Quality by Design in the RECOVERY Trial (Randomized Evaluation of COVID-19 Therapy)

New Study Will Test mRNA COVID-19 Vaccine in Eight sub-Saharan Countries

Clinical Researcher—December 2020 (Volume 34, Issue 10)


Brigid Mary Flanagan, BA, RN, CCRC, MSB


If you are not following the progress of the RECOVERY trial, you should be. It is a great example of quality by design (QbD) in action. The trial is a randomized controlled study to evaluate potential treatments for COVID-19. The lead investigator is Peter Horby, Professor of Emerging Infectious Diseases and Global Health at Oxford University in the United Kingdom.

The first draft of the protocol was available on March 10, 2020. It was submitted to the Medicines and Healthcare products Regulatory Agency (MHRA) and the appropriate ethics committee on March 13, received regulatory approval on March 16, and received ethics approval on March 18. The first patient was enrolled on March 19, and by April 1, 1,000 patients had been enrolled. As of this writing, 15,303 participants have been randomized at 176 sites.

The trial is designed to have the least possible impact on hospital personnel. It has already demonstrated that there is no clinical benefit from the use of hydroxychloroquine in hospitalized patients with COVID-19, and that low-cost dexamethasone reduces death by up to one third in hospitalized patients with severe respiratory complications of COVID-19.

This trial has demonstrated that when you build quality into the design of your trial, eliminate complexity, and have the buy-in of a broad spectrum of stakeholders, you can achieve success. This trial has changed clinical practice, including for pregnant women.

The Protocol

The design, conduct, and analysis of the trial is focused on issues that might have a material impact on the well-being and safety of study participants (hospitalized patients with suspected or confirmed SARS-CoV-2 infection) and reliability of results that would inform the care for future patients.{1} An additional factor to consider, in the context of this trial, is the well-being of staff, since SARs-CoV-2 is a transmissible disease.

A co-director of the study is Martin Landray, PhD, FRCP, Professor of Medicine and Epidemiology at the Nuffield Department of Population at the University of Oxford. In 2012, he argued for QbD to be factored into clinical trials whereby those responsible for the overall conduct of a trial would identify the critical aspects that, if not performed correctly, would threaten the protection of patients or the integrity of results.{2}

When developing the protocol, in early 2020, there were no approved treatments for COVID-19, the disease caused by the novel coronavirus which emerged in China in late 2019. The UK New and Emerging Respiratory Virus Threats Advisory Group advised that several possible treatments should be evaluated, including low-dose steroids, hydroxychloroquine, and lopinavir-ritonavir.

The inclusion criteria were unambiguous: Aged ≥ 18, admitted to hospital with proven or suspected SARs-CoV-2 infection (“suspected” infection was an early modification to the protocol). Physicians recognized the clinical syndrome, but there were delays in getting test results. They also recognized that not all tests are positive initially and that, if you are going to treat these patients, it makes sense to start treatment early, not two days later.

Eligible patients are consented by the admitting physicians. The patient is then randomized to one of several treatment arms, each of which will be given with the usual standard of care. A prescription for the assigned treatment is submitted to the hospital pharmacy, which manages supplies centrally. The main outcomes are death, discharge, the need for ventilation, and the need for renal therapy at 28 days post randomization.

It is an adaptive design trial, with the results being monitored on an ongoing basis by an independent data monitoring committee (DMC) to assess whether the randomized comparisons in the study have produced evidence that is strong enough to affect treatment strategies. Trial arms that demonstrate lack of efficacy are discontinued and new arms are added as other evidence emerges of potentially beneficial treatments. Of interest is that pregnant women are not excluded and there is no upper age limit for trial participants. The oldest participant to date has been 109 years of age.{3}

As of the end of September 2020, Version 9 of the protocol was open to enrollment. From Version 6 onwards, a factorial design (used to understand the effect of two or more independent variables) has been used so eligible and consenting participants can be randomized to one of the treatment arms in Randomization A and, simultaneously, to one of the treatment arms in Randomization B.{4} For Randomization A that can be:

i.     No additional treatment

ii.     Corticosteroids (pediatrics only)

iii.     Azithromycin

iv.     Intravenous immunoglobulin (pediatrics only)

For Randomization B, eligible patients will be randomly allocated between the following treatment arms (provided they have consented and there are no contraindications):

i.     No additional treatment

ii.     Convalescent plasma

iii.     Synthetic neutralizing antibodies (provided by Regeneron)

Patients with progressive COVID-19 as evidenced by hypoxia (oxygen saturation < 92% on room air) and an inflammatory state (C-reactive protein (CRP)≥ 75mg/L) can be randomized a second time to either:

i.          No additional treatment

ii.          Tocilizumab

The larger the numbers randomized, the more accurate the results will be. Tellingly, enrollment slowed down in the summer of 2020 as the numbers of patients being hospitalized decreased. Enrollment has picked up in September and October, as the number of infections began to resurge across the UK.

