The moment a reconciliation spreadsheet becomes your real workflow engine, the problem is no longer a missing interface. It is a missing owner for workflow state.
Every diagnostic lab has a version of the same spreadsheet. It starts as a quick daily check: orders from the EHR feed on one tab, accessions from the LIS on another, results waiting in the LIMS on a third. A few months later it has a status column no system knows about, a person who owns it, and a morning routine built around it. At that point the spreadsheet is running your lab's workflow, and none of your validated systems can see what it decides.
This rarely means the LIS, the LIMS, or the middleware is the wrong product. It usually means nobody decided which system owns each part of the workflow, so the gaps between them filled up with exports, interface error logs, and manual matching.
An LIS manages patient-centered clinical testing from order to report. A LIMS manages samples, batches, instruments, and lab-centered data. Middleware connects analyzers and applies instrument-level rules. An interface engine moves and translates messages. When a workflow spans several of these systems, one layer must own its state, or a spreadsheet will.
In this article, we will explore:
- How the LIS, LIMS, middleware, and interface engine differ, using the IHE laboratory actor model as a neutral reference.
- The specific boundaries where diagnostic lab workflows break, and the signs that a spreadsheet has become your workflow engine.
- How to assign a single source of truth for orders, specimens, results, reports, and workflow state.
- How to design a lab workflow state machine that maps to HL7 v2 result statuses and CLIA record requirements.
- What an interface engine, middleware, and lab workflow orchestration software should and should not own.
- Which CLIA, ISO 15189, GDPR, and EU IVDR requirements shape where each boundary sits.
- How to choose between configuring, extending, building an orchestration layer, or moving to a custom LIMS.
What Is the Difference Between an LIS, a LIMS, and Middleware in a Diagnostic Lab?
Vendor definitions blur these categories, and modern products overlap. A more useful approach is to describe each by the job it does in the workflow. The IHE Pathology and Laboratory Medicine (PaLM) technical framework gives a vendor-neutral vocabulary for that.
In IHE's Laboratory Testing Workflow profile, the Order Placer creates orders on the clinical side, the Order Filler is the central lab-side actor that splits orders into work orders and consolidates results, and the Automation Manager handles work with the lab's devices (IHE PaLM TF Vol. 1, Rev. 11.0). In the Laboratory Analytical Workflow profile, the Analyzer Manager is usually a middleware or an LIS, and it sends analytical work order steps to the Analyzer and collects results (IHE LAW profile).
| Layer | Primary job | Typical data it owns | IHE role it usually plays |
|---|---|---|---|
| Laboratory information system (LIS) | Patient-centered clinical testing from order to verified report | Accessions, patient results, verification, reports | Order Filler |
| Laboratory information management system (LIMS) | Sample-centered lab operations: batches, plates, instruments, QC, inventory | Samples, aliquots, batches, run data, QC records | Order Filler or Automation Manager, depending on the lab |
| Instrument middleware | Analyzer connectivity and instrument-level rules | Work order steps, raw results, autoverification and rerun rules | Automation Manager or Analyzer Manager |
| Interface engine | Moving, translating, and routing messages between systems | Message logs, mappings, acknowledgments | None; it carries transactions between actors |
| Workflow orchestration layer | Owning cross-system workflow state, timers, and exceptions | Case state, exception queues, hand-off audit trail | None; it coordinates actors |
Two points from this model matter in practice. First, the same product can play different roles in different labs: a genomics lab often runs its LIMS as the system that accepts orders, while a hospital outreach lab usually runs an LIS for that. Second, no IHE actor owns workflow state across the whole lab.
No IHE actor owns workflow state across the whole lab. That responsibility must be assigned intentionally.
Why Do LIS, LIMS, and Middleware Boundaries Break Down?
