This page is part of the International Patient Access (v1.0.0: STU 1) based on FHIR R4. This is the current published version in its permanent home (it will always be available at this URL). For a full list of available versions, see the Directory of published versions

Synchronizing Patient Records

Many applications that access patient records keep a synchronized copy of the patient record elsewhere.

To make this a safe process, application developers should code defensively. Notably, there are a few surprising reasons why a FHIR server may no longer return previously accessible resources:

• Not all FHIR servers preserve FHIR resource ids – all systems SHOULD, for all sorts of reasons, but not all systems do. Consequently, applications cannot maintain a copy of the downstream information.
• Similarly, some systems may permit the hard deletion of a resource (a record) from the origin server, making it unavailable from the API.
• The record was marked as “confidential” on the server side, and the policy is not to provide access to confidential information across the API.
• The record was created for the wrong subject and subsequently corrected. Therefore resource is no longer available for the original patient.
• The API service portal may have lost access to the underlying record store.
• The record may no longer be available to the production system if enough years (e.g., 5-7 years or longer) have passed.
• Patient records may be merged or unmerged. Different systems handle this differently.

Applications should note that there are challenges around synchronizing records - see a Hard FHIR Safety Problem: Synchronization for additional considerations and analysis on the above and other synchronization challenges.