In biostatistics, complexity rules

Jesse Berlin (center), ScD ’88, poses with Victor De Gruttola and Dianne Finkelstein of the Department of Biostatistics

November 12, 2013 — When it comes to statistical analysis, “context matters,” according to Jesse Berlin. “Different people look at the same data and come to different conclusions.”

This was one of the issues discussed by Berlin, ScD ’88, in a talk about challenges he’s encountered as a biostatistician on October 31, 2013 in FXB-G13 at Harvard School of Public Health (HSPH).

Berlin was on campus to receive the Lagakos Distinguished Alumni Award, which recognizes biostatistics alumni whose research in statistical theory and application, leadership in biomedical research, and commitment to teaching have had a major impact on the theory and practice of statistical science. The award was established to honor the career of Stephen Lagakos, an international leader in biostatistics and AIDS research, and former HSPH professor of biostatistics and chair of the department, who died in a car accident in 2009.

Berlin has worked in both academia and in industry. He served on the faculty of the University of Pennsylvania for 15 years, became senior director of statistical science at Johnson & Johnson in 2004, and now is vice president of epidemiology at Janssen Research & Development, LLC, a Johnson & Johnson subsidiary.

Berlin offered several examples of the complexity inherent in statistical analysis. For instance, research he co-authored in the late 1990s and early 2000s about balloon angioplasty raised questions about statistical methods that had been used to compare complication risks between a two-day “staged” procedure—in which imaging is done on the first day and surgery on the second—and a “combined” procedure in which imaging and surgery are done on the same day. Some studies had examined the issue using data from various medical centers but failed to adjust for potential differences between these centers. Berlin cautioned that it’s important for researchers evaluating medical procedures to consider so-called ‘operator effects’—that is, to take into account risks or benefits that may be associated with the surgeon who performs a procedure, or with the hospital where it’s performed.

He also talked about the current push for more transparency in clinical studies, so that research results can be reanalyzed for accuracy. (Harvard University researchers, including HSPH’s [[Michelle Mello]], are among those proposing ways to responsibly expand access to clinical trial data.)

“Whether you realize it or not, you’re making ethical decisions every day,” Berlin said. “I’m urging you to think hard about what you do and how you do it, and what the consequences might be, whether intended or otherwise. At J&J, when we’re having discussions around safety issues, someone is guaranteed to come up with the line, ‘If this were your daughter or wife or mother, would you be giving them this drug?’ It’s a good kind of compass, I think.”

Checking out the Department of Biostatistics' new Centennial display
Checking out the Department of Biostatistics’ new Centennial display

At a reception after Berlin’s talk, guests had a chance to see the Department of Biostatistics’ new Centennial display, which includes a timeline of the department’s history and accomplishments and portraits of two of its former chairs: [[Marvin Zelen]], Lemuel Shattuck Research Professor of Statistical Science, and the late Frederick Mosteller.

Paintings of former biostatistics chairs Frederick Mosteller (left) and Marvin Zelen
Paintings of former Biostatistics chairs Frederick Mosteller (left) and Marvin Zelen

 

There’s also a display that describes research by Lagakos and Zelen that found a link between contaminated wells in Woburn and a spate of illnesses and deaths in that city in the late 1970s and early 1980s—the subject of the book A Civil Action and the movie by the same name.

Karen Feldscher

photos: Emily Cuccarese