A Problem Statement That Names Its Own Solution
Northgate Community Hospital's emergency department complains that CT results arrive too late. The Define-phase team has drafted this problem statement: "CT turnaround time is too long and radiology is understaffed, which causes poor patient satisfaction in the ED." The charter scope reads: "Order entry to final report sign-off; all modalities; all shifts; inpatients and outpatients." No baseline exists yet. The radiology manager confirms the EHR holds an order timestamp and a report-finalized timestamp, and reports only that ED physicians want "results back fast." A nurse manager wants MRI and ultrasound added to the scope "while we're in it." The champion suggests the team simply interview the ED director and move on.
1. Critique the draft problem statement and rewrite it so it can survive a tollgate review.
Read the worked answer
What is wrong with it:
- "Too long" - no metric, unit, baseline, or comparison point. Long relative to what?
- "Radiology is understaffed" - an assumed cause and an assumed solution smuggled into the problem statement. If it were true, the project would be a staffing decision, not a DMAIC project. Cause belongs in Analyze.
- "Causes poor patient satisfaction" - a causal claim with no evidence and no measurement plan.
- No time frame, no magnitude, no location/segment.
Rewrite pattern - [metric + baseline + segment + time period + gap to target]:
"Between January and March 2024, 38% of stat CT exams ordered in the ED (n = 412) exceeded 60 minutes from the EHR order-entry timestamp to the report being finalized, against a target of no more than 10%. The gap is concentrated in the 19:00–07:00 window."
That statement is measurable (the EHR supplies both timestamps), bounded, has a baseline and a target, and states no cause. The team's first Measure task is then to confirm the 38% and break it down by shift, scanner, contrast use, and patient transport - the same data of course also test the 'understaffed' hypothesis later, in Analyze.
2. Evaluate the charter scope, and respond to the nurse manager's request to add MRI and ultrasound.
Read the worked answer
The scope is too broad in every dimension and will stall the project in Measure.
- All modalities: CT, MRI, and ultrasound have different workflows, different equipment bottlenecks, and different customer requirements; one Y cannot represent them.
- Inpatients and outpatients: different transport paths, different priorities, different demand patterns - mixing them buries the ED signal in routine volume.
- "Report sign-off": a genuine end point only if the radiologist read is part of the process you control; otherwise the team owns a step it cannot influence.
Fix the boundaries with a SIPOC. Suggested in-scope: ED stat CT orders, all shifts, one hospital, from EHR order entry to report finalized; process steps include order, protocol/scheduling, transport, scan, contrast if ordered, read, finalize. Out of scope: MRI, ultrasound, outpatients beyond the ED, image interpretation quality, and report content.
For the added modalities: do not expand. Put them in a parking lot with a named future owner. Expanding mid-Define invalidates the charter, doubles the baseline data collection, and dilutes the improvement effort. If leadership decides MRI is genuinely higher priority, that is a new charter with its own baseline and sponsor - not a scope creep on this one. Scope changes after charter sign-off require the champion's approval and a documented revision.
3. How should the team convert "results back fast" into a CTQ with an operational definition?
Read the worked answer
VOC → need → CTQ → operational definition → data source.
- VOC (verbatim): "I need results back fast." Interview more than the director: ED physicians, nurses, CT technologists, and (for turnaround perception) patients. Probe why - for a stat head-CT the physician's real need may be "decision to treat within 30 minutes," which is not identical to report finalization.
- Need: timely diagnostic information to support an ED disposition decision, segmented by acuity - stat versus routine, because the requirement is not the same.
- CTQ: CT turnaround time (TAT), measured in minutes, for stat ED CT exams. Supporting CTQs might include order-to-scan-start and scan-to-report, so the team can locate where time is lost.
- Operational definition (must be unambiguous): TAT = (timestamp the radiology report is finalized in the EHR) − (timestamp the ED order is entered in the EHR). Include: exams with an ED order and a completed, finalized report. Exclude: cancelled orders, patient unavailable/refused, repeat scans ordered separately (count separately). Unit: whole minutes. Recorded automatically by EHR extract, not by manual stopwatch.
- Spec/target and data plan: target ≤ 60 min for at least 90% of stat ED CTs, aligned to the charter; daily extract with a consistent query; verify a sample by chart audit to confirm timestamps are populated and not batch-entered at shift end.
The last point matters: if technologists finalize five reports at 06:00 in a single action, the timestamp measures data-entry habit, not turnaround - and that is a measurement-system problem the team must resolve before baseline.