Observation Unit Proposal: Eleven Percent Left Without Being Seen

The proposal treats a capacity problem as a bed-count problem, which is the core risk.

Operational risks:

• Staffing arithmetic. Fourteen beds requiring observation-level monitoring plus hospitalist coverage cannot be absorbed by nine hospitalists without either dropping their inpatient load or extending shifts. At a 14% RN vacancy, the unit will be staffed by pulling from inpatient floors, which relocates boarding rather than reducing it.
• No exit criteria. Observation units work when there is a defined protocol, a target length of stay, and clear disposition rules. Without them, the unit becomes a holding area and the measured boarding time improves while total length of stay worsens.
• Measure gaming. A 20% boarding reduction is a projection, not an analysis. Ask for the model: arrivals, admission rate, current discharge-by-hour profile, and the assumption about when beds are released.

Regulatory risks:

• Status and the two-midnight benchmark. Observation is not a reimbursement-neutral default. Misclassification in either direction invites denials and, at scale, False Claims Act exposure.
• Payer authorization. A dominant payer requiring prior auth beyond 24 hours means the unit's throughput is partly determined by a payer's phone queue. Build that into the length-of-stay target or the unit will back up daily.
• Bylaws. MEC approval and a defined scope of service are required; skipping that step exposes the hospital to a credentialing challenge.

Approve the unit conditionally, sequenced behind capacity-release fixes.

First 30 days - test the constraints cheaply:

• Measure discharge timeliness by hour and day. If discharges cluster in the afternoon, shifting even a fraction to before 11 a.m. produces beds faster and cheaper than converting 14.
• Quantify where admitted patients physically wait and for how long at each step.
• Run the payer authorization data: how many observation stays exceed 24 hours, and for which diagnoses.

Days 30–90 - build the unit properly:

• Written protocol with inclusion and exclusion criteria, a target length of stay under 24 hours, and mandatory reassessment points.
• Staffing plan funded by explicit reallocation, not by assumed productivity gains.
• MEC approval of the scope of service and hospitalist coverage model.

What would change the recommendation: if the discharge-timeliness work recovers enough capacity to cut boarding materially, defer the conversion and preserve the inpatient beds, because inpatient capacity is the scarcer asset in a 180-bed hospital during respiratory season.

Governance:

• MEC approval of the scope of service and the hospitalist coverage model.
• Board quality committee informed, because the unit changes how the hospital defines and measures crowding.
• Compliance and revenue cycle sign-off on status determination, with a documentation audit plan for the first 90 days.
• A named physician and nursing owner accountable for the unit's protocol.

Metrics, reported monthly with a pre-committed threshold for course correction:

• ED left-without-being-seen rate (target below 5%).
• Median boarding time for admitted patients, measured from admission decision, not from bed request.
• Observation length of stay, percentage exceeding 24 and 48 hours.
• Observation-to-inpatient conversion rate, and denial rate by payer.
• Unplanned 30-day returns among observation patients discharged home.
• Nurse overtime and agency hours on the unit.

Set a 6-month review with a pre-agreed stop rule: if boarding has not improved by at least 10% and observation length of stay exceeds 30 hours, the unit's bed allocation returns to the MEC for reconsideration.