An Admitted Patient's Wait Is Set by Inpatient Beds, Not the Emergency Room

A Waiting Room Is a Downstream Symptom

A person sitting in an emergency room for six hours is looking at the part of the building where the delay shows, which need not be the part where it is produced. When a visit is going to end in admission, there is a second candidate for the constraint: an empty bed of a particular kind, in a particular unit, two or three floors up, in a place that patient will never see.

The federal government has a name for the resulting condition. In GAO-09-347, Hospital Emergency Departments: Crowding Continues to Occur, and Some Patients Wait Longer than Recommended Time Frames (2009), the Government Accountability Office defined boarding as the situation in which “A patient remains in the emergency department after the decision to admit or transfer the patient has been made, for example because an inpatient bed elsewhere in the hospital is not yet available.”

That sentence contains the whole failure mode. A patient for whom the decision to admit has already been made is still sitting in a room built for someone who has not been examined yet. Nothing in the emergency department has broken; its exit is simply controlled from elsewhere in the building.

The Chain

The sequence runs in a fixed order, and the federal record does not cover it evenly. Three of the six stages have no published duration in the sources used here, and one of those three is the stage that holds the patient in place.

1. Arrival. Someone walks in, or an ambulance pulls up. No appointment exists, no advance notice is required, and the hospital learns of the arrival at the moment it happens. There were an estimated 155.4 million such visits in 2022, by the National Hospital Ambulatory Medical Care Survey's count.

2. Screening. Under 42 CFR 489.24, a hospital with an emergency department must, for an individual who comes to it and requests examination or treatment, “Provide an appropriate medical screening examination within the capability of the hospital's emergency department, including ancillary services routinely available to the emergency department, to determine whether or not an emergency medical condition exists.” How long that examination itself takes is not published in the federal sources used here. What is published is the length of the line in front of it, and that line is produced by stage 6 rather than by this stage.

3. Decision to admit. The screening ends with a determination that the patient needs to be in the hospital rather than sent home, and what remains is to find the room. This is where the chain narrows sharply: in the 2022 survey, 11.5 percent of emergency department visits resulted in hospital admission, so everything below applies to that fraction and not to the visits that end in discharge.

4. The bed is not free. A bed assignment requires an empty bed in a unit appropriate to the patient, of the correct type, with staff assigned to it, and it becomes available when the person already in it leaves. The federal definition quoted above places the release point here rather than in the emergency department. How long that takes is not published in the sources used here. What is published is a weekly count of how full the inpatient side is.

5. Boarding. The admitted patient stays where they are. GAO described the mechanical consequence directly: “When inpatient beds are not available for ill and injured patients who require hospital admission, the emergency department may board them, and these patients take up extra treatment spaces and emergency department resources, leaving fewer resources available for other patients.” How long a boarding episode lasts is the quantity this article is built around, and it is not published in the federal sources examined here, so the rest of this article is written without it.

6. Capacity is lost at the front. Each boarded patient subtracts one treatment space from the number available to people still arriving. The arrival rate is unchanged. The service rate has fallen. The queue that results is the one visible from the waiting room chairs, and it has a second form outside the doors: an ambulance crew waiting to hand over a patient is a crew that is not available for the next call, so a constraint on the inpatient floor removes a vehicle from the street. This queue is the part of the chain the national survey does measure: in 2022, 40.6 percent of visits involved the patient being seen by a physician, nurse practitioner or physician assistant in fewer than 15 minutes, while 0.6 percent involved a wait of six hours or more. Those figures describe the line, not what is holding it up.

The Inlet That Has No Calendar

Hospitals have several routes into an inpatient bed. Scheduled admissions sit on a calendar the institution controls; transfers are arranged in advance between two parties; emergency arrivals are neither, and the obligation that attaches to them begins on contact.

