Image
Laura Balzer
University of Massachusetts Amherst Assistant Professor of Biostatistics Laura Balzer graduated first in her class from the University of Vermont, where she studied applied mathematics, and first in her class from the University of Cambridge, England, where she earned a master’s degree in computational biology. As a PhD student at the University of California, Berkeley, she pursued the discipline that combined her love of numbers and commitment to improve the health and well-being of others: biostatistics.
Within biostatistics, Balzer focuses on causal inference, which aims to isolate cause-and-effect from other biasing factors. Says Balzer: “In my first course on causal inference, I saw how statistics and epidemiology could be used to identify the best practices and interventions to improve health, both at the individual level and more broadly at the population level. I remember thinking this is what I want to do with my life.”
Balzer’s work has already had great impact on the global effort to end HIV. And, Balzer’s expertise has been instrumental to the campus’s effort to navigate the COVID-19 pandemic.
Balzer is the primary statistician for three cluster randomized trials in East Africa: the SEARCH study to prevent HIV and improve community health, the SATURN study to improve care outcomes among HIV-infected youth, and the SPIRIT study to prevent tuberculosis. Unlike traditional clinical trials, these studies randomize entire groups, such as communities or clinics, to an intervention or control condition. Balzer notes, “these studies aim to learn about the effectiveness and the implementation of strategies at the population level.”
She has worked with the SEARCH team, composed of researchers from the US, Kenya, and Uganda, for more than 10 years. Last fall, SEARCH received a five-year, $23 million National Institutes of Health (NIH) grant to extend its study.
Balzer relishes being part of the large, multidisciplinary, multinational research. “Big changes in public health and medicine happen through collaborative teams who draw on each other’s knowledge and expertise,” she says. “When we learn from and trust one another, we can make real progress in preventing disease and improving community health.”
Work as a biostatistician on studies involving hundreds of thousands of people does not mean Balzer sees participants as numbers—far from it. As she explains, “In order to improve public health, we need to work hand-in-hand with the community. We have to engage with and care for each person—not just the disease currently afflicting them. We have to have compassionate conversations, meet people where they are, and try to eliminate any possible barriers to prevention, testing, and treatment.”
The first phase of the SEARCH study demonstrated that a strategy based on this community centered philosophy—partnering with communities to test entire populations for HIV and a range of other diseases and treating HIV-infected people compassionately and immediately—prevented HIV transmission and improved overall community health, including reductions in population-level mortality, tuberculosis, and uncontrolled hypertension.
SEARCH’s second phase demonstrated that if you let people self-identify as being at risk for HIV and they take preventative anti-HIV medications when they need them, there will be further reductions in HIV transmission. Current SEARCH studies focus on HIV prevention and treatment tailored to the hardest-to-reach groups, such as young adults and highly mobile populations.
Balzer points out that lessons learned from the HIV and AIDS epidemic apply to the COVID-19 pandemic, although she is disappointed that these lessons haven’t been put into wide practice. She says, “We have immense knowledge on how to meaningfully engage with communities and positively promote behavior that minimizes risk. But too often, the COVID-19 messaging and policies have been top-down and have focused on blaming-and-shaming and punishing ‘bad’ behavior.”
She hopes that the pandemic will lead to increased investment in public health infrastructure and education and a change in attitude. “The value of public health has been demonstrated locally, nationally and globally,” she says. “In the US, we have had an individual view of health, where people think mostly about how to protect themselves and their loved ones. We can see that’s not effective in large scale infectious disease settings; we need to think about protecting and promoting everyone’s health and well-being.”
As a member of the Public Health Response Team, Balzer is using her expertise to advise COVID-19 mitigation efforts for the UMass Amherst community. She leads the team that designed, implemented, and maintains the UMass Amherst COVID-19 dashboard, which communicates timely campus data on the pandemic. She also meets daily with campus leadership to review the dashboard, discuss national data and trends, proactively identify potential challenges, and plan adaptive public health measures. “There’s a lot of work to be done, but there’s also a lot of reason to be hopeful,” she says.
A lesson from Laura Balzer on causal inference
When two variables are associated or correlated, we want to know: does A cause B? However, there is a lot of messiness in the world, and it can be hard to parse out if A actually causes B. As a classic example, consider a data set of people with and without lung cancer and whether or not they have a lighter in their pocket; we will see a very high correlation between lung cancer and lighters. So does having a lighter cause you to have lung cancer? Or does having lung cancer cause you to have a lighter? No. There’s a common variable here, which is smoking; smoking causes lung cancer and smoking causes people to carry lighters. The observed relationship between lighters and cancer isn’t causal.
A lot of the work in causal inference is figuring out which variables to collect, how to best measure them, and finally how to optimally adjust for them in the analysis. To get as close as possible to understanding the underlying mechanisms, we need to be transparent about any assumptions we’re making and try to eliminate any assumptions that might not be supported by our knowledge.
Just because two things occur together does not mean that one caused the other, even if it seems logical.
For an example related to COVID-19, Balzer and colleague Brian Whitcomb, associate professor of epidemiology, recently published an article in the Conversation: https://theconversation.com/coronavirus-deaths-in-san-francisco-vs-new-york-what-causes-such-big-differences-in-cities-tolls-138399