many, many relatively reliable websites define ketonuria using blood concentrations. that is, they say that urine ketones over 3.5 is a medical emergency, but the ranges they are reporting seem to be the same as blood, which seems suspect.
i had to dig down the results a little, but i think this is a better definiton:
We categorized ketosis according to blood β-OHB levels as absent (<0.6 mmol/l), mild (0.6–1.5 mmol/l), moderate (1.6–3 mmol/l) or marked (>3 mmol/l). Ketonuria was semi-quantified as mild, moderate or marked for urine ketone levels of 1–4 mmol/l, 5–14 mmol/l and ≥15 mmol/l, respectively. The presence of ketoacidosis was assessed using both definitions suggested by the American Diabetes Association (ADA) and the Joint British Diabetes Society (JBDS) [15, 16].
at 3.9, that would put me in the mild category, and not be that concerning at all.
but if the blood results were 3.9, i'd be dead.
there's a reason i couldn't make sense of it - it was wrong.