Study concept and design is one of the most important steps in collaborative research. In collaboration with investigators across the United States and throughout the world, our researchers have designed hundreds of studies. We turn the clinical or behavioral investigator's concept into a clinical trial or observational study with appropriate hypotheses, design, measurements, logistical operations, and analysis plans.
Defining the primary and secondary hypotheses is the important first step that will guide the study design. The hypotheses should be clearly stated in terms of measurable quantities that clearly relate to the study concept. Our researchers collaborate with study investigators to translate the initially stated hypotheses into statements that can be quantified and tested to guide the study design.
Selection of Study Outcome Measures
Outcome measures should relate directly to the hypotheses, be measured with little bias and low variability, and be collected using appropriate methods. We collaborate with study investigators to ensure that the outcome measures quantify the study hypotheses, balance measurement cost with accuracy, and are temporally collected to ensure the measurement of correlation or causation. Our vast experience includes a wide array of clinical, imaging, and psychometric measurements.
Statistical Design, Power Calculation, and Randomization
We collaborate with study investigators to determine the statistical design (for example, random assignment or selection) that best supports the hypotheses to be evaluated. We conduct power calculations to compare competing designs to select a sample size allocation that yields adequate power without wasting valuable resources. Our experience in randomized trials includes factorial designs to simultaneously test two or more interventions, adaptive designs to more rapidly conclude studies, and crossover designs in which patients can serve as their own controls to increase power. We are also experienced in observational and epidemiological studies, such as case-control studies for rare events, population studies, and surveys.
In collaboration with study investigators, we prepare detailed study protocols to guide the uniform implementation of the study across the sites. We typically author the statistical sections of the protocol such as sample size justification, randomization or subject selection, interim monitoring, and statistical analysis plans. We provide administrative coordination and support for organizing, editing, and distributing working drafts of the protocol to the appropriate parties. Upon final review and approval, RTI issues the final protocol and/or revisions to the participating sites and maintains the final protocol for the convenience of the study investigators.
Study protocols include plans developed by the study investigators and our statisticians for the analysis of data. Analysis plans take into account the study hypotheses, types of measurements to be analyzed, and study design. The analysis approach depends on both the type of outcome (for example, continuous, categorical, or time-to-event) and the study design (for example, factorial, crossover, case-control, or population survey). The plan also accounts for the choice of randomization method, any stratifying or clustering, repeated measurements, and potential missing data issues.