Explainer: A physician-scientist on randomized control trials and improving the standard of care

We’ve all seen the recruitment posters on the subway or pamphlets in health care offices for medical trials. Maybe you’ve been asked directly by your doctor to be a participant in something called a randomized control trial. What is that, exactly? How can it improve future treatments?

The present-day standard of care for patients is set by national guidelines, which have been determined by committees of experts whose recommendations are based on research studies published in peer-reviewed journals. The research involves different types of studies. Some of the studies are observational, meaning the researcher simply watches and then collects the medical information. When it comes to a new medication or therapy, to be convincing, the research needs to be randomized and controlled.

Kellie Murphy, a professor and chair of the department of obstetrics and gynaecology, has sat on expert committees. She is also a researcher whose studies suggested treatments that have subsequently been adopted as a new international standard of care. Currently, she is recruiting participants for a randomized control trial (RCT), “Single Dose of Antenatal Corticosteroids (SNACS).”

Writer Matthew Tierney sat down with her to learn how an RCT works, why it’s considered the benchmark, and the challenge facing both researchers and trial participants in the age of AI and misinformation.

Are medical treatments always based on rigorous study?

New interventions can start after being subjected to randomized trials, but historically they haven’t always followed that path. Sometimes the interventions are implemented first, become commonplace, and people start to assume that they’re beneficial. It's not until trials are done that we learn they weren't helpful.

There are many examples throughout medical history. There is a notorious study brought up in clinical epidemiology classes on the treatment of gastric ulcers. Many years ago, it was thought that it was beneficial to freeze gastric ulcers, that it would cure them. Clinicians used an endoscopic device to freeze the ulcer, and companies started making devices strictly for this purpose. It wasn't until the randomized control trial came out that we learned not only did the ulcers not improve, they were associated with worse outcomes. That stopped the procedure.

In my field, another example is hormone replacement therapy or HRT. It was considered tremendously beneficial for a women’s bones and was prescribed to postmenopausal women even if they didn't have symptoms. In 2002, the women's health study was published. It trialled HRT versus a placebo and learned that along with the benefits were possible adverse effects. Physicians today still administer HRT but with much more consideration.

A randomized controlled trial (RCT) is considered the benchmark. How does it work?

A randomized control trial is essentially simulating research done in a laboratory setting in the wider population. You gather the sample size — the number of participants — you need and then randomly assign participants into two equally numbered groups. You administer to one half what the study intervention happens to be, and to the other half, called your control group, you give the consensus standard treatment.

Now, we know people are different: different heights, weights, ages, etcetera. But if you have a large enough group, all those differences will even out across the two randomly assigned subsets. For example, if you were to calculate the average age or average body mass in both groups, they’ll work out to be roughly the same.

In any research study, these types of variables are called ‘confounders.’ Age and weight are examples of known confounders because if we so desired we could identify them, we could statistically control for them. But there are also unknown variables, or unknown confounders, that we can’t identify and can’t control.

In theory, randomization nullifies the effects of all the known and the unknown confounders. It allows you to attribute differences in outcome between the two groups directly to the intervention you are testing.

Can you give us an example of the impact an RCT can have on future treatment?

The incredible thing about RCTs is that they can have a dramatic result and really change practice.

For example, in 2008 I published a paper that ultimately changed how physicians around the globe use steroids as treatment for preterm birth, meaning birth before 37 weeks. Steroids improve the mortality rate of babies born preterm. Nowadays, for those pregnant women at risk of preterm birth, we administer two doses, 24 hours apart, and then no more, regardless of subsequent symptoms.

The frequency of steroid treatment wasn’t always this way. This change came about because of our RCT.

To give some context: in obstetrics, it can be tricky to predict who is going to give birth early and who's not. Say somebody comes in at 25 weeks, they're bleeding, and we're worried about the placenta. They get a course of steroids. Everything settles and they do okay, but two or three weeks later, they come back with another bleed.

Is another course of steroids beneficial? In the 90s, the benefits of getting those two injections were so dramatic, people thought that we should start repeating doses for those who hadn’t yet delivered. And that is what happened.

But then several RCTs around the globe started up, including the large one I led in Canada, which had participants in 20 countries and 80 hospitals around the world. We wanted to determine whether subsequent steroid courses were truly beneficial. At the start, we were hoping they were, because NICU care is so expensive for preterm babies and hard to access. If a continued steroid course could mitigate that, wouldn't that be great?

In the end, my study found one course was just as good as four courses. There was no difference in severe outcomes between the two groups. In addition, we found that the babies who got repeated courses of steroids weighed less and, most importantly, had smaller head circumferences at birth. This put them at higher risk for neurodevelopmental disabilities in the future.

I’m currently running another RCT that looks at the steroid dosage, comparing a course of standard dosage to a half-dose. There's some data now showing a potential for adverse neurodevelopmental outcomes following steroid exposures. The hope is that after birth, the babies will experience similar benefits achieved by half the steroid exposure, which might be better for them long term.

A randomized control trial relies on consenting participants. What would you say to someone considering joining a trial?

Whenever you do a trial enrollment, recruitment can be challenging for the researcher. Even when they understand how important an RCT is, participants are sometimes not comfortable with the chance aspect of it.

I would say to them that you’re not signing up for substandard treatment. Research has shown us that people in clinical trials actually do better than people not in trial. Also, clinical trials are highly scrutinized and have strict oversite by ethical review boards and Health Canada. We simply don’t know which of the two groups might receive a better intervention. With any randomized trial there's that uncertainty, otherwise we wouldn't be doing the study in the first place.

I would definitely advise patients to be informed and be an advocate for their own health care. People have more information than ever at their fingertips — but in this day and age, disinformation makes that a double-edged sword.

Some websites are less credible and rely on biased sources. Information can sometimes be disseminated from a decentralized source, such as a Facebook group. And while there’s many strengths to AI, engines like ChatGPT are ‘people pleasers’ — they’ll always give you an answer whether there’s a credible one in the scientific literature or not. The references and sources that feed AI algorithms are not necessarily vetted and not necessarily true.

Randomized control trials remain our best path to safer, better treatments. To be a part of one is, I truly believe, an ethical act, and I thank all participants deeply. We literally could not do research without you.