My name is Clemma and I am writing about my experience with the Dexcom Seven continuous glucose monitoring system. I live in Minneapolis with my young son and my not so young husband. I was diagnosed with Type 1 diabetes 11 years ago, when I was almost 21 years old. I started pumping 7 years ago, first with a Minimed 508, then an Animas IR1200, and now with the OmniPod. Friday, June 29 I hooked up to my newest constant companion, the one and only Comrade Dex...

Thursday, July 26, 2007

Stupid User Experience: Rebound high

About a week ago I had a Stupid User Experience (SUE). See the screenshot of the Glucose Trend graph. Note that this is an example of enlarging a selected time frame for closer scrutiny.

I was especially exhausted this night, and I had gone to bed without bringing some source of sugar for a nighttime low, like I usually do.

A little before 11:00 pm, I got the buzz alerting me to a breach of the low threshold. I was tired and really didn't want to get out of bed to find some juice. I'm embarassed to admit it, but all I did was turn off my basal delivery for an hour and a half, hoping that would counter the insulin on board, and I would magically sleep through the night without going low. Well, you can see the result. Around 1:30 am, I got the buzz telling me I was headed up. I think the high peaked at about 240.

Why am I sharing this? Two points. One, this is a great example of the Dexcom doing its job. One of my motivations for buying this thing was the increasing number of mornings I woke up over 200. Without this monitor, I probably would have slept through the whole rollercoaster ride. I would have gone low without realizing it, and I would have rebounded without realizing it, waking up hours later with a high blood sugar and no idea what had happened. Without the high threshold I would not have taken a correction insulin dose at 1:30, and might have peaked even higher, certainly for much longer, before I woke up.

Dexcom also did its job alerting me in plenty of time to avoid the low in the first place, which brings me to the second point. One big problem about diabetes that is probably common to many chronic conditions: it's easy to get complacent, and having the CGMS doesn't magically protect me from extreme blood sugar swings. I feel silly needing these occasional reminders, these SUEs, that simply being alerted to a low isn't enough - I still have to be in charge, even when I'm too tired. I don't know if non-diabetics can understand, and I hope I'm not the only one out there who does things like this.

I also hope after posting this that no one shows up at my door and revokes my license to practice intensive insulin management. Turning off my basal delivery - what was I thinking? Next thing you know it will be Regular and NPH for me, just like in the dark ages. Nooooooo!

The main point is: thanks Comrade Dex for doing what you're supposed to. I dropped the ball. Nice textbook example of a rebound hyperglycemia though.

Wednesday, July 25, 2007

Sensor 3 down the tubes

Four days? FOUR DAYS? Yes folks, sensor 3 lasted a measly four days before biting the dust. I was in denial most of yesterday, but finally last night I couldn't pretend any more. I tried to restart it, but no dice. Well, for $15 per day sensor 3 was not worth the expense. I haven't started sensor 4 yet. Probably this evening. Funny how 10 days disappointed me for sensor 1. Now I'll be thrilled with 10 days for sensor 4.

I know of one person who is on day 12 of his first Dexcom Seven sensor and still going strong. Lucky guy.

Friday, July 20, 2007

Demise of sensor 2 and blog format

Sensor 2 died yesterday. It actually did pretty well up to the end, but then, as with sensor 1, it started asking for repeated calibrations. This time I attached myself to sensor 3 and waited 6 hours before starting the calibration. My pal Bernard and a few other people have suggested this approach to increase accuracy when the new sensor starts transmitting readings.

Only 9 days on sensor 2! So much for the 14 I hoped for. I will be curious to see how many days other people get out of the Dexcom 7 sensors. I know a few others who are using it now, but I think they are all still on their first sensor, or else they switched to the new one after 7 days, without waiting for the first sensor to die.

I am going to keep a chart of how long my sensors last. See the page element on the right. Hence the new blog template. When I uploaded that chart on the old template, half of it was cut off and I couldn't figure out how to fix it. This was one of the only formats that showed the whole thing. I'll try to make it more palatable later this weekend or next week. Well actually I'll get my husband to do it, because I'm very bad at that sort of thing.

I optimistically included up to 14 days in the graph, but maybe I'll be lucky if I ever get 11. Ah, well, no one said this endeavor would be cheap.

