Biostatistics researcher pursues gold standard in pharmacovigilance

Once a medical product — a vaccine, for instance — is approved for use in the general population, it’s not just left to fend for itself. It’s closely monitored for adverse reactions, a process called pharmacovigilance.

Given the obvious importance of such monitoring on public health, the fact that no gold-standard method for pharmacovigilance exists is a gap, one that UB biostatistician Marianthi Markatou wants to correct with work newly supported by the U.S. Food and Drug Administration.

The research is timely: “We need new methods for monitoring the safety of new medications and vaccines to combat COVID-19 as they reach the public,” says Markatou, primary investigator on the project and professor and associate chair of research and healthcare informatics in the Department of Biostatistics, School of Public Health and Health Professions.

She explains that pharmacovigilance is critical because many medical products are approved after testing in a relatively small number of patients compared to the number of people who will use the products once they’re on the market. As many more people use the medical products, new adverse effects can appear.

The adverse effects that pharmacovigilance reveals are not garden-variety side effects like fever or a sore arm from a vaccination. They are “very rare events, which are usually serious with medical implications,” Markatou points out. One such instance was when some patients developed Guillain-Barré Syndrome, a rare disease where the body’s immune system damages nerve cells, after they got the H1N1 flu vaccine.

Today, identification of adverse effects relies on numerous methods from a variety of fields — including pharmacoepidemiology, informatics and biostatistics — that attempt to show how frequent and how serious an adverse effect is.

Most current methods, however, are “just indications,” Markatou explains, “and you can’t trust one more than another.” The issues associated with these methods include identifying too many or too few potential adverse events. For instance, if a method identifies too many adverse events, it can affect the approval of new medications because final approval of a product depends on a product’s “safety profile.”

Perhaps the pharmacovigilance gold standard has been elusive to this point because the problem is “very complex. You are making a final decision on whether something is an adverse event and associating that with a product, which then could be pulled off the market,” Markatou says. Those decisions are currently based on scientific input and judgment calls, and have serious implications for people who use the products and for the companies that make them.

The new methods Markatou is evaluating aim to discover how to — with a high probability — reduce the number of events that appear to be adverse. Her project will promote the use of strong scientific methods and create new ones, ultimately sharpening and facilitating accurate monitoring. She also notes that technology has very recently become powerful enough to support this work.

“The method has to have a solid foundation and solid algorithm. But you can’t identify adverse effects only by algorithm — you also have to understand the science, as well,” she says, adding that the whole process requires input from experts in statistics, biomedical and pharmaceutical sciences, and other fields.

Ultimately, she says, developing a method that’s understandable is extremely important. “The safety of medical products continues to be a major public health concern, and adverse events are a serious concern worldwide.”

In addition to the FDA, Markatou is collaborating with Andrew Talal, professor of medicine, Jacobs School of Medicine and Biomedical Sciences, and Steve Allen of HealtheLink. Her team also includes graduate student Anran Liu and Saptarshi Chakraborty, assistant professor, both in the Department of Biostatistics.