The University of Iowa




CTSDMC Innovations

The Clinical Trials Statistical and Data Management Center is a collaborative group of individuals working on the cutting-edge of many developments in clinical trials. Clean, accurate, reproducible data is our number one goal, and we strive to produce it as efficiently and cost-effectively as possible. Some of our recent Innovations include:

The CTSDMC has a wide array of experience in innovative study design, including: 

  • Biomarker Discovery: The NN101 study of Spinal Muscular Atrophy (SMA) biomarkers, a prospective, multi-center natural history study, was one of the first multi-site studies to enroll infants with SMA along with healthy controls in a prospective fashion. Typically, natural history data in infantile onset SMA have been incomplete and retrospective, collected in a non-systematic fashion at specialized academic sites with no control patients. NN101 confirmed SMA diagnosis by genetic testing prior to enrollment, and met recruitment goals in a very difficult disease setting. This study collected longitudinal data on standard motor function assessments as well as potential physiologic and molecular biomarkers for disease progression. This study database is one of the first complete natural history databases for infantile onset SMA with healthy controls, and identified potential biomarkers for disease progression. Results of the study were published in Annals of Neurology, and ultimately used in the FDA approval of a novel medication for treatment of SMA. 
  • 3-in-1 Design: The CHAMP study, which involved three treatment arms (amitryptiline, topirimate and placebo), was one of the first multi-center Phase III trials conducted in the pediatric migraine population. The study design included pre-planned analyses for comparisons of the two active medications vs. placebo, and a head-to-head comparison of the two active medications. The CTSDMC developed an a-priori decision algorithm to determine the "best" first line treatment based on cumulative information across all three groups. The simulation studies indicated that the correct first line treatment was selected across a variety of plausible scenarios, and the study's power to determine the correct first-line treatment was greater when results were examined together in the three-arm study as opposed to separate two-arm studies.
  • Futility Design: The NN103 study of Rituximab in Myasthenia Gravis (MG) was a Phase II, placebo-controlled study examining the effect of Rituximab on composite endpoint (prednisone dose reduction and no clinical worsening as measured by the MG Composite scale). This study used a futility design, which tests the hypothesis that patients treated with Rituximab would achieve at least a 30% increase in rate of favorable endpoint. A traditional efficacy design would have required substantially more patients, an unfeasible scenario in this rare disease setting. The futility design used in NN103 was able to show whether or not further study in a Phase III setting would be likely to show clinically meaningful results. The CTSDMC, along with clinical experts on the study team, also developed a prednisone dosing algorithm for this study, and the study team handled central laboratory collection of B-cell and antibody data. Futility criteria for this study was met, suggesting that further study of Rituximab in a Phase III study is unlikely to produce clinically meaningful results. B-cell depletion from baseline to week 52 follow-up was observed in both groups, but the difference between the groups was not statistically significant.

Adaptive design allows modification to the trial or statistical procedures after its initiation without undermining the validity and integrity of the results. The goal of adaptive design methods is to make trials more flexible, efficient and fast. Adaptive design requires a pre-planned analysis while the trial is ongoing, and could be used to refine sample size, eliminate or refine outcome measures, or to stop the trial.

Dr. Coffey has extensive experience in adaptive design: 

  • McClure LA, Szychowski JM, Benavente O, Hart RG, Coffey CS. A post hoc evaluation of a sample size re-estimation in the Secondary Prevention of Small Subcortical Strokes study. Clin Trials 2016;13:537-44.
  • McClure LA, Szychowski JM, Benavente O, Coffey CS. Sample size re-estimation in an on-going NIH-sponsored clinical trial: the secondary prevention of small subcortical strokes experience. Contemp Clin Trials 2012;33:1088-93.
  • Coffey CS, Kairalla JA, Muller KE. Practical Methods for Bounding Type I Error Rate with an Internal Pilot Design. Commun Stat Theory Methods 2007;36.
  • Coffey CS, Muller KE. Properties of internal pilots with the univariate approach to repeated measures. Stat Med 2003;22:2469-85.
  • Coffey CS, Muller KE. Controlling test size while gaining the benefits of an internal pilot design. Biometrics 2001;57:625-31.

The CTSDMC recently completed an adaptive study, NN104, which used a modified continual reassessment method (CRM) design to determine the safety, tolerability and activity of 3K3A-APC in ischemic stroke. Results of this trial were published in the Annals of Neurology.

Our Data Management team begins reviewing the protocol as early as possible in order to brainstorm the most efficient and logical way to develop the Electronic Data Capture system. Core application modules are also used to increase efficiency in trial database development. During the process of Case Report Form development, the Data Management team often serves as interpreter between the steering committee, biostatisticians, and database developers. An effective approach in NeuroNEXT is to host an eCRF building meeting with the PI and the Clinical Coordinating Center in the very early phases of development. This approach leads to rapid finalization of CRFs, allows trial databases to be built more efficiently, and leads to faster study start-up.

The CTSDMC uses a multi-faceted strategy to ensure ongoing data quality and to improve overall efficiencies. Our approach blends traditional methods of data QC and cleaning with new applications we have developed for additional checks. More traditional approaches include real-time intra-form data logic checks and edit rule error messages programmed into our EDC, along with verification and documentation procedures that create an audit trail for corrections to database values requested by sites or monitors. In addition, the CTSDMC has implemented several core functionality systems to further promote excellent data quality and more efficient and accurate data entry.

Through our partnership with NINDS via the NeuroNEXT Network, the CTSDMC has incorporated many efficiencies into our Electronic Data Capture systems, including the use of Common Data Elements. The CDEs provide a more systematic approach to data collection in neurological disorders. This has reduced the average time from funding announcement to database release to 3.6 months within the Network.

The CTSDMC has been a member of the Clinical Data Interchange Standards Consortium (CDISC) since 2014 as part of our ongoing effort to produce the highest-quality data. 

The purpose of monitoring data in clinical studies is to ensure that the protocol is being followed at all sites, that the rights and safety of study subjects are being protected, and to confirm data integrity and quality. The CTSDMC uses a combination of on-site, remote, and centralized monitoring procedures to ensure that these objectives are met. Remote and centralized monitoring procedures are used to complement on-site monitoring and may include data management processes, reviewing reports, remote review of case report forms, remote access to electronic medical records, and review of other study documentation. The CTSDMC continues to develop remote monitoring procedures and technologies in order to stay at the forefront of this development in multi-site clinical trials. Remote monitoring techniques allow for less on-site monitoring within each study, increasing efficiency and ultimately reducing the financial burden of multi-site clinical research.

The CTSDMC has extensive experience in the statistical analysis of clinical trials. All analyses documented in a Statistical Analysis Plan and are independently verified by CTSDMC statisticians. 

Examples of other Statistical Analysis Plans can be found within the appendices of our recently completed work, published in the New England Journal of Medicine:

Since 2012, we have been the Statistics Core for the Parkinson's Progressive Markers Initiative (PPMI), which is an open source database presenting unique analysis challenges. The CTSDMC collaborates with the PPMI Steering Committee and Analysis Working Group to develop statistical analysis plans, perform statistical analyses, and write and review manuscripts and presentations. Due to the volume of variables collected in PPMI, we regularly deal with issues of multiple comparisons and have developed techniques for analyzing models with large numbers of predictors.  Additionally, we often see non-normally distributed variables, small sample sizes, and categorical variables with low counts, which preclude the most common parametric statistical methods. Our team utilizes different non-parametric methods to address these types of issues.  We document all decisions made about how to handle specific variables and data issues so that our methods are consistent across the different PPMI analyses.