Cascading panels or batteries?

This seems like a fine opportunity for me to get clarification on an issue that we (I) have had since the beginning with panels/batteries.

First I would like to know if we distinguish between these concepts in LOINC

My impression of the difference is:

Panels - set of assays that are usually performed and reported together (e.g. chemistry analyses)

batteries - set of tests or panels that are performed to achieve a particular conclusion, may include reflex tests that are performed based on the results of preliminary assays.

An example would be a bacterial or viral identification that begins with an isolation and then depending on the initial result is assigned a sries of identifying assays until the organism is characterized. Only proper subsets of the battery may be performed depending on the inital results.

Most of the issues I am currently dealing with are of the “battery type”. I would like guidance on how to define these.

The consensus from the LOINC gurus (Clem and Stan) is that LOINC has not made any distinctions between panels and batteries, considering them basically synonymous (see also Section 8 of the LOINC User’s Guide). LOINC has evolved its mechanism for representing and constructing enumerated sets (panels) with other attributes, including classifying the panel elements as:

  • Required (R) - The panel element is always expected to be reported when the panel is resulted.

  • Optional (O) - The panel element may not be reported with a panel result depending upon institutional policy or capabilities of the reporting lab.

  • Conditional (C) - The panel element is a key finding in the panel report and should be assumed to be negative, absent or not present if the panel result does not include data for this element.

At the HL7/LOINC Lab Orders meeting at NLM in December, the attendees discussed adding an additional node of Reflex (R) to this classification list, but were in agreement that the exact details for the reflex conditions couldn't and shouldn't be fully specified in the LOINC panel definition. We could would probably add as an attribute a text description of the reflex conditions (in whatever detail we can get) like we do for Conditional elements. This remains an area of future work. We've asked for labs to send us details about their most common test batteries (including the list of LOINC codes for the constituent elements) to help inform the creation of a more complete list of panels and to help uncover special issues. To date, we (Regenstrief) have received few replies to this, but would welcome more.

There are LOINC order codes which don’t have any components listed within them. There are also LOINC order codes that have components, but no relationships (No Optimal, Conditional or Required). Will those be remaining in the system, or do you think they’ll evolve as we get public input? An example is 50673-3 Insulin XXX challenge panel

Jim, are you working on the material we discussed at AMIA last November? Where you have a sample and the lab is instructed to run as much as necessary to identify the agent or chemical?