INTRODUCTION

After the emergence of SARS-CoV-2 in late 2019, the virus spread rapidly, resulting in the worst pandemic in nearly a century. Early in the pandemic, health care systems struggled with maintaining adequate supply of personal protective equipment (PPE), and infections in health care workers (HCW) were reported globally [1]. Without availability of protective vaccines, there was a need to identify therapies that might prevent infection and could be taken regularly by HCW who were at high risk of frequent exposures, like approaches taken with malaria and human immunodeficiency virus pre-exposure prophylaxis. As is common with new diseases, repurposed drugs offered immediate options for therapeutics and usually with a well-known safety profile, allowing for more rapid introduction into clinical trials and thereafter into clinical practice.

Chloroquine had been previously reported to have in vitro antiviral activity against SARS-CoV-1 and MERS-CoV, and both chloroquine and hydroxychloroquine (HCQ) showed similar in vitro activity against SARS-CoV-2 [2, 3, 4] . Thus, clinicians turned to HCQ early in the pandemic as a therapy that might have clinical benefit for treating COVID-19, the disease associated with SARS-CoV-2 infection, as investigators began testing the drug's safety and efficacy in the treatment and prevention of COVID-19. While vaccines are now available for the prevention of SARS-CoV-2 infection, access to sufficient quantities of vaccine remains challenging in some areas, and hesitancy towards vaccines and treatments further complicates the public health response. Thus, preventative therapies for SARS-CoV-2 infection remain relevant.

The Healthcare Worker Exposure Response and Outcomes (HERO) Registry (NCT04342806) was created as a community of HCW from across the United States to learn about issues impacting frontline workers and to offer opportunities to participate in research studies [5]. The HERO-HCQ trial was one of the first studies in the United States to test the safety and efficacy of HCQ as pre-exposure prophylaxis in frontline HCW. HERO-HCQ leveraged both the HERO Registry as well as its relationship with PCORnet®, The National Patient-Centered Clinical Research Network as a pragmatic and largely remote clinical trial, using a patient-facing online portal to capture frequent patient-reported outcomes [6]. The primary objective of HERO-HCQ was to evaluate the efficacy of HCQ in preventing SARS-CoV-2 clinical infection in HCW when taken daily. Secondary objectives of the study were to assess the efficacy of HCQ in preventing asymptomatic viral shedding of SARS-CoV-2 among HCW and to assess the safety and tolerability of HCQ in this study population. The protocol is available at https://heroesresearch.org/hero-hcq/.

Study Design

HERO-HCQ was a multicenter, double-blind, randomized, parallel-group study designed to evaluate the superiority of HCQ vs. placebo for COVID-19 pre-exposure prophylaxis in HCW.

The HERO-HCQ trial was reviewed by the Duke University School of Medicine Institutional Review Board and approved by the Western Institutional Review Board (Pro00105274). Additional details on the HERO-HCQ design can be accessed at Friedland et al., 2021.

Eligible participants were age 18 or older, working in a health care setting with potential exposure to patients with COVID-19, and provided informed consent. The main exclusions were prior diagnosis of COVID-19 infection or contraindications to HCQ [7].

Participants were randomized in a 1:1 ratio to receive HCQ or placebo at the level of the individual participant via the study portal. Randomization was stratified by clinical site using a permuted block design with varying block sizes. In the intervention arm, participants received an oral loading dose of study drug at 600 mg twice on the first day, followed by 400 mg daily for 29 days. Because of a lack of phase 1b data at the time, the dose was selected based on available in vitro studies that reported a wide range of 50% effective concentration (EC50) of chloroquine and HCQ for SARS-CoV-2 as well as known variability of absorption and of tissue distribution into the lung [8-14] . In the control arm, placebo tablets were administered using the same dosage schedule and number of tablets as the intervention arm. The placebo was similar in appearance to the study drug and packaged and labeled in a masked manner in compliance with regulatory requirements. All study drug doses were oral self-administrations. Study drug was supplied as 200-mg tablets, and each eligible participant was provided a quantity sufficient for 30 days.

