We want to identify patients with a history of Shigella in trinetx data so used a query with the ICD10CM code for shigellosis as well as the LOINC codes:
46455-2 (Shigella sp DNA)
80350-2 (Shigella species+EIEC invasion plasmid antigen H (ipaH) gene)
70242-3 (Shigella species+EIEC invasion plasmid antigen H (ipaH) gene)
since we ran the query, we learned two of LOINC codes (70242-3 and 46455-2) are categorized as “both” for order/observation. We understand this to mean that these two codes don’t necessarily indicate a positive test but could indicate a test was ordered. We would be very grateful to know if we are interpreting this correctly so we can understand how to best restructure our query if needed! Many Thanks in Advance, Chelsea
Hello Chelsea and thank you for the question. I would like to address this step-by-step.
First: LOINC terms represent what was being looked for; what the assay is specifically capable of discerning. Secondly, the attribute of Scale tells us what type of result format is coming when this assay is reported. Each of the three LOINC terms you refer to are Ordinal. In the knowledge base Knowledge Base – LOINC resides a definition under section 2.6 for Ordinal…Ordered categorical responses, e.g., 1+, 2+, 3+; positive, negative; reactive, indeterminate, nonreactive. I would like to point out that each of these LOINC terms means the nucleic acid of Shigella was being sought (two using the specific ipaH gene). Once the assay is run, only a negative/not detected or a positive/detected result value is coming out. These answer values may be tagged by SNOMED CT codes. I would recommend querying the trinetx data where the LOINC term is located alongside a “Positive” or “Detected” answer. You wouldn’t get false positives where the LOINC term resides alongside a “Negative” or “Not Detected” answer.
Finally, the order/observation column in the LOINC database is described in the Knowledge Base in section 9.1.3; “provide users with an idea of the intended use of the term by categorizing it as an order only, observation only, or both”. If something is tagged as an order, it infers to me to search for it within the data set where I know of orders to reside. If I’m using HL7 v2.5 messages, for example, I would be searching the OBR segment. The presence of “both” in the order/observation column means it may be identified as an order or as a result or both. There is NO INDICATION of what the answer value actually is; only what was being looked for.
