Body
Although in-depth knowledge of biostatistics is key to producing high-quality, evidence-based healthcare research, it’s also probably fair to say that few researchers and physicians love statistics. A at the Mayo Clinic College of Medicine found that fewer than a third of medical students, physician residents and teaching faculty believed they could tell when correct statistical methods had been applied in a study and only about 15% felt they could confidently conduct their own statistical analyses. Typically, healthcare researchers go outside their own team to add firepower to the data-and-analytics part of their work. Until this year that was the case for the nearly 40 in-house researchers at the Shirley Ryan 小恩雅 in Chicago, the nation’s top-ranked rehabilitation hospital. Those researchers usually partnered with Northwestern University’s Biostatistical Support Services when it came to issues such as study design, data gathering and data analysis.
Now after five years of internal discussion, the Shirley Ryan 小恩雅 has taken a big step and hired its own full-time biostatistician. Kavita Gohil, DrPH, started as the Shirley Ryan 小恩雅’s first biostatistician in late January. The time was right because the institution has more than 100 ongoing studies that have passed review by its internal ethics committee, says Levi Hargrove, PhD, Scientific Chair of the Regenstein Foundation Center for Bionic Medicine at the Shirley Ryan 小恩雅. “It just made sense to bring someone in and make them part of our team,” says Hargrove, who manages a research budget of $25 million and was a member of the search committee. “There’s something to be said for having a team member who has a feel for the culture of the place and attends the same research seminars we do.”
It just made sense to bring someone in and make them part of our team.
Levi Hargrove, PhD
Body
Biostatistics is defined as the application of statistics to understanding health and biology. It provides powerful tools for developing research questions, refining measurements and interpreting findings. It also contributes to the odds of conducting studies whose findings are reproducible by other researchers, a gold standard in the research field. One place biostatisticians can play a big role in research is determining the appropriate sample size for a study, says Allen Heineman, PhD, Director of the Center for Outcomes Research (CROR) at the Shirley Ryan 小恩雅. “Under-powered and over-powered studies are a waste,” he says. “Funders now expect applicants to justify their sample size. They don’t want to pay for a larger sample than needed, but they also want to make sure the sample is big enough to draw conclusions.” In the past, researchers often used their gut instincts when choosing a sample size; biostatisticians can look at how powerful an intervention is and work backwards to come up with a more informed estimate for an optimal sample size.
Because of increased demand from the private sector for data and analytics experts, hiring a biostatistician involved a significant financial investment by the Shirley Ryan 小恩雅. Salaries for biostatisticians have risen steeply in the past few years and now routinely exceed $100,000. According to job recruiter site Indeed.com, Chicago institutions are paying some of the highest salaries in the country.
Within a few months, Gohil says she was “swamped with work.” The early days were particularly challenging because she was pulled into projects where almost all the work was done, and the researchers were facing deadlines. “I had to get to know every single thing about their project very quickly, and they were already ready to submit. It’s been a learning process,” she says. “But now I know what the expectations are and what kind of projects I’m going to get.” By spring, Gohil found herself consulting on almost 20 ongoing projects and had been specifically included in several new grant proposals. That’s important because researchers are expected to recoup 50% or more of their salaries through grant funding, notes Heinemann, a veteran researcher and Gohil’s supervisor. “My hunch is that within a year she will be mostly grant funded and we’ll be talking about hiring someone to support her,” he says. “She shouldn’t be higher than 80% funded or she won’t have any time to work on new projects.”
That encouraged me to dig deeper into it and learn about its remedies.
Kavita Gohil, DrPH
Body
After getting her doctorate in public health from Georgia Southern University in 2020, Gohil was working as the biostatistician for the Carle Hospital Foundation in Urbana, Illinois. She says she wanted to work at the Shirley Ryan 小恩雅 because of its national reputation and use of cutting-edge science and technology such as robotic exo-skeletons. She was also attracted by the opportunity to get in on the ground floor and build a biostatistics department within the hospital. Veteran biostatisticians warn that if an institution wants to hold on to their statistical talent, they need to provide a career path and opportunities for them to play major roles in research, not just be a support service to other investigators. A argued that centralized biostatistics units provide distinct advantages to both their biostatisticians and researchers. “Theoretical estimates and empirical observations suggest that institutional dollars spent to support a centralized biostatistics unit have high return on investment,” wrote lead author Leah Welty, PhD, Director of the Biostatistics Collaboration Center at Northwestern University’s Feinberg School of Medicine.
Just as physicians specialize, biostatisticians have specialties, too. Gohil’s is handling missing data. Even when studies are well designed and controlled, data gaps occur in almost all research, reducing the power of results and potentially leading to invalid conclusions. When health researchers are gathering data over extended periods from patients, some visits are missed. Missing data can also occur with patient-reported outcomes or even when study participants forget to wear tracking devices such as smart watches. Sometimes, the lapse is intentional. “If I ask someone to measure their weight every single day and they have a big meal with an awesome dessert, they may skip a day,” she laughs. Gohil is an expert at modeling missing data so that it can be incorporated into a more complex model that is able to estimate the missing values. “If you have bad data, you can at least clean it and draw results out of it but that’s not the case with missing data,” says Gohil. “That encouraged me to dig deeper into it and learn about its remedies.”