FHIR R4 · X12 005010 · HL7 v2
Charles Trinh — Healthcare Interoperability | Las Vegas, NV · Las Vegas, Nevada · Henderson · Clark County, Nevada · North Las Vegas, Nevada

Five problems wearing one word
Interoperability covers at least five distinct problems discussed as though they were one: moving clinical data between systems, moving money between organisations, proving who a patient is, proving what a patient consented to, and proving that any of it happened.
Each has different standards, different failure modes, and a different owner inside the organisation. Programmes that treat interoperability as a single initiative produce a roadmap that satisfies nobody, because each layer needs a different person to say yes and a different measure of done.
The practical benefit of separating them is diagnostic. Every real failure in health data exchange is a failure at one specific layer, and knowing which layer takes an hour of tracing rather than a quarter of architecture review.
The wire
HL7 v2 carries the hospital's internal event stream and will for years. FHIR R4 is the modern surface for anything new. CDA carries documents that require narrative and attestation. The wire is largely solved and mostly boring, which is the correct state for a transport layer.
Boring does not mean maintenance-free. MLLP listeners without TLS, interfaces with no owner, and connectors that have been running since before anyone now employed arrived are all wire-layer problems that present as clinical problems when they fail.
The wire is where you should spend the least design effort and the most operational discipline. It rarely needs invention; it almost always needs monitoring, an owner, and a documented message flow with a real sample attached.
The transaction layer
X12 005010 moves eligibility, submission, remittance and prior authorisation. This is where money moves, and where defects cost cash rather than clicks. An error in an eligibility response is revenue; an error in a 278 is care delayed or denied.
The layer is mature, deterministic, and unglamorous, which is why it attracts less attention than FHIR and produces more avoidable loss. Enrolments lapse, acknowledgement loops go uninstrumented, and authorisation numbers fail to reach the claim they belong to.
Instrument it at the level of outcomes: submitted, acknowledged, accepted, rejected by class, paid. For any single day last month, on demand. An organisation that cannot produce those numbers does not have a claims pipeline, it has a file transfer with optimism attached.
The network
TEFCA and the QHINs, IHE profiles on the document fabric, direct messaging at the edges. The network is a routing and trust layer. It holds no data, and it cannot compensate for a matching failure at either end of a query.
This is the layer most likely to be built first and understood last, because a network connection is a visible deliverable that can be announced. What it delivers depends entirely on whether the participants are enabled and whether a query finds the right patient when it arrives.
The measure is not connectivity. It is the number of participants who have actually exchanged clinical content in the last thirty days, and the ratio of queries issued to responses that contained something a clinician could use.
The identity layer
Deterministic identifiers where they exist, scored probabilistic matching where they do not, and a human review path for the middle band. The identity layer decides whether everything above it is trustworthy, and it is the layer most often deferred because it produces no announcement.
The asymmetry is the design principle. A duplicate record fragments a history and causes a repeat workup. A wrong merge puts one patient's data in another patient's chart with full confidence behind it, and no downstream control repairs it because nothing was breached.
Build unmerge before merge. If a false positive is discovered after the records are combined and the system cannot separate them without destroying the audit trail, a correctable error has become permanent, and the organisation will choose to live with it rather than admit it.
The consent layer
Directives with periods, purposes, and categories, enforced in the data path, with the withholdings logged. Consent as a boolean cannot express what a patient actually decided, so organisations default to the strictest interpretation and the exchange delivers less than the patient expected.
Enforcement belongs before serialisation. Filtering after the fact means the data already left the boundary and the log records a leak rather than a control. And the log must record what was withheld, not only what was sent, because otherwise correct withholding is indistinguishable from a bug.
Sensitive categories make this concrete. Substance use records under 42 CFR Part 2 sit inside the same chart as everything else, so the enforcement point is the segment, evaluated at read time, per requester and per purpose. Record-level decisions are wrong in both directions.
Failure: building the visible layer
Network connections, endpoints and dashboards are announceable. Match quality and consent enforcement are not, and they are where the defects live. The result is an exchange that demonstrates beautifully and returns nothing in production.
It looks like success because the metrics available are the ones that were easy to collect: connections established, messages routed, endpoints live. None of them tell you whether a clinician received a record they did not already have.
The correction is to make the invisible layer produce a number. Queries issued against responses with clinical content, per source, per week. That single series will tell you more about the programme than every connectivity dashboard combined.