Trial Processes

In line with QbD principles, all trial processes have been greatly simplified to minimize the burden on frontline staff in busy hospital settings who have been stretched to the limit during the pandemic. The University of Oxford is the trial sponsor, with trial coordination coming from a Central Coordinating Office (CCO) staffed by members of two registered clinical trial units. The CCO oversees regional coordinating centers which, in turn, assist with the selection of local centers. The study is conducted at multiple hospitals within the local regions.{4}

The consent for participation is less than five pages long and is available in multiple languages reflecting the diversity of the UK population (e.g., Polish, Urdu, Bengali, etc.). Training is all available online and must be completed before a site is activated. Training requirements are dependent on one’s role in the study; for example, all are required to complete background training and those obtaining consent and/or performing randomization have additional training modules for those duties.

A one-page case report form (CRF) is completed online prior to randomizing a subject. A second CRF is completed at death, discharge, or at 28 days, whichever comes soonest. Information collected includes:

·       Vital status

·       Hospitalization status

·       SARS-CoV-2 result

·       Use of ventilation (number of days and type)

·       Use of renal dialysis or hemofiltration

·       Documented new major cardiac arrythmias

·       Use of any medications in the RECOVERY trial or other purported COVID-19 treatments (e.g., remdesivir)

Additional information is collected in the first 72 hours for the first 200 subjects randomized in Main Randomization B (no additional treatment vs. convalescent plasma and no additional treatment vs. synthetic neutralizing antibody):

·       Sudden worsening of respiratory status

·       Severe allergic reaction or other infusion reaction

·       Temperature >39C or ≥2C above baseline

·       Sudden hypotension

·       Clinical hemolysis

·       Thrombotic event

In addition, Serious Hazard of Transfusion (SHOT) reporting is conducted for all patients receiving convalescent plasma for the full duration of the study. All the information collected for the study is in the medical record and would routinely be documented absent trial participation.

For data and safety monitoring, the focus is on events that, based on a single case study, are highly likely to be related to the study medication (e.g., anaphylaxis, Stevens-Johnson syndrome, or bone marrow failure), where there is no other plausible explanation. Events that are the consequence of COVID-19 and common events which are the consequence of conditions which existed prior to randomization are exempt from reporting.

Monitoring is done remotely by the CCO. Onsite monitoring will only be considered if a training need is identified or the results of central statistical monitoring suggest there might an issue. No routine source data verification is taking place. Given that there is pandemic, site visits would not be appropriate, as they could increase the risk of spreading infection.{4}

The protocol, the informed consent, sample CRF pages, training materials, and patient information sheets are all available for public viewing on the RECOVERY website at www.recoverytrial.net.

The Stakeholders

The effort to date reflects the strong collaboration among the various stakeholders. These include the National Health System (NHS), the MHRA, the ethics committee, the Health Research Authority, and the Chief Medical Officers of England, Wales, Scotland, and Northern Ireland, who wrote to all physicians across the UK requesting their support with enrollment and encouraging a default position where every eligible patient is offered enrollment in a clinical trial.

Strong evidence requires large numbers (e.g., 2,000 subjects per arm).{5} The lead investigator had hoped to enroll 1,000 subjects a week, but that has not always been possible. However, the strong national (and international) coverage, as well as the recruitment successes to date, have resulted in requests for inclusion from other national health authorities; discussions are ongoing with Vietnam, Indonesia, and Nepal.

The aforementioned website has a section dedicated to patients. There is a video in which the co-investigators for the study describe each of the drugs being investigated as possible treatments for COVID-19 and a list of frequently asked questions.

Results to Date

The most consequential results to date have been that low-cost dexamethasone reduces death by up to one third in hospitalized patients with severe respiratory complications of COVID-19 and one fifth in other patients receiving oxygen alone. A press release issued in June 2020 indicated that 2,104 patients randomized to receive dexamethasone 6mg once a day, either by mouth or intravenously, were compared to 4,321 randomized to usual care alone. In the press release, co-investigator Landray said: “Since the appearance of COVID-19 six months ago, the search has been on for treatments that can improve survival, particularly in the sickest patients. These preliminary results from the RECOVERY trial are very clear—dexamethasone reduces the risk of death among patients with severe respiratory complications. COVID-19 is a global disease—it is fantastic that the first treatment demonstrated to reduce mortality is one that is instantly available and affordable worldwide.”{6} The trial results were subsequently published in New England Journal of Medicine (NEJM) on July 20, 2020.

