Teleradiology interpretations get treated as the whole story: a radiologist somewhere else reads a study, signs a report, and a coverage gap closes. The interpretation itself, the part that takes minutes, is also the smallest part of the workflow to get right.
What actually decides whether a teleradiology interpretation holds up is what surrounds it. Whether the priors that would change the read are there, and whether the report in front of the referring physician is preliminary or final, with someone accountable for each. How fast the read has to happen given what the study shows, and what happens the moment a finding needs attention right now. Whether the reading physician is allowed to interpret a study from that patient’s state at all, and who the signed report says did the work.
How a study physically reaches a remote reader, routing, prefetch, viewer parity, is its own infrastructure problem, and a real one. This piece assumes that infrastructure exists and stays on what happens once a remote reader actually has the study open: the clinical and accountability layer that decides whether the interpretation itself is defensible.
Why a Read Without Priors Is a Different Read
A current study read next to its priors and the same study read alone are not the same interpretation, even when the pixels are identical. Priors are how a radiologist tells a genuinely new finding from a stable one, a measurement change from measurement noise, an expected post-treatment appearance from something new. Remove them and some of those calls cannot be made responsibly at all.
The ACR Task Force on Teleradiology Practice is specific about what should happen instead of pushing a read through anyway. Interpretations, the task force writes, should be made “with complete availability of relevant collateral information, including previous imaging studies, electronic medical records, and details on the patient’s clinical symptoms and suspected diagnoses.” When a teleradiology arrangement cannot guarantee that access, the guidance does not lower the bar quietly. Instead, it directs the teleradiologist “to render a preliminary report only, outlining the limitation, which could be corrected in the final report.”
That single sentence connects two things a product team tends to treat separately. Whether priors reliably reach a remote reader is a retrieval problem, the same Query/Retrieve mechanics that move any study between DICOM systems. Whether the resulting report is preliminary or final is a clinical accountability decision made on top of that retrieval.
A platform that cannot promise priors arrive with the current study is not just slower. It pushes every reader on it toward the preliminary path more often than the workflow admits, because the alternative is finalizing a report without the comparison the interpretation actually needed.
Preliminary and Final: Two Reports, Two Kinds of Accountability
A teleradiology workflow routinely produces two reports for the same study, and treating them as the same document with different timestamps misses what each one is for. The ACR Practice Parameter for Communication of Diagnostic Imaging Findings defines the preliminary report as one rendered for directing immediate patient management. The parameter is specific that this report “very likely will contain limited or incomplete information” and “should not be expected to contain all the information subsequently found in the final report.” The final report is a different kind of document: “the definitive documentation of the results of an imaging examination or procedure.”
That difference in purpose creates a difference in duty. Whoever signs the preliminary report is accountable for directing immediate management on incomplete information, honestly labeled as such. Whoever signs the final report is accountable for the record that stands.
When the two reports do not match, closing that gap is not optional. The practice parameter requires that a significant variation in findings between the preliminary and final interpretations “should be reported in a manner that reasonably ensures receipt by the referring or treating physician/health care provider.” The discrepancy gets documented into the final report rather than quietly absorbed by it.
This is also where a workflow needs a defined process for producing an addendum, not an informal one. A final report that needs correcting after new history or a discordant reread surfaces does not get silently rewritten. It gets an addendum, tied to the original with its own timestamp. The record then shows what was known and signed at each point, instead of one report that looks like it always said the corrected thing.
DICOM’s own structured reporting mechanics exist partly to make findings like these queryable later. The addendum process itself, though, is a workflow discipline a reading group has to define, not something a viewer or archive enforces on its own.
Turnaround Expectations for Teleradiology Interpretations Change With Urgency, Not Distance
A common assumption is that teleradiology needs its own turnaround standard, something looser than on-site coverage to account for the distance. The ACR task force rejects that framing directly. Turnaround times for teleradiology interpretations should be set in accordance with accepted hospital and departmental requirements. The guidance asks for times “commensurate with other intradepartmental policies” that “should not be more or less stringent than for on-site radiology except for compelling patient-centered reasons.”
Distance is not a reason to read slower. Urgency is the variable that is supposed to move the clock.
That puts real weight on how a study’s priority survives the trip from order to remote worklist. A STAT flag has to actually reorder a reading queue, not just decorate an entry, and it has to keep meaning STAT once a study crosses from the acquiring site’s systems into a remote reader’s. A teleradiology arrangement that reads everything in arrival order, regardless of where the flag originated, has quietly adopted a turnaround standard the ACR guidance does not endorse. Nothing about the architecture will announce that it happened.
