Duplicate Triage Documentation in a Crowded ED

Frame the question first. The CNO has stated a solution ("fix the double charting"). A workflow analysis should test whether duplicate keying is the dominant cause of the LWBS rise, or merely a visible symptom of a bed-assignment bottleneck.

Data and methods

• Current-state swimlane map spanning registration, triage, bed assignment, and order entry, marking every handoff and every rework loop.
• Direct observation (not self-report) of at least three shifts per day-part, including two weekend shifts, using a time-motion form. Triangulate with system audit timestamps: time from triage-document-complete to encounter-create, and encounter-create to first order.
• Quantify the waste: 118 patients x 4.2 minutes is roughly 495 nurse-minutes per day, about 8.3 nursing hours, near 1.0 FTE of capacity; the 6.8-minute evening figure needs its own explanation.
• Structured interviews with triage nurses, registration, bed control, and two ED physicians to surface workarounds (e.g., verbal orders, sticky notes, a shadow spreadsheet).
• Variance analysis of the 22% LWBS cohort versus the 78% who stayed: arrival mode, ESI level, time of day, whether a bed was available.
• Flow chart of the request from patient arrival to order placement, annotated with decision points and delays.

Deliverable A single decision packet: current-state map, quantified non-value-added time, a Pareto of delay causes, three future-state options with cost, staffing, timeline, and risk, and a recommendation with the measures that would prove or disprove it. Separate what is value-added from the patient's perspective (assessment, vital signs) from what is pure rework (re-keying the same fields).

Caution. Do not accept the survey date as the driver. Joint Commission standards do not prescribe this workflow; framing the change as survey avoidance invites a cosmetic fix that the next shift will work around.

Generate options before choosing. At minimum three exist: (a) create a pre-admission or pre-encounter record at registration so orders can be placed at triage, likely configuration rather than custom code; (b) build the encounter-creation interface; (c) reduce re-keying through device integration or a one-way data pull from the ED system into the EHR.

Sequence by reversibility and speed, not by size.

1. Weeks 1-3: validate the registration-driven pre-encounter option. If configurable, it can be tested on one pod and rolled back. It also respects the constraint (nurses' triage steps do not change; registration gains a step). Cost is analyst time, not $38,000.
2. Weeks 1-6: instrument the baseline. Capture the metrics now, before any change, so later claims are credible. Without a baseline, a March improvement cannot be attributed to anything.
3. Weeks 4-17: run the interface build in parallel if the quick win proves partial, using the 0.5 FTE for specification and testing and asking IT to fund build hours from the capital request already drafted.
4. Ongoing: attack the evening peak specifically. The 6.8-minute figure at 19:00-23:00 suggests a staffing or bed-turn trigger, not an interface problem; an interface will not shorten a wait for a clean bed.

Justify the order on three grounds: patient harm risk (LWBS is a safety outcome, so a reversible change now beats a perfect change in April); cost and reversibility (a configuration change that fails costs weeks, an interface that fails costs $38,000 and 14 weeks); and learning (the pilot tells you how much of the 4.2 minutes is truly duplicate entry versus waiting).

Also state what you will not do: do not build a shadow documentation form or a nursing workaround that creates a third data source. That trades a visible problem for a silent one.

Three measure families, defined before go-live.

Outcome (what the CNO actually cares about)

• LWBS rate during 19:00-23:00, with a baseline of 22% and a stretch target over 8-12 weeks.
• Triage-to-first-order interval and door-to-provider time.
• ED length of stay for discharged and admitted patients separately.

Process (whether the change was adopted)

• Median minutes of re-keyed documentation per patient, measured by observation and by timestamp deltas, not by survey.
• Percentage of encounters with an order placed before bed assignment.
• Percentage of triage records with complete allergies and full vital signs.

Balancing (what could get worse)

• Registration data errors and duplicate MRN creation rate.
• Wrong-patient order events and patient-mismatch near misses.
• Nurse overtime and documentation after shift end.
• Vital-sign and allergy completeness at triage.

When to call it a failure despite a falling time metric.

A drop in duplicate-entry time is not evidence of success if it comes from skipping content. Pre-set the triggers: if allergy completeness falls below the pre-project level, if duplicate MRNs rise, or if wrong-patient events increase, stop and revert. Examine the mechanism, not just the number: nurses may be pasting a stale note, entering vitals only once into a device that never files them, or deferring documentation entirely.

Method. Plot all measures on statistical process control charts (p-charts for rates, XmR for time) with the pre-change period as baseline. A single month's improvement is noise; look for a sustained shift or a run of points beyond control limits. Report the outcome and balancing measures alongside the process measure every month, so a cosmetic win cannot be reported alone.