Sunday, May 14, 2017

To UCUM or not to UCUM



Those who read my blogs already know that I am fighting already a number of years for using LOINC as the coding system for lab tests in CDISC-SDTM, instead of the by CDISCdeveloped controlled terminology that is NOT uniquely describing lab tests, even not when in combination with LBMETHOD, LBSPEC, . It has even been reported that this has led to serious problems during review at the FDA whereas the use of LOINC would have avoided this.

When at the end of last year, the FDA then announced that it requires LOINC coding for lab tests for all studies starting as of 2018, I was enjoyed. On the other hand, I was ashamed that it was the FDA that made this step, i.e. that they came to wisdom before CDISC does: after all, it is (or should be) CDISC that is the standardization organization, not the FDA.

Another item where CDISC did not make any progress, is on UCUM. UCUM stands for Unified Code for Units of Measure. It is the worldwide accepted notation for units, also allowing to easily do unit conversions e.g. using RESTful web services. Currently the use of UCUM notation in SDTM is not allowed. Due to the discussions about the FDA mandating LOINC coding, the discussion about UCUM recently was flaring up again, also as each LOINC test for which there is a unit, does have it as an UCUM example unit:

LOINC codes and UCUM units for some typical blood tests.
UCUM units with a "sad" smiley are NOT in CDISC controlled terminology


There have been several requests to CDISC to at least allow the use of UCUM notation in SDTM variables like --ORRESU and --STRESU. It was turned down with the argumentation:
Changing what unit values appear in ORRESU and STRESU would be much more impactful (than the introduction of LOINC).


I have to disagree. Allowing UCUM notation (not mandating it) would have little impact on the process: sponsors can still use CDISC notation if they want, but can also finally use UCUM notation e.g. in the case the information comes from electronic health records (where the use of UCUM is mandatory) or even when coming directly from (modern) lab apparatus that have the applicable LOINC codes stored and that already deliver the results using UCUM notation.


At this moment, we have the ridiculous situation that when lab results are extracted from electronic health records, the UCUM unit needs to manually translated into the CDISC unit, which is not always possible but also error prone, and leads to a Babylonian confusion when comparing lab test results between studies at the FDA.


If you agree that CDISC should at least allow the use of UCUM units, drop a comment here, or better, send your comment to CDISC. If you contact me, I can provide you the applicable e-mail addresses.