GAO 27-108414: The Crisis Call Was Answered. Did the Veteran Connect to Care?
What GAO found
The Veterans Crisis Line provides immediate support by phone, text, and chat. Some people who use it also request follow-up care. For those requests, the Department of Veterans Affairs (VA) tracks whether suicide prevention coordinators at local VA medical facilities attempt to contact veterans. In its November 18, 2026 report, the U.S. Government Accountability Office (GAO) found that VA did not systematically track whether coordinators reached veterans or whether those veterans connected to care in accordance with their needs and preferences. GAO-27-108414
GAO recommended that VA systematically track both outcomes. It separately recommended identifying alternative emergency response resources in jurisdictions that had consistently declined to dispatch responders. VA concurred with both recommendations. This GovKM analysis focuses on the follow-up recommendation. GAO findings and recommendations
The Organizational Information Continuity challenge
Can the next coordinator tell whether the veteran was reached? A referral sent, a referral accepted, and a completed conversation are separate events. If the receiving facility’s acknowledgment and the coordinator’s contact record cannot be traced to the original request, staff taking over the case may have to reconstruct what happened before they can act. An AI assistant could read a completed outreach task as confirmation that a conversation took place. A record of an attempt supports only the claim that outreach was attempted.
Can the program verify a connection to care? A successful conversation may lead to an appointment, a different service, or more time for the veteran to decide. A scheduled appointment and a confirmed connection to care represent different outcomes. If reporting combines them, staff may overstate how many veterans received a service that matched their stated needs and preferences. An AI-generated summary could repeat that overstatement at scale or interpret missing evidence as proof that care never occurred. Both are prospective risks identified by GovKM.
GovKM’s proposed approach
- Assign ownership at the handoff. Have program leaders identify the authorized systems for the initial request, facility acknowledgment, contact attempt, successful contact, and outcome. Record the responsible role, event time, and next action under the applicable VA procedure. Preserve the link to the underlying record when a different facility or coordinator takes over. 44 U.S.C. § 3101 establishes the federal agency duty to make and preserve adequate documentation of agency functions, decisions, procedures, and essential transactions. The workflow is GovKM’s proposed way to support that duty.
2. Define what confirms each outcome. Have clinical and program leaders decide which authorized source establishes successful contact, what evidence confirms connection to care, and how to record whether the service aligned with the veteran’s needs and preferences. A verified encounter or an authorized confirmation from the veteran could support an outcome under approved criteria. Keep pending, declined, and unverified outcomes distinct. Show how many outcomes remain unverified alongside any reported rate so reviewers can see the limits of the measure.
3. Keep sensitive details in approved systems. Have VA privacy, security, and records officials approve the purpose, fields, access, and retention before linking records or enabling reporting. Use approved references and limited outcome fields for oversight where permitted. For covered systems of records, the Privacy Act limits information maintained about individuals to what is relevant and necessary for a qualifying agency purpose and requires appropriate safeguards. Have officials review disclosure authority and any additional confidentiality rules that apply to the records, including the protections for certain VA medical records. 5 U.S.C. § 552a · 38 U.S.C. § 7332
OrgAI’s proposed role is to help authorized users trace a reported outcome to the permitted evidence after a handoff or system change. That capability is in development. VA clinicians and authorized officials retain responsibility for care and disclosure decisions.
Verify that the outcome can be reconstructed
Start with synthetic cases. Give a reviewer a referral accepted by a facility, an outbound call that reached voicemail, an appointment that was scheduled, and a confirmed connection to care. Check that the records and any AI-assisted report distinguish those events, surface missing evidence, and withhold restricted details. Repeat the test after a simulated staff handoff. The reviewer should be able to identify who owns the next action and trace any claimed outcome to its supporting evidence.
The same approach can help a hospital or community service provider evaluate referrals under its own legal and policy obligations. Start with one handoff and ask whether the receiving team can verify what happened after accepting it.
Contact GovKM to discuss an Organizational Information Continuity assessment for referral handoffs, outcome evidence, or governed AI reporting.



