[Previous Months][Date Index][Thread Index][Join - Register][Login]
[Message Prev][Message Next][Thread Prev][Thread Next]

Re: [IP] Measure of control?

> In a message dated 05/28/1999 10:25:30 AM Central Daylight Time, 
> email @ redacted writes:
> << Yep, if I wanted to know how to adjust my control, that's what I'd
>  look for.  But the std. dev. and coeff. of var. will tell you if
>  you've got the kind of spread that you mention in your letter above.
>  I just wondered if there wer any other numbers I should be looking at
>  to gauge *overall* control. >>
> You must be an engineer <smile>  to want to do all these calculations.  Since 
> the "numbers" you get with calculations depend upon the tests that you do, 
> unless you are testing 24 times a day, you really don't have a way of 
> determing your overall control with just snapshots of the waking hours.  
> Isn't this what an a1c is for?
> Marilyn


But actually, that's the glory of statistics.  You don't actually have
to measure every member of a population in order to be able to tell
things about the population, you just have to sample.  And, from what
the guy at UC San Diego said (and he could be the janitor, for all I
know, but he comes across as knowing what he's talking about), you can
get a valid sample of the blood sugar "population" by testing 8 times
a day.

And that is what an A1c is for, but there are a couple of problems
with A1c's: the first is just that you may not have an A1c for the
period you're interested an (a month, say), and the second is that I
don't know, and I don't even know if it's known, whether an A1c will
come out normal if you have the sort of bimodal distribution suggested
earlier (e.g., you run 180 half the time and 40 half the time and so
average out well, to 110, but have lousy control).

Dave Breeden                                           email @ redacted
Insulin Pumpers website http://www.insulin-pumpers.org/
for mail subscription assistance, contact: HELP@insulin-pumpers.org