By the book and by the numbers

My hosts were astounded. “You actually read the manual? The whole manual?”

Well, yeah. Sure. Of course I did. Poor journalist I’d be if I didn’t do my homework first.

“I bet nobody else outside this building has read the whole manual!” said one management-type admiringly.

“I bet nobody else in the entire universe has read the whole manual,” muttered an engineer into his coffee cup.

Granted, the Asante Snap Insulin Pump User Manual isn’t a great work of literature, and you won’t see it on the New York Times best seller list anytime soon. (I have a dream: Someday, some forward-thinking company will hire me to write an entertaining user guide.) Still, I found some interesting  things in Asante’s 167-page Snap book.

The first thing that intrigued me was that as well as listing the precautions pumpers should take while undergoing various medical procedures (back away from the MRI scanner and no one gets hurt), Asante also details what precautions pump-wearing healthcare workers need to take to protect their gear in the workplace.

Being a most-of-the-time healthcare worker myself, I found this very appealing.

There’s a nice two-page, three-column chart that lays out the details. The first column lists the medical procedure in question. The second column gives pump-safety instructions if the patient is a pumper, and the third column lists pump procedures when the healthcare professional is the pumper.

For instance, ladies, for those you getting a mammogram, Asante advises that the mammogramee disconnect and leave her pump in the dressing room, while leaving her  cannula in place. Meanwhile, the mammogramographer can leave her pump on, so long as she’s behind the shield when she presses the button. Healthcare pumpers who work with MRIs, on the other hand, are cautioned never to enter the scanning room with their pumps on, and laser surgery techs are cautioned that some lasers will cause the pump alarms to go off! Can you imagine the fear that would strike the heart of an innocent sugar-normal getting laser surgery if your air-raid siren accidently went off?

Also of particular interest to me was something nearly hidden in the very back of the manual: The Snap’s delivery accuracy. Here I found some shocking honesty that I’ve rarely seen in pump manuals. Asante shows an analysis of the delivery accuracy of their system at several insulin volumes. The low volume delivery accuracy is depressing, to say the least.

At 0.3 units per hour, the accuracy of the Snap is plus or minus 10%. Oh, snap! Or should I say…Oh, fucking Snap! It’s not too inspiring to think that our basal delivery can vary so much, but at least they are honest and open about it. And for me, anyway, it also calls into question the real value of some other systems that let us set rates out to thousandths of a unit. Bio-usability aside, are we kidding ourselves about the value and accuracy of such minuscule delivery amounts?

More on how Snap stacks up to the competition in that regard in a minute, but first, as pumps are more than the sum of their basal: What about bolus delivery? How accurate is that on a Snap?

It depends on how big a bolus you are taking. For a coma-inducing 25-unit mega dose, it’s accurate to 3%. I guess that’s good enough to guarantee a successful suicide attempt. Perversely, for smaller, more realistic boli, a one-tenth unit bolus is only accurate to plus or minus 27%. Add that error to the error range of your fucking teststrip and it will be clear why seemingly identical circumstances net very different results from day to day. I’m so depressed thinking about it that I want to go drink some cinnamon whiskey, but the carb count on the nutrition label is also subject to the same basic error range.

These numbers are so… awful… that I decided to do some digging. Is this a pump problem or a Snap pump problem? I looked at the published data on the Med-T Revel, t:slim, and OmniPod, and found that… Well, I’m not sure what I found. Generally speaking, all three companies claim a plus or minus 5% accuracy on all basal and meal boli; but the other pump companies list their data differently. It’s like comparing apples to walnuts. For instance, OmniPod only gives their bolus accuracy data for volumes above half a unit, yet the pump is said to be able to deliver boli as small as 0.05 units, which is a volume ten times smaller. What’s the accuracy on that? The folks at OmniPod are mum on that subject

I suspect if we really knew how inaccurate all of our gear is we’d all become drunks. I’m beginning to see that diabetes control is a myth.

Back to Snap. Any other specs of interest? Not really. You can have up to four basal profiles (enough), but you have to switch among them manually. One of the things I miss most about my beloved Cozmo is the fact that it would automatically change basal profiles by day of the week. That let me have a clinic-day profile (more running around) and a writing day profile (more sitting on my ass) without having to remember to change every morning and then change back every evening. Each of the Snap profiles only allows ten steps. Hmmmm…. I know one endo who says that if you need more than four steps you need a new endo, but I’m a believer in smooth-curve basal changes. I have eight steps in my profile and I wouldn’t be insulted if someone felt they needed 24.

Speaking of steps and profiles, you can set eight different insulin-to-carb ratios in a day, but you are only allowed a skimpy three correction factors throughout the day. What were they thinking?

The Snap’s temp rates are limited to 24 hours, which seems sensible. The t:slim, which is currently living on my desk instead of on my pocket, as I’ve been double-checking some items on it as I compare the two systems, let me set a whopping 72-hour temp rate of zero units per hour. Before you wonder why on earth a pump would let you set a deadly temp rate, bear in mind that you might want to run, say, a +50% temp rate for three days when you had the flu. Snap’s temp rates range from 0% to 500% of programmed flow, but won’t actually let you choose the percentage you really want. Say what? That’s confusing and post-worthy by itself, so I’ll dig into that later this week.

By the way, the manual’s final word is that the Snap conducts more than 8 million “redundant safety cross-checks per day for both hardware and software functionality.” As there are only 86,400 seconds in a day, that means the Snap is preforming more than 92 checks a second. Busy little beaver, that Snap.

Well, actually that’s a lie. The part about the last word, not the part about beavers. I took a literary license. The last thing the manual really says is: “This page intentionally left blank.”

Not an ending worthy of great literature.

Next time on LifeAfterDx: Why at least one skeleton in the closet is a good thing