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

RE: [IP] What month were you dx'd?

Sam Skopp [mailto:email @ redacted] wrote:

> I'm just wondering how accurate all this information is? When we were 
> diagnosed and when symptoms first started to appear may be 
> months apart. 
> Even then, i've heard that the destruction of the islet cells 
> had probably 
> been going on for years prior. So, the exact month of 
> diagnosis may not 
> mean a whole lot. It's just when the outward symptoms started to be 
> recognized, not when the disease actually started.
> For more accurate information, one of our medical people may 
> want to chime 
> in here.

How about an epidemiologist/statistician?

 << Professional Mode = ON >>

What you are referring to is incidence or date of onset, something that gets
harder to determine as we age.  For some reason, symptoms follow pretty
quickly in children.  I don't know the mechanism, but for type 1 in
children, apparently the autoimmune process is a rapid one, and many (most?)
children are diagnosed in the ER.  For some reason the process for the
autoimmune reaction in adults is much slower - as long a two or three years.
It's just a guess on my part, but I suspect that those who seem to be type
1.5 or type weird are probably type 1 diabetics during this active immune
response period (honeymoon).

So . . . you are quite correct, Sam, that it is probably inaccurate to
assume that diagnosis date reflects incidence, although it may be pretty
close in children.  It's also important to remember that type 1 and type 2
diabetes are really two different diseases with common symptoms, and so it
really isn't meaningful to talk about diagnosis unless one is also
specifying what the diagnosis is/was.  This is known as case definition and
is one of the first steps in an epidemiological study.

Another aspect which makes this sort of survey probably inaccurate is that
it will be responded to only by those who have some interest in it and are
willing to take the effort to do so under their own initiative.  This is
known as volunteer bias.  Notice that several people have responded with
information, and that there seems to be a trend toward confirming the
hypothesis.  BUT . . . most people who were not diagnosed during the three
months will be less likely to respond.  As more and more people *do*
respond, the association will appear to get stronger and stronger until it
takes on a life of its own, while those who were not diagnosed in those
three months will lose interest (it isn't germane to them), and . . . well,
you get the picture.

In order to have validity, the data must be gathered in a way that is 1)
replicable; 2) representative; and 3) therefore generalizable.  Otherwise
the results can be applied only to those who respond, and no inference can
be made outside that small group of respondents.

All of these are reasons that most studies deal with prevalence (number of
extant cases) than with incidence (onset of disease).  Under some
circumstances, prevalence and incidence will be closely correlated . . . but
not always, and never with certainty.  When it is possible to ascertain,
incidence is usually the preferred factor.

James Handsfield, PhD, MPH
Division of Laboratory Systems
Public Health Practice Program Office
Centers for Disease Control & Prevention
email @ redacted / (770)488-8106
for HELP or to subscribe/unsubscribe, contact: HELP@insulin-pumpers.org
send a DONATION http://www.Insulin-Pumpers.org/donate.shtml