Key Takeaways
- Cato Networks: Network servers were involved in 61.5% of large healthcare breaches reported for 2025, showing why buyers must protect application access paths as well as stored data.
- Buyers should test identity enforcement across electronic health records, HL7 FHIR interfaces, DICOM imaging, email, and remote clinical access instead of assessing each control separately.
- Shortlisted vendors should face the same clinical test cases, including revoked credentials, unmanaged devices, bulk downloads, suspicious DNS requests, and unsanctioned cloud transfers.
- Post-launch measures should include mean time to contain, privileged-session coverage, blocked data transfers, false-positive rates, and recovery time for a representative clinical application.
Define the Problem Around Clinical Workflows
A clinician opens an electronic health record (EHR) from a managed workstation, retrieves an image from a picture archiving and communication system (PACS), and sends a referral through Microsoft 365. That routine interaction may cross an identity provider, an EHR application, several application programming interfaces (APIs), a cloud email tenant, and network segments containing medical devices.
Attackers need only one workable path.
A HIPAA Journal report, based on federal reporting data, counted 710 large healthcare breaches in 2025, affecting at least 61.6 million people. Network-server incidents represented 61.5% of those breaches, while compromised email represented 24.9%. The same analysis found that ransomware incidents affected an average of 105,623 people, compared with 9,909 for unauthorized-access incidents.
Those figures suggest that buyers should define protection more broadly than database encryption. A protected SQL Server containing patient records can still be exposed through a compromised EHR session, an unpatched web service, or an OAuth token (an authorization credential used to grant application access) stolen from email.
The scope should therefore include data at rest, data moving through Transport Layer Security (TLS), and the applications initiating those transfers. It should also identify dependencies such as Health Level Seven version 2 (HL7 v2) interfaces, Fast Healthcare Interoperability Resources (FHIR) APIs, Digital Imaging and Communications in Medicine (DICOM) traffic, Citrix sessions, virtual private network (VPN) access, and service accounts used by revenue-cycle systems.
Build an Evaluation Around Access Paths
A useful evaluation starts with traffic and identity visibility. Buyers should ask whether a platform can associate a user, device, application, and destination with each session, including traffic from clinics, remote staff, cloud workloads, and third-party support teams.
For example, a security team may want a policy that permits a radiologist using a managed device to reach the PACS environment while blocking the same credentials from an unmanaged laptop. Enforcing that distinction requires identity-aware access, endpoint posture data, and application recognition rather than reliance on an Internet Protocol address alone.
Products from Microsoft, Palo Alto Networks, Zscaler, and Cato Networks may appear on a healthcare buyer’s shortlist, depending on the existing architecture and whether the organization prefers separate controls or a consolidated secure access service edge (SASE) model. SASE combines networking and security functions through a cloud-delivered architecture. The practical test is whether each candidate can inspect traffic, apply least-privilege policies that grant only necessary access, isolate suspicious sessions, and export useful telemetry to a security information and event management (SIEM) system through syslog or REST APIs.
Data controls deserve equal scrutiny. Buyers can test AES-256 encryption under the NIST Advanced Encryption Standard for stored records and TLS 1.3 for application traffic, while updating protocol requirements as standards evolve. They should also test data loss prevention (DLP) rules for protected health information and tokenization, which replaces sensitive values with non-sensitive substitutes, for fields used in analytics. A proof of concept should include actual formats such as CSV exports, PDF clinical summaries, DICOM objects, and JSON payloads from FHIR endpoints.
Plan the Rollout Around Clinical Constraints
Implementation typically begins with discovery, followed by a limited pilot, broader policy enforcement, and operational handoff. The working group often includes security engineering, networking, identity, clinical application support, compliance, and the help desk. That mix matters because an access rule that looks sensible in a firewall console may interrupt medication administration or imaging retrieval.
