Nicholas G. Evans, PhD
We’re all waiting for and working toward the end of the novel coronavirus pandemic. There’s a lot of research on masks and tests for COVID-19, but the things we’re hoping for most are effective rescue drugs for the disease, and ultimately, a vaccine. While some countries have successfully suppressed the virus, and others are very close, it’s unlikely at this stage that COVID-19 will be controlled in the USA without a vaccine. To that end, there’s been a lot of talk of fast-tracking trials and new methods to test vaccines that cut down the time to approval. A question on people’s minds is whether, and when, researchers should stop trials early and move to approving a vaccine for broad distribution.
When to stop a trial
There are two main reasons researchers stop trials. The first, is that unexpected adverse events arise during a clinical trial. Adverse events are commonly recorded and reported with trial results, but serious and unexpected adverse events must be reported to the study sponsor, the study Data Safety Monitoring Board (DSMB) and/or Institutional Review Board (IRB), and usually the government agency responsible for approving new medicines.1
The more controversial reason a trial might be stopped is if the expected benefits of the drug, biologic, or vaccine being tested appear to outweigh the risks. These cases are less common but do happen. A high-profile trial of the drug remdesivir was stopped early at the recommendation of the DSMB because its expected benefits in treating COVID-19 were seen to be greater than the risks of its use.2 While study enrollment was incomplete and the trial had not demonstrated any significant improvement of the initial primary outcome—clinical improvement, the decision to stop the trial was made because the new primary outcome—median observation time in hospital—was seen to be significant.
Help or harm, the decision to discontinue a trial is one grounded in the ethical justification for clinical trials. One of the main reasons we conduct trials is if an intervention is one that is subject to equipoise, where we don’t have good reason to believe the intervention is going to be better or worse than a placebo (or the current “gold standard” of care, depending on the trial). If we know—or come to know—that our intervention benefits or harms patients, then the reason for the study evaporates. That’s because it is considered wrong to withhold a beneficial drug or vaccine from patients who would otherwise have access, or conversely, to intentionally expose someone to an intervention that either doesn’t work or harms them.
Stopping too soon
That said, knowing what works and doesn’t during a trial is tricky. In early stages, adverse events might signal that an intervention causes serious harm to patients, or they might be freak occurrences. There is, after all, a small chance that a one-in-one-hundred-thousand adverse event arises in one of the first hundred participants in a trial. Especially for trials in which there is a strong social benefit, it’s important to exercise caution before giving up too quickly.
If investigators are confident that an intervention works, however, it’s understandable they would want to immediately give patients that effective intervention over a placebo. But ending an appropriately powered study early risks giving patients a therapy or vaccine that only appears to work. Given that we don’t routinely track outcomes for patients who receive an approved therapy, stopping a trial can result in thousands or even millions of patients taking an ineffective or harmful therapy without oversight.
This is not a hypothetical. In 2001, a randomized trial of an intensive insulin regimen showed promising results. Investigators had designed the trial to facilitate early stopping, and when the results appeared overwhelmingly positive, they discontinued the trial on the basis of substantial benefit.3 A 2008 systematic review, however, showed that the regimen did not benefit patients, and, in fact, established an increased risk of hypoglycemia.4 Researchers have found that trials stopped at fewer than 200 total participants commonly overestimate treatment effects. Even trials stopped at more than 500 total participants show small overestimates of effects, creating risks for early stoppages even in large trials.
Fast-tracking a COVID-19 vaccine
In the context of COVID-19, an early stop to a trial could have serious implications. An ineffective drug mistakenly used because of a stopped study could harm (or, in the very best case, fail to benefit) thousands of patients. If it is a vaccine subject to stoppage, the failure to benefit could apply to hundreds of millions of people. That’s a huge risk to take—one that could seriously undermine trust in the entire project to vaccinate enough people to stop the pandemic.
It’s a given that researchers conducting clinical trials are going to want to get their interventions to market as soon as possible. The public is also highly concerned about this pandemic and its effects on themselves, their loved ones, and their communities. When planning out proposed stoppages, however, we can’t let ourselves pretend that our exceptional circumstances call for less rigor in testing. Researchers, as well as institutional review boards, should establish benchmarks for when to stop trials for benefit that are as rigorous as those for adverse events. Doing so will ensure that when we stop trials, we stop them for the right reasons.
1. Food and Drug Administration. "Guidance for Industry and Investigators-Safety Reporting Requirements for INDs and BA/ BE Studies, December 2012." (2016).
2. Beigel, John H., et al. "Remdesivir for the treatment of Covid-19— preliminary report." New England Journal of Medicine (2020). 3. Van Den Berghe, Greet, et al. "Intensive insulin therapy in critically ill patients." New England Journal of Medicine 345.19 (2001): 1359-1367.
4. Wiener, Renda Soylemez, Daniel C. Wiener, and Robin J. Larson. "Benefits and risks of tight glucose control in critically ill adults: A meta-analysis." JAMA 300.8 (2008): 933-944.