How to build a better insulin pump
Recently, Medtronic posed the following question on their Facebook page:
If you could add one feature to your pump, what would it be? We’re listening.
Something really impressed me about this question: the last two words. I like dialog, discussion, and patient outreach. It’s good to know they’re listening.
What didn’t impress me were some of the responses. While many of them were practical and thought-out, some of them were, to be honest, boneheaded. I won’t detail them here, as I’m not out to insult anyone, but part of the reason vendors don’t often open up the suggestion-box to the public like some of the responses you get are either illogical, impractical, or impossible (or require a cellular contract with data-plan — no thank you). Someone even suggested the pump include a cure “feature” — who wouldn’t want that? (Well, some might say the folks who make a living by treating diabetes, but I’m not one of those conspiracy theorists).
Still, there are some very simple, yet very beneficial improvements that can be made, and rather than just leave one comment amid a thousand (at last count) on a Facebook post, I’m going to take advantage of my blogger pulpit and talk about some of my ideas here. If you’ll indulge me…
By the way, I think Medtronic makes a fine insulin pump. I’ve mentioned earlier how it’s boxy design and low-res black-and-white display isn’t particularly sexy, but it’s still the smallest pump on the market (the new Asante Snap might challenge this), and I find it much easier to defend its features than criticize them. In otherwords, it’s got so many good features that are unmatched by the competition, and that’s why I’ve stuck with them all this time. As for all those technical difficulties I’ve had, well — they just put out a bulletin advising the proper way to fill a reservoir, and it turns out I’ve been doing it wrong all this time. More on that in another post…
But still, it can be better. For instance
The Correction Wizard
The pump does a good job at suggesting how to correct a high blood sugar. It simply takes my BG, beamed directly from my meter (!) and subtracts from it my target BG to figure out how much it needs to come down. It then takes that result and divides by my insulin sensitivity to figure out how many units (or fractional units) are needed to lower my BG from the target. Subtract the Active Insulin (IOB) from that, and you’ve got an answer. Simple math, really.
But my pump also has carb ratios stored in its brain — how much one carb will raise my BG. How nice would it be if it could tell me how far below my target I am, and then how many carbs I need to consume to bring me up to where I need to be? The “rule of fifteen” is arbitrary and goes back to the days when we used long-acting insulin and the concept of IOB was nonexistent. There’s a better way, let’s do it.
(Meanwhile, my “better way” consists of pretending to “bolus” for four carbs, then eight) until it tells me I need a bolus to cover it. Then I know how many glucose tabs to eat. But doing this while low is tedious and potentially dangerous.)
Of all the “safety” features built into the pump, either by Medtronic’s own discretion or mandated by the FDA, I’m surprised this one slipped through the cracks.
When setting a temporary basal, the pump always starts at 100%, or the “normal” basal. There’s then the option to scroll down to zero or scroll up to 200 percent of normal. If someone were to scroll up to 200% and hit the UP arrow, it would wrap around to zero. Similarly, scroll down to zero and hit DOWN, and you end up at 200%.
One: I can’t imagine why anyone would ever want to do this. The starting-point is in the middle, so the fastest way to get to the lowest value is to scroll down. Two: this is incredibly dangerous. Knowing just how easy it is to accidentally deliver ZERO basal instead of DOUBLE basal (or vice-versa) can have some serious repercussions. I use temp-basals of zero a lot… and as I go down, percent-by-percent, to zero, I often find myself passing the zero mark and then going back. There’s no reason for this wrap-around. Remove it.
And while we’re at it, why do we offer 100 different intervals between full-basal and no basal anyway? If somebody’s basal is, for sake of argument, a generous 10 units per hour, that means it can be adjusted to the one-tenth of a unit intervals. That precision is overkill, not to mention unnecessary given the SWAGgy nature of temporary basals anyway. Why not let me step-up and down in 5% intervals instead? That reduces the button-presses twenty-fold and makes setting them a heckuva lot easier.
One more basal “game”, and I recognize this one might be a stretch. I’m sure it would be harder to incorporate in the pump and get regulatory approvals, but a built-in Super Bolus option would sure be nice…
The Final Countdown
The CGM that’s built into my pump does a good job of alerting me that it will be due for a calibration in 30, 60, or 90 minutes (user definable). This is quite handy, as I’d much rather calibrate before my meal rather than find out that, an hour after, I’m due.