Failure: optimising for match rate
Match rate rewards the cheap error and punishes the expensive one. Any team measured on it will tune toward merging records, because merging raises the rate and the cost of the mistake is paid later by somebody else.
The correct framing is asymmetry with a staffing consequence: set thresholds for the expensive error, and fund the review queue that the resulting middle band produces. A matcher tuned to eliminate review is a matcher tuned to merge.
This is a governance decision disguised as a configuration value. It should be made by whoever owns the clinical risk, recorded with its rationale, and revisited when the review queue volume changes.
Failure: instrumenting nothing
No per-source error rate, no acknowledgement loop, no disclosure log, no record of what was withheld. Without instrumentation every failure is discovered by a clinician or a patient rather than by a monitor, which means it is discovered late and at the worst possible moment.
The minimum viable instrumentation is four numbers: queries issued, responses with clinical content, rejections by class, and withholdings by rule. Those four separate a working exchange from an exchange with a good dashboard.
Add one trace: a single patient through the full path, timestamped at every hop. An organisation that can produce that in an hour knows its architecture. One that offers a diagram instead has answered the question.
Conformance is a milestone, not an exam
The ONC (g)(10) suite, IHE Connectathon testing, C-CDA validation. All of them are most useful early, when a defect costs an afternoon and a commit rather than a release cycle and a customer conversation.
I have taken a platform to 317 of 317 subtests on that suite. What the process teaches is that the last fifteen percent of failures are never algorithmic: a required element left null, a profile not declared, a search parameter returning the right resource with the wrong include.
Schedule the run as a milestone with a date, early enough that findings can still change the design. Teams that run it in week two ship in month two, because they discover the shape of the problem while there is still room to move.
The ratio and the trace
One ratio and one trace, and everything else is commentary. The ratio is queries issued against responses with clinical content, reported per source. The trace is one patient through the full path, timestamped at every hop.
Per-source reporting matters because the problem is almost always concentrated. A blended number hides the single interface or registration site producing the duplicates, and aggregate dashboards will never localise it for you.
Run both monthly. The ratio tells you whether the programme is delivering; the trace tells you where it stops. Neither requires a steering committee, and both are answerable in an afternoon by anyone who knows the estate.
Build order that holds up
Identifiers first. Then matching instrumentation. Then unmerge before merge. Then consent as directives. Then segment-level enforcement for the categories that legally require it. Then the network connection, because by then everything the network depends on is already correct.
Most programmes do this in reverse, because the network is the visible deliverable and the identity work is invisible. The visible deliverable then propagates the invisible defect to every partner connected to it, which is how one matching configuration becomes an industry-wide duplicate problem.
I have built and operated this stack across a 51-hostname estate on a single AWS account under an active business associate agreement, with a 337-table clinical store, eight X12 005010 transaction sets, 104 clearinghouse enrolments across 53 jurisdictions, and a Fabric plus EVM ledger pair for consent and disclosure records. If your exchange returns fewer records than it should and nobody can say why, that is the first conversation.
The regulatory context, in one paragraph each
CMS-0057-F sets four API obligations for payers: patient access, provider access, payer-to-payer exchange, and prior authorisation. The deadlines have already begun to arrive, and the fourth obligation is the one that changes an operating model rather than adding a surface.
Information blocking inverted the default. Withholding is now the position that requires a documented exception, and the private exception is conditioned on consent or a legal permission rather than on organisational discomfort. An organisation that cannot name its exception and its condition for a specific withheld exchange is blocking and hoping nobody asks.
TEFCA provides the network hierarchy and the permitted purposes. It is a routing and trust layer that holds no data, and it cannot compensate for a matching failure at either end. Onboarding is a conformance exercise followed by an operational one, and the operational half is where programmes sit for months.
42 CFR Part 2 governs substance use records and requires consent with specific elements. Because the data sits inside the same chart as everything else, enforcement cannot be a record-level decision. It has to be a segment-level decision applied at read time, per requester, per purpose.
USCDI names the data elements that must be exchangeable. Mapping those elements onto resources is translation work, and every element you cannot source cleanly is a gap that will appear as a null in someone else's chart at the moment it matters most.
None of that changes the five layers. It changes who is entitled to a copy, how quickly, and what you must be able to prove afterwards about the decision you made.
The layers are the durable part. Regulations change on a schedule nobody controls, and each change arrives as a new obligation layered onto the same five problems. An organisation that has separated them absorbs the next rule as a modification to one layer; one that has not treats every deadline as a fresh programme.
