference lists of relevant articles were also screened (i.e. snowballing method). Two major pre-print servers (bioRxiv and medRxiv) were also searched for relevant not peer-reviewed articles from inception to 8 th June 2020. tool (Revised tool for Risk of Bias in randomized trials) was used for assessing randomized trials [6] . The ROBINS-I tool (Risk Of Bias in Non-randomized Studies of Interventions) was used for RoB of nonrandomized studies with comparison between relevant study groups [7] . The Newcastle Ottawa Scale (NOS) was used for assessing single-arm nonrandomized studies, without evaluating the -comparability‖ item [8] . For each domain we rated the overall RoB as the highest risk attributed to any criterion. We used the Robvis tool (visualization tool for risk of bias assessments in a systematic review) [9] for presenting the data as appropriate. The final RoB assessments are reported as either a plot or a table as per requirement (see below).

Data collection and management: The primary study outcomes were all-cause mortality at the longest reported follow-up for studies evaluating CQ/HCQ as treatment, and infection rate in prophylactic studies. Two authors (AC, MI) extracted information regarding study design, sample size, patients' characteristics, interventions and outcomes using a pre-piloted data extraction form in duplicate. If important data were missing or remained unpublished in any of the relevant studies, the authors were contacted by one of the reviewers with the aim of retrieving this data. The extracted data were used for both the description of the included studies and the qualitative analysis. 

Borba et al. conducted a block-randomization phase IIb trial, enrolling hospitalized patients with severe acute respiratory syndrome (e.g. fever, tachypnea, hypotension, altered mental status, oliguria) and suspected diagnosis of COVID-19 in a single centre in Brazil [18] . Patients were enrolled before laboratory confirmation of COVID-19. The study aimed to compare the efficacy and safety of two doses of CQ base (600mgX2/day for 10 days vs. 450mgX2/day on day 0 and 450mgX1day on J o u r n a l P r e -p r o o f days 1-4). Placebo pills were used to mask the treatment from the participants and researchers. All patients also received azithromycin and some also received oseltamivir. The trial was terminated prematurely after enrolling 81 patients of the 440 intended when unplanned interim analysis was requested by the independent data and monitoring board due to concerns regarding safety. Higher drug doses were found to be accompanied by higher rates of 13-day mortality (39% vs. 15%), QTc interval prolongation >500 milliseconds (18.9% vs. 11.1%) and ventricular tachycardia (2 patients versus none). The proportion of patients with detectable viral RNA levels was similar in the two groups.

France. The study included 1061 patients with a positive PCR test for SARS-CoV-2 who were admitted to either day-care or wards [19] . All patients received oral HCQ 200 mg X3/day for ten days and Azithromycin 500 mg on the first day, followed by 250mg daily for four days. All patients also underwent pre-treatment workup including electrocardiography and serum electrolyte testing to rule out the presence of contraindications to treatment. Good clinical outcomes, defined as survival, no ICU or hospital admission and negative nasal viral shedding were achieved in 91.7% of the patients by day 10. Adverse events occurred in 2.4% of the patients but none were cardiac. The same group of investigators published another article with similar methods and smaller sample size, without data overlap [20] . Molina et al. described a prospective case series of 11 patients. All the patients were treated with the same regimen proposed by the authors of the series of papers described above, but 80% of the patients had a positive PCR assay for nasopharyngeal swab specimens at 5-6 days [21] . PCR for SARS-CoV-2, treated with HCQ 400 mg (+ azithromycin 26%; + lopinavirritonavir 35%; both drugs 6%). An ECG was recorded after 5 days of treatment (QT corrected with Bazett formula). Proportion of patients with mild, moderate and severe (>500 ms) QT prolongation were 9%, 4% and 2% respectively. In 53% of cases who had an ECG off-therapy, the median increase of QTc was 18 ms [22] . treated with HCQ (11 with azithromycin). QT prolongation was detected in 38.5% of the patients, normalizing after treatment completion or discontinuation. One nonagenarian died of sepsis. [29] .

