A new method developed by researchers at Texas A&M University aims to improve the use of Restricted Mean Survival Time (RMST) analysis in clinical and epidemiological studies. RMST has been a popular tool in healthcare research for over 25 years, offering a simple way to measure the average survival time of patients after diagnosis or treatment. This new approach, however, addresses one of RMST’s biggest challenges: determining the ideal time threshold for analysis.
RMST is favored in clinical settings because, unlike other models such as Cox regression, it doesn’t rely on the assumption that the likelihood of an event happening remains constant over time. This makes it particularly useful for understanding survival outcomes in a range of fields, including healthcare, economics, and business.
However, RMST analysis faces a significant limitation: identifying the optimal time point, or “threshold,” for measuring survival between groups. The difficulty of choosing this threshold has led to less powerful statistical results, which can affect the accuracy of findings.
Gang Han, PhD, a professor of biostatistics at Texas A&M, led the team that developed a new method to solve this issue. The new approach uses a mathematical tool called the “reduced piecewise exponential model” to determine the ideal threshold time for RMST analysis. This is particularly crucial in medical studies, where the risk of events like disease progression can change at different stages of treatment.
Matthew Lee Smith, PhD, a health behavior professor at Texas A&M and a key contributor to the research, emphasized the importance of this method in studies where risk factors fluctuate over time. The team’s approach calculates the threshold time based on significant changes in hazard rates, providing a more precise way to analyze treatment effects between two groups.
The results of this new method were published in the American Journal of Epidemiology. In their research, the team tested the new method using simulation studies and real-world examples. These included clinical trials involving patients with non-small-cell lung cancer and people with mild dementia.
The new model outperformed traditional statistical methods, such as the logrank test, in detecting differences between treatments. In both real-world scenarios, traditional methods found no significant difference between treatments, while the new method clearly identified a superior treatment.
“This new method shows promising results, but more research is needed to explore its potential with more than two groups and additional variables like age, ethnicity, and socioeconomic status,” said Han. “We believe this model could offer a more powerful tool for analyzing time-to-event outcomes in clinical and epidemiological studies.”
In addition to Han, the research team included Marcia G. Ory, PhD, Regents and Distinguished Professor at Texas A&M, as well as Laura Hopkins, a doctoral student in epidemiology and biostatistics. External collaborators from Eli Lilly and Company and the H. Lee Moffitt Cancer Center & Research Institute also contributed to the study.
Related topics:
- This Week’s Major Food Recalls: Apple Juice, Salsa, And More Products Pulled From Shelves
- When to Worry About Hantavirus After Finding Mice in Your Home
- Why Cancer Patients In China Choose Traditional Chinese Medicine