Participants were prescreened through the HERO Registry, and eligibility was confirmed by the site by phone or in person [5]. Participants were able to electronically consent through the portal, which was done at the time of the site visit or in advance. There were two on-site visits – one at baseline and another at 30 days. Baseline assessments included a nasopharyngeal swab for SARS-CoV-2 and a blood sample to assess baseline SARS-CoV-2 nucleocapsid IgG antibody status. Weekly follow-up was performed remotely via standardized questionnaires utilizing the online portal. These questionnaires included screening for COVID-19 clinical signs/symptoms and self-reporting of COVID-19 testing and diagnosis. Additionally, participants were able to self-report medication changes, hospitalizations, clinical events, and adverse events. A call center provided support for missed visits to re-engage and remind participants to complete the questionnaires.

The second on-site visit at approximately 30 days was completed to assess study drug adherence and any subsequent clinical or safety events. A nasopharyngeal swab for SARS-CoV-2 PCR and a blood sample for SARS-CoV-2 nucleocapsid IgG antibody were obtained. Individual participants received study drug for 30 days and were followed for an additional 30 days for clinical events and patient-reported outcomes. An end-of-study virtual visit was conducted approximately 60 days after randomization via the direct-to-participant portal or call center to assess for any subsequent clinical or safety events.

Outcomes

The primary outcome was a composite of confirmed or suspected clinical infection with COVID-19 through 30 days, which was defined as new onset of fever, cough, or dyspnea and confirmed SARS-CoV-2 PCR positive test result via local testing or suspected COVID-19 disease without confirmation testing due to local restrictions or policies. Participants who developed symptoms of COVID-19 were expected to follow local clinical and/or employee health protocols for testing and management. Secondary outcomes were (1) viral shedding of SARS-CoV-2 at 30 days and (2) safety and tolerability as determined by subject-reported adverse events that met criteria per protocol for serious adverse events and HCQ-associated events of special interest; this latter group comprised fever, arrhythmia (ventricular), psychosis, angioedema, prolonged QT interval, secondary bacterial infection, and suicidal ideation.

Exploratory outcomes were (1) SARS-CoV-2 nucleocapsid IgG seroconversion at 30 days; (2) COVID-19 complications including hospitalization, ICU level care, or need for invasive ventilation; (3) days sick or lost work time; (4) self-reported health and well-being obtained from the Patient-Reported Outcomes Measurement Information System (PROMIS) Emotional Distress-Anxiety-Short Form [15], a Single Item Burnout Measure [16, 17, 18, 19], and the Patient Health Questionnaire (PHQ-2) [20, 21]; and (5) patient-reported clinical infections among household contacts and other impacts on HCW household.

Statistical Analysis

The original sample size of approximately 15,000 randomized participants was selected to yield high power for testing the primary outcome of clinical infection with SARS-CoV-2 assuming the usual risk of SARS-CoV-2 infection is 5%. This sample size was expected to provide greater than 80% power to detect a 1% absolute decrease (20% relative decrease) in COVID-19 infection rates between treatment arms. These calculations assumed a two-sided Type I error rate of 0.05 with 1:1 randomization. In October 2020, due to slower than expected enrollment and changing community attitudes about HCQ effectiveness, the study protocol was amended to reduce the total sample size to 2000, which provided 80% power to detect a 50% relative decrease in the risk of COVID-19 infections assuming a placebo group risk of 5%.

The primary analyses was conducted in the mITT population, which included participants with negative nasal swab at baseline (1,359 of the 1,360 subjects randomized). Statistical comparisons were performed using two-sided significance tests. The primary endpoint was clinical infection with SARS-CoV-2 through the 30-day period. Data collected during the 60-day follow-up were included for the safety analyses. For the primary outcome of clinical infection with SARS-CoV-2, comparisons between treatment arms were presented as differences in proportions with 95% confidence intervals (CIs) using the Miettinen-Nurminen method and a p-value calculated using the Fisher's exact test. A secondary analysis was based on a logistic regression model with an indicator variable for the treatment group. Supplemental analyses were conducted to (1) examine the differences using other methods for constructing the CIs [22] and (2) compute the common odds ratio using the Mantel–Haenszel test stratifying by enrolling site [23].