Proof
- ONC (g)(10) self-test PASS, 317 of 317 subtests, April 2026 — public suite URL
- 67 public hostnames, 51 serving, one AWS account
- 104 clearinghouse enrolments across 53 jurisdictions
- USCDI v7 proposal — Patient Digital Identity Credential — listed at isp.healthit.gov
- SAM.gov UEI MVKQW9DKKGB8 · CAGE 20KM5 · NPI 1194645564
- AWS Business Associate Addendum active since 18 August 2026
Engagement
Scoped in writing before work starts; fixed fee where the deliverable is defined. First call is 20 minutes and free. 702-559-3489 · charles@trinh.io
Talk to Charles
- Phone: 702-559-3489
- Email: charles@trinh.io
- Based in Las Vegas, Nevada · working remotely across the United States
How an engagement runs
1. Trace one real case
Follow a single patient or claim end to end with a timestamp at every hop, before any document is written.
2. Written findings
A findings list naming the defect, the system, the owner and the date.
3. Monthly review
Review the list monthly and check whether the dates held.
Frequently asked
What is this service?
This page describes charles trinh — healthcare interoperability — the scope, the failure modes it prevents and how the work is measured. It sits in the Interoperability category at trinh.io.
Where is the work delivered?
Engagements run from Las Vegas, Nevada across the valley — Las Vegas, Henderson, North Las Vegas, Spring Valley, Paradise, Winchester, Sunrise Manor, Enterprise, Summerlin, Clark County — and remotely anywhere in the United States. On-site work in the Las Vegas valley, remote delivery elsewhere.
What does it cost?
Standard rate $195 per hour, committed $165 per hour at twenty hours or more per week for three months or more; $1,600 standard day, $1,350 committed day. Scoped in writing before work starts, fixed fee where the deliverable is defined.
Who is Charles Trinh?
A technology architect in Las Vegas, Nevada, working in healthcare identity, consent and interoperability. Not a physician, not a medical practice, and not affiliated with any other Charles Trinh.
How does an engagement start?
With one trace of a real case and a written findings list naming the defect, the system, the owner and the date. The first call is twenty minutes and free: 702-559-3489.
Record
Profile
Healthcare technology architect working where federal regulation becomes production infrastructure. Built and operates a FHIR R4 clinical platform — SMART App Launch authorization server, patient and provider portals, SQLite — on a hardened single-host AWS estate of 67 public hostnames. The platform returned a self-test PASS on the ONC (g)(10) Standardized API test suite in April 2026, 317 of 317 subtests, publicly verifiable. Designed a consent architecture in which a permissioned ledger records patient grant, refusal and revocation as immutable events while all clinical data stays off-chain and encrypted per patient — no PHI on the ledger by design. Filed a proposed USCDI v7 data element with ONC on verifiable patient identity. Translates the Federal Register, HHS/CMS guidance and HITRUST requirements into architecture, controls and audit evidence.
Selected engineering scale
- Self-managed estate of 67 public hostnames, 51 serving, on one AWS account.
- FHIR R4 API with a SMART App Launch authorization server; patient and provider portals; an AI service layer; Hyperledger Fabric peer, orderer and CA; IPFS; a 337-table SQLite clinical store.
- Two permissioned distributed ledgers in production, one operating a consent registry that records grant, refusal and revocation with no PHI on chain.
- X12 005010 pipelines across eight transaction sets.
- Designed, built, deployed and operated end to end by one person.
Credentials and standards work
- Inferno (g)(10) self-test PASS, 317 of 317 subtests, April 2026 — the ONC Standardized API test suite. Standards validated: US Core 6.1.0, USCDI v3, SMART App Launch 2.0.0, Bulk Data 2.0.0. Public suite URL.
- USCDI v7 public comment submission, April 2026 — proposed Patient Digital Identity Credential under the Provenance data class, listed publicly on the ONC Interoperability Standards Platform.
- SAM.gov UEI MVKQW9DKKGB8 · CAGE 20KM5 · NPI 1194645564. AWS Business Associate Addendum active 18 August 2026.
Technical environment
AWS us-east-1 (compute, storage, messaging, SMS, email, voice) · Linux and systemd hardening · nginx · Fabric and EVM ledgers · FHIR R4 · X12 005010 · HL7 · SQLite · self-hosted mail, drive, search and monitoring · Git-based change control and CI release gates.