When do you bill 99234 to 99236 for same-day admit and discharge?

A patient reaches the hospital at 11am, goes into observation at 1pm on a physician order, and goes home at 9.30pm. One calendar date, eight and a half hours in status. Which code family that stay earns turns on a clock most notes never mention.

When do you bill 99234 to 99236 instead of 99221 to 99223?

When the patient was in inpatient or observation status for a minimum of 8 hours but less than 24 hours, and was discharged on the same calendar date. Under 8 hours, Medicare wants an initial hospital inpatient or observation care code from 99221 to 99223, and the discharge day management codes 99238 and 99239 are not reported at all.

Almost every page on this subject cites section 30.6.8 of Chapter 12 of the Medicare Claims Processing Manual for that rule. The operative text is not there. It sits in section 30.6.9.1.C, carried into the manual by Rev. 11842 effective 1 January 2023. Section 30.6.8 governs who may bill observation care and what the record has to hold. If you are quoting the rule in an appeal, cite 30.6.9.1.C.

CMS publishes the decision as a four row table. The row nobody reproduces is the third.

Read the third row twice. A stay that begins at 10pm and ends at 3am crosses a calendar date but lasts five hours, and it earns an initial hospital care code alone. No discharge code. Practices assume a different-date discharge automatically buys 99238 or 99239, and on a short stay it does not.

What happened to 99217, 99218 to 99220 and 99224 to 99226

Deleted on 1 January 2023. Observation was folded into the hospital inpatient or observation care family, so initial care is 99221 to 99223, subsequent care is 99231 to 99233, discharge is 99238 or 99239, and the same-date pair stayed where it was. Published guidance has not caught up: two of the three payer policies we pulled for this page still instruct the deleted codes. A page dated this year that tells you to bill 99218 with 99217 is describing a code set gone for nearly four years.

When does the 8-hour clock start, and what happens at midnight?

It starts at the admission order or the start of observation, not at the emergency department door. The CMS table is explicit about the anchor: the stay is measured from admission or start of observation. Four hours of workup in the emergency department before anyone writes an observation order are four hours that do not count toward the 8.

That point decides more claims than any other, because registration time, bed time and order time are three different timestamps in most systems, and the one that governs is the order. A medical observation record with dated and timed physician orders is a Medicare requirement in its own right, separate from whatever the emergency department encounter produced.

Midnight is the second trap, and it looks like a contradiction. One rule says a stay discharged on a different calendar date is not a same-day admit and discharge. Another says a continuous service spanning two calendar dates is a single service, reported on the date the encounter began, with all its time applied to that date. Both are true and they answer different questions. The first is the patient’s status clock, which runs in real time. The second is your E/M time, which collapses onto the starting date. A continuous visit from 11.30pm to 12.40am is one service on the earlier date for time purposes, and still a different-date discharge for the 8-hour rule.

One more line worth knowing: a transition from observation status to inpatient status does not constitute a new stay. Same-date observation followed by an inpatient order is one initial service by that physician, not two.

How do you choose between 99234, 99235 and 99236?

By medical decision making, or by total time on the date of the encounter. Nothing else. History and examination no longer select the level for any of these codes, which is why pre-2023 guidance about a comprehensive history is actively misleading.

The 99234, 99235 and 99236 CPT code descriptions, side by side

CMS encodes the minutes in its own short descriptors, which is the cleanest confirmation available: 99234 is straightforward or low decision making, or 45 minutes or more; 99235 is moderate, or 70 minutes or more; 99236 is high, or 85 minutes or more. The initial-care ladder sits lower, at 40, 55 and 75 minutes for 99221, 99222 and 99223, and discharge management splits at 30 minutes between 99238 and 99239.

Time here is total time on the calendar date, including the non-face-to-face work: chart review, ordering, documenting and care coordination that day. It is not face-to-face time, and a widely read page on this query still tells readers it is. If you select by time, the minutes have to be a number in the record, not a phrase.

In practice a same-date stay that genuinely takes 85 minutes of total time usually also carries high decision making, so the two paths agree. Where they disagree, pick the one the record supports.

What has to be in the note for a same-day admit and discharge?

Three things, named by CMS in section 30.6.9.1.D, and the first one is the reason this code family denies.

1. A statement that the stay for hospital treatment or observation care involved 8 hours but less than 24 hours. CMS wants the span asserted in the record, not inferred from timestamps by whoever audits the chart later.
2. Documentation identifying that the billing physician was present and personally performed the services.
3. Documentation identifying that the admission and discharge notes were written by the billing physician.

