Using HMORN's Virtual Data Warehouse From Two Health Systems to Identify Risk Factors for Abdominal Aortic Aneurysm
Background/Aims: Abdominal aortic aneurysm (AAA) is a leading cause of death in the United States, often undetected until rupture. At least 33% of ruptured AAA hospitalizations are among women, and 22% of AAA-related deaths occur in nonsmokers, individuals not covered by current screening guidelines...
Gespeichert in:
Veröffentlicht in: | Journal of Patient-Centered Research and Reviews 2015-04, Vol.2 (2), p.139-140 |
---|---|
Hauptverfasser: | , , , , , , , , , , , |
Format: | Artikel |
Sprache: | eng |
Schlagworte: | |
Online-Zugang: | Volltext |
Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
Zusammenfassung: | Background/Aims: Abdominal aortic aneurysm (AAA) is a leading cause of death in the United States, often undetected until rupture. At least 33% of ruptured AAA hospitalizations are among women, and 22% of AAA-related deaths occur in nonsmokers, individuals not covered by current screening guidelines. Identifying additional risk factors for AAA would allow for targeted screening of a larger at-risk population. We conducted a retrospective study of electronic health record and claims data from the HMO Research Networks (HMORN) virtual data warehouse (VDW) in two sites: Geisinger Health System (GHS) and Essentia Institute of Rural Health (EIRH).
Methods: We used an algorithm that includes CPT and ICD9 codes to classify individuals in the VDW as cases, controls or excludes. This algorithm had a positive predictive value of 94% and sensitivity of 100%. We extracted demographic, behavioral and clinical covariates, including comorbidities such as peripheral artery disease, diabetes, neoplasms, and pulmonary, kidney and cerebrovascular diseases. Individuals were excluded based on age, no visit in 5 years, genetic conditions or unspecified aneurysm site.
Results: We identified 2,133 AAA cases and 130,289 controls from GHS and 1,986 cases and 196,534 controls from EIRH. Risk factors were similar in direction and magnitude of effect and level of significance across sites, including the novel association of benign neoplasms with AAA.
Discussion: We leveraged the VDW to efficiently demonstrate the transportability of an algorithm for identifying AAA patients, expanding the sample size for studying AAA risk, and replicating our risk factor findings in a second institution. This work was funded in part by National Human Genome Research Institute as both sites are members of the Electronic Medical Records and Genomics (eMERGE) Network (U01HG006382 to Geisinger Clinic and U01HG006389 to Essentia Institute of Rural Health). |
---|---|
ISSN: | 2330-0698 2330-0698 |
DOI: | 10.17294/2330-0698.1193 |