I emailed Ed Uthman, a pathologist with a web site online who's very
generous with his time and knowledge.  Here's his reply about HbA1c:

Hi, Susan,

Each lab is required by regulators to establish its own reference ranges
using some sort of valid scientific method. In some cases, that means
testing a large group and doing descriptive statistics (e.g., mean,
standard deviation), then using those to establish reference ranges.
Other labs just use published standards and validate the range by
testing a select group of samples on both the lab's own machinery and
that of another reference lab. However it's done, you get a slight
variation of actual reference range values from lab to lab, but it's
usually pretty slight and doesn't make much clinical difference.

The main lesson is to not shop around from lab to lab. Find a good lab
and stick with it to follow each patient. Some doctors switch labs
frequently, looking for the best bargain, and they end up with a
confusing mush of data when trying to follow their patients.

Best regards,

Ed

Susan