Effect of Hydroxychloroquine in Hospitalized Patients with COVID-19: Preliminary results from a multi-centre, randomized, controlled trial.

Background: Hydroxychloroquine and chloroquine have been proposed as treatments for coronavirus disease 2019 (COVID-19) on the basis of in vitro activity, uncontrolled data, and small randomized studies. Methods: The Randomised Evaluation of COVID-19 therapy (RECOVERY) trial is a randomized, controlled, open-label, platform trial comparing a range of possible treatments with usual care in patients hospitalized with COVID-19. We report the preliminary results for the comparison of hydroxychloroquine vs. usual care alone. The primary outcome was 28-day mortality. Results: 1561 patients randomly allocated to receive hydroxychloroquine were compared with 3155 patients concurrently allocated to usual care. Overall, 418 (26.8%) patients allocated hydroxychloroquine and 788 (25.0%) patients allocated usual care died within 28 days (rate ratio 1.09; 95% confidence interval [CI] 0.96 to 1.23; P=0.18). Consistent results were seen in all pre-specified subgroups of patients. Patients allocated to hydroxychloroquine were less likely to be discharged from hospital alive within 28 days (60.3% vs. 62.8%; rate ratio 0.92; 95% CI 0.85-0.99) and those not on invasive mechanical ventilation at baseline were more likely to reach the composite endpoint of invasive mechanical ventilation or death (29.8% vs. 26.5%; risk ratio 1.12; 95% CI 1.01-1.25). There was no excess of new major cardiac arrhythmia. Conclusions: In patients hospitalized with COVID-19, hydroxychloroquine was not associated with reductions in 28-day mortality but was associated with an increased length of hospital stay and increased risk of progressing to invasive mechanical ventilation or death.


INTRODUCTION 53
Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), the cause of coronavirus 54 disease 2019 , emerged in China in late 2019 from a zoonotic source. 1 The majority 55 of COVID-19 infections are either asymptomatic or result in only mild disease. However, a 56 substantial proportion of infected individuals develop a respiratory illness requiring hospital 57 care, 2 which can progress to critical illness with hypoxemic respiratory failure requiring 58 prolonged ventilatory support. 3-6 Amongst COVID-19 patients admitted to UK hospitals, the case 59 was recorded on the occurrence of new major cardiac arrhythmia. In addition, routine health 126 care and registry data were obtained including information on vital status (with date and cause 127 of death); discharge from hospital; respiratory and renal support therapy. 128

Outcome measures 129
Outcomes were assessed at 28 days after randomization, with further analyses specified at 6 130 months. The primary outcome was all-cause mortality. Secondary outcomes were time to 131 discharge from hospital and, among patients not on invasive mechanical ventilation at 132 randomization, invasive mechanical ventilation (including extra-corporal membrane 133 oxygenation) or death. Subsidiary clinical outcomes included cause-specific mortality, use of 134 hemodialysis or hemofiltration, major cardiac arrhythmia (recorded in a subset), and receipt and 135 duration of ventilation. 136

Statistical Analysis 137
For the primary outcome of 28-day mortality, the log-rank 'observed minus expected' statistic 138 and its variance were used to both test the null hypothesis of equal survival curves and to 139 calculate the one-step estimate of the average mortality rate ratio, comparing all patients 140 allocated hydroxychloroquine with all patients allocated usual care. The few patients (2.1%) who 141 had not been followed for 28 days by the time of the data cut (22 June 2020) were either 142 censored on 22 June 2020 or, if they had already been discharged alive, were right-censored 143 for mortality at day 29 (that is, in the absence of any information to the contrary they were 144 assumed to have survived 28 days). Kaplan-Meier survival curves were constructed to display 145 cumulative mortality over the 28-day period. The same methods were used to analyze time to 146 hospital discharge, with patients who died in hospital right-censored on day 29. Median time to 147 discharge was derived from the Kaplan-Meier estimates. For the pre-specified composite 148 secondary outcome of invasive mechanical ventilation or death within 28 days (among those not 149 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 15, 2020. . https://doi.org/10.1101 receiving invasive mechanical ventilation at randomization), the precise date of starting invasive 150 mechanical ventilation was not available and so the risk ratio was estimated instead. Estimates 151 of absolute risk differences between patients allocated hydroxychloroquine and patients 152 allocated usual care were also calculated. 153 Pre-specified analyses of the primary outcome were performed in five subgroups defined by 154 characteristics at randomization: age, sex, level of respiratory support, days since symptom 155 onset, and predicted 28-day mortality risk (See Supplementary Appendix). One further pre-156 specified subgroup analysis (ethnicity) will be conducted once data collection is completed. 157 Observed effects within subgroup categories were compared using a chi-square test for trend 158 (which is equivalent to a test for heterogeneity for subgroups that have only two levels). 159 Estimates of rate and risk ratios (both hereon denoted RR) are shown with 95% confidence 160 intervals. All p-values are 2-sided and are shown without adjustment for multiple testing. All 161 analyses were done according to the intention-to-treat principle. The full database is held by the 162 study team which collected the data from study sites and performed the analyses at the Nuffield 163 Department of Population Health, University of Oxford. 164

