BACKGROUND
A medical referral is a letter written to explain a patient’s treatment progress for the subsequent healthcare professional to continue clinical treatment. Nevertheless, clinicians writing a referral often experience cognitive burdens from reviewing electronic health records (EHRs) under a time constraint, leading to an insufficient level of completion. Especially for chronic diseases like gout, errors are inclined to occur as these patients have a long course of medical history. While literature has highlighted the potential of summarization and visualization of EHRs to assist with such administrative tasks, little is known about the design of such support systems for clinicians.
OBJECTIVE
This study aims to (1) establish understanding of clinicians’ medical referral writing practices and the use of EHRs, (2) design and develop a system prototype (Dr.Aft) to support clinicians with writing medical referrals, and (3) evaluate its usability with medical specialists.
METHODS
To acquire understanding of clinicians’ workflow with medical referral writing, we conducted a preliminary study through observations of ambulatory care sessions and contextual inquiries on clinical summarization of patient EHR data. Afterwards, three design sessions with two clinical researchers were conducted to discuss clinicians’ needs when interacting with EHRs to write a referral and iteratively test possible system features. Based on the findings from the preliminary study and design sessions, we created a system prototype (Dr.Aft) which was evaluated by ten medical specialists. Through think-aloud activities and post-use interviews after using our prototype, the results were analyzed qualitatively by the researchers.
RESULTS
Findings from the design sessions highlighted main system features of Dr.Aft including (1) referral draft generation, (2) overview of patient medical history via text summaries and visualization, and (3) grouping patient visits into important clinical events. Evaluation with clinicians showed that Dr.Aft can be a practical tool for them when writing medical referrals by facilitating their inspection of patient medical history. However, several system issues, such as clinicians’ personalized preferences, discrepancy between presented medication data and actual workflow, and suspicion of information extracted from unstructured data were discovered as well.
CONCLUSIONS
This study introduces a case study of designing an EHR-driven clinician support system to aid healthcare providers’ efficiency handling administrative work like writing a medical referral. Our findings propose design implications for similar systems in need, recommend caution in utilizing unstructured medical data, and call for system flexibility to fulfill individual preferences. Future work is required to broaden its clinical scope to more complex diseases and a diverse pool of stakeholders like patients or caregivers.