Physician ordering patterns for routine mammogram screening in response to an electronic health record alert
- Author(s): GOWDA, INDIRA
- Advisor(s): Bell, Douglas
- et al.
Objective: Evaluate provider compliance with guidelines for routine screening of breast cancer before and after the implementation of a targeted “Best Practice Advisory” (BPA) alert, titled CareGaps, embedded in the routine health maintenance module in the UCLA electronic health record (EHR).
Design: The CareGaps BPA was deployed June 2019 in UCLA primary care clinics. We performed a difference-in-differences evaluation using a mixed effects logistic regression model. We compared mammogram order rates between our pre-intervention period and intervention period. These timeframes were broken into four study periods: Jan-May 2018, Jun – Oct 2018, Jan – May 2019 which were all included in our pre-intervention timeframe and Jun – Oct 2019 which was our intervention timeframe. Mammogram orders placed within two weeks of the encounter visit were considered related to the recent encounter and were considered for analysis. Provider characteristics such as specialty and years of practice and patient characteristics such as insurance and race were controlled for in the model.
Primary outcome: Encounter-based mammogram ordering rates were compared between the pre-intervention and intervention groups.
Secondary outcomes: We examined completion rates for breast cancer screening using the health maintenance status of “not due” as a proxy measure. We also examined the CareGaps orderset use between providers and looked at the ordering practices for mammography amongst providers exposed to the BPA.
The total number of encounters steadily increased over time and ranged from
26,232 to 32,257. Overall the likelihood of placing a mammogram order significantly increased by about 6% between our pre-intervention and intervention groups. (adjusted odds ratio = 1.07, p = 0.036). We did not find a significant impact of the BPA in mammogram completion rates. We did find however that use of the CareGaps orderset increased with deployment of the CareGaps BPA (OR 1.71, CI 1.43 – 2.07, p <.0001). We also found that those providers who were exposed to the BPA and chose to “address” the BPA by interacting with it such as acknowledging it or opening the CareGaps orderset were 25% more likely to place a mammogram order (X2 p<.0001).
Conclusions: Implementation of a targeted BPA for addressing health maintenance topics such as breast cancer screening can be effective for alerting those providers who have not previously addressed these topics with patients.