09
Technical Architecture & Hosting
Hosting models
| Hosting model | Where servers operate | Who operates them |
|---|
| Self-hosted | Hospital's own data center | Hospital IT |
| Epic Hosting | Epic-managed infrastructure | Epic |
| Third-party / cloud | Contracted hosting/cloud provider | Organization and provider |
| Community Connect | Another health system's Epic instance | Sponsoring health system |
Epic does not mandate Epic-hosted infrastructure. Epic Hosting is an available managed hosting option.
Core technologies
| Technology | Purpose | Cerner / Oracle Health analogy |
|---|
| Chronicles / InterSystems | Operational patient database | Millennium operational database |
| Hyperdrive | Newer Epic user interface | PowerChart clinician client |
| Hyperspace | Older Epic desktop client, being replaced by Hyperdrive | Legacy PowerChart client |
| Bridges | Traditional interface connectivity | Open Engine / FSI |
| Interconnect | APIs/web services | Oracle Health FHIR / Millennium APIs |
| Clarity | Relational reporting | Millennium reporting data |
| Caboodle | Analytics warehouse | HealtheEDW |
Interfaces, Integration & Data Exchange
Architecture: Epic Bridges is Epic's core HL7v2 messaging infrastructure. Interconnect exposes web services and APIs. A third-party interface engine may sit between Epic and other systems for transformation, routing, observability, or protocol mediation. Not every connection uses every layer. Epic HL7 interface catalog · Epic FHIR catalog
9.1 Integration building blocks — module or not?
| Item | Classification | What it does | Oracle Health / Cerner analogy |
| Bridges | Integration framework / infrastructure (not clinical module) | Configures and runs many HL7 v2 inbound/outbound interfaces; handles message processing and acknowledgments. | Millennium FSI / Open Engine |
| Interconnect | API/web-services layer (not clinical module) | Supports Epic web-service integration, including FHIR-based access. | Millennium API / FHIR services |
| open.epic / App Orchard | Developer ecosystem (not clinical module) | Public interface specifications, app/API registration and technical documentation. | Oracle Health developer ecosystem |
| Care Everywhere | Interoperability feature / capability | Cross-organization clinical record exchange. | CommonWell/Carequality and Oracle Health information exchange |
| External integration engine | Third-party middleware (not Epic module) | Optional message mapping, queues, monitoring, and connections to non-Epic systems. | Rhapsody, Mirth/NextGen Connect, Corepoint, Cloverleaf, etc. |
| Interface workqueues / error-monitoring tools | Operational features (not standalone clinical modules) | Review failed records, message exceptions, and reconciliation. | Millennium interface monitoring / integration engine queues |
9.2 How messages actually flow
External source/targetLab · PACS · HIE · Pharmacy · Payer · Device
⇄Optional enterprise integration engineTransform · route · queue · monitor
⇄Epic Bridges or InterconnectHL7 v2 / FHIR / services
⇄Epic workflows and recordsPrelude · EpicCare · Beaker · Radiant · Resolute
Placement: The middleware may run in a hospital data center, cloud environment, Epic-managed hosting arrangement, or external vendor environment. The exact network path, VPN/private links, interfaces, firewall rules, and tenancy are implementation-specific. (assumed) A typical hospital will keep a central interface team responsible for message lifecycle even when infrastructure is hosted elsewhere.
