Implementation of a fully remote randomized clinical trial with cardiac monitoring.

Background: The coronavirus disease 2019 (COVID-19) pandemic has challenged researchers performing clinical trials to develop innovative approaches to mitigate infectious risk while maintaining rigorous safety monitoring.
Methods: In this report we describe the implementation of a novel exclusively remote randomized clinical trial (ClinicalTrials.gov NCT04354428) of hydroxychloroquine and azithromycin for the treatment of the SARS-CoV-2-mediated COVID-19 disease which included cardiovascular safety monitoring. All study activities were conducted remotely. Self-collected vital signs (temperature, respiratory rate, heart rate, and oxygen saturation) and electrocardiographic (ECG) measurements were transmitted digitally to investigators while mid-nasal swabs for SARS-CoV-2 testing were shipped. ECG collection relied on a consumer device (KardiaMobile 6L, AliveCor Inc.) that recorded and transmitted six-lead ECGs via participants' internet-enabled devices to a central core laboratory, which measured and reported QTc intervals that were then used to monitor safety.
Results: Two hundred and thirty-one participants uploaded 3245 ECGs. Mean daily adherence to the ECG protocol was 85.2% and was similar to the survey and mid-nasal swab elements of the study. Adherence rates did not differ by age or sex assigned at birth and were high across all reported race and ethnicities. QTc prolongation meeting criteria for an adverse event occurred in 28 (12.1%) participants, with 2 occurring in the placebo group, 19 in the hydroxychloroquine group, and 7 in the hydroxychloroquine + azithromycin group. Conclusions: Our report demonstrates that digital health technologies can be leveraged to conduct rigorous, safe, and entirely remote clinical trials.

Introduction

Coronavirus disease 2019 (COVID-19), caused by infection with the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), has spread to more than 194 million people and caused more than 4 million deaths globally as of July 2021[1]. In the context of a critical need to stem the morbidity and mortality of the SARS-CoV-2 pandemic, the development and testing of therapeutics has proceeded at an unprecedented pace[2-5]. Hydroxychloroquine and azithromycin, two widely available drugs, were of intense interest at the outset of the pandemic given their potential efficacy against SARS-CoV-2 based on in vitro and early observational data[6-8]. Therefore, we designed a randomized, placebo-equivalent trial to test two therapeutic regimens—hydroxychloroquine and hydroxychloroquine combined with azithromycin—for efficacy in reducing disease progression in ambulatory patients with COVID-19. In light of concerns regarding potential cardiovascular toxicities related to hydroxychloroquine- and azithromycin-induced QT prolongation, this trial was designed to rigorously evaluate the cardiovascular safety of these regimens[9-14]. Although hydroxychloroquine has been shown to be ineffective for the treatment of COVID-19[15-17], our experience with digital cardiac safety monitoring provides important proof of principle for a technology that can be applied broadly to facilitate more pragmatic, “site-less,” and increasingly digital clinical trials of the future. Such remote trials may significantly improve access for patients who live in remote underserved places, work, have children, and/or lack transportation, which are common barriers to trial participation, and thus may help democratize clinical trials.

In this study, we utilized digital technologies that facilitated collaboration among scientists, implemented internet-based recruitment and enrollment, and leveraged rapid computer-assisted data analysis. Here, we describe our experience in the design and execution of this fully remote clinical trial, using digital technologies and infrastructure to provide real-time ascertainment of cardiovascular safety and risk in the evaluation of an investigational drug regimen for COVID-19.

There are three key elements of a digital clinical trial: recruitment and retention, patient-reported and -collected health data, and digital analytics[18]. Digital recruitment and analytics techniques are already in widespread use, as evidenced by several recent remote COVID-19 trials[19-23], but the incorporation of real-time digital health data collection in clinical trials has been limited.

We conducted a trial that incorporated real-time digital health data collection. The cardiovascular safety concerns inherent to hydroxychloroquine and azithromycin necessitated monitoring of the QT interval using the electrocardiogram (ECG). We hypothesized that patients could check their own vital signs and self-administer nasal swabs, a technique that is comparable in sensitivity to healthcare workers who were administered nasopharyngeal swabs[24], as well as remotely collect ECG tracings to monitor QT intervals. Previous trials of atrial fibrillation have relied on remote rhythm monitoring with Holter monitors or wearable extended recording devices to assess outcomes[25,26], but these modalities are limited in that they are adjudicated post hoc after being physically mailed to a processing facility. When QT interval monitoring is a relevant concern, active surveillance is typically required due to potential progression to torsades de pointes with continued dosing of the inciting agent after QT prolongation has developed[27].

This dictated the rapid development of a system of remote ECG collection, analysis, and reporting that was easy for participants to self-administer and would provide accurate, timely, and clinically meaningful data. While there are numerous consumer ECG devices on the market, the fact that multilead ECG tracings are necessary to make an accurate assessment of the QT interval[28,29] favored the selection of a device capable of simultaneous six-lead recording. We utilized the KardiaMobile 6L, the first device cleared by the US Food and Drug Administration for use in monitoring QT interval[30].

