by Eric Skjei
"existential" or "virtual" entity.

interfaces between their locations

quickly and completely.

should be on access and creating an integrated view

activation barrier to finding that information.

hyperacute skimming and screening mental function

precise bits of information they seek.

relational database and the longitudinal patient record.

"We don't read a discharge summary,
we don't read anybody's five-page history and physical,
10-page discharge summary,
two-page x-ray report or
op note--nobody actually reads those things."
Don Simborg, MD

Robert Beck, MD
'We really don't understand the business of clinical medicine in
the paper era.'

Tim Zinn
'Probably the greatest challenge that faces us in the
computer-based patient record is the different media that the
medical record will consist of...'

by Eric Skjei
The mythical medical record

Second of a two-part series on
automating the clinical medical record.

CAP Today
College of American Pathologists
Pathology/Laboratory Medicine/Laboratory Management
September 1993
Vol. 7 No. 9

"The medical record is a myth,"
declares Don Simborg, MD, CEO of
Bell Atlantic Healthcare Systems.

"I don't believe there is such a thing."

What's more, he adds,
"the whole concept is flawed...there's never been a single
physical entity called 'the medical record,'
and there's no reason to create one now."

Instead, argues Dr. Simborg, what we have always had is a kind of
ad hoc "record" that forms in the brain of the clinician a the
time he or she makes a decision, and it has always been based on
information from multiple sources.

To be sure, some of this information is indeed obtained from
"those pieces of paper called a chart or a medical record."

But much of it also comes from "the patient directly, at any
given time," and more comes from "calling the lab or looking at a
nursing note that is not in the chart or a temperature sheet at
the bedside or a cardex."

continued on page 34

In short, says Dr. Simborg, we must rethink our mental picture of
what we really mean when we talk about the "medical record."

"existential" or "virtual" entity.

Rather than "some kind of computer file sitting somewhere on some
computer," it is instead more accurately thought of as an
"existential" or "virtual" entity.

interfaces between their locations

quickly and completely.

Parts of this record typically reside in several locations and it
becomes a reality only to the extent that these components can be
digitized, interfaces between their locations established, and
their data exchanged across these interfaces quickly and
completely.

should be on access and creating an integrated view

In short, asserts Dr. Simborg, our focus in thinking about the
medical record should be on access and creating an integrated
view of information.

This is clearly a focus that is not incongruent with those
espoused by the Computer-based Patient Record Institute.

"We are hopeful that there will be access across systems,"
says Margret Amatayakul, the CPRI's executive director.

"If I'm in one hospital and my patient has laboratory work done
at a reference lab and was also seen in a couple of different
physician offices, and maybe even a hospital across the country,
I'd like to be able to capture that information, have access to
it."

For Amatayakul, the next questions that arise in this context
pertain to the standardization of medical codes and terminology,
an area that the CPRI is actively pursuing, and to the issue of
trust:
"Does the physician," she asks, "who's reviewing a history done
by somebody else feel comfortable that the questions were asked
in the same manner that he or she would have asked them in order
to elicit certain data?

Is the laboratory that he or she is getting results from the
same laboratory that or she is used to?"

Answering the question of trust is a function of how well we do
in meeting clinician needs, in building a new computer-based
medical record application that makes it possible for the
physician or nurse to deliver care easier, better, and above all,
notes Dr. Simborg, faster.

The paper disaster.
Mythical though it may be, there is no dearth of dissatisfaction
with the medical record in its current form, particularly in
regard to access--or rather, lack of it.

"The paper record is a disaster," says pathologist John Cate, MD
director of laboratory medicine at the
Medical University of South Carolina in Charleston.

"It's a disaster in this institution and in every other
institution I've ever been in."

Dr. Cate also directs and guides the MUSC lab's computer group,
chairs the medical records operation, and maintains an active
pathology practice there.

All too frequently, he says, he finds that laboratory tests on
patients referred to him for consultation simply aren't
available, for any number of reasons:
"There hasn't been time to put them in the chart,
they didn't get delivered to the right place,
the patient's been moved--it's a major problem."

Yes, agrees Robert Beck, MD, "the paper chart is horrible."

Dr Beck is a pathologist and vice president for
information technology at Baylor College of Medicine in Houston.

"Things are often not where you expect them to be,"
says Dr. Beck, "and whenever you really need the record it's
with the patient

continued on page 36

down in the radiology suite, or someone from dietary is
writing a note in it--it's generally not available."

And hsould you be so unlucky as to find that you need old data on
a patient, you may have to "go to some file drawer, hoping that
the record has been brought up from the records room,"
says Dr. Beck, only, as often as not, to come up empty.

activation barrier to finding that information.

The net result, of course, is that frustrated clinicians
"will as often as not just avoid looking in the old record
because of the activation barrier to finding that information."

And the result of that, in turn, is further duplication, waste,
and excess cost, as tests and procedures get repeated ad nauseam
with no clinical justification.