Critical Findings and the Documented Closed Loop
A finding that needs immediate action, a pneumothorax, a significantly misplaced line, cannot sit in a routine report queue waiting for a referring physician to open it. The same ACR practice parameter separates this kind of nonroutine communication into its own category and is specific about what closes the loop. It must be handled “in a manner most likely to reach the attention” of the treating or referring physician “in time to provide the most benefit to the patient”. Any electronic method used has to “provide documentation of receipt to demonstrate that the communication has been delivered and acknowledged”.
Delivered and acknowledged are two different events, and a workflow that only proves the first has not actually closed the loop. A message that went out is not the same as a message someone received and confirmed. The practice parameter requires documenting the method, the time, and who received it, kept in the report or the patient’s record rather than an informal log.
That matters more in a teleradiology arrangement than in a department where everyone works in one building. The reading physician and the treating team are often people who have never met. The documented handoff is the only evidence, after the fact, that a critical finding reached someone who could act on it.
Licensure and Credentialing: Why the Reading Site Is Not a Free Choice
Where a teleradiologist happens to be sitting is close to irrelevant to whether they are legally allowed to read a given study: what matters is where the patient is. States mandate and enforce medical licensure through their own legislation. The same ACR task force is direct about the consequence for a remote read: “medical practice is considered to occur at the location of the patient”. That typically means licensure in the state where the study was acquired, not necessarily the state the reading physician happens to be sitting in.
Credentialing runs on a related but separate track. A hospital normally vets every physician who reads for it directly, through its own medical staff process. Federal rules allow a narrower path for telemedicine specifically. Under 42 CFR 482.22, the governing body of the receiving hospital may have its medical staff rely on the credentialing and privileging decisions the distant-site hospital already made.
That reliance is not a default, and the conditions are the substance of the rule. It exists only through a written agreement with the distant-site hospital, and the agreement itself has to ensure each condition is met. The distant site has to be a Medicare-participating hospital, the physician has to be privileged there, and the distant site has to supply the receiving hospital a current list of those privileges. Licensure ties it back to the patient’s state: the physician “holds a license issued or recognized by the State in which the hospital whose patients are receiving such telemedicine services is located.”
Licensure by patient location and credentialing by proxy are two different mechanisms solving two different problems, and a workflow that treats reader qualification as one checkbox is collapsing a distinction the regulations keep deliberately separate.
The Attribution Trail on a Signed Report
A signed report has to say, unambiguously, who actually interpreted the study. That sounds obvious until you meet the failure mode the ACR has had to name explicitly. A radiologist signs a report initially read by someone else, without personally reviewing the images, and the task force calls that ghost reading. The College calls that conduct unethical and, depending on circumstances, fraudulent, when done “in a manner that causes the reader of that report to believe that the signing radiologist was the interpreter.”
Naming it plainly is the point: attribution on a radiology report is not a formality. It is a claim about who is professionally and legally responsible for what the report says.
The record that supports that claim is mundane by design. The ACR practice parameter’s suggested report format calls for the date and time of dictation, transcription, and finalization as distinct fields. A single date stamped on the whole document hides when each stage of authorship actually happened.
Those timestamps are a narrower record than the access logging a teleradiology platform keeps for every query, retrieve, and viewer session touching a study. The order and the signed report sit somewhere else entirely, with the radiology information system on the other side of the PACS seam, which stays the system of record for both. The attribution trail is about the report itself: who is named as the interpreting physician, when they dictated it, when it was finalized, and whether an addendum changed it afterward. Those three records answer different questions, and one does not substitute for another.
What This Means for Your Platform
None of the six requirements above are things a viewer renders or a network transports. They are workflow and record-keeping decisions sitting on top of whatever infrastructure gets a study from an acquiring site to a remote reader. Licensure and credentialing make up the one requirement on that list that no product feature decides. The other five turn into questions a buyer can put to a vendor, and a platform earns its place by making each answer easy to get right rather than easy to skip.
- Does the platform make clear, at the point of interpretation, whether relevant priors actually arrived.
- Does it distinguish a preliminary report from a final one structurally, not just by label, and support an addendum process rather than a silent overwrite.
- Does a priority flag survive the trip from order to a remote reader’s worklist intact.
- Does a critical-finding notification get logged as delivered separately from acknowledged.
- Does the final report carry the dictation, transcription, and finalization timestamps that make its own authorship traceable.
EBM mAIn PACS® sits at the infrastructure layer this piece assumes rather than at the licensure and credentialing decisions above it, which stay a partner’s and a reading group’s responsibility to get right. Native Query/Retrieve is the service that puts priors in front of a remote reader in the first place, and the EPS Pi edge appliance adds storage, routing, worklist, and reporting alongside it. UDE gives that reader a mobile reading workstation built for reading outside a fixed room.
What a partner builds or licenses on top of that, for reading-room and remote-access solutions, still has to answer the five questions above with a specific mechanism. Preliminary versus final accountability, the critical-findings loop, and the attribution trail on a signed report are workflow disciplines before they are features. They belong to the people and the process running on top of the infrastructure.