Costa Rica has already defined how health information gets exchanged: there are consumers, producers and custodians, and each role brings different obligations. The question you will be asked is not whether you want to connect, but which role you carry and what you are missing to meet it.
Once a country defines its interoperability architecture, connecting stops being optional. Your organization will have to answer queries, publish what it produces, or both — and prove it can do so before being let in. Most find out late: three or four capabilities were missing that never appeared in any project, and they surface during testing, when the date is already committed. Our job is for you to know that today.
Exchange is organized by role, not by type of institution. A single hospital usually carries all three at once, and the obligations are not a choice: they stack.
Needs to see information someone else generated — a result, a vaccination, an earlier episode — at the moment of caring for the patient.
Prove who is querying, on whose behalf and for what purpose; and that every query is on the record.
Nobody defined the purpose of use or where consent is recorded. The query works in testing and stalls in legal review.
Produces clinical information the rest of the ecosystem needs: laboratory results, notifications, imaging reports, vaccinations administered.
Publish conforming to the national profile, with terminology bound, at the moment the event happens — not in a batch at the end of the month.
The system generating the data cannot produce the profile, and the project finds out once that system has already been bought.
Holds the information and answers for it: who may see it, how long it is kept and what is handed over when the ecosystem asks. Sometimes it is the same party that produces it; sometimes it is a role of its own.
Be available when queried, hand over only what is authorized, and be able to reconstruct who accessed what, months later.
The repository exists, but there is no committed availability, no queryable audit trail and no written retention policy.
If your organization produces and also queries, you do not get to pick: both columns apply. And if it holds what it produces, all three do. That stacking is why connection projects run to twice the estimate.
Seven capabilities. Open the ones your role binds you to and you will recognize which you are ready for and which are not in any project your organization is running today.
Knowing that the record you hold and the one the ecosystem returns belong to the same person, even when the identifiers do not match.
It merges or splits the wrong histories. It is the error that escalates fastest to a clinical incident and costs the most to unwind.
You query the ecosystem's index instead of replicating it, with a written policy for the cases that do not resolve automatically.
Finding out what information exists about a patient and who holds it, before retrieving anything.
Without it you can only ask whoever you already knew had something — the exact opposite of why an HIE exists.
You query the ecosystem's document registry and retrieve only what the clinical decision at hand actually needs.
Declaring why you are querying and verifying there are grounds to do so, on every access rather than once at the beginning.
This is where projects that already worked technically come to a halt. Nobody signs off a production release with it open.
Purpose of use is defined per case, the legal basis is recorded, and it is bound to the authorization the request travels with — not to a checkbox on a form.
Handing over what you produce with the structure, mandatory elements and terminology the national guide requires.
What you publish gets rejected. Or worse: it is accepted incomplete and reaches whoever queries it that way, unnoticed until it matters.
It is validated against the profile before publishing and corrected at source. If the source system cannot produce it, an exposure layer solves it without replacing the system.
Being available to answer when the ecosystem asks, within a committed response time.
A custodian that does not answer stops being one in practice, and whoever was asking makes the clinical decision without that data.
Availability is committed as a measurable number, with monitoring, and the external query path is separated from the system that runs daily operations.
Being able to reconstruct, months later, who accessed what information, when and on what grounds.
It is the first thing you will be asked for when a complaint arrives. If the logs have rotated, the institutional answer is that nobody knows.
It is designed as a component, with defined retention and its own query path, and the event is emitted on every ecosystem transaction — not only in the local system.
The ecosystem knowing that the one asking is you, every log sharing the same time, and what was exchanged being provable.
Without clock synchronization the audit trail reconstructs nothing. Without signing, an exchanged document proves neither who issued it nor that it arrived intact.
Certificates per organization and per subsystem, clock synchronization, and signing and sealing on the exchanges that carry evidentiary weight.
That board is exactly the first deliverable of a gap assessment: these rows, filled in with what your organization already has, what is missing and what each gap costs.
No integration is a single call. This is the real sequence, and at every step there is something concrete your architecture has to be able to do.
Before asking for anything you have to settle who is being talked about. You query the ecosystem's index with the identifiers you hold and get back the one everyone else knows them by.
With identity resolved, you ask what information exists about that person: who holds it, what kind it is and how recent. Nothing is retrieved yet.
You pull the document or resource the decision at hand requires, not the whole record. Whatever you pull beyond that is now outside your control.
The same circuit from the other side: someone identifies the patient, sees that you hold information and asks for it. Your architecture has to answer in time, authorize per use case and trim the response.
