ART-HLC-009 ·
EHR Migration Economics: The Cash Trough Nobody Budgeted
A new electronic health record changes how a hospital gets paid before it changes anything else, and most plans fund the build while leaving the months after it unfunded.
Contents7 sections
Replacing a hospital's electronic health record is usually approved, planned and staffed as an IT project. The biggest risk in it is financial. The new system tends to work. What slips is billing, and a bill that goes out late gets paid late, for months.
Maine Medical Center shows how that plays out. For six weeks after it switched on a new electronic health record in 2012, nobody billed for a C-section. Babies were delivered, mothers went home, and the charges never reached a claim. The system had cost about $160 million. Six months after go-live the hospital reported a $13.4 million operating loss, and the chief information officer was gone (Healthcare IT News).
Nothing about that was a clinical failure. Clinicians charted in the new record, orders reached the lab, and results came back. One mapping between a delivery and a charge was wrong, and it sat in the part of the program that got the least attention.
Why billing slows down
An electronic health record, or EHR, is the system a hospital runs on: charting, orders, results, scheduling and the revenue cycle. The revenue cycle is the chain that turns care into a bill and a bill into cash. Replace the EHR and that whole chain gets replaced in the same weekend.
Every link changes at once. Registration screens are new on the first day, and so is charge capture, the step that turns a documented service into a billable line. Coders learn new work queues. Claims leave through a new connection to a clearinghouse, the company that checks claims and forwards them to insurers, and that connection has never carried real volume before. Each change is small. Together they slow the whole chain for months, and a slow chain hurts cash flow even when every dollar is eventually collected.
Payroll still runs every two weeks. That's the whole problem in one sentence: the money going out keeps its schedule, and the money coming in slips.
How long the money takes to come back
Recovery after go-live is well documented. Accordion, a firm that tracks health systems moving to Epic, breaks it into five measures, each with the time it takes to get back to its pre-conversion level (Accordion).
| Measure | Time to get back to normal |
|---|---|
| Clean claims and charging | 30 to 45 days |
| Coding | 45 to 60 days |
| Denial rate | 90 to 150 days |
| Accounts receivable days | 90 to 180 days |
| Cash collections | 120 to 180 days |
Table 1. Recovery runs in sequence, so the last line is the one a cash plan has to fund.
A clean claim is one the insurer accepts the first time it's submitted. Accounts receivable days, or AR days, measure how long the average dollar sits unpaid after the care was delivered. Read the table top to bottom. The front of the chain recovers in about six weeks, and cash at the end takes four to six months, three to four times as long.
Most plans fund the first number. A typical go-live budget pays for six weeks of at-the-elbow support and a command center. Its liquidity plan, if there is one, assumes the dip ends when the support staff go home. The gap between those two timelines is the cash trough, and it lands in the quarter right after the program declares success.
The published losses are what happens when nobody funds that gap. Ector County Hospital District reported a $77 million operating loss tied to receivables after its Cerner conversion. Covenant Health reported $60.9 million along with a 30 percent drop in productivity. Vanderbilt's operating income fell $60 million in the nine months after its Epic go-live (Becker's). Each of those is four to six times Maine Medical's loss. In every case, the software worked.
The contract is the smaller number
Budgets also get wrong where the money goes. Board conversations focus on the vendor contract, because it's the number with a signature on it. Public filings show it's the smaller part of the spend.
| Organization | Vendor contract | Total program | Contract share |
|---|---|---|---|
| NYC Health + Hospitals | $302.8M | $1,053M | 29% |
| Alberta Health Services | C$459M | C$1,400M | 33% |
| Manchester University NHS FT | £181M | about £400M | 45% |
Table 2. The signed contract runs from under a third to under half of what the program spends.
Sources: NYC Health + Hospitals and Healthcare IT News; Canadian Healthcare Technology and the Alberta Auditor General; HTN.
- nychealthandhospitals.org · hhc board approves contract for new smart electronic medical record…
- healthcareitnews.com · timeline nyc health hospitals 1 billion epic ehr rollout
- canhealth.com · alberta 459 million deal for epic emr
- oag.ab.ca · AHS WaveOne ConnectCare Feb 2020
- htn.co.uk · manchester university nhs signs 181m 15 year deal with epic
Most of the rest is people. The clearest cost breakdown in the public record is the University of Vermont Health Network's certificate of need, a filing the state regulator required before approving the spend. It put six-year total cost of ownership at $151.6 million. Outside staffing, meaning consultants and contracted builders, came to $36.4 million. The Epic licence came to $14.3 million. The consultants cost 2.5 times the software (Green Mountain Care Board).
Two more lines in that filing belong in every plan. Contingency was a flat 10 percent. And staffing came out close to even: about 85 legacy positions went away and 63 Epic support positions were created, a net reduction of 22. A replacement program that promises headcount savings is promising the wrong thing — the savings come from retired licences, infrastructure and interfaces.
Where the trough hides in an integrated system
Two years spent holding the ecosystem around an Epic implementation at a large multi-state integrated payer-provider system showed a second side of this problem. An integrated system owns a health plan as well as hospitals, so the same slowdown shows up in two sets of books. On the hospital side, receivables stretch out. On the plan side, claims from its own hospitals arrive late, which makes the plan's reserves look healthier for a quarter. A consolidated report can net the two together and show a system calmer than either half of it really is.
That program's architecture plan treated the revenue cycle as a set of interfaces to build and test. It scheduled nothing for the months after go-live, when those interfaces would carry slower and messier traffic than any test had produced. That was a scoping error. The interfaces worked; the plan simply had no line for what they would carry, and no owner for the months when it mattered. Revenue cycle was the workstream the architecture team stood furthest from, and it decided how the whole program looked in month four.
