Over the last fifteen years, the market for teleradiology services has undergone a transformation that most healthcare executives have welcomed without fully examining. After-hours and overflow radiology reads have been commoditized from below. Price compression has been significant. For procurement teams and hospital administrators, this looked like a straightforward win: the same service, lower cost, problem solved.
What actually happened is more complicated. The market stopped distinguishing between a reporting service and a medical service. And those are not the same thing.
That distinction is not semantic. It determines what you actually purchased, what liability you are carrying, and what happens to your institution when something goes wrong.
The Dashboard Problem
When organizations evaluate teleradiology vendors, they measure what is measurable. Turnaround time. Report volume. Cost per read. System uptime. These are real numbers. They are also the wrong numbers if you are trying to evaluate the clinical quality of what you bought.
Turnaround time tells you how fast a report was generated. It tells you nothing about whether the interpretation was correct, whether the clinical context was integrated, whether the protocol was appropriate for the question being asked, or whether a finding that warranted an urgent call was handled as such. A teleradiology operation can produce excellent TAT numbers while quietly accumulating clinical risk that will not surface until a patient is harmed or a claim is filed.
The commodity market selected for speed and volume. It did not price clinical accountability. Over time, buyers trained to evaluate on price and TAT got exactly what they selected for.
What a Reporting Service Delivers
A reporting service delivers a document. A radiologist, working remotely, reads the images and generates a report. The encounter closes. The report enters the system. From the vendor's perspective, the transaction is complete.
This model works adequately under normal conditions. Routine studies, clear findings, straightforward clinical questions. The radiologist interprets, the report is filed, the referring physician acts on it or does not. Nobody calls. Nothing escalates. The dashboard looks clean.
The problem is that teleradiology was not designed for normal conditions. It was designed for the margins. After-hours coverage. Overflow during peak volume. Cross-site reads when on-site capacity is insufficient. These are not routine scenarios. They are structurally the highest-risk studies in the system, arriving at the moments when institutional support is thinnest and clinical stakes are often highest.
A reporting service is not built for those moments. It is built for throughput.
What a Medical Service Does Differently
A medical service does something harder and less visible. It integrates clinical context into the interpretation. It applies protocol judgment before the study is acquired, not just after. It recognizes when the right response to an imaging finding is a phone call rather than a line in a report. It maintains the infrastructure to make that call, track it, and document it. It carries accountability for what the images mean, not just what they show.
The difference between these two models is almost invisible at procurement. Both deliver reports. Both have credentialed radiologists. Both can show you uptime statistics. The gap emerges at the edges: the missed finding on a night study, the incidental result that was documented but never escalated, the protocol error that produced a non-diagnostic study nobody flagged, the malpractice claim that turns on whether the radiologist had a documented process for urgent communication or simply filed a report and moved to the next case.
At that point, the price difference between the commodity provider and the medical service provider looks very different than it did at contract signing.
Where Diagnostic Errors Actually Cluster
Diagnostic errors in radiology are not randomly distributed across study types, times of day, or patient populations. The published literature on radiology error is consistent on this point: errors cluster around perceptual misses in complex or subtle cases, around failures to integrate clinical context into interpretation, and around communication breakdowns where a finding was identified but not acted upon appropriately.
These conditions are not evenly distributed across a teleradiology operation's volume. They concentrate in exactly the studies that teleradiology was built to absorb. The 2am trauma scan. The overflow MRI read on a patient with an atypical presentation. The cross-site study where the prior imaging is unavailable and the protocol was set by a technologist working without radiologist input.
The buyer who selected on price compressed the clinical margin precisely where the risk is highest. This is not a theoretical concern. It is the pattern that emerges when you examine the cases that result in claims, regulatory findings, and patient harm events in teleradiology operations.
The Credentialing and Accountability Gap
There is a structural accountability problem in commodity teleradiology that rarely surfaces in vendor conversations. When a radiologist employed by a teleradiology company reads a study for an institution, the question of who owns the clinical accountability for that interpretation is not always clearly resolved.
Malpractice coverage, institutional credentialing, and regulatory jurisdiction can point in different directions. The radiologist may be credentialed at the institution or may be operating under a broader agreement that has not been subjected to the same scrutiny as on-site credentialing. The indemnification chain in the service agreement may not align with the actual liability exposure. These are not hypothetical concerns. They are the questions that surface when a case is examined by a plaintiff's attorney or a regulatory body, and they are questions that procurement processes focused on price and TAT are not designed to ask.
A medical service model resolves these questions in advance. Credentialing is institution-specific. Accountability is documented. The quality framework is auditable. The radiologist reading the study at 2am has the same institutional standing as the radiologist reading elective studies during the day, and the organization can demonstrate that if it needs to.