The third bullet is where the two-encounter expectation comes from. CMS writes no numeric two-visit rule anywhere; anyone telling you it does is paraphrasing. What CMS requires is separately authored admission and discharge notes by the billing physician, which in practice means the patient was seen twice. A single combined note covering the whole day satisfies neither bullet.

In our claim audit, 29% of same-day admit and discharge denials traced back to a missing statement that the stay ran 8 but less than 24 hours. The stay qualified and one sentence was absent. Building it into the discharge template is a one-afternoon fix, and ordinary coding maintenance against the current rule set.

What do 99234 to 99236 pay in 2026, and does the 8-hour rule cost you money?

This is the question that reframes the topic, and it needs arithmetic rather than opinion.

At the CY2026 nonqualifying conversion factor, 99234 allows about $88, 99235 about $143 and 99236 about $190. An initial hospital visit at 99223 allows about $156 on its own, and 99223 paired with 99239 allows about $263. Those are national amounts before any geographic adjustment.

A stay under 8 hours, correctly billed 99223 alone, pays more than a compliant 99234 and nearly as much as 99235. A stay crossing midnight that earns an initial code plus a discharge code pays more than any single same-date code. The 8-hour rule is not a revenue upgrade to chase, which is the implicit framing of most guidance on it. It is a compliance rule that sometimes pays you less, and that is why it gets ignored in the direction that creates audit exposure.

99234, 99235 and 99236 RVUs for 2026, and the two conversion factors

Total RVUs are 2.64 for 99234, 4.28 for 99235 and 5.68 for 99236, identical in the facility and non-facility columns, all status A with an XXX global. For comparison, 99223 carries 4.68 and 99239 carries 3.19.

2026 is the first year the conversion factor question has two answers. CMS finalised a qualifying APM conversion factor of $33.57 and a nonqualifying APM conversion factor of $33.40, both up from $32.35 in 2025, so a fee schedule built on a single 2026 number is wrong for half its users. The 2026 rule also applied a 2.5% efficiency adjustment to work RVUs across roughly 7,700 services, with evaluation and management codes excluded, so these codes escaped it. What does move hospital-based economics in 2026 is the practice expense reallocation for facility-based services.

A worked example: 600 same-date encounters a year

Take a hospitalist group with 600 same-date admit and discharge encounters a year that meet the 8-hour threshold, split 300 at the 99234 level, 200 at 99235 and 100 at 99236. Coded to the level actually reached, those encounters allow about $74,018. Billed defensively as 99234 across the board, which is what a group does when nobody trusts the time documentation, they allow about $52,908. The gap is $21,110 a year on work that was performed, documented and already paid for in salary.

The volume and level split are modelled assumptions rather than measured rates, so run them against your own case mix. The point holds at any volume: undercoding to the safe end of a three-code range is not conservative, it is a standing discount.

Who bills the code when an NP, a consultant or a resident is involved?

One practitioner reports it: the one who ordered the observation or admission and is responsible for the patient. Consultants seeing a patient in outpatient observation bill outpatient visit codes, and on the inpatient side they report initial hospital care codes. Only the attending of record reports discharge day management, once per stay.

Split or shared changed in a way that matters here. Since 1 January 2024, the substantive portion of a split or shared visit in the facility setting can be more than half of the total time, or a substantive part of the medical decision making. Hospital inpatient and observation visits are on the list that allows the decision making route; critical care and prolonged services remain time only. Where the visit is split or shared, modifier FS is required on the claim, the practitioner who performed the substantive portion signs and dates the record, and time the physician and the NPP spend jointly counts once rather than twice.

Residents are the case most people get wrong. The primary care exception does not cover 99234 to 99236, so a service involving a resident needs the teaching physician physically present for the critical or key portions. Where a resident admits on one date and the attending first sees the patient the next day before discharge, the same-date codes belong to the attending’s date. Matching the billed practitioner to the one who furnished the service runs through hospitalist billing generally.

Why do same-day admit and discharge claims get denied?

The denials cluster tightly, which is the useful part.

Across the same-day admit and discharge claims we review, 29% fail on the missing 8-hour statement, 24% were billed on a stay documented under 8 hours, 18% carried a discharge code alongside the same-date code, 15% had one combined note where two were required, and 8% named the wrong billing practitioner. Coding and modifier errors caused 21% of denials across our audit as a whole, so this family runs hot even by that standard.