The regulation is specific about when that contact counts. Under the definition of “Comes to the emergency department” at 42 CFR 489.24(b), an individual has arrived when the individual “Is in a ground or air nonhospital-owned ambulance on hospital property for presentation for examination and treatment for a medical condition at a hospital's dedicated emergency department.” The same subparagraph draws the boundary on the other side: “However, an individual in a nonhospital-owned ambulance off hospital property is not considered to have come to the hospital's emergency department, even if a member of the ambulance staff contacts the hospital by telephone or telemetry communications and informs the hospital that they want to transport the individual to the hospital for examination and treatment.”

Property line, not front door. That distinction is what makes ambulance diversion a partial valve rather than a real one. A hospital may direct such an ambulance elsewhere when it is in diversionary status, which the rule describes in the same sentence as a state in which “it does not have the staff or facilities to accept any additional emergency patients.” The valve has a documented limit, stated in the next sentence: “If, however, the ambulance staff disregards the hospital's diversion instructions and transports the individual onto hospital property, the individual is considered to have come to the emergency department.”

The regulation also marks where the federal obligation stops, and it stops on a condition rather than at a moment. Subparagraph (a)(1)(ii) states that “If the hospital admits the individual as an inpatient for further treatment, the hospital's obligation under this section ends, as specified in paragraph (d)(2) of this section.” That closing phrase is a pointer, and the paragraph it points to carries the terms. Subparagraph (d)(2)(i) reads: “If a hospital has screened an individual under paragraph (a) of this section and found the individual to have an emergency medical condition, and admits that individual as an inpatient in good faith in order to stabilize the emergency medical condition, the hospital has satisfied its special responsibilities under this section with respect to that individual.”

Admission by itself is therefore not the release. Admission in good faith in order to stabilize the emergency medical condition is. The paragraph these terms sit in is headed Exception: Application to inpatients, and its second subparagraph works on a different axis again. It does not describe how a hospital discharges the duty; it removes a category of person from the section's reach altogether. “This section is not applicable to an inpatient who was admitted for elective (nonemergency) diagnosis or treatment.” Someone admitted for a scheduled procedure was never inside 42 CFR 489.24 to begin with.

A third subparagraph points at a different body of law: “A hospital is required by the conditions of participation for hospitals under part 482 of this chapter to provide care to its inpatients in accordance with those conditions of participation.” That is a separate and independent requirement rather than this one carried forward. Inpatient care answers to those conditions whether or not this section ever applied.

What none of these subparagraphs contains is an interval. They say when the section's special responsibilities are satisfied, and to whom the section does not apply, and they say nothing about how long an admitted patient then waits for a room. That silent interval is the one that produces boarding.

One thing the federal text quoted here does not contain is a deadline. Nothing in 42 CFR 489.24 states how quickly a hospital must take a patient off an ambulance stretcher once the ambulance is on its property. Whether such a limit exists is a matter of state and local rules, which vary and lie outside the federal provision examined here.

The Number That Looks Calm

Hospitals report bed counts to the Centers for Disease Control and Prevention through the National Healthcare Safety Network, so the upstream variable at stage 4 is visible at national scale. The reporting form defines it plainly. Field 2a is the “Total number of all staffed inpatient beds in the facility that are currently set-up, staffed, and able to be used for a patient.” Field 3a is the “Total number of staffed inpatient beds that are occupied. This reflects occupancy levels for beds reported in field 2a (All hospital inpatient beds).”

Across the 53 consecutive weeks ending 23 August 2025 through 22 August 2026, the national figure never left a band of roughly eight percentage points, averaging 73.21 percent. Its highest reading was 76.76 percent, on Wednesday 7 January 2026, when 567,559 of 739,371 staffed inpatient beds were reported occupied. Its lowest was 68.34 percent, on Wednesday 24 December 2025.

The peak is the interesting number. On the fullest day in the year of record, 171,812 staffed inpatient beds in the country were reported empty. That is not in tension with a full emergency department; it is the same situation seen at a different resolution.

Where the Measurement Runs Out

A national figure in the low seventies and a hospital that cannot place a patient are compatible for three documented reasons, each a limit of the measurement rather than a puzzle about the hospitals.

