Skip to content

Test Mapping

Tests and Panels in LIS

We assume that the connected LIS holds a register with tests and their properties such as expected result (quantitative, qualitative, etc.), minimum and maximum allowed values, reference ranges, etc. By referring term Test definition we will understand an entry in such a test registry. Following the common practice, we assume also that the LIS has definitions for panel examinations (profiles, batteries), e.g. Complete Blood Count, Lipid Panel, etc.

Tests in Laboratory Instruments

Basically, the instruments could be classified into two major groups - batch analyzers and selective analyzers. By batch analyzers we will understand instruments that make the same test(s) for all the loaded samples. The common example case are blood counters, which perform only CBC for any sample. On the other hand, selective analyzers are able to perform different tests for any given sample. An exemplary instrument is almost any classic (liquid reagent) biochemistry analyzer.

Batch analyzers are usually connected in an unidirectional way, i.e. they send results to the LIS and do not need orders. Selective analyzers usually need orders provided either manually entered in the instrument's software or provided by the LIS through a communication protocol.

Mapping

Every instrument identifies tests (assays, analytes) by its own coding system, e.g. GLU, 1205, etc. By test mapping we will refer to the linking between the instrument's test codes and the LIS's test identifiers, e.g. "the instrument's test GLU equals to the LIS test Glucose in serum with identifier 14771-0".

One to many (results)

There are cases, when instruments produce more than one value as a result of a single measurement. Usually, these are results in different units or presentations. Most of immunochemistry instruments for instance, report virology antibody tests both as quantity of measured antibodies (units per volume) as well as interpretation (positive, negative or gray zone). Other common case is urine strip readers, which normally report parameters as concentration (e.g. g/L) and arbitrary units (negative, +-, 1+, 2+, etc.).

The drivers of Iris are written in such a way, that the output of processing in terms of test identification are two values - test code and modifier. Modifiers are additions to test codes sent by the instruments, which acts as discriminators for differentiation of the results in different units or presentations.

Here is the example with a real-life urine strip analyzer (Sysmex UC-1000) for the parameter Bilirubin. You will have two results on every run as states the following example:

Sample barcode Test code Modifier Result
FS18A1 BIL -
FS18A1 BIL C 0

Depending of your needs, you could choose which result is suitable for your laboratory, and in some cases one could use both values.

One to many mapping

From the result processing perspective, Iris could have more than one results for a single analyte.

The following diagram explains the example above, using LOINC1 test codes.

flowchart LR
    BIL
    BLP[50551-1 Bilirubin - presence]
    BLC[68367-2 Bilirubin - moles/volume]
    BIL -->|Modifier: None| BLP 
    BIL -->|Modifier: C| BLC

Many to one (orders)

There are many cases where the LIS would have many tests which represent a single examination (analyte), e.g. concentration of a glucose in a serum, but with different clinical meanings. Therefore there will be many different test definitions for the same analyte. Probably the most common case is Glucose. The reason the LIS would have many definitions for this analyte could be different reference ranges, sample types (e.g. serum, capillary blood), etc. Usually that situation occurs in the response tests (measurement of the same analyte in a series of time points), where the LIS would differentiate results in time points by using different test definitions.

Many to one mapping

In Iris you may have many LIS tests mapped to a single instrument test.

Example and explanation

As an example case will be used Glucose tolerance test2. For the sake of brevity, we assume 2-point style of that examination panel. We assume that the LIS identifies tests according to LOINC1, but in real-life test identifiers could be anything, including database-generated keys, e.g. 12 and 15.

In the LIS, the laboratory order could be like that:

Sample barcode Sample type Test identifier(s) Test name(s)
A12B1 SERUM 14771-0 Fasting glucose in Serum
A13B6 SERUM 25668-5 Glucose in Serum, 2 hours post challenge

Let us have a biochemistry analyzer, which among other tests, measures glucose in a serum. The instrument is not aware of whether the sample is taken after overnight fasting or after taking procedure-defined volume of syrup. In this example, we assume the instrument identifies glucose test under the code GLU. So, from the instrument's perspective, the orders would be identical for all of the taken samples and would look like this:

Order number Sample barcode Sample type Ordered test(s)
156234 A12B1 SERUM GLU
161182 A13B6 SERUM GLU

Considering this example, it's getting clear that test mapping between instrument and LIS is of the type many-to-one, i.e. many LIS tests could point to a single instrument test.

flowchart LR
    GLU[GLU]
    GF[14771-0]
    G2H[25668-5]
    GNTH[other variations...]
    GF --> GLU
    GNTH --> GLU
    G2H --> GLU

Resolving

Test resolving is a process of translation from an LIS order to an Instrument order and vice versa.

From LIS to Instrument

When sample order is sent to the instrument, Iris will order distinct list of test codes, translated through test mappings with the following filtering:

  • Only examinations with status Pending will be included
  • Only examinations that "belongs" to the sample (i.e. can be tested from it) will be included

Obviously, only examinations that have a test mapping will be sent to the instrument.

From Instrument to LIS

Whereas forming order to the instrument is relatively easy, back resolving is more complicated, because of the non-deterministic nature of mapping (many to one). The problem is that Iris needs to realize which LIS test corresponds to a given instrument's test code. Using the example above, when the instrument returns examination results, Iris will receive something like this:

Order number Sample barcode Sample type Test code Result
156234 A12B1 SERUM GLU 5.6
161182 A13B6 SERUM GLU 11.2

So, the task here is to decide what GLU means - Fasting glucose (14771-0) or Two hours post challenge response glucose (25668-5). Iris will try two different strategies.

According to prior sent order strategy

Iris will look into its internal database and "see" what LIS code was the origin for ordered GLU for given sample. From the example, it is getting clear that GLU for sample A12B1 is ordered because LIS order has 14771-0 (Fasting glucose), whereas GLU for A13B6 is 25668-5 (Glucose in Serum, 2 hours post challenge).

Sometimes, Iris will receive results without prior orders. The reason could vary from the unidirectional driver (for instance drivers for almost all batch glucose analyzers) to manually ordered in instrument's software sample (probably because of a non-standard container).

So, if that step failed, Iris will try second approach.

According to LIS strategy

In this strategy, Iris will "ask" the LIS what examinations have to be performed for the given sample. In our example, LIS will "answer" that for sample A12B1 ordered test is 14771-0 (Fasting glucose). After look up into test mappings, Iris will "realize" that correct LIS test is 14771-0 (Fasting glucose). Similarly, for the A13B6 resolving will end with 25668-5 (Post challenge glucose).

If this strategy has also failed, then result will not be resolved at all. Note that if the LIS responds with more than one possible examination, Iris will take first found one.

Resolving limitation in many-to-one mapping

The obvious limitation for successful resolving in case of many LIS to one instrument test is single variation per sample.

Resolving limitation

The sample must have as ordered only one variation of the instrument's test, otherwise resolving will be non-deterministic and will fail.

  1. LOINC is an international standard for identifying health measurements, observations, and documents. ↩↩

  2. Glucose tolerance test is the most used response test where glucose is measured in two or more time points in different patient conditions. ↩