Boulware et al. conducted a randomized, double blind, placebo controlled trial in US and Canada, enrolling 821 asymptomatic non-hospitalized adults with selfreported high-risk or moderate risk of exposure to a positive COVID-19 case within 4 days of exposure. High risk was defined as household or occupational exposure at a distance of less than 6 ft for more than 10 minutes while wearing neither a face mask nor an eye shield. Moderate risk was defined as exposure while wearing a face mask but no eye shield [30] . The study had a power of 90%; the sample size was calculated based on a prior study that had actively monitored exposed cases [31] where 10% of close contacts developed COVID-19 plus an attrition rate of 20%.

Participants (recruited by social media) were assigned to HCQ (800 mg once, 600 mg 6-8 hours later, then 600 mg daily for 4 days) or placebo and followed up through emails. Data were provided by participants via a portal to an online database. The J o u r n a l P r e -p r o o f median age of the cohort was 40years, 66.4% were healthcare workers (HCWs) and the rate of adherence to the trial intervention was 79%. The incidence of new illness compatible with COVID-19 did not differ between those taking HCQ and those taking the placebo (11.8% vs. 14.3%; p=0.35). Participant-reported side effects were more common in those receiving HCQ (40.1%) than placebo (16.8%). Gastrointestinal side effects were the most common and no serious adverse reactions were reported. The proportion of patients hospitalized due to COVID-19 was similar among those using CQ/HCQ and among those who were not (55% [16/29] ) vs. 57% [29/51], p=ns) [34] .

Nine studies were retrieved from the search of pre-print repositories. These included one RCT [35] and eight non-randomized studies [36] [37] [38] [39] [40] [41] [42] [43] . The studies evaluated a total of 7702 hospitalized patients with COVID-19. We hereby describe only the RCT and the two retrospective studies that included the largest number of patients. A detailed description of all the studies is available in Table 1 . 

Among the included studies, the six RCTs [11, 13, 16, 18, 30, 35] were evaluated using the Rob2 tool [6] . Only two studies had low risk of bias [18, 30] .

Details regarding downgrading are provided in Figure 1 , and the weighted risk of bias is presented as a plot in Figure S1 (Supplementary Material 4) Fourteen [12, 14, 38, 39, 41, 42, 15, 17, 23, 28, 33, 34, 36, 37] nonrandomized studies were evaluated using the ROBINS-I tool [7] . The most frequent domain causing downgrading was confounding. Details regarding downgrading are provided in (Table S1 ).

We planned to perform quantitative synthesis if two or more studies were identified with a low risk of bias that also had sufficient homogeneity in study design, interventions and outcomes [10] We identified only two studies with a low risk of bias, and these differed in participants (ill inpatients vs. exposed outpatients), intervention J o u r n a l P r e -p r o o f (treatment vs. prophylaxis) and outcomes (mortality rate vs. incidence of compatible illness with COVID-19) [18, 30] . Therefore, particularly in light of the outcomes at stake, we decided to not perform quantitative synthesis of the data. Two of these trials included only 22 [13] and 30 [11] patients. A third included a large number of comparators but only 48 patients who actually received treatment [17] and in a fourth study only 75 patients were in the treatment arm [16] . The largest number of patients who actually received treatment in the published studies with comparators J o u r n a l P r e -p r o o f comes from one retrospective study (Rosenberg et al. n=1475 ) [15] and one prospective study (Geleris et al. n=1376 ) [12] , both at moderate risk of bias. Both provided data from the same geographical area (New York). The first (multicentre) did not collect data on other antiviral drugs co-administered in included patients [15] and the second (single-centre) had a composite primary outcome measure [12] .

Composite outcomes are often used to increase power, raising questions regarding outcome selection and interpretation [44] . The existing data regarding efficacy as treatment is also limited by issues that are not included in standard data quality tools.

