The Oxford Vaccine – How generalisable are the trial results?

Preliminary Oxford-AstraZeneca COVID-19 vaccine trial results were recently published in The Lancet. With growing public confusion pertaining to their precise meaning, this publication could not be timelier. Prior to the Lancet Publication, outlets such as New Scientist, the BBC and The New York Times had scrutinized the results of the Oxford Vaccine trial. The Oxford-AstraZeneca vaccine’s efficacy, dosages, and trial design were called into question as previous results published on other vaccines, such as those produced by Pfizer and Moderna, demonstrated greater efficacy. However, after closer reading, I’m not convinced that the Lancet Publication has put society’s apprehensions at bay.

The initial trial design included four studies, located in three different countries. Two phase 1/2 trials took place in the UK and South Africa, and two phase 2/3 trials took place in the UK and Brazil. Phase 1 trials are the first step in a clinical trial for a new drug seeking FDA approval, and typically the number of participants is small. If the drug is safe and effective in this small group of healthy participants, then it progresses on to phase 2, and eventually phase 3, where as many as 3000 participants are typically included. If the vaccine is deemed safe and effective in this population, then the vaccine is put forward to bodies such as the FDA for approval for use in the general population. In order to determine the safety of the Oxford vaccine, the results from all four studies were used. However, the efficacy of the vaccine, which has made controversial global headlines, only comes as a result of the phase 2/3 of the UK and Brazil studies. In addition, a few other issues pertaining to the study’s structure, demographics and outcomes appear to be evident. 

Double blinding inconsistency

Blinding, an important part in trial design which aims to reduce bias and improve study validity, was inconsistent between the studies. Double blinding in a clinical trial is the most reliable; neither the patient nor the researcher is to know which group the patients are in. This is done to minimise biases introduced into the study by the researcher, as well as any placebo effects. Double blinding was only used in the phase 1/2 studies done in South Africa. All other studies were single blinded; the patients didn’t know whether they were assigned to the experimental (COVID-19 vaccine) or control group. The authors provided no rationale for this discrepancy. 

Lack of Representative Population Demographics

The efficacy results of the vaccine are pooled from the UK and Brazil phase 2/3 studies. This resulted in rather homogeneous study population demographics; there was a lack of ethnic diversity, as more than 90% and more than 65% of trial participants were white, in the UK and Brazil studies, respectively.  Although the authors, in conjunction with advice from international regulatory bodies, claim that the pooling of studies from different countries would make their results more generalisable, we should look at the validity of this statement on a global scale. In addition to lack of ethnic diversity, 60% of participants were female, and generally those included in the published data were younger (<55 years old). Based on what is known about the virus, some of the most at-risk groups for severe illness and symptoms as a result of COVID-19 are ethnic minorities, older people, and men. While the authors have acknowledged these limitations in their publication, it’s clear that as the trials continue, it will be important to understand how the vaccine will impact these groups, and publish these results.  

Pooling of Results

In order to determine the efficacy of the Oxford-AstraZeneca vaccine, results were pooled from two study arms: those which received two standard doses, and those who received a low and standard dose. It was found that the low and standard dose yielded an efficacy of 90%, whereas the double standard dose yielded an efficacy of 62%, creating a combined overall efficacy of 70%. There has been much scientific debate and scrutiny online regarding the scientific rationale of doing so, and what the real-world implications are of combining the results of such different treatments to obtain a cumulative metric.  

Vaccine Effectiveness in Preventing Symptomatic COVID-19 – What about asymptomatic?

In the two combined studies (Brazil and UK), only the UK arm had a testing plan for asymptomatic infections. In the UK study, participants self-administered a weekly swab test for COVID-19, whereas no such test was administered in the Brazil arm. Estimates have suggested that up to 80% of COVID-19 cases are mild or asymptomatic, and therefore the results of this trial are puzzling: how effective is this vaccine really, if uniform testing for the virus in asymptomatic individuals was not considered in the analysis? 

Oxford-AstraZeneca has devised a vaccine in an unprecedented amount of time. Their results are promising, and surpass both the FDA and WHO minimum efficacy requirement of 50% to be licensed. While the Oxford vaccine has not yet been approved in the UK, it is expected to be so imminently as COVID-19 cases continue to rocket across the country. In cooperation with the Russian Sputnik vaccine, there will also soon be another trial whereby the vaccines are combined in effort to improve the efficacy of the Oxford vaccine.

Now that thousands of people globally have received the Pfizer and Moderna vaccines and approval of the Oxford vaccine is on the horizon, there is hope that we will see better control of the virus in the new year. However, with any and all science, it is important that we be critical of the claims being made, and question the scientific method used to arrive at such conclusions. With further work ensuring the generalisability of these results, especially towards the most vulnerable to the virus, the Oxford vaccine will surely join the ranks of the existing vaccines, and contribute to global efforts in reducing COVID-19. 

Image credit: wir_sind_klein / Pixabay