Sample size and decision to stop enrolment 165
As stated in the protocol, appropriate sample sizes could not be estimated when the trial was 166 being planned at the start of the COVID-19 pandemic. As the trial progressed, the Trial Steering 167 Committee, blinded to the results of the study treatment comparisons, formed the view that if 168 28-day mortality was 20% then a comparison of at least 2000 patients allocated to active drug 169 and 4000 to usual care alone would yield at least 90% power at two-sided P=0.01 to detect a 170 proportional reduction of one-fifth (a clinically relevant absolute difference of 4 percentage 171 points between the two arms). 172 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 15, 2020. . https://doi.org/10.1101 The independent Data Monitoring Committee reviewed unblinded analyses of the study data 173 and any other information considered relevant at intervals of around 2 weeks. The committee 174 was charged with determining if, in their view, the randomized comparisons in the study 175 provided evidence on mortality that is strong enough (with a range of uncertainty around the 176 results that is narrow enough) to affect national and global treatment strategies. In such a 177 circumstance, the Committee would inform the Trial Steering Committee who would make the 178 results available to the public and amend the trial arms accordingly. Unless that happened, the 179 Trial Steering Committee, investigators, and all others involved in the trial would remain blind to 180 the interim results until 28 days after the last patient had been randomized to a particular 181 intervention arm. 182 On 4 June, in response to a request from the MHRA, the independent Data Monitoring 183 Committee conducted a review of the data and recommended the chief investigators review the 184 unblinded data on the hydroxychloroquine arm of the trial. The Chief Investigators and Trial 185 Steering Committee concluded that the data showed no beneficial effect of hydroxychloroquine 186 in patients hospitalized with COVID-19. Therefore enrolment of participants to the 187 hydroxychloroquine arm was closed on 5 June and the preliminary result for the primary 188 outcome was made public. Investigators were advised that any patients currently taking 189 hydroxychloroquine as part of the study should discontinue the treatment. June 2020), 7513 (67%) were eligible to be randomized to hydroxychloroquine (that is 195 hydroxychloroquine was available in the hospital at the time and the attending clinician was of 196 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 15, 2020. . https://doi.org/10. 1101 the opinion that the patient had no known indication for or contraindication to 197 hydroxychloroquine) ( Figure 1 and Table S1). Of these, 1561 were randomized to 198 hydroxychloroquine and 3155 were randomized to usual care with the remainder being 199 randomized to one of the other treatment arms. Mean age of study participants in this 200 comparison was 65.3 (SD 15.3) years (Table 1) and 38% patients were female. No children 201 were enrolled in the hydroxychloroquine comparison. A history of diabetes was present in 27% 202 of patients, heart disease in 26%, and chronic lung disease in 22%, with 57% having at least 203 one major comorbidity recorded. In this analysis, 90% of patients had laboratory confirmed 204 SARS-CoV-2 infection, with the result currently awaited for 1%. At randomization, 17% were 205 receiving invasive mechanical ventilation or extracorporeal membrane oxygenation, 60% were 206 receiving oxygen only (with or without non-invasive ventilation), and 24% were receiving neither. 207 Follow-up information was complete for 4619 (98%) of the randomized patients. Among those 208 with a completed follow-up form, 1395 (92%) patients allocated to hydroxychloroquine received 209 at least 1 dose (Table S2) and the median number of days of treatment was 6 days (IQR 3 to 10 210 days). 13 (0.4%) of the usual care arm received hydroxychloroquine. Use of azithromycin or 211 other macrolide drug during the follow-up period was similar in both arms (17% vs. 19%) as was 212 use of dexamethasone (8% vs. 9%). 213

Primary outcome 214
There was no significant difference in the proportion of patients who met the primary outcome of 215 28-day mortality between the two randomized arms (418 [ (Figure 2). Similar results were seen across all 218 five pre-specified subgroups ( Figure 3). In post hoc exploratory analyses restricted to the 4234 219 (90%) patients with a positive SARS-CoV-2 test result, the result was similar (rate ratio, 1.09, 95% 220 CI 0.96 to 1.24). 221 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

