Online Program

Analysis of Electronic Health Records to Identify Contextual Factors Associated with Breast Cancer Screening Patterns (308063)

The policies and recommendations for mammography screening are riddled with confusion as several nationally recognized agencies have variation in their screening guideline recommendations. The most demanding recommendation states that women between the ages of 50 and 74 should receive a mammography screen on an annual basis while more liberal recommendations may defer to the primary care physician’s judgment. While regular mammographies may result in earlier detection of breast cancer thus allowing more effective treatment, there is also considerable risk of false positive results, if mammography is performed too frequently. Gaining a deeper understanding of factors that may influence a woman’s personal screening pattern is crucial. One way to elucidate such patterns is the secondary analysis of electronic health records (EHRs). EHRs from a large multispecialty healthcare system (1999-2013) were used in the development of a retrospective mammography screening cohort with 155,394 patients having 2+ mammograms during the follow-up period. EHR data was obtained corresponding to the completion of all screening mammograms as well as patient and provider characteristics. We developed an algorithm classifying a woman’s screening patterns across time (e.g. faithful pattern throughout follow-up, switching between annual and biennial, etc.). This classification enabled us to employ hierarchical polychotomous logistic regressions to quantify the effect of various contextual factors on the women’s screening patterns after adjusting for provider factors, race/ethnicity, calendar year, birth cohort, and/or patient-provider gender and language concordance status. Understanding the effect of a variety of contextual factors may influence a woman’s breast cancer screening pattern may enable more effective primary care delivery and improved early detection of cancer.