Project Catalyst
Project Catalyst began as an initiative from leadership to expand the offering of CarePort Guide. I was the only designer on a mostly internal team comprised of product managers, marketing specialists, researchers, and engineers. We began our product research by interviewing existing customers of Allscripts and CarePort, as well as users of other leading products in the industry (e.g. Epic). From these interviews we extracted the workflows and processes used today and the value and issues with these current methods. Below is the beta product concept for Catalyst that would eventually be named CarePort Transition.
Research: It can often take multiple healthcare-related professionals to provide care for a single patient. The complexity and compounding issues for a patient can often result in the need of additional specialized care, treatments, and services. The best provider of this care is for the patient, their family and their care team to decide. Once a decision has been made this kicks off a process of an exchange of information about the patient’s medical data. A comprehensive exchange is important so each health service can provide care to the greatest effect. This process is often referred to as a care TRANSITION. A common care transition that CarePort users need to coordinate is one between an acute setting and a post-acute setting, or, even more specifically between a hospital and a skilled nursing facility. This transition requires a REFERRAL or an application on the patient's behalf to a care provider for treatment after the hospital.
User scenario: Nanette Regan is an 78 year old widower who lives alone in a 2-story home. Her closest family, her adult age child Diane, lives about an hour away. Today, Nanette is being admitted to County General Hospital for a hip replacement surgery.
Starting a referral
User scenario: When Nanette’s procedure is complete and her hospital care team approves her for discharge she will need rehabilitation services. A skilled nursing facility is determined to be the most appropriate next level of care. There are many factors that can go into this decision. For Nanette a hip replacement surgery will limit her mobility and her home has stairs that are used on a daily basis. Also, specialized clinical care will be needed for a short time after the procedure and an in-patient facility can provide that more acutely.
Research: Determining the most appropriate LOC is a very important part of any transition because it directs the kind of the care that will be administered (e.g. facility-type, intensity, duration, clinical expertise, etc). By reviewing referral processes currently in practice we learned that the organization receiving the referral application is often also categorized by these levels of care. These categorizations speak to a key part of the care needed and are crucial for adhering to the patient’s care plan.
The discovery of this mental model led to one of the first decisions made about the structure of Catalyst’s UI — to partition referrals by level of care type. This mental model was supported by the fact that certain tasks are LOC-based (i.e. what information needs to be sent to a Skilled Nursing Facility vs. a Home Health Agency). Also Guide, which would need to be heavily integrated in Catalyst, could only run a search per LOC.
Design (left): The first thing you need to do as a user when starting a referral is select a level of care. This determines much of the content you will subsequently see. By limiting the meta data on each referral type we could keep the navigation between types per patient more streamlined to pick options. These options work in a first-time user’s scenario as well as navigation between in-progress type. This design was scaled down from initial concepts of a dashboard page and additional dropdown navigation within each referral type.
Design (right): Once you select your referral type you can start the referral from scratch for that patient or you can continue working on one that is currently in-progress. The recipients page (also the landing page) will give you the status of any referral requests that are out to providers. If nothing is currently active the UI shows an empty state where users are directed to on how to start selecting recipients for the referral.
Searching for care providers
User scenario: Corrine launches the search feature (also known as CarePort Guide). Nanette’s insurance information and the clinical service recommended for her rehabilitation are entered. The search location and radius is also adjusted. Diane, Nanette’s daughter, wants a facility close to her own residence so that visiting her mother is possible during the work week. The results returned are ordered by their post-acute network affiliation, then medicare star rating and then distance from search location. Corrine emails the search results so that Diane can review and select her top choices from a list of appropriate PACs in her area. Diane’s responses will appear in the catalyst recipients page. This has dual benefit. Corrine knows Nanette’s family’s wishes and it is recorded electronically in their system for future reference.
Research: Much of the original guide functionality was reused for Catalyst. Guide had active customers using and giving regular feedback on the product already. Also Guide was integrated into other referral management software and being used similarly as it would in catalyst.
Design: Two call to actions were added to the Guide UI for the Catalyst integration. One closed out the guide search page so that user could return to Catalyst. The other added, either individually or in bulk providers to the list of recipients for the patient’s referral. Because Guide was a more tested product at this point we wanted to keep changes to the UI minimal. Features like sharing a list of providers via email with patients and their families were kept in.
