What Actually Happens to Your Scan at 2 AM When the Hospital's Own Radiologist Is Asleep
Preliminary radiology interpretations explained: what happens to a scan overnight, who reads it, who signs the final report, and how to evaluate a partner.
A CT completes at 2:07 AM. The patient is back in the emergency department bay. The images are on the archive within ninety seconds. And the hospital's own radiologist is at home, asleep, forty minutes away, not on call until seven. So what actually happens to that scan between 2:07 AM and the moment somebody makes a decision about the patient? The answer is preliminary radiology interpretations — and the part most people get wrong is not how fast they are, but who is allowed to sign what at the end of it.
⚠️ Important scope note
For human patients at USA facilities, RAD365 provides preliminary reads only. Your facility's own licensed radiologist always completes and signs the final interpretation — RAD365 never does.
For veterinary and pet imaging there are no preliminary reads — veterinary studies are read as complete final reports by specialist veterinary radiology groups.
unique radiology job postings tracked March–May 2026 in AuntMinnie's job-market analysis
of those postings — roughly 1 in 5 — still unfilled after more than two months
higher radiologist attrition since 2020, per workforce tracking cited by the ACR's Neiman Health Policy Institute
the hour where coverage design stops being a policy question and becomes a clinical one
The gap nobody puts on the org chart
Every hospital has a staffing plan that works in daylight. The reading room is occupied, the worklist moves, the escalation path has a human at the end of it. The plan that matters, though, is the one covering 2 AM on a Tuesday in February — and at a large number of facilities that plan is either "the on-call radiologist wakes up" or, more honestly, "we find out in the morning."
This is not a diligence failure. It is arithmetic. AuntMinnie's 2026 radiology job-market analysis tracked 10,979 unique radiology job postings between March and May 2026 and found that about 1,470 of them — roughly one in five — were still unfilled more than two months later. The same analysis made a point that gets lost in national-average coverage of the shortage: the vacancies are concentrated in locations radiologists will not relocate to, not spread uniformly across the map. A rural critical access hospital and a coastal academic centre are not experiencing the same market.
Why the 2 AM gap is getting structurally harder
Two workforce facts sit underneath it. First, attrition: radiologists have been leaving practice at a rate roughly 50% higher since 2020, according to workforce tracking cited by the American College of Radiology's Harvey L. Neiman Health Policy Institute. Second, the supply curve: the Neiman Health Policy Institute's supply-and-demand modelling, published through JACR, projects that the radiologist workforce can remain constrained relative to demand for years even as raw headcount grows, because demand for imaging grows alongside it. That is the directional context for why overflow and after-hours preliminary capacity exists as a category at all — not a crisis headline, just a structural mismatch with a long tail.
Recruitment remains the right long-run answer. It is simply not an answer available this February. That is the gap preliminary read coverage was designed to fill.
What actually happens to the scan, step by step
- 2:07 AM — study completes. The technologist verifies the series and the study closes on the modality.
- 2:08 AM — routing. The study moves to the archive and out through the DICOM gateway to the covering worklist, with the clinical indication and relevant priors attached. In a badly configured chain, this is where minutes disappear.
- 2:09 AM — prioritisation. The study enters a queue ordered by acuity, not arrival. A suspected stroke does not sit behind a routine post-op chest.
- 2:1x AM — the read. A qualified radiologist reads the study and dictates a structured preliminary impression: the findings that change management now, stated plainly.
- Immediately — critical findings. Anything critical triggers direct voice contact with the treating clinician, logged and time-stamped. Not a message in a queue.
- Minutes later — delivery. The preliminary impression returns attached to the study in the facility's own system, where the ED physician reads it and decides.
- 7:30 AM — the final. The hospital's own licensed radiologist reads the study, completes and signs the final report. That signature never moves.
Stat callout
Roughly 1 in 5 of the 10,979 radiology jobs AuntMinnie tracked in spring 2026 was still open after two months — and the vacancies cluster in the places hardest to recruit into. For those facilities, overnight coverage is not a preference. It is the only lever with a timescale measured in weeks.
Preliminary versus final: the distinction that governs everything
| Preliminary interpretation | Final report | |
|---|---|---|
| Purpose | Support the decision being made now | Permanent diagnostic record |
| Who produces it (human/USA) | Covering radiologist | Facility's own licensed radiologist |
| Signature of record | No | Yes |
| Timing | Contracted window from study-available | Normal facility workflow |
| Critical findings | Direct voice contact, logged | Already communicated |
| Veterinary imaging | Available via vet teleradiology groups | Also available (e.g. RadsForVets) |
That last row is the one that surprises people. The human/USA boundary is a licensure and privileging boundary, not a capability one — which is why on the veterinary side a group such as RadsForVets reads all the way through to the signed final report, with its own veterinary peer review and QA track behind it.
