On a Tuesday morning in late 2024, a case manager at a mid-sized Washington state hospital submitted a prior authorization request through the CMS WISeR Model portal. The patient was a Medicare beneficiary, sixty-eight years old, post-surgical, requiring inpatient rehabilitation. The request was clinically straightforward. The documentation was complete. Under the legacy manual process — phone calls, fax queues, human reviewers working from the same clinical criteria — a decision like this one had historically arrived within forty-eight to seventy-two hours.
The WISeR decision arrived on day eleven.
By then, the patient had been waiting in an acute care bed that cost roughly four thousand dollars per day to occupy. The rehabilitation facility had provisionally released the reserved bed. The case manager had spent portions of six working days on portal troubleshooting, status inquiries, and documentation resubmissions triggered by system requests that did not correspond to the clinical record already on file. The authorization, when it finally cleared, said: Approved — medically necessary care. What it did not say — what no document in the WISeR interface was designed to say — was why it had taken eleven days to reach that conclusion, what had caused the two documentation resubmission requests, or who — or what — had reviewed the claim at each stage of its journey through the model.
She did not know, on day eleven, that she was inside an experimental system. She knew she was waiting. She knew her physician had told her the care was appropriate. She knew she was still in a hospital gown, in an acute care bed, in a facility that was not the one she was supposed to be in. She did not know that the walls around her were not walls at all — that they were the architecture of a demonstration model launched without binding turnaround requirements, without escalation pathways, without any mechanism to tell her what was happening or why. She did not know she had entered the machine. No one had told her. No informed consent document described the model's experimental status. No admission paperwork explained that her authorization would be processed by a system whose accountability structures had not been finalized before it went live on patients like her.
That is the shape of this story. The authorization cleared. The delay was eleven days. What accumulated in those eleven days — the carrying cost, the lost bed, the clinical risk of an extended acute stay, the hours of administrative labor — does not appear in any document associated with the WISeR Model. The machine processed the request. The machine does not record what the request cost.
What WISeR Was Supposed to Do
CMS has not published a definitive public expansion of the WISeR acronym in its Innovation Center program documentation as of this writing, and any rendering of the full name should be treated as provisional until confirmed against official model materials. What CMS has published is the program's intent: the WISeR Model is a demonstration program designed to test automated, data-driven prior authorization processing for Medicare fee-for-service claims. The policy rationale was legitimate, and it is worth inhabiting it for a moment before tracing where it failed — because the failure is more instructive when measured against a genuine need.
Prior authorization, in its current form, is one of the most thoroughly documented sources of care delay and administrative waste in the American healthcare system. The American Medical Association's 2023 prior authorization survey found that physicians and their staff spend an average of nearly fourteen hours per week completing prior authorization requirements. Sixty percent of physicians reported that prior authorization had led to a serious adverse event for a patient in their care — a hospitalization delayed, a complication that developed in the interval between request and approval, a patient who gave up and went without. Nurses are pulled from bedside care to resubmit documentation that was already submitted. Case managers spend working days on fax queues that were supposed to have been abolished by a prior generation of modernization efforts. A system that could process authorization requests faster, more consistently, and with less administrative burden was — and remains — a genuine need.
WISeR was supposed to be part of that solution. The demonstration model would use data-driven processing to adjudicate Medicare fee-for-service prior authorization requests with greater speed and consistency than the manual review infrastructure it was replacing. The friction would be reduced. The delays would shorten. The case manager spending six days chasing a fax confirmation would instead receive a portal determination in hours.
Instead, early operational data suggests the model has added a new layer to the labyrinth while removing the human callbacks and escalation pathways that, for all their inefficiency, gave providers somewhere to turn when a request stalled. The old system was slow and often irrational, but it was slow and irrational in ways that experienced revenue cycle staff had learned to navigate across years of daily contact with its specific dysfunctions. You knew which Medicare Administrative Contractor medical reviewer might accept a peer-to-peer on a borderline inpatient admission. You knew that faxing the operative report directly to the utilization management line sometimes broke a logjam that the portal had created. You knew, in other words, where the pressure points were — where a human being with clinical authority could be reached, argued with, persuaded.
WISeR sealed those pressure points behind an interface that returns status codes instead of conversations. The labyrinth was not demolished. It was renovated, its familiar corridors replaced with identical unmarked ones, and the guides who knew the old passages were not replaced.
According to reporting by Fierce Healthcare, Washington state hospital administrators have described the CMS WISeR Model as producing authorization delays two to four times longer than the manual prior authorization processes the model was intended to modernize. That is not a rounding error. That is a system producing the opposite of its stated purpose while generating no accountability data about why.
The Policy Failures Embedded in the Design
The failures that produced eleven-day authorizations on clinically routine inpatient rehabilitation requests are not random. They are the predictable consequence of specific decisions that CMS made — or failed to make — when it designed the WISeR rollout. Three structural omissions sit at the center of what went wrong.
