I find the urine strips are pretty consistent by the time I'm eating ketogenically enough that something would have to be wrong with me to not be in ketosis. Like, if I'm eating 60 grams of protein, the rest fat, 5 grams carbs--consistent deep purple. At that point, short of some sort of type II diabetes or something, I'm probably in ketosis.

Dehydration is just one problem with pee strips, another is what the kidneys are up to. One study (I think Owens?) from the 60's had people on a five-week fast. Constant urine ketones. Added about 8 grams per day of glucose--urine ketones disappeared, blood ketones remained as high as before. The hypothesis at the time--and I think it still stands--had to do with urine ph. When glutamine is used for glucose production by the kidneys, ammonia is produced. Very alkaline. The ph balance of the urine is regulated in the starved state by decreasing reuptake of ketones, ketones in the urine keep it from becoming too alkaline. That bit of glucose came with a reduction of gluconeogenesis from glutamine, so less ketones needed to be spilled because of the reduction in ammonia production.

Schwarzenegger has a section in his Encyclopedia of Modern Bodybuilding where he suggests using urine strips on a bodybuilding diet to stay just out of urine ketosis--back off on carbs until purple shows, then add them in slowly to be sure you're not wasting protein. I don't buy the premise, but it's an interesting throwback to earlier, lower carb bodybuilding. Doubtless a lot of people would only think they were out of ketosis by that method, still have blood ketones, but not urine.

We know something about why urine ketones don't always tell you much about blood ketones. There's probably stuff somewhere about why breath ketones can be misleading as well.