Project Summary

What is the research about?

Treatment for chronic diseases, especially in mental health, pain disorders, and opioid and substance abuse, is often sequential and requires individualized regimens. Randomized clinical trials (RCTs) provide high-quality evidence of treatment efficacy, but there is often a gap in the results from RCTs and clinical practice. Most RCT designs are relatively simple and do not address the sequential nature of treatment.

Sequential, multiple assignment, randomized trials are designs that can account for individualized sequences of treatments. Assignment to treatment at multiple time points of a patient therapy is based on a random process similar to a toss of a coin. The patient has no say in the therapy. In some trials, subjects may have a strong preference for one treatment and may not enroll in the trial or may drop out of the trial if they are not assigned to that treatment. Both of these situations cause problems in the interpretation of trial results.

Patient preference trials allow for a subset of the subjects in a trial to be assigned to their preferred treatment. The combination of patient preference and SMART has never been studied and this research will focus on the appropriate analytical methods for such a trial and under what conditions researchers need to limit the use of such trials.

Who can this research help?

Results from this research helps both patients and researchers. More patient preference SMARTs would allow patients with strong preference to participate in meaningful research and this in turn helps researchers because it widens the patient population in these trials.

What is the research team doing?

The research team is developing statistical methods to appropriately analyze patient preference SMARTs. This includes creating an online calculator that will help researchers determine how many subjects are needed in such trials. Finally, the research team is studying under what scenarios the analyses of a patient preference SMART break down. As an example, the analytical methods might break down if all of the patients prefer a single treatment at one of the time points of therapy.

Researchers in pain, substance use disorder, and patient advocates are in an advisory committee giving feedback to the research team.

Research methods at a glance

Goal : To develop statistical methods that increase efficiency of partially randomized, patient preference, sequential, multiple assignment, randomized trials.

Approach :

  • Developing novel methods to combine information from patients given their preference treatments and patients given a treatment based on chance
  • Using simulations to determine what the appropriate conditions are for use of patient preference SMARTs
  • Providing guidelines for sensitivity analyses of methods
  • Developing software to assist researchers plan patient preference SMARTs in terms of total sample size, proportion of patients who can be given their preference, and preference rates of individual treatments

Project Information

Kelley Kidwell, PhD
University of Michigan
$998,998 *

Key Dates

July 2021
January 2026

*All proposed projects, including requested budgets and project periods, are approved subject to a programmatic and budget review by PCORI staff and the negotiation of a formal award contract.


Award Type
State State The state where the project originates, or where the primary institution or organization is located. View Glossary
Last updated: March 15, 2022