Ebola and evidence: how will best intentions deliver the best outcomes?

Author: Professor Mike Clarke

After a year or more of the Ebola crisis in West Africa, how can we be confident that the care we provide for patients and practitioners in the next disease outbreak is effective?  Whether we’re interested in vaccines, re-hydration strategies, protective clothing, coping with bereavement, or any of the numerous dilemmas that will arise; reliable and robust evidence is vital to ensure our best intentions lead to the best outcomes.

This blog isn’t the top ten interventions to use in Ebola response, or the five emergency response interventions that do more harm than good. Rather, this is the top three ways to get the evidence that decision makers need:

(1) Systematically review what is already known.

(2) Randomise interventions when there is uncertainty in the choice between options.

(3) Update reviews of evidence to incorporate the new findings.

A key piece in this evidence has to be an estimate of the effects of the interventions, actions or strategies on the outcomes that matter to people making decisions and choices about their own care or that of others. The recently reported randomised trial of a new Ebola vaccine demonstrates what can be done [1].  We need to know the likely size of any difference in outcomes between the options available. We need to know what works and how much it works.

To achieve this, we have to overcome problems that can be caused by chance and bias. Conceptually, chance is easy to deal with. Just study more people. This might be done through studies that are large enough in their own right and by bringing similar studies together in a systematic review. Not only will this allow us to compare and contrast the different studies, it might also allow us to combine their findings, boosting the power of the analysis and the precision of the estimate. Bringing all the studies together will also reduce bias that might come from placing too much emphasis on any single study, especially if that emphasis is driven by the results of the study.

We also need to minimise bias within those studies. That’s where randomisation comes in. When more than one option is available and there is uncertainty about which is better, a random process like flipping a coin is used to decide which of these acceptable options is allocated to which people.  Differences in their outcomes will be due to the option they were allocated or the imbalances in their characteristics that are due to chance. Again, we cope with chance by including more people to minimise the possibility that the difference in outcomes is due to factors other than our planned interventions, actions or strategies.

Faced with a crisis like Ebola, perhaps it feels unethical or inappropriate to determine treatment by flipping a coin. Perhaps our immediate reaction is that we should just act, and do what we think is best. But, in the absence of good evidence that what we do really is best, how ethical is it to act in that way? If there are options, and there is uncertainty between them, flipping a coin might be the fairest way to choose. And, flipping a coin in the context of a randomised trial will gather the data that can resolve that uncertainty for the future.

So, what can we do in preparation from another outbreak of Ebola? We need to ensure that people have access to the evidence that has already been generated and ideally in the form of systematic reviews. Once we have determined the key areas of uncertainty, we need to be ready with the randomised trials that could resolve key areas of uncertainty. Our aim is to ensure that best intentions deliver the best outcomes.

For evidence-based resources to help with Ebola response, please visit Evidence Aid.

Evidence Aid is working with partners to compile systematic reviews, important articles and other resources for decision-makers that are relevant to the Ebola crisis.  Please contact Claire Allen if you have further evidence you can contribute.

1. Henao-Restrepo AM, et al. Efficacy and effectiveness of an rVSV-vectored vaccine expressing Ebola surface glycoprotein: interim results from the Guinea ring vaccination cluster-randomised trial. Lancet 2015 [Epub ahead of print 3 August 2015] http://www.thelancet.com/journals/lancet/article/PIIS0140-6736(15)61117-5/abstract

This blog was written by Professor Mike Clarke, from the Northern Ireland Network for Trials Methodology Research at Queen’s University Belfast. He is one of the founders of Evidence Aid, an international initiative to provide decision makers in disasters and other humanitarian emergencies with the knowledge they need to make well-informed decisions and choices.

Views expressed through our blog are those of the author(s) alone and do not necessarily reflect the official position of Evidence Aid.

Read the Evidence Aid collection of resources for Ebola