Billing Holds That Stop Unready Infusion Claims

Infusion Billing in a Nutshell

Infusion billing turns each infusion a patient receives into an insurance claim: the drug and its administration (the nurse's time; supplies are usually bundled in). Many infusion centers buy these drugs up front, often thousands of dollars per dose, and are paid back only when the insurer pays the claim. While accurate billing keeps infusion care financially sustainable, the traditional method relies on spreadsheets of payer rules, authorization trackers and clearinghouse portals that are prone to errors and overwhelm small billing teams.

What is Endue

Endue delivers an infusion operations platform that helps infusion operators and home infusion pharmacies manage referrals, intake, scheduling, inventory, dispensing and billing in one place, so patients get on therapy faster and teams can focus on care, not clicks. Endue integrates with EMRs such as Epic and NextGen through API or HL7, with pharmacy systems such as CareTend and CPR+, and with claims clearinghouses to ensure accurate data. The company's broader vision is to be one platform for all things infusion, from the first referral to the final payment.

Where we started with infusion billing

Endue released its Billing MVP in 2024, so infusion operators could run their revenue cycle (RCM) in the same platform they use for care: from charge capture and claim creation to submission and payment posting. Our goal as a billing solution is to make preparing and submitting infusion claims simpler and faster.

Endue's billing module had a clean interface and key features like automatic claim creation from completed visits, 837P claim files, multiple locations and tax IDs, clearinghouse submission, payment posting from 835 remittance files, basic billing setup, and role-based permissions. However, the module experienced low adoption: six months after launch, only 12 of the 41 customers who bought the Billing suite were submitting claims through Endue. Most of the others exported claims to fix them in their old clearinghouse portal, and 11 still sent everything to an outside billing company.

Problem space

Current customers and prospects are blocked on adopting Endue Billing.

Opportunities

There is an opportunity to investigate customer adoption issues and identify product enhancements to better serve our target audience: independent infusion centers and home infusion pharmacies whose billing is done by a small in-house team (one to five people) or by an outside billing company working in Endue on their behalf.
How can we enhance Endue Billing to provide a more efficient way of preparing and submitting infusion claims that get paid the first time?

My role

As a Product Designer in a company without a dedicated PM for Billing, I partnered with our engineering lead and our in-house billing specialist, a certified coder with 12 years in infusion billing, to define project scope and objectives. My responsibilities included managing the end-to-end design process, conducting user research and validation with our Operator Advisory Council, documenting requirements, creating high-quality UI designs, contributing to our design system, planning sprints to track progress, and aligning sales, customer success and implementation teams with the product direction. Because our users work with patient data, all research used de-identified data covered by our business associate agreements (BAAs). Every billing rule in this project was reviewed by our billing specialist before it shipped.

Solutions discovery

Based on research, usability testing, self-reported requests and feedback from our sales, customer success and implementation teams, I identified nine potential features to explore. With only seven engineers in the company, and 1–2 engineers per feature pod, we had to stay scrappy and considerate. Our team scored every feature with the RICE model (reach, impact, confidence and effort) so we could compare them objectively and agree on the trade-offs together. These were the priorities for the next two quarters:

1. [Billing manager / Automation] Adding Billing Holds will stop visits that aren't ready from being sent as claims, reducing rework and preventable denials, and route each problem to the team that can fix it.
2. [Billing manager / Orchestration] Adding a Drug catalog will let teams set how each drug is billed (NDC, HCPCS code, billing unit and administration code class) without depending on the EMR's drug list being clean.
3. [Biller / Accuracy, Compliance] Adding Charge checks will compare every claim line with the nurse's infusion record and the drug catalog, and stop lines the documentation doesn't support before they're billed.
4. [Scale, Home Infusion Readiness] Adding home infusion claim types (per diem and home nursing for commercial plans, Medicare home infusion therapy services) will let home infusion pharmacies and larger operators bill everything in Endue.

Score = Reach × Impact × Confidence ÷ Effort. Billing Holds doesn't have the widest reach (Charge checks does), but it has the highest impact and the strongest evidence. Home infusion claim types scored close to the next candidates; we chose it because six prospects in the pipeline couldn't buy without it. Pharmacy-benefit (NCPDP) claims were split out to a later release.

Billing Holds project

To show how we delivered projects at Endue, let's take a closer look at the Billing Holds project. It has a clear scope and direction, and it came up in 61 of 214 recorded sales and onboarding calls, more than any other billing topic. In addition to leading the end-to-end design process, I managed the project plan for the whole Billing enhancement effort.

