Guided by data

F. Perry Wilson, MD, MSCE, loves medical research, and he is on a mission to get others to appreciate it too. “Like an artist or musician who wants people to understand their language,” he said, “I want the public to know that with the right tools, they can grasp medical research concepts and judge the quality of published studies for themselves.”

A nephrologist with a hospital-based practice, Wilson is an associate professor of medicine and director of Yale’s Clinical and Translational Research Accelerator (CTRA). While he credits inspiring professors for turning him toward nephrology, the specialty’s quantitative nature also suits him. “I’m a data geek,” he said, “and the kidney is the only organ that allows you to sample both the input and the output without getting invasive.”

Wilson is also The Methods Man — an educator with an extensive social media presence. For his regular series of educational videos on Medscape (The Impact Factor), Wilson analyzes the strengths and weaknesses of a variety of published studies with real-life implications. Recent topics have ranged from racial disparities in pediatric post-surgery mortality rates to how physicians might mitigate rising drug costs to whether people with more sex partners are at increased risk of cancer. Over the past several months, Wilson has added a slew of topics related to the COVID-19 pandemic.

“What I began to find a couple of years ago,” he said, “is that certain studies that spread—excuse the expression—virally, are often flawed.” Figuring out what makes problematic studies so appealing to the general public led Wilson to develop not only his weekly broadcasts but also an entire course available for free on Coursera: “Understanding Medical Research: Your Facebook Friend Is Wrong.”

Especially relevant as medical news dominates headlines, the seven-week course uses familiar experiences, animated graphics, and pop-up quizzes to lead viewers through the thinking and methodology behind medical research. Wilson explains in accessible language how misinformation occurs through faulty study designs, bias, and bad reporting, including the conflation of correlation with causality. “Going to primary sources is essential,” he said. “I know that a lot of people say, ‘I can’t read that stuff. It’s like stereo instructions translated from German and Japanese and back to English.’ Yes, there is jargon. But once you learn what the jargon means, if you can think rationally, you can develop the skills to examine and understand scientific studies.”

Two related factors lead some people to embrace bad science, Wilson explains. The first is confirmation bias—the tendency to accept information that supports one’s existing belief structure and reject information that challenges it. The second is motivated reasoning—the tendency to believe studies that justify conclusions we want to reach. “For instance,” he said, “we all want a quick and easy cure for COVID-19. As a physician and parent, I want one too.”

Under normal circumstances, these tendencies may not matter so much—if you read that a certain food enhances or diminishes health and you eat more chocolate or fewer eggs, the world isn’t going to feel the impact. “But,” Wilson said, “with coronavirus, people who are convinced there is a cure around the corner or that the disease is fake are not going to take precautions necessary to stop the spread. Confirmation bias and motivated reasoning have done societal damage in ways I have not seen before.”

Moreover, Wilson points out, even those of us with the best intentions need to be aware that by the time news of a medical study shows up in a Facebook news feed, it has been filtered through several layers of reporting—think of the kids’ game of “telephone.” Plus in our 24-hour news environment, every study is a new news story—often with sensational headlines that contradict each other from one week to the next. “The newest study is not necessarily the truest one,” he said. “No single study is definitive. All the data has to be considered together.”

As a clinical researcher himself, Wilson, who has his medical degree from Columbia and a master’s in clinical epidemiology from the University of Pennsylvania, focuses on interventional data science—using real-time data analytics to create clinical interventions. His primary interest is acute kidney injury (AKI)—the abrupt decline in kidney function seen in 10% to 15% of hospitalized patients. A current major study examines the effect of an electronic alert for AKI on provider prescribing decisions when the patient is on a medication that is toxic to the kidneys. Another study looks at the impact of building the statistical risk of death within a year for heart failure patients into the electronic medical record. Does such information shift provider decisions about treatment plans, and are patient outcomes affected?

“Rather than just pulling data, analyzing it, and writing papers, we use our analyses to feed back into patient care,” Wilson said. “And, as I tell my students and everyone else, I have to make sure I recognize and account for my own biases. You have to form judgments based on the data you have, not the data you want.”