The Protocol Governance Problem
One of the least visible failure modes in teleradiology is protocol governance. In a well-run on-site radiology operation, the radiologist has authority over imaging protocols. Studies are acquired according to protocols that reflect current clinical evidence, institutional patient population, and equipment capability. When a protocol is wrong for the clinical question, the radiologist can intervene before the study is acquired.
In many teleradiology arrangements, this authority does not transfer cleanly. The radiologist receives images that were acquired according to protocols they did not set and may not have reviewed. If the protocol was inappropriate for the clinical question, the resulting study may be non-diagnostic or misleading. The radiologist can note this in the report. They cannot undo the acquisition. The patient may need to be reimaged, with additional cost, delay, and radiation exposure where applicable.
At scale, across a multi-site teleradiology operation, protocol drift is significant. Studies are acquired inconsistently. Image quality varies. The radiologist's ability to deliver a high-quality interpretation is constrained upstream by decisions they were not part of. This is an operational and clinical governance problem, not a technology problem, and it is one that commodity teleradiology models are not structured to address.
The financial cost of this problem is also underappreciated. Non-diagnostic studies require repeat imaging. Repeat imaging generates additional cost, additional scheduling burden, and in modalities involving ionizing radiation, additional patient exposure. When these events are tracked and attributed correctly, the apparent savings from a lower per-read rate frequently do not survive the full accounting. The institutions that have done this analysis honestly tend to reach the same conclusion: the cost of inadequate protocol governance is not visible in the teleradiology invoice. It is distributed across the rest of the operation.
What the Right Architecture Looks Like
The organizations that have built teleradiology as a medical service rather than a reporting service share certain characteristics. Radiologist involvement in protocol governance is not optional. Credentialing is institution-specific and auditable. Urgent communication pathways are documented and tracked. Quality frameworks include peer review, discrepancy tracking, and regular audit. The infrastructure exists to answer hard questions about any specific study: who read it, what their credentials were, what protocol was used, whether any finding warranted escalation, and what happened next.
This architecture costs more to build and more to operate. It also produces a fundamentally different risk profile for the institution purchasing the service. The premium is not for faster reports or a more sophisticated user interface. It is for the accountability structure that determines what the service actually is when it matters.
The Risk Transfer Nobody Priced
Cheap teleradiology is not a bargain. It is a transfer of risk. The institution that selected on price has not eliminated the clinical risk inherent in remote radiology coverage. It has transferred that risk in a direction that is rarely explicit at procurement: toward the patient who receives a substandard interpretation, and eventually back toward the institution when that interpretation is examined.
The organizations that understand this are the ones that have been through the examination. A claim that turns on whether the overnight radiologist had a documented urgent communication process. A regulatory review that asks whether the radiologist reading cross-site studies was credentialed at the specific institution. An adverse event that traces back to a protocol that produced a non-diagnostic study nobody flagged.
At that point, the price difference between what they purchased and what they should have purchased is not a line item. It is the finding.
A Note on How to Evaluate What You Have
If your organization is currently operating under a teleradiology arrangement, the evaluation questions are straightforward. Can you audit the credentialing status of the radiologists reading your studies at any given moment? Is there a documented urgent communication protocol and a record of its use? Who governs your imaging protocols and when were they last reviewed? What does your indemnification chain look like and has it been reviewed by someone who understands radiology liability specifically?
If those questions are difficult to answer, you have a reporting service. You may have been told you have a medical service. The difference will be visible when you need it to be.
This is the problem Boridy Imaging Advisory was built to solve. If you are designing, evaluating, or restructuring a teleradiology arrangement and want an independent clinical perspective on how it is structured, I am glad to have that conversation.
The AI Question
A newer dynamic is entering the teleradiology conversation. AI-generated preliminary reports are being positioned by some vendors as a solution to the throughput and cost pressures that drove commoditization in the first place. The technology is real and in some contexts clinically useful. The risk framing deserves scrutiny.
An AI-generated preliminary report is a tool. Like any tool in a clinical environment, its value depends entirely on the governance structure around it. Who controls the template. Who audits the output. Whether the radiologist is exercising independent clinical judgment on the study or reviewing and approving a document the system generated. Whether the accountability for the final interpretation sits clearly with the radiologist or is diffused across a human-machine workflow that is harder to examine when something goes wrong.
Done well, with proper radiologist control and a robust audit infrastructure, AI assistance can reduce cognitive load and support quality. Done poorly, it becomes another layer of commoditization. The preliminary report optimized for throughput, approved under time pressure by a radiologist who did not meaningfully engage with the images, is not a better product. It is a faster version of the same risk.
The question to ask any vendor offering AI-assisted teleradiology is the same question you should ask about any teleradiology service: where does the clinical accountability sit, and can you demonstrate it when you need to? If the answer is unclear, the technology is not the issue. The governance is.