I had the pleasure of interviewing Lisa Shieh, a Hospitalist at Stanford University, about her team’s recent publication in JHM Why providers transfuse blood products outside recommended guidelines in spite of integrated electronic best practice alerts. The study found that many best practice alerts (BPAs) “overrides” were due to protocolized behaviors, resident/nurse practitioner/physician assistant ordering, or subjective indications, such as “symptomatic anemia”.
What is your background and how did you become interested in using BPAs to enhance quality?
I am a hospitalist and Medical Director of Quality for the Department of Medicine at Stanford University Medical Center, and have a longstanding interest in how the appropriate use of BPAs can enhance quality efforts. Our medical center has been on Epic EMR for many years, and over time we continually strive to optimize our EMR clinical decision support to try to achieve the right amount of support, without invoking alert fatigue. In this particular case, we were interested in figuring out why so many providers were transfusing patients who had hemoglobin levels above an evidence based threshold (e.g. >8mg/dL or >7mg/dL depending on the problem list conditions).
How should hospitalists be using BPAs to advance quality in hospitals?
Introducing BPAs has to be done very carefully, balancing the impact of the intervention with the potential for inappropriate amounts of “noise”. It should also incorporate as much information as possible from the medical record, to ensure it is only firing on the appropriate patients (e.g. those without active bleeding).
Were there any unexpected findings in your study?
We were a little surprised to see that nurses or NP’s were ordering blood products (either through attending orders or protocols), and that so many residents actually explained why they were transfusing above appropriate thresholds (through the free text comment in “other”). Since our blood utilization significantly improved despite providers “overriding” the alert, we do think clinical decision support (CDS) can have a “later” impact on provider behavior. It may not always change the behavior during that ordering episode, but may impact future ordering behavior as the provider learns from the CDS.
Where does this take you with respect to future research efforts?
We feel strongly that BPAs can be a very powerful vehicle to enhance evidence-based behavior and therefore higher quality care by intervening at the point of care. For future studies, we are very interested in using more robust QI research methods to prove the benefit of CDS, such as using randomization methods.
In conclusion, all hospitalists should be attuned to how CDS can improve evidence-based clinical practice, but it needs to be carefully implemented and monitored to be effective. Future research should focus on how the clinician-CDS interface can result in the most effective workflow.