Summary

  • The Phoenix failure was not simply a long queue. VA's inspector general found that thousands of veterans seeking primary care were absent from the official electronic waiting list, while schedulers used practices that made recorded waits shorter than the waits some veterans actually experienced. The resulting data could not reliably show demand, delay or clinical risk.
  • Capacity pressure, ambiguous date rules, an antiquated scheduling system and performance targets all mattered, but none made concealment inevitable. The decisive control failures were the absence of end-to-end queue reconciliation, weak validation independent of facility management, incomplete audit trails, and escalation channels that did not turn repeated warnings into timely corrective action.
  • Evidence about harm requires care. The inspector general found access barriers and troubling lapses, including cases in which delay negatively affected patients, but said it could not assert that the absence of timely quality care caused the deaths reviewed. Allegations, medical causation, administrative findings, criminal liability and civil compensation therefore cannot be collapsed into one verdict.
  • Congress expanded community care, mandated wait-time publication and penalties for knowing falsification, while VA revised scheduling policy, trained staff and pursued new technology. Later audits nevertheless found inconsistent wait-time calculations, unreliable community-care measures and incomplete modernization planning. Closure of recommendations is evidence of completed actions, not by itself proof that veterans' real waits are accurately and durably controlled.

The warning existed before the scandal

Phoenix is often narrated as a sudden revelation in spring 2014. The deeper accountability lesson begins earlier. A scheduling metric is only valid if the start event, end event and population are consistently defined. VA measured a patient's wait using a “desired date,” meaning the date on which the veteran or clinician wanted care. If staff instead entered the next appointment VA could offer, the recorded interval approached zero even when the veteran had already spent weeks asking for care. If a request never reached the electronic waiting list, it disappeared from the denominator altogether.

The weakness was documented before Phoenix became a national controversy. In 2012, the Government Accountability Office reported that VA outpatient wait-time measures and scheduling oversight needed improvement. GAO found unclear policy and training around desired dates, inconsistent implementation, staffing weaknesses and unreliable reported waits. At every medical center GAO visited, at least one scheduler recorded the desired date incorrectly. Some clinic staff said dates were changed to show waits within performance goals. VA agreed with the recommendations, but agreement did not create a reliable control at the point of entry.

That prior warning changes the assessment. Phoenix was not a novel software defect that no one could anticipate. Senior ownership already included defining the date, training the person who entered it, allocating scheduling resources and checking whether reported performance matched source evidence. A control system should have compared patient requests, referral orders, phone records, enrollment applications, pending consults, official waiting lists and completed appointments.

It should have treated a request without a scheduled appointment or waiting-list entry as an exception requiring resolution, rather than as an event outside the metric.

The 14-day access goal also blurred a vital distinction. A goal can guide capacity planning; it cannot certify clinical safety or create appointments. When a hard target is attached to management evaluation without an equally strong data-integrity control, staff can face a choice between reporting operational reality and satisfying the metric. That does not excuse falsification. It explains why assurance must be designed for the incentive environment it is meant to govern.

The trigger was a missing population, not merely a bad average

The central official record is the VA Office of Inspector General's final Phoenix report, number 14-02603-267. It followed whistleblower allegations and a large investigative response. OIG reviewed 3,409 patients assembled from electronic, paper, hotline, congressional and other sources. It found about 1,400 veterans appropriately included on the electronic waiting list for a primary-care appointment and more than 3,500 additional veterans on unofficial lists or otherwise at risk of not obtaining requested appointments.

Those categories are more important than a single headline average. An official queue gives managers at least a chance to age requests, identify clinical priorities and add capacity. A paper or locally maintained queue can sever those functions. A veteran may be waiting, yet the central record shows no wait. A facility may appear to improve because unrecorded demand is not measured. Reviewers may select samples only from official data and therefore never see the people most at risk of omission.

