Ever heard of a bad apple? Every organization and profession has them. Lawyers, doctors, policeman, accountants. You name it, life is full of ‘em. Those of you in the Puget Sound area may know of just one bad apple, James Joubert, M.D. (Maybe it’s French?). I did some digging on this ‘ole boy. At first, VA said hehe, we got nothing on him, then finally they came back with the goods.
“No Records”
On August 26, 2026, The Veteran’s Spectator filed a Freedom of Information Act request for records concerning James I. Joubert, M.D., NPI No. 1538259619. We said in the first paragraph of the request exactly who he was and exactly where he worked. He is a VA employed provider at the VA Puget Sound Health Care System, Seattle VA Medical Center.
We asked for the ordinary machinery of his job. The VA Forms 21-2507 and exam requests tasking him with work. Records to show how many Compensation and Pension examinations, Disability Benefits Questionnaires, and medical opinions he was asked to produce, by month and by type. Copies of the opinions themselves, with every Veteran’s name, Social Security number, claim number and date of birth redacted. We said so in the request, twice. Insufficient-exam returns. Quality reviews. Training records. Clinical privileges. Local policy on how examiners at that facility are told to approach a nexus opinion.
In other words, anything that would reveal an apple rotten to its core.
Six days later it was over.
“MDEO provided a negative response. Dr. Joubert is a VHA employee and does not complete contracted examinations. Therefore, we are issuing a no records response.”
Read that twice, because the whole of this story is in it.
The Veterans Benefits Administration sent a request about a Veterans Health Administration physician to the Medical Disability Examination Office—the office that administers contract examinations. MDEO looked and, correctly, found nothing, because contract examiners are not VA employees and Dr. Joubert is. VA then took that answer “he works for us,” so he isn’t in the contractor file and converted it into a finding that no records exist anywhere in the Department.
Well, no duh. The search criterion, by VA’s own description, was his name. Not his NPI number, which we gave them. Not “DBQ,” not “Compensation and Pension,” not “medical opinion,” not “nexus.” His name, typed into the one system guaranteed not to contain him.
FOIA Request No. 26-26184-F, filed by The Veteran’s Spectator on August 26, 2026, seeking records concerning James I. Joubert, M.D., NPI No. 1538259619, VA Puget Sound Health Care System. 10 pages.
DownloadOf course, we appealed
We appealed the same day.
Under the FOIA statute, and this is the part the VA hates, you have to make a reasonable search. If you are asking for records about Z, the VA can’t open up the top drawer of the filing cabinet that holds A-C, slam it shut and say “yeah, nothing about Z in here.” Well, you forgot to check the bottom drawer, you know, the one where they stuffed all those claims back in the day.
So VA waived the white flag and on September 2, VHA’s FOIA office transferred the request to the VA Puget Sound Health Care System, i.e., to the facility where the man actually works. So we kept going back and forth a little with the VA. Can you clarify what you are looking for, etc.
And then, finally, this week, 64 days later, VHA finally gave up part of the goods.
One search inquiry, down the hall, to the Chief of Compensation and Pension. 139 pages. The records had been sitting in the office of the service that employs him the entire time. Who would’a thunk?
The complete correspondence: VA’s September 1, 2026 “no records” decision; the September 1 administrative appeal to the Office of General Counsel; VHA’s September 2 transfer to VA Puget Sound; the September 8 clarification letter; the September 13 response; and the November 4, 2026 closure letter. 27 pages.
DownloadWhat is in the 139 pages
The first page is a letter dated January 17, 2024, from the VA Puget Sound credentialing office, approving the renewal of his clinical privileges. Not much there.
The privileging packet that follows says what the privileges are for, and the language is worth keeping in mind for everything that comes after:
Not for therapeutic purposes. Nobody in this arrangement is anybody’s doctor. And in my experience, no one is going to him for therapy anyhow.