The first is the absence of any binding, enforceable performance benchmark for authorization turnaround time. Medicaid managed care regulations under 42 CFR § 438.210(d) require plans to establish timeframes for prior authorization decisions — seventy-two hours for urgent pre-service requests, fourteen calendar days for standard pre-service requests, with extensions requiring enrollee notification. Medicare Advantage organizations operate under similar CMS requirements. The 2024 CMS final rule implementing the Improving Seniors' Timely Access to Care Act — the most significant recent federal prior authorization reform — goes further still, mandating twenty-four-hour turnarounds for urgent requests and seventy-two-hour turnarounds for standard requests on Medicare Advantage plans, effective 2026. The regulatory direction of travel is unambiguous: faster, not slower; binding, not aspirational.
Medicare fee-for-service, operating through the WISeR demonstration, entered this experiment without equivalent guardrails. A hospital submitting an authorization request through the WISeR portal has no regulatory right to a decision within any particular timeframe. The portal may show Pending Review for eleven days, or fourteen days, or longer. There is no clock. There is no penalty provision. There is no regulatory tripwire that converts an unreasonably delayed authorization into a deemed approval. The absence of a mandatory performance benchmark is not a technical oversight — it is the foundational policy failure from which all others follow, because a system without a clock can fail indefinitely without triggering any formal accountability response. The patient in the acute care bed on day eleven has no regulatory claim. The hospital absorbing four thousand dollars per day in carrying cost has no formal remedy. The case manager has no escalation right. There is nowhere to go.
In the legacy manual system, the absence of a regulatory clock was a known problem. Advocates and revenue cycle professionals had spent years pushing for faster turnaround mandates — the same advocates whose work eventually produced the 2024 MA rule. WISeR did not inherit their progress. It inherited the underlying problem and then stripped out the informal compensating mechanisms that had partially addressed it.
The second failure is the opacity of the review pathway itself. When a prior authorization request moves through traditional utilization management — even a highly automated process — experienced billers and case managers can usually determine, through escalation calls and peer-to-peer requests, what stage of review the request has reached and what specific documentation gap, if any, is holding it up. The information is not always accurate. The reviewers are not always reachable. But a pathway exists, and the pathway has identifiable nodes.
WISeR, as implemented, offers no such pathway. Hospitals interacting with the portal receive status updates — pending, additional information required, under review — without the specificity required to act on them efficiently. An additional information required notification that does not identify which document is required, under which clinical criterion, reviewed against which version of the coverage policy, is not information. It is a delay mechanism wearing the costume of information. The case manager who receives it cannot go directly to the relevant section of the medical record and resubmit the specific item that is missing, because the notification does not say what is missing. She must instead infer — from her knowledge of the clinical criteria, from her experience with similar requests, from whatever she can reconstruct about the reviewer's probable logic — which of the dozens of documents in the record is the one the system is waiting for. She is doing interpretive work that the system was supposed to do. She is, in the precise sense, solving a puzzle whose solution is known to the machine and withheld from her.
The documentation resubmission loops that Washington state hospitals have described — submitting records, receiving a request for additional documentation, submitting additional documentation, receiving another request — are the operational signature of a review process that either lacks the specificity to communicate its requirements clearly or is structured in a way that generates iterative requests as a feature rather than a flaw. The distinction matters, but not as much as the effect: each iteration consumes time, and time, in prior authorization, is always paid for by the patient who is waiting.
The third failure is the absence of a mandatory escalation pathway. Before WISeR, a provider facing a stalled authorization had imperfect but real options. A peer-to-peer review — a conversation between the treating physician and the insurer's medical reviewer — was not always available and was not always productive, but it existed as a formal mechanism. A call to the Medicare Administrative Contractor's provider relations line could surface a human being who understood the clinical context. A Medicare beneficiary could call 1-800-MEDICARE and initiate a complaint that created at least a nominal paper trail.
WISeR has no mandatory peer-to-peer pathway. It has no documented escalation process for requests that exceed a reasonable review period. When a hospital case manager told Fierce Healthcare that she had no way to find out why a request was delayed or who was reviewing it, she was not describing a frustration with a poorly designed user interface. She was describing the deliberate absence of a fundamental accountability mechanism — a design choice, not a bug, embedded in the model before it went live.
She was also describing, though she did not use this word, the experience of the patient one floor above her: the sixty-eight-year-old woman who did not know the system she had entered, who did not know the exit was not where anyone had told her it would be.
