Yeah. Before insulin was discovered, Type I diabetes was a death sentence.
You would effectively starve to death within a year of symptoms showing up, regardless of how much you ate. (IIRC, actual starvation could prevent/slow the progress in some way)
However, once you've been on insulin therapy for a while, eventually you'll be in trouble within hours of insulin becoming insufficient. (An especially big problem for pump users - people using long-acting insulins like Lantus probably will have 1-2 days before they're in serious trouble after stopping administration of insulin.)
This reminds me of rumors of studies a decade or so ago involving administering long-acting insulin to diabetics in their "honeymoon period" (After diagnosis and starting insulin therapy, in many cases a diabetic's requirements for injected insulin will drop to near zero after not too long, but this only lasts for a few months after it starts) - reducing load on the pancreas seemed to prolong the period, allowing them to rely on their pancreas to handle meals and such.
Of interest is the "52 people between the ages of 19 and 45 that have received a diagnosis of type 1 diabetes within the previous three months" - That's a VERY rare category of people. The most interesting is that 3 months is typically within that "honeymoon period". Diagnosis of Type I diabetes that late in life is very uncommon (which is why Type I is often called juvenile diabetes). There's also the fact that this might be far less effective on diabetics who have had the disease for years, who basically have no remaining beta cells. (In most cases, Type I diabetes in mice is artificially induced - in humans the root cause is that the immune system attacks beta cells, however, this might allow at least some of the cells to survive the onslaught by preventing a failcascade due to the cells being overworked.)