Then come the workload logs, and this is where you start to understand the scale of what one physician’s signature carries. Take the single complete month in the production, August 2026. The log shows five in-person examinations and 43 “ACE” reviews (records review only), in which the examiner never sees the Veteran at all and renders an opinion from the paper. 48 exam and opinion assignments, in one month.
Behind them is a scheduling log titled “Joubert — FORM 2507 DAILY REVIEW 6.1.26 to 9.10.26.” It covers fourteen weeks. In those fourteen weeks, exam requests were routed to this one neurologist in Seattle from regional offices in Philadelphia, Houston, Cleveland, Waco, Muskogee, Phoenix, San Diego, Montgomery, San Juan, Providence, Hartford, Salt Lake City, Atlanta, Los Angeles, Oakland, Seattle, Lincoln, Winston-Salem, Nashville, Chicago, St. Paul, St. Louis, Huntington, Detroit, Des Moines, Jackson, Portland, New York, Newark, Milwaukee, Little Rock, Denver, Buffalo, Boston, Boise, and Sioux Falls. 36 offices. The scheduling notes are brisk and identical: “SCHEDULE 2 HOURS WITH DR. JOUBERT.” “SCHEDULE 4 HOURS WITH DR. JOUBERT.” “TBI HUB — NEURO IMO.”
A Veteran in San Juan does not know that the man deciding whether his traumatic brain injury is related to his service is in Seattle and has two hours blocked for him. He never will. That is not an abuse; it is how the system is built. But it is worth saying out loud, because the reason this matters comes next.
Not to mention that when we review a file, truly, most times it is a 6-8 hour endower, and that usually just scratches the surface. But I guess I won’t go on a tangent about that.
The 139 pages released by VA Puget Sound Health Care System on November 4, 2026: clinical privileges and credentialing, monthly workload logs, the Form 2507 daily review for June 1 to September 10, 2026, the completed and insufficient-exam trackers, CACHE quality review scores, and Talent Management System learning history. Veteran identifying information was redacted by VA. 139 pages.
DownloadKeeping it 100, every time
VA grades these examinations. The quality system is called CACHE, and the production contains forty of Dr. Joubert’s scored reviews, running from February 17, 2023, to April 27, 2026. Sleep apnea, headaches, peripheral nerves, TBI initial, TBI review, ENT, respiratory.
39 of the forty scored 100.
The 40th, on October 20, 2025, scored 86.
We want to be careful here, because a quality score is not a verdict on whether an opinion was right. CACHE measures whether the report is complete and internally consistent. Whether the boxes are filled, the questions answered, the rationale present. It does not and cannot measure whether the rationale is sound. A perfectly formatted opinion that reaches an indefensible conclusion scores a hundred.
Which is precisely the problem. Over the same stretch of years in which VA’s quality system was giving this examiner a perfect score on nearly every report it looked at, the Veterans Court was (and is) vacating Board decisions built on his work. Those two facts cannot both be measuring the same thing. One of them is measuring the paperwork.
The other tracker
There is a second set of numbers in the production, and it points the other way, which is why we are including it.
VA keeps an “Insufficient Tracker” — a log of examinations that the Veterans Benefits Administration sent back to the medical side as insufficient. The production contains 26 such entries for Dr. Joubert between October 21, 2025, and July 27, 2026: sleep apnea, migraines, Parkinson’s disease, back conditions, peripheral nerves, TBI, ear conditions.
In all 26, the column headed “INSUFFICIENT WARRANTED?” reads NO.
The tracker’s comment field explains most of them in the same handful of words, over and over: “Insufficient not warranted; issue = pdf DBQ viewing in VBMS.” The benefits side could not open the document. So, it marked the examination insufficient. So, the medical side cancelled the insufficiency and sent it back. So, the two halves of one Department spent nine months arguing about a PDF viewer, and in every one of those 26 files a Veteran was waiting.
How rich.
One entry is worth reading in full, because of what it says about how little the two sides can see of each other:
A rating activity somewhere returned an examination because it could not establish that the person who signed it was a physician. The medical side had to write back and point at the signature block.
