At 7:10 a.m., the practice can reach its phones but not the EHR. The backup dashboard is green. Nobody knows who can request a restore, whether multifactor authentication depends on the unavailable network, or where today’s medication messages should wait.
The missing item is not another backup. It is an executable plan.
Takeaways
- Name the data, applications, dependencies, and owners inside the recovery boundary.
- Test restoration, not just backup completion.
- Define a safe emergency mode for the clinical day.
- Record gaps and retest after material change.
This outline supports planning. It does not establish HIPAA compliance, clinical downtime safety, or a recovery guarantee.
Draw a recovery boundary
Start with the electronic protected health information the practice creates, receives, maintains, or transmits. Then add the services needed to reach it.
RECOVERY BOUNDARY
Application or data set: __________________________________
Business and clinical owner: ______________________________
Hosting and backup owner: _________________________________
Identity, network, device, and power dependencies: ________
Last verified restore: ____________________________________
Downtime method and safe limit: ___________________________
Recovery priority and reason: _____________________________
Vendor escalation route: _________________________________
A backup that needs an unavailable identity provider is not yet a usable recovery path. A restored database without the attachments or audit history the practice relies on may be incomplete. Ask what is backed up, what is excluded, how versions are selected, and who can authorize each step.
Build the plan around three required jobs
HHS’s administrative-safeguards guidance describes the contingency-plan standard and its specifications. Translate the three required jobs into operations.
| Job | Practice decision | Evidence after a test |
|---|---|---|
| data backup | what is copied, where, by whom, and how failure is noticed | backup event, exception, scope, retained version |
| disaster recovery | how lost data and systems are restored | request, authorization, restore result, reconciliation |
| emergency mode | how critical work continues while protections and systems are disrupted | approved downtime steps, role roster, controlled records |
Testing and revision procedures and application-and-data criticality analysis are addressable specifications. “Addressable” does not mean optional by default. The practice evaluates what is reasonable and appropriate, implements an appropriate measure, or documents why an alternative meets the standard.
Make recovery order a clinical decision
A vendor may rank systems by infrastructure dependency. A practice also needs to rank patient-facing work. List the first few hours of a disrupted day: urgent messages, medication questions, results needing review, scheduled visits, referrals, billing, and routine filing. Assign a safe temporary route and a maximum tolerable delay through the practice’s clinical-governance process.
Do not invent universal recovery-time promises. Record the practice’s target, the basis for it, the vendor commitment if any, and the test result as separate fields. When those numbers disagree, the gap belongs in the risk register.
The EHR downtime plan provides a clinical-day worksheet. The HIPAA risk-assessment template helps connect recovery gaps to risk treatment.
Run a restore with invented data
Choose a test record that contains no patient information. Create it, confirm that the backup scope should include it, and have an authorized person request a restore into an approved test location. Record:
- who noticed and approved the request;
- the backup version selected;
- identity and network dependencies encountered;
- whether the restored content was complete and readable;
- how the test copy was secured and removed;
- every exception, owner, due date, and retest.
Then run a tabletop with the application unavailable. Staff should locate controlled downtime forms, reach the role roster, handle one urgent message, record one visit, and reconcile the invented records after service returns. Do not send a real patient message or alter a live chart during the exercise.
HHS’s ransomware fact sheet places contingency planning within the response to cyber disruption. NIST SP 800-34 Rev. 1 provides a broader seven-step planning method for federal information systems. It is useful planning reference material, not a statement that every NIST step is a HIPAA requirement for a small practice.
Keep one revision trail
TEST AND REVISION LOG
Scenario and date: ________________________________________
People and systems participating: _________________________
Restore or emergency-mode result: _________________________
Unmet dependency or unsafe ambiguity: _____________________
Interim control: __________________________________________
Permanent owner and due date: _____________________________
Retest evidence: __________________________________________
Plan version approved: ____________________________________
Patient Square is one part of a practice’s service map, not its contingency plan owner. Before adoption, ask which data can be exported, what support and recovery terms are documented, and how the selected deployment behaves when connected services are unavailable. Check the current US security page and agreement, then test the practice-owned workflow. A green vendor dashboard is a signal; witnessed recovery is evidence.