What happened in Chattanooga
In February 2026, CommonSpirit Health moved to replace Diagnostic Imaging Consultants — the local radiology group that had served Memorial Hospital in Chattanooga, Tennessee, for decades — with a remote teleradiology company based in India, according to an open letter from frontline clinicians published in the Chattanoogan. The cutover happened on June 17, 2026. It did not go well.
"In the days following the June 17, 2026 transition, clinicians report serious radiology delays, technical failures affecting access to imaging, and workflow breakdowns," the letter states. Internal communications from administration, it adds, "acknowledge a substantial imaging backlog and have directed staff to prioritize certain patients, divert imaging to outside facilities, and have non-radiologists perform preliminary 'wet reads.'"
The specifics the clinicians describe are stark: STAT emergency scans taking up to six hours to be interpreted, inpatient studies on deteriorating patients taking more than a day, and "a troubling rise in vague, inconclusive reports — along with diagnostic errors they fear have already misguided patient care in some cases." Per Radiology Business, some delays affected stroke patients in the emergency department.
Patients noticed too. One long-time patient told Chattanooga's local Fox affiliate that her ultrasound report in December was available in the patient portal the same day — but after a return visit in June, she had been waiting weeks for results. In early July, the health system issued a public apology: "We deeply regret any inconvenience this may have caused our patients... you deserve timely and clear information." It did not cite a reason for the delays.
Note the timeline. This was not a rushed, overnight switch: the decision came in February, the cutover in mid-June — roughly four months of lead time. The system still hit a backlog severe enough that non-radiologists were doing preliminary reads within days. That is what makes this case worth studying: it shows how radiology outsourcing fails even when there is time to prepare.
Not an isolated incident
The Chattanooga episode is a sharp local example of a national pattern. The VA's National Teleradiology Program — one of the largest outsourced-read operations in the country — was the subject of an Office of Inspector General review that, as reported by U.S. Medicine, documented a growing trend of interpretation delays. In fiscal year 2024, the program met its 1-hour STAT benchmark on 78.9% of roughly 1.08 million stat studies, and its 48-hour routine benchmark on 69.3% of 363,830 routine studies. In summer 2024, when staffing couldn't keep up with demand, program leaders cut accepted routine volume by 15% and stopped interpreting stat studies during weekday business hours.
Meanwhile, baseline turnaround pressure is rising everywhere. A Journal of the American College of Radiology study of nearly 2.6 million outpatient imaging studies from Medicare claims found interpretation turnaround times jumped 113% between 2014 and 2023 — from about 2 hours 11 minutes to about 4 hours 37 minutes on average, with CT rising the most (up 318%), as covered by Radiology Business. The authors' conclusion: the radiology workforce "has reached maximum capacity in that it cannot absorb additional workload." Any outsourcing vendor is hiring from that same exhausted pool — which is exactly why a contract that looks fine on paper can fail in the queue.
Why outsourced reads slow down: five failure modes
Teleradiology is not inherently broken — it covers gaps that many hospitals could not otherwise fill. But transitions like Chattanooga's fail in predictable ways. Five mechanisms show up again and again:
1. The onboarding cliff
On cutover day, 100% of volume shifts to radiologists who have never read for the site: they don't know its scanners and protocols, its referring physicians, or where the priors live. Throughput drops exactly when the queue is fullest. In Chattanooga, serious delays appeared within days of the June 17 switch — despite months of lead time.
2. Credentialing and licensing lag
Every radiologist rendering reads for a facility must be licensed for the jurisdiction and credentialed at that facility. A vendor roster that looks deep on paper can shrink to a fraction of its size at go-live, because only part of it has cleared credentialing. The effective reading pool is the credentialed pool — not the brochure.
3. Queue economics without enforceable SLAs
Outsourced studies enter a multi-client worklist the hospital does not control. When vendor demand exceeds vendor capacity, someone's studies wait — and without turnaround-time SLAs backed by financial penalties, there is no mechanism forcing it not to be yours. The VA's program shows what capacity stress looks like even with benchmarks: it simply stopped taking stat studies during business hours.
4. IT integration failure
The Chattanooga letter cites "technical failures affecting access to imaging" alongside the delays. PACS-to-vendor interfaces, prior-study migration, HL7 results delivery back into the EMR — each is a single point of failure, and each tends to be tested at demo volume, not at full production volume on day one.
5. Loss of local knowledge and escalation paths
When the group down the hall is gone, so is the phone call about an ambiguous critical finding. The letter describes a rise in "vague, inconclusive reports" — a familiar symptom when readers lack local context and clinicians lack a direct line to the person who read the study.