Secondary outcomes 222
Allocation to hydroxychloroquine was associated with a longer time until discharge alive from 223 hospital than usual care (median 16 days vs. 13 days) and a lower probability of discharge alive 224 within 28 days (rate ratio 0.92, 95% CI 0.85 to 0.99) ( Table 2). Among those not on invasive 225 mechanical ventilation at baseline, the number of patients progressing to the pre-specified 226 composite secondary outcome of invasive mechanical ventilation or death was higher among 227 those allocated to hydroxychloroquine (risk ratio 1.12, 95% CI 1.01 to 1.25). 228

Subsidiary outcomes 229
Information on the occurrence of new major cardiac arrhythmia was collected for 698 (44.7%) 230 patients in the hydroxychloroquine arm and 1357 (43.0%) in the usual care arm since these 231 fields were added to the follow-up form on 12 May 2020. Among these patients, there were no 232 significant differences in the frequency of supraventricular tachycardia (6.9% vs. 5.9%), 233 ventricular tachycardia or fibrillation (0.9% vs. 0.7%) or atrioventricular block requiring 234 intervention (0.1% vs. 0.1%) ( Table S3). Analyses of cause-specific mortality, receipt of renal 235 dialysis or hemofiltration, and duration of ventilation will be presented once all relevant 236 information (including certified cause of death) is available. There was one report of a serious 237 adverse reaction believed related to hydroxychloroquine; a case of torsades de pointes from 238 which the patient recovered without the need for intervention. 239

DISCUSSION 241
Although preliminary, these results indicate that hydroxychloroquine is not an effective treatment 242 for patients hospitalized with COVID-19. The lower bound of the confidence limit for the primary 243 outcome rules out any reasonable possibility of a meaningful mortality benefit. In addition, 244 allocation to hydroxychloroquine was associated with an increase in the duration of 245 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 15, 2020. . https://doi.org/10.1101/2020.07.15.20151852 doi: medRxiv preprint hospitalization and an increased risk of requiring invasive mechanical ventilation or dying for 246 those not on invasive mechanical ventilation at baseline. The results were consistent across 247 subgroups of age, sex, time since illness onset, level of respiratory support, and baseline-248 RECOVERY is a large, pragmatic, randomized, controlled platform trial designed to provide 250 rapid and robust assessment of the impact of readily available potential treatments for COVID-251 19 on 28-day mortality. Around 15% of all patients hospitalized with COVID-19 in the UK over 252 the study period were enrolled in the trial and the fatality rate in the usual care arm is consistent 253 with the hospitalized case fatality rate in the UK and elsewhere. 7,29,30 Only essential data were 254 collected at hospital sites with additional information (including long-term mortality) ascertained 255 through linkage with routine data sources. We did not collect information on physiological, 256 electrocardiographic, laboratory or virologic parameters. 257 Hydroxychloroquine has been proposed as a treatment for COVID-19 based largely on its in 258 is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 15, 2020. . https://doi.org/10. 1101 The primary concern with short-term high dose 4-aminoquinoline regimens is cardiovascular 271 toxicity. Hydroxychloroquine causes predictable prolongation of the electrocardiograph QT 272 interval that is exacerbated by co-administration with azithromycin, as widely prescribed in 273 COVID-19 treatment. [16][17][18] Although torsade de pointes has been described, serious 274 cardiovascular toxicity has been reported very rarely despite the high prevalence of 275 cardiovascular disease in hospitalized patients, the common occurrence of myocarditis in 276 COVID-19, and the extensive use of hydroxychloroquine and azithromycin together. The 277 exception is a Brazilian study which was stopped early because of cardiotoxicity. However in 278 that study, chloroquine 600 mg base was given twice daily for ten days, a substantially higher 279 total dose than used in other trials, including RECOVERY. 34,35 Pharmacokinetic modelling in 280 combination with blood concentration and mortality data from a case series of 302 chloroquine 281 overdose patients predicts that the base equivalent chloroquine regimen to the RECOVERY 282 hydroxychloroquine regimen is safe. 35 Hydroxychloroquine is considered to be safer than 283 chloroquine. 15 We did not observe excess mortality in the first 2 days of treatment with 284 hydroxychloroquine, the time when early effects of dose-dependent toxicity might be expected. 285 Furthermore, the preliminary data presented here did not show any excess in ventricular 286 tachycardia (including torsade de pointes) or ventricular fibrillation in the hydroxychloroquine 287

