A better sensor?

So I’m in the middle of a brief affair. More than a one night stand, but less than a relationship of any sort. Kinda of like a wild weekend with a Swedish Stewardess who is going to be flying home on Monday.

Yep. I’m test-driving the Dexcom Seven Plus CGM.

And it’s about fucking time.

As a reminder to all of you, I was standing on the beach when the very first CGMs crawled out of the primordial soup onto dry land. I was there from the very beginning of the Continuous Glucose Monitoring revolution. I’ve worn every model of CGM ever made, except Dexcom. I was never able to get my paws on one, and never had the resources to just go out and buy one. Until now. I was finally able to score a loaner and one single solitary seven-day sensor.

But like the two-timing fink I am, I’m still carrying-on with my Med-T Girl at the same time. It is a lot of gear with two sensors, two monitors, and my insulin pump to boot. But it is very interesting. I guess all men like to have two women fighting over them.

I’ve got a lot to say about the Dex, but I hesitated to say anything at all because a brief affair isn’t really enough time to get to know this girl. She might be really sweet, or she might be a total bitch. I’m not really sure what her inner-woman is yet.

So please bear in mind that my observations in this post aren’t from a properly long review period. Technology honeymoons are tough. Sometimes you love a new device at first, then it sours on you over time. Other times, you come to love a machine that you nearly kicked to the curve, even to the point you can’t do without it.

Also my observations are based on one single sensor. It is impossible to know if its performance is typical or atypical.

And I also realize that I’m accustomed to my Med-T Guardian, having worn it 24-7-365 for … well, forever. It is very easy to get into habits and assume that what you are used to is the right way.

OK, so I guess I’ve qualified my statements ad-nauseam, and at this point and need to cut to the chase. The Dex Seven day sensor is an awesome girl married to a complete dolt. I’m very, very, very impressed with the performance of the sensor, and very, very, very unimpressed with the device it talks to.

The monitor is worse than junk. It is a complete piece of crap. It is large, bizarrely shaped, and strictly right-handed. It has to be plugged into a wall to be recharged; and a single charge does not last the life of a single sensor. If you run out of juice in the field you are just plain screwed. By comparison, Guardians use a single universally available AAA battery that will last several sensors.

The belt case for the Dex can only be called a joke. I gave up on it and carried the monitor in my pocket.

On the plus side, the screen is large, bright, and easy to read. The menus are simple, logical, and (kudos!) plain English. Learning to run the device is easy-peasie. In a negative, strictly personal, pet-peeve of mine, I’m bugged that I have to turn the damn thing on. Can you imagine if nurses in Intensive Care Units had to turn on heart monitors to see if the patient’s hearts are still beating?

Anything that monitors my body for life-threatening trouble should be always-on to be glanced at. Remember the red L.E.D. digital wrist watches in the late 70’s? They were waaaaay cool, but they are not around anymore. Why? ‘Cause you had to press a button to see what time it was, something that you did not have to do since the invention of the wrist watch. Who has time for that?

I like to be able to glace at my CGM to see where I’m at. Of course other’s can see it too. Rio likes to check out my “trace” and Deb has been known to steal a glance at my CGM now and then. I was even standing at the nurse’s station about a week ago and Eve cast a glance at me and said, “nice line.”

Huh?

“Nice line.” She pointed at my waist, “on your CGM.” I was having a great low-carb day and my six-hour trace looked like a dead man’s.

That is one of several things I don’t like about Med-T’s combined pump and CGM. It turns off. Presumably they were nervous about the faster battery drain of leaving the screen on leaving someone pumpless. As if we are too stupid to carry a AAA battery, or find one before we go DKA.

So for some Dex-details, starting at the start: insertion. The Dex sensor comes in a large bag and looks like some sort of terrifying dental tool. It is a disposable, one-shot insertion device. I don’t like the size of the thing, but damn, does it work great! It is a four step or so process that sound and looks worse than it is. The manual somehow doesn’t really get across how it works and left me afraid I’d screw up my one-and-only sensor. But like many things it life, in practice it was really pretty simple. It generates less trash than the Abbott Navigator, but more than the Med-T with its re-usable Sen-setter.

Disclaimer: I find the Sen-setter to be a piece of crap and I put in my Med-T sensors by hand, simply not an option with the Abbott and Dex, which must be put in with the one-shot inserters.

