I just ran some reports and it looks like my sensor average for the month was 133 mg/dl. While I remembered things being really great at the end of February, my actual 30 day average for Feb was 132 mg/dl. So, really, with the good and bad all averaged together, things are roughly the same.
My sensor data says that I have been:
26% above 110 mg/dl
73% between 70-110 mg/dl
1% below 70 mg/dl
My standard deviation (SD) indicates how much swinging I'm doing and it shows 34.9, meaning that my range of deviation is about 35 points around my target. This is okay, but when I get things as tight as I want, I hope to see my standard deviation in the 20s. Some people think this number is more important than the sensor average.
General observations from March...
- I was thinking and blogging that daily consumption Yerba Mate tea was behind my best days. I'm not totally sure that it was the sole player now. I have drunk it every day for the last 3 weeks and still had some high-ish and unpredictable days. I think the main player might have actually been insulin sensitivity caused by regular exercise. The week that I blogged about it was right at the end of a very consistent 2 weeks of daily exercise, strength training and sprinting. I think my muscles were metabolizing more glucose, and causing my levels to hang lower. More recently, with less exercise, I've been seeing a need for more insulin and numbers averaging higher. Lesson learned.
- During Spring Break I have found that my AM dawn phenomenon is always focused around the time I drink my 2 cups of strong coffee, regardless of wake up time. I have learned that if I bolus .5 units for it, I can keep my AM numbers pretty flat. Before I figured that out, I was trying to do it with a basal change, but it was too slow to do the trick. Now, my basal still accounts for DP, but I'm having less lows later from it being too aggressive.
- One thing that I think is good... I'm seeing a bit faster response to insulin. When I give a correction bolus I am often seeing change happen within 40 minutes. I also find that when I am closer to the target range, things are more responsive.
- Chicken seems to be the protein that gives the biggest glucose spike. I'm still having a hard time timing my insulin to match different types of protein and their effect on my BG. I have started to use the square bolus and dual wave bolus, but can't say that I have have it worked out perfectly yet.
Caveat: One thing that may have impacted my numbers is that I realized that the box of sensors I was using were expired and I wasn't getting very accurate data from them. I stuck (pun intended) with 2-3 sensors from that box, but found that they would be pretty far off at times. When I switched to a box with a valid expiration date I started getting fairly predictable results again. I had found that both highs and lows were sometimes undetected because of faulty sensors, but, because I didn't trust them, I usually caught them quickly with meter tests.
It takes you 40 minutes to see changes from a correction bolus? Yuck! If I don't see the rise stop within 40 minutes of a correction bolus on the Dexcom, I figure either the Dexcom sensors not working very well, or it's time for another shot.
ReplyDeleteOf course, I usually correct aggressively and just plan to snack if the correction was high.
That standard deviation looks pretty good to me. I really wish that it was clear exactly what kind of deviation mattered in terms of developing diabetes control. Like, is it my lifetime deviation that matters? Quarterly? Daily? Is the issue more that my blood sugar shouldn't be subjected to rapid change, or that it shouldn't vary much?
Like, is a week in which I spent one solid day with a fairly flat blood sugar at 190 and the rest of the days at a flat 100 (haha) worse than a week in which everyday I bounced between 100 and 190 or is it better? If I spent a day with blood sugars drifting slowly up and down in the 100- 160 range, does my body care that this is different from a day in which I zoomed fast between 100 and 160? The standard deviation will be the same; the standard deviation cares only about how far the numbers were from average and not in what order they occurred; but I think my body cares, as evidenced by how awful I feel during the up and down zooms.
The coffee thing sounds like a good thing to have figured out. Think like a pancreas has some suggestions about caffeine and diabetes but I don't remember them because I don't consume any caffeine ever.
How do you get an average that is so much higher than your median? If you actually spent the majority of the time between 70 and 110, your median blood sugar should be between 70 and 110. Your average is 22 points above 110. When a median is far below the mean, the distribution has to have a very dramatic skew in the upper ranges, and while it's no stretch to believe that your numbers have such a skew, I don't think it's that big. I think there's an error in your computations somewhere.
I don't spend much time in the 70-110 range particularly because I consider the lower half of that too low for me. I spent 28% of the time in March and in February in that range, with another 4% below that range and 68% above that range. My standard deviation for March was 52 compared to a standard deviation of 48 in February. I like to look at the standard deviations of my individual days too; averaging that, my average daily standard deviation is 42 for March.
Hi Jonah,
DeleteI like your questions about standard deviation and the speed of changes. My CDE said that fast changes are bad for us, and that we should try to use things like basal settings and slower insulin deployment to make things more smooth. I feel like everything is always slow for me, but I wish I spent less time out of range when something goes awry.
I wonder if my "time spent" percentages might have been set by my alarm limits instead of my correction targets. My high alarm is set for 150 mg/dl, so that would make sense if it says I'm spending most of my time at an average of 133 mg/dl and that I'm "in range" 73% of the time. I think that's about right. I'd like my data to make more sense... but, I guess just knowing my average and standard deviation tells me the most. I'll have to check it closer before I assume that the data spit out by the pump software is making sense.
I like your statistical skills! You use the right terms and everything! I'm a novice, but data is fun to use.