By Ahmed Jemaa, Co-Founder and CEO, Medrecords AI
LinkedIn: Ahmed J
LinkedIn: Medrecords AI
Every medical record has two lives. The first is clinical. A note is written, an order is placed, a result comes back, and the record moves between clinicians who need it to treat a patient. A decade of federal policy has targeted that first life, and records now travel between EHRs far better than they did in the Meaningful Use era.
The second life starts later, sometimes years later, when the same record is needed to decide something other than care: a personal injury claim, a workers’ compensation dispute, a long-term disability appeal, an independent medical examination, a malpractice case. That second life is where I work, reviewing records alongside physicians, attorneys, and nurse consultants. It is also the life our interoperability framework was never designed for.
The national framework, TEFCA, authorizes six Exchange Purposes: treatment, payment, health care operations, public health, government benefits determination, and individual access services. Only one of those six, government benefits determination, serves decisions outside health care itself, and today it is built around a single agency.
That lane proves the point. The Social Security Administration receives about 2 million disability claims a year. By SSA’s own estimate, as relayed by ONC, an initial claim can take more than 200 days, and SSA spends over $500 million a year collecting and creating medical evidence. Under the new Government Benefits Determination use case, ONC says records that took weeks or months to collect could arrive in minutes or seconds. When the second life gets a defined purpose, a standard query, and an accountable network, the problem shrinks dramatically.
Everyone else still lives in the fax era. An IME physician on a workers’ compensation file, an attorney on a spinal injury claim, a private disability carrier, and the nurse consultant who builds their chronologies all request records the old way: an authorization or subpoena, a release-of-information queue, and a package that arrives weeks later as a scanned PDF. Productions routinely run to several thousand pages per claimant, with no consistent order and no way to tell whether anything is missing. Duplicate pages commonly make up 10 to 20 percent of a production in the cases I review, and I have seen them reach 23 percent. On a 5,000-page file, that can mean a thousand pages someone is paid to read twice.
The costs land on both sides of the request.
For HIM and release-of-information teams, every second-life request is a manual event. Staff interpret authorizations, decide what belongs in the production, and track deadlines. The distinction between the legal health record and the broader designated record set, which HIM professionals have wrestled with for years, still gets applied by hand, request after request. OCR has now reached 55 enforcement actions under its Right of Access Initiative, a reminder that slow, manual release processes carry real regulatory exposure.
For the people who inherit the record, the damage is quieter but larger. When an EHR is flattened into a PDF, it loses the structure that made it useful: encounter context, the link between an order and its result, the timestamps that show what was known and when. Reviewers then spend hours rebuilding a timeline the source system already had. A single missing page can shift an impairment rating or a claim outcome, and nobody can say with confidence whether the package was complete.
None of this means providers should open the doors wider. Second-life requests carry different consent, minimum necessary, and privacy obligations than treatment, and they deserve careful governance. But careful governance is exactly what a defined exchange purpose provides. SSA’s lane runs on patient authorization, purpose-of-use controls, and audit trails. Lacking a lane does not make other disclosures safer, only slower, costlier, and harder to audit.
HIM leaders are well placed to push three moves.
First, treat second-life disclosure as a design requirement. When evaluating EHR exports and release workflows, ask whether a production can leave with its structure intact: a machine-readable index, such as FHIR DocumentReference or XDS metadata, carrying encounter dates, source facility, and page provenance alongside the PDF.
Second, standardize completeness. A simple manifest listing what was requested, what was produced, and what was excluded and why would eliminate much of the back-and-forth that follows every production.
Third, press for the next exchange purposes. SSA’s Government Benefits Determination SOP is a working template for consent, identity, and directed queries. State workers’ compensation systems are natural candidates to follow.
The fax is being phased out of care delivery for good reasons. It would be a shame if it survived another decade in the one place where a record’s accuracy is tested hardest: when someone has to decide what actually happened.