9.3 Interface catalog — clinical and operational
| Interface / message | Common standard | Direction relative to Epic | Typical source → destination / use |
| ADT: A01 / A02 / A03 / A04 / A08 and others | HL7 v2 ADT | In or out | Prelude/Grand Central ↔ outside registration, census, billing, bed, and ancillary systems; admit, transfer, discharge, updates. |
| Orders | HL7 v2 ORM; OML where applicable | Commonly out | EpicCare/Beaker → external lab, radiology, cardiology, reference laboratory. |
| Results / observations | HL7 v2 ORU | Commonly in | Lab/RIS/device → Epic results and clinician chart; match to patient and order. |
| Clinical documents and transcriptions | HL7 v2 MDM; CDA/C-CDA where applicable | In or out | Transcription, documentation repository, referring providers → chart/HIM; or notes pushed out. |
| Appointments and scheduling | HL7 v2 SIU | In or out | Cadence ↔ scheduling/notification, transport, ancillary systems. |
| Patient identity matching | HL7 v2 / FHIR / EMPI services | In or out | Cross-system MRN, encounter, demographics, and enterprise identity reconciliation. |
| Medication information | HL7 v2 RDE/RAS/RXA as applicable | In or out | Epic medication orders/administration ↔ dispensing, automation, external medication systems. |
| E-prescribing | NCPDP SCRIPT | In or out | Prescriber ↔ pharmacy, network, prescription messages and renewals. |
| Clinical observation devices | HL7 v2 / vendor device gateways | Commonly in | Bedside monitors, infusion pumps, vitals aggregators → Epic documentation. |
| Images and imaging metadata | DICOM; HL7 v2; FHIR ImagingStudy metadata | Both | Modality/PACS/VNA ↔ Epic/Radiant; image pixels are commonly retained in PACS/VNA while reports and links surface in Epic. |
| Care Everywhere / HIE exchange | FHIR, CDA/C-CDA, exchange networks | Both | Other health organizations ↔ Epic clinical records. |
| External clinical app | FHIR R4 + SMART on FHIR / OAuth 2.0 | API read/write as supported | Approved app ↔ Epic clinical and demographic resources; endpoint support varies. |
9.4 Revenue-cycle X12 EDI — claims, remittances, Medicare
| Transaction | Purpose | Typical direction (provider perspective) | Epic / operational destination |
| 837P | Professional/physician electronic claim; CMS-1500 electronic equivalent | Provider → payer/clearinghouse | Resolute Professional Billing; claims engine/clearinghouse |
| 837I | Institutional/hospital electronic claim; UB-04/CMS-1450 electronic equivalent | Provider → payer/clearinghouse | Resolute Hospital Billing; claims engine/clearinghouse |
| 837D | Dental claim | Provider → payer/clearinghouse | Dental billing workflows, if configured |
| 835 | Electronic remittance advice (ERA): paid, denied, adjusted; payment posting data | Payer/clearinghouse → provider | Resolute remit processing and posting |
| 270 / 271 | Insurance eligibility inquiry / response | Provider ↔ payer | Registration / coverage verification |
| 276 / 277 | Claim status inquiry / response | Provider ↔ payer | Billing follow-up and claim status |
| 277CA | Claim acknowledgment / acceptance or rejection | Payer/clearinghouse → provider | Claims status/error workqueues |
| 999 | Implementation-level EDI acknowledgment | Receiver → sender | EDI/interface processing monitoring |
| 278 | Prior authorization / utilization review transaction | Both | Authorization workflows |
| 834 | Benefit enrollment / maintenance, especially payer operations | Employer/sponsor → payer | Tapestry or health plan systems, when applicable |
| 820 | Payment order / remittance for premium payments (payer context) | Depends on payer role | Tapestry premium billing, where used |
Medicare: Medicare Fee-for-Service institutional and professional claims ordinarily use 837I and 837P through Medicare Administrative Contractors (MACs), subject to Medicare billing rules, enrollment, edits and companion guides. CMS-1500 (professional) and CMS-1450/UB-04 (institutional) are the corresponding paper claim formats. An 835 is a remittance transaction, not the claim itself. See Epic X12 interface inventory and CMS EDI information.
