Discharge before noon is the wrong target

Discharge before noon is the most widely adopted throughput metric in American hospitals and one of the least useful. Not because early discharge is bad — it is good — but because the metric is trivially gameable and measures the wrong end of the process.

What gets optimized is the discharge order timestamp. Physicians round earlier and write conditional orders. The percentage climbs. Leadership reports improvement. And the emergency department still boards patients from two in the afternoon until midnight, because the bed did not actually become available when the order was written.

The interval that matters

The useful measure is the gap between the discharge order and the patient physically leaving the room, and then between departure and the bed being ready for the next patient. That is where the capacity lives, and it is almost always a three-to-six hour window that nobody owns.

Inside that window: waiting for transport, waiting for a prescription, waiting for a family member who was told 'sometime this afternoon,' waiting for durable medical equipment, waiting for a skilled nursing facility to confirm a bed, waiting for a nurse who has four other patients to complete discharge teaching. Each of these is a separate operational problem with a separate owner, and none of them is affected by moving the order earlier.

What to measure instead

Three intervals, tracked by unit and by day of week:

Order-to-departure. If this exceeds three hours, you have a discharge execution problem, not a physician rounding problem. Break it down by cause — the top three causes usually account for most of the delay and are usually fixable.

Departure-to-bed-ready. This is environmental services capacity and communication. It is frequently the cheapest hour to buy back in the entire hospital, and it is almost always understaffed at exactly the time discharges cluster.

Discharge clustering. Plot departures by hour. If your distribution is a spike between 3 and 6 p.m., the problem is that discharge work is being done after rounds rather than in parallel with them, and every downstream process is being handed a batch it cannot absorb.

The intervention that actually moves it

Identify the discharge the day before. Not a prediction — a decision, made on rounds, that this patient goes home tomorrow unless something changes. That decision lets everything with a lead time start: transport booked, prescriptions sent, family told a specific hour, equipment ordered, teaching started.

Almost all discharge delay is lead-time delay. The processes are not slow; they are started too late. A metric that pushes the order earlier does nothing about that, because the order was never the constraint.

If you want a single number to replace discharge-before-noon, use the percentage of discharges identified on the prior day's rounds. It is harder to game, it is causally upstream of everything else, and when it moves, beds actually appear.