Efficacy and safety outcomes of proposed randomized controlled trials investigating hydroxychloroquine and chloroquine during the early stages of the COVID‐19 pandemic

Aims To assess whether randomized clinical trials (RCTs) proposed to evaluate the treatment of patients with COVID‐19 with hydroxychloroquine (HQ) or chloroquine early in the pandemic included plans to measure outcomes that would translate into meaningful efficacy/effectiveness and safety outcomes. Methods The World Health Organization International Clinical Trials Registry Platform database was searched for registers of RCTs evaluating HQ or chloroquine, alone or in combination, compared with other treatments for patients diagnosed with COVID‐19. The final search was performed on 8 April 2020. Results Among 51 registered RCTs (median sample size 262; interquartile range: 100, 520), 34 (67%) reported a clinical outcome, 12 (24%) a surrogate outcome, and 5 (10%) a combination of clinical and surrogate outcomes as primary endpoints. Six (15%) trials included the World Health Organization scale for clinical improvement as a primary clinical outcome. Clinical improvement and mortality accounted for 45% of the unique domains among 18 clinical outcome domains of efficacy. Twenty‐four (47%) RCTs did not describe plans to assess safety outcomes; when assessed, safety outcomes were determined in generic terms of total, severe or serious adverse events. Conclusion The RCTs investigating HQ or chloroquine during the early stages of the COVID‐19 pandemic included heterogeneous and insufficient approaches to measure efficacy/effectiveness and safety relevant to patients and clinical practice. These findings provide insights to inform clinical and regulatory decisions that can be drawn about the efficacy/effectiveness and safety of these agents in patients with COVID‐19. Trialists need to adapt quickly to the research progress on COVID‐19, ensuring that core outcome measures are assessed in ongoing RCTs.


| INTRODUCTION
On 31 December 2019, a cluster of cases of atypical pneumonia of unknown aetiology was reported in the city of Wuhan, China, and later identified as being caused by a novel coronavirus. 1 The ability of the novel severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2; hereafter referred to as COVID- 19) 2 to infect human hosts and be transmitted among individuals rapidly evolved into a global pandemic with over 52.2 million confirmed cases in 235 countries as of 13 November 2020 (https://www.who.int/emergencies/ diseases/novel-coronavirus-2019). Although the majority (80%) of the COVID-19 cases develop a mild condition, a smaller percentage (15%) of patients require hospitalization and some develop a severe condition (5%) that requires mechanical ventilation in the first 24 hours of hospital admission. 3 Clinical complications such as profound acute hypoxaemic respiratory failure and sepsis 4,5 have led to a total of 1 286 063 deaths worldwide (https://www.who.int/emergencies/ diseases/novel-coronavirus-2019; last updated on 13 November 2020).
Worldwide, the capacity of the healthcare systems to offer care for patients diagnosed with COVID-19 depends on the capability of intensive care units (ICUs) and emergency departments (EDs) to accommodate the additional requirements brought about by the increased patient volumes during the pandemic. This is of critical relevance considering the high ICU occupancy commonly seen in many locations 6 and the long-recognized ED overcrowding and its negative consequences on patient outcomes. [7][8][9] Moreover, early in the pandemic, the management of patients with COVID-19 was supportive, and recovery time was estimated at around 3-6 weeks for critically ill patients. 10 In this challenging scenario, it is not surprising that there has been a frantic search for effective treatments. Early in April 2020, there were 788 entries of registered COVID-19 trials on the World Health Organization International Clinical Trials Registry Platform (WHO-ICTRP). By the end of August 2020, the WHO-ICTRP database had registered 3 times more entries.
Clinical trials provide vital evidence to establish the efficacy and safety of new interventions or new indications for existing interventions. To be informative, however, they have to be designed and implemented following standards to ensure valid and meaningful evidence. 11 Meaningful evidence involves defining outcomes comprising the potential benefits (efficacy/effectiveness) and harms (safety) of the treatments under investigation. 12 Moreover, efficacy/effectiveness outcomes should represent clinically meaningful results that directly measure how a patient feels, functions or survives. 13 Alternatively, trials may report surrogate outcomes, which may provide evidence for benefit that encourages further research. Trials assessing surrogate outcomes may be smaller, completed faster and be less expensive; however, surrogate outcomes may or may not predict clinical results and translate into meaningful evidence of efficacy/effectiveness. 13,14 Likewise, safety outcomes are essential in defining the value of a treatment intervention for healthcare providers, patients, and health systems. Despite the importance of finding a treatment that effectively mitigates or cures patients diagnosed with COVID-19, it is critical to appropriately define and detect the potential adverse events of the treatment options under investigation. There are guidance and legal requirements for clinical trial protocols to plan the data collection of adverse events, whether applying systematic or nonsystematic assessment approaches. 12,15,16 The objective of this study was to assess whether the randomized clinical trials (RCTs) of COVID-19 treatments registered on the WHO-ICTRP included definitions and data collection plans to produce evidence on meaningful efficacy, effectiveness and safety outcomes. We focused on RCTs as the highest level of evidence and those evaluating treatments with hydroxychloroquine (HQ) or chloroquine. Traditionally, these immune-suppressants have been used to treat autoimmune diseases such as rheumatoid arthritis and inflammatory bowel disease; however, they have also been approved for the treatment of malaria since 1955. 17 In the context of COVID-19, these drugs received widespread support as effective treatments following demonstration of in vitro viral activity against SARS-CoV-2 and a potential viral load reduction in a case series report. 18,19 During the early stages of the pandemic, HQ and chloroquine received emergency use authorization 20 by the Food and Drug Administration in the USA and overwhelming social media and leadership support, resulting in medication shortages. Evolving research has shown that HQ and chloroquine do not reduce the mortality of patients with COVID-19. 21 Nonetheless, there is continued interest in high-quality RCTs on the efficacy and safety of these and other drugs in the treatment of patients diagnosed with COVID-19 infection. 21 Furthermore, familiarity with HQ and chloroquine have created a push for expedited clinical trials.

