We appreciate the interest in our work of Drs Lynch and Berse. With the increased recognition of the hereditary disposition of some cancers, there has been great interest in utilizing testing for genetic abnormalities such as BRCA mutations. We analyzed use of BRCA testing in women with ovarian cancer and women and men with breast cancer from June 2009 through July 2013.

We concur with the difficulty in identifying genetic testing in cancer patients using claims data. Our analysis focused on women with ovarian cancer and women and men with breast cancer. As stated in the article, to identify newly diagnosed patients we selected those who underwent primary cancer-directed surgery within 6 months of the first recorded claim for the cancer of interest. As MarketScan is not linked to tumor registry data, we acknowledge that we cannot identify all patients who were diagnosed with new cancers. We chose a conservative approach to attempt to select a cohort of patients who were likely recently diagnosed with cancer and thus most likely to undergo genetic testing. We searched for testing codes in the period 6 months before and after the first cancer claim. Sensitivity analyses were performed using a window of 12 months before or after diagnosis (limited to the cohort from 2010 through 2012) and the results were largely unchanged.

The second comment raised by Drs Lynch and Berse relates to the codes used for genetic testing. As stated in the manuscript, we used available Current Procedural Terminology ( CPT ) and Healthcare Common Procedure Coding System (HCPCS) codes. These codes were derived from published reports by commercial payers of coding recommendations (version updated in October 2014 used for analysis). The available CPT codes were for nonspecific molecular analysis and were used until December 2012 ( CPT 83890-83894, 83896-83898, 83900-83906, 83912). Additionally, during this time period, more specific HCPCS codes (S3818-S3823) were in use until April 2012 at which time these codes were deleted. Lastly, in 2012, CPT codes specific for BRCA analysis ( CPT 81211-81217) were introduced. Our analysis utilized all of these codes. As an example, among women in our data set with ovarian cancer diagnosed in 2009 who underwent testing, 70.5% had a HCPCS code and 42.2% had a CPT molecular analysis code (many patients had both). None of these women had one of the 2012 CPT codes. In contrast, for women diagnosed in 2013, 94.4% had coding using the updated 2012 CPT codes.

We also appreciate the comment about the difficulty in evaluating guideline-concordant care. We acknowledge in the “Comment” section that we lack data on personal and family history as well as tumor characteristics that would help guide the appropriateness of testing. The overall goal of our analysis was to descriptively report the use of BRCA testing. We acknowledge that there may be both undercapture of some testing and that patient and physician preferences that we cannot capture undoubtedly influenced care. However, it should be noted that national guidelines recommend universal BRCA testing for men with breast cancer and women with ovarian cancer. While guidelines for testing in women with breast cancer are based on risk stratification, clearly a large number of ovarian cancer patients and males with breast cancer did not received guideline-recommended care.

As interest in comparative effectiveness research increases, the implementation of well-defined, clear, and concise coding for molecular tests should be an imperative. Accurate coding will greatly facilitate both research and quality improvement initiatives.