Thursday, July 19, 2007

Dexcom Seven Software: Exporting Data

Apparently the really exciting thing the Seven software provides is the ability to export data for your own manipulative enjoyment. I find it hard to believe they didn't make this possible with the three-day system, but at least they got it right for the Seven.

Exporting the data is easy. From the main Dexcom Software menu, select Tools --> Export Sensor / Meter Data:

You can export all readings, or choose a subset. You can select sensor or meter readings (you have to export them separately and merge if you want them in the same file). You can choose an XML file, or tab-delimited TXT file. Defaults are all readings from the sensor in an XML file. Here is the dialog box for exporting data:

Here is an excerpt of the XML file containing my sensor readings:

Here is an excerpt of the TXT file containing my meter readings:

These files require some sophistication to manipulate, but it's easy to import them into Excel or another software program once you get the hang of it.

Dexcom Seven Software: Daily Statistics

Daily statistics gives you just that – lots of descriptive statistics broken down by day. You can display any number and combination of days that you choose. The statistics include a pie chart similar to those in the Glucose Distribution output, and a row of statistics similar to the first column of the Hourly Statistics table.

Here is my Daily Statistics output for 5 days, starting when I initiated my second sensor:

Here is a close-up showing how you can select certain groups of days to view:

Nothing really new to discuss here.

Tuesday, July 17, 2007

Dexcom Seven Software: Glucose distribution graph

The glucose distribution graph presents selected glucose values by time of day and categorized glucose level. Here I have selected one day of readings:

The colors indicate percent of sugars in the low range (blue, default definition is less than 80 mg/dL), percent in the target range (green, default 80 mg/dL - 130 mg/dL), and high (red, default is greater than 130 mg/dL). This target range is at odds with the 80-140 default target range in the hourly statistics graph, but perhaps I am mistaken in one or the other. You can customize the target range.

The pie charts break the readings down by time of day in the following categories: All day totals, early morning, before breakfast, after breakfast, before lunch, after lunch, before dinner, after dinner, and late evening. You can customize the hours for each of these categories. Categories can overlap. You can also create new categories. For instance, I was able to create a category called "Pre-dawn phenomenon" and define it as the hours from 11:00 pm - 4:00 am.

The bar chart on the right shows the percent of total readings for the specified time period by glucose category. Default categories are less than 40-55, 56-70, 71-100, 101-130, and so on. I don't think these categories can be changed.

You can choose to display statistics for any number of consecutive days, or types of days, for instance weekeneds or Mondays. You can even select groups of days, like "Tuesdays and Thursdays".

My computer is back!

Finally, finally, finally. Oh I am so dependent upon this thing. Now that my computer is fixed, I can put up pictures of the other software graphs and the exported data. Stay tuned...

Tuesday, July 10, 2007

Demise of sensor #1, and test strips


Yesterday I think my first sensor died. It was a bad blood sugar day to begin with, which didn't help since Comrade Dex doesn't do as well with the highs. After breakfast I went up to the 250-300 range, and darned if I couldn't get it to come down for what felt like hours. Then there was some up and down action, and finally I wrote the day off as a loss.

I had some accuracy issues with the receiver when my sugar was really high, but that's nothing new and I didn't take it as an indication of sensor death. Later in the day, though, I had some long signal gaps, and then a couple hours where it kept asking me for a calibration finger stick. I probably did 4 calibrations before I gave up. About 30 minutes later I did get some more readings, but then it asked for calibration again so I just ripped the darn thing out.

Two points I want to make about this.

First, this sensor lasted just a few hours past 10 days. I am disappointed, I was hoping to get 14, but there you have it. That comes out to just under $6.00 per day. If it had been a 3-day sensor, it would have been just under $3.50 per day.

Second, when I finally admitted the sensor had probably failed and took it out, it was almost 10:00 pm. There was no way I was going to stay up for 2 hours waiting for the new sensor to warm up, so I slept sans Dex, and put in a new sensor this morning. I guess I could have let the alarm wake me up to calibrate at midnight, but for a variety of reasons I didn't want to do that. I hadn't thought of this wrinkle in letting the sensor go until it fails - if it fails late at night you might not realize it, because I don't think it vibrates or anything when there is a signal gap or when it says it needs calibration. Maybe in the future I can learn to read the warning signs of imminent failure, and avoid nights where Comrade Dex doesn't have my back.