Subgroup analyses were planned for age, sex, race and ethnicity. For each subgroup analysis, a logistic regression model was estimated, with additional terms identifying subgroup membership and intervention by subgroup interaction. The statistical comparisons of serious adverse events and events of special interest were based on chi-square tests and Fisher's exact test. All analyses were conducted using SAS Version 9.4 software. The Duke Clinical Research Institute served as the Statistical and Data Coordinating Center.

Patient and Public Involvement

HCW were engaged in the HERO program and trial through membership in HERO governance, including participation in the steering committee and subcommittees. HCW stakeholders reviewed enrollment materials and the study protocol, and advised on trial conduct throughout the study. An independent data safety monitoring board, which included an HCW representative, met regularly and monitored participant safety and study performance. The protocol was publicly shared and is available on heroesresearch.org.

RESULTS

The HERO-HCQ trial start-up timeline was 4 weeks from concept to first participant randomized (Supplemental Figure 1). Between April 2020 and November 2020, a total of 1360 participants were enrolled and randomized from 34 US sites participating in PCORnet (Figure 1, see Supplemental Table 1). One participant had a positive SARS-CoV-2 PCR test at the time of the baseline visit and was excluded from the primary analysis population. Overall, 92.9% and 92.3% of randomized participants completed their PCR and serology tests, respectively, at their Day 30 visit with no significant difference by treatment group. The Day 60 visit was completed by 89.0% of total participants.

Baseline Participant Characteristics

The mean age of the HCW in the study population was 43.6 years, 65.3% were female, 90.8% reported being White race, and 5.8% self-reported as Hispanic or Latino ethnicity (Table 1). The median body mass index was 27.1 kg/m2, and 33.2% of the population was considered obese (BMI ≥ 30 kg/m2). The most common comorbidities were hypertension (14.6%), asthma (9.9%), and diabetes (4.0%). Among the most common HCW locations were the emergency department (14.0%), ambulatory or outpatient care (9.5%), inpatient medical unit (8.5%), emergency medical services (8.1%), intensive care unit (7.9%), inpatient surgical unit (6.8%), and dedicated COVID-19 unit (5.7%). Among the enrolled participants, the most common occupation/employment characteristics were registered nurse (26.2%), physician (21.3%), nurse practitioner (5.2%), and paramedic (5.2%). Twelve (0.9%) participants were positive for SARS-CoV-2 nucleocapsid IgG at study enrollment.

Primary Endpoint

There were a total of 94 primary endpoint events during the 30-day follow-up period. Most of these endpoints were suspected clinical infection (n=85) rather than confirmed clinical infection with COVID-19 (n=9) (Table 2). The most common presenting symptoms were cough (86.2%), fatigue (68.1%), headache (66.0%), and muscle aches/joint pain (51.1%). There were numerically fewer primary endpoint events in the HCQ group (41 [6.0%]) compared with the placebo group (53 [7.8%]); however, this difference of -1.8% (95% CI -4.6% to 0.9%) was not significant (Fisher's exact p=0.20) (Supplemental Table 2). A secondary analysis based on a logistic regression model yielded similar results (OR 0.75, 95% CI 0.49 to 1.15, p=0.18). Among the participants with confirmed clinical infection, there were numerically fewer in the HCQ group (3 events [0.4%]) compared to the placebo group (6 events [0.9%]), and the difference was not significant (0.45%, 95% CI -1.54% to 0.50%) (Figure 2A,B). Supplemental analyses using the Mantel–Haenszel method, which stratified by enrolling site, yielded a similar estimate of the common OR (0.69, 95% CI 0.45 to 1.05). However, there was evidence of heterogeneity at the site level for the primary endpoint (p=0.011).

Subgroup Analyses

The prespecified subgroups for the primary endpoint are shown in Figure 3. All subgroups except for the youngest age group (18-35 years) showed estimated odds ratios <1.0 favoring HCQ, but none were significant with confidence intervals overlapping 1.0. Additionally, there were no statistically significant findings for the subgroup interaction tests suggesting a lack of evidence for heterogeneous treatment effects.”