9.5 Example end-to-end workflows
| Scenario | Sequence | Key control point |
| External lab test | Epic order → outgoing HL7 order → lab LIS → incoming HL7 results → Epic chart/In Basket routing | Patient/order identifiers, ACK, result reconciliation |
| Hospital registration | Prelude/Grand Central → ADT → downstream department systems → ACKs and census reconciliation | MRN/visit consistency, duplicate and missing messages |
| Radiology imaging | Epic order → Radiant/RIS ↔ modality/PACS → report/result → Epic | Order/accession matching; DICOM objects and final report may travel separately |
| Professional claim | EpicCare Ambulatory visit → coding/charge capture → Resolute PB → 837P → clearinghouse/MAC/payer → 999/277CA → 835 → payment/denial posting | Claims acceptance, rejects, remit balancing and denial follow-up |
| Hospital claim | Inpatient encounter → charges/coding → Resolute HB → 837I → payer/MAC → status and 835 → posting | DRG/coding, charge reconciliation, claim edits |
| External patient-facing application | App authorization → SMART/OAuth → Epic FHIR endpoints → permitted patient/clinical data | Scopes, consent/authorization, access logging |
| Patient message | Patient MyChart message → Epic In Basket → physician or staff pool; staff may initiate Secure Chat | Pool rules, assignment, follow-up; Secure Chat ≠ patient MyChart inbox |
9.6 Interface deployment and validation checklist
| Stage | Activities / deliverables | Accountable team (typical) |
| Inventory and ownership | Define producer/consumer, system of record, message direction, business owner, SLA | Integration lead + application owner |
| Specifications | Version, trigger events, segment fields, code sets, mappings, sample payloads | Epic interface analyst + vendor analyst |
| Connectivity | Endpoint, network routes, TLS, MLLP/SFTP/HTTPS as applicable, credentials, allowlists | Network/security + interface engine team |
| Message handling | ACK/NACK, retries, de-duplication, ordering, queue limits, dead-letter/error handling | Interface engineers |
| Testing | Unit, end-to-end, negative/error cases, high volume, downtime/recovery, integrated testing | Hospital QA + Epic + third party |
| Cutover | Endpoint swap, feed activation, backfill and duplicate controls, command-center support | Cutover lead + interface teams |
| Production operations | 24×7 monitoring where critical, queue thresholds, on-call routing, reconciliation, change control | Hospital integration ops / managed provider |
9.7 Who supports an interface?
| Layer | Usually responsible | What gets escalated |
| Epic workflow configuration | Hospital Epic application analysts | Orders, routing, encounter/charge build |
| Epic Bridges / Interconnect | Hospital Epic interface analysts, supported by Epic TS | Interfaces, build, application-level logs/behavior |
| Third-party interface engine | Hospital integration team or outsourced vendor | Mappings, transforms, queues, endpoint connections |
| External application | Lab, PACS, payer, pharmacy or vendor IT | Source payload, target processing, downstream data quality |
| Network / hosting | Hospital IT, Epic Hosting, or contracted cloud/provider per model | DNS, certificates, firewall, service outages, network paths |
| Revenue-cycle reconciliation | Billing operations, clearinghouse, EDI team | 837 rejections, 835 unposted remits, 277CA status, payment balancing |
9.8 Integration references
Epic official HL7 v2 integration catalog · Epic official FHIR integration catalog · Epic official X12 catalog · CMS: electronic billing and EDI. Exact interface support depends on licensed application, version, configuration and the connected endpoint.
Ambulatory workflow
Cadence scheduling → Prelude / Welcome registration → EpicCare Ambulatory visit, notes and orders → diagnostics / pharmacy → Resolute Professional Billing → MyChart
10
Epic Implementation Methodology, Engagement & Support
Epic follows a Foundation-based implementation approach with standardized content, workflow validation, configuration, testing, readiness and optimization. The exact proprietary internal methodology and templates are not publicly available; this is an evidence-based summary of published Epic implementations.
Big bang is not unique to Epic, and phased implementation is not unique to Cerner. A published Intermountain Cerner Millennium study describes phased regional deployment with big-bang activation inside each region. The defensible conclusion is that Epic leadership publicly promotes simultaneous enterprise activation more strongly—not that either vendor only supports one rollout pattern. Intermountain Cerner implementation study. Epic's own implementation executive said large health systems can launch a first phase within roughly 16–18 months, with later launches about six months apart. April 2024 report.
