An ER and hospitalist group's first four weeks: enrollment, not coding, was the leak

The group and the starting point

The client is a California emergency medicine and hospitalist group that staffs a hospital emergency department and its inpatient units. Its physicians document in the hospital's systems, so every account reaches billing as a registration feed from the hospital, followed by the physician's chart.

Credentialing work started in March 2026. Billing went live in August: we took a backlog of dates of service from August 15 to 23, registered the first accounts on August 19, began the daily inflow on August 24 and sent the first claims on August 25. The first weekly governance review used the data as of September 11, roughly four weeks in.

At that cut the inventory held 1,787 account lines, 1,737 unique accounts. The ER coding log alone carried 1,534 rows. 1,186 claims worth $410,027.23 had been submitted. No electronic remittances had posted yet, so the numbers below describe throughput and risk, not collections.

What the first four weeks measured

Front-end registration was not the problem. 98.8 percent of accounts were registered; 22 needed clarification from hospital registration, and 20 of those 22 were missing the physician's name in the hospital feed. That is a fixable data field, not a staffing gap.

Coding throughput was 78.8 percent of inventory lines. The rest were waiting for a reason. 89 accounts were held for documentation, 63 of them because the note was still in draft or not finally signed, and three physicians accounted for 77 of the 89 holds. Another 214 accounts were queued, 201 of them inpatient, ICU or medical-surgical encounters on the hospitalist side.

On the back end, 93.3 percent of submitted claims had been confirmed received by the payer. Of 79 claim exceptions, 55 worth $19,477 were enrollment problems and none were coding problems. The group's denial log told the same story: 45 of 46 remittance denials were for provider enrollment.

Enrollment was the real constraint

The enrollment roster covered 38 physicians plus the group across 14 payers, 546 lines in all. Only 118 of those lines were billable at the cut: 79 with an effective date and 39 with a local Medi-Cal plan that does not require a contract for this service. That is 21.6 percent of the roster.

Submitted claims fell into three enrollment groups. Claims to the plan that needed no contract totaled $147,858. Claims to closed-panel plans, whose networks are closed to the group, totaled $85,712. California law still requires a Knox-Keene plan to reimburse providers for emergency services until the patient is stabilized, so these claims are worked as out-of-network emergency claims rather than written off. Claims where the physician's enrollment was still in process totaled $81,038. Each group needs a different route: the first can be paid now, the second is an out-of-network emergency claim and must be worked that way, and the third waits on the payer's effective date.

What we did

We set up the work so each problem lands with the person who can fix it. Registration gaps go back to the hospital's registration staff as named clarification requests. Documentation holds are listed by physician and by note status, so the group can see that three physicians hold most of them. Claims for a physician who is not yet effective with a payer are held and routed rather than sent to deny.

Every Tuesday the group sees one governance pack: registered, coded, held, submitted and received, holds by physician, exceptions by cause and claims by enrollment status. Before the first call we re-parsed every headline number independently from the five source files and tied it to the team's own totals; all 46 checks matched.

We also put the open decisions in front of the group's leadership rather than deciding them on its behalf: a chart completion window, how to handle 24 patients who left without being seen, the approach to closed-panel plans, adding four plans that accounted for 52 claims to the enrollment roster, co-signing contracts covering 78 roster lines, ownership of 47 inpatient accounts whose attending physicians are not on the roster, and patient statements for 56 self-pay accounts plus 8 redirected Medi-Cal accounts.

What it means for other ER groups

Two lessons carry over. First, measure enrollment as a revenue line, not a back-office task. In this group, a coding team working at full speed could not move claims that the payer would deny for enrollment, and the roster showed it before the denials did. Medicare lets physicians bill retrospectively for only up to 30 days before their effective date, and only when circumstances precluded enrolling in advance, so the enrollment calendar is part of the revenue calendar.

Second, separate the reasons an account is not billed. A single 'unbilled' number would have hidden that 20 of 22 registration gaps were one missing field, that 63 documentation holds were unsigned notes, and that 201 queued accounts were hospitalist encounters with their own timing. Each has a different owner and a different fix. Results from the following weeks will be reported once remittances post.