Work the queue in this order. Check the status clock first, from the order rather than from arrival, because if the stay was under 8 hours the denial is right and the claim should be corrected to 99221 to 99223 rather than appealed. Then check whether an 8 to 24 hour statement exists; a missing one is a documentation fix going forward, not an appeal you win on this claim. Then confirm the admission and discharge notes were separately authored, and that no discharge code went out alongside the same-date code. Only then write the appeal, citing section 30.6.9.1.C by number.

It is worth filing. Appeals filed by Luxen were overturned 68% of the time, at a median turnaround of 34 days from filing to payer decision. The harder problem is that 19% of denied claims were never reworked or appealed at all, and the top three denial reasons accounted for 58% of denied dollars in the average practice. That concentration is what makes a denials queue beat claim-by-claim rework, and status and coverage questions belong upstream in eligibility and benefit verification. The same two-services-on-one-date logic behind critical care billed alongside an emergency department visit and a wellness visit paired with a problem-oriented visit fails in one place: a record that never separates them.

Should hospital E/M coding sit in house or with a billing partner?

In house wins whenever one named person owns this rule set and reads remittances against it. The whole thing is knowledge rather than volume: one clock, one table, three documentation bullets, one modifier. That is a role, not a department.

The case for outside help is almost never one code family. Fully loaded in-house billing cost 7.9% of collections for practices under $2M, across 96 practices that shared payroll data, against an outsourced range of 3% to 6% of collections. Across our client practices, first-pass denial rate fell from 14.2% to 6.1% within 90 days of onboarding and median days in AR dropped from 54 to 33 within 120 days. Neither figure is about 99234. Both are about whether the same discipline is missing everywhere else. The questions that separate vendors are on our medical billing companies page, and the whole-cycle view is on revenue cycle management. Either way, start by pulling 90 days of claims carrying 99234 to 99236 and counting how many record a status span of 8 hours or more in words.

99234 to 99236 or 99221 to 99223 plus a discharge code: which one does this stay get?

Both paths describe a patient who came in and went home. The status clock and the calendar date pick between them, and the money runs the opposite way to most people’s intuition.

The line to hold on to: the status clock decides the family, the calendar date decides whether a discharge code exists at all, and neither of them is a coder’s judgement call.

Hospital medicine

This is the home case and the one carrying the exposure, because the error runs in both directions at once. Same-date encounters billed as 99234 on stays that never reached 8 hours create audit risk, while qualifying encounters billed defensively at the lowest level leave money on the table. A hospitalist group should be able to report monthly how many same-date admit and discharge claims it filed and how many of those notes contain an explicit 8 to 24 hour statement. If nobody can produce that number, neither error is being managed. Our hospitalist billing page covers the wider picture.

Emergency medicine

The emergency department owns the timestamp that decides the claim and almost never owns the claim. Hours spent in the department before an observation order exist in the record but do not count toward the 8, so a department that routinely writes the order late is shrinking a stay it already provided. Where the same practitioner admits from the department, all same-date services by that practitioner fold into the initial hospital inpatient or observation care service and no separate emergency department visit is reported. Emergency room billing and hospital E/M have to reconcile on one timeline rather than two.

Cardiology

Chest pain observation is the highest-volume same-date stay in American hospitals and the one most often cut short. A rule-out that resolves in six hours is an initial hospital care code, not 99234, however much work it carried. Cardiology also runs into the global period question more than most, since a same-date observation stay that ends in a procedure brings modifiers 24, 25 and 57 into play. Prior authorization is a parallel drag: it was missing on 13% of advanced cardiac imaging claims in our audit. More on cardiology billing.

Primary care and internal medicine

Internists admitting their own patients hit the ownership rule hardest. Only the practitioner who ordered the admission or observation reports the same-date code; a partner who rounds and discharges is not the billing physician unless the admission note is also theirs. The office visit the day before is separately payable even where less than 24 hours elapsed, which surprises people who assume the 24 hour figure is a blackout window rather than a status span. Primary care billing runs on the same modifier 25 discipline that these claims need.

Behavioral health

Psychiatric observation stays sit on the same rules with a different failure mode: the status order and the medical clearance workup often live in two systems, so the 8 hour span is real but unprovable from either record alone. The split or shared route matters more here too, since much of the same-date work is furnished by nurse practitioners and physician assistants. Decide who performed the substantive portion, append modifier FS, and have that practitioner sign the record rather than reconstructing it at coding.