We do not report this as a mark against Frenchy. We report it because it is the texture of the thing, because any honest account of a VA file has to include the pages that complicate the story, and because a system in which the claims side and the medical side cannot read each other’s documents is a system that is going to produce bad outcomes no matter who is holding the pen.
What they would not give us
VA produced the workload. It produced the quality scores. It produced forty pages of his Talent Management System learning history. Pages numbered 104 through 143 of a 295-page record, showing privacy training, HIPAA training, workplace civility, suicide risk identification, falls prevention, antimicrobial stewardship, and, scattered among them, 12 Disability Medical Assessment course completions: musculoskeletal examinations, mental health examinations, Gulf War general medical, military sexual trauma, cold injury residuals, aid and attendance, foot, hand and fingers, heart. And one course titled “DMA: Medical Opinions — Update 2022,” completed January 4, 2023.
Then, in the same letter that enclosed all of it, VA said this:
Please be advised, I can neither confirm nor deny records responsive to your request exist.”
The refusal to say whether records exist at all, citing the privacy exemption. On training. In the same envelope as the training records.
VA also withheld the names of other providers appearing in the logs, and the stated reason deserves to be quoted in full, because it is an unusually candid description of what these numbers are for:
And on the last item, any local policy, memorandum or instruction telling examiners at VA Puget Sound how to approach a nexus opinion, or the “at least as likely as not” standard, or a recommendation to sever or reduce service connection, the answer was a clean no records, with seven words of explanation:
We intend to find out what the National Guidance says. That is the next request.
Excuse me Mr. Secretary, what about Charles Gardenhire?
Now the part that is not about paperwork.
Charles F. Gardenhire served on active-duty from September 1950 to September 1953, and again from January 1955 to March 1956. Somewhere in the first enlistment he contracted encephalitis. It is as nasty as it sounds, an inflammation of the brain. That is putting it lightly. He was treated as an in-patient with antibiotic therapy in 1951. He finished his tour. He went home. He came back and enlisted a second time, and he left the service again in 1956, he said, because of the headaches.
In December 1965, VA granted him service connection for a history of encephalitis and rated it zero percent, effective August 1965. He did not appeal. Nearly fifty years later, in 2013, he filed a motion alleging clear and unmistakable error in that 1965 decision, and in July 2016 the Board of Veterans’ Appeals agreed with him: the effective date was wrong. The correct date was November 16, 1953 — the day after he came home from his first enlistment.
So, the rating period at issue in this case opens in 1953. VA eventually assigned ten percent from November 16, 1953, to January 14, 1955, and ten percent again from March 17, 1956, to September 29, 2009, at which point the rating jumps to one hundred.
Mr. Gardenhire’s position, for the remainder of his life, was that the 10 percent was wrong. That the encephalitis had been active and that its residuals had been with him the whole time. He described numbness in his lower lip. Headaches. Fatigue. Shortness of breath. Memory trouble. Sleeplessness. He testified he had not slept more than four hours at a stretch in sixty years. Anxiety. Agitation. Dizziness and falls. Slurred speech.
The relevant regulation is not a hard one to apply. Diagnostic Code 8000 rates encephalitis at ten percent for minimum residuals and one hundred percent for active febrile disease. And the note that governs the whole family of codes says this:
Subjective residuals will be accepted. The rating schedule anticipated exactly this problem — a brain infection whose leavings are headaches and dizziness and fatigue, none of which show up on a scan — and wrote the answer into the regulation seventy years ago.
There is one more fact that matters enormously. VA lost his treatment records. On March 14, 2013, the VA issued a formal finding his records were known only to Jesus, and in a September 2018 letter it told him that the records “cannot be located and therefore are unavailable for review,” that “all efforts to obtain the needed information have been exhausted,” and that further attempts “would be futile.” When the government loses the evidence, its obligation to consider the benefit of the doubt is heightened. O’Hare v. Derwinski, 1 Vet. App. 365, 367 (1991).
