How does imaging actually move through an emergency department?

How is an ED order different from an outpatient order?

An outpatient order is written against a question that has been stable for weeks and can wait for the right protocol. An ED order is written against a decision the physician has to make before the patient leaves the department. Same modality, same body part, different job, and the history field carries far more weight.

The reader cannot see the patient. Everything else follows from that. In outpatient work the deficit is tolerable, because the question has usually been sitting for weeks and the priors are usually in the archive. In the ED the reader is helping sort a patient whose story changed twenty minutes ago, working from an order field that says abdominal pain. Abdominal pain is not a clinical question. Concern for perforation in a patient with two days of pain and a rigid abdomen is a clinical question, and it changes which series get worked hardest, which reconstructions get pulled, and how much time the reader spends at the lung bases before calling a study negative.

Timing cuts both ways. The study is acquired fast, often before labs are back and sometimes before anyone has settled what the differential is, so the images arrive carrying a question that was already stale when the technologist started the scan. Then the department moves on. By the time the interpretation returns, the patient may have improved, deteriorated, or been reassessed by a second physician who never saw the original order and does not know what it said. Some departments let the physician update the indication after acquisition. Most do not, because nobody owns that field once the order is signed. Find out who at your facility can edit it, and whether the reading side ever sees the edit.

The urgency label has the same problem. A study ordered routine at the top of a shift can become the study the whole department is waiting on by the time it is acquired, and the label almost never follows that change. Decide who is allowed to move a study up after acquisition, and decide how the reading side finds out it moved. If the answer is that nobody can, say so in plain words to the physicians who order. Otherwise they keep assuming that a phone call quietly re-sorts a worklist somewhere, and that assumption holds right up until the one night it does not.

What does the ED need from the read, and what does the final report add?

The ED needs the answer to one question: does this finding change what happens to the patient in the next hour. Admit, discharge, transfer, operate, or call a consultant. The final report answers a wider set of questions, for the record, for coding, for the clinic that sees the patient next week, and for whoever reviews the encounter later.

Watch what an emergency physician actually does with a read. They are looking for the finding that flips the decision, and they will often act on a two sentence answer delivered over the phone. The rest of the report is not useless to them, it is just not the thing they needed at that minute. The final report is written for a wider audience: the admitting team reading it at midnight, the surgeon deciding on an approach, the coder, the clinic that inherits the patient after discharge, the quality committee that pulls the chart next spring. Those readers do need the incidentals, the qualifiers, and the comparison the ED never asked about.

The awkward part lives in the space between the two. A study can be entirely negative for the question that was asked and still contain something that needs a follow-up appointment nobody has made. The ED physician saw the answer to their question and moved on, correctly. The final report carries the other finding into a chart the patient has already left. Departments handle this badly and know it. The usual fix is a named list: which categories of incidental finding get flagged to the ED before discharge, which get routed to a follow-up owner, and which simply live in the report. Write that list with the ED, not for it, or it will be ignored by the people it governs.

Say the awkward thing out loud in your own policy. Most emergency departments run on preliminary interpretations for disposition and treat the final as a document that arrives later for somebody else. That is a reasonable operating model, and arguing with it wastes a meeting. It becomes a problem only when nobody has written down who that somebody else is, which is the state most departments are actually in. Ask a few colleagues who reads the final on a patient who has already gone home, and count how many different answers come back before someone says they will check. The model is not the defect. The unnamed owner is.

Why is trauma routed separately from the rest of the department?

Because the recipient, the sequence, and the acceptable latency are all different. A trauma activation has a team assembled before the patient arrives, a surgeon who wants specific answers in a specific order, and a route where the read may be delivered by voice into a room rather than sent to an inbox. Treating it as a fast ED study breaks it.

A trauma activation inverts the normal order of a reading. The team wants the things that change what happens in the next ten minutes, and it wants them in the sequence the resuscitation needs, not the sequence in which a report is written. There is a surgeon standing in the room waiting on a narrow question. Meanwhile the study itself is large, arrives in pieces, and may still be reconstructing while the first series are already open. A reader who works it like an ordinary study, from the top down, will deliver a complete interpretation at the exact moment it stops being useful.

Because the answer usually goes into the room by voice, trauma is where the record breaks. The reader says what they see, the team acts on it, and the resuscitation absorbs everyone's attention for the next hour. Nothing gets typed. Later the chart holds a report finalized well after the decisions were made and no documentation that the verbal read happened at all, who gave it, or what it said. When somebody reconstructs the case, the timeline reads as though the team acted with no imaging information. This is not a rare edge case. It is the normal outcome unless the pathway assigns a specific person to write the verbal read into the chart, with a name and a time, while the trauma is still running.

So write the trauma route as its own document. Which studies it covers, who reads them, in what order the findings are wanted, how the reader is reached, who is in the room, who records the verbal read, and what happens when the activation and a stroke alert land in the same ten minutes. That last one produces the most disagreement and gets settled last, usually after it has already happened once. Settle it first. A pathway that has never been tested against a second simultaneous emergency has not been tested against anything.

What does a reader need from an ED order to be useful?

One line of decision, one line of story, and one line of who to call. The reader is not asking for a history and physical. They need to know what you are trying to rule in or out, when it started, what was already done to the patient, and who picks up the phone at this hour.