What a funded plan includes
Five things separate a plan that survives the trough from a typical go-live budget. All five are financial or contractual, which is why they tend to belong to nobody until the first month-end close after go-live shows the gap.
A liquidity line comes first, sized to the end of Table 1 instead of the start: four to six months of slower cash, agreed with finance before cutover. It needs a baseline, measured before go-live, because every recovery target is a percentage of it. Accordion's targets show what good looks like: claim acceptance above 97 percent, charges captured within 24 to 48 hours of service, denials under 10 percent, and cash at 95 percent or more of expected revenue.
The old receivables need an owner. Legacy AR, the money still owed on claims billed from the old system, is usually collected inside that system instead of being converted. That means keeping the old system licensed in read-only mode until the balance is worked, and placing the oldest accounts with collection vendors before go-live.
Daily charge reconciliation is the cheapest of the five, and it's the one that would have caught Maine Medical. It's simple: each day, compare the charges each department should have produced with the charges it actually captured. A labour and delivery unit that bills no C-sections all week shows up on the first day. CommonSpirit reports that facilities it has converted now run above 105 percent of their pre-conversion charges, and credits charge reconciliation for the result (Becker's).
Then there's the archive. Converting every year of history into the new system is expensive, and keeping the old EHR running beside it costs even more. A read-only archive that opens from inside the new system costs about a tenth of running a second EHR, according to one health system CIO (Harmony Healthcare IT). That saving only shows up when the old system is actually switched off. In one KLAS study, just 7 of 36 organizations had done it (HIT Consultant). Shutdown dates belong in the contract, with penalties attached, because a read-only licence nobody owns renews itself every year and quietly eats the saving the archive was bought for.
- harmonyhit.com · enterprise wide ehr migrations between industry leaders cerner and…
- hitconsultant.net · klas 2026 data archiving healthcare it insights
Payback takes years, and the plan should say so. Los Angeles County's own analysis of replacing its Oracle EHR with Epic found higher one-time cost, operating savings within three years, and net savings only after about ten (LA County). WMCHealth's CIO expects Epic to cost less than the current systems toward the end of 2028, a year after a December 2027 go-live (Becker's). A business case that shows savings in year one has counted the licences it retires and skipped the quarter it has to finance.
- file.lacounty.gov · 1210925 06.10.26EpicBoardMemo FINAL
- beckershospitalreview.com · epic is about to retire 88 applications at wmchealth
The number the board sees in month four
The best published result shows the trough can be planned for. Ensemble reports a 21-hospital revenue-cycle conversion that returned to baseline in six days. Before go-live it ran weekly readiness meetings on patient access, discharged-not-final-billed accounts and clean-claim rate; after go-live, those meetings were daily (Ensemble). That's a vendor describing its own work, so six days is the best case rather than the benchmark. Still, organizations that measure the trough from the first day tend to find it shallower.
Every program has a month four. The clinical build is done, the command center has closed, and the steering committee has moved on to optimization. That's the month AR days peak and cash bottoms out, and it's the month a board decides whether the program was worth it. Whether that meeting reads as a crisis or a forecast was settled a year earlier, in a spreadsheet, by whoever thought to add a line for how long the money takes to come back.
Glossary
| Term | Definition |
|---|---|
| Accounts receivable (AR) | AR is money owed to the hospital for care already delivered. It's the balance that stretches out during a conversion. |
| AR days | AR days measure how long the average dollar of receivables waits to be paid. The number takes 90 to 180 days to get back to normal after go-live. |
| At-the-elbow support | These are trained staff who stand beside clinicians in the first weeks after go-live. They're usually the largest line in a go-live budget and the first to be withdrawn. |
| Certificate of need | Some states require this filing before a hospital makes a large capital investment. It's the most detailed public source of EHR program costs. |
| Charge capture | Charge capture turns a documented service into a billable line. A mapping error here produces care that never gets billed. |
| Clean claim | A clean claim is one the insurer accepts on first submission. Its rate is the first revenue-cycle measure to recover. |
| Clearinghouse | A clearinghouse checks claims and forwards them to insurers. A new EHR usually means new clearinghouse connections. |
| Contingency | Contingency is budget held back for the unplanned. The UVM filing carried it at a flat 10 percent. |
| Cutover | Cutover is the period, usually a weekend, when the old system stops and the new one starts. Every revenue-cycle change lands inside it. |
| Denial | A denial is a claim the insurer refuses. Denial rates take 90 to 150 days to recover after go-live. |
| Discharged not final billed (DNFB) | DNFB counts patients who have gone home but whose bill hasn't gone out. The pile grows after go-live, and it's one of the first readiness measures to watch. |
| Electronic health record (EHR) | An EHR is the system a hospital runs on, from charting and orders to scheduling and billing. Replacing it replaces the revenue cycle at the same time. |
| Go-live | Go-live is the day the new system goes into use. It starts the clock on every measure in Table 1. |
| Integrated payer-provider system | This kind of organization owns both hospitals and a health plan. A conversion slows both sets of books at once. |
| Legacy AR run-off | Run-off means collecting the old system's receivables inside that system instead of converting them. It keeps the old system alive in read-only mode until the balance is worked. |
| Read-only archive | A read-only archive stores historical records that clinicians open from inside the new EHR. It replaces a second live system at a fraction of the cost. |
| Revenue cycle | The revenue cycle is the chain from registration to payment that turns care into cash. It decides the financial outcome of a conversion. |
| Total cost of ownership (TCO) | TCO is everything a system costs over its life, staff included. It's the number that shows the contract as the smaller part of the spend. |
Keep reading
More field notes
This piece is part of the MAX Research Collective library. Browse the rest, or connect on LinkedIn.