@TiaG, us it possible his last 30 days were a lot better in control?
Don’t know if you remember these couple threads that happened a few weeks ago, when someone was arguing false interpretations of A1c articles because they were misunderstanding the arguments. In one of them, there was good mathematical modeling of A1c that showed the last 30 days of the A1c measurement to be much more significant in the overall A1c:
HbA1c: do the latest BG samples count for more?
I also ran around the same time into an article that argued with good math that each person has a different average half life for red blood cells, and that it should be taken into account for A1c measurements. The differences were quite significant.
As for us, we tend to trust GMI more than A1c. But this is not a good thing in your case 