10.0.3 Important distinction: Epic versus Oracle Health/Cerner
| Pressure point | Practical consequence | What the evidence supports |
| Foundation-first build | Departments are pushed toward common, prebuilt workflows rather than independent, local configuration. | Epic describes its Foundation System as a default design based on customer practices; the degree of local variation is a governance decision. |
| Shared enterprise activation | Scheduling, medication management, nursing, orders, labs, interfaces, coding and revenue cycle must work across departments at once. | Observed in big-bang implementations; not required for every Epic customer. |
| Fixed readiness milestones | Late changes can disrupt integrated testing, training materials, conversion rehearsals and cutover planning. | Implementation-management inference; the exact project controls vary by contract and organization. |
| Centralized change authority | Local department preferences may be rejected to protect a system-wide common build. | Common enterprise-governance effect, not proof that Epic forbids every customization. |
| Limited staggered learning | A large activation leaves less opportunity to use one department's live experience before another department converts. | Contrast described in staged-implementation case studies. |
| Go-live issue concentration | Support queues, command center, clinical informatics, and interface analysts all face problems simultaneously. | Explicitly described by Vanderbilt's implementation leadership. |
10.0.2 Why the approach can feel draconian
| Health system | Activation model | Specific evidence |
| Intermountain Health | Big bang | 33 hospitals across six states transitioned together in September 2025. Source |
| Vanderbilt University Medical Center | Big bang | More than 18,000 physicians, nurses and staff switched on one day. Its informatics leader said issues from simple to complex occurred simultaneously; initial go-live support lasted about five weeks. Vanderbilt retrospective |
| Northwell Health | Regional waves | First major go-live November 1, 2025; leadership called a big-bang rollout across 28 hospitals and hundreds of ambulatory sites disruptive. Feb. 2026 report |
| UPMC | Phased waves | UPMC selected a phased transition to reduce risk to patient care, integration, data conversion, and readiness while consolidating nine EHRs. Source |
| WMCHealth (planned, March 2026) | Big bang | Public RFP questions and answers specify a single enterprise instance, all facilities in scope, and a big-bang go-live. RFP Q&A |
10.0.1 Actual Epic implementation examples
| Dimension | Big-bang / enterprise activation | Phased / wave activation |
| Scope | Multiple sites and tightly coupled clinical, ancillary, and revenue-cycle functions start on the same date. | Sites, regions, service lines, or functions transition on different dates. |
| Benefit | Unified chart, workflows, registration, orders, and billing from day one; avoids running two operational models for an extended period. | Smaller immediate blast radius; later waves can incorporate lessons learned. |
| Risk | Simultaneous clinical, operational, integration, training, and revenue-cycle issues; very large command-center demand. | Dual-system reconciliations, cross-system handoffs, temporary interfaces, and prolonged cutover staffing. |
| Staffing | Intense concurrent build, validation, training, at-the-elbow support and command-center coverage. | Support resources distributed by wave, but teams may support recently activated sites while building the next wave. |
| Decision pressure | Early, cross-department agreement on standardized workflows and a common cutover date. | Repeated readiness decisions and transitional operating procedures. |
Verified finding (November 2026): Epic frequently favors a big-bang activation, with multiple hospitals, ambulatory clinics, and integrated applications switching together on one date. At Epic's 2025 Users Group Meeting, CEO Judy Faulkner explicitly described why she believes big-bang installs are often better: phased activation generates temporary workflows and repeated transitions. This is a preference, not a mandatory rule: large Epic clients also deploy in waves. Becker's, Oct. 2, 2025.
10.0 Enterprise-wide go-live versus phased activation: why Epic feels more rigid
10.6 Sources and evidence strength
10.7 Standard implementation stages and ownership
| Criticism | Evidence / context |
| “Epic dictates the workflow” | A hospital's own published guiding principles prioritize “Foundation, Foundation, Foundation” and adherence to standardized workflows. |
| “They won't customize before go-live” | A 2013 consultant critique described a model plan that limits extensive customization until after activation. This is a criticism, not a universal Epic policy. |
| “The hospital needs far more people than expected” | The same critique alleged staffing plans could underestimate needs by 20–30%; this is a historical consultant claim, not a verified current benchmark. |
| “They are extremely controlled, but it works” | Modern project schedules show tightly coordinated build, testing, readiness and cutover checkpoints; the trade-off is less flexibility during deployment. |
10.5 What the criticism actually says
| Deployment scale | Published benchmark | Interpretation |
| Smaller health systems | 9–12 months, commonly | Epic implementation executive statement (2024) |
| Larger health systems | First go-live 16–18 months, subsequent waves ~6 months apart | Epic implementation executive statement (2024) |
| Community Connect clinics | Can use shorter affiliate rollout schedules | Depends on host standardization, interfaces and conversion |
10.4 Timing and model
| Checkpoint | Operational meaning | Evidence |
| Build waves 1–4 | Incremental configuration according to a master schedule | UMC / TTUHSC implementation timeline |
| Readiness waves 1–4 | Progressive clinical, operational and system validation | UMC / TTUHSC implementation timeline |
| Integrated testing | End-to-end workflows across applications and interfaces | UMC / TTUHSC implementation timeline |
| Parallel revenue-cycle testing | Check charges, claims and financial results before conversion | UMC / TTUHSC implementation timeline |
| TDR | Test workstations, devices, printing, scanning, access | CTG and published project schedule |
| GLRA | Escalate unresolved readiness risks at scheduled milestones | CTG and published project schedule |
UMC / Texas Tech's published implementation schedule identifies four configuration waves, four readiness waves, three rounds of integrated testing, two rounds of parallel revenue-cycle testing, Technical Dress Rehearsal, and readiness assessments at 120, 90, 60 and 30 days before go-live. These are specific documented project practices; not every Epic contract must use identical counts.
