Chris Franck is an associate professor in the department of statistics at Virginia Tech.

I met Dean Follmann during a visit to the Biostatistics Research Branch of the National Institute of Allergy and Infectious Diseases. A former student invited me to spend the day there and give a talk in March 2023. I was astonished during my visit, as I figured the group would be specific in focus and hugely preoccupied with COVID. But instead of a narrowly focused group deep into their own weeds, I met the most incredible collection of statisticians.

There were textbook authors, theoreticians, application-oriented methodologists, statistical consultants, adjunct college professors, and even a National Institutes of Health program officer.

They advise researchers about grants and design and review clinical trials. They analyze data. They publish academic papers. The group is broadly focused on allergies and infectious diseases, including COVID, but they do a lot of different types of statistical work, even more than I have mentioned here.

Everybody was super friendly, and I found the many conversations I had extremely engaging. It was sort of the perfect day of statistics. I grew curious to learn more about Dean, who leads the group and coordinates the many complexities of such a work environment, so I asked him for a follow-up interview, which I present here.

Dean Follmann

When and how did you discover the statistics field?
I started out studying psychology. After a while, I grew dissatisfied with that. There were a lot of theories but not a lot of evidence or proof to determine which was right. As I was taking psychology classes, I also took a statistics class. At first, I really hated statistics, because I didn’t understand it. They had a big X and a little x, and they meant different things. I thought, “What’s up with that?”

I struggled through it, and then I decided a statistics minor would be a nice addition to a bachelor’s degree in psychology. And then I had a professor named Richard Kryscio, who taught me regression analysis, and everything just kind of clicked. Suddenly, I could really understand what the equations meant by relating them to plots and geometry. This ignited a curiosity in me, and I just wanted to keep learning more.

That one class really turned it on to me. I could see things differently and see that there was more and more I could do and more and more I could learn. I then went to Carnegie Mellon, which I loved and gave me the foundation to go anywhere.

You currently serve as the chief of the Biostatistics Research Branch at the National Institute of Allergy and Infectious Diseases. Can you tell us about that job?
Yeah. So, I have two hats. The first is as the branch chief and the second is as a working statistician. I’ll talk about the working statistician part first. The NIH [National Institutes of Health] has been a great place to work for me. I have a very supportive boss who gives a lot of leeway and allows for a lot of creativity. I also have many collaborators.

I view our mission at NIH in a broad sense, which is to improve human health.

Various projects come up, and I always think about the best way to approach them in a solid and straightforward way. But I also always wonder if there is a better way. Is there a cool approach that could be developed that would be statistically interesting and fill a knowledge gap? I collect these little ideas, and they get turned into my research. More specifically, there are a lot of tasks we do. We collaborate, design studies, do analysis of small data sets, review protocols, provide advice, etc. Different collaborators will want to know about a statistical method or whether a design is sound, things like that.

I view our mission at NIH in a broad sense, which is to improve human health. We collaborate with researchers around the country and at different universities, as well as with different institutes, depending on where projects lead. Within NIAID [National Institute of Allergy and Infectious Diseases], there are divisions, and there’s an intramural division, which is rather like a university where you have people working in different departments, doing basic research or applied research. We collaborate with intramural researchers to design studies, develop new assays, analyze clinical trial data, and so on. The other part is extramural research, which is the part of NIH that funds outside groups to do research. These outside groups will propose studies that go through scientific review here.

We’ll also participate in ongoing studies that require independent monitoring. And so we get involved in the setting up of committees to do that, and we need to understand the statistical aspects of monitoring. When I was first at NIH, I spent a lot of time working on the statistical issues related to monitoring test statistics over time. So, that leads to standardizing test statistics so they behave like Brownian motion and leads to boundaries that control the type I error rate. There’s a lot of interesting math related to that.

You mentioned the other hat you wear is as the director of the biostatistics group. How do you balance research and administrative duties? How has this changed over the course of your career?
I try to avoid administrative work. I’m more interested in mentorship. I like to work on projects with people in the group. We have a bunch of people in the branch, and I want to give them opportunities to do what they want to do.

My job is to line up the right project or opportunity with the right people.

While we do a variety of things, including monitoring clinical trials, we also have people who are interested in the ethics of clinical trials. We have a person in the group who’s interested in developing biostatistical capacity in Sub-Saharan Africa. Some members are interested in helping train Sub-Saharan African statisticians. And we also have a member of the group who’s developing research capacity to do overseas trials in outbreak settings. They set up a data coordinating center that ran a trial identifying the first treatments for Ebola virus disease. That’s all sort of aligned with improving public health, and I’m supportive of that.

And, like different people in the group, I want to sort of play matchmaker, give them opportunities they are interested in. I think that’s a major part of what I do—and to be available and listen to what people say. And, you know, try to create a good environment for people. My job is to line up the right project or opportunity with the right people.

It sounds like there is a lot going on in your group! What skills are needed in an environment like this?
We have collaborators who work on infectious disease and immunology. Skills you really need are good technical statistical training, including the theory behind the software programs you might run or the analyses you might do. You need the ability to extend things and develop new methodologies.

Being able to communicate is very important in this institute, as well. We work with a lot of scientists who need to understand what we’re doing, and so being able to communicate on their level is critical. This requires statisticians to understand immunology, vaccinology, infectious diseases, and biology.

I think there are opportunities and flexibility in this group, so ideally you want people with some initiative who can make the most of this environment. You also want people who fit in and complement the group.

