athenahealth internship
During the course of a summer, I researched and designed enhancements for the internal tool that manages the clinical content on athenahealth’s EHR software. These enhancements shipped and improved the overall usability and satisfaction of the tool.
UX DESIGN
UX RESEARCH
WEB
Role
UX Designer & Researcher
Team
4 Developers, 1 Product Manager
Timeline
June 2021–August 2021 (10 weeks)
Tools
Figma, Mural, Microsoft Teams
Responsibilities
Interviews and Analysis, Research Report and Presentation, Hi-Fi Mockups, Usability Testing, Future Project Proposal
Background
athenahealth is a healthtech company that offers a suite of software tools to support clinicians in providing high-quality care. One of these tools is athenaNet, a robust electronic health record (EHR) system that houses over 170 million patient records.
Included in patient records are results from screening questionnaires. These are questions about a patient’s health used to make assessments about certain medical conditions.
The athenaNet EHR software.
Example of the Oceans Healthcare Depression Screening Questionnaire.
challenge
To manage these screening questionnaires, athenahealth uses an internal tool called the Clinical Content Builder, a form-builder and data management platform.
My job was to understand the Builder’s limitations and design improvements.
A demo of the drag and drop form builder functionality
RESEARCH
Research Methods
SECONDARY
In order to understand the Clinical Content Builder and the context surrounding it, I employed the following secondary research methods:
Reading documentation about the Clinical Content Builder (internal)
Reading documentation about athenaNet (internal)
Researching Form.IO functionality (external)
Through this inquiry, I learned:
• How there came to be a need for the Clinical Content Builder
• How its functionality supports the clinical content on athenaNet
• Why Form.IO came to be the solution for the tool
PRIMARY
In order to understand the workflow and needs of the Clinical Content Team (the main users of the Clinical Content Builder), I conducted four moderated interviews with members from differing roles.
Role: Clinical Content Manager
Main Responsibility: Managing the Content Team’s projects
Role: Clinical Content Project Analyst
Main Responsibility: Building out questionnaires
Role: Manager of Patient Safety
Main Responsibility: Performing quality assurance of questionnaires
Role: Clinical Content Associate
Main Responsibility: Researching relevant questionnaires
The Clinical Content Team members’ differing responsibilities gave me an understanding of the complete process of managing screening questionnaires.
research findings
The stages in black involve the Clinical Content Builder specifically, and are where my research findings are focused.
The Team has to roll out screening questionnaires during inconvenient hours.
Rollouts contractually have to happen between 1 AM and 6 AM. This process leaves room for human error as team members have to roll out the questionnaires, validate that they went out correctly, and problem-solve if they didn’t. This is a lot to do while sleep-deprived.
The Builder doesn’t automatically display subscores that communicate vital assessment information.
The Builder allows the Clinical Content Team to define subscores (i.e. they can name what parts of a total score correspond with different things), but it does not automatically display subscores.
The team works around this by manually listing out all the possible subscore outcomes using the component that allows for certain phrases to display if certain conditions are met. This burdens the Team with a repetitive task that wastes precious time and leaves room for human error.
The screening questionnaires are listed in no particular order.
This makes it difficult to find the ones you need. Consequentially, performing quality assurance on questionnaires can be time-consuming. Additionally, sometimes there are duplicate questionnaires on the list,which could lead to working on two different drafts at once.
List of screening questionnaires in the Builder
Subscore outcomes manually listed out
The Builder doesn’t support highly-requested questionnaires that use complex scoring methods.
The Builder doesn’t support subtraction, multiplication, division, averaging, or calculations that use multiple subscores. This means the Clinical Content Team is limited in the types of questionnaires they are able to publish onto athenaNet, which impacts their ability to meet the needs of clinicians.
The team works around this by providing a single input field that allows physicians to record a patient’s score. This workaround burdens physicians with A) having to find and access the correct questionnaire and B) having to perform sometimes complex calculations accurately.