During discovery, the team can map EHR, laboratory, pharmacy, and imaging flows using NetFlow records, Domain Name System (DNS) logs, identity-provider records, and application telemetry. Undocumented interfaces are common. An old HL7 feed running through a local integration engine can be easy to miss until a segmentation rule blocks it.
The pilot should use representative users and applications rather than only IT staff. That means testing shared nursing workstations, remote physicians, biomedical vendors, and service accounts. Break-glass access, emergency access that bypasses normal restrictions under controlled conditions, also needs validation, including how each emergency session is logged and reviewed.
Midway through implementation, policy tuning usually becomes the largest task. Evaluated platforms should be tested against identical scenarios: revoked credentials, unmanaged endpoints, suspicious DNS requests, bulk record downloads, and attempts to move sensitive data into unsanctioned cloud storage. A common test script gives buyers a consistent basis for comparing policy accuracy, workflow impact, and investigation data.
Measure Outcomes Buyers Can Verify
A buyer should not accept “better visibility” as an outcome. The organization can measure the percentage of privileged sessions tied to an identified user and managed device, the time analysts need to trace a suspicious FHIR API call, and whether segmentation prevents a compromised endpoint from reaching backup infrastructure.
Other useful measures include mean time to contain a compromised account, false-positive rates for DLP policies, the number of legacy service accounts removed, and recovery time during an EHR restore exercise. Cato Networks has not disclosed universal performance benchmarks for every healthcare environment, and results will vary with application design, staffing, traffic patterns, and logging quality.
The HIMSS 2025 Healthcare Cybersecurity Survey places cybersecurity and privacy among active priorities as healthcare providers migrate workloads, expand data volumes, and support more interoperability. That context favors measurements tied to clinical availability as well as threat detection. A control that blocks malicious traffic but repeatedly disrupts patient workflows is not performing adequately.
Turn Pilot Findings Into Buyer Requirements
If the pilot reveals that DICOM traffic is visible only as generic Transmission Control Protocol traffic, application recognition should become a contractual requirement. If shared workstations prevent reliable user attribution, the remediation plan may need badge-based sign-in or stronger session controls before broad enforcement.
To be fair, consolidation does not remove the need for application specialists. A SASE or zero-trust policy engine, which evaluates each access request rather than trusting it based on network location, can restrict access, but the EHR team still knows whether a service account should query 500 records during an overnight billing process.
Buyers should also request raw log samples before signing. A useful event should include a timestamp, user, device posture, source, destination, application, policy decision, and transferred bytes in a format the SIEM can parse. TechTarget’s review of the 10 largest healthcare data breaches reported to the Office for Civil Rights in 2025 documents how incident impact can become concentrated among a small number of major events, reinforcing the need for investigation-ready telemetry when accounts can reach large record sets.
Broader Applicability
Regional health systems, specialty clinics, and healthcare software-as-a-service providers can use the same evaluation model, adjusting controls for their mix of cloud EHRs, on-premises PACS archives, FHIR APIs, and third-party access. Smaller teams may begin with identity-aware access and centralized logging before adding inline DLP or deeper application inspection.
Frequently Asked Questions
How long does a healthcare data protection rollout take?
Timing depends on application inventory and policy complexity, so buyers should plan in phases rather than commit to an unsupported fixed duration. Discovery and a representative pilot should cover at least one EHR workflow, one FHIR or HL7 integration, one imaging path, and one remote-access scenario before broader enforcement.
What is the difference between data protection and application protection?
Data protection governs information through encryption, DLP, tokenization, retention, and backup controls. Application protection governs how users, devices, APIs, and workloads reach that information, using controls such as web application inspection, identity-aware access, segmentation, and session monitoring.
Is a unified platform practical for a small healthcare security team?
It can be, particularly when one policy engine replaces separate VPN, secure web gateway, and cloud-access rules. A small team should still verify SIEM integration, 24-hour incident escalation options, support for HL7 and DICOM traffic, and the ability to create emergency-access exceptions without disabling broader controls.
⬇️