Index Home About Blog
From: (Steve Harris
Subject: Re: Infectious Hospitals?
Date: 25 Jun 2004 12:51:06 -0700
Message-ID: <> (David Wright) wrote in message

> >Once again: The CDC has reported that 80,000 hospital pts die EACH
> >YEAR from iatrogenic infectious diseases. That just HAPPENS to be 80%
> >of the 100,000 reported by the IOM.
> So it does.  Whether that's of any significance whatever is another
> question entirely.


Indeed. The CDC generally reports these things with the more correct
term of "nosocomial" infections. Which suggests the more neutral and
correct idea that they happened from a new infection which appeared
after being admitted to a hospital (nosocomial used to refer to
infection while under any medical care, but these days it means more
specifically to the CDC an infection you came down with, within 3 days
after being admitted to hospital).

"Iatrogenic" is sometimes used as a (bad) synonym for "nosocomial",
but we need to discourage this because "iatrogenic" literally means
"caused by the doctor." It's an appropriate word for something caused
by a doctor's mistake, but it's not an appropriate word for an
infection picked up in a hospital, which may have been due to doctor's
mistake, but also could be due to a nursing mistake or (more usually)
to nobody's mistake.

The majority of people who get ill and/or die from "nosocomial"
infections are badly ill and immunosuppressed people who would get
infected and often die no matter WHERE they were, even a sterile
bubble (since a fair fraction of septic people die from their own GI
flora or mouth flora causing UTIs or pneumonias). These are people
like the very elderly, the pre-term neonate, the patient who has no
bone marrow from chemo, or the guy with AIDS. We've all seen them.

Anybody who's done much traveling will understand that we live in a
sea of bacteria all the time, no matter where we are or what we do.
If you travel, whether you eat the local food and water or not, it
takes about 3 days for your system to start reacting to the strange
bugs of the new land, and if diarrhea is all you get, you're lucky.

Well, it's the same for the immunocompromised. They usually die with
or from the germs in the environment of whereever they are when they
get immunocompromised. If we shipped them all the yelxol's bedroom,
they'd die from the flora there. Instead, however, we tend to send
people to hospitals for treatment, so that when they happen to die of
infections, it's an infection that arose after hospitalization. The
CDC therefore puts them down as "nosocomial infection deaths."

Note please that nobody has to prove that an infection even came from
hospital bugs for the CDC to term it "nosocomial". Although of course
hospital bugs are often involved. But (as noted) the bugs can come
from anywhere. Nosocomial just means the infection appeared newly
within 3 days after admission. It really says nothing about ultimate
causes, which are (of course) extremely complex and multifactorial.

To sum up, *iatrogenic* infections caused by germs on a doctor's
improperly washed hands in a hospital, are a subset of hospital
infections caused by poor hygeine and other correctable problems,
which are a subset of new infections caused by hospital flora for
various reasons (some of which have nothing to do with medical
mistakes and are unavoidable), which are a subset of all infections
(from hospital flora or not) that arise in people admitted to
hospitals within the previous 3 days--- which is all that "nosocomial
infection" means.

I would think that anybody presuming to write a book on this subject
would have mastered these distinctions, but then the world is filled
with fools who want to publish.


From: (Steve Harris
Subject: Re: Infectious Hospitals?
Date: 25 Jun 2004 21:33:50 -0700
Message-ID: <> (yelxol) wrote in message

> 1. General question: Do you find the CDC's report(s) concerning
> iatrogenic cross-infections to be credible?


I know of no CDC report on "iatrogenic cross-infection." The CDC,
through its NNIP system, regularly reports on "nosocomial infection
rates, and mortality and morbidity associated with them, but of course
it's not the same thing. As I explained. It's difficult even to
estimately attributable mortality, which is the deaths caused BY the
infections, rather than the underlying disease whose treatment led to
the infection.

> 2. Specific question: Do you 'disbelieve' (or question) the CDC data
> that reports 80,000 U.S. hospital pts die each year from iatrogenic
> cross-infection?


The CDC reports no such thing. The experts on the subject, some of
whom work for the CDC, sometimes make estimates of attributable
mortality as a result of nosocomial infections, which are infections
aquired by patients while hospitalized (half of them in the ICU).
I've included one such attempt in the link below. Attributable
mortality estimates for nosocomial infections range from 26,000
patients a year to 105,000 patients a year, depending on what
assumptions you make about hospital infection rates, fraction of these
which are blood infections, and fraction of blood infections which
cause death, in cases where the patient wouldn't have died anyway.
Nobody keeps track of *all* these cases, and sampling is all we have
to go on. Nor does anybody know the fractional causation for
infection-associated death, exactly. Many patients die WITH
infections, but how can we know for sure that they died OF infection?

As for sampling, even the CDC NNIP program now only includes 300 or so
of the nation's hospitals, and infection experience varies widely
between them.  Some of this is due to the hospitals, and some is due
to the populations that use them. For example, LDS Hospital in Salt
Lake City has one of the lowest nosocomial infection rates in the US.
But I can tell you from personal experience it's because LDS is a
hospital full of relatively affluent, clean-living whitebread Mormons
with good social support, not because the doctors there wash their
hands there more than elsewhere.

Can you cut nosocomial infection rates in ICUs by increasing
hand-washing (by both doctors and nurses)? Yes, some. One study
suggests you can cut them by about a quarter, with a 25% increase in
washing. This could save up to 1000 lives a year, according to the
authors below. But it's a coincidence that the two numbers are about
the same, and we can't extrapolate to say that (say) 50% increase in
washing will give us 50% decrease in infection, because we don't know
the shape of the dose-response curve for hand washing. Like most
things in life, doubtless it saturates at some point. We KNOW many
infections come from the patient's own flora. So we really don't HAVE
any of the answers you pretend to know. To know what fraction of
infections (and thus infectious deaths) are due to cross
contamination, we'd have to cut cross contamination to zero somewhere
for a large group of patients, and nobody has ever thought of a way to
do so. Or prove it if they did.

Again as explained, a nosocomial infection is simply one that occurs
while the patient is in the hospital. Most of the fatal ones are due
to some medical procedure, like placement of a foley catheter or
central venous catheter. But only a fraction of them are preventable.
A typical nosocomial death might involve a patient in the ICU who dies
from sepsis as a result of skin flora (enterococcus or candida)
invading the bloodstream and gaining a foothold as a result of
presense of a central venous catheter of some kind. But that is not to
say that these infections are necessarily a result of "cross
contamination", let alone cross contamination by doctors (iatrogenic).
The bugs may come from the patient's own skin. Unless you can figure
out a way to sterilize the skin of every patient who has a central
venous cather placed, much of this will continue to happen. Actually,
the best bet for preventing it lies in development of antibiotic
impregnated catheters on which bugs have a harder time growing.

Steve Harris

Index Home About Blog