AuthConnect
Moved prior-auth tracking out of spreadsheets and into the EHR, across 7 markets
- Role
- Lead designer; trained the two designers who took it over
- Scope
- Product design, research
- Outcome
- Deployed in 7 markets; targeted $7.76M less in write-offs, Jan–Jun 2024
78% of physicians say patients abandon treatment because of prior authorization struggles with insurers (AMA, 2024). At Ascension, it often started with a surgery booked at short notice: the staff at the surgeon’s office who submit the authorization heard about it late or not at all, and the hospital wasn’t reimbursed.
The hospital and the clinics worked in isolation. The hospital runs on Cerner and the surgeons’ clinics on athenaOne, each with its own scheduling and billing. The staff who process authorizations were tracking cases on paper, sticky notes and Excel.
I was the lead designer on AuthConnect, working with a UX researcher, engineering teams and our sponsor, a clinician. We built it inside athenaOne with athenahealth’s Forge design system, syncing with Cerner. The work started conversations across teams, and I became the designer the clinical tools and clinical transformation groups turned to. When I moved to clinical transformation, I trained the two designers who took AuthConnect over.
AuthConnect is live in 7 markets, with a target of $7.76M less in write-offs tied to prior authorizations from January to June 2024.
Where it broke
We mapped the surgical process end to end and found a failure point at almost every step:
- Scheduling. The OR scheduler books the procedure in Cerner.
- The face sheet. The surgeon carries a printed patient face sheet back to the office and hands it to the surgery scheduler. That can take days, and sometimes it never arrives.
- Insurance. The scheduler finds the patient in athenaOne, but the insurance there doesn’t always match Cerner, so authorizations get requested for the wrong plan.
- Codes. Incorrect ICD-10 or CPT codes hold up approval.
- Visibility. Once the authorization is in athenaOne, the revenue-cycle team still can’t see its status, and the gap with the surgeon’s office causes more delays.
What staff told us
We worked closely with experienced staff who process surgical authorizations. We mapped how they worked, ran card sorts to design their ideal worklist, and wrote user stories from the interviews.
Three things came out of it. Tracking cases across paper, stickies and spreadsheets led to inconsistencies, missed requests and general work fatigue. Staff wanted more information on screen, because they work through many cases at once. And a big part of the job was reconciling data that disagreed from one EHR to the next.
Four decisions
Designed in athenaOne’s own language
Ascension has its own design system, Helix, and there was a question of whether to use it. We chose athenahealth’s Forge instead, so AuthConnect looks and works like the rest of athenaOne, and staff aren’t pulled out of the system they spend their day in. That meant learning athena’s and Cerner’s design systems in depth, and that expertise became the foundation for the microapps that followed.
athena’s notes as the source of truth
We didn’t want to keep a second database, and athena’s records were the ones people trusted. The revenue-cycle team could already see athena. So staff open AuthConnect straight from athenaOne, and every change, whether the system made it or a person did, is written back to athena as a note. Notes flow both ways, so the record stays the same wherever someone looks.
That was a product decision, and my job was to turn it into UI. Working closely with engineering, I defined the status model (Not Started, Pending, Obtained, Closed) and the format of every note: when, what status, who, and what happened. The cost was structure. Anything we wanted to track had to fit inside a note.
A dense table instead of a kanban board
A kanban board looked simpler. In the card sorts, staff chose the dense table. They scan a lot of cases at once and need the details side by side. We kept the table and cut it down to the data elements they said mattered.
Reconcile before you submit
Incorrect or incomplete requests are the most common reason an insurer delays or denies an authorization. AuthConnect gives staff the most accurate data feeds to fix conflicts before a request goes out, and the authorization status stays in sync both ways with athenaOne and Cerner.
A second version for radiology
Radiology had the same problem in a different shape. Every radiology order sat in an athena bucket, marked with a Task Authorization Override (TAO), until someone worked its authorization. Some orders didn’t need one. The buckets couldn’t be sorted or prioritized, couldn’t track authorization status, and couldn’t tell new orders from aged ones. So some staff printed a report and worked off paper. New orders kept coming, the report had to be reprinted, and notes were lost or copied over by hand.
The radiology version starts from the order in athenaOne instead of a face sheet. Staff get a live list of every order and its authorization status that they can sort, assign and reassign. Notes stay in sync with the athenaOne record, so nothing is lost when the list changes. Automatic notes record every status change. Text macros came from watching staff type the same notes over and over; a short code now fills in the whole note. Managers get a view of each person’s workload at every stage.
What changed
AuthConnect is live in 7 markets. For patients, fewer late or denied authorizations mean fewer delays to their care.
I also designed another microapp, Seamless Transitions, which recorded encounters to keep care continuous.
What I’d do differently
We were answering immediate requests, and working in sprints kept us shipping. It also left little time to test with staff between releases. We tracked write-offs, but not the staff’s side, like time per case, missed requests and how the work felt. Next time I’d build both into the plan. I’d also A/B test the design-system choice, AuthConnect in Forge against a version in Helix, to see whether staying inside athenaOne’s look actually helped staff.
The bigger change would be scope. There was an exploration of Redwood, the design system behind Oracle Health’s new EHR. I’d push to design for that platform, and to rethink what seamless transitions of care look like across all the microapps at once.