Integration projects usually start by connecting systems in pairs: EHR to LIS, LIS to LIMS, LIMS to pipeline, LIS to billing. Each connection can pass its own tests, yet the workflow across them still fails. The breaks cluster at the same boundaries.
| Boundary | What breaks | What it looks like on the floor |
|---|---|---|
| Ordering system to LIS or LIMS | Unmapped test codes, missing requisition fields, duplicate resubmissions | Orders parked in an interface error log; staff re-keying from the requisition |
| Order to specimen | Specimen arrives before its electronic order, or the order arrives with no specimen | Unmatched accessions; a daily match-up export |
| LIS to LIMS (or LIMS to bioinformatics pipeline) | Two systems each think they own the sample or its status | Status disagrees between screens; manual status updates |
| Middleware to LIS | Reruns, dilutions, and holds handled differently in each system | Results released from one system while held in the other |
| Result to destination | Message accepted by the transport but not filed to the right patient or order | Clinicians call asking for results the lab marked as sent |
| Correction propagation | Corrected results reach some destinations but not others | Corrected reports tracked in email |
| Lab to billing | Billing derives codes from a different version of the order | Denials traced back to an order change nobody propagated |
Each row has the same root cause. A piece of workflow state, such as whether an order is accepted, whether a specimen is usable, or whether a result has reached its destination, exists in two systems or in none.
Interfaces copy data between systems. They do not decide which copy is right.
Signs a Reconciliation Spreadsheet Has Become Your Lab Workflow Orchestration Engine
Spreadsheets do not appear because teams are careless. They appear because people are covering a real gap, and they do it well enough that the gap stops being visible. Check your lab against these signs:
- A daily export compares orders in one system against accessions or results in another.
- Someone maintains a status column that no system of record contains.
- Unmatched or rejected messages sit in an interface error log that only one person checks.
- Corrected or amended reports are tracked by email or a shared file.
- Test code mappings live inside interface channel scripts that few people can read.
- Billing or customer service finds missing orders before the lab does.
- Turnaround time is calculated from spreadsheet timestamps rather than system events.
If several of these apply, your lab already has a workflow orchestration layer. It is just made of people, files, and inboxes, so it has no access control, no validation record, and no audit trail. CLIA requires the lab to keep records such as the positive identification of each specimen and the date and time it was received (42 CFR 493.1283(a)). When you reconcile those facts in a spreadsheet, the official record and the working record have already drifted apart.
Which System Should Be the Source of Truth for Orders, Specimens, and Results?
Every entity in the workflow needs exactly one system of record. Other systems can hold copies, but they should never correct those copies on their own. Write this matrix down before changing any interface.
| Entity | System of record | Systems holding copies | Conflict rule |
|---|---|---|---|
| Patient identity and demographics | Registration or ordering system, or the lab's own patient index for outreach work | LIS, LIMS, billing | Match on more than name and date of birth; route mismatches to a named role, never auto-merge |
| Test order as requested | Ordering system (EHR or portal) | LIS, billing | Keep the requested version alongside the version the lab accepted |
| Test order as accepted | LIS, or LIMS where it accepts orders | Middleware, billing, pipeline | Changes travel as order control messages, such as CA for cancel, not as silent edits |
| Test catalog and code mappings | LIS or LIMS test master | Interface engine, EHR order catalog | Version mapping tables; send unmapped codes to an exception queue, never to a default |
| Specimen identity and chain of custody | One system only: the LIS or the LIMS | Middleware, pipeline | Mint the accession ID once; every other system references it |
| Verified results | The system that verifies them | EHR, portal, billing | Downstream systems never edit; changes arrive only as corrections from the owner |
| Report document | Reporting system | EHR, portal | Version every report; keep superseded versions |
| Workflow state | Orchestration layer, or the single core system if only one is involved | Dashboards | One owner; every other view is read-only |
The last row is the easiest to leave blank, and it is where spreadsheets take root. When an order touches four systems, no single one can answer what state it is in unless it is given that job.
