Project Summary
One of PCORI’s goals is to improve the methods that researchers use for patient-centered outcomes research. PCORI funds methods projects like this one to better understand and advance the use of research methods that improve the strength and quality of comparative effectiveness research.
What is the project about?
A randomized controlled trial, or RCT, is often the best way to learn if one treatment works better than another. RCTs assign patients to different treatments by chance. But RCTs aren’t always feasible. When researchers can’t conduct RCTs, they can conduct observational studies instead. In these studies, researchers use data from patient records, like insurance claims, to see what happens when patients and their doctors choose treatments.
But current methods for analyzing data from observational studies don’t account for times when a health event doesn’t occur because the patient dies. When death makes an event impossible to observe, it is called a semi-competing risk. Semi-competing risks can bias study results. For example, death is a semi-competing risk for cancer recurrence; if a patient dies, their cancer can’t return. But if the person had lived, the cancer might have come back.
Data from observational studies are often clustered. For example, patients may be clustered within hospitals. When this occurs, analyses must take the clustering into account. In this study, the research team is developing new methods to analyze clustered data from observational studies when death is a semi-competing risk for the outcome of interest.
How can this project help improve research methods?
Results may help researchers compare treatments in observational studies with outcomes that have semi-competing risks such as death.
What is the research team doing?
First, the research team is creating new methods to analyze clustered observational data that include outcomes that are subject to semi-competing risks. The team is testing the methods using simulated data. Next, the team is comparing the new methods to current methods.
Second, the research team is creating new methods to match patients and hospitals on their traits. The new methods measure average treatment effects when semi-competing risks exist. The team is also developing a new way to address missing data, such as data on cause of death or patient traits, in studies with outcomes that are subject to semi-competing risks. The team is using simulated data to test the new methods.
Finally, the research team is testing the new methods using registry data on treatment for head and neck cancer in older adults.
Research methods at a glance
Design Element | Description |
---|---|
Goal |
|
Approach | Hierarchical regression modeling, nonlinear estimating equations, simulations, propensity score weighting, novel multiple imputation |