A man says he has had headaches since 1953. The regulation says accept it. The records that might have corroborated it were destroyed or misplaced by the agency now asking him to prove it.
Mr. Gardenhire died in July 2021, with the appeal pending. His widow, Petra L. Gardenhire, was substituted in September 2021 and has carried it ever since.
Board of Veterans’ Appeals decisions of December 4, 2024, and March 18, 2026, in the appeal of Petra L. Gardenhire, in the case of Charles F. Gardenhire, Docket No. 220422-238079. Both were vacated by the United States Court of Appeals for Veterans Claims. 65 pages.
Download“A medical expert who specializes in Neurology”
In December 2021, the Board remanded the case with an instruction that it had the authority to give and that VA has the obligation to follow. Quoting the remand directive:
Section 5109(a) is a specific tool. It authorizes the VA to obtain “an advisory medical opinion from one or more independent medical experts who are not employees of the Department.” The entire point of the statute — the only reason it exists — is to put the question in front of somebody outside the building.
The Board ordered an outside neurologist.
What VA obtained, on January 19, 2022, was a medical opinion from, none other than, its own full-time Compensation and Pension neurologist at VA Puget Sound: Dr. James I. Joubert.
It gets narrower than that. Dr. Joubert had already written an opinion in this case, in November 2020. In its December 2021 remand, the Board had expressly found that opinion — along with the October 2020 and January 2021 opinions — inadequate, and said why:
So, the Board found the November 2020 opinion inadequate, ordered an independent expert, and VA went back to the same examiner who wrote the inadequate opinion.
And in the January 2022 opinion, according to the Board’s own summary of it, that examiner “noted that they stand by the opinion provided in November 2020.”
He stood by the opinion the Board had already thrown out. Because I said so.
The weight of the medical literature
Here is how the January 2022 opinion disposed of the case. Every symptom Mr. Gardenhire reported was assigned to something else. The lower lip numbness, oh that’s a problem with the third branch of the trigeminal nerve, somewhere outside the skull. The headaches, oh that’s occipital neuralgia, not service connected. The fatigue, memory loss, anxiety, agitation and insomnia, oh that’s PTSD. The memory loss, again, oh that’s vascular dementia, on the strength of an MRI. The fatigue, again, that’s congestive heart failure and renal failure. The shortness of breath, dizziness and slurred speech, you guessed it, heart failure and hypertension. For sure not that claim that goes back to 1953.
Joubert had an answer to everything, to ensure the claim would not be paid back to 1953.
The authority offered for this, over and over in the Board’s recitation of the opinion, is a single phrase. The findings are supported by “the weight of the medical literature.” By “the weight of the medical literature and the examiner’s expertise concerning human neuroanatomy.” By “the evidence of record and the weight of the medical literature.” By “the objective evidence of record and weight of the medical literature.”
The Board’s summary of this one opinion invokes the medical literature at least seven times.
It does not identify a single article, study, textbook, author or source. Not one.
That’s not a quibble about citation format. When the Appellant finally got the chance to say so, in a November 2025 submission to the Board, it was the center of the argument. Again in September 2026 the VA signed yet another joint motion conceding the point:
An unnamed literature cannot be checked, cannot be rebutted, and cannot be wrong. That is the whole function of invoking it. It converts a medical opinion into a credential. This is so because a neurologist says it is so, and a credential is not evidence.
I will say the next part in my own voice, as the editor of this publication and as someone who has sat across from this examiner’s work in more than one claims file.
This is the pattern I have seen from Dr. Joubert again and again. The conclusion arrives first, the reasoning is a gesture at authority, and the authority is himself. I said no, and I am a neurologist. That is an argument a first-year philosophy student is taught to recognize and reject by name, and in VA practice it carries the day for years at a time because the only two institutions positioned to test it — the rating board and the quality review system — are not built to test it. The rating board is not medically trained. CACHE counts whether the box was filled in. Nobody between the examination room and the Court ever asks the one question that matters: based on what?
