User-Centered Design and Implementation of a Clinical Deterioration Risk Prediction Tool for Providers of Cancer Outpatients
Event Type
TimeWednesday, October 12th3:30pm - 4:30pm EDT
LocationPoster Gallery
DescriptionOncology patients are at particular risk of clinical deterioration from treatment toxicities and/or medical error, especially patients who are elderly, medically underserved, of low socioeconomic status, and/or have multiple comorbidities. Despite these concerns and the rise of clinical decision support tools in healthcare settings, few systems aim to proactively protect patients from unplanned treatment events (e.g., visits to the emergency department, major changes to the treatment plan). With funding from AHRQ, we are developing a new surveillance-and-response system to protect patients from unexpected clinical deterioration. Based on FitBit, EHR, and patient-reported data, our system predicts a patient’s seven-day risk of experiencing an unplanned treatment event. Following the human-centered design process, our team developed two mock-up interface designs to communicate the predicted risk scores to the clinical team. We conducted formative usability testing with 7 oncology clinicians (i.e., physicians and nurses) to gather user feedback on the mock-ups. Each session lasted approximately 1 hour and was audio-recorded and transcribed for analysis. We conducted deductive content analysis to systematically identify recommended changes to the mockups. Clinicians recommended improvements to our description of how the system predicts deterioration to build clinicians’ trust in the model. Clinicians also suggested adding links to the patient’s notes and past encounters to support workflow. The resulting design recommendations will improve the design of the risk communication system and can be translated to the design of other tools to support patient safety in high-risk outpatient populations.