How to Design a Lab Workflow State Machine Across LIS, LIMS, and Middleware
A state machine defines the states a case can be in, the events that move it from one state to the next, and who is allowed to trigger each move. Designing one forces decisions about the boundaries that interfaces leave open. Where possible, align states to HL7 v2 result status values (table 0123) so every system speaks the same language (HL7 v2.5.1 table 0123).
- Order received
- Specimen received
- In testing
- Awaiting verification
- Final reported
- Delivered
- Order on hold
- Specimen rejected
- Preliminary reported
- Corrected
| State | Owner | Entered when | HL7 v2 signal |
|---|---|---|---|
| Order received | Orchestration layer or LIS | A valid order is accepted and acknowledged | Result status O (order received, specimen not yet received) |
| Order on hold | Exception queue with a named role | Validation fails: missing field, unmapped code, or suspected duplicate | Application acknowledgment AE or AR, or an internal hold |
| Specimen received | LIS or LIMS (one of them) | Positive identification matched; receipt date and time recorded | Result status I (specimen received, procedure incomplete) |
| Specimen rejected | LIS or LIMS | Acceptability criteria fail; condition and disposition recorded | Result status X if the order is cancelled |
| In testing | Middleware, LIMS, or pipeline | Work order steps sent to analyzers or the bioinformatics pipeline | Result status I |
| Awaiting verification | LIS or LIMS | Results stored but not yet verified | Result status R |
| Preliminary reported | LIS or LIMS | Verified early results released | Result status P |
| Final reported | LIS or LIMS | Results verified and released | Result status F |
| Corrected | LIS or LIMS | An error is found after release | Result status C |
| Delivered | Orchestration layer | The destination confirms it processed the result | Application acknowledgment AA |
The last row exposes a common gap. HL7 v2 separates an accept acknowledgment, which only confirms the receiving system committed the message to safe storage, from an application acknowledgment, which reports the result of processing it (HL7 v2.5.1 chapter 2). A lab that treats a transport-level ACK as proof of delivery can mark a result delivered when it never filed to the patient's chart.
Two more design details prevent common failures. HL7 v2.5.1 keeps the ORM general order message for backward compatibility only, and points to OML for laboratory orders (HL7 v2.5.1 chapter 4), so new interfaces should not copy old ORM assumptions without review. And every transition into Corrected should trigger delivery to every destination that received the earlier version, because CLIA requires prompt notification and corrected reports when errors are found in reported results (42 CFR 493.1291(k)).
Interface Engine vs. Middleware vs. Lab Workflow Orchestration Software: What Each Should Own
Teams often put workflow logic wherever it was easiest to add at the time. Over the years, that scatters business rules across channel scripts, middleware rules, and LIS configuration. The table below sets clear responsibilities.
| Layer | Should own | Should not own |
|---|---|---|
| Interface engine | Transport, parsing, translation, routing, acknowledgments, retries, message logs | Clinical rules, long-lived workflow state, or business logic hidden in channel scripts |
| Instrument middleware | Analyzer connectivity, work order steps, autoverification rules, reruns, QC holds | Patient-level order acceptance or cross-system reconciliation |
| LIS or LIMS | Test master, accessioning, results, verification, reports | The state of work happening in other systems |
| Workflow orchestration layer | Cross-system state, exception queues with owners, timers and turnaround alerts, hand-off audit trail | A second copy of clinical results or verification decisions |
Note that "lab workflow orchestration software" often refers to robotic scheduling for automated benches. In a diagnostic lab's information architecture, the orchestration layer coordinates orders, specimens, results, and exceptions across systems. It may be a module of a modern LIS, a dedicated service, or a carefully scoped custom component, but it must have an explicit owner.
Interface engine choices also carry licensing risk worth reviewing. NextGen moved Mirth Connect from a dual open-source and commercial license to a single commercial license with version 4.6, announced on March 19, 2025, and named 4.5.2 as the last published open-source version (NextGen). Before deciding to upgrade, stay, or migrate, inventory how much workflow logic lives inside your channels, because that logic moves with you.
