Clinical trials are becoming increasingly expensive, and patients are more and more reluctant to enroll and to persist with the protocol, further increasing costs. We propose to develop Bayesian statistical methods and software for more efficient, ethical, and affordable clinical trials. These methods incorporate all sources of knowledge (structural constraints, expert opinion, and both historical and experimental data), thus reducing sample size and typically leading to increases in statistical power and reductions in cost and ethical hazard, since fewer patients need be exposed to the inferior treatment. The methods are also better able to adapt to unanticipated changes that inevitably arise as the trial progresses.
In Aim 1 we will develop novel methods to exploit available historical data.
In Aim 2 we look at the more general problem of combining inference across related subpopulations, when the subpopulations are not exchangeable. Here our approach is based on random clustering of subpopulations. Finally, Aim 3 is concerned with the development of user-friendly computer code that can be publicly distributed. The end result will be clinical trials that come to conclusions more quickly with fewer patients randomized to inferior treatments. Our methods and software should have immediate impact on practice not only for early phase studies, but for large confirmatory trials as well. In the longer term, improvements in trial compliance and accrual should also obtain.

Public Health Relevance

of this work to public health lies in its ability to randomize fewer patients to control groups in clinical trials, since reliable historical information on such groups will be brought to bear. Patients are thus treated more ethically, and trial conclusions are obtained more quickly. The work has a potentially large impact on small studies of interventions in cancer, alcohol, and many other areas where sample sizes are small yet other sources of information exist that can accelerate the process while maintaining scientific integrity.

Agency
National Institute of Health (NIH)
Institute
National Cancer Institute (NCI)
Type
Research Project (R01)
Project #
5R01CA157458-02
Application #
8435381
Study Section
Biostatistical Methods and Research Design Study Section (BMRD)
Program Officer
Dunn, Michelle C
Project Start
2012-06-01
Project End
2015-05-31
Budget Start
2013-06-01
Budget End
2014-05-31
Support Year
2
Fiscal Year
2013
Total Cost
$285,406
Indirect Cost
$52,535
Name
University of Minnesota Twin Cities
Department
Biostatistics & Other Math Sci
Type
Schools of Public Health
DUNS #
555917996
City
Minneapolis
State
MN
Country
United States
Zip Code
55455
Murray, Thomas A; Hobbs, Brian P; Lystig, Theodore C et al. (2014) Semiparametric Bayesian commensurate survival model for post-market medical device surveillance with non-exchangeable historical data. Biometrics 70:185-91
Ohlssen, David; Price, Karen L; Xia, H Amy et al. (2014) Guidance on the implementation and reporting of a drug safety Bayesian network meta-analysis. Pharm Stat 13:55-70
Muller, Peter; Quintana, Fernando A; Rosner, Gary L et al. (2014) Bayesian inference for longitudinal data with non-parametric treatment effects. Biostatistics 15:341-52
Zhang, Jing; Carlin, Bradley P; Neaton, James D et al. (2014) Network meta-analysis of randomized clinical trials: reporting the proper summaries. Clin Trials 11:246-62
Hobbs, Brian P; Carlin, Bradley P; Sargent, Daniel J (2013) Adaptive adjustment of the randomization ratio using historical control data. Clin Trials 10:430-40