Discharge orders written but not executed

1. System failure identification

   A discharge order is written in the morning and the patient leaves in the evening. The bed is counted as occupied throughout, though the clinical decision to release it was made hours earlier. Capacity that exists on paper is unavailable to the emergency department all afternoon.

2. Data pipeline diagnostics

   Bed status is derived from physical occupancy. The order lives in a different system with no link to the capacity view, so intent and state are never compared against each other. The two systems answer different questions and are never asked the same one.

3. Predictive computational model

   Measure the interval between discharge order and physical departure, broken out by the step that consumed it: transport, pharmacy, paperwork, family arrival. The delay is not one delay. Segmenting the interval turns one opaque delay into four addressable queues.

4. Workflow integration

   Route each component to the team that owns it rather than publishing a single composite figure. A combined number tells nobody which of four queues to work on. Aggregate reporting spreads accountability until nobody holds it.

Back to the case study database