What the Revenue Cycle Knows That the Model Does Not Count
The prior authorization process, whatever its stated purpose, functions in practice as a revenue protection mechanism for payers. A delay is not a neutral outcome. A delay is, from the payer's perspective, a deferred liability — and in some proportion of cases, a deferred liability that resolves itself when the patient deteriorates, is readmitted at higher acuity, or simply gives up. For Medicare fee-for-service, the payer is CMS itself, which creates a structural tension that the WISeR Model's architects either did not perceive or chose not to name: the agency designing and administering the authorization review model is the same agency that benefits, in the short term, from deferred authorization expenditures.
Revenue cycle professionals who work in hospital business offices understand this dynamic with a clarity that policy documents rarely acknowledge. The costs that WISeR's eleven-day delays impose do not disappear. They redistribute. The hospital absorbs the carrying cost of the occupied acute care bed — four thousand dollars per day, multiplied by the days between when the patient was clinically ready for discharge and when the authorization cleared. The rehabilitation facility absorbs the revenue loss from a reserved bed that sat empty. The attending physician absorbs the time cost of reauthorization documentation that displaces clinical work. The patient absorbs the clinical risk of an extended acute stay — the hospital-acquired infections, the deconditioning, the psychological weight of waiting in a place you were supposed to have left.
None of these costs appear in CMS's demonstration model cost accounting. The model measures what it chooses to measure: authorization volume, perhaps denial rates, perhaps turnaround time averages that smooth over the cases at the tail. What it does not measure — provider administrative burden, patient harm from delayed care, downstream utilization costs from extended acute stays, the cost of beds occupied by patients who should be elsewhere — is precisely the territory where the damage accumulates. A model that measures only its outputs and not the collateral costs of its failures is not an accountability mechanism. It is a documentation of its own preferred version of events.
The billing office sees the rest. The case manager sees the rest. The patient on day eleven sees nothing, because no document in the system was designed to tell her.
The Architecture of a Fix
The following are not aspirational gestures toward reform. They are the minimum accountability structures that should have been conditions of the WISeR Model's authorization to operate — the scaffolding that the demonstration was launched without, and whose absence produced the outcomes now being reported from Washington state hospitals.
The first and most fundamental correction is a deemed-approval provision tied to a hard clock. Congress should mandate — through legislative amendment or, failing that, through regulatory action under CMS's existing Section 1115 and Innovation Center authorities — that any prior authorization request processed through the WISeR Model is deemed approved if a final determination is not issued within seventy-two hours for urgent requests and seven calendar days for standard requests. The choice of seven days is not arbitrary. The 2024 CMS final rule implementing the Improving Seniors' Timely Access to Care Act established a seventy-two-hour standard for standard requests in Medicare Advantage plans, effective 2026. A demonstration model marketed as technologically superior to manual review — and operating within the same Medicare program — has no principled basis for claiming a more permissive standard than the one CMS has already determined is achievable for MA plans. Seven calendar days is, if anything, generous. It is offered here not as the ideal but as the regulatory floor below which a deemed-approval provision would activate.
The deemed-approval mechanism exists in analogous form in several state insurance codes governing Medicaid managed care plans. Its logic is simple and its deterrent function is direct: if the reviewing entity fails to act within the time it has been given, the default resolves in favor of the patient and the treating physician. This is not a radical concept. It is the application, to a federal demonstration model, of a principle that state insurance regulators have already determined is compatible with functional utilization management. The objection that deemed approvals would undermine the model's cost-control function is undermined by the obvious response: a cost-control function that requires eleven days to adjudicate a clinically routine inpatient rehabilitation request is not functioning. The deemed-approval provision does not eliminate cost control. It imposes a deadline on its exercise.
The second correction addresses the opacity of additional information requests. Every documentation request generated by the WISeR portal must identify, with regulatory-grade specificity: the clinical criterion under review, the coverage policy and policy version number being applied, the specific document or data element that is missing or insufficient, and the timeframe within which the resubmission must be received to avoid denial. This is not an onerous standard. It is the standard that a good-faith utilization management process already meets when it functions correctly. The specificity requirement has a second-order function beyond its immediate utility: it creates an auditable record. A documentation request that cannot specify what it is requesting — that cannot identify the criterion, the policy version, the document — is a documentation request generated by a process that does not know what it is doing. Requiring specificity surfaces that dysfunction. It also provides the legal basis for the companion rule: an additional information request that fails to meet the specificity standard does not toll the authorization clock. The machine does not get to pause the countdown by issuing a request it cannot justify.
The third correction is the restoration of the peer-to-peer pathway that WISeR's design eliminated. Every prior authorization request processed through the model must be subject to a mandatory peer-to-peer review pathway, available upon request by the treating physician, within twenty-four hours of the request. The reviewer on CMS's side of that conversation must be a physician in the same specialty as the treating physician, or a physician with documented clinical experience in the relevant service category. This standard is not novel. It is the standard that CMS has historically applied to Medicare Advantage plans and that the broader utilization management field has recognized as a requirement of good-faith clinical review. Peer review that is not conducted by a peer is not peer review; it is a performance of review. The WISeR Model, by eliminating the pathway entirely, did not even preserve the performance. It substituted a status code for a conversation and offered no mechanism for the physician to contest the code's implicit clinical judgment.