The independent DMC concluded that there was no beneficial effect of hydroxychloroquine in patients hospitalized with COVID-19. A separate June 2020 press release {7} indicated 1,542 patients randomized to hydroxychloroquine were compared to 3,132 randomized to usual care alone. There was no significant difference in the primary endpoint of 28-day mortality and no evidence of beneficial effects on duration of hospital stay or other outcomes. Those results were subsequently published in the NEJM on October 8, 2020. The conclusion was the same for the lopinavir-ritonavir arm, and that arm was discontinued. Those results were published in The Lancet on October 5, 2020.


The RECOVERY trial is a great example of what can be accomplished when you have buy-in from all stakeholders. There is no negotiation on the contract, no payments to investigators, and no recruitment targets.{8} Indemnity is addressed in the protocol (the university has a specialist insurance policy in place which will operate in the event of any participant suffering harm because of their study participation).

The protocol objectives are clear; the primary objective is to estimate the effects of study treatment on all-cause mortality within 28 days of randomization. The secondary objectives are to investigate the effect of study treatments on the duration of the hospital stay, the need (and duration) of ventilation, and the need for renal replacement therapy. There are no tertiary or exploratory endpoints. Trial procedures are greatly streamlined, and randomization is done online. The confirmation of the allocated treatment can be downloaded and printed. Data entry is minimal. Follow-up information is collected at one timepoint only (Day 28) and can be done by phone, in person, or electronically.

Usually in studies, participants are assigned a study number and the trial sponsor does not know their identities. In this case, the participant’s NHS number is collected along with some other personal details. The informed consent explicitly states that the researchers may request additional medical information that is maintained in local or national records for up to 10 years following the scheduled follow-up period. As the long-term sequelae of COVID-19 are unknown now (10 months into the worldwide pandemic), and eligible participants may be receiving investigational agents, this reservoir of data could potentially be a great source of information. This is one of the benefits of a national healthcare system.

Other points of interest are the very broad inclusion/exclusion criteria. It is refreshing to see that participants are not excluded because their body weight exceeds 30kg/m², they are pregnant, or they are more than 65 years of age. With respect to co-morbidities, if, in the opinion of the investigator, study participation would put the patient at significant risk, they should be excluded. As the trial is for hospitalized patients and there are no known cures, participation in the RECOVERY trial may be a patient’s best option for making a good recovery.


The RECOVERY trial is a great example of QbD in action. For people who struggle with the concept of QbD, this is an opportunity to learn from one of the original proponents for QbD in clinical trials, Professor Martin Landray. All the materials are available for access by the public, including the protocol, informed consent, patient information materials, sample CRF pages, and the statistical analysis plan. This transparency is to be commended.

The protocol is easily understood (33 pages in total). The informed consent is short and to the point. Trial processes have been simplified. Data collection is minimal. The numbers enrolled will depend on the duration of the pandemic. A DMC is reviewing the data on an ongoing basis to determine the effect of treatments on mortality that is strong enough to affect clinical practice. Most importantly, the RECOVERY trial has broad support from all a variety of stakeholders, including the MHRA and NHS clinicians, in the UK.

To date, dexamethasone (first made in 1957, off-patent and widely available around the world) is the only treatment identified to have benefits for patients receiving respiratory support. Those results have changed the management of hospitalized patients worldwide.

As research professionals, we can all learn something from this trial. The pandemic has been a catalyst for change in how trials are conducted. Let us hope we can keep that momentum going forward.


1.     https://www.recoverytrial.net/files/recovery-protocol-v9-1-2020-09-18.pdf

2.     Landray MJ, et al. 2012. Clinical Trials: Rethinking How We Ensure Quality. Drug Information Journal 46(6):657–60.

3.     http://www.francesoir.fr/politique-monde/interview-exclusive-martin-landray-recovery-hydroxychloroquine-game-over-uk

4.     https://www.recoverytrial.net/files/recovery-protocol-v9-1-2020-09-18.pdf

5.     Letter dated 06 May 2020 from NHS Chief Medical Officers to their Colleagues.


6.     https://www.recoverytrial.net/news/low-cost-dexamethasone-reduces-death-by-up-to-one-third-in-hospitalised-patients-with-severe-respiratory-complications-of-covid-19

7.     https://www.recoverytrial.net/news/statement-from-the-chief-investigators-of-the-randomised-evaluation-of-covid-19-therapy-recovery-trial-on-hydroxychloroquine-5-june-2020-no-clinical-benefit-from-use-of-hydroxychloroquine-in-hospitalised-patients-with-covid-19

8.     https://www.recoverytrial.net/for-site-staff/site-set-up-1/recovery-trial-faqs-for-study-sites

Brigid Mary Flanagan, BA, RN, CCRC, MSB, (bflanagan@orielresearchservices.com) is Managing Director of Oriel Research Services in Drogheda, Co. Louth, Ireland.