The beds are not interchangeable. The same weekly record separates them. On Wednesday 19 August 2026, adult staffed inpatient beds were 76.92 percent occupied and pediatric beds 60.33 percent, while intensive care beds stood at 70.22 percent. An adult waiting downstairs cannot be placed in an empty pediatric bed, and a patient who needs intensive care cannot be placed in an empty medical-surgical one. Aggregating the categories creates empty beds that exist arithmetically and not operationally.

The reading is one weekday. The NHSN Hospital Respiratory Data protocol states that “Single-day snapshot data elements: the 24 data elements within the reporting form sections for Bed Capacity and Occupancy and Prevalent Hospitalizations are required to be reported as single day snapshots (specifically, Wednesday) of the reporting week.” Every point on the chart above is one Wednesday, so the series cannot show an hour of day, and the calendar can move it: the two lowest readings in the 53 weeks, 69.21 percent and 68.34 percent, fall on the Wednesday before Thanksgiving and on Christmas Eve.

The denominator moves, and the percentage by itself will not say why. Staffed beds are not a fixed inventory. The same reporting form instructs facilities to “Include any surge/hallway/overflow beds that are open for use for a patient, regardless of whether they are occupied or available.” A bed opened in a corridor enters the count. Within the year of record that count moved a long way: 739,371 staffed inpatient beds on Wednesday 7 January 2026 against 670,840 on Wednesday 19 August 2026, a fall of 68,531 beds, or 9.3 percent. Two Wednesdays make the arithmetic plain. On 24 December 2025, 497,447 beds were occupied and the figure read 68.34 percent. On 19 August 2026, 500,873 were occupied, 3,426 more, and the figure read 74.66 percent. The percentage rose 6.32 points on a numerator that had barely moved, because the denominator had fallen by 57,039. Both are single Wednesdays and one is Christmas Eve, so the pair shows what a ratio can conceal rather than measuring demand.

This is a limit of the headline number, not of the data behind it. The record publishes numerator and denominator as separate fields, and the dataset description states that it carries metrics on capacity and occupancy “with corresponding metrics indicating reporting coverage for a given reporting week.” Anyone who opens the row can tell a change in occupancy from a change in the bed count. Anyone who sees only the percentage cannot, and the percentage is the part that travels.

A fourth limit is the largest. None of these fields measures boarding. The series records how full the inpatient side is on a Wednesday, not how many admitted patients spent that Wednesday in an emergency department treatment space.

Where This Doesn't Apply

This chain describes one route to a long emergency department stay. Several common situations do not run through it.

Visits that end in discharge. The 11.5 percent admission share named at stage 3 is the outer boundary of this article. Visits that end in discharge never reach stage 4, and for those patients the inpatient bed supply is not the binding constraint. Their time in the department is governed by triage acuity, imaging and laboratory turnaround, and staffing inside it. The same 2022 tables report that 8.0 percent of visits lasted less than one hour and 33.1 percent lasted two hours but less than four; those durations are set within the department, not upstairs.

Departments that are not full. Subtracting treatment spaces only produces a queue when arrivals are competing for them. A department with open rooms can hold admitted patients without a visible waiting room effect.

Surges driven by arrivals. A mass casualty incident, a severe respiratory week or a regional facility closure raises the arrival rate directly. The queue then forms at the front door for its own reasons.

Small and specialized facilities. Critical access hospitals, freestanding emergency departments and specialty hospitals have different bed structures and transfer patterns, and the NHSN series reports jurisdictions, not buildings.

What the Chain Predicts

Two consequences follow from the ordering above, and neither needs a duration to hold. The first is positional: by the federal definition, a boarding episode ends when an inpatient bed becomes available, so its endpoint sits outside the emergency department, and a change confined to the emergency department acts on the stages around that endpoint rather than on the endpoint itself. The second is about reading the record. A national occupancy chart can look calm in a week when emergency departments are holding admitted patients overnight, because it aggregates across bed types, across thousands of hospitals and across the hours of one Wednesday, and because it does not count boarded patients at all.

The waiting room is the display. The mechanism is upstairs.