NetNews Usenet Archive 1993 #3/NN_1993_3.iso/spool/sci/med/24382

Internet Message Format | 1993-01-24 | 3.1 KB

Path: sparky!uunet!gatech!pitt!ky3b!ky3b.pgh.pa.us!km
From: [e-mail] (Ken Mitchum)
Newsgroups: sci.med
Subject: Re: CXR, screening, and anecdotes
Message-ID: <[e-mail]>
Date: 24 Jan 93 23:21:02 GMT
References: <[e-mail]>
Sender: [e-mail]
Organization: KY3B - Vax Pittsburgh, PA
Lines: 46

In article <[e-mail]>, [e-mail] (Gordon Rubenfeld) writes:
|>
|>   Just to add to this interesting thread.  It raises a number of very
|> important issues about searching for disease in a healthy person,
|> statistical vs "real" lives, and health economics.
|>   One would traditionally discuss here the ideal nature of a screening
|> test's sensitivity, specificity, and "cost" (financial and otherwise) of
|> the necessary follow up tests when the screening test is positive.  Does
|> the screening test really help people live longer or just find them sooner
|> (lead-time bias)?  Can newer, generally more expensive, technology be
|> recommended as a screening test when it has not been tested that way?  For
|> example, CT scans have been well studied for defining the location and
|> stage of lung cancer, but recommending them as a screening test without
|> data is foolish and VERY expensive. These are all fascinating and
|> technical questions that lend themselves to quantitative responses.

The best screening tests are cheap tests that identify very treatable diseases
and have few false positives. The CBC (blood count) is one such test. Tests
that are cheap but have a lot of false positives can be very expensive in the
long run: the common test for occult blood in the stool ("guaiac") is one
example, because a positive result leads to expensive tests that are often
negative: I saw one estimate that it cost several hundred thousand dollars
in negative GI tests to discover one GI malignancy this way. The hooker, though,
is that this cost is spread out over a lot of patients, and the guy who discovers
he has cancer at an early stage that can lead to curative surgery benefits.

Some cheap tests identify things we don't know what to do with, such as
serum calcium tests, which lead us to identify people with primary hyperpara-
thryoidism many years before they have symptoms. What to do with these people
is not clear at all.

Somtimes "new" technology leads to identification of things we don't know what
to do with as well. When we started doing CT scans of the abdomen we started
seeing a lot of what looked like adrenal tumors in people with absolutely no
symptoms or findings to suggest this (CT was done for something totally unrelated
to adrenals). What do we do in these cases?

|>   Anecdotes should not be ignored.  They should be studied, and when
|> disproved, then ignored.  Anecdotes prey on the insecurity engendered by
|> incomplete data and the tails of the gaussian distribution.

Anecdotes are just that, and most physicians accept them in perspective. However,
personal experience is always going to affect one's perspective, even when one
knows better. That is why no two physicians think exactly the same.

-km