But the larger problem with the paper record is that it is not a
good model for the way physicians actually absorb information.

Even when they can find them, "physicians don't read the
documents that are in the charts," explains Dr. Simborg.

"We don't read a discharge summary,
we don't read anybody's five-page history and physical,
10-page discharge summary,
two-page x-ray report or
op note--nobody actually reads those things."

hyperacute skimming and screening mental function

precise bits of information they seek.

Instead what clinicans do is flip from page to page, searching
for specific and relevant informationusing a kind of hyperacute
skimming and screening mental function that helps them filter out
all the clutter and zero in on the precise bits of information
they seek.

Occasionally, pertinent information leaps off the page.

The clinican stops and reads further, looking for more detail.

But that is the exception.

If you could listen in on it, the mental process would sound
something like this, says Dr. Simborg:
"What was my last note,
what did I say I was going to do?

Oh I was going to order a chest x-ray.

OK, let's see, x-ray normal, seen in ER.

What's the diagnosis?

What meds?"

Our challenge now, in Dr. Simborg's view, is to recreate that
same highly precise, intuitive, selective function
electronically, making it possible for the "virtual" medical
record to be scanned and skimmed, while also continuing to make
available in the backiground all the additional detail needed for
those rare moments when the physician wants to dig further.

For Sunquest CEO Sidney Goldblatt, MD, where we are now is
face-to-face with the fact aht "there really isn't a good model
out there that all of us can follow" in building a computer-based
patient record.

And we are slowly realizing that, without that model, hampered by
the limitations of the paper paradigm, we are likely to make the
mistake tha always seem to accompany any significant paradigm
shift.

Consider the first typewriters, suggests Dr. Goldblatt.

Because they were manufactured by sewing machine makers, they
came equiped with wholly useless treadles.

Because the first cars were built by buggy makers, they too came
with features, like the buckboard, that were curiously
anachronistic.

We too, Dr. Goldblatt feels, run the risk of falling into this
same trap as we try to "define an electronic utility to replace a
very complex system based on a handwritten record," a record
transcribed by a variety of experts, like nurses, pharmacists,
and lab technicians, and "burdened with a large body of rules
from within the profession, from state and federal government
bodies, and from payers."

In the absence of a consensus about that new model,
Drs. Goldblatt and Simborg and others urge us to keep basic realities in mind.

First and foremost, they argue, is the growing viability of the
open computing environment in health care.

Second is the closely associated need for continued attention to
further development of standards and protocols for medical
nomenclature and the exchange of data among
best-of-breed systems, particularly on a regional--
and perhaps ultimately national--basis.

In addition, we must bear in mind the growing presence and power
of the networked PC and workstation, a concomitantly expanding
role for commercial relational databases in the health care
system architecture of the 1990s, and the fact that clinical data
comes in forms other than numbers and letters.

BLOBs of new media.
"Probably the greatest challenge that faces us in the
computer-based patient record is the different media that the
medical record will consist of--voice, data, and image,"
says Tim Zinn, vice president of marketing at Houston-based
Commmunity Health Computing.

"Our data standards have begun to evolve with HL7, IEEE Medix
and others," says Zinn, "but there also has to be a lot more
work done in understanding better standards for these other
media."

As good clinical systems have evolved, he points out, an
inevitable interest has arisen in "tying them all together around
a relational database," and that interest is now being
intensified by an increasingly strong desire to accommodate these
new media, to see them become "media" bases, rather than just
(alphanumeric) "data" bases.

"The key to enterprise-wide access to voice, data, and image,"
he says, "is the relational database or data repository."

But voice and image files impose special demands on information
system, particularly in terms of their larger size, and thus
additional requirements for faster transmission and larger
storage technologies.

To solve some of these issues, CHC is, according to Zinn,
looking to an emergin information technology known as BLOBs
(Binary Large OBjects).

BLOBs are a tool facilitated by object-oriented relational
databases that "allow the system to compress and compact highly
dense objects--like image an dvoice files--and store them for
subsequent retrieval and display just as you would data."

CDC is also, Zinn adds, looking closely at complementary
technologies, including digitized voice entry, hand-held devices
for physical and history-taking, and expert system pixel analysis
and pattern recognition to assist radiologists and physicians in
analyzing radiology and pathology images.

37
Ultimately, he agrees, "the big key to health care sytems today
is not so much where you put the data--although relational
databases will help you put it somewhere--but access to that
data."

The relational database is also a key building block in
Dr. Beck's vision of the clinical information system of the near
future.

In Dr. Beck's view, this system will comprise three components:
"scholarly information, clinical infomration filtered through a
middleware relational database system, and aggregated information
such as health outcomes or regional quality assurance data."

This increasing emphasis on a clinical relational database has
significant implications for pathologists concerned with system
architecture, design, and implementation.

