HA 645 Unit 5 quality improvement analysis example

Reducing Left-Without-Being-Seen Rates in a Community Hospital Emergency Department

Student Name

Healthcare Administration Program, Herzing University

HA 645: Quality Improvement and Patient Safety in Healthcare

Instructor Name

March 2, 2026

Department Profile, Problem Statement, and Measure Set

Northfield Regional Medical Center is a composite 214-bed community hospital whose emergency department holds 28 treatment spaces and recorded 46,800 registrations in the most recent fiscal year, an average of 128 arrivals each day. Volume is not distributed evenly across those hours. Between 3:00 p.m. and 11:00 p.m. the department receives 47 percent of its daily arrivals against 38 percent of its staffed provider hours. The consequence shows up in a single figure. Of the 11,842 patients who registered between January 1 and March 31, 462 left before a provider evaluated them, a left-without-being-seen rate of 3.9 percent. Median door-to-provider time over the same quarter was 47 minutes, and the ninetieth percentile was 112 minutes.

Departments often treat that figure as a service complaint. It is a safety exposure. A patient who registers with chest discomfort and leaves after 90 minutes carries an unevaluated condition out of the building, and the department retains no reliable way to learn what happened next. The Joint Commission (2012) frames patient flow as a leadership responsibility precisely because delay converts into clinical risk before anyone records an adverse event, and the Agency for Healthcare Research and Quality (2011) treats crowding as a system property rather than a staffing accident. Framing the problem this way also settles who owns it. Intake design, provider scheduling, and room turnaround sit with administration, not with the individual clinician standing in triage at 7:00 p.m.

The improvement work therefore carries three families of measures rather than one, following the structure the Institute for Healthcare Improvement (n.d.) sets out. The outcome measure is the left-without-being-seen rate, expressed monthly as a percentage of all registered arrivals. Two process measures track the mechanism: median door-to-provider minutes, and the share of arrivals placed in a treatment space within 30 minutes of registration, which stood at 41 percent at baseline. Two balancing measures test for harm done elsewhere: the rate of unscheduled return visits within 72 hours that end in admission, at 1.2 percent, and worked provider hours per 100 visits, at 21.4. Every figure comes from registration and disposition timestamps in the electronic health record, extracted monthly with no sampling.

Applying the Model for Improvement

The department applied the Model for Improvement (Langley et al., 2009) rather than a full Lean deployment, because the problem was already localized and the constraint was known. The aim statement reads as follows: reduce the left-without-being-seen rate from 3.9 percent to 2.0 percent or lower, and the median door-to-provider interval from 47 minutes to 30 minutes, for all patients registering in the emergency department, by December 31. The improvement group was deliberately small, with the department operations manager as sponsor, a charge nurse, the registration supervisor, one emergency physician, an environmental services lead, and a data analyst holding rights to the timestamp extract. Naming the analyst mattered as much as naming the physician, since no change could be judged without a monthly chart produced the same way each time.

Three changes were tested in sequence, each kept small before any spread. The first placed a provider in triage from 3:00 p.m. to 11:00 p.m. on the three highest-volume days, tested across two seven-day periods, so that arrivals received an initial evaluation while waiting for a treatment space. The second authorized immediate bedding whenever an open space existed, moving full triage to the bedside instead of holding patients in a queue for it. The third set a 20-minute room turnaround target, supported by a direct page to environmental services at the moment of disposition rather than a call placed once the space was already empty. Each cycle ran with a written prediction, and the second cycle was revised twice before it held.

Judging those cycles required a chart, not a comparison of two averages. Monthly values for each measure were plotted on annotated run charts, with the date of each change marked on the horizontal axis, and read using the shift and trend rules described by Provost and Murray (2011). That discipline protected the group from two familiar errors: declaring victory after one favorable month, and abandoning a change during the ordinary variation that follows any disruption to routine. It also fixed the standard of evidence in advance. The group agreed before the first cycle that six consecutive monthly points below the baseline median would count as a signal, and that anything less would be treated as noise.

Results, Balancing Measures, and the Recommendation

Across the two quarters that followed, April 1 through September 30, the department registered 12,391 patients and recorded 231 who left before evaluation, a rate of 1.86 percent against the 3.9 percent baseline. Median door-to-provider time fell to 29 minutes and the ninetieth percentile to 74 minutes. The share of arrivals roomed within 30 minutes of registration rose from 41 percent to 68 percent. The run chart carries six consecutive monthly points below the baseline median for the outcome measure, which meets the signal rule the group set before the first cycle. Applying the baseline rate to the later volume, roughly 252 patients completed care who would otherwise have been expected to leave.

The balancing measures are where the analysis earns its keep. Unscheduled returns within 72 hours ending in admission moved from 1.2 percent to 1.3 percent, 142 patients to 161, with no shift or trend on its run chart; that movement sits inside ordinary variation and does not support a claim that faster intake pushed patients out prematurely. The second balancing measure did move. Worked provider hours per 100 visits rose from 21.4 to 23.1, an increase of about 211 hours across the period. At the department standard blended rate the added labor costs roughly $40,000, against an estimated $45,000 in contribution margin from visits previously lost. The intervention is close to cost neutral, not free.

The recommendation is to hold provider in triage staffing at its tested hours and to fund it from recovered volume rather than from a new position request, with the immediate bedding rule written into standing department practice and the turnaround page kept in place. Sustainment requires an owner and a threshold, not an expression of commitment. The department operations manager reports all five measures monthly to the hospital quality committee, and two consecutive months above a 2.5 percent left-without-being-seen rate triggers a review of intake staffing within ten business days. Spread to the system's second campus is deferred until that site produces its own baseline, because the constraint there is inpatient bed availability rather than front-end intake.

References

Agency for Healthcare Research and Quality. (2011). Improving patient flow and reducing emergency department crowding: A guide for hospitals (AHRQ Publication No. 11(12)-0094). U.S. Department of Health and Human Services. https://www.ahrq.gov/

Institute for Healthcare Improvement. (n.d.). Science of improvement: Establishing measures. https://www.ihi.org/

Langley, G. J., Moen, R. D., Nolan, K. M., Nolan, T. W., Norman, C. L., & Provost, L. P. (2009). The improvement guide: A practical approach to enhancing organizational performance (2nd ed.). Jossey-Bass.

Provost, L. P., & Murray, S. K. (2011). The health care data guide: Learning from data for improvement. Jossey-Bass.

The Joint Commission. (2012). R3 report issue 4: Patient flow through the emergency department. https://www.jointcommission.org/