I recently came across a post on the Council for Foreign Relations’ Development Channel that asked: Are Randomized Controlled Trials a Good Way to Evaluate Development Projects?
This is an incredibly important question because, as the authors note, “International donors have spent well over $2 trillion in development assistance over the past five decades, but there remains significant uncertainty about what works and doesn’t work to reduce poverty and grow economies.” Aid accountability has become an increasingly salient prerogative of many donors, yet showing that aid gets where it is supposed to and achieves desired goals continues to be a challenge because we often propose complex interventions whose outcomes may not be linear (for instance, providing school books or desks should not be equated with learning, but it is much more easily quantifiable) and may be longer-term rather than short-term outcomes.
The randomized controlled trial is a mainstay of biomedical research because it allows us to compare the effects of an intervention – a drug, a therapy, a process of care – in one group (the intervention group) and compare it with the results of another population who receives either nothing (often in the form of a placebo), or receives the usual treatment, which we assume to be inferior to the new intervention. By controlling for a number of variables, we try our best to make sure that the two populations are comparable (equally matched numbers of men and women, in the same age range, with similar burdens of disease, etc.) and that they are representative of the population in which we would like to use this tested intervention in, assuming the results show promise.
Issues of methodological and interventional complexity aside, it’s important that we consider the social contract between researchers and research subjects that allows us to conduct this research in order to advance biomedical science. While the underlying principle of the RCT is to advance medical science, we can’t do so without respect for our patients and our research subjects. That is, we can’t perform experiments on people without respecting their basic human rights, including things like their right to autonomy and their right to not participate in research if they don’t want to.
Taking it one step further, we compare our new interventions against the gold standard in order to show an improvement. If no treatment exists, then providing a placebo may be considered acceptable as a control arm; however, when an effective treatment exists, denying patients that treatment in order to provide a placebo to compare your intervention against, would not. For example, if you want to demonstrate a new formulation of insulin is effective, you would have to compare your new intervention against existing types of insulin. You couldn’t deny patients something that they need or that we know is good for them just to show that your new formulation or treatment protocol works better than nothing.
The reason behind experimentation and the use of randomized controlled trials is that we are answering a question about the intervention, and we are unsure whether the intervention is better than existing treatments, or doing nothing. In short, there is a general uncertainty as to whether a treatment will be beneficial. This is known as clinical equipoise in research ethics.
In international development, we are often uncertain of the models for appropriately scaling up development initiatives, but it would be unfair to say that we are uncertain of the many of the interventions. For instance, we know that funding education or immunizations are important and successful interventions for children or that ensuring access to clean drinking water reduces water-borne illnesses. These are not questions of science, they are questions of political will that allow the absence of these interventions to persist.
What is often unclear is not whether interventions, per se, are effective in a global context, but rather what models best allow them to be scaled up and run sustainably. Testing models of implementation in a randomized controlled trial is notably different from testing the effectiveness of the intervention, itself. So, when we know that the intervention is effective, can we justifiably randomize a population to a model that would result in it being unlikely that they would receive it? I don’t believe that we could.
If we do determine that randomized controlled trials are the way that we should proceed in international development, then I argue that the same safeguards need to be in place for conducting these trials with new medical innovations: they need to be approved by a competent research ethics board, there needs to be an oversight committee capable of determining throughout the study whether the intervention or the control is disproportionately placing participants at risk (and if it is, they need to have the authority to stop the trial/study/intervention), and there needs to be an ethical obligation to publish negative findings. If we are going to apply rigorous scientific methods to international development, we need to apply all of these methods and not only the ones that suit us and our information needs best.
And of course, it’s worth noting that not everything that we know about effective interventions has been proved through a randomized controlled trial. I suspect that the authors of this cheeky paper in the British Medical Journal are still waiting for participants for their randomized controlled trial on parachute use to prevent death and major trauma related to gravitational challenge…
Gordon C S Smith, & Jill P Pell (2003). Parachute use to prevent death and major trauma related to gravitational challenge: systematic review of randomised controlled trials BMJ, 327 DOI: 10.1136/bmj.327.7429.1459