A due-diligence checklist for any radiology outsourcing contract
The clinicians' letter asks for "objective data comparing imaging turnaround times before and after the transition." That data should exist before any contract is signed. Whatever vendor a hospital evaluates, these are the questions that separate a managed transition from a backlog:
| Contract area | What to require | Red flag |
|---|---|---|
| Turnaround-time SLAs | Per-priority-tier targets (STAT / urgent / routine) measured per study, with financial penalties for misses | TAT expressed only as an average, or as a goal with no penalty |
| Credentialed capacity | Named count of radiologists licensed and credentialed for your facility on day one, per modality and subspecialty | A total roster number with credentialing "in progress" |
| Cutover plan | Phased volume shift with the incumbent group overlapping the new vendor; go/no-go criteria at each phase | Hard cutover of 100% of volume on a single date |
| IT integration | End-to-end testing at production volume: PACS transfer, priors access, results back into the EMR | Interface testing scheduled for the week of go-live |
| Critical findings | Named escalation path and maximum notification time for critical results; direct clinician-to-reader contact | Critical findings handled through a generic ticket queue |
| Quality assurance | Published discrepancy-review process, peer review rates, and reporting of QA metrics to the hospital | QA described as internal and not shared |
| Exit and contingency | Defined contingency coverage if SLAs are missed, and exit terms that don't strand the hospital | No plan B beyond "divert imaging to outside facilities" |
Every red flag in that table maps to something that reportedly happened in Chattanooga: the hard cutover, the technical failures, the backlog managed by diverting patients and improvising wet reads. A more detailed walk-through of transition mechanics is in the guide on switching from teleradiology without breaking coverage, and current turnaround norms are collected in the CT turnaround-time benchmarks.
The structural fix: outsource the drafting, not the read
Underneath the checklist sits a structural question: when a hospital outsources interpretation wholesale, it trades a staffing problem for a control problem. Turnaround now depends on an external company's recruiting, credentialing pipeline, and multi-client queue — the exact variables the Chattanooga clinicians say broke, and the ones the VA program has struggled with at national scale.
There is another way to divide the work. The most time-consuming part of a radiologist's read is not the judgment — it's producing the report. AI systems built on foundation models can draft a complete, structured CT report covering all visible findings, so the hospital's own radiologists review and sign instead of dictating from scratch. In xAID's workflow, the AI drafts the report, an in-house radiologist reviews every preliminary, and the study comes back ready-to-sign — the final signature stays with the hospital's own reading radiologist.
That division of labor changes the failure modes. There is no credentialing cliff, because the readers of record are the radiologists the hospital already credentials. There is no multi-client queue deciding whether a STAT head CT waits six hours, because prioritization stays on the hospital's own worklist. And there is no loss of local knowledge, because the physician signing the report is the same one clinicians can call. The comparison of AI-drafted reporting vs traditional teleradiology covers the trade-offs in detail — but the core point is simple: capacity can be bought; control has to be kept.
Frequently asked questions
What happened with the teleradiology transition in Chattanooga?
In February 2026, CommonSpirit Health moved to replace Diagnostic Imaging Consultants — the local radiology group that had served Memorial Hospital in Chattanooga for decades — with an India-based teleradiology company. In the days following the June 17, 2026 cutover, frontline clinicians reported serious radiology delays, technical failures affecting access to imaging, and workflow breakdowns, including STAT emergency scans waiting up to six hours for interpretation. The health system publicly apologized to patients in early July.
Why do outsourced radiology reads slow down after a transition?
Common failure modes include the onboarding cliff (full volume shifts to the new vendor on day one, before its radiologists know local protocols and systems), credentialing and licensing lag (only part of the vendor's roster may be credentialed at the facility at go-live), queue economics (studies enter a multi-client queue the hospital doesn't control, often without enforceable turnaround-time penalties), IT integration failures between PACS/RIS/EMR systems, and the loss of local escalation paths — there is no radiologist down the hall to call about a critical finding.
What should a hospital check before signing a radiology outsourcing contract?
At minimum: turnaround-time SLAs with financial penalties (per priority tier, measured per study, not as an average), the number of radiologists actually licensed and credentialed for each facility at go-live, a phased cutover with the incumbent group overlapping the new vendor, tested PACS/RIS/EMR integration with access to priors, a named escalation path for critical findings, published QA and discrepancy-review processes, and a contingency plan if the vendor misses SLAs — including exit terms.
Is there an alternative to outsourcing radiology reads entirely?
Yes. Instead of moving the whole interpretation to an external queue, hospitals can keep final reads with their own radiologists and outsource the drafting work. AI-drafted preliminary reports — reviewed by a radiologist before delivery and signed by the hospital's own reading radiologist — cut per-study reporting time without handing turnaround control to an outside vendor's staffing situation.
Source: Radiology Business coverage of the CommonSpirit Health (Chattanooga) teleradiology transition; the frontline clinicians' open letter in the Chattanoogan (June 25, 2026); patient reporting by Fox Chattanooga; VA OIG findings via U.S. Medicine; turnaround-time trends from the Journal of the American College of Radiology, as reported by Radiology Business. Figures are rounded as reported.