A checklist for the facility that has never bought this before
- Define the start timestamp in writing. Order placed, acquisition complete and study-available are three different clocks producing three different averages.
- Ask for the 95th percentile overnight, per modality and priority class. The average is the number that never caused a complaint.
- Require per-study attribution. You should be able to see who read a study and when, without asking twice.
- Get the critical-finding protocol documented, including who is called, how quickly, and how the call is evidenced.
- Require discrepancy reporting against your own finals, delivered on a schedule rather than on request.
- Check the routing before blaming the reading. Prior-fetch failures and worklist misprioritisation are cheaper to fix than capacity, and they cap the benefit of any capacity you add.
That last point is worth dwelling on. A meaningful share of "slow reads" turn out to be slow routing, and the fix lives in imaging IT rather than in staffing — the same operational layer that sits behind RAD365's human radiology operations support.
About the author
Trisha Seal writes on radiology operations and imaging workflow for RAD365. RAD365 operates preliminary read coverage for USA facilities — structured preliminary impressions delivered against contracted, auditable turnaround windows, with documented critical-finding communication and per-account percentile reporting — and does not provide preliminary reads for veterinary imaging. For human patients, the facility's own licensed radiologist always completes and signs the final interpretation.
Find out what your 2 AM actually looks like
Talk to RAD365 about overnight and overflow preliminary read coverage measured against your own 95th-percentile data.
Discuss preliminary read coverage →Preliminary radiology interpretations: frequently asked questions
What a Preliminary Interpretation Is
What is a preliminary radiology interpretation, exactly?
It is a documented, structured first read of a completed imaging study, produced by a qualified radiologist before the facility's own radiologist issues the signed final report. It records the findings that matter for immediate management — the bleed, the free air, the occlusion, the fracture, the line position — and flags anything critical for direct communication. It is a clinical checkpoint, deliberately delivered early, not a shortened version of the final report.
How is a preliminary interpretation different from a final radiology report?
Three ways: authorship, permanence and purpose. The preliminary interpretation is produced by the covering radiologist to support the decision being made right now; the final report is completed and signed by the facility's own licensed radiologist and becomes the permanent diagnostic record of the study. The preliminary read informs treatment in the moment. The final report is the document the facility stands behind, bills against and archives.
What is a preliminary MRI or CT report used for?
It is used to make the next decision. Admit or discharge. Activate the stroke pathway or stand it down. Call the surgeon now or wait for the morning list. Transfer the patient or keep them. On cross-sectional studies in particular, where a single exam can run to thousands of images, the preliminary impression is often the difference between a clinician acting at 2 AM and a clinician waiting until the reading room opens.
Who is legally allowed to sign off on a final radiology report in the US?
A physician licensed in the state where the patient is located, credentialed and privileged at the facility that performed the study, and appropriately qualified to interpret it. That is a facility-and-state matter, not a vendor matter. It is precisely why RAD365 does not sign finals for human patients in the USA — that signature belongs to your radiologist, and the entire preliminary model is built around preserving it.
Does a preliminary read carry the same legal weight as a final read?
No. The final signed report is the report of record. A preliminary interpretation is a documented clinical communication that supports urgent decision-making and forms part of the record of what was known and when — which is why it should always be time-stamped, attributed and retained. Treating it as equivalent to a final report is a governance error; treating it as informal or undocumented is a bigger one.
The Overnight/Overflow Process
What happens between the time a scan is taken and a preliminary interpretation is delivered?
The technologist completes and verifies the study; it routes from the modality to the archive and out through the gateway to the covering worklist; priors and the clinical indication travel with it; the study lands in a prioritised queue where the acuity of the order determines position; the radiologist reads it and dictates a structured preliminary impression; critical findings trigger direct voice contact rather than a queued message; the impression returns to the facility's system attached to the study. Most of the elapsed minutes in a badly performing chain are routing and reconciliation, not reading.
How fast can a preliminary radiology interpretation typically be turned around overnight?
Contracted windows for emergent studies are usually set in the tens of minutes from study-available, with stat and stroke pathways tighter still and routine overnight work given a longer window. The number that matters is not the advertised average but the 95th percentile at 3 AM on a busy night, measured from a start timestamp both parties have defined in writing. Ask for that figure specifically; an average conceals the exact nights you are buying coverage for.