Silly me, I thought I'd use fewer test strips once I started the Dexcom Seven. In fact, I used more test strips than normal in the past 10 days. I hope this is because I am getting to know the system, and I need to test more often to learn when it is and isn't accurate. In the past 11 days I have used 81 test strips. I think I've done 4 today (including the 2 for initial calibration of sensor #2), so that's 77 test strips in the 10 days that I had my first sensor. Seven to eight strips a day isn't unusual for me when I'm exercising, but I don't exercise much these days, and I was using closer to 6 per day before I started the CGMS. I hope as I get to know the Dexcom Seven better, the number of test strips will go down, but I guess it doesn't matter too much. Insurance isn't paying for the Dexcom, but they will pay for test strips. Interestingly, the daily cost of test strips vs. daily cost of a sensor that lasts for 10 days aren't too far apart. Funny that so many insurance companies won't consider CGMS coverage, then. Oh wait, did I say funny? I meant horrible, stupid, cruel, and shortsighted.

High and low alert thresholds

After 11 days I am starting to develop my method. Today I will write about how I use the high and low alerts. I imagine all this is very similar to the 3 day system.


The default setting for the high threshold is 200 mg/dL. I want to avoid reaching 200, not treat myself once I'm there, so I immediately lowered it to 160. You can only work in 20 mg/dL increments for the high threshold. I've had some accuracy issues with high blood sugars in certain situations, and I quickly realized that even 160 was too high. I'll write a separate post about accuracy, but what is working best for me is to set the high threshold as low as it will go (140 mg/dL), and hope the receiver will alert me before my actual blood sugar rises over 200. When the alarm goes off, I decide whether I need a finger stick to confirm the reading and/or calibrate the meter, and whether I need a correction bolus. Then I reset the high threshold to 160 mg/dL and repeat the process until I see the curve start to trend back down. One really great thing about CGMS is you can see when the rising blood sugar starts to plateau, and reverse back down. This really helps avoid stacking correction boluses.

This works fairly well. The accuracy problems I'm experiencing, though, lead to frequent instances in which the Dexcom shows me stable at a reasonable blood sugar level, but a finger stick shows we well above my target range (for instance, yesterday the receiver said I was stable at about 115, but the finger stick put me at 173). I call these "Type 2 errors" (that's statistics Type 2, not diabetes Type 2), i.e. times when the Dexcom fails to alert me to a bad blood sugar. The other type of error, "Type 1", or times when Comrade Dex warns me that I'm out of range when the finger stick shows me at a decent level, hasn't happened in the past 10 days that I can remember. When I can get back on my computer I'll check my records and I'll keep count of Type 1 and Type 2 errors over time. Maybe I'll correlate them with how many days I've been using the sensor. Oh, the games I get to play.


I use the same technique with the low threshold, but with better success because so far the Dexcom Seven is much better at catching lows than highs. In the past 10 days, I haven't had a single serious low blood sugar. I think I dipped into the 40's a couple times, but that was as much due to me experimenting with the system as anything else.

The default low threshold is 80 mg/dL. On tight control, I drop below 80 regularly and it's no big deal, so I usually keep it set at 70 mg/dL. You can work in 10 mg/dL increments with the low threshold. When I get the alarm at 70, I might or might not do a confirmation finger stick, calibrate the meter, and eat or drink something. Then I reset the treshold to 60 and repeat the process. There is also an alert at 55 mg/dL that is hardwired, and the user cannot change it.


The low threshold is working great for me. It has definitely prevented serious hypoglycemia episodes. I give it an A.

The high threshold works pretty well when my sugar is rising quickly, but works very poorly when my sugar is on a slow sustained rise. If this problem is widespread and not just a fluke of my body chemistry, my suggestion to Dexcom is to allow even lower high thresholds. I would really like to set it at 120 mg/dL. I think at that level I could catch most of the highs I want to prevent. For now, I give the high threshold/alert system a C, or even a C-.

Monday, July 9, 2007

Computer problems

Whew! Computer problems mostly solved, but I still can't open or manipulate MS Office documents. I've been using Word to get my screenshots, so unfortunately I can't put up any more software graphs until...tomorrow? I hope?

