PCN349 Data linkage between tumor registry and electronic medical records for patient journey study in lung cancer
Zhang Y, Simoff M, Ost D, Wagner O, Lavin J, Nauman E, Hsieh MC, Wu XC, and Shi L. PCN349 Data linkage between tumor registry and electronic medical records for patient journey study in lung cancer. Value in Health 2020; 23:S86.
Value in Health
Objective: Lung cancer is the leading cause of cancer death in the US, with most patients diagnosed at advanced stages. Patient pathways prior to cancer diagnosis remains unclear. This study linked tumor registry and electronic medical records data to analyze varying clinical pathways to lung cancer diagnosis. Methods: Research Action for Health Network (REACHnet) and Louisiana Tumor Registry (LTR) data from 2013 to 2017 were linked using a Privacy Preserving Record Linkage (PPRL) algorithm. REACHnet provides longitudinal clinical data extracted from electronic health records, including Ochsner Health system and Tulane Medical Center. LTR, a statewide population-based registry, collects information of demographics, tumor characteristics, cancer stage at diagnosis and initial treatment of cancer cases. ICD-9 and ICD-10 diagnosis codes were used to define patients with primary lung cancer from REACHnet. ICD-O-3 codes were used to select patients with primary lung cancer from LTR. Patients with solitary pulmonary nodule (SPN), low-dose computer tomography (LDCT) or chest computer tomography (CT) identified between July 2013 and December 2017 diagnosed with primary lung cancer were included in REACHnet sample. Patients excluded were 1) SPN with pleural biopsy as first biopsy, 2) lung cancer as secondary diagnosis, 3) non-LA residents, 4) diagnosed by Death Certificate and 5) diagnosis on autopsy. Results: From 2013 to 2017, 2,860 patients with primary lung cancer were identified in LTR. Among 30,561 patients obtained from REACHnet, 2,929 (9.58%) patients had a recorded diagnosis of primary lung cancer. Furthermore, 1,964 (67.05%) of 2,929 received primary lung cancer diagnosis by linkage to LTR records using the PPRL algorithm. The linked database had lung cancer characteristics (e.g., stages), and health resource utilization prior to diagnosis (e.g., patterns of clinical work-up). Conclusion: Data linkage between tumor registry and electronic medical records allows researchers to investigate clinical pathways and health outcomes in lung cancer.