2004-2007: I used the pump with Novolog and Humalog insulins (at different times).
2007, March: I went off the pump and switched to injections of long-term insulin, including both Lantus and Levemir
2008, January: Found and started training in CrossFit
2009, June: Returned to using the pump.
From 2004-2007, and indeed from 1996-2001, a combined total of 8 years of using an insulin pump, my basal rates remained consistently at or above the convenient unit of 1.0 units/hour. I vary that rate depending on sleep and activity schedules, so that I might go to 1.5 from 05:00 - 07:30 (historically). I spent more time at 1.1 and 1.2 and 1.3 units/hour, but during more active times and when I was training, I could spend 70% of my time at 1.0 units/hour.
In changing from Pump W/ Novolog, to Injections W/ Levemir (March 2007), I could not compare the effectiveness of Levemir to Novolog; the context and process muddled this comparison, and I never really saw it. I then had to change my injection schedules over time, and my injection amounts, and I never saw a clear change due to my adoption of CrossFit as my training methodology.
(My training methodology prior to January 2008 had been 70% distance running, 20% swimming, with small bouts of bodyweight strength training in the form of pushups, dips, handstands, rock climbing.)
Now I've been CrossFitting for 1.5 years, and I've started on the pump again, and we have the AFTER PICTURE: A basal rate of 1.0 is clearly too high. I am going low constantly, repeatedly. I am now being forced to lower my basal rates, the dominant rate is probably going to be 0.8 units / hour.
My basal metabolism is markedly more insulin sensitive due to my crossfit training.
The implications for insulin dependent diabetics are significant, meaningful; but they are even more profoundly meaningful for type 2 diabetics and pre-diabetics. The control sample was not "couch potato with high glycemic diet"; the control sample was extremely fit, healthy distance runner, swimmer, climber.