10.3 Actual implementation sequencing: documented example
| Workstream | Typical deliverables / activities | Responsible teams |
| Governance & scope | Executive sponsors; steering committee; issue escalation; scope decisions; master schedule; project controls | Hospital leadership, PMO, Epic implementation team |
| Foundation & design | Foundation walkthroughs; workflow validation; gap and exception decisions; future-state process maps | Operational workgroups, physician/nursing champions, app analysts, Epic |
| Build waves | Application configuration; security roles; order sets; documentation templates; charge capture; build trackers | Certified app analysts, operational SMEs, Epic TS |
| Technical | Hosting/infrastructure; workstation inventory; printers/scanners; interfaces; devices; identity/access; downtime | ECSA, infrastructure, interface, security and vendor teams |
| Data conversion | Patient matching; appointments; allergies, medications, histories; mapping; mock loads; reconciliation | Conversion and clinical validation teams |
| Testing | Unit/functional tests; integrated testing; Day-in-the-Life; revenue cycle parallel tests; defect triage | App teams, operational SMEs, billing, QA |
| Readiness | Technical Dress Rehearsal (TDR); device validation; training records; GLRA checkpoints | Project leadership, IT, training, operations |
| Activation | Cutover checklist; appointment conversion; command center; at-the-elbow support; issue prioritization | Joint Epic/hospital/contractor go-live organization |
| Stabilization | Hypercare; support handoff; defect backlog; optimization governance; upgrade planning | Hospital application support, Epic TS, operations |
10.2 Epic-specific project workstreams and deliverables
Critical distinction: Foundation is not “zero customization.” Peer-reviewed interviews with EHR implementation leaders describe extensive parameter-level configuration as necessary, while over-customization increases maintenance difficulty. Epic's rigid reputation reflects an implementation operating model, not an absolute ban on customization.
| Control | What occurs in practice | Why it feels restrictive | Trade-off / benefit |
| Foundation System default | Begin with Epic's configured standard workflows rather than blank-sheet design. | Local practices must justify deviations; departments cannot simply recreate their old EHR. | Limits unnecessary build complexity and accelerates deployment. |
| Workflow standardization | Operational committees select common workflows, often across many hospitals. | Clinicians or sites lose some local autonomy. | Consistent care processes, training and analytics. |
| Decision deadlines | Workgroups must make and record decisions to keep successive build waves moving. | Late objections and requests can be deferred to optimization. | Protects interdependent testing and go-live schedule. |
| Cross-functional governance | Clinical, revenue cycle, IT and operational representatives must approve design. | A single department may not control a shared build decision. | Prevents downstream conflicts and fragmented configuration. |
| Required training and role validation | Analysts, super users and end users receive role-specific preparation. | Time away from patient care and mandatory training obligations. | Reduces operational disruption at activation. |
| Readiness gates | Technical dress rehearsals, integrated testing and 120/90/60/30-day readiness reviews assess preparedness. | Open issues are visible to executives and can force corrective action. | Surfaces patient safety and financial risks before cutover. |
| Centralized cutover | Command center, issue triage, go-live schedule and stabilization are coordinated. | Limited freedom to change workflows during activation. | Maintains a controlled transition. |
| Customization constraints | Organizations may tailor Epic extensively but must manage build, test and upgrade implications. | Desired changes are often challenged or pushed to post-live optimization. | Contains cost and long-term maintenance burden. |
Epic implementation is Foundation-first, governed, highly scheduled, and adoption-intensive. The characterization “draconian” is an industry criticism, not Epic's formal methodology name. The documented mechanisms behind that impression are:
10.1 Why Epic implementations are perceived as rigid
| Stage | Epic implementation practice | Primary ownership |
|---|
