Designing Digital Order Sets for a Form Building Application
What is this project about?
A new strategy to speed up Form Building and create usable, digital Order Sets.
(November 2015 to present)
Think Research develops clinical content and builds Order Sets that clinicians use on the floor to treat patients. Order Sets are documents that contain groups of orders that apply to a specific diagnosis.
Clinicians have traditionally written down orders for patients using paper and pen in an unstructured format. That often resulted in errors that could pose a risk to patient safety. Those carrying out the orders not only had to find the information they needed within a written document of orders, but also had to decipher the writing of clinicians.
In order to improve this workflow, Think Research developed Order Sets, which are highly structured documents that allow clinicians to digitally document orders. Think Research, together with clients, built Order Sets in Microsoft Word and then converted those documents into fillable Order Sets.
The Overarching Problem
The technology Think Research used to convert Word Documents into fillable Order Sets involved a long and arduous ‘scoring’ process. An internally developed application, Databridge, automated the conversion process to a certain degree. Document Developers had to run through hundreds of steps for each form and ensure formatting was perfect.
In addition, once a document was scored - if there were any errors, the Document Developers had to go back to the original Word document, fix the issues and start the scoring process again. Often, the scoring process took hours, and sometimes was left to run overnight.
This caused a bottleneck if clients requested a change to one of the Order Sets they were using on the floor. As Think Research onboarded more clients, each with a library of hundreds of Order Set documents, this solution was no longer scalable.
This project involved a number of internal stakeholders to make everything work. The team working on this problem included the CTO, the VP of Engineering, a Senior UX Designer, and an engineering team of 4, and myself. In addition, the President and Founder, an ICU physician who pioneered Order Sets was involved, as well as stakeholders such as Document Developers, and the Research and Development team.
Our teams knew that we still had to make use of Microsoft Word, as this was the best way to collaborate with clients to build Order Sets that fit their organization’s needs. The main goal of the new strategy was to reduce the time it takes to create or modify an Order Set.
In order to do this, the team needed to re-think the entire workflow. We not only needed to support the development of documents, but also manage versions of those documents and deploy them to sandbox and production environments in order to collaborate with clients and publish completed digital Order Sets to our current platform for client use. In addition, the company wanted to improve the gathering of usage data from these Order Sets to inform the need for content updates, and help clients understand the practices of physicians using Order Sets.
A Senior UX team member, Anna Kop, spent time shadowing our Research and Development and Document Development teams to understand their workflows and tools used to bring an Order Set to life.
- Microsoft Word
- Standard software used for document creation. Documents are distributed to clients to build Order Sets that meet their organization’s needs
- Adobe Acrobat
- Word documents converted to Acrobat to be ingested in internal scoring tool, Databridge
- Databridge (Internal tech used to score documents - manual and automated processes)
- Document Developers manually add fillable fields to the PDF which will be converted to a fillable Order Set
- Once the fillable Order Set is created in Databridge, automated scoring occurs to create the final, fillable Order Set
- TxConnect 1 & 2 (Internal tech used to store documents in a library)
- Research and Development teams also use a Tx2 to upload and store developed Standard Reference Order Sets. These act as a best practices reference library for clients
- Tx 1 is a document repository of working and completed Order Sets used by Think and clients. Clients can download Order Sets from the Reference Library to help build their own Order Sets or Upload/Download their own Order Sets to modify or use during digital platform downtime)
Databridge - Creating fillable fields
Interviews were conducted with document developers and the internal team to understand pain points and opportunities.
Main pain points for Document Developers
- Creating Order Sets is a long, time consuming process
- Many steps involved across several different document types
- Steps often need to be repeated from scratch if a mistake is made
- Documents are complex and often many pages long
- No way quick way to make changes to an order set after it has been scored
- Any time a change is made to an order set, it needs to be re-scored in Databridge – process is unnecessarily time-consuming
- Client may only notice a change is required after the order set has been completed, scored and deployed to Sandbox for the client to test it out
- No version control
- Versions and dates are added manually to the title of the Word document
- Clients may mistakenly download an older version of an order set instead of a newer version
Main pain points for Clients (Hospital Admins)
- Unable to obtain rich data on form usage
- Fillable HTML fields are placed on top of image-based representations of the form
- Turn around time for change requests is slow
I conducted research to understand how other digital Order Sets building platforms approached this problem, and to understand how we might adapt digital technology to build our forms. During my research, I looked at the functionality of each builder, and the user interface elements.
For each platform, I reviewed:
- How documents were built
- Whether it was drag and drop or click to add
- How elements could be moved within the layout and deleted from the layout
- Document properties and setups
- Properties that could be set on each component type
- Styling options, including fonts, alignment, colour, and borders.