Every time a relevant clinical event occurs it is published conforming to the profile, with terminology bound, at the moment it happens. The monthly batch no longer works.
Each of the transactions above emits an audit event, at both ends and on the same clock. That is what later lets you answer a complaint with facts.
Entering a national ecosystem is not a technical agreement between two teams: there are requirements proven with evidence. Worth knowing before you commit to a date, not after.
That what you publish meets the implementation guide in force, with mandatory elements and terminology bound.
Validation report against the profile, with test cases covering real scenarios and not just the happy path.
That the codes you send exist, are current and belong to the ValueSet the guide requires.
Automated terminology validation at the publication point, and a written plan for what happens when a ValueSet changes version.
Organization and subsystem certificates, and authorization per use case rather than general access to the system.
Registration in the ecosystem, proof of mutual authentication and a matrix of what each consumer may ask for.
A record of every access, with defined retention and the ability to answer a specific query.
A live demonstration: who viewed this person's record in the last six months, answered in minutes.
That what was exchanged can be attributed to whoever issued it and proven unaltered, in the cases that require it.
Signing and sealing of documents or resources, validated on the receiving end. Meddyg provided technical assistance to the Banco Central de Costa Rica to enable JAdES signing of FHIR resources.
Passing the ecosystem's test cycle with data resembling the real thing, before touching production.
A test environment with synthetic data, an agreed set of cases and evidence of every transaction executed end to end.
Six stages. The first settles your role, because everything else — scope, cost and date — follows from it.
We establish which roles your organization carries, service by service, and which concrete use cases it enters the ecosystem with. Without this, any estimate is a guess.
We test the capabilities your role requires against what your architecture does today, and separate what has to be built from what only has to be configured. This is where the date becomes realistic.
We define which component talks to the ecosystem, which transactions it supports, how it authenticates and what gets trimmed from each response.
What is missing gets built, and it is validated against the national profile at source rather than at the end. Where the source system cannot produce the profile, an exposure layer solves it instead of a replacement.
We run the test cycle end to end, including the scenarios that fail: the patient that does not resolve, the document that does not conform, the other end that does not answer.
We stand by the first real transaction and leave the monitoring, the auditing and the procedure for when the one failing is the other end.
Material you estimate, procure and demonstrate compliance with — not a slide deck.
The capabilities your role requires, what you already have, what is missing and what each gap costs.
Which component talks to the ecosystem, which transactions it supports and under what identity, authorization and audit controls.
The technical and regulatory requirements with the evidence that proves them, ordered to be submitted as is.
The connection scenarios, failures included, automated so they can be re-run on every change.
Which use case goes first, with which dependencies, and in what order so something reaches production early.
What gets measured, what raises an alert and what to do when the end that stops answering is the other one.
Each of these projects resolved a specific role inside an exchange ecosystem, and stayed in operation.
Proofs of concept with the National Digital Government Agency to validate FHIR resource transactions over the national platform, within Costa Rica's functional interoperability demonstration: secure exchange between different systems, without centralizing the data.
Design, production deployment and operation of the first FHIR server receiving notifiable disease, vaccination and laboratory result notifications from EDUS/CCSS, and later from other health facilities.
Part of the Meddyg team took part as consultants in designing the first HL7© FHIR© implementation guide for regional AEFI (adverse events following immunization) notification, and in designing and implementing the infrastructure that would receive those notifications.
Implementation of PACS architectures and diagnostic workstations for medical image exchange under the DICOM standard, in public hospitals of the Dominican Republic's National Health Service.
The models and profiles we name on this page — OpenHIE, IHE, HL7 FHIR — are the method we work with, not clients. When a case belongs to a client, we say so by name.
It serves both those who already have a connection date and those who do not yet know which role they were assigned.
Social security funds, ministries and hospitals the national ecosystem has already assigned a role — and often a date.
Clinics, hospitals and networks that need to take part without opening up their network or rebuilding their record system.
Producers of results who will be required to publish conforming to a profile, at the moment the event happens.
Teams that have to answer how long connecting takes and what it costs, and today have nothing to estimate with.
Manufacturers whose product has to meet the national profile for their client to be able to connect.
Initiatives that need to connect dozens of organizations on the same terms, rather than negotiating one by one.
If you recognize three or more, what you are missing is not a project — it is a diagnosis. That first step is measured in weeks, and it is what stops the date from being a bet.
We tell you exactly which capabilities your role requires, which you already have, which are missing and in what order to build them — with the evidence the ecosystem is going to ask you for.