Δαφνη Ευτυχια Αρθουρος

[This s the entry that I was going to have posted a little after 16:00 yesterday, but got rather dramatically interrupted while I was typing the footnotes.]

Urgh. Really not feeling well. Did finally get a decent amount of sleep in one day [Thursday night and Friday morning], but at inconvenient times (of course *pout*) and broken up into smaller chunks. Anyhow, I am currently feeling a lot of pain in various places and having trouble with stairs [even more so now], but this [was] the most awake I'[d] felt all week. It [felt] like I almost [had] my own brain again.

One of the sleep-chunks ended with a dream about calculus. About having a calculus test in the morning that I wasn't prepared for and was worried about oversleeping and missing. Except that evert few minutes of the dream, it switched whether I was student or teacher, worried about taking the test, or about being there to give the test. And I was in a different house than mine, but the roof leaked there too.

A while back an airline lost my mother's glucometer and her insurance gave her a different brand when they replaced it, so she had a lot of no-longer-the-right-kind of test strips which are the right kind for one of my meters. Thus, at the moment, I've got more strips than I need, some of which have expiration dates a decent ways into the future, some expiring soonish, and many past their expiration dates. I've been wondering just how bad the 'expired' strips are. Sometimes when getting to the end of a vial of still-current test strips I'll crack open a vial of expired ones and compare results on the same drop of blood, and if they're close, I'll go ahead and use that batch of old strips next. So far I've only once gotten a result far enough from the result on a not-expired strip to convince me to set that expired batch aside.

But then I started wondering how repeatable the results are anyhow, and how far apart the numbers have to be to indicate that one of the strips isn't right. Alas, probability & statistics is my weakest topic in math, so I don't have an intuitive grasp of what the numbers I can find on the web mean, but with this huge surplus of test strips at the moment I could do my own experiment to get a rough idea with my own live data.

Due to side effects of the "give away the razors, charge for the razor blads" business model of the glucometer test strip companies, and how that interacts with insurance, free clinics, etc., I have four meters (two are the same model). One of those (the easiest to use[*], naturally) I have no test strips for. The other three are a pair of Accucheck Aviva and a OneTouch Ultra2 -- the OneTouch is what I have the big surplus of strips for. The strips I have for the Avivas are from the same manufacturing batch -- same expiration date, same calibration code. The OneTouch strips are from several different batches. Shortly after waking, before taking any of my meds or eating anything, I ran several tests with the Ultra2 using strips from three different batches. Twenty minutes later, when I felt like going downstairs, I tried with the Aviva meters (I still hadn't taken any drugs or eaten anything, but I did have a couple ounces of water ... and just being awake a little longer may have had some effect (though looking at the data, I doubt it)).

Notes:

1. Container opened for the first time today
2. Container opened for the first time 2-3 weeks ago
3. Container opened for the first time ~2 months ago
4. Container opened for the first time ~2 months ago but not opened as often as the container in note 3

Observations: I'd expected results using strips from the same container to vary by a few points, and I had wondered whether the length of time the drop of blood was outside my skin would matter (I was not committed enough to this experiment to punch ten separate holes in myself), but I found the jump from 121 to 131 on successive tests a couple seconds apart interesting, and it's also interesting to see that the largest variation was between two strips from the same batch, not between a current strip and an expired one.

I do realize that to really understand how reliable these meters are and how much to trust test strips past their expiration dates, I should be testing at several different blood sugar levels[**], or perhaps testing with manufacturer-supplied calibration solutions[***]. It's possible that at higher glucose concentrations the differences between readings from current and expired test strips will be more pronounced; from this mini-experiment I can't tell that one way or t'other, only that they're close to each other in the Glenn-is-hungry range. Also, I only did "does the same batch give me the same numbers" tests on current strips, so it may be that the expired ones vary more from test to test than the current ones (maybe I'll do that part of the experiment later .. unless I get distracted by some other idea). This isn't SCIENCE science; this is informally-check-your-assumptions science. (It's still science, just not really good science.)

Now this is, of course, too small a sample to draw really meaningful conclusions from, but it does suggest a couple of things for me to keep in mind when I try to make sense of how my body reacts to different foods, or wonder why my morning reading is a little higher than usual. Mostly that "a little [higher|lower] than usual" could just be expected fluctuations in the instrument rather than meaningful differences in my blood sugar level. I'm not at all surprised at the performance of the expired strips, as my guess was that the expiration dates are conservative limits on how long the manufacturer is sure the strips will be good for, and any one vial of strips may or may not still be reliable for quite a while longer. (So I'll still need to compare results from expired strips I'm thinking of using to not-expired strips to determine whether that vial is still useable or not.) The second test with a strip from that batch with calibration code 17 was from the vial I mentioned earlier as having set aside because it produced a measurement out of line with a current strip a couple of weeks ago. It looks like maybe those strips don't need to be tossed out after all.

Anyone is, as usual, welcome to comment on this, but folks with greater Statistics-fu than I have are especially invited to do so. As are folks who actually know stuff about these meters, how they work, and the published performance stats.

[*] If anybody's wondering: the easiest to use glucometer that I've seen so far (not that I've seen all that many so far), is the Bayer Ascencia Contour. It also works with a particularly small drop of blood, though I think the Accucheck Aviva may have it beat slightly in that regard (both take noticeably less blood than other meters I've used).

[**] So far, in myself, I've seen blood glucose levels ranging from 59 to 190 mg/dL, with readings above 160 being fairly rare. So I should repeat this experiment some day when my sugar is up in the 160-180 range, and then talk someone else who routinely gets higher numbers into doing the same experiment above 200, just to see whether the results diverge more when the numbers are higher. Or I could just get distracted by some other project...

[***] Ever notice that the meters' instruction booklets all give guidelines for how often you should check the meter's calibration with the test fluid (which itself becomes unreliable, according to the manufacturers, three months after you first open the bottle), but insurance plans that supply test strips and lancents don't include calibration fluid? (And some drug stores have to special-order it?) I'm seeing a disconnect here.