Skip to main content

Multi-Clinic Record Access: Test Isolation and Authorised Sharing

To evaluate multi-clinic record access, first define which organisations, locations, roles and care relationships may share information. Then test both permitted access and access that should be refused, using synthetic records across search, encounters, documents and exports. Hiding a clinic in a menu or using hard-to-guess record numbers is not sufficient evidence of separation. Shared records can be appropriate when explicitly authorised; accidental sharing and accidental denial both need investigation.

By Bractis · Updated · 8 min read

Owners of clinic groups, independent practices sharing clinicians, and administrators comparing multi-location EMR access models.

Write the sharing policy before testing the software

“Keep our clinics separate” can mean several things. One owner may want central registration across branches but separate financial reporting. Two independent practices may share a visiting doctor while keeping staff access distinct. A specialist may need an authorised referral summary without access to every record at the referring clinic.

List the organisations and locations, who controls each record, which people work across them and what information each role needs. Identify deliberate sharing routes, such as a reviewed referral or an agreed group-wide chart. Record the clinical purpose and the applicable consent and organisational process without treating a software toggle as a complete legal basis.

Separate patient identity from record access. Linking two registrations to the same person can help identity management, but it does not automatically answer whether a receptionist at one clinic may see the other clinic's notes. Conversely, separation should not prevent a properly authorised clinician from receiving information needed through the agreed care pathway.

OWASP's authorization guidance recommends least privilege, default refusal and permission checks on each request. These are technical evaluation principles. They do not decide the clinic's lawful sharing arrangements or prove that a particular product applies the intended policy.

Source: OWASP: Authorization Cheat Sheet

Follow the information beyond the patient list

Start with the screens staff actually use, then ask the vendor to demonstrate the less visible routes through which information travels. Search suggestions, document previews, downloaded files, reports, notifications and exports may be implemented differently from the main encounter screen.

A role that cannot open another clinic's chart should not discover restricted diagnoses in a search snippet or receive its report attachment through an unrelated notification. Equally, a clinician expressly authorised to work across branches should be able to complete the agreed task without staff sharing another person's password.

OWASP's multi-tenant guidance discusses isolation across storage, caches and background processing as well as requests. For a clinic buyer, this supports asking for consistent boundaries across the whole workflow. It is not a requirement that every branch use a physically separate database.

Ask how expired links, removed staff and changed roles are handled. A document downloaded earlier may continue to exist outside the application; revoking access does not recall that copy. The operating plan therefore also needs appropriate devices, handling practices and staff offboarding.

On a narrow screen, scroll the table horizontally to read all columns.

Positive and negative access cases to agree with the vendor
Synthetic caseExpected result to defineWhat to inspect
Clinic A receptionist opens an A registrationPermit only the fields and actions assigned to that roleSearch result, registration and any clinical detail exposed
The same user follows a Clinic B record linkRefuse if no B access or sharing authority existsResponse, preview, browser history and absence of restricted details
A doctor authorised at A and B changes locationPermit the agreed work with the destination roleRole differences, chart context and document association
A referral summary is deliberately sharedDeliver only the approved material through the agreed routePatient match, recipient, scope and record of the transfer
Staff membership is removedPrevent subsequent unauthorised actions under the defined session policyOpen tab, refresh, new request and downloadable links
A manager exports an authorised reportInclude the approved population and fields onlyRows, headers, totals and files produced by the export

Use a small, explicit acceptance matrix

Create an inventory of synthetic users, organisations and records that cannot be confused with live patients. Use a vendor-approved test environment and agreed procedures. Do not probe unrelated organisations or try to bypass a real patient's access controls to gather evidence.

For an original planning example, take two organisations, three representative roles and four information surfaces. Testing each combination produces 2 × 3 × 4 = 24 planned cases. This is only a starting matrix: it does not cover every endpoint, permission, care relationship or software version.

Suppose 20 cases pass, two fail and two are not run. The results reconcile to 24, but there are only 20 demonstrated passes. The two unrun cases cannot be counted as secure, and the two failures need an explanation and retest. A percentage should not obscure a serious disclosure in one failed case.