What is a typical day like for statisticians at NIAID?
Let me give you a few examples. Some days, there will be a big meeting, for example, a data safety and data safety monitoring board meeting where an independent group will come together and look at data from an ongoing clinical trial. So, on a day like that, you would spend a chunk of time beforehand reading reports, trying to understand what’s going on or whether there are issues that need to be addressed. Then, at the meeting, you will discuss the data and maybe make recommendations to modify the trial or have everything continue as it is.

I feel there are situations in which statisticians should lead teams, situations in which we should be the ones driving the science. Typically, statisticians talk about giving the results of their analyses to decision-makers. I think sometimes we need to be the decision-makers.

People in the group also go to FDA [US Food and Drug Administration] advisory committee meetings where the format is similar. You prepare for several days for a controversial issue related to a drug that’s going up for licensure. So, this committee will meet, and we’ll discuss the pros and cons, the strength of evidence for the drug benefit versus maybe a safety signal, and then answer specific questions related to that. That’s a kind of typical day. I would say they’re not super common, but they’re not uncommon, either.

We also do methodological work. Right now, I’m obsessed with how hazard functions relate to frailty, so I’m reading books and papers but mostly thinking about equations and writing things on the board, typing LaTeX, and then running some R code to get clues about how these things behave. And talking to other branch members with similar interests. Obsessions like this happen every so often, and it’s pretty great while they’re around. That is another kind of typical day.

The third type of typical day involves data analysis, so running and debugging code and writing a report based on that, briefing collaborators, and iterating.

Finally, there are meetings. Not a ton of those, but there are some meetings.

I think there’s opportunity here to be influential, maybe more so than other places. I feel there are situations in which statisticians should lead teams, situations in which we should be the ones driving the science. Typically, statisticians talk about giving the results of their analyses to decision-makers. I think sometimes we need to be the decision-makers.

You have an impressive publication record. I see you have a paper called “Efficacy and Safety of the mRNA-1273 SARS-CoV-2 Vaccine” with almost 10,000 citations since 2021. What has it been like on the front lines of biostatistical work related to COVID?
It has been surreal and amazing. You know, I work at the National Institute for Allergy and Infectious Diseases, so we study infectious diseases. And then COVID hit and there was an incredible amount of work to get done. We were involved in the early treatment and vaccine studies, which continued throughout the pandemic. We had to set up everything very, very quickly. There was no time to fret, basically, which was kind of liberating. I specifically was involved in the Operation Warp Speed vaccine trials. They moved at an incredible pace, and we had to wrestle with all sorts of decisions one after another after another.

We had to set up everything very, very quickly. There was no time to fret, basically, which was kind of liberating.

For example, we originally planned to follow placebo and vaccine people for two years to see how well and how long the vaccine worked. But that didn’t really make any sense because the volunteers on placebo should get the vaccine as soon as its efficacy was established, since they had shown through their contributions in the trial that the vaccine was efficacious.

While it seems placebo vaccination destroys the opportunity to learn how long the vaccine works, it actually doesn’t. The simplest way to understand this is that even if the placebo group has received the vaccine in a randomized trial, you still have a fine randomized contrast between those who just got vaccinated versus those who were vaccinated a while ago. So, if the event rate is higher in those who were vaccinated a long time ago, the vaccine must be losing efficacy. That was the basic intuition that led to a formal statistical method.

Normally, when I have an idea like that, I develop it thoroughly, maybe taking a year to work it out completely. With COVID, I quickly realized that wasn’t going to work, so we ended up engaging a big group of clinicians, statisticians, and vaccinologists who were involved in Operation Warp Speed so we could have a united voice on this particular topic. Moving so fast and engaging a large heterogenous group was new and awkward for me, but the situation required it.

I felt what our group was doing in response to the pandemic was important. It was scary at times, but there was a lot of energy. I feel we fed off each other and our camaraderie provided support for each other. I feel the group did a fantastic job in response to COVID-19—I felt like we were built for that moment.

Do you still collaborate with that team?
Yeah, it’s interesting. The paper you cited is the seminal paper that evaluated the Moderna vaccine, and I continue to work with that team three years out. We’re analyzing data from the trial and thinking of new analyses and experiments that can be performed. We respond to how the virus changes, but, more broadly, to have different methods that will suit other vaccines. I now work on Ebola virus vaccines, and some of the things I’ve learned from COVID apply to Ebola. At some point, COVID-related work will be mostly in my rear-view mirror, but it is taking a while. I met some great scientists and want to continue those connections as best I can.

One final thing about COVID. One of my jobs is to provide advice to NIH leadership. At the start of the pandemic, I got a terse email basically saying, “Here’s a study. What’s going on?” I wasn’t sure what to do, so I made this little rule: Give your best answer in 23 hours. This lets me prepare something, sleep on it, and then make sure I’m good with it.

When these requests arrive, I drop everything. I read the study and usually read other papers to understand what methodology they were using so I can explain the study well. But I feel it’s my role to go beyond that and come up with insight or larger context.

One of the last requests was to evaluate a study in Boston Public Schools in which a mask mandate was changed to a local decision. You could look at the COVID rate before and after the masks were lifted in the schools, so it was a kind of natural experiment. For the larger context, I went online and read the high-school newspapers from Waltham and Chelsea to try to glean clues about behavior related to masking and if this might cause biases. Waltham lifted the mandate rather quickly, while Chelsea kept the mandate.

The Waltham article included interviews with students with some for and others against the lifting. Chelsea was different. I ended up reading a report about how the Chelsea area was hit first and worst and how overcrowding, pollution, chronic diseases, and social mixing from jobs contributed to COVID having a major impact on the Chelsea community. Not surprising, the high school kept the mandate. I didn’t end up explicitly using this in my response, but I knew the context better, which made me confident about my analysis.