Statistical Analysis and Reporting
With over 150 statisticians on staff -- most with advanced degrees -- we are world-renowned for the breadth and depth of our statistical expertise. We routinely use a wide variety of advanced statistical methods for data analysis. Equally important is our collaboration with study investigators on manuscripts, publications, and presentations.
Analysis Methods
We use generalized linear mixed models, hierarchical linear models, multilevel models, nonlinear mixed models, or generalized estimating equations when observations are not independent owing to repeated measurements, cluster randomization, or familial relationships. For time-to-event data such as death, we use survival techniques, including the Kaplan-Meir method, life tables, parametric survival regression, Cox proportional hazards regression, accelerated failure time models, and frailty models. Structural equation modeling (SEM) is a powerful technique that accounts for nonlinearities, latent variables, correlation between independent variables or error terms, and the modeling of interactions. Many different analysis techniques fall under the framework of SEM, including confirmatory factor analysis, path analysis, latent class and latent profile analysis, and longitudinal techniques such as latent growth curve analysis.
DSMB Safety and Efficacy Reports
Reporting to the data safety and monitoring board (DSMB) on safety and data integrity, and on the results of interim analyses for ongoing clinical trials, is a priority for us. These reports provide a summary, by site and in aggregate form, of the progress of subject recruitment and retention, data integrity, safety issues including the reporting of adverse events (AEs) and serious adverse events (SAEs), and the results of various outcome measures and interim analyses. The substance, timing, and number of interim analyses are generally set in the protocol and therefore may vary from study to study. Reports are developed in concert with the DSMB and are tailored to the specific study. These reports may have data and figures on treatment comparisons and on statistical information to facilitate sequential monitoring of the study and conditional power calculations.
Publications and Presentations
Our deep resources ensure that abstract, presentation, and manuscript preparation runs smoothly and that the results are of the highest possible quality. Our statisticians, epidemiologists, and other staff have a strong record as primary authors and coauthors on publications and presentations. We work in a multidisciplinary research culture that is conducive to joining our proficiency in statistical and methodological techniques used in study design and data analysis with study investigators' experience in the clinical or behavioral topics of interest. Together, our team analyzes data and generates abstracts and manuscripts based on study results. Our staff also includes document preparation specialists and editors who assist with professional production of manuscripts and presentation materials.
Statistical Consultation
Throughout the course of a collaborative research study, our statisticians provide statistical consultation in areas ranging from protocol planning through final data analysis and publication. Typical examples include
-
Working with investigators to frame the statistical questions
-
Identifying appropriate study design elements and statistical techniques for addressing the research questions
-
Determining any unique statistical challenges presented by the design and the data structure for data analysis and formulating solutions
-
Collaborating with the investigators to select covariates, build models, and complete data analysis using appropriate statistical methods
-
Providing results in a clear and concise manner easily interpreted by nonstatisticians
Contact us for more information
- Rick L. Williams
- Susan D. Pedrazzani