Methods

Design and oversight

In this multicenter, double-blind placebo-controlled trial, participants were randomized into three arms in a 1:1:1 fashion to hydroxychloroquine + placebo (folic acid), hydroxychloroquine + azithromycin, and a placebo-equivalent control (ascorbic acid + folic acid). The complete methods have been previously published[17]. The study was registered at ClinicalTrials.gov (NCT04354428).

The trial was conducted entirely remotely (Fig. 1), from enrollment through data collection and follow-up, with the first enrollment occurring on 16 April 2021 and the last enrollment date being 28 July 2021. Five US institutions enrolled participants. Written informed consent was obtained from all participants. The study was approved by the Western Institutional Review Board with reliance agreements from the collaborating institutions. An external and independent data and safety monitoring board provided oversight.

Fig. 1

Timeline of participant study drug administration and data collection with schematic demonstrating data processing and storage.

Visual representation of the study protocol. Study drugs were taken on protocol days 1–10. ECGs, mid-nasal viral swabs, symptom surveys, and vital signs were self-collected on protocol days 1–14, while only ECGs, symptom surveys, and vital signs were obtained on days 21 and 28. On days 2, 4, 9, 14, and 28, planned investigator-initiated contact was undertaken to collect subjective data and encourage adherence. Vital signs were collected twice daily. ECGs were uploaded to the secure web portal where study coordinators pushed them to the core ECG laboratory and then recorded the returned interpretation, as described in the “Methods” section. Viral swabs were sent to the core virology laboratory. All data were warehoused in REDCap[33].

Inclusion and exclusion criteria

Individuals between the ages of 18 and 80 who tested positive for SARS-CoV-2 via polymerase chain reaction (PCR) assay within the prior 72 h were eligible for inclusion provided they possessed access to the internet for participation in video conference visits and to complete study data. Exclusion criteria included a personal or family history of long QT syndrome, concurrent use of other QT interval prolonging drugs (e.g., ondansetron, citalopram), and heart failure with New York Heart Association Class II or worse symptoms (full inclusion/exclusion criteria described elsewhere)[17].

Recruitment and data collection

Nationwide social media advertising was employed for recruitment. Participants were screened via the web interface, secure video conference, telephone, or text message. Electronic informed consent was obtained in English or Spanish through a secure video conference. A randomization sequence was utilized to assign households 1:1:1 to the three arms of the trial, stratified by high- or low-risk cohort, as previously described[17]. A courier delivered study drugs, a self-monitoring kit including a six-lead ECG monitor (Fig. 2; KardiaMobile 6L, AliveCor®, Mountain View, CA), an oxygen saturation (SpO2) monitor (Vive Precision), and an oral thermometer (Adtemp IV), as well as mid-nasal swabs for SARS-CoV-2 RT-PCR testing[31] to each participant. On days 1–14 and on days 21 and 28, participants self-collected mid-nasal PCR swabs, obtained vital signs including temperature, respiratory rate, heart rate, and oxygen saturation, completed daily symptom surveys (modified Flu-PRO[32]), and uploaded an ECG (Fig. 1). Participants packaged and returned mid-nasal swabs using provided prepaid shipping labels and containers. Data were collected and curated using REDCap electronic data capture tools hosted at the University of Washington[33]. The overall workflow is described in Fig. 3.

Fig. 2

Electrocardiogram acquisition and workflow.

Six-lead electrocardiogram acquisition by the patient using the AliveCor® KardiaMobile 6L device (A). Note: rotating the device 180° results in limb lead reversal. Implementation of remote QTc monitoring (B). ECG is collected by the participant, uploaded to a secure web portal, and transmitted by study coordinators for adjudication at a core laboratory. QTc results are transmitted to study coordinators for safety ascertainment by study clinicians.

Fig. 3

Graphical overview of recruitment, screening, logistics, and data acquisition.

A multipronged recruitment effort was utilized, and screening leveraged secure digital health technologies.

ECG acquisition and analysis

Participants were provided with written and verbal instructions regarding the use of the KardiaMobile 6L device, including the collection of ECGs and submission of the data to the secure internet interface. To use the device, participants downloaded the Kardia application on their personal devices (smartphone or tablet). If a participant’s device was incompatible[34], a smartphone was provided by the study. For Spanish-speaking participants, the default language of the Kardia™ application was switched to Spanish. Each participant was given instructions by study staff on the day of kit delivery via secure video conference. During this visit, a study coordinator helped participants download the application and set up an account, and link the device to the study. The coordinator demonstrated this process via video conference, requested the participant replicate the steps, and verified the collection of an ECG from the participant in real time. This also served as the baseline ECG.

