LAUSR

PURPOSE Genomic testing improves outcomes for many at-risk individuals and patients with cancer; however, little is known about how genomic testing for non-small-cell lung cancer (NSCLC) and colorectal cancer (CRC) is used in clinical practice. PATIENTS AND METHODS In 2012 to 2013, we surveyed medical oncologists who care for patients in diverse practice and health care settings across the United States about their use of guideline- and non-guideline-endorsed genetic tests. Multivariable regression models identified factors that are associated with greater test use. RESULTS Of oncologists, 337 completed the survey (participation rate, 53%). Oncologists reported higher use of guideline-endorsed tests (eg, KRAS for CRC; EGFR for NSCLC) than non-guideline-endorsed tests (eg, Onco typeDX Colon; ERCC1 for NSCLC). Many oncologists reported having no patients with CRC who had mismatch repair and/or microsatellite instability (24%) or germline Lynch syndrome (32%) testing, and no patients with NSCLC who had ALK testing (11%). Of oncologists, 32% reported that five or fewer patients had KRAS and EGFR testing for CRC and NSCLC, respectively. Oncologists, rather than pathologists or surgeons, ordered the vast majority of tests. In multivariable analyses, fewer patients in nonprofit integrated health care delivery systems underwent testing than did patients in hospital or office-based single-specialty group settings (all P < .05). High patient volume and patient requests (CRC only) were also associated with higher test use (all P < .05). CONCLUSION Genomic test use for CRC and NSCLC varies by test and practice characteristics. Research in specific clinical contexts is needed to determine whether the observed variation reflects appropriate or inappropriate care. One potential way to reduce unwanted variation would be to offer widespread reflexive testing by pathology for guideline-endorsed predictive somatic tests.