Let’s say, during a checkup, you find out your blood pressure is too high. You follow your doctor’s advice to overhaul your diet and lace up your running shoes a few times a week. Unfortunately, lifestyle changes don’t do the trick, which means it’s time to try medication. Your doctor chooses one that most patients tolerate well, explaining that many different drugs are used to manage blood pressure, and finding the right one may take trial and error. You’re supposed to come back for a follow-up visit in a few months — or sooner, if you have a bad reaction to the drug. If your blood pressure doesn’t normalize, your doctor will revise your treatment plan.
So you take your new blood pressure pill every morning, as instructed. You experience some dizziness, and tingling in your fingertips, but the side effects seem too minor to warrant a call to the doctor. At your next appointment, your blood pressure is lower, but not quite low enough. Your doctor says you can stick with the drug a little longer, or just try something else now. For no particular reason, you decide to keep taking the same one and hope for the best.
That’s the way most of us are used to landing on a medication. But what if finding the right treatment didn’t involve so much guesswork? What if it were possible to know exactly which drug would level out your blood pressure the fastest? That’s the goal of a pilot program at Stanford University called Humanwide.
In 2018, primary care doctors recruited a diverse group of existing patients to wear blood sugar monitors, blood pressure cuffs and Bluetooth-enabled scales at home. These smart devices would enable the doctors to remotely monitor patients’ health over the course of a year. Humanwide patients also underwent DNA testing to assess their risk of cancer and heart disease, and to identify which types of medication would be more and less likely to work for them, based on their unique genetic makeup.
“We used the person in front of us, with all the streams of data available through wearable devices and genetic tests, to craft the most effective treatment for a given patient.”
“We used the person in front of us, with all the streams of data available through wearable devices and genetic tests, to craft the most effective treatment for a given patient,” says Dr. Megan Mahoney, a physician and medical school professor who led Humanwide.
In personalized medicine (also called precision medicine), a patient’s treatment plan is informed by their specific physiology and genetics. While the approach is gaining traction, it’s still largely confined to experimental programs, like Humanwide. So, as of now, personalized trials are only available to a small percentage of patients.
One barrier to widespread use is technology. To bring personalized medicine into exam rooms, doctors need reliable and precise data-collecting technology, says Dr. James Peacock, a cardiologist and electrophysiologist in New York who studies personalized trials. There are some wearable devices, like FitBit, that monitor and track patient health data. But for the most part, technology to help doctors understand their patients’ health outside the clinic isn’t quite there yet. “This is what will allow us to be able to treat a patient differently,” Peacock says.
Part of the case for personalized trials, proponents say, is that they help address a longstanding flaw in clinical research: It’s not diverse enough.
Any and every medication (and medical device) must go through four phases of clinical testing in order to secure FDA approval and hit pharmacy shelves. The purpose of a clinical trial is to show that a drug is safe and effective. An “effective” drug means one that works the way it’s supposed to for a designated patient population.
Historically, clinical trials have included a disproportionate number of white men, while older patients, people of color and women have been underrepresented. Individual characteristics, such as genetic makeup, current drug regimen and even gut bacteria, can affect how people respond to medication. A trial could show that a new drug helps adults with insomnia fall asleep faster. But if three-fourths of the adult insomniacs studied were middle-aged men and zero were women of color, then how widely applicable are the results? It’s hard to know.
One way to overcome this particular study-design problem would be to recruit thousands of participants for every trial. (Typically, a trial might involve any where between few hundred and a few thousand participants.) But Mahoney suggests measuring patients’ individual responses to treatments could be a more efficient and cost-effective solution than supersizing every trial.
Let’s say a doctor sees a 48-year-old Vietnamese woman with hypertension. Knowing that the data from hypertension drug trials is probably drawn from a heavily white, male population — Mahoney says it’s well known among doctors that clinical trials usually aren’t diverse — the doctor could monitor the patient’s response to a treatment over time. “Through the personalized trial design, we would be able to understand which medications would be effective for her, which is much more feasible than [running] a larger trial,” Mahoney says.
One way to make this more feasible is to share information, which means increasing both your dataset and broader access to data. Karina Davidson, a psychologist and professor of behavioral health at Columbia University, has teamed up with the National Institutes of Health to create an electronic platform to help doctors manage personalized trials and the data they generate.
“It’s useful for anything that’s a behavior, a symptom, a condition or a chronic disease that changes across time.”
“Personalized trials help a person, with science, figure out what works best for them,” Davidson says. Currently, she and her colleagues are working with study participants with chronic conditions like lower back pain, fatigue and high blood pressure. Personalized trials identify the best treatments, pharmaceutical or otherwise, when there are lots of options to choose from.
“It’s useful for anything that’s a behavior, a symptom, a condition or a chronic disease that changes across time,” says Davisdon. It wouldn’t make sense, though, for something like a broken leg, which you can either treat with a splint or not.
In the case of lower back pain, study participants try several different methods of pain management, such as yoga and massage, over the course of 12 weeks, and electronically rate their symptoms for each one. The therapies are “controlled” — patients try them one at a time, for the amount of time required to see results. “You know exactly what it is that’s making you feel better,” Davidson says.
Personalized trials can also help patients identify treatments that aggravate their symptoms or cause side effects. Davidson compares the approach to an elimination diet: If you have a stomach ache every day and want to figure out which food group is bothering you, you’d need to stop eating each potential food group or allergen — dairy, wheat, and soy — on its own, long enough for results to emerge. When your stomach aches subside, the culprit should be obvious.
In a clinical scenario, a personalized trial could also help a doctor understand which medication, if a patient’s taking more than one, might be not-so-helpful. Say an elderly patient is on 13 drugs and starts having hallucinations. A doctor could take the patient off each drug until the hallucinations stop. Without a personalized approach, a doctor might just add on another medicine to treat the hallucinations — which could end up causing more side effects in the long run.
There’s another benefit to personalized clinical trials. Beyond allowing doctors to work with patients individually, Davidson’s research will give doctors more diverse data about how different treatments can affect more than just the “average patient.” Mahoney could treat her 48-year-old patient with the information from Davidson’s personalized trial research, and rule out medications that could cause undesirable side effects. “Out of this research, there is a goal of being able to identify a precision therapy for specific populations,” Mahoney says.
But Davidson isn’t focused on how doctors can use troves of data at the moment. Instead, she wants to give doctors the tools to take a more individualized approach to medicine with their own patients, in a one-on-one setting. “My ambitious dream is that we offer this to every patient,” she says. “I’ve moved to the radical position that the universe is the person right in front of me, so I want to know what that person needs.”