Scoring method options in the Builder
Instructions on questionnaires can be hard to read.
The component used for writing out questionnaire instructions doesn’t allow customization. Its default formatting impacts the readability of longer instructions because there’s no way to differentiate between paragraphs.
This formatting issue risks making physicians work harder to understand given instructions, and potentially having them understand and then administer the questionnaire incorrectly, which risks patient health.
Patient Prompt Component Formatting in Builder
Manual roll out options in Builder
ROLLOUT paINPOINT
BUILD PAINPOINTs
REVIEW PAINPOINT
PROBLEM DEFINITION
My research findings indicate there’s an overall lack of efficiency throughout the process of managing screening questionnaires.
How might we improve the Clinical Content Builder so that the Clinical Content Team can build, review, and roll out screening questionnaires efficiently?
DESIGN
constraints
A couple weeks into the project, the engineering team realized there were technical constraints that would keep us from pursuing solutions to subscore displaying and scoring limitations. These scoring enhancements that couldn’t be prioritized were included in a project proposal left behind for a future team to work on.
DESIGN SOLUTIONS
BUILD Stage
Problem: Subscore Non-Display
Solution: Displaying
Adding the subscore(s) next to the total score and corresponding guidelines allows clinicians to quickly see the most pertinent assessment information at once. It’s also important to display the subscore names (as defined by the questionnaire) so there’s no confusion if there’s more than one.
Problem: Limited Scoring Calculations
Solution: Expanding
We needed to expand the types of scoring calculations able to be done in the Builder to include subtraction, multiplication, division, averaging, and subscore calculations . My proposed solution involved inputting a mathematical expression (whatever the questionnaire scoring model calls for) and allowing for previously defined subscores to be used as variables in that expression, a functionality similar to how Wolfram Alpha performs computations. This solution allows for the most mathematical flexility and considerably widens the pool of questionnaires supported by the Clinical Content Builder, helping the Clinical Content Team efficiently meet the needs of clinicians.
Prototype demo of expanded scoring calculations
Problem: Instruction Readability
Solution: Legibility and Hierarchy
The patient prompt form component allows for two levels of information: the initial instructions (primary level text) and any additional information (body copy). Enhancing readability requires implementing text legibility and visual hierarchy principles.
Paragraph spacing allows for differentiating between groups of ideas within the body copy, and generally avoiding giant blocks of text.
Bolding the primary level text allows for differentiating between the primary and secondary level of information.
REVIEW STAGE
Problem: List View Findability
Solution: Organizing
While there were many functionalities that could have been pursued, such as a search bar or defaulting the order to most recent questionnaire, our limited amount of time meant we had to prioritize features that would be most impactful. Filtering & sorting functionality would directly address the most problematic consequences of the findability issues within the list view.
Sorting names alphabetically allows for easily finding a desired questionnaire and any potential duplicates
Filtering by status allows for easily finding relevant questionnaires that are awaiting quality assurance
ROLLOUT STAGE
Problem: Inconvenient Rollout Times
Solution: Scheduling
Because there’s no way to change the time that rollouts have to happen, my proposed solution facilitates scheduling rollouts in advance. To account for changes in timelines or business needs, it also allows for rescheduling and canceling individual or multiple rollouts.
Scheduling a roll out in advance allows team members to not have to work during non-traditional hours.
Mass-updating rollouts allows the Team to address any changes that affect multiple questionnaires all at once.
Prototype demo of rollout scheduling and updating
REFLECTION
Company Impact
I exceeded the org’s goal for design work to be shipped that summer (Goal: 1, Actual: 3). I also left behind design documentation for 2 additional enhancements to the Builder. Finally, I received positive feedback from the Clinical Content Team about the shipped enhancements:
What i’ve learned
I’ve learned how to work on a cross-functional scrum team, including how to collaborate with a Product Manager on research, data synthesis and scoping. I’ve also learned how to work with a design system and research guidelines for how to use design elements in a way that parallels previously established interactions.
