HL7 has always included an abnormal flag with observations. In FHIR, it’s the Observation.interpretation. One of the reasons for doing this is because reference ranges can vary by so many factors (ethnicity, gender, age, timing of specimen, lab equipment used, etc.). The notion of “abnormal” is not so much a universal truth; rather, it’s a representation of whether the system reporting the observation believers it should be reported as abnormal.
We didn’t include abnormal flags in the early days of OpenMRS, but, if I had a time machine, I would have included it. The same goes for Observation.status to allow us to properly handle pending results, observations with exceptions (inadequate specimen), etc. If someone could come along and add status and abnormal attributes for obs, I think we would be very happy to add them.
The challenge is the central & often large obs table. Making changes to the obs table isn’t hardly possible for some larger implementations. This is where we need the third leg to this stool: a viable upgrade path for changes to large tables. For example, imagine an upgrade helper module that creates a second obs table with the new attributes and, in a non-disruptive (bandwidth-limited) way, begins copying obs to the new table and keeping any changes in sync (maybe with stored procedures?). A big implementation installs this module and a week later all their data has been copied into the obs2 table, so only a brief downtime is needed for the helper module to replace the original obs with the new populated obs table.
If you can give me (1) an Obs.abnormal, (2) an Obs.status (so status doesn’t get left behind), and (3) a viable upgrade path that doesn’t exclude larger implementations, I’m sold. 