To collect ECG tracings, participants were instructed to open the Kardia™ application on their mobile device, select “record EKG,” and then hold the KardiaMobile 6L device with their thumbs on the anterior silver spaces and the back of the device resting on their bare left knee or ankle for at least 30 s (Fig. 2). The Kardia™ application automatically identified poor quality or noisy tracings at the end-user level, prompting participants to repeat acquisition. Participants were identified within the portal by their study ID, and each site was able to track and monitor their own participants in order to contact those with adverse events (AEs) necessitating study drug discontinuation.

Standard ECG readings from KardiaMobile provide a rhythm strip but not QT intervals. Therefore, we created a real-time system to adjudicate QT intervals at a central facility with results returned to sites. The study facing secure Kardia web interface was queried several times daily and ECG submissions were tracked. The QT interval reading was requested on the first ECG collected each day (Fig. 2B). This action resulted in the transmission of deidentified ECGs to a central core laboratory (Mayo Clinic Heart Rhythm and Physiologic Monitoring Laboratory, Mayo Clinic, Rochester, MN) where the QT interval was measured by certified rhythm analysis technicians. Fiducial points were tagged on the tracing using an automated algorithm (QRS onset, T wave offset, etc.) and the RR interval, QT interval, and corrected QT interval (QTc) (Bazett and Fredericia corrections) were calculated. The ECG technologist then manually adjudicated these points and adjusted the fiducial points and confirmed the measurements. The six-lead tracings were displayed in a “QT dashboard” user interface that was developed for the purpose of this study. Each participant’s heart rate QTc values were returned to the investigator within 1 h of uploading in order to be available at the point of care. If the QT interval was not able to be calculated, the result was read as “0” and another ECG was requested from the participant.

Qualitative data collection

For the purpose of this paper, the authors (J.M., A.S.) conducted interviews and corresponded extensively with study staff regarding qualitative experience with this paradigm. We specifically focused on practical challenges encountered at each step of the trial and the ad hoc solutions to overcome them. The results of this process are reported in the experience section of this manuscript and inform our discussion.

Outcomes

The primary outcomes (development of lower respiratory tract infection, hospitalization, or death attributable to COVID-19) and secondary outcomes (time to cessation of viral shedding and time to resolution of COVID-19 symptoms) of the study are presented separately[17]. The data described in this paper include demographics, protocol adherence, QT prolongation-related AE), and qualitative experience derived from interviews with investigators and study coordinators.

ECGs were monitored for development of QT prolongation, defined as a QTc value ≥500 ms or an increase of 60 ms or more above baseline, either of which constituted a grade 3 AE and necessitated study drug hold while obtaining a prompt follow-up ECG assessment. Participants were contacted to collect an additional ECG. If the follow-up ECG confirmed a prolonged QT interval, the study medication was permanently discontinued.

Table 1

Characteristics of per-participant ECG collection.

Statistic	Value
Total patients	219
Total ECGs	3245
Median ECGs per patient	17
Mean ECGs per patient	15.9
Standard deviation	3.9
Range of ECGs per patient	1–32

Table 2

Average ECG protocol adherence (proportion of expected study days with ECG) stratified by demographic characteristics.

ECG protocol adherence by demographic characteristics
	No. (%)	Mean adherence (±SD)
Age
≥60	19 (8.7)	92.1% (22.7)
<60	199 (91.3)	84.0% (15.3)
Sex assigned at birth
Female	121 (55.5)	86.3% (20.6)
Male	97 (44.5)	82.5% (24.2)
Race
American Indian or Alaskan Native	38 (17.4)	86.2% (20.7)
Asian	10 (4.6)	88.1% (20.5)
Native Hawaiian or Pacific Islander	3 (1.4)	70.8% (20.1)
Black	23 (10.6)	77.7% (27.6)
White	112 (51.4)	81.2% (18.2)
Other	29 (13.3)	73.3% (30.5)
Prefer not to say	3 (1.4)	98% (3.6)
Ethnicity
Not Hispanic/Latinx	153 (70.2)	87.6% (18.0)
Hispanic/Latinx	65 (29.8)	77.7% (29.1)
Preferred language
English	198 (90.8)	85.8% (20.9)
Spanish	20 (9.2)	73.1% (31.4)

±SD standard deviation.

Table 3

Patient-level adverse events related to QTc prolongation.

Randomized arm	ECG QTc ≥60 ms change from baselinea,b	ECG QTc >500 msa,b	Total QTc-related adverse eventsb	Total participants
Ascorbic acid + folic acid	2 (2.5)	0 (0)	2 (2.5)	80
Hydroxychloroquine + folic acid	19 (29.2)	2 (3.1)	19 (29.2)	65
Hydroxychloroquine + azithromycin	7 (9.5)	0 (0)	7 (9.5)	74
Total	28 (12.7)	2 (0.9)	28 (12.8)	219

aColumns are not mutually exclusive (all ECG QTc >500 ms were also ≥60 ms change from baseline).

bValues are the number of participants followed by the percent of total patients in the row within parentheses.