The Dex sensor needle is much, much, much smaller in diameter. Insertion was virtually painless. It is similar in length to the Med-T, so it anchors nicely into your flesh. The Abbott system inserts rather shallowly, increasing the risk of it not staying in place with vigorous activity.

The transmitter is not rechargeable (boooo, hisssss) and expensive to replace. They do guarantee it for a year, but this approach pisses me off. Med-T did this on both the first and second generation CGM devices, and then stopped making the sealed transmitter/battery components; turning my first CGM into a $2,700 paperweight within 18 months of buying it in cash.

The Dex transmitter is small, and snaps into the sensor base using a built-in plastic crow bar. Even though the transmitter is actually smaller than the Med-T transmitter, the skin landscape required for the sensor site is actually larger due to the shape of the transmitter cradle and the size of the sticky-patch that holds it on your skin. I needed two overlapping IV3000s to secure the site, while I only need a single one to secure my Med-T Guardian site.

Now comparing the Dex’s monitor only to Med-T (as Abbott’s Navigator seems to be disappearing from the Planet—not officially dropped as a product, but very low profile right now); one of the largest differences is in the shape of the screen. The Med-T uses a rectangle and the Dex uses a square. This makes a much larger difference than you’d expect; but it is impossible to say which is better. The same data displayed on the two different screens looks very different. Depending on your point of view, the square either exaggerates excursions or the rectangle minimizes the appearance of change.

To illustrate this Mercator Projection Effect I created a fake blood glucose trace. The “data” is the same. The left-to-right size is the same. On the bottom one I stretched the up-and-down diminution to show you how the same information looks different on the different shapes of screens.

I think you could adjust to either. I always leave my Guardian on at the six hour screen; an option I would not have should I switch to Dexcom, it does not stay on, and ALWAYS reverts to the three-hour screen when you press a button. I choose the 6-hour screen on my Guardian as the 3-hour Med-T screen always shows a straight line unless the shit has really hit the fan. The small data period has been stretched out too much by the rectangle. Over time, I’ve gotten used to thinking in the perspective of six hours and it seems logical to me. The Dex screen, being taller, actually shows nice visual rate of change data over three hours; but at the same time, on the 6-hour screen can make a leisurely excursion look like a disaster. The other thing I don’t like about three hour screens is that the duration of action of fast-acting insulin is 4-hours and I think it is important to always be thinking about your BGL in the perspective of how it relates to your last shot, and how much more insulin might still be in play. Dex also features a 1-hour and 24-hour screen in addition to the three and six. The Med-T does not have a 1-hour display.

My biggest single bitch about the Med-T Guardian is its petite little voice and the fact it won’t squawk and vibrate at the same time (OK, so maybe that is two bitches). Imagine my delight at hearing a very loud Dex alarm during the setup of the monitor. And it vibrates too! Well, at least it is supposed to. The loaner I got will only utter a peep on the set up screen. It has not once used its voice when it gives a real alarm. Maybe I got a lemon, but still a bummer.

The Dex screen shows a trace made up of little + signs, one for each sensor reading. As the Dex is designed to ignore the occasional funky reading, the trace lines are erratic, with occasional holes in them, which I find disturbing. Top right is your current sensor glucose with an Abbott-style compass rose arrow indicating (in case you are blind) the rate of change. Actually, I kind of like this feature. Below that, the current time. Below that, the “hour screen,” telling you how many hours are being displayed. On the bottom right is the battery strength (ohmyGod! Find a plug! Quick!) and a telemetry icon letting you know the transmitter and the monitor are talking to each other.

The reason I emphasized transmitter, is that the Dex monitor will report telemetry with the transmitter sitting on your desk, unconnected. Med-T transmitters will only report telemetry when hooked up to a properly functioning sensor. This means that positive telemetry on a Dex monitor has nothing to do what-so-ever with the sensor’s ability to function. If you’ve somehow pulled out a Med-T sensor in a moment of wild passion (hey, I’m told it happens) you’ll lose telemetry and know where to look for the problem. Not so with the Dex. So Dex-wearing Diabetics are not allowed to have moments of wild passion.