| METHODS
We downloaded the COVID-19 WHO-ICTRP database (https://www. who.int/ictrp/search/en/) on 8 April 2020, at 10:30 GMT−6. We filtered studies on the database according to the intervention (HQ or chloroquine) and the study design (randomized vs nonrandomized). All the retrieved registers were screened and reviewed for data What is already known about this subject • There is an urgent need for effective and safe treatments that reduce the morbidity and mortality of patients diagnosed with COVID-19.
• Chloroquine, and its derivate hydroxychloroquine (HQ), have been identified as potential therapies described in the scientific literature and social media.

What this study adds
• The results show that the initial randomized controlled trials (RCTs) proposed to investigate the efficacy/effectiveness of HQ or chloroquine include a heterogeneous set of clinical outcomes domains.
• These early planned RCTs failed to include sufficient and structured approaches to detect adverse events that are relevant to patients and to inform clinical practice.

| Data extraction and sources
One author (D.J.) extracted data using a standardized form. Quality control was performed by re-extracting data from 15% of the included trial registries. Information on the trial ID, scientific title, date of registration, recruitment status, patient population and funding sources were extracted from the WHO-ICTRP database. Information on the country where the trials were planned to be conducted was extracted primarily from the WHO-ICTRP database and completed using the data from the trial registry when appropriate.
The trial's registry was accessed and provided additional information to characterize the RCTs according to: • Number of participants planned to be recruited; • Age and sex of the participants planned to be recruited; • Intervention and comparison treatments, including doses and administration schedules; • Treatment duration; • Efficacy/effectiveness outcomes defined as primary endpoints; • Safety outcomes, i.e. adverse events; • Timeframe for the assessment of the efficacy/effectiveness and safety outcomes; • Mode of data collection of safety outcomes.
The efficacy/effectiveness outcomes were classified as clinical (e.g. improvement or recovery from respiratory symptoms) or surrogate outcomes (e.g. viral load, biomarkers). The mode of data collection of the adverse events was classified as systematic assessment when specific ascertaining methods to detect the occurrence of adverse events were described by the use of checklists, questionnaires, or laboratory tests at regular intervals, and as nonsystematic assessment when the detection methods relied on the spontaneous report of adverse events by clinicians or participants. 15

| Data analysis
The characteristics of the RCTs were summarized according to clinical or surrogate outcomes and ascertainment methods to detect adverse events. Simple proportions were reported for dichotomous outcomes.
For continuous data, means and standard deviations or medians and interquartile ranges (IQRs) were reported, as appropriate. Comparisons between continuous variables were made using t-test and reporting the mean difference (MD) with 95% confidence intervals (CI). Finally, outcomes planned to be measured in the RCTs were compared to the primary outcome defined in the master protocol for COVID-19 studies published by WHO on 18 February 2020, 22 which comprised a composite measure of clinical improvement and/or survival measured on an ordinal scale ranging from 0, uninfected, to 8, dead.