Two included RCTs were terminated early, one for concerns regarding side effects in a high CQ dose arm [18] and another due to a decline in cases [16] . These two studies were therefore underpowered for their primary outcomes. Underpowering has also been shown to be a major issue in studies on antiviral medications in patients with COVID-19 [45] . From the specific aspect of critical care, at least two of the studies included patients who were mostly not critically ill. One a-priori excluded patients with organ failure [16] and the second excluded patients who died within 24 hours of presentation to hospital [12] . Finally, all of the published studies reported allcause mortality at various timepoints, rather than attributable mortality and none reported on the rate of withholding and withdrawal of care. This issue is particularly poignant since one of the studies reported that among the patients included almost half died without intubation [12] . The possibility of crisis standards of care is very valid in the context of a pandemic, and patients that do worse may have been more likely to receive both compassionate medication and expectant care.

Prolongation of QTc is a consistent finding with CQ/HCQ, suggesting that patients receiving treatment with CQ/HCQ require in the least periodic electrocardiographic assessment or, better yet, continuous monitoring. Higher doses J o u r n a l P r e -p r o o f of these drugs may be associated with higher risk of harm and side effects however this remains uncertain given that the study suggesting so was terminated very early [18] . Importantly, the association between CQ/HCQ and QT prolongation and arrhythmia was not adjusted for the presence of conventional risk factors for QT prolongation such as older age, electrolyte disorders, cardiac disease, genetic predisposition and other QT prolonging drugs in any of the studies.

Drug induced prolongation of the QT interval is a known risk factor for Torsades de Pointes but the exact electrographic markers of a tendency towards this arrhythmia remain unknown. Treatment with Azithromycin has been associated with induction of short-coupled polymorphic VT irrespective of QT prolongation [46] .

Conversely -the effect (of CQ/HCQ) on the QTc (is) driven entirely by prolonging the repolarization and regardless of QRS, as evident by the corresponding JTc prolongation‖ [24] .

Reports of QTc prolongation and arrythmias during combination treatment of CQ/HCQ with azithromycin have been anecdotal thus far and mostly based on case reports [47] . The initial recommendation to combine the two drugs was based on the outcome of six patients [48] and has not been supported by any real evidence of benefit. One included nonrandomized large study suggest an independent association between the combination HCQ with azithromycin and higher risk of cardiac arrest compared to no drug (OR, 2.13, 95% CI 1.12-4.05) [15] and so it is probably time to rethink the efficacy and safety of this approach. More detailed study of the coupling interval of the arrhythmia-initiating beat may perhaps contribute to differentiate between ventricular arrythmias caused by the two drugs. Evidence from available data suggests that monitoring baseline and subsequent (e.g. daily) ECG during treatment especially in high risk patients with known risk.

We identified only two studies with low risk of bias (one for treatment and one for post-exposure prophylaxis) and our included studies varied widely in term of patients' characteristics, outcomes definitions, interventions and design ( Table 1) .

Our early decision to proceed with quantitative synthesis only for the primary outcomes and only if the data would clearly lend itself to such analysis may seem conservative. However, given the stakes at hand such an approach may save lives as it should lead to better monitoring and research and, hopefully in the interim, Table 2) .

Finally, we identified only one trial on post-exposure prophylaxis with HCQ.

Although this trial seems appropriately powered and has a low risk of bias, delayed initiation of treatment (usually ≥3 days) and the use of self-reported outcomes make it 

Pharmacologic Treatments for Coronavirus Disease 2019 (COVID-19): A Review

Hydroxychloroquine in the management of patients with COVID-19: the need for an evidence base

A systematic review on the efficacy and safety of chloroquine for the treatment of COVID-19

Chloroquine for COVID-19: rationale, facts, hopes

Including non-randomized studies on intervention effects

RoB 2: A revised tool for assessing risk of bias in randomised trials

ROBINS-I: A tool for assessing risk of bias in non-randomised studies of interventions

Critical evaluation of the Newcastle-Ottawa scale for the assessment of the quality of nonrandomized studies in meta-analyses

Risk-of-bias VISualization (robvis): An R package and Shiny web app for visualizing risk-of-bias assessments

Synthesizing and presenting findings using other methods | Cochrane Training

A pilot study of hydroxychloroquine in coronavirus disease-19 (COVID-19)

Observational Study of Hydroxychloroquine in Hospitalized Patients with Covid-19

Clinical efficacy of hydroxychloroquine in patients with covid-19 pneumonia who require oxygen: observational comparative study using routine care data

