BlackLeafwatch the watchmen
Department of Veterans Affairs -- disability claims processing under the PACT Act

VA's own watchdog found errors in 61% of PACT Act claim denials

Summary

A VA Office of Inspector General statistical review of veterans' toxic-exposure disability claims denied under the PACT Act estimated that 61 percent of nonpresumptive claim denials from May through August 2023 contained at least one processing error, some of which could have cost veterans benefits they were owed. A separate internal VBA review in fiscal year 2024 found comparably high error rates, and claims processors told investigators the guidance they're required to follow is scattered across dozens of frequently updated documents.

By Augustus · July 16, 2026

The PACT Act -- the 2022 law itself calls perhaps the largest expansion of veterans' health care and benefits in its history -- opened a new path to compensation for veterans whose conditions can't be granted automatically ("presumptively") because they aren't on a pre-approved list. For these "nonpresumptive" claims, a caseworker must instead verify the veteran took part in a toxic exposure risk activity (TERA) -- a category 's own procedure calls "extremely broad," covering everything from solvents to burn-pit air pollutants -- and route the claim through a medical exam before ruling on it.

A Office of Inspector General statistical review of exactly these denials, examining a random sample of 60 decisions made from May through August 2023, found 37 had at least one processing error -- a projected 61 percent of the roughly 122,822 such denials issued in that four-month window alone. The mechanism of harm isn't abstract: in one sampled case, denied a veteran's Agent Orange-linked prostate claim a week after receiving it, without ever ordering the toxic-exposure exam the claim was entitled to.

The step VA added, and the four ways to skip it

Before the PACT Act, a nonpresumptive claim was decided on the paper record alone. Now, when a veteran claims -- or the record implies -- a condition tied to toxic exposure, 's Veterans Benefits Administration (VBA) must run a specific sequence: identify that the claim touches TERA, check the veteran's exposure history in the Individual Longitudinal Exposure Record (ILER, a joint Defense Department- database covering more than six million veterans' deployment and occupational-hazard data), request a medical exam and opinion linking the condition to that exposure, and then discuss all of it -- explicitly -- in the written decision the veteran receives. Each step is a place a rushed or under-guided caseworker can drop the thread, and 's sample shows all four failure points in use: exams skipped when required, the ILER lookup skipped, the required internal memorandum documenting exposure left uncompleted, and -- the single most common failure -- the decision notice itself never mentioning that toxic exposure was considered at all.

Nonpresumptive PACT Act denials with an error
61%
OIG's estimate for May-Aug 2023 denials, 90% confidence range 46%-74%
Estimated decisions with 1+ error
74,366
out of an estimated 122,822 in-scope denials over the same 4-month window
VBA's own FY2024 site-visit error rate
70%
for "TERA Conceded" cases; 40% for "TERA Not Conceded" cases, 40 cases reviewed
OIG's projection: how many of ~123,000 denials had an error
Estimated in-scope population vs. estimated decisions with at least one processing error, May-Aug 2023 review period
Estimated denials in scope
122,822
Estimated denials with 1+ error
74,366
Source: VA OIG Report 23-03357-156, Appendix B, Table B.2 (p. 27)
View data as table
OIG drew two random samples of 30 decisions each -- one from denials with no exam/medical-opinion request, one from denials that had one -- and weighted the combined results to project across the full in-scope population. Source: Appendix B, Tables B.1-B.2.
Sample reviewed60 decisions
Sampled decisions with 1+ error37 of 60
Projected error rate (90% CI)60.5% (46.3%-73.6%)

A second, independent count lands in the same range

's 90-percent confidence interval for the error rate runs wide -- 46.3 percent to 73.6 percent -- because a 60-decision sample can only narrow an estimate so far; that width is a function of sample size, not evidence the true rate sits at either edge. What steadies the finding is that VBA checked its own work independently and landed in the same neighborhood: in a fiscal year 2024 site-visit review, the agency's own consistency-and-oversight staff examined 40 PACT Act toxic-exposure cases and reported error rates of 70 percent for claims where exposure had been conceded and 40 percent for claims where it had not. Two different review teams, working from two different samples in two different years, both found VBA getting more than a third -- and by 's count, likely a majority -- of these decisions wrong.

Where the process actually breaks

Combining 's two sampling groups, the review team logged 86 distinct errors across the 60 decisions examined -- some decisions had more than one. The largest single category, by a wide margin, was a decision that simply never discussed toxic exposure at all: 24 of the 86 errors were rating decisions where TERA should have been addressed and wasn't, leaving the veteran with no way to know whether their exposure claim was actually weighed or just overlooked. A missing case-tracking flag (17 errors) and a skipped exposure database check (8 errors) followed. The category flags as most directly capable of costing a veteran their benefits -- a required exam or medical opinion that was never requested -- accounted for 11 of the 86, including the Agent Orange prostate case above.

