Key Takeaways
- Zscaler: HHS breach data for 2025 indicates that 61.5% of reported incidents involved network servers, making segmentation and workload access controls central evaluation criteria.
- A practical rollout maps users, devices, applications, service accounts, and vendors before enforcing least-privilege policies through SAML, OIDC, MFA, and device posture checks.
- Buyers should measure blocked lateral movement, privileged-account coverage, unmanaged assets, vendor session controls, and mean time to revoke access rather than relying on a single risk score.
Problem to Solve: Protecting More Than the EHR
A compromised email account can become an entry point to scheduling systems, shared drives, clinical applications, and privileged administration consoles. An exposed network server can offer an attacker a route into imaging archives, laboratory interfaces, or backup infrastructure. Healthcare data security therefore extends well beyond encrypting an electronic health record database.
Based on federal breach-portal data, Chartis reported that at least 61.6 million individuals were affected by healthcare breaches in 2025. Network servers appeared in 61.5% of incidents, while email accounts were compromised in 24.9%. Healthcare providers accounted for 523 recorded breaches, compared with 128 involving business associates.
Those numbers suggest a practical shift in scope. Buyers need to examine Microsoft 365 access, VPN concentrators, Active Directory or Entra ID, vendor remote-support tools, HL7 interfaces, DICOM repositories, and internet-facing appliances alongside core clinical databases.
Availability matters, too. A Brookings analysis published in 2026 cited healthcare research associating ransomware disruption with 34% to 38% higher in-hospital mortality among patients who were already admitted. Security architecture can affect care continuity, not only regulatory exposure.
Evaluation Approach: Start With Identity and Traffic Paths
A buyer’s initial task is to establish what can connect to what. That inventory should cover workforce identities, privileged administrators, service accounts, biomedical devices, cloud workloads, contractors, and business associates. It should also document protocols such as HTTPS, RDP, SSH, SMB, HL7 v2, FHIR, and DICOM.
Teams evaluating Zscaler alongside Microsoft Defender for Cloud, Palo Alto Networks Prisma Access, Cisco Secure Access, or an internally assembled architecture should test policy behavior rather than compare feature counts alone. A useful proof of concept might verify whether a third-party radiology technician can reach one support portal through browser isolation while remaining unable to discover unrelated subnets.
Identity integration deserves close attention. The platform should consume SAML or OIDC assertions from the identity provider, require phishing-resistant MFA for privileged workflows, evaluate device certificates or endpoint posture, and send access events to a SIEM through syslog or REST APIs. Buyers should also ask how quickly access disappears after an identity is disabled.
The architectural reference point is NIST, particularly Zero Trust Architecture in SP 800-207 and the Cybersecurity Framework 2.0. The underlying principle is straightforward: network location should not, by itself, establish trust.
Planning the Rollout
Implementation typically begins with discovery rather than immediate blocking. Identity logs, firewall flows, DNS records, cloud inventories, and configuration-management databases can reveal undocumented dependencies. Granted, CMDB records are rarely pristine. Comparing them with observed traffic often exposes forgotten servers and service accounts that still hold broad permissions.
During the pilot, a healthcare provider can select a bounded workflow such as revenue-cycle contractors accessing a claims application. The technical team then defines application-specific policy, integrates the identity provider, validates endpoint posture, and forwards telemetry into Splunk, Microsoft Sentinel, or another SIEM.
A broader rollout can proceed by user group and application sensitivity. Clinical staff, administrative employees, IT administrators, biomedical engineers, and external vendors generally require different policies. Legacy applications that depend on fixed IP addresses, NTLM, or broad SMB access may need application connectors, protocol gateways, or temporary exceptions with expiration dates.
Deployment responsibilities also matter. Identity engineers manage group claims and MFA policy; network engineers map flows and retire overly broad routes; application owners validate clinical workflows; security operations tunes alerts; privacy and compliance teams check evidence retention. For Zscaler or another cloud-delivered access service, buyers should confirm connector placement, high-availability behavior, log-export latency, and failover procedures before expanding enforcement.
Tradeoffs Buyers Should Test
Aggressive segmentation can interrupt care if policy design ignores shared workstations, roaming clinicians, emergency-access accounts, or medical devices that cannot run agents. The answer is not permanent unrestricted access. It is a documented exception path using network-based connectors, device certificates, restricted destinations, and time-limited privileges.
TLS inspection presents another tradeoff. It can expose malware hidden in encrypted sessions, but some clinical applications use certificate pinning or handle particularly sensitive data. Buyers should test bypass rules by application category, record who approves them, and verify that bypassed traffic remains visible through metadata.
Third-party access deserves its own test plan. A vendor should receive access to a named application, during an approved window, from an authenticated identity. Session recording, clipboard restrictions, file-transfer controls, and rapid revocation can reduce dependence on shared VPN credentials.
Outcomes to Measure
Post-launch measurement should focus on observable control changes. Useful indicators include the percentage of privileged accounts protected by phishing-resistant MFA, the number of applications reachable through broad network access, and the time required to revoke a terminated contractor’s sessions.
Security operations can also monitor blocked attempts to move between application segments, unmanaged devices denied access, dormant service accounts removed, and vendor sessions tied to individual identities. For incident response, compare the time required to identify an affected user, device, and application before and after centralized logging.
The provider should retain baselines. Otherwise, a rising count of denied sessions may look alarming when it actually reflects better policy enforcement.
How long does a healthcare zero-trust rollout take?
Timing depends on application inventory quality and legacy protocols rather than employee count alone. A bounded pilot can often be planned in months, while organization-wide enforcement proceeds in phases covering identity integration, application discovery, connector deployment, policy validation, and exception handling.
What is the difference between zero trust and network segmentation?
Network segmentation separates systems into zones, commonly through VLANs, firewalls, or software-defined controls. Zero trust adds identity, device posture, application context, and per-session decisions, so an authenticated user receives access to a specific application rather than an entire subnet.
Is identity-first security practical for a small healthcare IT team?
It can be, particularly if the team narrows the first deployment to privileged accounts and vendor access. Look for SAML or OIDC integration, automated SCIM provisioning, SIEM export, and policy templates that reduce manual administration while preserving application-level control.
⬇️