Order templates are where good intentions go to die. A dropdown offering pain, trauma, follow-up, and other will produce orders that say pain, because at three in the morning the physician is taking the fastest path through a screen while holding several other tasks in their head. That is not carelessness, it is the interface doing exactly what it was designed to do. If your indication field is a picklist, the reads you get back will be as thin as the picklist. A free text line with a two word prompt gets better results than a long taxonomy, and it takes less time to fill.

The other half is what a reader can do with a poor order, which in most arrangements is very little. There is often no fast, documented way to ask the ED a question about a study in progress. The reader either guesses at the intent or stops to make a call that costs several minutes and lands with whoever is nearest the unit phone. Give the reading side a defined route to ask, with a defined recipient, and log it. The questions turn out to be fewer than people fear, while what a wrong guess costs never shows up in any log, which is exactly why the guessing continues.

• The decision in play, stated as a decision: rule out perforation before calling surgery, not abdominal pain.
• Onset and trajectory: when it started and whether it is getting worse.
• Mechanism when there is one, including direction of force and what was struck.
• What has already been done: contrast given elsewhere, a reduction attempted, a line placed, a drain in situ.
• Known conditions the images will otherwise look wrong without: a prior resection, a shunt, hardware.
• Whether a prior study exists anywhere, and whether it can be reached, which are two separate questions.
• Who to call, by name or by role, and where that person physically is at this hour.

Where is the line between an urgent finding and a critical one, and who gets called?

In the ED the distinction gets compressed, because the physician who ordered the study is usually a few steps away and reachable. The real question is not urgent against critical. It is whether the patient is still in the department, and whether the person who ordered the study is still on shift. Both answers change who gets called.

Tiering lists written for outpatient imaging translate poorly into an ED. A finding that would trigger an immediate call from a clinic study can be the expected finding in a patient already on a monitor with a surgeon at the bedside. Calling it in adds nothing and trains the department to treat calls as noise. The reverse happens too: something unremarkable in isolation matters a great deal in a patient about to go home with a prescription. The reader usually cannot tell which situation they are in, because the order does not say. A line about intended disposition solves more of this than any refinement of the tiering list.

Who gets called is a roster problem, and EDs have a specific version of it. The physician who ordered a study overnight may be gone by the time a finding surfaces on a later study, in an addendum, or in a comparison that arrived at seven. The name in the ordering provider field then belongs to someone who left the building. What the reader needs is the role currently responsible for that patient in that department, written as a role rather than a name, because roles survive shift change and names do not. The charge nurse desk usually knows the answer, and the electronic record usually does not.

Agree in writing on what happens when the reader reaches someone who is not a physician. A unit clerk who writes the finding on a note has not closed a loop, and the reader who hangs up believing otherwise will be surprised later. If your department wants the charge nurse to be an acceptable recipient because that person can act and can find the physician quickly, name the charge nurse in the policy. If your department does not want that, say so with the same clarity, because otherwise the reader decides it alone at two in the morning with a phone in one hand. Unwritten assumptions about who may take the call are what reviews keep turning up.

Why do ED imaging results go missing at shift change?

Because the results pathway is keyed to a person, and at shift change the people change while the studies do not. Anything still in flight at handoff depends on a verbal sign-out the reading side never hears. Results that arrive after the patient has been discharged land in an inbox belonging to someone who went home hours ago.

Here is the failure, in the form it actually takes. A patient is worked up, the ED physician gets what they need, and the patient goes home. Hours later a final report is signed with something in it that was not in the preliminary, or the finding was there and the discharge happened first. The report routes to the ordering provider field, which holds the name of the physician who worked that shift. That physician is asleep. The next time they open the inbox they are on a different shift with a different queue, and this result is one line in a long list. The primary care office never received the study because the study was never theirs. Nobody did anything wrong at any step, which is precisely why this pathway survives review after review.

The second failure is quieter. A verbal or preliminary interpretation is given, the physician acts on it, and it never reaches the chart in a form anyone can find. The disposition note says the scan was negative. The signed report, when it arrives, says something slightly different, and there is no record of what was actually communicated at the moment the decision was made. Both documents look fine on their own. Together they cannot be reconciled, and the person trying to reconcile them is usually doing it months later with no memory of the shift. Requiring the ordering physician to record who read it, when, and what was said costs one line.

The fix is procedural and nobody enjoys it. Pending imaging goes on the handoff list by study and patient location, not as a general remark that some scans are outstanding. The incoming physician acknowledges each one. A named role owns the queue of results that arrive after discharge, and that role is one that is always filled rather than the physician who happened to sign the order. Test it by pulling a handful of results that landed after a discharge and asking who opened them and when. The first time a department runs that check, the answer is uncomfortable and the meeting goes quiet.

• Every pending study named at handoff, with the study and the patient location, not a general remark that scans are outstanding.
• A named role, always filled, that owns results arriving after the patient has left.
• A rule for findings communicated close to shift change: they go to the incoming physician as well.
• A place in the chart for the verbal read, recording who gave it, who took it, and when.
• A defined route to the patient's follow-up clinician once the ED no longer owns the patient.
• A periodic check of results that landed after discharge, with the time each one was opened.

Sources and scope

• American College of Radiology, practice guidance on communication of diagnostic imaging findings
• American College of Emergency Physicians, policy statements on imaging in emergency care
• American College of Surgeons Committee on Trauma, standards for trauma center verification and resuscitation care
• Radiological Society of North America, professional education on radiology reporting and communication
• HL7 International, standards for clinical order and result messaging