What do our customers hope to accomplish

I wanted a foundation of issues and requirements before designing anything, so I started with what customers had already told us. I searched 214 recorded sales and onboarding calls and 96 billing requests for mentions of holds, authorizations and claims going out too early.

Next, I asked our sales team which customers had paid for Billing and compared that list with product analytics. The gap was clear: 23 of the 41 customers created claims in Endue, then exported them to their old clearinghouse portal to fix them by hand. I also listed 18 prospects in the sales pipeline whose buying decision depended on billing.

I sent a short survey to our Operator Advisory Council, a group of 50+ infusion operators who help shape the product, to learn how they bill today and who would help design the feature. 31 people answered and 16 agreed to work with me. I interviewed all 16: 9 billing managers, 4 home infusion billers and 3 prior authorization coordinators, because authorization problems are fixed by intake, not by billing. I also shadowed two billing teams for half a day each. In both, the billing manager kept a printed list of authorizations close to running out and checked every day's claims against it by hand.

One question came up in every interview with prior authorization coordinators: “Why don't you just stop the infusion from being scheduled?” Endue's intake already checks authorization when an infusion is scheduled. But authorizations change between scheduling and the infusion chair: an auth ends before the visit, the approved visits or drug units run out, or a dose change needs a new approval. Billing Holds is the safety net for what changes after scheduling, and every hold is also a signal to fix intake.

Jobs to be done

[JTBD] As a billing manager, I want to configure which completed visits are ready to be sent as claims so that no effort is wasted in Endue on claims that will be denied or that shouldn't be sent yet: Every completed infusion flows into billing, but not every visit is ready to become a claim. When unready visits show up in the billing queue, billing managers keep side lists of payer rules and expiring authorizations, and have to tell billers which claims not to touch. They also check every batch before it's submitted, to make sure an expired authorization or a missing modifier isn't about to turn a $6,000 drug claim into a denial.

Common hold rules include:

• Authorization (e.g., the visit is after the auth's end date, or the approved visits or drug units are used up).
• Documentation (e.g., the nursing note isn't signed, or the infusion stop time is missing).
• Payer and plan (e.g., a Medicare Advantage plan that just changed its authorization rules).
• Drug (e.g., a new biosimilar whose coverage hasn't been confirmed yet).
• Drug source (e.g., manufacturer free drug or white-bagged drug: bill the administration only).
• Claim amount (e.g., claims over $10,000 need a second review).
• Coverage changes (e.g., the patient's plan changed and benefits weren't re-verified).
• Clinical trial visits (e.g., need trial billing modifiers and the study number).
• Timely filing (e.g., fewer than 30 days left to file).

[JTBD] As a biller, I want to only see visits in my queue that are actually ready to bill, so that I can submit with confidence: On the flip side, it's slow and risky for billers to work around claims they've been told not to send, especially when a held claim can go out by mistake with one click.

[JTBD] As a prior authorization coordinator, I want to see the held visits that need my action, soonest deadline first, so that I can get authorization while it's still possible: Coordinators work authorizations for upcoming infusions all day. A visit that was already given without valid authorization is urgent, because payers grant retro authorization only in narrow cases and only soon after the date of service.

What outcomes are we trying to drive

Expectations

Current customers and prospects are blocked on adopting Endue Billing unless they can stop unready visits from being sent as claims. For people already billing in Endue, we expect Billing Holds to reduce rework and preventable denials, without delaying or hiding money.

Key metrics

I defined metrics before designing so we would know whether Billing Holds worked, and where to look if it didn't. They read as a funnel. If billing managers don't open Hold rules, we have a discoverability problem. If they open it but don't create rules, the rules are confusing or the right ones are missing. If they rely on holding visits one by one, a rule type is missing. If denials don't drop, we're holding the wrong things. And if denials drop but write-offs rise, we're only hiding denials, not preventing them. That last check is why the table has a guardrail row.

Baseline (Q1, submitting customers): 12 of 41 customers submitting through Endue, 88% clean claim rate, 3.1 authorization denials per 100 claims. Targets one quarter after launch: 60% of submitting customers with an active rule, under 1.5 authorization denials per 100 claims, median hold under 3 days, and fewer authorization write-offs than before holds.

