Emergency department throughput
Find the hours your ED is losing.
OmniFlowED lets emergency physicians log exactly where a workup stalls: which consult, which lab, which read, which bed. A case is identified only by zone, room and time. Administration gets the pattern. No patient data, no blame, no chart review.
Free during pilotPHI-free by designBuilt by a practicing ED physician
The gap
Throughput delays get discussed in the hallway. Then they're gone.
Every physician on shift knows where the department is stuck. Almost none of it reaches the people who could change it, because the only tools available were built for something else.
The EHR stops at the order
It records when an order was placed and when it came back. It has no field for the consult that was paged three times, or for the patient who waited on a gurney because there was one tech overnight.
Incident reporting feels punitive
Safety event systems ask who was involved. Physicians read that as naming a colleague, so systemic delays go unreported and the department's real bottlenecks stay invisible.
Averages don't name a cause
Median length of stay tells you the department is slow. A fix needs the next level down: which of six failure modes, on which shift, in which zone.
Why it exists
Built by a practicing ED physician.
OmniFlowED was built on shift. The taxonomy is the set of delays that actually recur in a working department, and the two hard rules (PHI-free and non-punitive) are there because a tool that breaks either one stops getting used by week two.
It is currently in pilot at a small number of departments and free for pilot sites.
Two rules that never change
- No PHI, ever. A case is identified only by zone, room and time. There is no field for name, MRN, date of birth, age or chief complaint, and the optional note runs a live guard that flags anything resembling one.
- No individuals in the analytics. Every administrative view is aggregate and centers a contributing factor, never a clinician.
For the first time I walked into a flow meeting with numbers instead of stories.
How it works
Two screens on shift. A pattern by the end of the month.
A physician logs the delay where it happened
Pick the category, pick the subtype, enter the two stage timestamps. Duration computes itself. The case is identified by zone and room. There is no field for a name, an MRN, a date of birth, or a chief complaint, anywhere in the product.
The report names a systemic cause, not a person
Every entry carries a contributing factor drawn from a controlled list: no inpatient beds, single overnight tech, pager unanswered, transport backlog. Attribution is the physician's choice, report by report: named or anonymous.
Administration reads the aggregate
The overview ranks delays by contributing factor, category, zone and shift, flags recurring patterns, and shows a needs-attention queue with a status workflow. Everything exports to CSV for your existing quality review.
For physicians
Report the system without reporting a colleague.
- Anonymous or named, your call, chosen per report rather than once at sign-up.
- Nothing to chart. A delay log is not a safety event, a peer review, or part of the medical record.
- Two steps, mid-shift: controlled lists and quick-log shortcuts instead of a free-text essay.
- You can see your own record. Every entry you file stays visible to you, with its status.
For administrators
Numbers for the capacity meeting.
- Ranked contributing factors: the systemic causes, ordered by how much time they cost.
- Zone and shift breakdowns: where and when the department actually breaks down.
- Recurring-pattern flags: the delays your physicians say happen every week, counted.
- No PHI to govern. Nothing in the dataset identifies a patient, which shortens every review.
The taxonomy
Six categories, grounded in the throughput literature
Each category tracks its own stage timestamps, so the delay duration is measured rather than estimated, and the same event is recorded the same way by every physician in the department.
| Category | Stages tracked | Example subtypes |
|---|---|---|
| Provider & consult | Requested → acknowledged → at bedside | Initial evaluation, cardiology, surgery, psychiatry |
| Laboratory | Ordered → drawn → resulted | CBC and chemistry, troponin, cultures, type and screen |
| Imaging acquisition | Ordered → performed | X-ray, CT, MRI, ultrasound |
| Imaging read | Performed → final read | CT, MRI, and the resident-to-final read gap |
| Boarding & disposition | Decision → bed or transfer ready | Floor, telemetry, ICU, transfer out |
| Other throughput | Start → resolved | Registration, transport, pharmacy, EVS, IT and EHR downtime |
Before you ask
The questions we get first
How long does logging a delay take?
Two screens. Category and subtype come from controlled lists, the duration is computed from the two stage timestamps you enter, and quick-log shortcuts pre-fill the categories a physician files most often.
What will our IT department need to do?
Nothing to install and no EHR integration. OmniFlowED runs in the browser on the devices your department already uses. Accounts are email-verified, facility-scoped and enforced by row-level security, and SSO is available for pilot sites that require it.
Who owns the data?
You do. Your department's reports export to CSV on demand at any time, including if you end a pilot, and nothing in the dataset identifies a patient.
Run a pilot in your department.
Tell us your zone layout and we'll configure it. Pilot sites pay nothing, and you keep your data as CSV whether you continue or not.