Health care professionals working in a range of roles and settings can benefit from strengthening their clinical research skills and learning to examine data from different perspectives. That’s one of the lessons recently learned by scholars in Harvard Medical School’s Postgraduate Foundations of Clinical Research program. As part of their coursework for this six-month certificate program, the participants took part in a project where they were grouped into multidisciplinary teams and challenged to discover new insights using existing datasets. This experience helped them hone their investigative thinking skills and practice using statistical analysis techniques.

Such in-depth clinical research training lays the groundwork to develop a new generation of independent investigators, while at the same time enabling clinicians to advance their careers through a greater understanding of research evidence, according to Kenneth Christopher, MD, SM, faculty director of Postgraduate Medical Education at Harvard Medical School and co-director of the Foundations of Clinical Research program.

This hands-on practice also demonstrates the real-world value that a foundation in clinical research can bring to researchers, patients, and the medical field as a whole, adds Jamie Robertson, PhD, MPH, director of innovation in surgical education at Brigham and Women’s Hospital, who also shares the role of co-director of Foundations of Clinical Research.

Challenging Students to Apply Research Skills

The Foundations of Clinical Research scholars were asked to come up with an original research question that could be answered using data available through the National Institutes of Health’s Framingham Heart Study. The study consists of data points collected from more than 15,000 participants followed over the past three generations in the hopes of gaining a deeper understanding of cardiovascular disease and related risk factors.

For this assignment, each team had only two hours to take the study data and examine it through unique angles to determine the statistical relationship with key risk factors, then briefly present their findings using a slideshow format to the larger group. The idea was to identify possible behaviors or conditions that can be associated with a heart attack, stroke, or other negative cardiovascular outcomes.

“It was very impressive to see what scholars could accomplish in such a short amount of time,” Christopher points out.

He says that the scholars had an opportunity to draw on their coursework in epidemiology, biostatistics, statistical programming, and study design to understand existing findings in a new light.

Examining Data from Different Research Angles

Using the Framingham Heart Study data, the scholars explored a number of topics. For instance, one group looked at the relationship between obesity and stroke risk. They had expected that obesity would be independently associated with stroke, but determined through their data analysis that hypertension and diabetes were stronger indicators of stroke likelihood than obesity alone. Their findings revealed that patients who are obese are also more likely to have hypertension and diabetes, which they believed helps explain the relationship.

Other groups looked at the association between smoking and stroke. For instance, they considered the number of cigarettes smoked per day and how this correlated with the likelihood of stroke. The scholars used existing data sets to determine how many cigarettes were smoked per day by adults in the study, adjusting for age, hypertension, and other factors to try to narrow in on the association between number of cigarettes and stroke risk. Other scholars used a linear regression to further explore the cigarette/stroke relationship, determining that although smoking itself was associated with stroke, the number of packs smoked per day didn’t seem to affect this risk.

Another group explored the link between having a blood pressure reading of more than 140 systolic and the risk of experiencing a stroke. They also looked at the results of patients with diabetes and high blood pressure and determined that those with diabetes whose blood pressure was over 140 had much higher odds of having a stroke than their counterparts. In addition, another angle scholars explored was the connection between diabetes and the amount of time to stroke. Finally, one group discovered that a history of previous stroke increased the risk of experiencing a second stroke.

Identifying Common Research Analysis Flaws

While all these angles were informative, however, some of the scholars encountered important flaws in their formulas and calculations that skewed their conclusions.

For instance, the types of common mistakes the student-scientists made included grouping different categories together without separating out different risk factors, making it hard to identify the actual relationship; not including all of the variables tracked, so the numbers didn’t add up properly or didn’t represent the true picture; not comparing two comparable groups to be able to accurately identify the differences that occur; focusing on too many different scenarios (both adjusted and unadjusted) that made it difficult to get a clear understanding of the risks; and combining different types of data analyses (such as logistical regression and linear regression), rather than sticking with one methodology.

Preparing a New Generation of Clinical Researchers

Overall, the exercise presented a valuable opportunity for participants to apply the knowledge gained through the coursework and to see how it can play out in a meaningful way. Both Christopher and Robertson also agree that the project illustrated the challenges and benefits inherent in designing and implementing a research study. Their hope is that this experience will prepare scholars to effectively explore situations and data sets from different perspectives. This can provide them with a strong foundation to build on as they continue to develop their clinical research careers.

Written by Lisa D. Ellis