| RESULTS
Among 927 clinical trials registered on the WHO-ICTRP database as of 8 April 2020, 72 registrations were identified as RCTs investigating the use of HQ or chloroquine for COVID-19 infection and considered potentially eligible for this study ( Figure 1). Among these 72 registered trials, 2 were duplicate entries of the same trial in more than 1 clinical trial registry, 7 trials had been cancelled and 12 entries related to trials testing prophylactic interventions. Therefore, 51 registered clinical trials were included for analysis. An updated search (conducted on 24 August 2020) revealed approximately 300 registered trials involving treatments with HQ or chloroquine that would be potentially eligible. Table 1 summarizes the characteristics of the clinical trials planned to investigate the use of HQ or chloroquine to treat patients diagnosed with COVID-19. The RCTs proposed to test the hypothesis of whether these drugs could be beneficial for people infected with SARS-CoV2 started in February 2020 when 12 trials were registered.
In the following month of March, the number of trials registered tripled. All trials planned to include adults of both sexes, and 3 trials Considering the dosing schedule of all treatment arms of either HQ or chloroquine, maximum treatment duration ranged from 7 to 14 days.
Twenty trials reported at least 1 arm with a variable dosing administration schedule of HQ or chloroquine. Considering all treatment arms with a variable dosing schedule, treatment duration varied from 5 to 16 days. Fourteen registered trials did not report information on the treatment duration. One trial (2%) reported plans to monitor adherence and 2 trials (4%) reported funding support from sources with potential commercial interest (data not shown).
Forty-five of the RCTs (90%) were planned to be conducted in a single country (Table S1) with a median sample size of 262 (IQR: 100, 520). Among the trials planned to be implemented in a single country, China was the main location (16; 32%) followed by the USA (5; 10%). Five (10%) of the registered RCTs were designed to be conducted in multiple countries; 1 trial did not provide information on the location where the RCT was planned to be implemented. Overall, the proposed clinical trials anticipate recruiting a total of 37 303 participants, among outpatients and inpatients, to be randomized to receive a variety of experimental and comparison treatments with HQ, chloroquine or other agents in diverse combinations and dose schedules (Table S2). Only 14 (27%) of the registered trials reported the number of patients being recruited to the treatment and comparison arms; among these trials, a total of 1138 patients would receive HQ or chloroquine alone or in combination with other drugs (data not shown). Table 3 summarizes the type of outcomes described in the registry of the RCTs and the related assessment timeframe. One-third of the clinical trials included in their registry information a surrogate outcome to be measured as a primary endpoint; the remaining trials (34; 67%) described plans to assess 1 clinical outcome as a primary endpoint. The timeframe of outcome assessment varied substantially among the designs of the RCTs. Trials planning to measure only clinical efficacy/ effectiveness outcomes described timeframes of assessment ranging from 5 to 120 days (median 15; IQR: 15,28). Trials planning to measure only surrogate outcomes defined timeframes of assessment ranging from 3 to 56 days (median 15; IQR: 15, 28). The RCTs planning to evaluate a clinical outcome were compared with trials planning to assess a surrogate outcome; however, no statistical difference in timeframes for outcome assessment was identified (MD 6.3; 95% CI: −10.51 to 23.12; P = .45). Among all 51 registered RCTs describing at least 1 clinical or surrogate efficacy/effectiveness outcome, 13 (26%) did not report a timeframe for outcome assessment.
The WHO scale for clinical improvement was described in the outcome assessment plans of 6 (15%) RCTs, and 16 (41%) trials mentioned clinical improvement among the primary outcomes without using the WHO scale or without detailing how it was planned to be measured. Overall, 18 different clinical outcomes were described among trials with at least 1 clinical efficacy/effectiveness outcome defined in the trial registration (Table S3). Clinical improvement and mortality accounted for 45% of the unique clinical outcome domains proposed to assess the efficacy/effectiveness of HQ or chloroquine treatment in patients diagnosed with COVID-19. Twenty-one different surrogate outcomes were identified in the registered RCTs planning to measure at least 1 surrogate outcome, with viral load and virological clearance accounting for 36% of the surrogate outcomes.
Twenty-four (47%) of the registered RCTs did not describe plans to assess a single safety outcome. Among the trials including a description of at least 1 safety outcome (n = 28), most (25; 89%) T A B L E 1 Characteristics of the randomized controlled trials for the treatment of COVID-19 with hydroxychloroquine or chloroquine (n = 51) n (%) did not report the method to be implemented for the detection of adverse events. The timeframe for the assessment of safety outcomes was not defined in 13 (46%) of the trials reporting plans to measure at least 1 safety outcome. The timeframe for the assessment of the safety outcomes ranged from 7 to 120 days (median 28; IQR: 14, 30).

