JANUARY 20, 2017 BY MOLLY SKUBAK
Data Paths: An Update from the PaTH Informatics Team
The Geisinger Health System joined PaTH as a new site just over a year ago, bringing a wealth of experience with using "real-world" health data in research to improve health care and health outcomes. At Geisinger Health System, the clinical informatics efforts are led by site Principal Investigator H. Lester Kirchner, PhD, and the Phenomic Analytics and Clinical Data Core team, or Core, in the Department of Biomedical and Translational Informatics. The Core team works with various data structures to facilitate research studies on diverse topics from analyses of how health care is delivered to using genomic data to develop treatments for people with genetic disorders.
The PaTH Common Data Model is only one of the numerous data models that the Geisinger team manages, including:
Two of these data models, the PCORnet Common Data Model and HCSRN Virtual Data Warehouse, incorporate data from electronic health records (that is, patient charts), billing records, and health insurance claims records. These data sources are carefully combined in an “Extract, Transform, and Load” process, in which data are extracted from their original data sets, converted to a format that will fit with other data, and loaded into a new secure database. This is a complicated process, but the Geisinger team has impressively combined these claims and EHR data from multiple systems to create unified encounters and avoid duplicate records. This critical process needs to be clearly conveyed to researchers in order to ensure that they carry out accurate analyses using these data – and the Core team has invested a lot of effort into making sure that information is conveyed clearly.
All of this work by the informatics team at Geisinger to combine claims and health record data pays off when it comes to asking patient-centered research questions. While both data sources tell about diagnoses, they carry complementary information. For example, from the health record data (but not claims data), researchers can understand how blood pressure and body weight change over time, and what tests and medications are ordered by a patient’s health care team. Claims data (but not health record data), can provide insight into what tests were actually carried out and what medications were filled. Both data types convey information about patients’ diagnoses. The ability to analyze claims and electronic health record data together greatly increases the types of questions that can be answered using real-world health data from the Geisinger Health System.«—- Back To News