How to link patient care databases from multiple sources?

Challenge

Researchers seeking to conduct patient-centered outcomes research (PCOR) and clinical effectiveness research (CER) have used a variety of data sources to further their work. Depending on the research question being examined, data resources have included administrative data, such as claims and electronic health records (EHRs), primary data collection from randomized controlled trials, independent surveys, and related data from institutions that researchers may be affiliated with.

These data are very difficult for the typical PCOR researcher to find and access; it is typically difficult and sometimes impossible to combine them to obtain a comprehensive view of patient care. Notable challenges include a lack of information about medical practitioners and health care organizations, missing data on social determinants of health, inability to link geographically, and restrictions in accessing the data.

Solutions

The Agency for Healthcare Research and Quality (AHRQ) seeks to build upon its considerable portfolio of data assets for researchers, which currently includes expansive state-level claims data and novel work in the construction of national all-payer claims databases and provider databases. The goal of this effort is to construct innovative health care databases for 3 states that are representative at the state level and combine multiple sources of individual-level administrative data, including state all-payer claims databases, potentially EHRs, a comprehensive collection of provider-level data, and an expansive set of community-level social determinants of health data.

The outcome will be the creation of de-identified databases for the chosen states that provide a panoramic view of patient care across care settings and are representative of health care insurance populations at the state level. The databases will be made available to researchers for PCOR and CER.

To create these prototype databases, Westat is teaming with partner Pantheon and providing critical technical and research expertise to all aspects of the database design and construction. Westat will provide all statistical expertise needed for the complex database creation tasks, which include record linkage (probabilistic and deterministic), sampling, weighting, confidentiality testing, and de-identification.

Westat will also provide policy and research expertise by working with AHRQ to decide which data sources to use, how to harmonize variables across data sources and over time, and how to best modify the databases to maintain confidentiality while maximizing the usability of the databases for PCOR and CER.

Westat will also apply its research expertise to conduct a case study using the new databases for a peer-reviewed journal.

Results

The Panoramic View of Patient Care Through Data Innovations and Linkages databases will provide innovative products for researchers to examine all facets of health care across payers and populations. Beyond that, these prototypes will serve as the groundwork for how databases can be constructed that link all-payer claims databases with other sources of individual-level administrative data, provider-level data, and community-level data to provide a panoramic view of patient care.

The databases created through this effort will catapult what is possible in patient-centered outcomes and cost-effectiveness research. The methodologies developed and knowledge gained through the acquisition of data and construction of the databases will set the stage for an array of new comprehensive research databases.