Sunday, July 8, 2007

Dexcom Seven software: Hourly statistics graph

Hourly statistics give you a deluge of information and some cool looking boxplots. You can customize many details, and it's too confusing to summarize here. Read on.

Here is the hourly statistics window, covering all my blood sugar readings for the past 8+ days:
I don't know how clear the picture is if you click on it, but it is showing data from all days, all hours, all glucose readings. Nothing held back, nothing filtered.

As you can see, I am having some issues in the late afternoon.

This funny chart at the bottom gives lots of statistics broken down by hour of the day. The first row gives the number of 5-minute averages used to calculate the numbers for each hour. The next row gives the average glucose reading for that hour. For those of you who are into the rest of the stats, the table also gives minimum, quartiles (25, 50, 75), maximum, standard deviation, estimated standard deviation (highest and lowest 25% classified as "outliers" and removed), standard error of the mean, and percentage coefficient of variation. I'm not going into what all that means here, but I am a biostatistician, and if you ever want an explanation about a certain type of statistic, contact me and I'm happy to enlighten you.

The cool thing about the hourly statistics is that you have a lot of control over how things are categorized, and then you can display the information according to category. For instance, here are the default definitions of hypoglycemia, low blood sugar, target range, high blood sugar, and hyperglycemia:

These definitions don't quite work for me, so I can customize them to my personal goals:

The rows in blue indicate categories that I edited.

Now I can filter my graph to only show readings that are categorized as hyperglycemia:

You can see some hyperglycemia after breakfast, some after lunch, and most of it after dinner. Now I get to target my dinner boluses to reduce the highs. Maybe in a month I can use this chart to show progress.

Saturday, July 7, 2007

Dexcom Seven Software: Installation and Glucose Trend graph

I hear the software released with the 3-day system really sucks. I have been playing with the 7-day system software this morning. Here are my thoughts in no particular order. I will clean up this post as time goes on.


1. Software takes a long time to install, but it's easy. Also easy to upload. Plug in receiver, plug into USB port on computer. People who don't know much about computers should be able to figure it out.

2. If you have more than one family member using a Dexcom, you can maintain more than one person's data on the software. This is probably geared towards health care professionals (Dexcom sent a copy of the software to my endocrinologist).

Here is the screen that pops up when you launch the software:

3. Easy to upload data. Just select the patient from the menu and click the upload button. It takes a few minutes to upload the information from the receiver. The information is still on the receiver after it transmits to the computer.


There are four graph types generated by the software: Glucose Trend, Hourly Statistics, Glucose Distribution, Daily Statistics.

Glucose Trend graphs are just what you'd think. They show the glucose values from the Dexcom and the One Touch Ultra over time. You can choose the day and time to start the graph, and you can display values for up to 7 days from the start time. Here is my first day of readings:

The blue dots are the 5-minute averages from the Dexcom sensor. The red diamonds are the calibration readings from my One Touch Ultra meter. The yellow bars represent signal gaps when the receiver failed to receive one or more 5-minute averages from the transmitter. The dashed lines are the high and low blood sugar alert thresholds as they were set when I uploaded the data. The green shaded area is my specified target glucose range. I can change this interactively using the software.

The second and third diamonds on the left are my initial calibration finger sticks. The first diamond shows that when I tested I was pretty low, so I treated that and waited before calibrating the meter. The blue circles start a few minutes after the initial calibration.

As you can see from the first day, the calibration finger sticks were pretty close to the Dexcom readings.

Here is a chart showing three days:

Finally, you can select a small segment for a close-up view:

Here you see some problems with gaps and accuracy. I am still learning how best to live with Comrade Dex, but I am getting better. One day soon I'll do a post on accuracy.

Friday, July 6, 2007

Dexcom Seven is waterproof...sort of

"Waterproof" is one of the big selling points of the Dexcom Seven. Apparently with the 3 you have to use an extra adhesive patch every time you get wet. This seems like a pain, mostly because I know I would forget the patch any time I needed it away from home. I am like that.

After seven days of showers, the patch sticks to my skin very well. My insulin pump patch starts to pull away after a few days of showers. The Dexcom patch is still on there as tightly as ever. I haven't noticed any problems with water leaking in between the transmitter and the pod. I usually have the receiver sitting too far away to pick up signal, which means while I am in the shower I don't get readings, but my showers aren't long enough for that to be a problem.

