ED Flow Under a No-New-FTE Constraint
A 350-bed hospital's emergency department has door-to-provider time of 52 minutes (target 30), left-without-being-seen (LWBS) of 4.6% (target under 2%), and 18% of admitted patients boarding more than 4 hours. The ED has 42 beds and peak arrivals from 11 a.m. to 7 p.m. A rapid improvement event is scheduled in two weeks. The ED director wants to focus on triage speed; nursing wants more staff; the hospitalist group says the real bottleneck is inpatient bed assignment, controlled by the bed management office. Last year's lean event produced a standard-work document that was never audited. The CMO asks you to 'fix the front end' with no new FTE.
1. How do you scope the rapid improvement event so it addresses the actual constraint rather than the loudest complaint?
Read the worked answer
Use time-stamped flow data to locate the constraint before choosing a scope.
- Map intervals, not opinions: arrival→triage, triage→room, room→provider, decision-to-admit→bed assignment, bed assignment→transfer. Compare medians and 90th percentiles by hour of day.
- Test the boarding hypothesis: if 18% of admitted patients board more than 4 hours during peak arrival hours, boarded patients occupy ED beds and starve the front end. Front-end triage fixes cannot overcome that. Apply flow logic: when boarding consumes capacity, reducing triage time shifts the queue but does not shorten total length of stay.
- Include the right owners: bed management, hospitalists, environmental services, and transport must be in the room. An event that excludes bed management cannot fix a bed-assignment constraint.
- Respect the CMO's constraint but reframe it: 'fix the front end' should mean 'fix the front-end experience,' which may require changing inpatient discharge timing and bed assignment rules - not adding FTE.
Scope proposal: one value-stream map covering arrival through inpatient bed, with PDSA tests on discharge-before-noon and bed-assignment rules, plus a smaller triage test. Avoid producing another standard-work document without an audit and measurement plan.
2. What measures and sustainment mechanisms would you put in place so this event does not repeat last year's unused standard work?
Read the worked answer
Balance flow measures with guardrails and make the work visible daily.
Measures:
- Door-to-provider (median and 90th percentile), LWBS, ED length of stay, boarding hours after admission decision.
- Balancing measures: 72-hour returns, patient experience, staff overtime, and inpatient discharge time.
- Display on SPC charts, not monthly averages, so the team sees whether changes produced a shift.
Sustainment:
- Daily huddle with a visual board showing the constraint and today's tests.
- Standard work with a named owner and a weekly audit with feedback - the missing element last year.
- Leadership rounding on the gemba, and a defined escalation path when boarding exceeds a threshold.
- Assign a process owner for each changed step and review at 30, 60, and 90 days.
If a test fails, document why and adjust. The goal is a learning system, not a binder.
3. Nursing demands more staff and the hospitalists blame bed management. How do you manage the stakeholder conflict?
Read the worked answer
Make the data the shared arbiter and separate positions from interests.
- Acknowledge both: staffing and bed assignment are plausible contributors. Neither group is wrong to raise them.
- Use a shared aim: 'Reduce ED length of stay and LWBS for our patients' - not 'prove whose fault it is.'
- Test, don't debate: run a two-week flex-staffing test at peak hours and a parallel bed-assignment rule change, then compare. Small tests settle arguments faster than meetings.
- Clarify decision rights: bed management owns assignment rules; the hospitalist group owns admission decisions and discharge timing; nursing owns unit readiness. Write down who can change what.
- Escalate only what cannot be resolved: if discharge-before-noon requires physician behavior change, take it to the COO or chief medical officer with the data.
Keep the tone blameless. The facilitator's job is to protect the test from being killed by the loudest stakeholder before it produces evidence.