OIG also compared a sample of 226 veterans' actual experience with national data reported for Phoenix. The national data showed an average wait of 24 days for a first primary-care appointment; OIG calculated an average of 115 days. The report described practices that included using the next available appointment date as the desired date to create an apparent zero-day wait, canceling and rescheduling appointments, relying on paper lists, closing consults without appropriate clinical review and changing clinic-utilization information. These were not all necessarily performed by the same person, for the same reason or with criminal intent.

They were, however, control patterns capable of separating operational truth from reported performance.

The report also said an annual facility-director certification of compliance with the scheduling directive had been waived for fiscal 2013. Certification alone is weak if leaders can sign without evidence, but removing it eliminated a named attestation point at the same time that date interpretation and local practices remained vulnerable. A stronger certification would require reconciliation results, exception counts, audit-log review, patient callbacks and independent sampling—not a general statement that policy was followed.

How the scheduling system could hide delay

There were several pathways from a veteran's request to a misleadingly short or nonexistent wait. A new enrollee could ask VA to contact them, yet not be placed promptly into the scheduling workflow. A scheduler could receive a request and hold the name outside the electronic waiting list until an appointment became available. A clinician could request a follow-up or specialty consultation, but the desired or clinically indicated date could be transcribed incorrectly. A pending consult could be canceled without adequate clinical review.

An appointment canceled by the clinic could be recreated with a new start date, erasing part of the elapsed time.

Each pathway illustrates a different ownership boundary. The clinician owns the clinical need and urgency. The scheduler owns faithful entry and documented contact attempts. Clinic management owns capacity, template availability and exception escalation. Facility leadership owns the integrity of local performance reporting and the decision to seek outside care or added resources. Network and national leaders own policy, comparable measurement, independent validation and action when a facility cannot meet demand. Technology leadership owns the auditability and usability of the system, but cannot define clinical truth on its own.

An adequate control would preserve immutable timestamps for the first request, enrollment, referral, scheduling attempts, appointment offer, patient acceptance or declination, cancellation, rescheduling and completed encounter. It would retain who made each change and why. The timer would not reset merely because work moved between teams or systems. A declined appointment would record the offered date and the veteran's stated preference without transforming a VA capacity delay into patient-caused delay. A canceled consult would require an authorized clinical disposition rather than an administrative cleanup code.

The OIG later told Congress that audit controls in VistA were not enabled when investigators began examining Phoenix and said VA enabled that capability at Phoenix and nationwide at its request. Audit logs do not prevent an incorrect entry, but they make reconstruction and deterrence possible. If logs are disabled, former employee accounts are deleted, or changes overwrite prior values, an investigator cannot reliably distinguish error, policy ambiguity, supervisory direction and intentional concealment.

Capacity pressure was real, but it was not a data policy

Phoenix served a growing veteran population with constrained primary and specialty-care capacity. The final OIG report attributed unreliable wait information to multiple causes, including staffing and appointment shortages, increased demand, the old scheduling system and ethical lapses in leadership and management. That multi-causal finding matters. Treating the episode as only individual misconduct would leave structural capacity and measurement problems intact. Treating it as only a resource shortage would erase choices to use unofficial queues and inaccurate dates.

Capacity must be made visible before it can be managed. A clinic template should show available slots, protected urgent capacity, provider leave, room constraints and expected demand. A backlog should be segmented by service, urgency, age and veteran contact status. When demand exceeds the safe threshold, a named executive should approve a recovery plan: additional VA sessions, temporary staffing, redistribution across facilities, telehealth where clinically suitable, or authorized community care. The escalation should preserve the true queue instead of waiting for an appointment to become available before counting the patient.

VA's immediate public response illustrates both operational action and evidentiary limits. In August 2014, the department described outreach, accelerated appointments, non-VA referrals and actions responding to OIG. It reported contacting more than 5,000 Phoenix veterans identified on unofficial or electronic lists, scheduling appointments and making community referrals. These figures document a response at a point in time. They do not establish that every veteran received clinically appropriate care, that the original queue was complete, or that the same omission could no longer recur.