Let's talk about swimming. The Dexcom promotions literature says you can swim. The manual says the site is waterproof to a depth of 3 feet for up to 30 minutes. Now, I guess this is fine if I'm sitting in a wading pool for a brief period, but 3 feet for 30 minutes doesn't sound like waterproof to me. If I want to swim laps for an hour, is it ok at 1.5 feet for 60 minutes? What if I play water polo? I don't, but there are diabetic people out there who do. I don't know, 3 feet for 30 minutes sounds more like water resistant, not waterproof. When I was at a pool last week with my son, I didn't get in because I was afraid of going too deep or being in too long and killing my first sensor. I don't want to run any trial and error testing, because these dang things are so expensive.

My other problem with the waterproof claim is that the manual states very clearly that the receiver is not at all waterproof. Must not moisten the receiver! OK, so what good is the waterproof transmitter when I'm swimming, or just hanging out near a pool or lake while my son plays in the water, if I have to keep the receiver too far away to receive signal? Hanging out at the edge of the pool with my son for two hours, unable to use the Dexcom because water is flying everywhere, really defeats the purpose.

My one-week conclusion is that the Seven is much improved over the original for taking showers, having a water balloon fight, or getting caught in a rainstorm - as long as the receiver stays dry. I think it falls short of the mark for swimming. I'd be too nervous about ruining the sensor, and without the receiver nearby I wouldn't have access to the trends, especially the lows that can come with exercise, which is the whole reason for having this dang thing in the first place. In summary: better, but still room to improve.

Why Dexcom?

Once I decided to buy a CGMS, I had to choose from the available models. There are currently three games in town, although Abbott's Navigator should be coming soon. Besides Dexcom, Minimed has two systems - the Guardian and the Paradigm Real Time system (or whatever it's called). The latter is integrated with Minimed's latest insulin pump. I have an Animas IR1200, and my insurance company is not interested in buying me a new insulin pump yet, so the Paradigm was not an option. I chose semi-arbitrarily between the Guardian and the Dexcom. From my research on the internet, it seemed like each system had pros and cons, some people love one, some people love the other, and you can't really know how a system will work for you until you try it. Dexcom had cheaper startup costs, and the Dexcom rep called me back much more quickly. Some people think I'm silly, but I put a lot of stock in the behavior of sales reps to evaluate a company. That was a big reason I chose the Animas pump over Minimed in the first place. After I met with Tim the Dexcom sales rep, and the product looked good to me, I decided to go for it.

Initially they were going to give the the 3 day system for a week, then upgrade me to the 7, but someone behind the scenes decided to give me the 7 right off the bat. I got trained via a 20 minute phone call, and that was that.

Thursday, July 5, 2007


In these early days of CGMS, these dark times before widespread insurance coverage, people have all sorts of reasons for choosing to fork out thousands of dollars each year for this technology. Besides medical research suggesting that people who use CGMS have improved glucose control, here were my reasons:

1. Unexpected highs, often overnight, which meant they lasted for hours before I caught them. After 11 years I have pretty much figured out my dawn phenomenon, but I'm not as adept at bolusing for high-fat Indian food right before I go to bed.

2. Severe hypoglycemia unawareness. The week before I started my Dexcom, one day I was feeling a little spacy, and when I checked my blood sugar was 28. Huh??? Plus, I almost never wake up low at night, so I can only assume I sleep through the hypoglycemia and that some of my morning high readings are rebounds. I do have my husband sleeping next to me, and I love him dearly, but he is not one of those magical people who can wake up out of a deep sleep because my breathing changes and he senses trouble. I know someone whose husband can do that, the lucky duck, but the love of my life sleeps through just about everything. I needed some mechanical backup.

3. To avoid unexpected nighttime highs I had to wake up at least once each night to test my sugar. I started doing this when I got pregnant in January 2005, which means I have not slept straight through a night for 2 1/2 years.

4. My husband and I are considering another pregnancy, and 15 finger sticks each day, including 2-4 times each night, just isn't feasible now that I have my son running around, sleeping with us at night, and just making it generally impossible to have my second pregnancy be all about me.

