(U.S. Deprescribing Research Network) Novel Methods for Estimating the Effects of Deprescribing using Observational Data

Abstract Project Title: Novel Methods for Estimating the Effects of Deprescribing using Observational Data Taking a large number of medications is common among older adults and increases the risk of harmful medication side effects. “Deprescribing” is the process of stopping medications under the supervision of a doctor or other healthcare professional when medications are more likely to cause harm or when they no longer provide enough benefit. Many older adults and their healthcare professionals believe that more evidence is needed before deprescribing can occur. Randomized trials are one of the best ways to generate that evidence and learn if stopping a medication is helpful and not harmful. People would be assigned to stop their medications or continue them by chance in trials. A problem is that trials are often too expensive, take a long time to complete, and may have trouble getting people to participate. In those cases and others where randomized trials are not possible, researchers can instead conduct observational studies. In observational studies, we can still learn whether deprescribing will be helpful and not harmful. The difference from trials is that older adults and their doctors or healthcare professionals choose (rather than leave it to chance) whether to stop the medication. There are some challenges to using observational studies though. One of the most important challenges is that it can be hard to measure when somebody has stopped a medication because of how the data are generated. Our research team believes that a new statistical method may help to solve the problem and assist researchers with obtaining the correct information from observational studies, very much like from randomized trials. Our research team plans to conduct two studies. In the first study, we will use a computer program to create special data to make sure that the new statistical method actually works well. The first study will confirm whether the method can help researchers like us to more accurately measure when older adults stop medications in observational studies. In the second study, we will apply the new statistical method in real-world data that many other researchers use for their observational studies of stopping medications. The second study will help show us whether the statistical method can work well in more complicated data from the real world. It will also show us if we can obtain similar results from observational studies and randomized trials of deprescribing. If the new statistical method we plan to study works well, the results from observational studies will be more accurate and useful. Researchers, older adults, and healthcare professionals will then have more data about whether stopping medications is safe and helpful for different health conditions and medications. Researchers can then figure out and help to design the best ways to deprescribe medications. Scope of Work Matthew Duprey, PharmD, PhD, BCCCP is an Assistant Professor in the Department of Pharmacy Science and Practice at the University of Kentucky College of Pharmacy. As Principal Investigator for this project, he will work to accomplish the two aims set forth. He will direct the proposed research, including leading the study design, data management, creation of relevant variables, validation of the novel methods, and analysis. He will lead the development and dissemination of abstracts and peer-reviewed scientific manuscripts based on this work, both at the Annual Network Meeting of the U.S. Deprescribing Research Network and other national conferences. He will also be responsible for holding regular meetings with the research team. Finally, he will lead the development of one or more future larger grants based on this work. Daniel Manion, MSc is a data management analyst in the Department of Pharmacy Practice and Science at the University of Kentucky College of Pharmacy. As the primary data analyst and programmer, he will work directly with Dr. Duprey to contribute to various aspects of data programming and developing the data analytic files. Mr. Manion will also assist with carrying out the simulations, statistical analyses, and data visualization for manuscripts and presentations. Finally, Mr. Manion will review all of Dr. Duprey’s code to ensure high accuracy and the ability to disseminate the software code publicly to other researchers.