Association of Treatment With Hydroxychloroquine or Azithromycin With In-Hospital Mortality in Patients With COVID-19 in

Hydroxychloroquine in patients with mainly mild to moderate coronavirus disease 2019: open label, randomised controlled trial

Low dose of hydroxychloroquine reduces fatality of critically ill patients with COVID-19

Effect of High vs Low Doses of Chloroquine Diphosphate as Adjunctive Therapy for Patients Hospitalized With Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) Infection

Early J o u r n a l P r e -p r o o f treatment of COVID-19 patients with hydroxychloroquine and azithromycin: A retrospective analysis of 1061 cases in Marseille, France

COVID-19 patients with at least a six-day follow up: A pilot observational study

No evidence of rapid antiviral clearance or clinical benefit with the combination of hydroxychloroquine and azithromycin in patients with severe COVID-19 infection

Association of Hydroxychloroquine with QTc Interval in Patients with COVID-19

Risk of QT Interval Prolongation Associated With Use of Hydroxychloroquine With or Without Concomitant Azithromycin Among Hospitalized Patients Testing Positive for Coronavirus Disease 2019 (COVID-19)

QT Interval Prolongation and Torsade De Pointes in Patients with COVID-19 treated with Hydroxychloroquine/Azithromycin. Hear Rhythm

Incidence and Determinants of QT Interval Prolongation in COVID-19 Patients Treated with Hydroxychloroquine and Azithromycin

Assessment of QT Intervals in a Case Series of Patients With Coronavirus Disease 2019 (COVID-19) Infection Treated With Hydroxychloroquine Alone or in Combination With Azithromycin in an Intensive Care Unit

Chloroquine-induced QTc prolongation in COVID-19 patients

The Effect of Chloroquine, Hydroxychloroquine and Azithromycin on the Corrected QT Interval in Patients with SARS-CoV-2 Infection

Experience of short-term hydroxychloroquine and azithromycin in COVID-19 patients and effect on QTc trend

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

Active Monitoring of Persons Exposed to Patients with Confirmed COVID-19 -United States

HyPE study: hydroxychloroquine prophylaxis-related adverse events' analysis among healthcare workers during COVID-19 pandemic: a rising public health concern

Continuous with SARS-CoV-2: Insights from a large healthcare database analysis

Baseline use of hydroxychloroquine in systemic lupus erythematosus does not preclude SARS-CoV-2 infection and severe COVID-19

Efficacy of hydroxychloroquine in patients with COVID-19: results of a randomized clinical trial

Hydroxychloroquine and azithromycin plus zinc vs hydroxychloroquine and azithromycin alone: outcomes in hospitalized COVID-19 patients

Treatment Response to Hydroxychloroquine, Lopinavir-Ritonavir, and Antibiotics for Moderate COVID-19: A First Report on the Pharmacological Outcomes from South Korea

Outcomes of hydroxychloroquine usage in United States veterans hospitalized with Covid-19

Hydroxychloroquine is associated with slower viral clearance in clinical COVID-19 patients with mild to moderate disease: A retrospective study

Experience with Hydroxychloroquine and Azithromycin in the COVID-19 Pandemic: Implications for QT

Outcomes of Hydroxychloroquine Treatment Among Hospitalized COVID-19 Patients in the United States-Real-World Evidence From a healthcare network for design assistance to complete this project

Hydroxychloroquine and Tocilizumab Therapy in COVID-19 Patients -An Observational Study

Doxycycline and Hydroxychloroquine as Treatment for High-Risk COVID-19 Patients: Experience from Case Series of 54 Patients in Long-Term Care Facilities

Composite Outcomes in Randomized Trials: Greater Precision but with Greater Uncertainty?

Remdesivir for COVID-19: challenges of underpowered studies

Azithromycin Causes a Novel Proarrhythmic Syndrome

Azithromycin, cardiovascular risks, QTc interval prolongation, torsade de pointes, and regulatory issues: A narrative review based on the study of case reports

Hydroxychloroquine and azithromycin as a treatment of COVID-19: results of an open

J o u r n a l P r e -p r o o f