What lab middleware does: rules, autoverification, QC and instrument management
Lab middleware is often described as "the box that connects analysers". That undersells it. In a well-run lab, middleware is where the rules live: which results can go straight through, which must wait for a scientist, which are blocked because QC failed. This guide explains each job middleware does and what to look for.
Written and reviewed by the Catenix team. How these guides are written and checked.
In brief
- Middleware's first job is connectivity: speaking each analyser's protocol and delivering clean results to the LIS or EHR.
- Its second job is rules: checking each result against ranges, flags, QC status and operator authorisation before it moves on.
- Autoverification means results that pass every defined rule can be released without manual review, under rules the laboratory director approves and validates.
- Middleware also manages QC (Levey-Jennings charts, Westgard rules, lockout) and instruments (status, maintenance, connectivity health).
- Catenix provides these jobs for clinics and small labs, and can feed an existing LIS or act as the working record where there is none.
Take it with you: POCT connectivity buyer's guide · POCT readiness checklist (PDF, no form).
Job one: connectivity
Every analyser model has its own interface: protocol (ASTM E1394, HL7 v2, POCT1-A2 or proprietary), transport (serial or network), test codes, flags and quirks. Middleware holds a driver for each, keeps the connections alive, answers host queries, and converts results into what the receiving system expects. It also stores results when the receiving system is unavailable and forwards them later. Analyser not on the list? We build the driver, usually within a couple of days. See ASTM to HL7 conversion, our analyser connection guides for 130 analysers, and the open connectivity dataset.
Job two: result rules
Before a result leaves the middleware, it can be checked against rules the lab defines:
- Range checks: is the value inside the analytical measuring range, or does it need dilution and rerun?
- Flag checks: did the analyser flag an error, interference, haemolysis, lipaemia or a short sample?
- QC status: did the most recent QC for this test on this analyser pass?
- Operator authorisation: is the person who ran the test competent and authorised for it?
- Critical values: does the result need urgent communication and acknowledgement before release?
- Delta checks: has the result changed implausibly since the patient's previous value?
- Consistency checks: do related results make sense together?
Each rule either lets the result through, holds it for review, or blocks it. The rules are the lab's, written and validated by the lab; the middleware applies them consistently and records what happened.
A worked example: one result through the rules
Take one chemistry result and the rules an illustrative lab has written and validated:
| Check | The lab's rule | This result | Outcome |
|---|---|---|---|
| Measuring range | Value within the analyser's reportable range | Within range | Pass |
| Analyser flags | No error, interference or sample-quality flag | Haemolysis flag present | Hold for review |
| QC status | Last QC for this test on this analyser passed today | Passed at 08:05 | Pass |
| Operator | Competency current for this analyser and test | Current | Pass |
| Delta | Change from previous result within the lab's limit | No previous result | Not applied |
One hold is enough. The result goes to a review queue with the reason shown, and the scientist decides: release with a comment, rerun, or request a new sample. The record keeps every check, its outcome and who resolved the hold. Without middleware, the haemolysis flag would sit on the analyser screen or a printout, and the value could be typed into the record without it.
Autoverification
Autoverification is the release of results without manual review when every rule passes. In high volume labs it saves hours a day and focuses scientists on the results that need them. CLSI guideline AUTO10 describes how to design and validate autoverification rules. The principles for any lab:
- The laboratory director approves the rules.
- Each rule is validated with real and constructed results before use.
- Results that fail any rule go to a person.
- The system records which rules applied to every released result.
- The rules are reviewed when methods, analysers or ranges change.
Small labs often start with every result reviewed, then automate the clear-cut ones once they have data. Middleware should support both without changing systems.
Job three: quality control
Analysers run QC material through the same path as patient samples. Middleware recognises QC runs, files them against the right lot and level, plots them on Levey-Jennings charts, applies Westgard multi-rules, and can lock out a test on an analyser when QC fails until a repeat passes and someone has reviewed the failure. That link between QC status and patient release is the single most important thing an inspector checks. See our Levey-Jennings and Westgard guide and QC lockout guide.
Job four: instrument management
Middleware sees every analyser, so it is the natural place to show which are connected, which have stopped sending, which have QC due, and which are out of service for maintenance. Good middleware also holds maintenance, calibration and temperature records per instrument and shows downtime. See device operations.
Job five: the audit trail
For every result, middleware should be able to show the raw message from the analyser, the converted message sent on, the acknowledgement received, the rules applied, the QC status at the time, the operator, and who released it and when. That trail answers the question "how do you know this result is right?" in seconds rather than days. See the audit trail.
Where middleware sits in small and large labs
In a hospital, middleware usually sits between dozens of analysers and the LIS. The LIS owns patients, requests and reports; the middleware owns connectivity, rules, QC and autoverification, and passes released results up. The LIS team and the middleware team agree which system holds each rule, so nothing is checked twice or not at all.