Initial scope and iterations

MVP rules must-haves

• Authorization: This is the most common reason customers hold claims. I suggest we use whether authorization is required (from the payer and drug policy), the authorization's end date, and the visits and drug units it approves, instead of a simple “auth on file” checkbox. It's more precise, and it catches visits after an auth ends or beyond what it approved. Visits and units are tracked separately: an ocrelizumab authorization, for example, approves milligrams of drug (1,200 units of J2350), not a number of visits.
• Payer and plan type: Uses payer data we already receive from intake, so a rule can target “UnitedHealthcare Medicare Advantage” without also holding “UnitedHealthcare commercial.”
• Drug (therapy or HCPCS code): Uses the Drug catalog from priority 2.
• Documentation: Hold a visit until the nursing note is signed and the infusion start and stop times are recorded.
• Every hold needs a reason code, an owner and an ending: Billing teams already sort their work by short hold codes. Each rule sets a reason code, the queue it goes to, and a filing alert (30 days before the payer's limit by default, adjustable per payer contract). Every hold must also end in a recorded outcome: authorization obtained (enter the number and the hold releases itself), billed for denial and appeal, or written off with a manager's approval. Holds release automatically when their condition clears, such as a signed note.
• Charges are posted, only the claim is held: Held visits still post their charges, so held money shows up in unbilled reports instead of disappearing until someone remembers it.
• Individual visit hold or release: Rules never cover every case, so billers need to place any visit on hold with a reason code, or close a hold by hand with an outcome and a note. Ideally they can paste a list of visit IDs to do it in bulk.
• Held visits should be removed from billers' Charge review queue entirely and sent to the team that can fix them, as opposed to staying in the queue but locked.

Test before activating

• To build confidence that a rule holds the right visits, billing managers should be able to test it on recent completed visits before turning it on, and see every visit it would hold and why. Ideally the results can be searched by patient name or visit ID.

What we left out: “Claim amount” was a common request, but it's about who reviews a claim, not whether the claim is ready. It became a review threshold in the Ready to submit step instead of a hold rule. Drug source, coverage changes and clinical trial billing (which needs trial modifiers and the study number, not just a hold) are documented for later.

For nearly every element of Billing Holds, I had our design partners test iterations of wireframes and clickable prototypes. Each session followed a validation study and interview guide based on IDEO's Design Kit. Eight billing managers tested the rule editor, and four prior authorization coordinators tested the holds queue. Participants completed realistic tasks on de-identified sample data, such as “Hold visits whose authorization has ended, then close the hold for V-10244 after the payer approves a retro authorization,” and explained in their own words what they expected the system to do.

Applying visual design

Once 7 of the 8 billing managers (88%, above our 80% bar) had completed every rule task without help, and all 4 coordinators could find and close their holds, I moved to visual design. Billing teams spend whole days in these screens, so I designed for density and speed: work queues with counts in the navigation, short reason codes, deadlines in days, and every number lined up for scanning. Below are the final designs for the Hold rules list, the rule editor, and the Billing holds queue.

Final deliverables for full scope of improvements

Additional screens for the other three priorities. The last one, Ready to submit, replaced the MVP's single “Submit batch” button and shipped in the same release; the Billing home dashboard in the Results section came in a follow-up release.

Results and what's next

One quarter after launch, we measured the metrics we defined before designing. All four priorities shipped together, so the adoption and win-rate changes belong to the whole release. The change we attribute to Billing Holds is the drop in authorization denials.

Usage. 25 of the 27 submitting customers opened Hold rules, and 22 activated at least one rule. Authorization was the most used rule (all 22), followed by Documentation (17).

Guardrail. Median time on hold was 2 days. Of the 1,140 holds closed that quarter, 77% were released and billed after the problem was fixed, 6% were billed to get a denial that could be appealed, and 9% were written off; 8% were still open. Write-offs for missing authorization fell by about a third compared with the quarter before, so the drop in denials came from preventing them, not from hiding them.

Revenue. Holds or authorization checks came up in 4 out of 5 billing demos, and we won 6 of 9 head-to-head deals against an incumbent billing system, compared with 3 of 10 in the two quarters before the release.

The screen below is what one customer's billing manager sees each morning.

That panel confirmed our next priority. Free drug came up in the very first research quote, and the template was already marked “coming soon,” but the number of manual holds moved Drug source from “later” to “next.” Charge checks will use it to bill only the administration when the drug was free or white-bagged.

What I'd do differently: unsigned notes were the second most common hold, as research suggested, but I underestimated how easy they were to prevent. The fix belongs in the nursing workflow (a reminder before a nurse closes the visit), not in billing. I would have shadowed nurses as well as billers, and I would have tested the holds queue with coordinators earlier, before the rule editor was finished.