E. Toy1, C. Y. Peterson1, K. A. Ludwig1, T. J. Ridolfi1 1Medical College Of Wisconsin,Department Of Surgery – Division Of Colorectal Surgery,Milwaukee, WI, USA
Introduction: Designing high-quality clinical studies is often limited by poor understanding of study cohorts and populations, information that is difficult to obtain easily and efficiently, thus leading to poor recruitments and under-powered studies. The Informatics for Integrating Biology and the Bedside (i2b2) Cohort Discovery Tool provides an easy to use, self-service way for researchers to query an externally maintained database that draws information from the electronic medical record, stored in the i2b2 Clinical Translational Research Informatics Data Warehouse (CTRI-CDW). Search criteria can include nearly any piece of information in the medical record from demographics, diagnoses and procedure codes, laboratory test results, pharmacy orders and dispenses, as well as free text within clinical documents. In response to a query, the tool can return an approximate number of patients matching the search criteria without revealing identifying information. Searches can be stored online and de-identified information is released after IRB approval. Over eighty academic medical centers nationally have the i2b2 Cohort Discovery Tool integrated with the electronic medical record. The aim of the current project was to assess the performance of this tool when compared to a manually curated internal colorectal surgery database.
Methods: Both the manually curated internal database and the i2b2 Cohort Discovery Tool were queried by procedure name and/or corresponding for procedures occurring from February 2008 to April 2014 performed by a single surgeon. Procedures and CPT codes were grouped into three categories: 1. Low anterior resection ( CPT codes 44145, 44207, 44146, 44208), 2. Abdominal perineal resection (CPT codes 45110, 459395) and 3. Segmental colectomy (CPT codes 44140, 44141, 44143, 44144, 44160, 44204, 44206, 44205). Number of patients identified within each cohort were then compared.
Results: For low anterior resection, the internal database identified 83 patients while the i2b2 Cohort Discovery Tool identified 155. For abdominal perineal resections the internal database identified 28 patients while the i2b2 cohort Discovery Tool identified 97. For segmental colectomy, the internal database identified 176 patients while the i2b2 Cohort Discovery Tool identified 662 patients.
Conclusion: The i2b2 Cohort Discovery Tool outperformed the manually curated internal database on all queries, in part due to an early empahsis on neoplasms within the internally maintained database. The i2b2 Cohort Discovery tool has the potential to replace prospectively maintained departmental databases and has the potential to foster large multi-institutional studies as it is available in over eighty academic medical centers nationwide.