Every hospital administrator has seen the CMS Hospital Compare data. What fewer have done is sit with the OP-18 numbers long enough to understand what separates the top quartile from the median — and whether that gap is even closeable with a standard operations playbook.
What CMS OP-18 Actually Measures
CMS Outpatient Quality Reporting measure OP-18 captures median ED length of stay for patients who are subsequently admitted to the hospital. This is a critical distinction. OP-18 is not a measure of the quick ESI-4 and ESI-5 visits that cycle through in 90 minutes. It is measuring the cohort that stays — the higher-acuity patients, the ones requiring workup, imaging, labs, and bed placement — and it is asking: how long does the ED hold those patients before they reach an inpatient bed?
The measure is reported in minutes. National median performance has historically hovered around 260–300 minutes for admitted patients in publicly available CMS Hospital Compare datasets. Top-quartile performers — typically large academic medical centers or health systems that have invested heavily in capacity management infrastructure — run closer to 180–220 minutes. The gap between median and top-quartile performance is therefore roughly 60–90 minutes of admitted patient waiting time per encounter. Across a 300-bed ED doing 60,000 annual visits with a 20% admission rate, that represents 12,000 patient-hours per year sitting in a performance gap that the top-quartile peer has already closed.
The Metric's Blind Spot: What OP-18 Doesn't Capture
OP-18 is valuable precisely because it is publicly reported and nationally benchmarked. But it has a structural blind spot worth understanding before you build a dashboard around it.
The measure captures only admitted patients. Patients who are evaluated, treated, and discharged home — even those who waited three hours for an ESI-3 chest pain workup that turned out to be musculoskeletal — do not appear in OP-18 at all. Nor do LWBS (left without being seen) patients or LWOT (left without treatment) patients, whose departures actually reduce your reported LOS by removing slow-cycling patients from the denominator.
This creates a perverse incentive that every ED medical director knows about but rarely acknowledges publicly: a high LWBS rate can artificially improve your OP-18 number. If your sickest, most time-consuming admitted patients stay, but a meaningful fraction of your ESI-3 and ESI-4 patients give up and leave, the denominator skews. We're not saying OP-18 is a bad measure — it captures one of the most consequential ED throughput failures with high construct validity. We're saying it should always be read alongside CMS OP-22 (LWBS rate) and your internal median ED LOS for discharged patients as a full picture of ED performance.
Decomposing the LOS Clock: Where Time Actually Goes
A useful mental model breaks the admitted patient ED LOS into four distinct intervals, each with its own operational owner:
- Arrival to triage: From patient registration to ESI assignment. Best-practice target is under 15 minutes. In high-volume EDs without a split-flow fast track, this interval frequently extends to 25–40 minutes during surge hours.
- Triage to provider: From ESI assignment to first physician or advanced practice provider evaluation. This interval varies the most across facility types and staffing models. Median community hospital performance is 30–50 minutes; high-performing EDs achieve 15–25 minutes via front-end vertical patient models or team triage.
- Decision to admit (DTA): The interval between the treating provider's decision that the patient needs inpatient admission and the formal order entry for admission. This is often where operational waste accumulates invisibly — bed control may not receive notification for 10–20 minutes after a verbal DTA.
- Admit to bed (ATB): From the admit order to the patient physically arriving in an inpatient bed. This is almost entirely determined by downstream inpatient capacity — whether beds are available, whether beds are clean, and how quickly EVS and patient transport can execute. National data suggests this interval averages 60–120 minutes in hospitals without active capacity management programs and can be reduced to 30–50 minutes with structured bed management workflows.
When a hospital's OP-18 is high, the instinct is often to examine the triage-to-provider interval. The data rarely supports that instinct. In the majority of cases, the longest intervals are DTA and ATB — the back half of the clock — driven by inpatient capacity constraints, not ED process failures.
What Top-Quartile Hospitals Are Actually Doing Differently
Consider a plausible scenario that illustrates the operational pattern: a 280-bed regional health system in the upper Midwest, operating an ED that sees approximately 48,000 visits annually. Before a formal capacity management program, their OP-18 was in the 310-minute range — near national median. Their arrival-to-triage and triage-to-provider intervals were actually competitive: 12 minutes and 24 minutes respectively. The problem was entirely in ATB, which averaged 105 minutes, driven by a bed control process that relied on nursing unit charge nurses calling down when beds were ready rather than a centralized notification system.
After implementing a real-time bed management dashboard with automated ADT-feed-driven alerts, their ATB came down to approximately 58 minutes over a 12-month period. OP-18 followed proportionally. This is not an unusual outcome; it is the modal outcome for hospitals that correctly diagnose where their LOS clock is accumulating time.
Top-quartile hospitals share a few distinguishing practices: (1) They have real-time visibility into bed status, including beds in EVS cleaning and expected time to clean. (2) They have daily bed huddles at predictable times — typically 7 a.m., 11 a.m., and 3 p.m. — at which expected discharges and pending admissions are reconciled. (3) Their bed control function is a dedicated role, not something nursing supervisors manage as a secondary responsibility. (4) They track DTA as a discrete metric, not just ATB.
The 90th Percentile Problem
Median ED LOS is the reported number, but the 90th percentile is often the more operationally important signal. A hospital can have a 240-minute median with a 90th percentile of 420 minutes — meaning one in ten admitted patients spends seven hours in the ED. Those patients are typically the highest acuity, the most complex, and the most likely to require ICU-level beds that are constrained. They are also disproportionately visible to nursing staff and to patients themselves, shaping HCAHPS perceptions far more than median metrics suggest.
Health systems that focus exclusively on median LOS reduction often find that the 90th percentile remains stubborn. The patients at the extreme end of the distribution require different interventions: geographic placement protocols for high-acuity patients, escalation pathways when ATB exceeds a threshold, and dedicated bed planning for ICU-overflow scenarios.
EHR Data as the Foundation
The LOS clock intervals described above — arrival to triage, triage to provider, DTA, ATB — are all derivable from ADT and nursing documentation data in your EHR. In Epic environments, these timestamps are surfaced through Cogito and available for operational reporting. In Cerner Oracle Health environments, equivalent data flows through Discern Analytics or the Cerner Health Quality Measures module. The data exists. The challenge is normalizing it into a consistent operational view that bed control coordinators, nurse managers, and CNOs can act on in real time — not retrospectively through a report that arrives 24 hours later.
This is the gap Mediflowly is designed to close. If your organization wants to understand where your OP-18 performance stands relative to top-quartile benchmarks and which interval is driving your gap, reach out to request a mapping session — we'll review your EHR data environment before the first conversation.
Benchmarking Realistically
One caution: national OP-18 benchmarks include the full distribution of US hospitals, from 50-bed critical access facilities to 1,000-bed academic medical centers. Comparing a 180-bed community hospital's OP-18 against a 900-bed academic system's performance is not operationally useful. Meaningful benchmarking compares hospitals within the same peer group — similar bed count, similar case mix, similar admission rate. The IQR (Inpatient Quality Reporting) program's publicly available stratified data provides some of this context, but the stratifications are coarse. Detailed peer benchmarking typically requires either commercial data products or a health system analytics team with access to CMS claims data at a more granular level.
The more actionable use of OP-18 benchmarks is directional: if your performance is above the national median, you have confirmed that improvement is achievable, because hospitals with similar characteristics are already performing at that level. The benchmark validates the target; your own internal interval data tells you which lever to pull first.