| 1. Initiation | Governance, charter, scope, executive sponsorship, project plan | Hospital leadership + Epic implementation leads |
| 2. Foundation & discovery | Review Foundation System / standard workflows; identify local variation and decisions | Operational workgroups + Epic |
| 3. Design & build | Workflow validation, configuration, security, application build waves | Hospital certified application analysts + Epic TS/IS |
| 4. Data & interfaces | Legacy conversion, HL7/FHIR integrations, external systems, device setup | Hospital technical teams + vendors + Epic |
| 5. Testing | Application testing, integrated workflows, billing/claims, conversion validation | Application analysts + clinical/business owners |
| 6. Readiness | Training, Day-in-the-Life, technical dress rehearsal (TDR), go-live readiness assessments (GLRA) | Hospital training/operations/IT + Epic |
| 7. Go-live | Cutover, on-site support, command center, triage, defect prioritization | Hospital + Epic + external partners |
| 8. Stabilize & optimize | Resolve defects, measure adoption, hand over into operational support | Hospital support organization + Epic |
Terminology: Foundation System, workflow validation, build waves, integrated testing, Day in the Life, TDR, GLRA, command center, optimization. Methodology classification: organizational implementation approach, not an Epic software module.
Evidence: Stanford implementation phases · UMC/TTUHSC implementation timeline · Epic vendor services.
14
Support Organization & Operating Model
Support is shared across customer operations, customer IT, Epic employees and optional consulting or technology partners. Hospital support is not exclusively staffed by Epic.
| Layer / Team | Staffing model | Role / escalation |
|---|
| L0. Super users / education | Hospital operations | Answer workflow questions, coaching, adoption |
| L1. Service desk | Hospital in-house or outsourced | Ticket intake, access, basic troubleshooting and routing |
| L2. Certified Epic application analysts | Hospital staff or contractors | Application build, In Basket routing, clinical workflow, billing, Cadence/Prelude and other module support |
| L2. Interface & integration | Hospital integration team, vendors or managed services | HL7/FHIR interfaces, interface engine, APIs, connected applications; coordinate with Epic Bridges/Interconnect |
| L2. Infrastructure & ECSA | Hospital IT, Epic Hosting or other host | Client infrastructure, servers, environment operations, networking and device issues |
| L3. Epic Technical Services (TS) | Epic employees | Epic-specific technical/product issues escalated by customer analysts |
| Implementation Services (IS) | Epic employees | Project and workflow implementation support during rollout |
| Project/PMO & clinical governance | Hospital + Epic + contractors | Priorities, change control, operational decisions, release readiness |
| Boost / staff augmentation | Epic Boost or outside consultants | Additional Epic-focused PM or technical capacity when contracted |
| Third-party vendor support | External vendor | Vendor-owned application, interface endpoint or device problems |
| Community Connect host | Large hosting health system | Application support, upgrades and service desk for subscribed clinics; model varies by sponsor |
Typical escalation
User → hospital help desk → module/application analyst → Epic TS (if an Epic product issue). External-interface defects also route to the hospital integration team and the third-party vendor. Infrastructure incidents route to the responsible hosting team.
Integrated into other support systems?
Yes. A hospital may use its enterprise ITSM / ticketing platform, change-management and monitoring systems, and its own integration engine. Epic product escalation and vendor collaboration remain distinct from the hospital’s internal service-desk workflow. No universal ITSM vendor or mandatory ticketing platform is implied.
Support-by-hosting-model
| Model | Clinical/application support | Infrastructure support |
|---|
| Hospital self-hosted | Hospital analysts + Epic escalation | Hospital infrastructure teams |
| Epic Hosting | Hospital analysts + Epic escalation | Epic hosting operations; local devices remain hospital responsibility |
| Community Connect | Sponsoring health system support (contract-dependent) | Hosting health system; local networks/devices generally remain practice responsibility |
Evidence: Epic official support roles · Epic Boost · NYP Community Connect handoff · AdventHealth Community Connect support and upgrades.