But when the three-day sensor ends, there is no such warning. My SENSOR END alert comes up — without warning — and I’m stuck. It would be so nice to be alerted that my sensor is about to expire in an hour, so I can attend to it before I’m blindsided by its absence. I feel naked without a functional CGM, and that warning would at least let me grab a bathrobe, so to speak, so I’m not too vulnerable. (I’m not even talking about the alleged workarounds to the sensor’s end-of-life).
Riding the Waves
This is a big pet-peeve of mine.
When I deliver a bolus, 90% of the time it’s a Normal bolus. Just a regular kind, where all the insulin is delivered at once. To do this, after putting the carbs into the pump, I press “ACT” three times – first to confirm the number of carbs, second to accept the “Estimate Details”, and third to select Normal Bolus. Because 90% of the time, that’s the selected option.
But the other times? If my last bolus was a Square- or Dual-Wave, then it defaulted to that option, not the Normal one – and I probably didn’t notice it. The next step is to confirm the number of units of insulin and press ACT, then the bolus is delivered. So after pressing ACT, I turn my attention away from the pump and towards something else (like my delicious meal). However, if a Square- or Dual-wave was selected, it’s now prompting me for a time duration, but I didn’t notice.
I’ve missed many a meal-bolus because of this flaw (yes, I call it a flaw). If the same menu option was always selected, this wouldn’t happen. But by defaulting to different selections at different times, things happen that aren’t expected. I call this another safety concern.
The Suspends is Killing Me!
Related to the above (in more ways than one) is another default that changes following a Square- or Dual-wave bolus.
When pressing ACT from the home screen, the first option is for Bolus, the second is Suspend. The default (highlighted) option is the first. So when I’m getting ready to bolus, I simply press ACT three times: one to bring up the menu, two to select Bolus, three to select Bolus Wizard. But…
…if there is an extended bolus already in progress, the highlighted option is SECOND, not FIRST. I’ve found myself many times suspending my pump (and cancelling the extended bolus) rather than adding another Normal bolus on top of the one already in progress. Since the extended boluses are generally for pre-party “grazings” or long drawn-out meals (where the carbs just keep on coming, and can’t accurately be measured beforehand), I tend to want to Normal bolus on top of Extended boluses quite often.
True, if I accidentally suspend, I could go to the bolus history, see how much of the insulin was already delivered, then re-start a Square Wave for the rest, but I can only give it durations in half-hour intervals, whereas it might have had 15 or 40 minutes left before I accidentally suspended it.
But I doubt most people use the Square and Dual-waves nearly as much as I do, so this “fix” would probably not get too much attention.
Integrate the pump and CGM
This is what I wrote about during DBlog Week in 2012. The pump and CGM share a case, a display, and a few buttons, but there is nothing truly integrated about it. If it were, I wouldn’t have to set my high-thresholds artificially high to mask the postprandial-high nuisance alerts. I could keep it at 140 mg/dl throughout the day, and if the sensor read a higher value AND saw I had just bolused and there’s a ton of insulin on-board, it wouldn’t have to scream bloody-murder. Because there’s really nothing that needs to be done.
Likewise, when the pump is predicting a low, I do want it to consider the insulin on-board, not just the trend over the last fifteen minutes or so, because (as we all know) insulin is the leading cause of low blood sugars – and then scream bloody murder if appropriate. Just looking at the trend-line is inadequate; as any financial adviser will caution you: past performance does not guarantee future results.
I’m not looking for a closed-loop system that alters insulin delivery, just some intelligence behind the CGM alerts.
And finally — this is one of the common responses to that Facebook inquiry, and we can all thank Tandem for this one — it would be really nice to have the Insulin-on-board (aka “Active Insulin”) in an easy-to-find location; if not on the home screen, then the top of the status screen, not buried halfway down the second page between the time of the last meter reading and the last bolus. It’s an important value to consider, and I think lots of pumpers are paying closer attention to it these days.
Note: I have no relationship with Medtronic, or any other pump manufacturer for that matter, other than that of a customer. However, if they would like me to help advance these “wish-list” items into real, implementable features, it would be a dream-come-true for me. I’ve got expertise in technical-writing and specification-development, and I’m full of great ideas. Contact me.