The right performance unit is therefore not the number of appointments created. It is the disposition of every validated request. For each person, VA should be able to show care completed, appointment accepted, clinically appropriate alternative arranged, veteran declined after informed contact, request determined invalid with reason, or active escalation. Aggregate acceleration claims should reconcile to that person-level ledger. Otherwise, a campaign can move visible cases while leaving hidden demand unresolved.

What the evidence does and does not say about deaths

The most serious public allegation was that veterans died while waiting for care. OIG reported that a whistleblower alleged 40 deaths but did not provide a list of 40 names. The office broadened its review and found unacceptable, troubling lapses in follow-up, coordination, quality and continuity. It discussed 45 patients: 28 negatively affected by delays, including six who died, and 17 whose care deviated from the expected standard independently of delay, including 14 who died. Yet OIG stated that it could not assert that the absence of timely quality care caused those deaths.

That language is neither a finding that delay was harmless nor proof that delay caused every adverse outcome. Clinical causation requires patient-specific evidence: condition, urgency, natural disease progression, what treatment would likely have occurred, when it could have occurred and whether a timely intervention would probably have changed the result. Administrative evidence can establish that follow-up failed or a wait was unacceptable without resolving medical causation under a civil or criminal standard.

The House Veterans' Affairs Committee examined that boundary in its September 2014 hearing on scheduling manipulation and the OIG final report. Members questioned the investigation's scope, patient-record selection and wording about death. Witness testimony and committee questioning are essential records of oversight, but they are not a court judgment. Claims made in opening statements, questions, prepared submissions and answers retain different evidential weight.

For families, the institutional obligation extends beyond a public statistic. OIG recommended case review, appropriate response to possible injury and consultation with regional counsel about disclosures to patients and families. A serious remedy process should preserve records, provide clinical disclosure where warranted, explain claim routes and separate care improvement from legal defense. Public reporting should state how many cases were reviewed and what categories of action followed, while protecting patient privacy. Without that bridge, “recommendation closed” can obscure whether an affected person received an answer.

Incentives changed the meaning of the metric

Phoenix leaders emphasized access improvement through performance goals. OIG found that the facility director's “Wildly Important Goal” effort produced an inaccurate or unsupported portrayal of primary-care access and third-next-available appointments. After the interim report, VA removed the 14-day scheduling goal from employee performance contracts. The important control point is not that measurement or incentives are inherently improper. Healthcare systems need goals. The danger arises when a measure is rewarded before its validity is independently tested.

No bonus formula should pay on a self-reported access measure that the same management chain can alter. Before an access score affects compensation, promotion or facility rating, an independent analytics function should certify population completeness, timestamp lineage, cancellation treatment and sample agreement with patient contact. The certifier should report to a level capable of challenging facility and network leaders. Material exceptions should suspend use of the metric, not be averaged away.

A balanced score also needs counter-metrics. Reported short waits should be viewed beside the number of unbooked requests, canceled and rescheduled appointments, pending consults, unanswered calls, third-next-available appointment, staffing vacancies, community-care referral age, complaints and patient-reported access. When one measure improves while its counter-measures deteriorate, the change demands investigation. That pattern is more useful than assuming any isolated metric represents experience.

Congress made the integrity obligation explicit in the Veterans Access, Choice, and Accountability Act of 2014. Among other provisions, the law required publication of appointment wait times, scheduling-system review, training, expanded access options and policies penalizing employees who knowingly submit or require false wait-time or quality data. Knowing falsification is a narrower category than poor policy, inadvertent error or inadequate capacity. A fair system must preserve that distinction while ensuring that ambiguity is corrected rather than used indefinitely as a defense.

Whistleblowers were a necessary control outside the hierarchy

The Phoenix disclosures showed why ordinary management reporting was limited public evidence. Employees could observe paper lists, unsafe delays, improper cancellations or weak triage long before a national dashboard showed a problem. A protected disclosure channel is therefore not an optional ethics program; it is an independent sensor for a system whose formal data may be unreliable.

