There is a recognizable genre of federal finding in which every box has been ticked and nothing intended has been accomplished. The VA Office of Inspector General’s new roll-up review of ten inpatient mental health inspections is a fine specimen of the form, and the veterans on the receiving end of it are the ones paying for the genre.
Released October 8 as report 26-01435-02, the review is not a fresh inspection. It synthesizes ten Mental Health Inspection Program reports on Veterans Health Administration facilities in Kentucky, West Virginia, Michigan, Georgia, Massachusetts, Wisconsin, New York, Pennsylvania, Virginia, and Florida, with inspections conducted between January 2024 and August 2025. At each site, inspectors pulled 50 randomly selected records of veterans discharged from acute inpatient mental health stays and measured them against a 90 percent compliance benchmark. Two things kept failing: whether the discharge paperwork could be understood, and whether anyone wrote down that a prescriber explained the risks of a new medication.
Complete, Offered, and Unreadable
Credit where it is due, because the report gives it. All ten facilities documented completed discharge instructions in every sampled record, and copies were offered to veterans or caregivers at rates between 94 and 100 percent. Medication lists were nearly universal. Follow-up appointments were generally scheduled before discharge. The institution is, in short, excellent at producing a document.
What the document says is another matter. Nine of the ten facilities missed the benchmark on describing the follow-up appointment location in easy-to-understand language. Lexington, Martinsburg, and Tampa scored zero. Ann Arbor scored 2 percent and Milwaukee 4. Only Salem, at 94 percent, cleared the bar. The OIG found instructions full of abbreviated or technical clinic names, the sort of label that makes perfect sense to the scheduler who typed it and none to the veteran holding the page.
Medications fared no better. Eight of ten facilities failed to state the purpose of each prescribed drug, with Milwaukee at 4 percent and Martinsburg at 10. Eight of ten also left difficult medical abbreviations in the dosing instructions, with Martinsburg at 18 percent and Augusta at 23. Picture a veteran leaving a psychiatric unit with new prescriptions and a follow-up to keep, handed a sheet that does not say what the pills are for, abbreviates the instructions for taking them, and names the clinic in a dialect of the building’s own. The OIG warns that this raises the risk of confusion, medication errors, and missed appointments, and cites research tying poor discharge communication to preventable errors and readmissions.
The Conversation That Left No Trace
VHA Directive 1004.01(3) requires prescribers to document an informed consent discussion for new treatments. The OIG checked whether that happened before newly prescribed central nervous system medications were started. Nine of ten facilities fell short of 90 percent. From the bottom: Lexington 18, NY Harbor 19, Ann Arbor 20, Philadelphia 22, Salem 34, Milwaukee 50, Leeds 70, Augusta 73, Martinsburg 82. Tampa recorded 100.
A fair caveat belongs here, and the OIG supplies it: the review measured compliance on paper and did not assess direct effects on veteran care. A missing note is not proof of a missing conversation. But the chart is where policy says the conversation must be recorded, and at four systems no more than 22 percent of sampled charts recorded it. Without that documentation, the OIG says, there is limited assurance that veterans had enough information to make informed decisions. That is a restrained way of describing a medication started on a veteran in acute mental distress with nothing in the file to show anyone explained the risks.
The Tampa Problem
Tampa is the most instructive result in the report, mainly because it is so inconsistent. The system documented risk-benefit discussions in 100 percent of sampled records, stated medication reasons in 96 percent, and kept abbreviations out of dosing instructions in 100 percent. It scored zero on telling veterans where to go for follow-up care in plain language. Ann Arbor cleared 90 percent on medication reasons and abbreviations while managing 2 percent on locations and 20 on risk-benefit. No facility cleared every measure. Salem, the only Virginia site in the review, was the lone pass on appointment locations and then posted 34 percent on risk-benefit documentation.
Inference, labeled as such: the competence here appears to be local and unshared. The basis is Tables 1 and 3 of the OIG report, where the same facility tops one measure and bottoms out on another. The reasoning is that if the failures stemmed from a single national template or a single national training gap, scores would move together across sites. Instead they vary by facility and by element, and the OIG’s summaries of corrective actions describe facility-level template and education fixes. Somewhere in the VA, a team already knows how to do each of these things. The report does not say whether anyone has tried to carry that knowledge elsewhere.
The individual inspection reports were published between September 2024 and May 2026, each time with local leaders implementing corrective actions, and the same deficiencies kept turning up at the next site. The OIG concludes that findings this consistent across geographically varied systems are not isolated and warrant system-wide action. It also says it did not assess whether VHA’s later national initiatives have fixed anything, since the inspections came before or near the start of VHA’s transformation initiative.
Concur, With Fine Print
Acting Under Secretary for Health John Figueroa concurred with all three recommendations. Target dates are December 2026 for appointment locations and risk-benefit discussions, and January 2027 for medication purposes and abbreviations. The OIG called the action plans acceptable.
The plans reward close reading. For appointment locations, VHA will develop a memo instructing facilities to use patient-friendly clinic names, and says solutions for facilities that have moved to the Federal Electronic Health Record are still being investigated. A memo, in other words, with a footnote reserved for the sites where the memo may not apply.
On risk-benefit discussions, VHA describes an initiative to strengthen informed consent for psychotropic medications, including standardized patient education materials, signature consent requirements for higher-risk prescribing scenarios, and new documentation tools in the electronic health record. Inference: VHA regards the existing documentation requirement as inadequate for at least some drugs. The basis is VHA’s response in Appendix A. The reasoning is that agencies do not add signature requirements to a process they believe is working. The same response says the policy rewrite will reinforce the existing requirement that practitioners hold a risk-benefit discussion before any new treatment. The requirement existed, went undocumented at nine of ten inspected sites, and will now exist again in a new format.
The medication-purpose response invites a narrower inference. The OIG recommended that discharge instructions include the purpose of each medication listed. VHA’s reply stresses that the review rests on acute hospital stays, and says Pharmacy Services education will emphasize newly prescribed medications and those relevant to the reason for admission. Inference: the plan may aim at a subset of the medications the recommendation covers. The basis is the wording of Recommendation 2 and VHA’s reply to it. The reasoning is the gap between “each medication” in one text and “newly prescribed” in the other. The OIG accepted the plan, so perhaps the gap is smaller than it looks.
Two smaller notes. The OIG records that it received an initial VHA memorandum on September 1 and a revised one on September 28, and the report does not say what changed. And the methodology section discloses that the inspection team used Microsoft Copilot Chat to review inspection data, checked the output against source material, and retained responsibility for the content. It also uses the term “super intelligence (SI)” in a sentence where “AI” would be expected. We report the wording without drawing conclusions from it.
The veterans in the sample left an acute psychiatric stay with paperwork they could not fully read and, in many cases, a record that does not show they were told what their new medications could do. VHA’s target dates run from December to January, and the OIG says it will follow up until the actions are complete. Everyone, in other words, will be monitoring something. The compliance failures documented here are not isolated: at the VA Nebraska-Western Iowa Health Care System, a Legionella case went unaccounted for in a required water-safety assessment and an infection-control fix declared complete in 2023 had not held by 2025. At the VA West Texas Healthcare System in Big Spring, a test-results policy the agency was required to have in place by 2024 carries a target completion date of March 2027. And a separate OIG audit found that VHA’s three-day specialty care standard for mental health was missed at a rate that stretched waits to ten days.
