Audit Log

Characterizing the Patterns of Electronic Health Record-Integrated Secure Messaging Use - Cross-Sectional Study

EHR-integrated secure messaging is increasingly used for clinical communication, yet relatively little is known about how it is used and who its primary users are. In this study, we characterized secure messaging users and their messaging behaviors within a large health system. We found that secure messaging was widely used by a diverse range of healthcare professionals, with many users interacting with over 20 messages per day. The short message response times and high messaging volume observed highlight the interruptive nature of secure messaging, raising questions about its potentially harmful effects on clinician workflow, cognition, and errors.

Anesthesia Clinical Workload Estimated From Electronic Health Record Documentation vs Billed Relative Value Units

We compared EHR-based workload with reimbursement in anesthesiology and found that payments for anesthesia services are likely poorly calibrated with clinical workload, largely by not recognizing the physical and cognitive effort of caring for the sickest patients. This likely penalizes academic and safety net hospitals the most. Our method for measuring clinical workload from EHR audit log data could be used to measure the time and intensity of clinical work more objectively to better inform healthcare policy.

Characterizing the macrostructure of electronic health record work using raw audit logs - an unsupervised action embeddings approach

Audit logs have great potential for studying the EHR-based workflows and work habits of physicians and other healthcare professionals. However, one major barrier is the granularity of the data, which makes identification of discrete clinical tasks difficult. In this paper, we describe a novel unsupervised approach using the comparison and visualization of EHR action embeddings to learn context and structure from raw audit log activities, which can be useful for task identification and annotation.

Temporal Associations Between EHR-Derived Workload, Burnout, and Errors; a Prospective Cohort Study

Physician burnout is common and has conseqeunces for the health of physicians and their patients, yet the temporal evolution and causal contributors to burnout are not well-understood. Here, we conducted a prospective study to measure the monthly evolution of burnout in intern physicians and its association with clinical workload and wrong-patient errors. Burnout was highly correlated with recent workload; interns who worked more hours and took care of more patients had more burnout. However, burnout was suprisingly elastic; interns on lighter rotations were able to recover. We think these findings have implications for intern scheduling.

Effect of clinician attention switching on workload and wrong-patient errors

Clinical work environments are filled with competing demands for clinicians' attention, resulting in attention switching between different tasks up to 150 times an hour. The consequences of such fragmented work are not well-understood, partially because measuring attention switching has traditionally been laborious observational work. We developed a novel scalable method of measuring clinician attention switching using passively collected EHR audit log data. As a case study, we applied this method to critical care clinicians. We found that ICU work was highly fragmented, and increased attention switching was associated with decreased clinician efficiency and increased errors.

Predicting physician burnout using clinical activity logs - Model performance and lessons learned

Physician burnout is widespread, especially during the COVID-19 pandemic, and has serious consequences for the health of physicians and their patients. Burnout needs to be measured before it can be improved, but the current ways to measure burnout involve physicians filling out surveys, a request that engenders little enthusiasm. We set out to develop a model that could identify burnout in physicians from passively collected EHR log data. However, we ran into several challenges; our experiences are described in this manuscript.