Article Title

Data Mining via the Electronic Medical Record to Identify Inherited Colorectal Cancer in an HMO Patient Population

Publication Date



family history colorectal cancer, data mining electronic medical record


Background/Aims: A family history of colorectal cancer (CRC) and/or endometrial cancer, especially under age 50, is suggestive of Lynch syndrome, the most common form of inherited colorectal cancer. Limited family history (FH) data are currently collected on our members by nongenetics providers and recorded in the FH tab of our electronic medical record (EMR). We used these data as a potential “screening tool” for inherited CRC.

Methods: We queried our EMR (Epic, Verona, WI) system to identify individuals meeting the following criteria: 1) age 25–50 years, 2) no personal history of CRC, 3) FH of at least two relatives with CRC, and 4) accessed our electronic patient portal at least once in the last year. We identified 444 members who met all criteria and had not already had a genetics consultation. All received a secure email requesting additional personal and FH via a secure electronic questionnaire (created by using HP Process Automation). Survey respondents provided information on: a) known mutation in a cancer susceptibility gene in the family, b) FH of colon, endometrial and other cancers, and c) personal/FH of polyps. Based on their responses, members were informed that: a) genetics would schedule a consultation to discuss their FH, or b) they do not appear to be in increased risk for a cancer syndrome.

Results: At present 106 (24%) of those contacted completed the survey, and 96 (91%) of them meet our moderate/high-risk criteria. Over half have been scheduled/seen and genetic testing is in progress for some. Males and Hispanic/Latino members were underrepresented in both the original cohort and in the group of respondents. Non-Hispanic Caucasians were overrepresented in both groups.

Discussion: The underrepresentation of males and Hispanics in both the original cohort and the group of respondents suggests that gender and racial/ethnic background play a role in the likelihood of the patient reporting a FH of cancer. The high proportion of high-/moderate-risk responses suggests that those with more suspicious histories may be more motivated to complete the survey. Our results will help us determine if mining limited FH data in the EMR is an effective method for identifying inherited CRC syndromes




March 27th, 2015


April 28th, 2015