Regulatory Requirements That Shape LIS and LIMS System Boundaries
Regulations do not prescribe an architecture, but they define facts the architecture must prove. For US labs, that means CLIA. For UK and European labs, ISO 15189:2022 accreditation, GDPR and UK GDPR, and, for in-house devices, the EU IVDR add their own expectations.
| Requirement | What it expects | Boundary implication |
|---|---|---|
| 42 CFR 493.1241 | A written or electronic request from an authorized person; requisition data transcribed or entered accurately | Validate requisition and order data at intake, before it reaches the LIS or LIMS |
| 42 CFR 493.1283(a) | Positive specimen identification, receipt date and time, condition and disposition of unacceptable specimens, testing records with personnel | One system owns specimen identity and its state history |
| 42 CFR 493.1291(a) | Systems that move results and patient data accurately and reliably to the final report destination, including interfaced reports | Delivery confirmation belongs to the workflow as well as the transport |
| 42 CFR 493.1291(k) | Prompt notification and corrected reports, keeping both original and corrected versions | Corrections propagate to every prior destination |
| 42 CFR 493.1105 | Retention periods for test records and reports | State history and message logs stay retrievable, not just the latest status |
| ISO 15189:2022 clause 7.6 | Control of data and information management, including information systems management, downtime plans, and off-site management | Documented ownership, validated changes, and downtime procedures for each system |
| GDPR Article 5(1)(d), retained in UK GDPR | Personal data kept accurate and, where necessary, up to date; inaccurate data erased or rectified without delay | Identity mismatches resolved through a defined process, not left in a spreadsheet |
| EU IVDR Article 5(5) | In-house devices exempt only under conditions, including EN ISO 15189 compliance and no transfer to another legal entity | Keep result-interpretation logic that may qualify as a device separate from workflow orchestration |
A spreadsheet can hold the right answer and still fail every row of this table, because it cannot show who changed what, when, and on whose authority.
Build, Buy, or Extend: Choosing an Architecture for Lab Workflow Orchestration
Replacing the LIS or LIMS is rarely the first move. Most labs fix boundary problems by moving logic to the layer that should own it.
| Option | Fits when | Watch for |
|---|---|---|
| Configure the LIS or LIMS | One core system already handles most of the workflow, and the gaps are rules inside it | Vendor configuration limits; upgrades that overwrite custom scripts |
| Extend instrument middleware | The problems are analyzer-side: autoverification, reruns, QC holds | Patient-level logic creeping into instrument rules |
| Add a workflow orchestration layer | Several systems share one workflow and exceptions are handled in spreadsheets or email | Building a second LIS by accident; keep results in their system of record |
| Move to a custom LIMS | No commercial LIMS fits the test menu or workflow, as is common in genomics and specialty labs | Validation scope and long-term ownership |
| Replace a core system | The system cannot meet the requirements at all, or is end of life | Cost, migration risk, and the same boundary problems reappearing on new software |
Whichever option fits, run new interfaces and workflow logic in parallel with the current process until outputs match, and keep downtime procedures current. Under CLIA, procedure changes still need laboratory director approval before use, and under ISO 15189 information system changes need to be controlled and documented.
How NonStop Engineers LIS, LIMS, and Middleware Integration for Diagnostic Labs
NonStop builds the connective layer between ordering systems, LIS and LIMS platforms, instruments, pipelines, reporting, and billing for diagnostic labs, including clinical genomics, clinical chemistry, toxicology, and pathology labs. The capabilities below are published on NonStop's own pages.
- Interface-engine channels per system pair. NonStop's LIS and LIMS integration services design Mirth Connect channels between each system pair, using HL7 v2 (ADT, ORM, ORU) and FHIR R4 resources such as ServiceRequest, Observation, DiagnosticReport, and Specimen. NonStop has built bidirectional integration layers for LabWare, STARLIMS, and LabVantage, alongside EHR-side work on Epic and Cerner, and maps ASTM/LIS2-A instrument connections so EHR-facing work does not break analyzer interfaces.