In a clinic or small lab, there is often no LIS. Middleware then sits between the analysers and the EHR, or acts as the working record itself, with registration, review and reports. The jobs on this page are the same in both cases. What changes is where the patient record lives and which system sends the final report. Decide that first, and the rest of the design follows from it.
What to look for
| Capability | Question to ask |
|---|---|
| Drivers | Do you support each of our analysers today, by model? |
| Rules | Can we define range, flag, QC, competency, critical and delta rules ourselves? |
| Autoverification | Is every auto-released result traceable to the rules that released it? |
| QC | Is QC filed automatically, charted, and linked to patient release? |
| Instruments | Can we see connection health and maintenance per analyser? |
| Delivery | Which LIS and EHR routes do you support? |
| Audit | Can you show the raw analyser message behind any result? |
Common mistakes with middleware rules
- Autoverifying before you have data. Start with review of everything, then automate the clear-cut cases once you can show how they behave.
- Rules nobody owns. Each rule needs an owner, a reason and a review date.
- Copying another lab's rules. Rules must be validated on your analysers, your methods and your patients.
- Holding too much. If almost every result is held, reviewers stop reading the reasons. Tune the rules.
- Silent changes. A rule change is a change to a validated system: approve it, test it and record it.
Questions to ask any middleware vendor
- Which rules can we write ourselves, and which need the vendor?
- Can a QC failure or a lapsed competency hold results automatically?
- Does every held result show the reason, and is the person who resolved it recorded?
- Can we see which rules applied to any released result?
- How are rule changes approved, tested and recorded?
See Catenix device connectivity for how analysers connect, and ask whether your analyser connects inside ChatGPT or Claude through our AI assistant connector.
Where Catenix fits
Catenix is lab and point-of-care middleware that grew into a full platform. It connects analysers over HL7 v2, POCT1-A2 and ASTM E1394, and sends HL7 v2 or FHIR R4 on to clinical systems, through an on-site gateway, files QC with Levey-Jennings charts and Westgard rules, holds results when QC fails, an operator's competency has lapsed, or a critical result has not been acknowledged, shows analyser health and maintenance, and keeps a wire-level, tamper-evident audit trail aligned to ISO 15189:2022. It feeds an existing LIS or EHR, or acts as the working record for clinics and small labs. Catenix applies the lab's rules; it is connectivity, workflow and record-keeping software, does not interpret results and is not marketed as a medical device.
Questions people ask
What is the difference between middleware and a LIS?
A LIS manages patients, requests, reports and the lab record. Middleware sits between analysers and the LIS, handling connectivity, rules and QC. Some platforms, including Catenix, cover both for smaller labs. See LIS vs LIMS vs middleware.
Is autoverification safe?
When rules are approved by the laboratory director, validated before use, and every failing result goes to a person, autoverification is a recognised practice described in CLSI AUTO10.
Do small labs need middleware?
If results are typed from the analyser into another system, yes. Middleware removes transcription and links QC to every result.
Who writes the rules?
The lab. Middleware applies them consistently and records which rules applied to each result.
What is a delta check?
A rule that compares a patient's new result with the previous one and holds it for review if the change is larger than the lab's limit for that test. It helps catch mislabelled samples and analytical errors.
What should happen when a result is held?
It should appear in a review queue with the reason, and the system should record who resolved it and what they did: release, rerun, recollect or cancel.
Can middleware release results automatically?
Yes, if the lab writes, validates and approves autoverification rules, and every result that fails a rule goes to a person. Many small labs keep manual review for everything at first and automate later.
Sources and further reading
Read next
More for coordinators.
Analyser connection guides
Step-by-step guides to connecting point-of-care and lab analysers to your LIS or EHR over HL7, POCT1-A2 and ASTM, with bidirectional, wire-level...
Read →LIS vs LIMS vs lab middleware
The difference between a LIS, a LIMS and lab middleware, in plain terms: who uses each, what they store, and which a clinic or small lab actually...
Read →Interface engine vs lab middleware
When a general HL7 interface engine such as Mirth Connect makes sense for a lab, when purpose-built lab middleware is faster and safer, and how to...
Read →Take it with you
The buyer's guide, in your inbox.
Nine pages on what POCT connectivity is, the standards in plain English, the cost of doing it by hand and eight questions to ask any supplier.
Get the POCT connectivity buyer's guide
Nine pages, vendor-neutral: what connectivity is, the standards in plain English, the cost of doing it by hand and eight questions to ask any supplier. The PDF also opens directly from the downloads page; leave an email only if you want it in your inbox.
See it done in software.
A 30-minute walkthrough on a live tenant, with your analysers and your allowable-error limits.