But a channel works only if investigators are independent, cases are triaged promptly and retaliation is remedied. The Office of Special Counsel later described Dr. Katherine Mitchell's disclosures about emergency-department triage and staffing in Phoenix. Its 2015 closure statement criticized VA's internal accountability response after VA substantiated serious concerns but did not discipline responsible officials. Those issues were related to Phoenix patient safety, not identical to the primary-care wait-list mechanism. Their relevance is institutional: repeated safety escalation failed to produce timely corrective and personnel action.

Phoenix concerns also persisted after the first scandal response. In 2017, OSC published the outcome of another Phoenix whistleblower disclosure concerning specialty-care waits and cancellations. The underlying VA reviews documented an average daily population waiting more than 30 days, long psychotherapy waits, thousands of cancellations in a sampled week and cases of possible or actual harm from delay. OSC transmitted the findings and whistleblower comments to the President and Congress. This was a disclosure-referral outcome, not a judicial finding that every allegation was proven or that named managers had criminal intent.

An effective escalation architecture should allow anonymous and identified reporting, immediate clinical-risk referral, evidence preservation and status updates to the discloser. Managers accused of wrongdoing should not control the investigation. Corrective action for the patient and corrective action for the employee are separate tracks; one should not wait for the other. Leaders should also be measured on whether they surface bad news early, not merely on whether their unit avoids substantiated complaints.

Discipline required evidence and due process

Public pressure after Phoenix generated demands to fire responsible officials quickly. Accountability, however, is not speed alone. A defensible personnel action needs a clear charge, reliable evidence, consistent penalty and the process required by law. If the government removes an employee on a theory it cannot prove, reversal can weaken both the particular remedy and confidence in the system.

Former Phoenix director Sharon Helman was removed in 2014. The later Merit Systems Protection Board final order explains a critical boundary. VA initially proposed removal for failure to provide oversight but rescinded that proposal without final action. Her eventual removal was sustained on different charges involving gifts and lack of candor. The decision is evidence that removal occurred and survived the adjudicative path described in the order; it is not a finding that Helman was personally guilty of criminal wait-list manipulation or that the rescinded oversight charge was adjudicated.

The distinction is often lost in retrospective accounts. Institutional accountability may establish that a director owned a failed control environment even when individual criminal intent is not proved. Administrative discipline may address conduct or performance under one standard; criminal prosecution requires proof of statutory elements beyond a reasonable doubt; civil liability uses another framework. Each remedy must identify its respondent, charge, evidence and disposition.

Later GAO work showed that VA's broader misconduct process remained unreliable. Its 2018 review of employee misconduct and accountability found fragmented information systems, missing documentation, inconsistent senior-official investigations and instances in which proposed discipline was not imposed. GAO's data also showed employees who made identified disclosures to OSC received adverse disciplinary actions at substantially higher rates than peers. The review did not decide that every action against a whistleblower was retaliatory. It showed a risk pattern and a records problem that prevented VA from demonstrating consistent accountability.

Criminal and civil enforcement had narrower outcomes than the scandal narrative

The Phoenix investigation involved OIG, the FBI and federal prosecutors. OIG said it would present potential criminal violations to prosecutors and refer substantiated manipulation without criminal intent for administrative action. That is a pipeline, not a conviction count. Referral means evidence was submitted for legal assessment; declination may reflect proof, intent, jurisdiction or prosecutorial-discretion issues and does not necessarily validate the underlying practice.

Helman's criminal case illustrates the scope boundary. In March 2016, the U.S. Attorney's Office announced that she pleaded guilty to making a false federal financial disclosure. She admitted failing to report gifts from a former VA official associated with a consulting and lobbying firm. The release said investigators did not intend to pursue additional criminal charges at that time. The conviction was real, but it concerned financial disclosure—not falsification of Phoenix appointment dates, a patient's death or the existence of an unofficial queue.

Civil enforcement elsewhere in the VA network demonstrates that wait-time representations can create a monetary remedy when tied to government contracts. In 2020, a contractor agreed to pay $1.85 million to resolve False Claims Act allegations that it failed to schedule timely appointments at two Minnesota clinics and changed requested dates to make waits appear shorter. The Justice Department's settlement announcement expressly says the claims were allegations and there was no determination of liability.

