Lies, damn lies and statistics

Samuel’s pump comes with a USB interface that allows us to download the data to my laptop. It also includes some software that allows us to analyse that data. My natural interest in numbers means I like analysing Samuel’s pump data. We have been downloading it for the last few weeks, and I have found I like the following three reports the best.

Nikki just said she thought that this post’s headline was a bit strong, but it is an old saying that implies believing statistics can be a bad thing. Or that certainly, they can be spun to prove almost any point. So a bit of caution is always councelled. But on the whole, the pump data will tell the tale of Samuel’s previous couple of weeks control of his diabetes.

Here you can see that he has settled down nicely again after his tonsillitis over Christmas and the New Year. Averages can be a bit misleading, but his mean bG for the last fornight of 6.0 is encouraging. Perhaps more important is that the standard deviation (SD) is 2.3. Online advice tells us that 2 SD’s should be less than his mean bG, if you are looking for good control. If this is a good guide (and I will be checking with the consultant next time we see her), then Sam’s has been well controlled for the last two weeks.

My overall impression before looking at the results was that Sam had been running low for the last couple of weeks. And the bottom graph shows this to be true. I’m reasonably comfortable with this however because Samuel is blessed with very good awareness of when he is going low. It’s quite hard to see from the graph but each black cross is a blood test and he has lots of them bunched together on school days. Because he knows when he is going low however, very few of these lows are below the red line, which is a hypoglycemic event of 2.7 or less.

It would be a lot harder if Samuel was unaware of when he was going low, that’s for sure! And for that, I am very thankful.