On the left of the screen is BGL from 50 to 400 in 50-point increments. The body of the screen is blank with only two horizontal dotted lines showing your high and low thresholds; and this leads us nicely into another weakness of the Dex system. The Guardian lets you, basal-rate-style, to choose different high and low thresholds for different times of day. This is damn handy, as you might want to have tighter glucose control during the day without being woken up a night because your sugar just went over 150, or whatever.

Dex forces you to choose one high and one low threshold 24-7. And speaking of alarms, the Dex is less advanced in this respect than either the Med-T Guardian or the Abbott Navigator. Dex has no predictive alarms. I was so surprised by this I flipped through the manual several times and even double checked the internet; and the sad fact is that the Dex Seven only has threshold and rate of change alarms. How… primitive.

I think, given the weaknesses of CGM and interstitial fluid, predictive alarms are a must. This is the spice that make the recipe work. Also, given that the Dex shares the every-five-minute sampling time that Med-T uses; a predictive alarm is all that much more needed. Navigator (which does have predictive alarms) could probably have gotten away without one, as they have every-sixty-second sampling, but the Dex system is sorely lacking without one.

Now, ignoring these issues for the moment, let us just talk about the accuracy of the Dex sensor. Yeah, it rocks.

The two CGMs trended much the same over the week, but the Dex seemed more sensitive to change. If I was going up rapidly, the Dex seemed to pick up on it more quickly than the Med-T sensor did. Same was true for lows. The Dex was more responsive to both lows and the rebounds for the corrections. In 20 minutes they were always neck-in-neck; but in the heat of battle I have to tip my hat to the Dex. It responded more like the one-minute-sampling Abbott Navigator in lows, but unlike Navigator, Dex reports accurately on the high end as well.

I’ve been wearing the current Guardian model for almost three years and I’m pretty impressed with it overall, but I have noticed it does run a bit late to the party. If I’ve gone low the CGM reports the numbers accurately about 10-15 minutes later (which is OK, as the predictive alarm has already given me a heads up before I ever got low). Guardian is also slow on reporting a rise rate after treating a low. The Dex is faster to report, but far, but still leaves me bush-whacked with no advance warning. It is also much faster to report the rebound.

Within 25 minutes either system is accurately reporting the overall situation.

The performance of the Dex is so noticeably better that if the device were not so crappy I’d switch teams.

But… you have a gun to your head now… do you want the super sensor with the lousy machine; or the good sensor with the super machine? I want my cake and be able to eat it too (and then be informed about the high blood sugar that follows).

Sorry, Dex, a super sensor with a poor monitor does NOT trump the good sensor with a super machine. As a system, the Guardian still delivers more. Just having an awesome sensor is not enough. We don’t rely on sensors, we rely on systems that include a sensor.

I should mention that the Dex system has a very nice software, but that like the monitor screen, uses colored dots to show data points. On modal-day type overlays of data it is visually confusing to the point of nausea.

Now I do need to talk about the whole issue of calibration, as this is one of the great make-or-break elements of using CGM. The Abbott device is the domineering parent, strictly regulating when one can and when one cannot calibrate, so much so that the device suffers the devastating consequence of not working at all if it is not 100% satisfied.

Med-T systems need to be calibrated in calm water, when not much is changing, and seem to thrive on fewer, rather than a greater number of calibration finger sticks. I am totally convinced that most of the trouble people have with Med-T CGMs are caused by calibration issues.

The Dex is a whole different kind of animal, and in this one regard might make it the better choice for the majority of diabetics needing CGM. Yeah, you read that right.

Here’s the deal: you can calibrate a Dex at any time, even when your blood sugar is changing rapidly, without freaking it out. I don’t know how they did it, black magic, I suspect; but it is true.

I’ve also noticed that when you give the Dex a fingerstick, it serves not only as a calibration, but also a correction. Like all CGMs, the Dex “trends” well. In other words if it says you are going up, you are most likely going up. If it says you are going up very fast, you are most likely going up very fast. Now, one should not expect, however, that the number on the CGM monitor is going to match the number on your meter.

Remember that the CGM is only checking every five minutes. It isn’t really continuous at all. And of course it is also not checking blood sugar, it is checking glucose in the water between your cells. And on top of that, meters really aren’t all that good in the first place. In fact, the first calibration that kicks off a Dex sensor run requires two finger sticks which the device averages (you can steal this gem from the Dex play book, even if you use a Med-T device, but if you are using a Navigator you can only calibrate with the built-in FreeStyle meter).