It did not concern Phoenix, but it shows a different accountability route: contractual scheduling duties and payment claims can support civil enforcement even where criminal intent is not proved.

For Phoenix patients, remedies could also include institutional disclosure, corrective treatment, administrative claims or litigation depending on individual facts. The available official record does not support one aggregate civil-damages figure or a claim that every delayed veteran had a compensable injury. A credible account should therefore resist replacing missing outcome data with a dramatic total. The accountability gap itself is worth reporting: the public can see extensive investigation and policy action more readily than a unified patient-level ledger of disclosures, claims and resolutions.

Choice expanded access but created a second scheduling chain

Congress responded to the access crisis by creating the temporary Veterans Choice Program and providing substantial funding for eligible veterans to obtain non-VA care. The policy addressed a basic control failure: when VA capacity cannot provide timely care, the veteran should have an alternative rather than remaining in a hidden queue. Yet outsourcing the appointment does not end VA's accountability. It adds handoffs among a VA clinician, medical-center staff, a third-party administrator, a community provider and the veteran.

GAO's 2018 review of the Veterans Choice Program found that VA's scheduling process was not designed consistently with the statutory 30-day requirement and could permit much longer waits when each allowed process interval accumulated. GAO also found limitations in the information used to monitor access. These findings do not mean community care was never faster or clinically appropriate. They show that moving a referral outside VA can reproduce the same measurement problem if the timer resets at each handoff.

End-to-end community-care control should begin with the clinician's indicated date or the veteran's request, preserve the authorization date, track transmission, provider acceptance, contact attempts, scheduled date and completed care, and return clinical records to VA. A referral should not be counted as access. Neither should an authorization or an appointment offer that the veteran never receives. VA should identify which organization owns every stalled stage and automatically escalate aged referrals.

The VA MISSION Act later established a more unified, permanent community-care framework. The enacted legislation and history matter because timely-access eligibility became part of a lasting mixed delivery system, not a short emergency workaround. But statutory availability is not implementation proof. Network adequacy, authorization accuracy, provider acceptance, scheduling and record return all determine whether a veteran receives care sooner.

New date rules improved clarity but did not eliminate judgment

VA revised scheduling policy after Phoenix, separating concepts that the earlier “desired date” label had combined. Clinically indicated dates were to reflect the provider's judgment about when care should occur; preferred dates reflected the patient's preference. Clearer definitions reduce ambiguity, but the same data field and manual transcription can still introduce error. If a clinician does not enter a date, if a scheduler substitutes availability, or if a veteran's preference is recorded after the fact, the resulting wait measure remains vulnerable.

GAO's 2016 review of newly enrolled veterans' access demonstrated why the clock must start before an appointment is created. In its sample, 60 of 180 newly enrolled veterans had not seen a primary-care provider, and many had not been scheduled according to policy. For the 120 who were seen, GAO calculated 22 to 71 days from the request for VA contact to the visit. VA later enhanced reporting and revised its directive, actions GAO accepted as implementing the recommendation.

The report still provides an enduring design rule: count the entire period from a veteran's request, including time before the scheduling system contains a bookable appointment.

Policy compliance should be tested by replay. An auditor samples a veteran request and independently reconstructs the correct start date, urgency, offered dates, contacts and completion from references. The reconstructed wait is compared with the dashboard. Differences are classified as definition, transcription, system, workflow or possible misconduct. Results are stratified by facility, clinic and scheduler, with trends reported to an independent owner. A policy document without this feedback loop cannot demonstrate reliable use.

Transparency can mislead when methodologies change

Publishing wait times was meant to let veterans and overseers see facility performance. Public data can create pressure for improvement, but comparison is valid only if methods are disclosed and stable. Completed-appointment waits, pending-appointment waits and next-available appointment measures answer different questions. Starting from request date, clinically indicated date, preferred date or appointment-creation date can yield materially different results. A single number without its population and clock definition invites false confidence.

