Not a One Size Fits All

Methods

Medical studies can be complicated, often requiring out-of-the-box thinking and techniques. Based on the research questions and data for the study, our biostatistical and epidemiological teams properly design studies and analyze the data. For the past two decades, CDRG has worked with Medicare Research Identifiable Files, including the 20% sample and specific 100% cohort samples. Our understanding of and proficiency with the Medicare data allow us to move beyond basic methods into advanced modeling.

Strategy

In surveillance research, some of the methods we use is employing Bayesian models to produce stable event rate estimates for small groups/regions. In trend analysis, we also use penalized spline generalized additive models to assess the “event” trend over time. Additionally, in provider profiling, we use random effects modeling and Bayesian statistics and techniques to stabilize estimates for small providers. In post-marketing research, we use nested case-control studies to evaluate the safety and benefit of the study drug, and we use the difference-in-difference method to evaluate the effect of policy changes. We also use two-part modeling to compare medical costs and healthcare resource utilization and we use random forest and LASSO regression for variable selection in prediction models. We also use simulation to evaluate which statistical methods we should choose, based on the unique case we're working with.

Additionally, comparative effectiveness research (CER) using observational data is a major aspect of our studies. An important task for CER with observational data is to adjust for biases from multiple sources. We apply state-of-the-art methodologies, such as propensity score methods, marginal structural models, doubly robust methods, and instrumental variable analysis, to best correct biases as necessary.

Since our strategies are tailored to fit your specific project, these are only some of the methods we employ. We tap into additional techniques as needed to fit your specific project needs.

Meeting Expectations 

From start to finish, we strive to exceed expectations, and communication is the key component of our successful collaborations. Whatever the size of the project, we will work to determine your needs and the most useful method of data presentation to meet your objectives. 

Project deliverables vary depending on the research question and goals, but typically numeric results are presented in Excel files that include a table of contents tab for ease of navigation.

Reports and Presentations

Some projects require a simple summary, others a full-scale report. Summaries are part of the Excel worksheet. They include a methods tab for quick reference to information about how the work was executed, and a definitions tab documenting definitions and algorithms used in the project. Complete project reports are delivered in Word format. They include background, methods, results links to the Excel worksheet, conclusions, limitations, and suggested next steps. 

Slides of the project for presentation to internal sponsor groups are also routinely created, catering to your specific audience.

Publications 

If warranted, we will work with you to publish the work in a peer-reviewed medical journal. As one of our goals is to improve knowledge, our contracts include wording allowing us to publish the research studies, given an acceptable amount of review time by the sponsor, on our own.

Browse a list of our published work/medical articles:

CDRG Publications

Have questions about how we can help you with your next project?

Contact Us