What triggers a hospital to request a preliminary read instead of waiting for its own radiologist?
Usually one of four situations: nobody is on site or on call overnight; the on-call radiologist is already reading and a queue has formed; volume has spiked beyond what the roster can absorb; or a vacancy, leave period or subspecialty gap has opened. AuntMinnie's 2026 job-market analysis is instructive on the last one — of 10,979 unique radiology job postings tracked between March and May 2026, roughly 1,470, about one in five, were still unfilled after two months.
How does a preliminary read get reconciled with the facility's final report the next morning?
The preliminary interpretation stays attached to the study. The facility's radiologist reads the study independently, sees the preliminary impression, and issues the signed final report. Where the two agree, the final simply supersedes the preliminary as the record. Where they differ, the discrepancy is logged and routed through the facility's discrepancy process. A functioning programme reports concordance rates back to the facility rather than waiting to be asked.
What happens if a preliminary interpretation and the final report disagree?
The final report governs, and the clinical team is notified promptly if the change affects management — that notification is the part that must never be left to chance. The discrepancy is then documented and reviewed. Rather than being a failure of the model, this loop is the model working: discrepancy tracking is measurable, auditable evidence of quality, and it should feed a formal peer review and QA programme rather than sitting in an inbox.
Scope, Compliance, and the Human/Pet Distinction
Does RAD365 provide final reads for human patients?
No. For human patients at USA facilities RAD365 provides preliminary reads only. The facility's own licensed radiologist completes and signs every final interpretation. This is a fixed boundary of the service, stated in the contract, not a limitation that can be negotiated away for a particular account.
Why do preliminary reads exist as a distinct category from full teleradiology reads?
Because the two solve different problems. A full reading arrangement transfers responsibility for the diagnostic record to an outside group. A preliminary read arrangement adds capacity at the hours when a facility lacks it while leaving the diagnostic record, the signature, the billing and the clinical relationship entirely in the facility's hands. Many hospitals want the second thing and specifically do not want the first.
How does veterinary reporting differ from human preliminary reads?
Yes, and the difference is scope rather than process. Veterinary imaging is not governed by the state medical licensure and facility-privileging framework that applies to human patients, so on the veterinary side a dedicated group such as RadsForVets provides the complete final report rather than a preliminary read — see veterinary radiology services for how that works, including its own veterinary peer review and QA track.
What credentials do radiologists need to provide preliminary interpretations?
Ask for the specifics rather than accepting a general assurance: medical qualification and radiology training, subspecialty experience relevant to your case mix, documented onboarding to your protocols and priority definitions, and a named, auditable roster rather than an anonymous pool. Any facility should be able to see who read a given study and when, on request.
Are preliminary reads regulated the same way as final diagnostic reports?
The obligations differ, but the governance should not slacken. Patient data handling, HIPAA safeguards, audit logging, retention and critical-result communication all apply in full. What differs is that the final diagnostic report carries the signature, the licensure requirement and the billing consequence — which is why the facility's radiologist owns it, in every case, for human patients.
Evaluating a Preliminary Reads Partner
What should a hospital ask before signing on with a preliminary reads provider?
Six questions, in writing. What is the start timestamp used to measure turnaround? What is the 95th percentile overnight, not the average? Who is on the roster, and can we see attribution per study? What is the critical-finding communication protocol and how is it evidenced? What is the discrepancy rate against our finals and how is it reported? And what does the escalation path look like at 3 AM when the queue backs up?
How common is it for US facilities to rely on outsourced preliminary reads today?
Common enough that after-hours coverage is a routine line in imaging budgets rather than an exception. The reason shows in the workforce data: radiologist attrition has run roughly 50% higher since 2020 according to workforce tracking cited by the American College of Radiology's Harvey L. Neiman Health Policy Institute, and AuntMinnie's 2026 analysis found the shortage concentrated in locations radiologists will not relocate to rather than spread evenly across the country. Those facilities cannot hire their way out on a useful timescale.
What's the real cost tradeoff between hiring more radiologists and using a preliminary reads service?
Recruitment gives you permanent capacity but requires a candidate to exist, accept and relocate, and it carries a fixed cost that persists through quiet months. Preliminary read coverage buys capacity only in the hours where the gap is, and can be in place in weeks rather than quarters. Most departments end up doing both — recruiting for the long run and covering nights and overflow now, because the vacancy data says the long run is genuinely long.