What the study found
Researchers at Poznan University of Medical Sciences in Poland, led by Patryk Konieczka, ran a retrospective cohort study of 13,533 emergency department CT exams performed between July 2025 and June 2026, comparing scan-to-complete-report CT scan turnaround time during periods with and without an on-site CT radiologist. The study, published in the European Journal of Radiology, defined off-site periods as nonholiday weekdays from 3 p.m. to 7 a.m., plus weekends and public holidays — the hours most hospitals already staff differently — as reported by Radiology Business.
The unadjusted numbers are stark: a median reporting time of 81 minutes when no radiologist was on-site, versus 58 minutes when one was. After adjusting for patient, temporal, workload, and exam characteristics — controlling for the fact that off-site hours also tend to be busier, more complex, and later at night — the absence of an on-site radiologist was still associated with roughly 34% longer turnaround time (Konieczka et al., European Journal of Radiology, 2026).
The effect wasn't uniform. It was smallest for routine noncontrast head CT — about 28% longer without on-site coverage — and largest for abdominopelvic CT, at nearly 46% longer. The authors also ran sensitivity analyses: redefining coverage by the radiologist's actual reporting location rather than the staffing roster, and isolating exams performed near shift-change boundaries, produced gaps of similar or larger size, suggesting the roster-based 34% figure is, if anything, a conservative estimate.
Why "on-site vs. off-site" is the wrong frame
It's tempting to read this as a case for physical presence: put a radiologist in the building, and reports move faster. But look at where in the workflow the delay actually accumulates. Image acquisition time doesn't change based on who's reading the study. What changes is everything downstream of the scan completing — how quickly the radiologist picks it up, how quickly a question from the ED gets answered, how quickly a first-pass read becomes a structured, complete report.
An on-site radiologist compresses that chain because they're reachable in person: faster callbacks on ambiguous findings, faster coordination with the ordering physician, fewer handoffs between a remote reading queue and the local care team. None of that is intrinsic to geography — it's a proxy for how quickly a draft report gets produced and finalized once images land. A radiologist reading from off-site at 3 a.m. isn't slower because they're elsewhere; they're slower because the structural supports that compress drafting and communication time aren't there.
That reframing matters for staffing models, because physically relocating radiologists to every site, every overnight shift, isn't realistic for most hospitals and imaging groups. The more useful question is: what else closes the same gap in drafting speed, without requiring a body in the building?
The gap by exam type
The size of the turnaround-time gap tracked with how much manual drafting work a study requires — simple studies lost less time, complex ones lost more:
| Exam type | Adjusted TAT gap without on-site coverage |
|---|---|
| Noncontrast head CT | ~28% longer |
| All ED CT exams (overall) | ~34% longer |
| Abdominopelvic CT | ~46% longer |
Head CT for stroke or trauma rule-out is a relatively fast, templated read — there's less room for coverage model to matter. Abdominopelvic CT involves more structures to review, more incidental findings to characterize, and more report text to draft, so the exam type most dependent on radiologist time is also the one where a coverage gap costs the most minutes.
Where AI-assisted drafting fits
If the real driver is drafting and communication speed rather than geography, AI-assisted report drafting addresses the same bottleneck from a different angle. Instead of compressing the chain by having a radiologist physically present, a structured draft report is generated as soon as the images are available — for the straightforward head CT and the more involved abdominopelvic study alike — so the radiologist's time goes to review and correction rather than dictation from a blank page. In xAID's workflow, that draft gets an in-house radiologist review on every preliminary before it reaches the client's reading radiologist ready-to-sign — so overnight or off-site coverage doesn't have to mean a slower report, whether or not anyone is in the building when the scan finishes.
Frequently asked questions
Does on-site radiologist coverage reduce CT scan turnaround time in the ED?
Yes, according to a 2026 retrospective study of 13,533 emergency CT exams. Median scan-to-report turnaround time was 81 minutes when no radiologist was physically on-site, versus 58 minutes when one was. After adjusting for patient, workload, and exam characteristics, the absence of an on-site radiologist was associated with roughly 34% longer turnaround time.
How much faster is CT reporting with an on-site radiologist?
In the study, the unadjusted median gap was 23 minutes (81 vs. 58 minutes). The adjusted effect — which controls for case mix and timing — was about 34% longer turnaround without on-site coverage, with the size of the gap varying by exam type.
Which CT exam types see the biggest turnaround time gap without on-site coverage?
The gap was smallest for routine noncontrast head CT, at about 28% longer without on-site coverage, and largest for abdominopelvic CT, at nearly 46% longer. More complex studies, which typically require more image review and more communication with the ordering team, showed the largest turnaround time gaps.
Can AI-assisted reporting close the on-site vs. off-site CT turnaround time gap?
The study itself did not test AI; it compared physical presence, not drafting tools. But the mechanism it describes — slower turnaround when review and drafting aren't immediate — is exactly what AI-assisted report drafting targets, by producing a structured preliminary draft as soon as images are available, independent of whether the reviewing radiologist is in the building or reading overnight from off-site.
Source: Konieczka P, et al., "After-hours on-site radiologist availability and emergency CT reporting: a roster-based cohort with staffing-boundary analysis," European Journal of Radiology (2026), doi.org/10.1016/j.ejrad.2026.113332, as reported by Radiology Business. Figures are rounded as reported.