arm. 288
The findings indicate that hydroxychloroquine is not an effective treatment for hospitalized 289 patients with COVID-19 but do not address its use as prophylaxis or in patients with less severe 290 SARS-CoV-2 infection managed in the community. Treatment of COVID-19 with chloroquine or 291 hydroxychloroquine has been recommended in many treatment guidelines, including in Brazil, 292 China, France, Italy, Netherlands, South Korea, and the United States. 36 In a retrospective 293 cohort study in the United States, 59% of 1376 COVID-19 patients received 294 hydroxychloroquine. 22,37 Since our preliminary results were first made public on 5 June 2020, 295 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 15, 2020. . https://doi.org/10.1101/2020.07.15.20151852 doi: medRxiv preprint the U.S. Food and Drugs Administration has revoked the Emergency Use Authorization that 296 allowed hydroxychloroquine and chloroquine to be used for hospitalized patients with COVID-297

19, 38 and the World Health Organization (WHO) and the National Institutes for Health have 298
ceased trials of its use in this setting on the grounds of lack of benefit. The WHO has recently 299 released preliminary results from the SOLIDARITY trial on the effectiveness of 300 hydroxychloroquine in hospitalized COVID-19 patients that are consistent with the results from 301 the RECOVERY trial. 39 302 303 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 15, 2020. The views expressed in this publication are those of the authors and not necessarily those of 388 the NHS, the National Institute for Health Research or the Department of Health and Social 389 Care (DHCS). 390

Conflicts of interest 391
The authors have no conflict of interest or financial relationships relevant to the submitted work 392 to disclose. No form of payment was given to anyone to produce the manuscript. All authors 393 have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. 394 The Nuffield Department of Population Health at the University of Oxford has a staff policy of not 395 accepting honoraria or consultancy fees directly or indirectly from industry (see 396 https://www.ndph.ox.ac.uk/files/about/ndph-independence-of-research-policy-jun-20.pdf). 397 398 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 15, 2020. 19. 2020. https://www.who.int/news-room/detail/04-07-2020-who-discontinues-hydroxychloroquine-486 and-lopinavir-ritonavir-treatment-arms-for-covid-19. 487 488 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 15, 2020. . https://doi.org/10. 1101  Results are count (%), mean ± standard deviation, or median (inter-quartile range).* No children 492 (aged <18 years) were enrolled. †Includes 6 pregnant women. † † SARS-Cov-2 test results are 493 captured on the follow-up form, so are currently unknown for some. All tests for difference in 494 baseline characteristics between treatment arms give p>0.05. The 'oxygen only' group includes 495 non-invasive ventilation. Severe liver disease defined as requiring ongoing specialist care. 496 Severe kidney impairment defined as estimated glomerular filtration rate <30 mL/min/1.73m 2 . 9 497 (0.6%) patients allocated to hydroxychloroquine and 9 (0.3%) patients allocated to usual care 498 alone had missing data for days since symptom onset. 499 500 Table 2: Effect of allocation to hydroxychloroquine on main study outcomes 501 RR=rate ratio for the outcomes of 28-day mortality and hospital discharge, and risk ratio for the 502 outcome of receipt of invasive mechanical ventilation or death. CI=confidence interval. 503 * Analyses exclude those on invasive mechanical ventilation at randomization. For the pre-504 specified composite secondary endpoint of receipt of invasive mechanical ventilation or death 505 the absolute risk difference was 3.3 percentage points (95% CI 0.3 to 6.3). is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 15, 2020. . https://doi.org/10.1101 randomized to convalescent plasma vs control (1 [0.1%] patient allocated to hydroxychloroquine 514 arm vs 5 [0.2%] patients allocated to usual care) in accordance with protocol version 6.0. 515 Among the 167 sites that randomized at least 1 patient to the hydroxychloroquine comparison, 516 the median number randomized was 20 patients (inter-quartile range 11 to 41). includes patients receiving non-invasive ventilation. The method used for calculating baseline-531 predicted risk is described in the Supplementary Appendix. One further pre-specified subgroup 532 analysis (ethnicity) will be conducted once data collection is completed. 533 534 535 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 15, 2020. . https://doi.org/10.1101  is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 15, 2020. . https://doi.org/10.1101/2020.07.15.20151852 doi: medRxiv preprint . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 15, 2020. Total recruited # n=11197 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 15, 2020.  Figure 2: Effect of allocation to hydroxychloroquine on 28−day mortality . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 15, 2020. . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 15, 2020. . https://doi.org/10. 1101