Write expected results before performing a case. Include both a legitimate task that should succeed and a nearby task that should be refused. If every attempt is denied, the system may be unusable rather than correctly configured. If every attempt succeeds for an owner account, the trial has not tested reception or restricted clinical roles.

Preserve enough evidence to reproduce the result: software environment, test role, synthetic record reference, action, expected result, observed result and date. Avoid copying sensitive production configuration or credentials into a general review document.

Investigate the failure without turning staff into security testers

When restricted information appears, stop the affected test and preserve the synthetic evidence. Ask the vendor to investigate the boundary and identify the scope of any related issue. A sales explanation such as “users normally do not click there” does not answer why the request was permitted.

Do not treat a random-looking URL, hidden button or unpredictable file number as an access decision. These can reduce accidental discovery but do not establish that the requester is entitled to the information. The technical team should explain how the relevant authorisation is enforced.

Distinguish a wrong record association from an access-control failure. A report attached to the wrong synthetic patient may appear to an otherwise correctly authorised clinician. The permissions may work while identity matching fails. The review should route the problem to the correct investigation and retest both association and access where necessary.

The opposite failure matters too. If legitimate shared-care access is blocked, use the approved alternative route and investigate the configuration. Staff should not solve it by reusing an owner password, exporting the entire database or maintaining an uncontrolled second copy of every record.

Recheck when the organisation or role changes

A clinic group changes over time: a branch opens, an associate leaves, ownership changes or a new reporting integration is added. Each change can alter the sharing model. Keep the policy and test cases close to the operational decision so the software configuration does not lag behind it.

For a new location, verify which existing staff inherit access and which do not. For a visiting clinician's departure, settle outstanding clinical responsibilities while removing permissions no longer justified. Preserve authorship and the appropriate record history; deactivating a person should not rewrite who documented an encounter.

Ask what evidence the vendor can provide about changes, review practices and independent assessment, while keeping its scope explicit. A certificate or a passed demonstration can inform a decision, but neither guarantees that every clinic-specific permission is correct.

Review a sample of real operating decisions through authorised internal processes once the arrangement is in use: who approved a new role, why a report was shared and how an access problem was resolved. This guide's synthetic matrix is an acceptance aid, not a penetration test, legal opinion or security certification.

Twenty successful cases do not erase two failures

The illustrative 24-case matrix has 20 passes, two failures and two unrun cases. Keep all four unresolved results visible. Prioritise the consequence of each failure and retest after correction; do not describe the whole multi-clinic system as proven secure from this limited sample.

Define and test clinic record boundaries

Use these prompts in your clinic’s own notes. Record the result, responsible person and next action for each item.

Sharing model
Define independent organisations, branches, role purposes and deliberate sharing routes.
Information surfaces
List search, encounter, preview, download, notification and export paths relevant to the clinic.
Acceptance evidence
Create synthetic positive and negative cases with expected results and reproducible evidence.
Change ownership
Assign responsibility for new branches, changed roles, departures and unresolved access findings.

Questions doctors ask

Does every clinic branch need a separate database?

Not necessarily. Different technical designs can enforce appropriate boundaries. Define the authorised sharing model and test its implementation instead of treating database count as proof.

Is hiding another clinic's records from the menu enough?

No. Evaluate the permitted behaviour of direct record access, search, previews, downloads and exports through an agreed synthetic trial.

Can one doctor have different permissions at different clinics?

Yes, the operating arrangement may require that. Demonstrate the destination role and information scope rather than assuming one universal permission set.

Does passing this acceptance matrix certify a product's security?

No. It provides evidence for the specified users, records and surfaces in that environment. Broader security assurance and legal assessment require separate work.

Sources and scope

Updated on . A buyer's operational acceptance framework. It is not permission to test unrelated systems, a legal determination about sharing or a security certification. No independent clinical or legal review is claimed.

Try the workflow
with Bractis.

Agree a synthetic two-organisation trial with reception, clinician and manager roles. Test an allowed task and a refused task across search, records and files, then inspect the evidence and any unresolved findings.

Request a clinic demo