Friends:
Thanks for all the input. The subject was also discussed at the Vocabulary Committee and offline. We think we got a good compromise between avoiding too much back and forth, making life easy for the ETLer (avoid splitting and merging) and for the analyst (the one who works with existing source codes and the one trying to capture information spread over several concepts).
To remind you, this is about deciding how to represent certain facts: analytes which are outside the normal range. We can treat them as pre-coordinated Conditions (e.g. “neutropenia”) or Measurements, either pre-coordinated (“neutrophil count below reference range”) or post-coordinated (“neutrophil count” and “below reference range”). The problem came up when mapping such source codes, typically ICD10 family diagnosis codes.
Here it is the resolution:
- Procedures and their results (e.g. “EKG: anterior ischemia”) are not affected by this.
- Concepts without a specific analyte (e.g. “liver enzymes”) are not affected by this.
- LOINC and Nebraska Lexicon concepts, whether pre-coordinated or not, standard or not, are not affected by this.
Affected are pre-coordinated source concepts (ICD10 family, Read etc.), their mapping to pre- or post-coordinated SNOMEDs and their Domains.
- All pre-coordinated SNOMED concepts falling into this class will undergo a revision of their Domain assignment as either Condition or Measurement.
- The source concepts mapped to pre-coordinated SNOMED Condition concepts will assume the same Domain as per mapping rules.
- The source concepts which have equivalent pre-coordinated SNOMED Measurement concepts (analyte and result in one) will be mapped to these and assume the same Domain as per mapping rules.
- In rare cases, when there are no pre-coordinated SNOMED Measurement concepts, they will be mapped to post-coordinated SNOMED Measurement (analyte) concepts and adequate value concepts (the result).
- At some point in the future, we may switch all mappings to post-coordinated SNOMED Measurement concepts and the adequate values. (There is no such a thing as post-coordinated Condition concepts).
How do we decide whether something is a Condition or Measurement? The rules are somewhat imprecise and arbitrary and constitute a gray zone. The Vocab team will just have to call it. But roughly, if the aberration of an analyte is over a period of time, typically requires repeat testing and needs treatment of its own, rather than the underlying cause, it will be a Condition, otherwise a Measurement. For example, neutropenia takes some time to develop and resolve, the WBC is checked every other day and it definitely needs attention - Condition. The concept “PSA level above normal range” is a single measurement and the underlying condition (prostate cancer, prostatitis) may need treatment or not, but the PSA itself is irrelevant - Measurement.
Once we have implemented that, you will find some concept either in the Condition or Measurement tables, depending on their assigned Domain. But here is the deal: If they are Measurements, they are pre-coordinated (for now). There may be also post-coordinated records, typically coming from the lab tests. So, when you want to select patients with, say, an elevated PSA, you will have to look for both: pre-coordinated concepts (“Elevated PSA”) and postcoordinated ones (“PSA level in blood” - “Elevated” or a number+unit).
