Highly Efficient Clinical Trials (HECT)

Randomized clinical trials (RCTs) are the gold standard for making sure health interventions are both safe and effective. There’s just one problem. RCTs tend to be time-consuming and expensive—and sometimes the results are too context-specific to provide satisfactory answers to the most important questions.

However, advances in quantitative sciences are opening up new possibilities for highly efficient clinical trials (HECT) that generate the maximum possible knowledge with the minimum possible investment of resources, including money, people, and time.

Computational simulations are key to planning efficient trials. They help researchers understand the ramifications (in terms of cost, relevance, time, and validity) of different choices they make as they design an RCT. These choices can include the dose or package of interventions to test, key outcomes to look for, specific populations to focus on, and important technical details (for example, two-arm vs. multiple arm, cluster vs. individual, multi-arm vs. factorial).

In global health, researchers face an additional problem: there’s often a lot of uncertainty surrounding the disease they’re investigating and the setting in which they’re investigating it, which means it’s especially difficult to optimize RCT designs.

Sometimes, during the course of a trial, information comes to light that suggests that the trial could be less expensive, shorter, or otherwise different and better than originally planned. However, since researchers aren’t allowed to change the design of most RCTs in the middle of the experiment, they’re stuck spending more and waiting longer for less helpful results.

Adaptive clinical trials, a relatively new concept borrowed from the pharmaceutical industry, let researchers modify trial design, based on interim analyses of the data, while the trial is still in progress. In the vast majority of cases, changes to trial design are based on pre-determined statistical rules, to make sure investigator bias or other arbitrary factors can’t influence the RCT inappropriately.

What follows is a list of examples of how adaptive and related methods can improve RCTs.

Sample size re-estimation:

Investigators often realize, in hindsight, that they either under- or overestimated the sample sizes required for their RCT. Using adaptive trial design, they have the opportunity to correct these errors midway through the trial. In under-enrolled trials, they can recruit more patients to get useful results. In over-enrolled trials, they can release some patients from the study, saving money and exposing fewer people to risk.

Seamless designs

Typically, trials are done in phases, with a delay between phases. Seamless trials save time by allowing trials to move immediately from one phase to the next. For example, a seamless phase II/III trial of 9-valent HPV vaccine tested three dosages (low, medium, and high) in phase II to select the dosage to be tested; it then moved immediately into a phase III trial against the standard-of-care 4-valent HPV vaccine.

Response adaptive randomization

At the beginning of a trial, there should be genuine uncertainty about the benefits of the intervention being tested. If researchers were sure a treatment worked, it would be unethical to randomize patients into a placebo group. Over the course of the trial, however, this uncertainty can decrease. As it grows more statistically certain that a treatment works, investigators can assign more patients to the group that gets the treatment, while still maintaining the trial’s statistical integrity.

Adapting to external changes

Occasionally, the external context in which an RCT is being conducted changes in a way that forces a change in trial design. For example, a recently published RCT in Kenya was investigating the effectiveness of using SMS messages to help HIV patients adhere to treatment. However, during the middle of the trial, Kenya’s national HIV program changed to a test-and-treat-all policy. Adaptive trial design allowed researchers to modify the study so that it fit the new context and still answered the questions they were asking.

Platform clinical trials

In these multi-arm RCTs, individual treatments can be dropped and new treatments added during the trial. In the conventional paradigm, which only accommodates two or three arms at a time, you would have to run several separate head-to-head trials to distinguish between, say, five relevant interventions. A platform trial can test and compare all interventions, both to each other and to a placebo or the standard of care. Platform trials have been proven to reach definitive conclusions much faster while limiting the number of patients at risk.

Perpetual clinical trials:

Platform trials can also evolve into perpetually running trials that address multiple questions over time. Perpetual trials save money and time that would otherwise be spent on getting multiple individual trials up and running, they build expertise among staff, and they strengthen trust with the patients’ communities.

Highly Efficient Clinical Trials have enormous potential to deliver more knowledge—faster—to save lives. However, they are new, and the global health community has to learn how to use them—and how to trust them to deliver results that are just as (or even more) valid than traditional RCTs. A first step is building interdisciplinary teams of clinical, methodological, and statistical experts to change the paradigm together.