What went wrong: 86 errors found across 60 sampled denials
Combined explicit + implicit-claim error instances, both sampling strata
TERA not discussed in rating decision
24
TERA claim attribute missing
17
Required exam/opinion not requested
11
TERA not discussed in favorable findings
11
Required ILER exposure lookup skipped
8
Required TERA memorandum missing
6
Exam requested when not needed
4
Higher-level review skipped duty-to-assist error
4
Examiner not given relevant evidence
1
Source: VA OIG Report 23-03357-156, Appendix C, Tables C.1-C.2 (pp. 30-32)
View data as table
TERA not discussed in rating decision24The veteran's own decision notice never says whether toxic exposure was weighed
TERA claim attribute missing17Case record lacks the tracking flag that identifies it as a toxic-exposure claim
Required exam/opinion not requested11Can directly cost a veteran benefits they're entitled to
TERA not discussed in favorable findings11
Required ILER exposure lookup skipped8The DOD-VA deployment/exposure database was never checked
Required TERA memorandum missing6
Exam requested when not needed4Burdens veterans and wastes exam funding on unnecessary appointments
Higher-level review skipped duty-to-assist error4
Examiner not given relevant evidence1

Why: guidance scattered across dozens of documents

VBA's Adjudication Procedures Manual is meant to be the one place claims processors go -- explicitly designed, per VBA's own prologue language, to be "reasonably complete" and "easy to understand and apply." It isn't, for PACT Act nonpresumptive claims: when began its review in October 2023, a separate internal VBA web page carried 43 links to the actual governing guidance -- an 82-page standard operating procedure, eight other policy documents, nine general-information documents, 19 job aids, and a 94-page FAQ -- and that guidance kept changing under processors' feet, with roughly four dozen prior versions archived as rescinded and replaced. Interviewed in January 2024, claims processors told that finding the applicable rule across all those locations was time-consuming, and some said the difficulty may have contributed directly to the error rate went on to measure.

The under secretary for benefits told Congress as much in May 2023, calling PACT Act claims "a massive new benefit" that added "new complexity to the work process... a new theory of entitlement [with] different effective dates and different conditions that are covered." VBA has run two of its own internal reviews since -- one in early 2023, one across FY2024 -- and both surfaced errors; neither, per , was followed by the root-cause analysis that would explain why they keep happening.

What VA has promised, and by when

issued three recommendations, and VBA concurred with all of them: review and correct every processing error the team found on the cases it sampled and report back on the results; work with the Secretary's office, the Office of Field Operations, the Office of General Counsel, and the Board of Veterans' Appeals to fold PACT Act guidance into the single Adjudication Procedures Manual; and evaluate whether its existing controls for nonpresumptive TERA denials are strong enough, adding new ones where they aren't. None of the three commitments carries a completion date in 's response -- "concurred" here means VBA agreed to act, not that it has. 's own practice is to keep a recommendation open, and keep checking, until it receives evidence the action was actually taken; as of this review's publication, none had closed.

  • 's statistical sample projects that 61% of the roughly 122,822 nonpresumptive, toxic-exposure PACT Act claim denials issued from May-Aug 2023 contained at least one processing error -- an estimated 74,366 decisions, some of which could have cost veterans benefits they were entitled to.
  • A second, independent VBA internal review in FY2024 found comparably high error rates -- 70% for 'TERA Conceded' cases, 40% for 'TERA Not Conceded' cases -- corroborating 's finding from a different sample.
  • The single most common error, in 24 of 86 instances logged, was a decision that never discussed toxic exposure at all, leaving veterans no way to know whether their exposure claim was considered or simply skipped.
  • traces the error rate partly to guidance spread across dozens of frequently updated documents rather than VBA's own manual; VBA concurred with all three of 's recommendations but has not attached completion dates, and had closed none of them as of publication.

Figures here run through 's review period (October 2023-May 2025) and reflect denials from a single four-month window, May-August 2023; has not published a comparable statistical estimate for later periods. has continued processing PACT Act claims since, including the broader backlog-reduction push covered separately by BlackLeaf's reporting on the 's overall disability claims backlog -- that reporting concerns claim volume and speed, a different question from the accuracy of denials measured here.

Sources(2) ▾
  • Department of Veterans Affairs Office of Inspector General, Better Controls Needed to Accurately Determine Decisions for Veterans' Nonpresumptive Conditions Involving Toxic Exposure Under the PACT Act (Report 23-03357-156) (2025-09-30)The 's statistical review of how VBA processed denials of nonpresumptive, toxic-exposure-related PACT Act claims from May-Aug 2023. Source of the 61% (60.5%) estimated error rate, the 122,822-decision in-scope population, the sample design (60 decisions, 37 with errors), the error-category breakdown (Appendix C), VBA's own FY2024 internal error findings (70%/40%), the dispersed-guidance finding, the three recommendations, and the site-visit locations. vaoig.gov
  • United States Congress / U.S. Government Publishing Office, Public Law 117-168 -- Sergeant First Class Heath Robinson Honoring our Promise to Address Comprehensive Toxics Act of 2022 (PACT Act) (2022-08-10)The statute itself -- confirms the PACT Act's official short title, enactment date (Aug. 10, 2022), and public law citation (136 Stat. 1759), which the report's framing of the law's scope and toxic-exposure provisions rests on. govinfo.gov
Weekly digest: the most-read systems, in brief. Mondays.

Comments

Always open. Logged-in readers can annotate paragraphs in place.

Loading comments…
or log in to comment under your account