Alum Wins Biostatistics, Data Science Student Paper Competitions.

School of Public Health alum Isabelle Weir (SPH’19) has received two awards for papers she authored on statistical methods and randomized controlled trials.

Weir, who is currently a research associate/senior biostatistician at the Center for Biostatistics in AIDS Research at Harvard T.H. Chan School of Public Health, won the 2020 Lifetime Data Science Student Paper Competition hosted by the American Statistical Association (ASA) in July, and she also received SPH’s 2020 Biostatistics Doctoral Applied Research Paper Award in May.

She presented the first paper on August 4 at the ASA’s Lifetime Data Science Section Student Award Session, which was held virtually this year during the organization’s annual Joint Statistical Meeting. Titled “Counterfactual Mediation Analysis with Multistate Models for Surrogate and Clinical Time-to-Event Outcomes,” the paper was the final component of Weir’s doctoral dissertation at SPH, and was co-authored by Jennifer Rider, assistant professor of epidemiology, and Ludovic Trinquart, associate professor of biostatistics, who also served as Weir’s primary advisor.

In the paper, the researchers introduced a novel method to evaluate surrogate endpoints in randomized trials with time-to-event surrogate and clinical outcomes, utilizing 18-year follow-up data from the Scandinavian Prostate Cancer Group Study Number 4.

“In this trial, we assessed time to metastasis as a surrogate outcome for time to prostate cancer death,” says Weir. The researchers found that a considerable proportion of the treatment effect on prostate cancer death was explained by the treatment effect on time to metastasis. Using a validated surrogate outcome can “enable researchers to answer important clinical questions in a shorter time frame,” she says.

The second paper, which was published in the journal Annals of Oncology, examined researchers’ interpretation of time-to-event outcomes, as well as their understanding of hazard ratios (the most commonly used measure of treatment effect for time-to-event outcomes). The SPH doctoral award that Weir received for this work is presented annually to a graduating PhD biostatistics student to recognize the leadership roles of biostatisticians in biomedical research.

“This paper provides empirical evidence that misinterpretation of the hazard ratio is common.” says Weir, who surveyed 469 authors of cancer RCTs. “We found that 47 percent of participants misinterpreted the hazard ratio, with 40 percent equating it as a reduction in absolute risk. That’s a problem because often the hazard ratio is further from the null effect than the relative risk and such misinterpretation may lead clinicians to an overly optimistic interpretation of treatment benefit.”

Better education of the hazard ratio is crucial, says Weir. For other statistical reasons and for ease of interpretation, authors should include measures based on the restricted mean survival time (a measure of average survival time over a given timeframe) as a supplemental or alternative measure for RCT results, she says.

In her current role, Weir continues to design, implement, and analyze RCTs for tuberculosis treatment. She says she appreciated the opportunity to work with colleagues across department during her time at SPH.

“The Biostatistics Department offers a very well-rounded curriculum, covering many applied topics while also emphasizing theoretical math,” says Weir. “I received excellent mentorship from my advisor and other department faculty during my dissertation. I am grateful for their support and these two awards.”

—Jillian McKoy