That is my opinion of his work, formed from his work. The facts in this article are from the documents, and the documents are below. Readers can form their own.
Twice to the Court
The Board adopted the January 2022 opinion, called it “adequate and complete,” and denied the claim on December 4, 2024.
Mrs. Gardenhire appealed. On October 9, 2025, the parties filed a joint motion agreeing that the Board had erred. It had relied on the October 2020, November 2020 and January 2021 opinions without ever explaining how it could do so after having found those same opinions inadequate.
The Board tried again. On March 18, 2026, it denied the claim a second time, on the same opinion.
Mrs. Gardenhire appealed again. On September 21, 2026, the parties filed a second joint motion, and this one reached the thing underneath:
The joint motions for remand filed in Gardenhire v. Collins, Nos. 25-0473 (October 9, 2025) and 26-2988 (September 21, 2026). Both were granted. 11 pages.
DownloadTwo trips to Court. Nearly 5 years since the remand that was supposed to fix it. A Veteran dead since 2021. A widow still carrying his claim. A claim the VA has been working on since November 16, 1953.
And the VA’s position on all of it was that it had no records.
What we are doing next
This FOIA is not finished.
VA produced the counting and withheld the work. We have the logs that show 48 assignments in August 2026 and 36 regional offices in fourteen weeks, and not one of the opinions those numbers represent. We asked for the DBQs and the medical opinions themselves, redacted of every Veteran identifier. We offered the redactions in the request itself, before anyone asked us to. The November 4 letter answers that category by announcing, in a single sentence, that it is “withholding all records” under the claims confidentiality statute and the privacy exemption. It then encloses a hundred and thirty-nine redacted pages. Both things cannot be true, and neither the letter nor anything in it explains what, in a medical rationale with the Veteran’s name removed, invades anybody’s privacy.
That is what is being appealed. A medical rationale with the Veteran’s name taken off it is not the Veteran’s private information. It is the government’s reasoning, offered in support of a government decision, and the public that pays for it is entitled to read it.
We are also asking for every nexus opinion he has authored, and for the national guidance that VA Puget Sound says it follows.
Mrs. Gardenhire’s case goes back to the Board, for the third time, on a record that still contains the January 2022 opinion. Somewhere in Seattle, a full-time Compensation and Pension neurologist with a wall of perfect quality scores has four hours blocked out tomorrow for a Veteran in a city he has never been to.
Excuse me, Mr. Secretary. That claim starts in 1953.
Godspeed.
Documents below. All exhibits are VA records produced under FOIA or filings made on the public docket of the United States Court of Appeals for Veterans Claims. Veteran identifying information in the FOIA production was redacted by VA.
The complete file
FOIA Request No. 26-26184-F, filed by The Veteran’s Spectator on August 26, 2026, seeking records concerning James I. Joubert, M.D., NPI No. 1538259619, VA Puget Sound Health Care System. 10 pages.
DownloadThe complete correspondence: VA’s September 1, 2026 “no records” decision; the September 1 administrative appeal to the Office of General Counsel; VHA’s September 2 transfer to VA Puget Sound; the September 8 clarification letter; the September 13 response; and the November 4, 2026 closure letter. 27 pages.
DownloadThe 139 pages released by VA Puget Sound Health Care System on November 4, 2026: clinical privileges and credentialing, monthly workload logs, the Form 2507 daily review for June 1 to September 10, 2026, the completed and insufficient-exam trackers, CACHE quality review scores, and Talent Management System learning history. Veteran identifying information was redacted by VA. 139 pages.
DownloadBoard of Veterans’ Appeals decisions of December 4, 2024, and March 18, 2026, in the appeal of Petra L. Gardenhire, in the case of Charles F. Gardenhire, Docket No. 220422-238079. Both were vacated by the United States Court of Appeals for Veterans Claims. 65 pages.
DownloadThe joint motions for remand filed in Gardenhire v. Collins, Nos. 25-0473 (October 9, 2025) and 26-2988 (September 21, 2026). Both were granted. 11 pages.
Download