In 2022, VA OIG issued a management advisory on consistency and transparency in wait-time calculation. It found that since 2014 VHA had used different methodologies, particularly different start dates, and had not always clearly disclosed the basis. Some methods deviated from the scheduling directive and previously announced measures. OIG did not conclude that every published number was intentionally false. It concluded that inconsistent calculations and incomplete disclosure impaired transparency.

A trustworthy public table needs a versioned methodology, effective date, numerator, denominator, exclusions and known limitations. Historical series should not splice unlike measures without marking the break. Facilities should publish distributions—not only averages—including medians, upper percentiles and counts beyond clinical thresholds. Pending requests and people not yet scheduled must remain visible. Results should be reconciled to patient experience surveys and secret-shopper or callback samples so that a technically correct computation does not mask an incomplete population.

Transparency also needs correction procedures. When a data defect is discovered, VA should identify affected periods and facilities, preserve the earlier release, publish corrected values and explain impact. Silent replacement prevents users from understanding whether apparent improvement reflects care, methodology or cleanup. The objective is not a perfectly smooth trend; it is an auditable account of change.

Modernization is a governance project, not a software purchase

Phoenix exposed limitations in VistA scheduling, but technology replacement is not a sufficient remedy. A new interface can still encode an ambiguous start date, allow work to sit outside the system, or reward a misleading measure. Modernization must begin with a controlled business process and data model. It must then prove that the software enforces or records that process across VA's many facilities and specialties.

The latest broad implementation evidence remains cautionary. GAO's 2025 report on appointment-scheduling modernization described the collection of scheduling and wait-time tools VA still uses and evaluated modernization planning. GAO found that VA had not fully met key practices for a comprehensive integrated master schedule, among other planning weaknesses. It noted facility and veteran challenges and the coexistence of legacy and newer systems. The report is not a finding that current VA data are universally false.

It shows that the control environment remains fragmented and that modernization outcomes are not yet demonstrated by a completed enterprise plan.

A safe transition requires dual-system reconciliation, migration tests, role-based access, immutable audit events and explicit ownership of locally purchased tools. Every request created in an online portal, call center, consult system or local application should receive a durable enterprise identifier. Interface failures should create visible exceptions. Training should be competency-tested with realistic scenarios, including clinic cancellation, patient declination, urgent referral, no available slot and community-care transfer.

Modernization milestones should report control outcomes rather than only deployment. Useful evidence includes the percentage of requests with complete timestamp lineage, unmatched request counts, audit-log coverage, correction rates, unbooked backlog age, system-to-system reconciliation and user error patterns. A facility going live is not the same as a veteran gaining timely access. The proof is whether fewer requests are lost, dates are more reproducible and clinically risky waits trigger action sooner.

Community-care metrics remain part of the same accountability test

The MISSION-era model makes internal and external access inseparable. VA must know both how long a veteran waits inside a facility and how long a community referral takes. Otherwise, an internal queue can appear shorter when work is transferred, while the veteran continues waiting outside the measurement boundary.

GAO's 2024 review of statutory community-care timeliness measurements found that VA's approach could not produce a required time point—the date a provider accepts a referral—and that four of 15 related metrics were unreliable because they depended on that date. GAO recommendations focused on accurate measurement and reporting. This is implementation evidence years after Phoenix: the problem has migrated from the desired date inside VistA to missing handoff data across organizational boundaries.

The control remedy is contractual and technical. Community providers and administrators should return acceptance and scheduling events in a standard format, with validation and late-data flags. VA should distinguish no provider available, provider declined, veteran unreachable, veteran requested later care and VA authorization delay. Contract performance should use end-to-end veteran outcomes, with rights to inspect source data and remedies for inaccurate reporting. But contracts must not pressure vendors to recode delays merely to avoid penalties; independent sampling remains necessary.

Veterans also need usable transparency. A public average cannot tell an individual whether a particular specialty has capacity, whether a community referral was accepted or who to call when it stalls. Operational notice should show referral status, responsible party and next action without exposing clinical information. An escalation channel should be able to see the same authoritative record as the scheduler, not start another disconnected list.