Month of registration
The timeframe of safety outcome assessment was planned to be longer in comparison with the assessment of clinical outcomes (MD −9.8; 95% CI: −26.08 to 6.56; P = .23). The generic terminologies total, severe and serious adverse events accounted for 41% of the unique domains reported in at least 2 registered trial (Table S4). Twenty-five different domains were reported by only 1 clinical trial in which the assessment of at least 1 safety outcome was described. cines. 23 Given the potential severity of the COVID-19 infection, the need to find a mitigating or curative treatment is beyond urgent. In this study, we found that early RCTs proposed to evaluate the clinical efficacy/effectiveness and safety of HQ or chloroquine in the treatment of patients diagnosed with COVID-19 are designed to collect data that vary substantially in terms of the outcome domain used to determine the evidence base upon which these drugs will be judged. Moreover, data on safety outcomes are overlooked or only superficially included among the outcomes planned to be measured in these trials. Finally, essential information related to dosing schedules, treatment duration and timeframe of outcome assessment were frequently missing in the description of the RCTs. Overall, this analysis yielded 3 major areas of concern.

| DISCUSSION
4.1 | Selection of efficacy/effectiveness outcomes The severity of these adverse effects range from mild to severe; occasionally, these agents have been found to cause death.
Since we can anticipate a set of adverse events that is highly relevant to patients and clinical practice, the proposal of RCTs should contain plans to systematically assess fully defined adverse events according to appropriate timeframes. 12,16,29 For instance, QTc prolongation and drug-induced arrhythmias such as torsades de pointes, are of concern in critically ill patients with COVID-19 30 and should be carefully ascertained. Monitoring QTc through electrocardiographic tracings regularly would represent a systematic approach to the problem, even if monitoring is required to be performed remotely for safety reasons. The systematic assessment of adverse events can improve the accuracy of estimates within trials 31 while also minimizing bias. 32 Finally, the assessment of defined anticipated adverse events, together with their seriousness, severity and duration, would be more informative than the mere documentation of generic events.
In this study, we showed that retinopathy, cardiac and dermatological adverse events, and hypoglycaemia were planned to be assessed in a single clinical trial among the 51 trials that had been registered to evaluate the treatment with HQ or chloroquine for patients diagnosed with COVID-19. Outcomes of safety were not included among the outcomes defined in several of the proposed trials (24, 47%), while the remaining RCTs reported a nonspecific approach for observing safety outcomes. Based on these results, and the fact the many adverse effects are rare in small clinical trials, we are concerned that the evidence on the harms of these investigational drugs to patients diagnosed with COVID-19 may be likely to be incomplete and biased.

| Missing information
The comprehensive and prospective registration of clinical trials has been internationally supported since 2004 as a way to reduce the selective publication of studies and the selective reporting of outcomes. 33 Since the early years of clinical trial registration, ensuring that the registered data are complete and accurate has been a challenging objective of multiple enforcement mechanisms, including legal requirements. 29,33 Remarkably, approximately 1/3 of the registered RCTs included in this study had at least 1 piece of missing information, either related to treatment dose, duration, timeframes of outcome assessment or the lack of definition of a safety outcome. This is of particular concern amid the current pandemic scenario where the rush to test any potential helpful drug may pose a risk that low-quality evidence may be used to support clinical decisions with unpredictable impacts on patients and the health system.

| Limitations
We reviewed the information provided by all clinical trials focused

COMPETING INTERESTS
There are no competing interests to declare.

DATA SHARING
The data used for analysis are in the public domain. Nevertheless, the full dataset analysed can be requested from the authors.

NOMENCLATURE OF TARGETS AND LIGANDS
Key protein targets and ligands in this article are hyperlinked to corresponding entries in http://www.guidetopharmacology.org, the common portal for data from the IUPHAR/BPS Guide to PHARMACOLOGY.