"Anyone who is fighting the battle to preserve lab data inside an
LIS," Dr Beck believes, "is fighting the wrong battle."

relational database and the longitudinal patient record.

The real challenge "is in influencing and helping direct the shape
of the relational database and the longitudinal patient record."

His fellow pathologists, he argues, have a lot to gain from
participating actively in the development of these clinical data
repositories and should "start thinking about repositioning the
LIS as an LDS, a Laboratory Data Server."

Thinking this way, he argues, will mean seeing the laboratory as
"making and packaging data to fire not to a user but to another
system, the clinical data repository, whose 'middleware' then
communicates with the ultimate end user and to which that user
addresses ad hoc queries, for example."

One implication of this senario is that it immediately becomes
less important for the laboratory per se to worry about physician
viewing functions:
"Cum reports are no longer terribly interesting," is the way
Dr. Beck puts it.

Parenthetically, Dr. Beck notes that for all its current
deficiencies, the medical record may ironically turn out to be
the idel application to serve as a prototype for building
clinical systems that are truly flexible an dresponsive across a
wide spectrum of institutional and clinical needs.

After all, he points out, the mediclal record is the one
application that "is acted upon by scores of people with
different levels of training and expertise, different reasons
for wanting to look at various parts of it," and yet it is also
the application that potentially is of greatest value to
clinicians and others actively engaged in delivering and
coordinating care.

First steps.
Embryonic projects are under way at a handful of institutions
that are intended to put these convictions into practice and
shape the medical record of the next decade, if not beyond.

For example, at the Medical University of South Carolina,
Dr. Cate and associate Victor Del Bene, MD, are spearheading one
of the earliest installations of
Bell Atlantic's StatLAN Oacis Gateway++Release 5.0.

As well as chairing the MUSC clinical information systems
steering committee, Dr. Del Bene is an internist and
infectious disease specialist.

According to Dr. Del Bene, Oacis translates these principles into
"smart" summarizations of the several ancillary functions that
can relate to a given patient's medical record.

By smart summarization, Dr. Del Bene explains, he means that
clinicians will not have to see all the data available for that
patient.

In an x-ray, for example, "MUSC radiologists have agreed that the
summary can indicate whether it's normal or abnormal or
unchanged," so the clinican can view that summary and then
decide whether or not to "drill down further and read the first
impressions."

Oacis users at MUSC are also able to review results in real-time
graphical form, customized to their practice patterns and
preferences.

"For instance," explains Dr. Del Bene, "I have many patients with
infections who are on toxic drugs, so I have arranged my
personal screen to show me, among other things, hematocrit,
white blood count, creatinine, potassium, and so forth."

For a specific type of patient, Dr. Del Bene can track these

continued on page 39

critical values "just by looking at one screen."

Moreover, "every time data is added it is automatically graphed
and automatically placed inside or outside a band showing normal
ranges."

In addition, he adds, the system includes an early-warning
device.

"When I call up my roster of patients in the morning,
Oacis tells me whether or not something has been added to any
part of the database, tells me when it was added--minutes or
hours ago, and tells me if it's a critical value...in short,
allows me to scan a lot of data and see when something has been
looked at, when something has been added--tells me right off the
bat if I need to get to it right away, an example being a
critical value, or whether I don't need to go any further down
because the value is normal."

High stakes and an open mind.
The payoff for successful development of a new, more functional
model for medical record access is considerable.

"The practice of medicine is unnecessarily costly," observes
Dr. Simborg.

Medical recordkeeping may account for as much as 25 cents of
every health care dollar spent in this country, far higher than
the cost in other countries for health care systems that deliver
care of generally comparable quality.

Saving billions of dollars would be sufficient justification for
redesigning and upgrading our medical record process.

But the stakes are in fact much higher.

"I also believe that the outcomes of health care
could be much better for less expensive care,"
Dr. Simborg adds.

"We are going to literally remake the care process," says
Sunquest's Dr. Goldblatt.

Much of the way clinicians behave in the management of patients,
Dr. Goldblatt observes, is less exct than it might be precisely
because it is still necessarily focused on the paper record, such
as it is, around unindexed files and loose disciplines in
terminology.

And as that changes, "we're going to see a different level
of rigor in the way medicine is practiced, in the way doctors
and nurses hold each other accountable--we're going to make
health care a great deal more powerful."

"We really don't understand the business of clinical medicine in
the paper era," is Dr. Beck's view.

"Until we can measure and view the proces of clinical medicine
better, we're not going to udnerstand it better, and automation
is part of the solution."

And as we do this, cautions Dr. Goldblatt, it's worth keeping an
open mind about the right solution.

"Ultimately," he says, "we're not going to know what their
benefits really are until we begin to deploy these products."

Vendors can learn a great deal from each other,
"just as we all learn from the user," and
"there's room for a host of approaches to this problem,"
he concludes. [ ]

Eric Skjei is a freelance writer in Stinson Beach, California.