|
Creation of Public Use Files: Lessons Learned from the Comparative Effectiveness Research Public Use Files Data Pilot Project
|
| Presenter(s):
|
| Erkan Erdem, IMPAQ International LLC, eerdem@impaqint.com
|
| Sergio Prada, IMPAQ International LLC, sprada@impaqint.com
|
| Abstract:
We describe the lessons learned from the creation of Basic Stand Alone (BSA) Public Use Files (PUFs) for the Comparative Effectiveness Research Public Use Files Data Pilot Project (CER-PUF). CER-PUF is aimed at increasing access to CMS claims data sets through the creation of public use files that: do not require user fees and DUAs, have been de-identified, and provide analytic utility to researchers.
We describe the steps taken in the project to strike the right balance between data utility and privacy protection. We draw lessons learned from three tasks: (i) the creation of each PUF involving design of the sample data, analysis of variables, analysis of de-identification strategies, risk analysis, and documentation, (ii) environmental scan including stake-holder interviews, case-studies of de-identified individual level public use data, and literature review and legal analysis, and (iii) review of the needs of researchers and statistical de-identification methods that are acceptable to them.
|
|
The World is Not Flat: When to Use Relational Databases in Evaluation and Research
|
| Presenter(s):
|
| Todd Ruitman, Cobblestone Applied Research & Evaluation Inc, todd.ruitman@cobblestoneeval.com
|
| Rebecca Eddy, Cobblestone Applied Research & Evaluation Inc, rebecca.eddy@cobblestoneeval.com
|
| Namrata Mahajan, Cobblestone Applied Research & Evaluation Inc, namrata.mahajan@cobblestoneeval.com
|
| Abstract:
The problem: Some evaluators try to store their data in programs based on comfort level rather than appropriateness. We intend to tackle one aspect of the problem by sharing our organization's evolution from using flat file programs like Microsoft Excel to using a relational database (i.e., Microsoft Access) to store some of our evaluation information or data. For example, the evaluation of educational programs can require multiple levels of information that needs to be stored for quick access. District level information (e.g., address, superintendent), school level information (e.g., teacher names, standardized testing schedule), teacher level information (e.g., years teaching, email address), and student level information (e.g., assessment score, gender). Combing all of this information in one flat file is inefficient. We will discuss strategies for organizations to integrate relational databases into their data management systems and tips on choosing the right programs to store study data.
|
| |