Report turnaround time is the metric that gets the most attention in radiology performance discussions. Department heads track it, hospital administrators set targets for it, and when it goes in the wrong direction, radiologists are usually the first people asked to explain why. That framing misidentifies where most of the delay actually lives.
When you break the TAT clock down into its component intervals, the radiologist's reading and dictation time is often a surprisingly small fraction of total elapsed time from study order to signed report. The other intervals are system, process, and staffing problems. A radiologist working faster does not close those gaps. Here is where the time actually goes.
1. Study Queue Position and Worklist Order
The single largest controllable contributor to TAT variability in most departments is not reading speed. It is how long a study sits unread before a radiologist picks it up. A FIFO (first-in, first-out) worklist means that a time-sensitive study arriving in the middle of a packed queue waits behind every study ordered before it. In a department running 150-plus studies per shift, a study ordered at 10 AM might not be read until 1 PM or later if the shift started at 7 AM with a full queue.
Manual prioritization exists but introduces its own delays. A technologist who spots something on a scout image can flag a study for the radiologist, but this depends on the technologist being available, being confident enough to escalate, and the radiologist seeing the flag quickly. This chain breaks down under volume pressure, which is exactly when it matters most.
Automated triage that repositions studies based on AI-detected findings can close this gap without adding a step for anyone. The study arrives, the pre-read runs in the background, and if a flag-worthy finding is detected, the study moves up the worklist before the radiologist's cursor even gets near it. The elapsed time from study completion to radiologist attention shrinks independent of how fast anyone works.
2. Prior Study Retrieval
Comparison to prior studies is a standard part of radiology reporting. A lung nodule that is stable over 24 months is a very different finding from a nodule that is new or growing. Reading without the prior means either giving a less informative report or flagging the absence of comparison, both of which create downstream friction.
The problem is that retrieving prior studies from archiving systems, particularly when the patient has been seen at another institution or when the department has fragmented storage across legacy PACS archives, can add 10 to 20 minutes to effective reading time. The radiologist is not reading. They are waiting for DICOM data to move across a slow archive link or manually requesting an outside study from medical records.
This is a workflow architecture problem. Departments that have invested in a vendor-neutral archive with aggressive pre-fetching of relevant priors based on scheduled orders substantially reduce this delay. Departments that have not are asking radiologists to absorb wait time that has nothing to do with their reading capability.
3. Dictation and Report Completion Friction
The actual dictation-to-signed-report interval is where AI drafting has its clearest impact. The steps between reading a study and having a signed report involve: activating the dictation system, choosing or loading the correct template, dictating findings and impression, reviewing the transcript for speech recognition errors, and finalizing. For a normal chest X-ray, this chain takes somewhere between three and seven minutes for most radiologists.
Three to seven minutes per study, multiplied across 150 studies in a shift, is 450 to 1,050 minutes of dictation work. That is the range where pre-populated AI drafts that the radiologist edits rather than originates reduce per-study time meaningfully. A well-structured draft for a routine normal study that requires light editing takes 60 to 90 seconds to verify and finalize. The savings are real and compound across a high-volume shift.
The friction that does not get solved by AI drafting: report finalization that requires administrative steps, addendum workflows for critical finding communication, and the interruptions that fragment dictation sessions. These are process design issues. A radiology department that has a clear critical finding communication protocol with defined response time expectations handles interruptions predictably. Departments that do not have a defined protocol deal with ad hoc calls during reading sessions that break concentration at unpredictable intervals.
4. Shift Structure and Coverage Gaps
Radiology TAT has a structural peak-and-trough problem that individual reading speed cannot address. Order volume in most acute care settings peaks in the morning and again in the late afternoon as admissions and outpatient imaging concentrate. Reading staff are typically scheduled in full shifts that do not align with the order peaks. The result is that studies ordered during the order peak accumulate faster than they are read, and studies ordered during the mid-shift trough get read faster than they arrive.
This creates a batch effect. The morning order peak builds a queue. The radiologists read through it during the morning shift. TAT for the studies ordered at 8 AM may be short. TAT for the studies ordered at 11 AM, after the queue has built, is longer because those studies are sitting behind a two-hour backlog. From the outside, the TAT numbers look erratic. From the inside, the pattern is entirely predictable given the scheduling structure.
Coverage gap periods, particularly overnight and weekend, create the most extreme TAT delays. A non-urgent CT ordered at 9 PM in a department with overnight radiology coverage only for urgent reads might wait until the next morning. This is not a reading-room problem. It is a coverage structure problem. Teleradiology solutions exist for overnight coverage and have been adopted by many smaller departments, but even with teleradiology, the priority logic for what gets read overnight matters for how long non-urgent studies wait.
The Takeaway for TAT Improvement Programs
TAT improvement initiatives that focus on encouraging radiologists to read faster, set individual performance targets, or optimize dictation technique are addressing the smallest piece of the problem. The measurable gains come from reducing queue wait time through better prioritization, reducing prior retrieval delay through pre-fetching and archive architecture, reducing per-study dictation time through AI-assisted drafting for routine studies, and matching coverage structure to order volume patterns.
None of those improvements are things a radiologist can implement by working differently. They are system changes. The departments that have moved their TAT numbers meaningfully have done it by treating the non-reading intervals as seriously as the reading interval. The reading room is not the bottleneck. The bottleneck is everything upstream of it.