Secondary Endpoints

Four participants had a positive SARS-CoV-2 PCR test at the Day 30 visit, with each treatment arm having two positive cases (0.3% viral shedding rate at 30 days for both groups, p=1.00). Similarly, there were few seroconversions, defined as having a negative SARS-CoV-2 nucleocapsid IgG at entry and a positive IgG at Day 30, with two (0.3%) participants in the HCQ arm and five (0.8%) in the placebo arm having evidence of seroconversion. There were no deaths reported during the study period.

Participants in the HCQ arm reported lower levels of emotional distress and anxiety during the 30-day treatment period based on the worst recorded PROMIS Short Form T-scores (49.8 vs. 51.1 for a difference of -1.3 points, 95% CI -2.2 to -0.4, p=0.007). The percentage of participants reporting the PHQ-2 score ≥ 3 was not different between the two treatments (6.0% for HCQ vs. 7.0% for placebo; p=0.50). Levels of burnout were not different between groups over the treatment period (p=0.065); however, numerically more participants in the HCQ arm reported no symptoms of burnout (21.6% vs. 18.0%) compared to placebo.

Adherence to Treatment and Study Drug Discontinuation

At day 30, self-reported adherence (i.e., taking the drug for 29 days) was 94.4% for HCQ and 95.7% for placebo (p=0.32). Permanent discontinuation rates were 4.1% for HCQ and 2.7% for placebo. Permanent discontinuation due to adverse events was more common in the HCQ group (12 of 28 discontinuations) than the placebo group (3 of 18 discontinuations).

Adverse Events and Safety

Adverse events of special interest (fever, ventricular arrhythmia, psychosis, angioedema, prolonged QT interval, secondary bacterial infection, and suicidal ideation) over the 60 days of follow-up were similar across groups (7 subjects in HCQ arm and 8 in placebo). At 60 days, serious adverse events were reported for 3 participants (0.4%) in the HCQ group and 2 (0.3%) in the placebo group. Of those serious adverse events, 2 for the HCQ group and 1 for the placebo group resulted in hospitalizations (Table 3).

DISCUSSION

Statement of Principal Findings

The study was not powered to detect a small beneficial effect and the test of the primary endpoint does not provide evidence of a benefit for HCQ for PrEP in a high-risk HCW population.

Strengths and Weaknesses in Context

The original study design was powered to show a 20% relative treatment effect assuming a 5% event rate in the placebo arm. However, due to slowed enrollment early in the study, the study was amended to decrease the sample size and hence the power, increasing the detectable relative treatment effect to 50%. Thus, the study was not powered to detect a small treatment effect. This outcome was not unique to this randomized trial; a 2021 analysis concluded that among the early COVID-19 studies, only 5% were both randomized and adequately powered [24].

While the partially remote nature of the trial was novel and improved feasibility during a pandemic, it also resulted in the limitation that we did not have laboratory confirmation for COVID-19–like illness. Early in the pandemic, in some regions, testing was not performed per local policies in HCW with suspected infection and mild or moderate symptoms. Per the study protocol, these events were defined as suspected cases and were combined with the confirmed cases in the primary composite outcome. This resulted in few confirmed COVID-19 infections; thus, our primary outcome was primarily comprised of suspected COVID-19 clinical infections. While the study did not have frequent PCR testing, testing at entry and end-of-study showed low cross-sectional asymptomatic rates of viral shedding in the study population. Similarly, the seroconversion rate was low (0.6%) over the 30-day intervention period. These low rates of asymptomatic shedding and seroconversion suggest that SARS-CoV-2 infection was lower than expected in the study population and may be overestimated by our composite primary outcome definition.

Further, the trial was designed to evaluate the clinical efficacy of HCQ for the prevention of SARS-CoV-2 with dosing based on prior in vitro data on activity against SARS-CoV-2, smaller trials, or observational data and dosing strategies already used safely in medical practice.  As such, we did not perform a dose-finding study nor incorporate testing of HCQ plasma concentrations. There are strengths to highlight in the HERO-HCQ trial that might inform future clinical trials. As with many trials, there was a need for rapid development and execution; HERO-HCQ went from initial discussions with the funder to first participant enrolled in 1 month. Although there was a need to move expeditiously, HERO-HCQ did not trade speed for validity. Many studies during the pandemic have had limited impact due to significant study design limitations, particularly lack of randomization and blinding to intervention arm. HERO-HCQ is a randomized, double-blind, placebo-controlled trial. In part to respond to the need to limit in-person activities and to design a trial that was patient-centered and pragmatic, HERO-HCQ used direct-to-participant recruitment and a participant-facing portal to capture patient-reported outcomes. As recruitment and retention remain significant challenges for many clinical trials, we believe these strategies should be considered more frequently in trial design across more disease states.