The peer-to-peer requirement has an additional significance in the context of automated review. If the WISeR Model is using data-driven criteria to make or recommend coverage determinations, the treating physician has no current mechanism to introduce the clinical particulars that automated criteria do not capture — the patient's functional baseline, the comorbidities that make a standard protocol inappropriate, the clinical judgment that has been exercised at the bedside over days or weeks of care. The peer-to-peer pathway is not merely an appeals mechanism. It is the only point in an automated review process where the specific human patient, with her specific clinical presentation, can be introduced into a system otherwise operating on abstracted data.
The fourth correction is independent auditing of WISeR turnaround and outcome data. CMS should be required to submit authorization turnaround data to an independent audit conducted by the HHS Office of Inspector General no less than annually, disaggregated by service type, provider type, geographic region, and request outcome. That audit should include a specific assessment of whether the model's authorization decisions are consistent with CMS coverage policies and with the decisions that would have been reached under the manual review process it replaced. The audit results should be published in full, without redaction for proprietary model parameters.
The argument against this audit is the argument that has been used against every prior authorization transparency requirement: that the model's decision logic is proprietary or too complex to adjudicate against a manual-review counterfactual. That argument should be rejected. CMS is a public agency. The WISeR Model is operating on Medicare beneficiaries using public funds. The beneficiaries did not consent to serve as subjects in an accountability-free experiment. If the model's decision logic cannot be audited against the coverage policies it claims to apply, it should not be operating. The audit requirement does not demand that the model be simple. It demands that the model be accountable. Those are different things, and only one of them is optional.
The Patient Who Does Not Know She Is Inside the Machine
On day eleven, the authorization cleared. Approved — medically necessary care. The case manager logged the determination and began the discharge coordination she should have been able to begin nine days earlier. The rehabilitation facility scrambled to recover a bed that had been provisionally reassigned. The attending physician signed the transfer order. The patient, who had been told on day three that the authorization was pending and on day seven that additional documentation had been requested and on day nine that the request was under review, was told on day eleven that she could go.
She did not know, leaving, what had happened to her. She did not know that the delay was not a clerical error, not a miscommunication, not a flaw in her specific claim that more complete documentation would have prevented. She did not know that it was the output of a system designed, launched, and operated without the accountability structures required to prevent it. She did not know that the walls of the WISeR portal — the status codes, the resubmission requests, the under review notifications that told the case manager nothing — were not the malfunctioning parts of a system that was otherwise sound. They were the system. The system was working as it was built to work.
Poe understood that the most effective horror is not the monster at the door. It is the realization, arriving gradually and then all at once, that you have been inside the monster for some time — that the walls that seemed like walls are something else, that the house you entered in good faith is the haunted thing itself, and that the exit you were told about is not where they said it would be. The Medicare beneficiary whose inpatient rehabilitation authorization sits in a WISeR portal queue does not experience this realization. That is not because the realization is wrong. It is because no document in the system is designed to produce it. The forms do not confess. The status codes do not explain. The machine approves the request on day eleven and offers nothing — no accounting, no apology, no record of what the eleven days cost — because it was not built to offer those things. It was built to process the request. The request was processed.
What accumulated in the interval between submission and approval — the carrying cost, the lost bed, the clinical risk, the hours — does not appear in the WISeR Model's outcomes data. It appears in the hospital's financial records, in the rehabilitation facility's occupancy logs, in the attending physician's documentation of an extended acute stay, in the patient's memory of nine days in a hospital gown waiting to be told she could leave. The machine has no field for any of that. The machine does not know it happened.
The CMS WISeR Model prior authorization delays that Washington state hospitals are now documenting are not a malfunction. They are the output of a design that was allowed to go live without the guardrails that patient safety and provider accountability require: no binding clock, no specificity requirements, no peer-to-peer pathway, no independent audit mechanism. The model can be fixed. The four structural corrections described above are not comprehensive reform. They are the minimum — the absolute floor of accountability that should have been in place before a single Medicare beneficiary's care was routed through this system.
Below that floor, what we are left with is a machine that processes requests and records its own approvals and has no language for what it costs, and no mechanism for the person inside it to understand where she is, and no clock that anyone has agreed to keep.
The clock is running. Or rather — and this is the precise and terrible point — it isn't.
Source note: This analysis is based on Fierce Healthcare's reporting, "AI-powered prior authorizations for Medicare have greatly delayed care, Washington state hospitals say", and on publicly available CMS prior authorization policy materials.