- Order validation at the boundary. Intergenix, NonStop's pre-built, production-tested order quality software, checks every order at intake against completeness, validity, and the lab's business rules. It reports failures at the field level, routes failed orders to a structured resolution queue, and deduplicates before validation so duplicate retries never enter the queue. NonStop estimates it cuts about 80% of integration effort and eliminates 8 to 12 hours a week of manual re-entry.
- Requisition intake with human confirmation. SmartReq reads mixed printed, handwritten, and checkbox requisitions, validates fields against a schema, flags low-confidence fields for confirmation, and logs every extraction and edit before the order flows into validation and the LIMS. NonStop estimates about 90% of intake time eliminated.
- Safe cutover. New interfaces run in parallel with the existing process until output matches, followed by phased cutover and a real-time interface health dashboard.
- Custom builds where products do not fit. When no commercial LIMS fits, NonStop builds a custom LIMS owned by the lab, with audit trails, role-based access, and HL7 v2 and FHIR R4 integration built in. When the gap is a cross-system workflow component, NonStop builds it as custom software for the lab.
In a published engagement with a U.S. genetic testing leader, NonStop modernized test order management with FHIR and HL7 interoperability, moved sample management to a LIMS-driven architecture for traceability, and automated reporting pipelines, cutting turnaround time by 55% while the platform stayed live. NonStop's work is engineered for HIPAA, SOC 2, GDPR, and the CLIA and CAP environment; the laboratory retains certification and accreditation.
Frequently Asked Questions About LIS vs. LIMS vs. Middleware and Lab Workflow Orchestration
What is the difference between an LIS and a LIMS?
An LIS manages patient-centered clinical testing from order to verified report. A LIMS manages sample-centered lab operations such as batches, instruments, QC, and run data. Many genomics and specialty labs use a LIMS to accept orders, while hospital and outreach labs usually use an LIS, so the right split depends on the lab's workflow.
What does middleware do in a clinical laboratory?
Middleware sits between analyzers and the LIS or LIMS. It sends work order steps to instruments, collects results, and applies instrument-level rules such as autoverification, reruns, and QC holds. In IHE terms, it usually acts as the Automation Manager or Analyzer Manager.
Is an interface engine the same as lab middleware?
No. An interface engine moves, translates, and routes messages between systems such as the EHR, LIS, LIMS, and billing. Instrument middleware manages analyzer work and results. Workflow rules should not be hidden in either one.
What is lab workflow orchestration software?
In a diagnostic lab's information architecture, workflow orchestration software owns a case's state across systems: which step it is in, what exception is blocking it, who must act, and whether results have reached their destination. It can be a module of an LIS, a dedicated service, or a scoped custom component.
Which system should be the source of truth for lab results?
The system that verifies the results, usually the LIS or LIMS. The EHR, portal, and billing systems hold copies and should receive changes only as corrections from that owner, never edit results themselves.
How do you stop orders and results from falling out of sync between the LIS and LIMS?
Give each entity one system of record, mint the accession ID once, route validation failures and unmatched messages to an exception queue with a named owner, and track workflow state in one place rather than in exports or spreadsheets.
Does an HL7 acknowledgment prove a result reached the patient's chart?
Not necessarily. In HL7 v2, an accept acknowledgment only confirms the receiving system stored the message. An application acknowledgment reports whether it was processed. Delivery tracking should rely on the application-level response.
Should a diagnostic lab replace its LIS or LIMS to fix integration problems?
Usually not first. Most boundary problems are fixed by assigning ownership, moving logic out of interface scripts, and adding exception handling. Replacement makes sense when the system cannot meet requirements or is end of life.

Mahendra works as a Sr. Bioinformatics Engineer at NonStop. He builds genomic pipelines, agentic AI systems, and AI-powered products for genomics and life sciences. He has 10+ years of experience in bioinformatics and genomics software. He studied Biotechnology at Mumbai University.

