Since you’ve been collaborating on multidisciplinary teams as a biostatistician, what has been the most high-impact research you’ve worked on?
I’ll answer from two different perspectives—first when your research appears in a high-impact journal and second when your research impacts public policy.
In studying the outcome of mortality and health care expenditures post-bariatric surgery, along with my colleague Matt Maciejewski, I built rigorous studies using electronic health record data that led to publications in JAMA and JAMA Surgery. Unlike most, we had access to a patient’s BMI and healthcare expenditures prior to surgery, allowing for a precise comparison. Our results showed that surgery didn’t reveal any cost savings as previously thought (published in JAMA Surgery).
We’re now at the tail end of a third study that investigates long-term mental health post-bariatric surgery and we’re preparing to work with Kaiser Permanente on our fourth grant researching health care costs following bariatric surgery. We hope to discern if there are certain subgroups with more cost effective outcomes than others and to learn what’s driving increased costs—for example, is it complications from the original bariatric surgery, plastic surgery to remove excess skin, or eligibility for a knee transplant?
From a policy impact perspective, I’ve been collaborating with Courtney Van Houtven, PhD, a health economist, on a project funded by the VA Caregiver Support Program’s national office. Our work has been instrumental in executing their caregiver program, and we’ve provided feedback directly to the national leadership and the Secretary of the VA. Senators also used our research when developing the MISSION Act, which expands VA services, including the caregiver program, to pre-9/11 veterans.
What do you find are the benefits of working on multidisciplinary teams, and are there techniques or characteristics that make working on them a success?
You have to appreciate working across silos, feel comfortable focusing on multiple content areas, and enjoy taking on a ‘big picture’ role, which can include sharing methodological ideas with different research teams and translating the language from one study to another.
Most recently, I designed a way to model cost data with a lot of zeros, which we used in our JAMA-published bariatric surgery study. Using inpatient costs as an example, previously we would model cost data with a large number of zeros in two separate parts—whether or not a patient had inpatient costs and separately, those with a range of positive costs. My model combines the 2 components and accounts for the large proportion of zeros, but the interpretable output is the same as what you’d receive from a standard regression.
Many think a statistician does analysis at the end of a project, but my preference is to be involved from the initial idea to the last published paper. The PI and I will first discuss the best way to approach answering the research question—based on the data, can we can answer it at all, and if not, what question can we ask that’s as close to the original as possible? A well-trained statistician will think about the implications of every small data decision.
It’s said that research outcomes are more innovative when working on a multidisciplinary team, with your research in mind, can you describe a project when this happened?
I’m currently working on a study with Courtney Van Houtven, a health economist and Nicki Hastings, an MD and geriatrician. We’re studying patients treated by a geriatric-specific primary care medical home (care delivery model where a patient’s treatment is coordinated through their primary care physician) compared to a traditional primary care model. In real-time, we prospectively matched and contacted patients as they transitioned to the geriatric-specific model and asked for their perspective on their care outcomes. The team’s individual areas of expertise were crucial to our success—Nicki for how and when to identify patients on a national scale and for her expertise on the transition process and geriatric care; Courtney, for the types of question to ask that evaluate the transition; me for the biostatistical plan and matching methodology; and all three of us for our rigorous study design.