Exploration of PCORnet Data Resources for Assessing Use of Molecular-Guided Cancer Treatment.

PURPOSE Examine the ability of PCORnet data resources to investigate molecular-guided cancer treatment. PATIENTS AND METHODS Patients (N = 86,154) had single primary solid tumors (diagnosed 2013-2017) from hospital oncology registries linked to the PCORnet Common Data Model (CDM) at 11 medical institutions. Molecular and anatomic test procedures and oral and infused therapies were identified with Current Procedural Terminology (CPT) and Healthcare Common Procedure Coding System (HCPCS) codes, RxNorm Concept Unique Identifier, and National Drug Codes from CDM tables. Chart review (2 institutions, n = 213) for advanced colorectal cancer and Medicare claims linkages (7 institutions, n = 1,731) for breast cancer explored options for increasing electronic data capture. RESULTS Molecular testing prevalence detected via analyte-specific molecular CPT/HCPCS codes was 5.5% (n = 4,784); for the nonspecific anatomic pathology codes, for which only some testing is performed to guide therapy selection, it was an additional 44.8% (n = 38,610). Molecular-guided therapy prevalence was 5% (n = 4,289). Testing and treatment were most common with stage IV disease and varied across cancer types and study institutions (testing, 0%-10.4%; treatment, 0.8%-8.4%). Therapy-concordant test results were found in charts for all 36 treated patients with colorectal cancer at the 2 institutions, 3 (8.3%) of whom received treatment outside the institution. Breast cancer Medicare claims linkage increased rates of identified testing from 62.7%-98.9% and treatment from 3.9%-8.2%. CONCLUSION Although a minority of patients received molecular-guided therapies, the majority had testing that could guide cancer treatment. Claims data extended electronic data capture for therapies and test orders but often was uninformative for types of test ordered. Test results continue to require text data curation from narrative pathology reports.