5. Now that I have a kid, the potential consequences of severe hypoglycemia unawareness are much more frightening. They were easy to dismiss when it was just me and random hapless pedestrians at risk, but if I don't feel a bad low coming on, and I'm alone with my son...I can't even bear to think about it.

6. Now that I have a kid, the cumulative effects of sustained high blood sugars scare the pants off me.

Let me say more about those last two. Before my son was born, I was vaguely afraid of diabetes complications and I was vaguely afraid of death. I felt that a long life was preferable to a short life, all things considered. Post son-birth, this preference for long life and good health became a crushing need. I must see this little person turn into a big person. I must be there to help launch him into adulthood. I want to be around to help him in the way my parents are still around to help me. I want to drink sangria on the porch with my 50 year-old baby boy. Of course, he might decide as an adult that he wants nothing to do with me, but heck I'd at least like to be around to give him that choice.

I'd liked the idea of a CGMS, but I knew insurance wasn't covering them yet, and I thought I'd give it a few years and see if I could get it paid for. I think it was one morning in May, when I skipped my middle of the night sugar check (still didn't sleep all the way through since my little leech sleeps in bed with us and still nurses), I woke up and my sugar was over 300, and it hit me, really hit me, that these highs and lows are very dangerous. I turned to my husband and said I needed one of those continuous glucose monitor thingies, no matter how much it cost. Other than being an excessively sound sleeper he's a really good guy, and he agreed with me. So I started to research the various options.

Comrade Dex and me

A couple people have asked if I am blogging about my new Dexcom 7 continuous glucose monitor system (CGMS). I have never blogged in my life, but since I am apparently the first person to receive this device commercially (or so the sales rep kept telling me), maybe I’ll give it a shot. I haven’t read other Dexcom blogs, so I’ll probably be redundant to other people with the 3 day system, but I will do my best to be original.

First off, I never used the 3 day system, so I can’t compare the 7 to the 3. I just purchased it a couple weeks ago, and they gave me the 7 right off the bat. I am on my first sensor, and it’s only day 6. At $60 a pop, I will use each sensor for as long as possible.

Starting off the experience was the sales rep meeting. Tim is a very nice guy and gave a good demonstration. He was able to answer my basic questions, but was fuzzy on some of the details. For instance, when giving the spiel about how the higher price of the 7 day sensor ($60 vs. $35 for the 3 day) actually translates into lower per-day costs, I pointed out that since most people get far more than 3 days out of each sensor, that is only true if the 7 day sensor lasts proportionally longer than the 3 day sensor. He seemed thrown by that one. I also asked if the sensor was actually different, or if they just asked the FDA to let them market the original sensor for 7 days. He didn’t know. Some things in the 7 system are different – for instance, the transmitter is redesigned to fit more tightly to the pod, allowing 7 users to shower and swim without using a patch. That doesn’t seem to have anything to do with 7 days, though. He did say it’s more accurate than the 3 day, but I don’t know if that’s a difference in the sensor, or improvements in the transmitter and receiver. Since I am only on day 6 of my first sensor, I don’t know how long the 7’s will last, but my guess is that the per-day cost will be higher than the 3. Since I am a nerd, here is a chart to show what I mean:

For anyone who has been using the 3 day sensor, a rough guide is to take the number of days you usually get per sensor, double it, subtract a few, and that is about how many 7-sensor days you need to get the same per-day cost. If your 3-day sensors usually last 10 days, you have to use the 7 sensors for almost 18 days to break even. If you get 21 days out of the 3, you have to use the 7 for a whopping 36 days before the 7 starts to get cheaper. So yes, the 7 is cheaper than the 3 if you only use it for the specified number of days, but we all know most people use them as long as they last, so it is disingenuous to claim the cheaper costs as a selling point for the 7. Maybe the improvements in the new system will help people not resent paying more for the 7, but heck, let’s be honest about the expense, since most of us are paying for all or part of these systems ourselves.

All this said, I don’t want to come down too hard on these companies. I know a lot of money went into development of Comrade Dex, and I am very grateful to have it. It feels like a miracle to spend time with my toddler and not worry (quite so much) about crashing while I’m taking care of him. Now if only I can figure out how to get it to sense my high blood sugars at night, but that’s another story.