Antibiogram
A doctor’s primary concern is to treat their patients and if possible, use targeted treatments. However, in emergency medicine, doctors don’t always have the time or data to refine their antibiotics choices. This results in frequent misuse and overuse of antibiotics in the emergency department setting.
How can we empower doctors in the emergency room to be more confident in choosing targeted treatments and counteract this habit of over-prescription?
Our proposed design is a web-based desktop tool that can quickly provide antibiotic options based on coverage and specificity, among other filters.
With the Antibiogram, doctors can enter their suspected syndrome. Then, they can explore most frequent pathogens and antibiotics options sorted by coverage and specificity. These lists can be further refined with filters. By selecting a specific pathogen or antibiotic, they can dive deeper into individual coverage.
The Team
We were a team of 5: myself, a project manager, back end developers, and a front end developer. Throughout the project, I was responsible for problem framing and definition, rapid concept prototyping and testing, and usability testing.
We worked under the Design Lab at UCSD, whose mission is to advance best practices in human-centered design and address complex major societal issues through research, education, and community interaction.
Initial Framing
To better understand the ER environment and the problems doctors face in their practice, I used a combination of a literature review, user interviews, and contextual inquiry.
My flowchart of one slice of the basic doctor workflow, referenced by the team for ideation. Doctors have access to patient info like current symptoms and medical history before running their clinical exams. After they perform their exam, they review their data and create a diagnosis and treatment plan.
To help focus my team’s meetings and ground ideation sessions, I visualized the doctor’s journey with an individual patient and synthesized my findings into some key problems and goals:
Current antibiotic tools are impractical in the ER because they require often unavailable knowledge of specific bacteria. Lab testing to get this data can take up to 3 days. As a result, doctors are pushed towards more common, broader treatments. How might we help doctors get access to antibiotic information they can apply in their practice?
Specialists see patterns between pathogen occurrence and antibiotic coverage and known factors like syndromes, demographics, and medical histories. How might we communicate these patterns to ER doctors?
In reality, doctors are going through this workflow with multiple patients at a time so they’re jumping between tasks every 5-10 minutes at most. How might we encourage doctors to quickly and safely tend towards targeted treatments?
Computers are the most used in and outside the patient room. Designs must support a desktop environment.
Concept Prototyping
After a comparing other medical software and a brainstorming session, we focused on a form-style concept where doctors would fill out patient data and receive antibiotic recommendations based on the entered information. We felt that basing the tool on available patient data would be more relevant to clinical practice and allow for more targeted suggestions. We narrowed down on the form entries to take into account time constraints.
Early testing of the form-style concept. Doctors fill out patient demographics, medical history, symptoms, and syndrome information. Then, they receive the most likely pathogens and suggested treatments, either in separate tabs or a one-page display.
Organizing and narrowing down the form entries with post-it notes to optimize time efficiency in the tool.
I drafted a paper prototype to test with users and discuss with our team’s medical advisers. The main takeaways:
Concern over the time and work it takes to get to a recommendation.
Conflict between the tool and doctor’s mental model. They tend to start broad and filter when relevant; our form-style concept was the opposite order.
Instead of a specific recommendation, doctors needed access to more applicable data to make the better decisions themselves.
Balsamiq prototype with changes from concept testing feedback. Flow allows for quick scan through one page of options just based on diagnosis. Filters are available afterwards if doctors find them relevant.
Usability Testing
I ran usability testing sessions with a functional prototype built by my team to be able to include real data. To evaluate how well the design fits into their practice, I supplied a few example patient cases for them to evaluate:
Doctors were able to explore within seconds of use. This workflow was very well received.
Most doctors were able to choose treatments that matched the actual prescriptions used for the example cases.
Some essential filters need to be added for more medical accuracy.
Reflection
Problem framing is essential to being able to reach the right solutions. But if you don’t nail it the first time, you need the flexibility to be able to circle back and realign. This is what allowed us to rethink our design early on and create a tool that fit our users’ need for greater confidence, time efficiency, and accuracy.
I also think that we compared with other tools too early in the process, and that hindered our creativity during ideation and brainstorming. In the future, I want to be more conscientious and strategic about comparative analysis in my process.