RE: den.list-d Digest V2005 #14

["John McConnell" <wysowl@msn.com.au>]

Malcolm Macpherson wrote,
I see very few opportunities for
us to do that any better by deeper analysis of our data streams.

Is your rate of medication errors stable?  In most hospitals, it is.  If it
is stable at your hospital, what are the implications for management?
If you plot re-admits, are the data stable?  How about the infection rate?
How about the level of admissions caused by medical misadventure?
In most hospitals, 80 to 90% of medical misadventure cases can be traced to
10 to 20% of the medical practitioners.  Is this true for your hospital?
In most hospitals, the death rate in the ER is reasonably stable.  What does
this mean?
If the birth rate is stable, what clues does this give me about staffing
approaches and levels?  Can we use this information to improve the level of
care and still control costs?

Many attempts by well meaning doctors to control blood sugar levels in
diabetics within a narrow band has led to classic over-control (Nelson
Funnel Experiment).  If you would like some example charts for
over-controlled blood sugar, drop me a message.

Do your discharge rate data spike nearly every Monday?  If they do, an
opportunity exists to reduce LOS without reducing the level of care.

How do the error rates for each of the patient care pathways compare?  Does
it matter?  Will such data help us to understand where to focus our efforts
to improve the hospital?

Which haemophilia patients need Factor 8 as a prophylactic and which should
receive Factor 8 only to treat a bleed?  Do our clinical people know how to
do these analyses and to find statistically valid answers to such questions?
About ten years ago when Factor 8 was in short supply, doctors were asked to
administer Factor 8 only when the patient had suffered a bleed.  Data
analysis revealed that in nearly every case where the patient was likely to
suffer a spontaneous bleed, more of this precious material was used when
patients had their prophylactic treatment replaced by treatment “when and as
required”.   Patient wellness declined…long term damage to joints is a real
possibility…and more, not less Factor 8 was needed to treat the same
patients.  These analyses was done by an industrial engineer to convince the
clinicians that they should return his son to prophylactic treatment.  The
analyses took about an hour.

If you really want to determine what data analyses might be helpful, why not
contact somebody whose area of expertise is statistical analysis.  If this
person had a little understanding of health care, so much the better.  Were
your job mine, I would contact Dr. Doug Shaw, of the CSIRO.

Cheerio!

John