Globally, there remains a role for preventative therapeutics against SARS-CoV-2 infection, particularly in regions where access to or acceptance of preventative vaccines remains low [25, 26]. Furthermore, additional preventative options may become more relevant as novel variants or subvariants emerge with vaccine immune escape. In addition to HERO-HCQ, other similar randomized clinical trials investigating the safety and efficacy of HCQ as pre-exposure prophylaxis in HCWs have been completed and were not able to demonstrate that HCQ significantly reduces the risk of confirmed or clinically suspected SARS-CoV-2 infection [27-32]. Therefore, although this study was not powered to detect a large effect size, taken together, these studies do not support a role for HCQ for pre-exposure prophylaxis for COVID-19 in HCW.

Conclusion and Future Research

The prophylactic use of HCQ by HCW was safe but not effective to prevent COVID-19 clinical infection. This is one of several negative studies assessing HCQ for prevention of COVID-19, all of which were not powered for <50% efficacy. Due to ongoing interest in HCQ globally, a meta-analysis of all published randomized, placebo-controlled trials could provide more definitive evidence through a pooled analysis.

Acknowledgments

The authors thank the site investigators, study teams, and participants who made this project possible, and who contributed to scientific research during a time of immediate need.

Competing interests: All authors have completed the ICMJE uniform disclosure form and declare: support from the Patient-Centered Outcomes Research Institute (PCORI) for the reported work; Vir Biotechnology (SC); AstraZeneca, GSK, Novartis, Pulmatric, Sanofi-Aventis, Genentech, Teva (MC); Vir Biotechnology (KA); Centers for Disease Control, Infection Control Education for Major Sports, LLC, UpToDate, AHRQ, NIAID (DA); NIH (KA); Pfizer (EO); Abbott (RR).

Funding: This project was funded by the Patient-Centered Outcomes Research Institute (PCORI), Contract Number COVID-19-2020-001.

Ethical approval: The HERO-HCQ trial was reviewed by the Duke University School of Medicine Institutional Review Board and approved by the Western Institutional Review Board (Pro00105274).

Author contributions: The senior author (AH) obtained funding from the Patient-Centered Outcomes Research Institute (PCORI) to conduct the study (Contract Number COVID-19-2020-001). AH and the lead author (SN) affirm the manuscript is an honest, accurate, and transparent account of the study being reported; that no important aspects of the study have been omitted, and that any discrepancies from the study as planned have been explained. SN, LC, EF, AF, RO, CW, KA, and AH conceived, designed, and oversaw the study, and led manuscript development. JG, HM, JS, and KA were responsible for data management and statistical analysis. All of the authors participated in data collection and acquisition; reviewed the manuscript for important intellectual content; and gave administrative, technical, or material support. The lead author had full access to all the data in the study and takes responsibility for the integrity, transparency of the data, and the accuracy of the data analysis. The corresponding author attests that all listed authors meet authorship criteria and that no others meeting the criteria have been omitted.

Data sharing: Data will be made available upon reasonable request and per the PCORI guidelines for open science and transparency.

References

[1] CDC COVID-19 Response Team. Characteristics of health care personnel with COVID-19 — United States, February 12–April 9, 2020. MMWR Morb Mortal Wkly Rep 2020;69:477-81. https://doi.org/10.15585/mmwr.mm6915e6.

[2] Kalil AC. Treating COVID-19—off-label drug use, compassionate use, and randomized clinical trials during pandemics. JAMA 2020;323:1897-8. https://doi.org/10.1001/jama.2020.4742.

[3] McCreary EK, Pogue JM. COVID-19 treatment: a review of early and emerging options. Open Forum Infect Dis 2020;7(4):ofaa105. https://doi.org/10.1093/ofid/ofaa105.