The Senior UX designer, Anna Kop began sketching out the structure of an Order Set and mapping out the pieces that teams commonly use to build this type of document. Atoms included form fields used to build orders: checkboxes, radio buttons, text boxes, and dropdowns. Other components included Modules, to group orders together with a title.
Anna mapped out all of the combinations of orders that builders might need to create an Order Set and the properties that orders and components may need, ie. setting a field as mandatory, or including a clinical flag to alert clinicians. We worked together to map out the information architecture of the Order Set builder, to allow for new form creation, visualization of the document library, and the document builder.
Working closely with the engineering team, we began ideating on a drag and drop WYSIWYG builder, where our internal teams could build Order Sets for clients in a quick and flexible way, without the need for any tooling except for the Word Document template built with the client and integration with our internal document library, TxConnect. Documents in FormDesigner are in HTML format, allowing for robust data collection.
Sub-ProblemUnable to obtain rich data on form usage
With the current architecture, Think Research was only able to deliver high-level data to clients, using visual heat mapping to denote which orders were being used. Analysis of this data was very manual in nature, and required a lot of time to visually review document and order usage. Due to the form fields being placed on top of an image-based form, obtaining rich data was not possible. We needed to architect the new Order Set forms to be able to provide us with more complete data on Order Set usage at a granular level.
With the change in how Digital Order Sets were going to be built, using HTML for the entire form, we also needed to align to the Word Document templates created by Think Research R&D team and used by clients to make the new digital Order Sets as readable and usable as possible.
One of my tasks was to ideate on the fillable Order Set architecture and visual design, and to create a digital form template that was usable for clinicians, met the needs of the components required to build forms, and also informed data capture.
Considerations & Requirements
- Patient safety is paramount
- Clinicians do not want to spend time filling forms - they want to spend time treating patients; Order Sets must be readable and easy to use
- Structure and layout of orders on the Order Set can inform treatment
- Users must be able to move through the Order Set fields using mouse or keyboard
- Digital Order Sets must be printable
- Ability to include required/mandatory fields and help the user understand that mandatory fields must be filled
Ideation, Round 1
In order to build data-rich HTML forms, I was tasked to ideate on the visual display of form elements within the Order Set. I created a working HTML/CSS proof of concept of the form structure and components. User testing was conducted to understand how best to lay out the forms and provide emphasis on elements that are most important for clinicians.
User Testing Feedback, Round 1
- Modules headers need to be clearly distinguishable from other modules
- Consider using a coloured background with Module Titles instead of the thick line
- Font size is too small
- Increase font size for orders - The Institute for Safe Medicine Practices (ISMP) recommends a minimum 11pt font size
- Use real data to better represent what we might see on these Order Sets
- Columns of ½, ⅓, full row width is awkward in some places
- Review layouts and document structure to understand number of columns used within real documents
- Orders with labels above are difficult to read, some have no labels
- Ensure every input has a label or placeholder associated with it to inform data required to be entered
Ideation, Round 2
User Testing Feedback, Round 2
- Related orders and components are too far apart
- Elements within an order should be closer together so it is clear that they are associated
- Placeholder text can be misconstrued as a filled order
- Consider using labels for all inputs instead
- Clinicians prefer dense data and will not want to paginate to see all of their orders; less clicks!
- Reduce vertical spacing on entire Order Set
- As a physician navigation through a frequently used doc, they skip over content. Picking out an asterisx through a lot of content is hard & it will be easily missed until document submission. If required fields are more obvious with a quick scan, this will prevent errors
- Required fields - make them more apparent beforehand with styling
Ideation, Round (n)
After the implementation of this product, Order Set building time decreased significantly, once Document Developers became more familiar with the new application. The response time for client requested changes changed dramatically, as Document Developers could quickly make changes without having to re-score the entire document from scratch.
The coded structure of the Digital Order Sets allowed clinicians to be able to find the orders they need easily, and fill Order Sets quickly. With change management assistance from our Client Success team on the new forms, clinical usability was increased. In addition, the architecture of the Digital Order Sets provided Think Research with rich data regarding which parts of the Order Set are being used, which is used, in turn, to inform whether clinicians are following best practices.
FormDesigner has been an ongoing project for over 4 years. Throughout that time, numerous rounds of user feedback and ideation has occurred to refine the WYSIWYG builder, and to improve form building usability, features and functionality and design. This application served as the cornerstone of a suite of in-house tools used to create many different types of forms with varied structures, requirements and use cases in clinical and other settings.