Uptake of Clinical Decision Support in the Electronic Health Record Varies by Physician Attitudes and Characteristics

Background: Clinical decision support (CDS) is an effective tool for increasing guideline-concordant care and is included in the criteria for achieving Meaningful Use of electronic health records (EHR). The PECARN prediction rules for emergency department (ED) management of pediatric head trauma offer risk estimates useful for decisions about use of cranial computed tomography (CT). Children seen in the ED for head trauma are particularly vulnerable to radiation exposure during imaging, and CDS could be a useful tool to translate knowledge into practice, thereby reducing unnecessary imaging. Objectives: To examine emergency physician (EP) attitudes toward clinical prediction rules (CPRs) and their usefulness, and to study whether incorporating the PECARN prediction rules into the EHR would change clinical practice. Methods: We electronically surveyed EPs within an integrated delivery system with an integrated EHR. Respondents were randomly assigned one of two self-administered questionnaires, both based on a hypothetical clinical vignette of a child with head trauma in which a CT is not recommended by the PECARN rules. Both versions asked for EP's most likely evaluation prior to any description of CPR; however, one version included CDS (a list of the PECARN criteria) and the other did not (NCDS). The remainder of the questionnaire was the same in both versions and assessed provider characteristics and knowledge of and attitudes toward CPRs and their incorporation into the EHR. Results: The survey response rate was 339 (68%) with 51% responding to the CDS version. Initially, 25% of respondents on either version indicated they would order a head CT (between-versions, p=0.43). After decision support in the CDS version, 70% of respondents changed their management decision to “no imaging”, versus 5% in the NCDS version (p=0.003). In regression analyses, we found that EPs were less likely to change their decision based on CDS if they practiced for >15 years (p= 0.03) and agreed with the opinion “CPRs are too difficult to use” (p=0.04). Conclusions: A majority of EPs who originally would have ordered a CT changed their clinical judgment after receiving decision support via the EHR. Negative attitudes toward CPR usefulness in general and greater years in practice decreased the likelihood of CPR uptake. This variation in provider perspective needs to be carefully addressed in settings where decision support is being incorporated into EHRs. Targeted electronic CDS provides an opportunity to safely decrease imaging utilization according to clinical guidelines, a goal of particular interest in vulnerable populations.