November 2007 | Back to Table of Contents

Pulse

Bamboozled by Biostatistics

Medical educators are trying to make biostatistics less daunting and more meaningful to practice.

When Colin West, M.D., Ph.D., entered medical school at the University of Iowa in 1992, first-year students were required to take a biostatistics seminar. The 12-hour course covered some of the basic formulas and procedures for running statistical tests. But because anatomy, physiology, and other topics seemed far more relevant to patient care and because biostatistics was graded pass/fail, it wasn’t a priority for many students.

West was different. Having majored in statistics as an undergraduate (he tested out of the med school course), he wanted a career that combined biostatistics with medicine. So he earned a doctorate in biostatistics. Today, the assistant professor of medicine at Mayo Clinic is in charge of Mayo Medical School’s evidence-based medicine (EBM) curriculum, which includes a biostatistics component.

Concerned that medical schools were still making biostatistics an afterthought, West became interested in clinicians’ views of biostatistics. “Every time this has been studied, the numbers suggest that clinicians report having no comfort whatsoever with statistical concepts. Every time a formal knowledge assessment is given—they’re asked to do a multiple choice exam or answer statistical questions in some examination format—they fail miserably,” he says.

A recent Yale University study of 277 residents in primary care specialties confirmed that. Researchers found that when tested on their statistical knowledge and ability to interpret results, residents got 41 percent of the answers correct, on average. Fellows and general medicine faculty with research training did better, scoring 72 percent correct. More important, 75 percent of respondents indicated that they did not understand all the statistical concepts they encountered when reading journal articles; yet 95 percent felt it was important to understand such concepts. The findings were published in the September 5, 2007, Journal of the American Medical Association.

West did a similar study at Mayo Clinic, surveying 301 medical students, internal medicine residents, and internal medicine teaching faculty. In the August 2007 Mayo Clinic Proceedings, he reported that 87 percent said they thought a better understanding of biostatistics would help their career, yet fewer than 18 percent felt their training in biostatistics was adequate for their needs. “The survey suggested very strongly that clinicians are uncomfortable with the field; but they’re also uncomfortable with their discomfort,” he says.

How Does it Relate?
Why the queasiness about statistics—especially when physicians work with numbers every time they calculate risk or interpret test results for patients? West isn’t sure. “We use numbers every day; yet as soon as someone labels those numbers as statistics, suddenly this panic response kicks in,” he says.

West believes physicians’ and students’ feelings about biostatistics may go back to the way it is often taught. In his study, only 20 percent of respondents said they felt their biostatistics coursework was taught effectively. “Traditional stats teaching has been ‘Here are a whole bunch of formulas, and here’s your data, and here’s what you do.’ It’s very stepwise,” West says. “It can be difficult to relate this to the care of patients.”

Karyn Baum, M.D., an associate professor of medicine and special assistant to the dean for educational program development at the University of Minnesota Medical School, thinks similarly. Baum, who graduated from the University of Michigan Medical School in 1995, admits she struggled with a required biostatistics course because the concepts didn’t seem relevant to medical practice. “It didn’t teach me how to appraise evidence in a set pattern or how to search the medical literature,” she recalls. However, by the time she finished her residency in 1998, EBM was becoming an accepted practice. “And biostatistics is one tool you need for evidence-based practice,” she says.

Changing Curriculum
As medical schools began adding EBM to their curriculum, they’ve had to figure out how to get students interested in—and not afraid of—biostatistics. But incorporating such lessons into an already packed curriculum and in a way that’s meaningful to future physicians has been a challenge for faculty at Mayo and the University of Minnesota.

When West took over the EBM curriculum at Mayo three years ago, students spent less than six hours of class time on biostatistics. “In terms of number of hours in relation to other topics in the curriculum, this was not an emphasis,” he notes.

West says students now spend about 24 hours toward the end of their second year learning about EBM—how to write good clinical questions about a patient’s condition, diagnosis, and treatment; search the literature for articles on the topic; determine whether the results of studies are indeed valid and applicable to the patient; and then translate the literature into a language patients can understand. It’s the last two steps—examining the research and using it to help patients make decisions—that require an understanding of statistics. “We try to give them a background in statistical principles integrated into that model,” he says. “However, that integration is not as advanced as we would like it to be.”

At the University of Minnesota, educators are also trying to create a more integrated EBM curriculum that makes biostatistics relevant. Currently, first-year students on the Twin Cities campus are exposed to biostatistics during a six-week segment on managing clinical information. They learn basic concepts such as how to calculate sensitivity and specificity and use them to analyze a medical paper. “But the concepts aren’t as real to students as they could be because they don’t have to see a patient that afternoon,” Baum says.

Baum is developing a self-directed curriculum that will cover topics that transcend specialties. Her idea is to take pieces such as biostatistics that are taught at different points throughout the four years and integrate them. Students will learn concepts online, then do assignments in a clinical setting. “Although it’s really important to know how to analyze statistics, what really matters is understanding how new or old medical literature should be applied to your patient or group of patients,” she says.

Baum says she may adapt a self-paced online biostatistics course that is being taught for the first time this fall on the medical school’s Duluth campus. Students there learn to assess the impact of risk factors on disease, interpret effectiveness of lab tests, and learn concepts related to univariate statistical designs, nonparametric tests, and multivariable tests. “It prepares them for EBM and critical appraisal of the literature, which is the second part of the course. Ninety-two percent of our students go into clinical practice, so they will spend a fair amount of time reading journal articles,” says Rick Hoffman, Ph.D., course director and associate dean for education and curriculum on the Duluth campus.

Because medicine is evolving so quickly, “we’re being drowned in study after study after study,” he says.

And that’s the main reason why physicians are saying they need to feel less intimidated by biostatistics. They want to be able to determine whether the evidence presented in a report is valid and relevant to their patients and then use it in a way that builds trust and improves care.

Having a grasp of biostatistics is “a great mechanism for explaining to a patient why you’re making the decision that you are,” Baum says. “There’s nothing like saying, ‘If I take 15 people like you and put you all on a drug for six years, we’ll avoid three heart attacks. But I don’t know if you’re one of those three’… That’s pretty understandable to people.”—Kim Kiser