[4] Wang M, Cao R, Zhang L, Yang X, Liu J, Xu M, et al. Remdesivir and chloroquine effectively inhibit the recently emerged novel coronavirus (2019-nCoV) in vitro. Cell Res 2020;30:269-71. https://doi.org/10.1038/s41422-020-0282-0.

[5] Forrest CB, Xu H, Thomas LE, Webb LE, Cohen LW, Carey TS, et al.; HERO Registry Research Group. Impact of the early phase of the COVID-19 pandemic on US healthcare workers: results from the HERO Registry. J Gen Intern Med 2021;36:1319-26. https://doi.org/10.1007/s11606-020-06529-z.

[6] Forrest CB, McTigue KM, Hernandez AF, Cohen LW, Cruz H, Haynes K, et al. PCORnet® 2020: current state, accomplishments, and future directions. J Clin Epidemiol 2020;129:60-7. https://doi.org/10.1016/j.jclinepi.2020.09.036.

[7] Friedland A, Hernandez AF, Anstrom KJ, Chen-Lim ML, Cohen LW, et al. Design of the Healthcare Worker Exposure Response and Outcomes (HERO) research platform. Contemp Clin Trials. 2021;109:106525. https://doi.org/10.1016/j.cct.2021.106525

[8] Liu J, Cao R, Xu M, Wang X, Zhang H, et al. Hydroxychloroquine, a less toxic derivative of chloroquine, is effective in inhibiting SARS-CoV-2 infection in vitro. Cell Discov 2020; 6(16). https://doi.org/10.1038/s41421-020-0156-0

[9] Yao X, Ye F, Zhang M, Cui C, Huang B, et al. In Vitro Antiviral Activity and Projection of Optimized Dosing Design of Hydroxychloroquine for the Treatment of Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2). Clin Infect Dis 2020;71(15):732-739. https://doi.org/10.1093/cid/ciaa237

[10] Keyaerts E, Vijgen L, Maes P, Neyts J, Van Ranst M. In vitro inhibition of severe acute respiratory syndrome coronavirus by chloroquine. Biochem Biophys Res Commun. 2004;323(1):264-268. https://doi.org/10.1016/j.bbrc.2004.08.085

[11] Vincent MJ, Bergeron E, Benjannet S, Erickson BR, Rollin PE, et al. Chloroquine is a potent inhibitor of SARS coronavirus infection and spread. Virol J. 2005;2:69. Published 2005 Aug 22. https://doi.org/10.1186/1743-422X-2-69

[12] Fox RI. Mechanism of action of hydroxychloroquine as an antirheumatic drug. Semin Arthritis Rheum. 1993;23(2 Suppl 1):82-91. https://doi.org/10.1016/s0049-0172(10)80012-5

[13] Tett SE. Clinical pharmacokinetics of slow-acting antirheumatic drugs. Clin Pharmacokinet. 1993;25(5):392-407. https://doi.org/10.2165/00003088-199325050-00005

[14] McChesney EW, Conway WD, Banks WF Jr, Rogers JE, Shekosky JM. Studies of the metabolism of some compounds of the 4-amino-7-chloroquinoline series. J Pharmacol Exp Ther. 1966;151(3):482-493.

[15] Pilkonis PA, Choi SW, Reise SP, Stover AM, Riley WT, Cella D; PROMIS Cooperative Group. Item banks for measuring emotional distress from the Patient-Reported Outcomes Measurement Information System (PROMIS®): depression, anxiety, and anger. Assessment 2011;18:263-283. https://doi.org/10.1177/1073191111411667.

[16] West CP, Dyrbye LN, Sloan JA, Shanafelt TD. Single item measures of emotional exhaustion and depersonalization are useful for assessing burnout in medical professionals. J Gen Intern Med 2009;24:1318-21. https://doi.org/10.1007/s11606-009-1129-z.

[17] Rohland BM, Kruse GR, Rohrer JE. Validation of a single-item measure of burnout against the Maslach Burnout Inventory among physicians. Stress Health 2004;20:75-9. https://doi.org/10.1002/smi.1002.

[18] McMurray JE, Linzer M, Konrad TR, Douglas J, Shugerman R, Nelson K. The work lives of women physicians results from the physician work life study. The SGIM Career Satisfaction Study Group. J Gen Intern Med 2000;15:372-80. https://doi.org/10.1111/j.1525-1497.2000.im9908009.x.