OK, so where I was going with all of this is that CGMs frequently have a different opinion of your blood sugar than your meter does. Usually they are close, but every once and a while they can get quite far off.

Guardians use a daisy chain of calibration sticks, each new stick contributes to some sort of algorithm that keeps the ship sailing in the right general direction. Now when the sensor seems to have drifted off course, a calibration nudges it in the right direction, but does nothing radical to the readings. This can, in theory, leave it reading somewhat inaccurately for a significant period of time. But it is very stable.

The Dex, on the other hand, almost treats calibration sticks like correction sticks. If it is off by 30 points a correction makes a significant change in the sensor readings. To me it seems like the Dex algorithm is very heavily weighted towards the most recent fingerstick while the Med-T algorithm is comparing a number of fingersticks. The Med-T sensors seems to get better and better the longer they run, and this may have something to do with the accumulated calibration knowledge of the sensor over time. Dex is good, but I didn’t sense a change in performance one way or another over the life of the run.

The weakness in this approach, of course, is that a “bad” fingerstick is going to disproportionally effect the performance of the sensor. Oh, I suppose I should mention that right now I’m using a AgaMatrix WaveSense Presto meter and strips. (And I just got a new leopard/cheetah print meter case that is either very Safari Macho or Women’s under wear—depending on who you ask.) Presto strips are heaps more accurate than many options out there. I had noticed that my Guardian started behaving much better when I started using the Presto system too. So I guess the bottom line for CGM users is why spend all that money on sensors and then try to guide them through the night with a crappy meter.

So if you have a Dex, based on my very limited experience with it, I think I would encourage you to calibrate more often rather than less often. Which is the opposite from the advice I usually give. I guess I’d also worry much less about what conditions I calibrate under, which is great for kids or other “brittle” folks who suffer a lot of glycemic ups and downs.

One other cool thing that I do like about the Dex is that when you go to enter a calibration stick it displays the current sensor glucose, while the Med-T always starts at 100. This is a nice, user friendly extra, as often the two are close. It gets old scrolling up from 100 all the time. Bottom line on this: it is faster to enter a manual fingerstick in to the Dex than it is to enter the same fingerstick into the Med-T.

But, again, the great performance and flexible calibrations are not enough to make me switch. The system suffers greatly from the lack of a predictive alarm, the lack of ability to stay turned on for quick evaluation, the need for a wall plug to recharge, the lack of a rechargeable transmitter, the lack of variable threshold alarms, and the poor build quality of the monitor. When I first heard the nicely loud alarm during set up, I thought that this feature alone might be enough to make me change, but in operation it has never worked, so that added to my concerns about the physical quality of the monitor, and its lack of potential for functional longevity.

Also in the mix are the poor layout of controls for lefties (although in fairness, the Med-T device is two-handed and the Dex can be run with the right hand alone). Too bad the engineers didn’t incorporate some sort of sensor to “know” which side up the monitor is and then flip the screen.

But Med-T has some stiff competition here. If Dex comes out with a monitor that addresses these issues before Med-T launches their next generation (which seems over due, but then again the economy sucks) the Dex could become the clear choice.

All of that said, I did put a very young Type-1 on the Dex, the small sensor size being 50% of the equation (she’s a 32-pound skinny little thing, I dreaded the thought of putting the whopping Med-T sensor into her tiny little body. Hell, her arms are almost as small as the Med-T sensor needle!). The other 50% of my decision coming from the fact she NEVER has stable blood sugar to calibrate a Guardian.

I like to have a lot of tools in the tool chest. I’m very impressed with the Dex, and I can’t really say that about the Abbott Navigator. The Navigator has some fine features, but it is more like dating a Neanderthal. Fine device, back then. Also the loooooong spool up time almost killed one of my best friends who had a DKA crisis in the time it took the damn sensor to come online.

But Dex. The Dex I’ll add to my tool kit. And if she comes back from Sweden, I wouldn’t mind hooking up with her myself for another week or two.


(And before you all start comment-spamming me about my overdue CGM book please know that I am taking 16 days off around the end of the year to finish it. After all, who wants to see their diabetes educator three days before Carbmass?)