Accountability law can create an office without creating trust

Phoenix also reshaped the federal personnel framework. The Department of Veterans Affairs Accountability and Whistleblower Protection Act of 2017 established an Office of Accountability and Whistleblower Protection and revised authorities and protections. The law created responsibilities and procedures; it did not resolve every tension among rapid action, due process, investigative independence and employee trust.

An accountability office should be assessed through case-quality evidence. How quickly does it triage a patient-safety disclosure? Does an investigator outside the implicated chain examine it? Are interim protections available? Are substantiated findings linked to corrective action and tracked through disposition? Are proposed and final discipline both recorded, including reasons for change? Are whistleblowers informed and protected from retaliatory work assignments, ratings or access restrictions?

The office should not become another destination that fragments evidence. OIG, OSC, professional standards, human resources, regional counsel and clinical quality teams have distinct authorities, but a case can touch all of them. A shared, access-controlled case map should preserve referrals and ownership without improperly merging investigative files. Senior leadership and Congress need aggregate visibility into aging, outcomes and recurring control themes; individual employees need confidentiality and fair process.

What durable proof would look like

The OIG page shows all 24 Phoenix recommendations as closed and implemented, with dates extending through April 2017. That is meaningful evidence that VA supplied actions accepted for closure: patient reviews, wait-list work, training, routine scheduling-quality review, management action, call monitoring, system updates and independent validation mechanisms. It should not be dismissed. Nor should it be treated as permanent assurance. Recommendation closure tests whether an agreed corrective action was completed, not whether the risk remains controlled under later demand, staffing, software and policy conditions.

Durable proof requires recurring evidence at four levels. At the patient level, every request has end-to-end timestamps, clinical priority, documented offers and a final disposition. At the clinic level, demand, capacity, backlog and cancellation patterns reconcile, with aged exceptions escalated. At the enterprise level, definitions are versioned, audit logs are complete, local systems are inventoried and independent samples reproduce published results. At the remedy level, disclosures, patient reviews, disciplinary referrals, legal cases and corrective actions have traceable outcomes without overstating what each process established.

Boards, Congress and the public should also see adverse indicators. The number of omitted requests found by reconciliation, corrected timestamps, unexplained cancellations and referrals returned for missing data should be published alongside wait-time improvements. A mature control system does not prove reliability by reporting zero exceptions. It proves that exceptions are detected, investigated and reduced without disappearing from view.

The central question after Phoenix is not whether VA can produce a wait-time number. It is whether a veteran's first request survives every handoff until care or an accountable disposition, and whether an independent reviewer can reconstruct that journey from evidence. Phoenix became an accountability test because the institution managed the representation of access while losing sight of people outside the represented queue.

The durable remedy is a system in which operational pressure becomes visible demand, bad news travels upward, whistleblowers can speak safely, remedies retain their legal boundaries and no performance target can erase the time a veteran actually waited.

That proof should be tested under stress, not only during a scheduled audit. A sudden provider absence, rapid enrollment growth or specialty backlog should cause the same controls to expose demand, preserve original dates and activate alternatives. Random callbacks should ask veterans when they first sought care, what dates they were offered and whether the recorded disposition matches their account. Results should be compared with system timestamps and investigated when they diverge. Congress should receive trend data that separates improved capacity from changed definitions, transferred referrals and removed records.

Facility leaders should be able to explain every aged exception without editing its starting point. Only repeated agreement among references, patient experience and independent reconstruction can show that the lesson of Phoenix has moved from policy language into daily operation.

Finally, assurance must follow vulnerable requests across time. A veteran who moves, misses a call, changes clinical priority or returns after a cancellation should not become a fresh case that erases the earlier wait. The record should retain linked episodes and explain why one ended and another began. That continuity is especially important when several technically valid transactions can conceal one prolonged effort to obtain care. Longitudinal review turns appointment data back into a patient journey and makes accountability resistant to administrative resets.