[19] Dolan ED, Mohr D, Lempa M, Joos S, Fihn SD, Nelson KM, et al. Using a single item to measure burnout in primary care staff: a psychometric evaluation. J Gen Intern Med 2015;30:582-7. https://doi.org/10.1007/s11606-014-3112-6.

[20] Kroenke K, Spitzer RL, Williams JB. The Patient Health Questionnaire-2: validity of a two-item depression screener. Med Care 2003;41:1284-92. https://doi.org/10.1097/01.MLR.0000093487.78664.3C.

[21] Arroll B, Goodyear-Smith F, Crengle S, Gunn J, Kerse N, Fishman T. Validation of PHQ-2 and PHQ-9 to screen for major depression in the primary care population. Ann Fam Med 2010;8:348-53. https://doi.org/10.1370/afm.1139.

[22] Garner W. Constructing confidence intervals for the differences of binomial proportions in SAS®. Available at: https://www.lexjansen.com/wuss/2016/127_Final_Paper_PDF.pdf.

[23] Cochran WG. Some methods for strengthening the common X2 tests. Biometrics 1954;10:417-51. https://doi.org/10.2307/3001616.

[24] Bugin K, Woodcock J. Trends in COVID-19 therapeutic clinical trials. Nat Rev Drug Discov. 2021 Apr;20(4):254-255. https://doi.org/10.1038/d41573-021-00037-3.

[25] Lazarus JV, Wyka K, White TM, Picchio CA, Rabin K, et al. Revisiting COVID-19 vaccine hesitancy around the world using data from 23 countries in 2021. Nat Commun. 2022;13(1):3801. https://doi.org/10.1038/s41467-022-31441-x

[26] Teperowski Monrad J, Quaade S, Powell-Jackson T. Supply, then demand? Health expenditure, political leanings, cost obstacles to care, and vaccine hesitancy predict state-level COVID-19 vaccination rates. Vaccine. 2022;40(45):6528-6548. https://doi.org/10.1016/j.vaccine.2022.08.050

[27] Abella BS, Jolkovsky EL, Biney BT, Uspal JE, Hyman MC, et al. Efficacy and safety of hydroxychloroquine vs placebo for pre-exposure SARS-CoV-2 prophylaxis among health care workers: a randomized clinical trial. JAMA internal medicine 2021;181(2):195-202.

[28] Grau-Pujol B, Camprubí-Ferrer D, Marti-Soler H, et al. Pre-exposure prophylaxis with hydroxychloroquine for COVID-19: a double-blind, placebo-controlled randomized clinical trial. Trials. 2021;22(1):808. Published 2021 Nov 15. https://doi.org/10.1186/s13063-021-05758-9

[29] McKinnon JE, Wang DD, Zervos M, Saval M, Marshall-Nightengale L, et al. Safety and tolerability of hydroxychloroquine in health care workers and first responders for the prevention of COVID-19: WHIP COVID-19 Study. International Journal of Infectious Diseases 2022;116:167-73.

[30] Naggie S, Milstone A, Castro M, Collins SP, Lakshmi Set al. Hydroxychloroquine for pre-exposure prophylaxis of COVID-19 in health care workers: a randomized, multicenter, placebo-controlled trial (HERO-HCQ). medRxiv 2021

[31] Rajasingham R, Bangdiwala AS, Nicol MR, Skipper CP, Pastick KA, Axelrod ML, et al.; COVID PREP Team. Hydroxychloroquine as pre-exposure prophylaxis for COVID-19 in healthcare workers: a randomized trial. Clin Infect Dis 2021;72:e835-43. https://doi.org/10.1093/cid/ciaa1571.

[32] Rojas-Serrano J, Portillo-Vásquez AM, Thirion-Romero I, Vázquez-Pérez J, Mejía-Nepomuceno F, Ramírez-Venegas A, et al. Hydroxychloroquine for prophylaxis of COVID-19 in health workers: a randomized clinical trial. Posted May 16, 2021. https://www.medrxiv.org/content/10.1101/2021.05.14.21257059v1. https://doi.org/10.1101/2021.05.14.21257059