What’s a post-acute network? PACs that have a formalized relationship with the acute setting that is licensing CarePort. Usually our customers want these PACs highlighted because they have been vetted and approved as providers of quality care.
What’s a medicare star rating? CMS (Centers for Medicare & Medicaid Services) regularly review the quality of US post-acute care providers and assign a grade. In the case of SNFs they rate from one to five stars with five being the highest rating.
Preparing and sending a referral
User scenario: As she waits for a response, Corrine will send the referral application to her purposed SNFs. Coordination of these transitions can take time and it is ideal for the patient to be admitted at an acute care facility for a few days. Acute facility are very expensive and also have a lot of other patients. Limiting all patients exposure to other illness is important. Corrine inputs data about Nanette, reviews the clinical information and then sends the referral to the selected recipients.
Research: Users can be in different stages of the application process with providers at different times. This results in a need for flexibility in editing and sending the referral. There can only be one version of the referral last sent/saved but recipients will not receive an updated one until they are sent the application again.
A lot of patient and clinical info is inputted and stored in Epic resulting in a user resistance to working outside of Epic. Users, very understandably, do not want to input something twice. As a result, integrating with Epic was considered an important feature. Most of the data inputted in the “information” and “clinical” pages was pulled from Epic. Some of it could be then edited in Catalyst, although clinical data could not.
Design: These series of screens nicknamed the “checkout flow” is where a user inputs and reviews the data needed for a referral application. The information needed varies depending on the level of care. This flow is separate from the other main tasks such as searching for providers, adding them as recipients for the referral and communicating with the provider after the referral is sent. Special attention is needed when sharing patients’ personal health information (PHI) because a misuse of data is not only violating for the patient but is also a liability with the law under HIPAA.
Once a version of the referral is sent it is saved and can be sent again to new or existing recipients. Users can review and edit the referral as well as select who will receive it before each submission. A few UI points:
The idea of the checkout flow influenced the tabular navigation that could act as CTAs as well as a count of pages before completion. Referral applications usually need to be updated over time. Editing and resending or just resending can be easily achieved with navigation.
“Information” tab is split by the data that can’t be edited but isn’t clinical documentation and information that can be edited or needs to be inputted from scratch. The exact value of these inputs were not clear because they could vary so wildly depending on many factors including LOC, location of the hospital, etc. So for a beta what seemed safest and fastest was to split the screen by editable and not editable. Reevaluation would be needed upon use by actual users in real scenarios.
“Clinical” information is usually pulled from Epic and shared as a PDF so no inputs could be pulled individually for review or editing. We did hear user feedback that being able to review the information even though it is not editable in Catalyst was important.
“Recipients” tab contains the same information as it does outside the checkout flow. It is modified to fit the new context but the similarities help reduce cognitive load.
Versioning is something that was thought to be not important enough for MVP. The user can only review a referral application in the state it was last sent. But recipients of those applications will not receive a new applications unless checked before submission.
Tracking responses in Catalyst
User scenario: After sending the referral application to 3 potential post-acute providers for Nanette, Corrine can now track any responses or messages on the recipients page. One provider has declined and another has asked for more information about Nanette. Corrine can open up the activity panel for that provider and send a message back. She can also see that Diane has responded with her top selection. Corrine will have to ask Diane if she has any other selections because hers unfortunately declined to take Nanette for care. They will probably have this conversation via email or phone. In the meantime, Corrine will continue to work to place Nanette until her transition date comes.
Research: I had one user interviewee say to me that she didn’t want to click on anything that she would rather just keep scrolling. Healthcare workers have to use a lot of different software so when asked to learn the UI of a new one it is especially fatiguing. Of course we didn’t want to build an endlessly scrolling page (how would you know where anything is?) but I also wanted to ease the transition of going to a new page.
Design: A split screen design turns the dashboard-like overview of the recipients screen into a worklist and detail page. “New” activity is identifiable on both screens so that you don’t need to go back and forth. All activity between the user and the provider is documented on the detail overlay.