JGIM solicits reviews of new books from its readers. If youwish to review a book, please submit a letter of interest thatidentifies the book in question (title, author, and publisher) toRobert Aronowitz, MD, Book Review Editor, JGIM, Veterans Af-fairs Medical Center (JGIM-111), University and Woodland Av-enues, Philadelphia, PA 19104; telephone (215) 823-4470; fax(215) 823-4450.

JGIM

BOOK REV IEW

721

Rationalizing Medical Work: Decision-Support Techniques

and Medical Practices.

B

Y

M

ARC

B

ERG

, MD, P

H

D. Cambridge,

Mass: MIT Press; 1997. 256 pages. $30.00.

About 40 years ago, against a background of “increasing interest

in the use of electronic computers as an aid to medical diagnostic

processing,” a landmark article by computer scientists Ledley and

Lusted appeared in

Science

analyzing the reasoning processes

inherent in medical diagnosis in computational terms. By 1970,

the growing optimism was evident in William Schwartz’s

New

England Journal of Medicine

article, which predicted that the

computer would “fundamentally alter the role of the physician”

and “profoundly change the nature of medical manpower recruit-

ment.” The computer, Schwartz declared, would take note of the

history, physical findings, and laboratory data and alert the phy-

sician to the most probable diagnosis and the most appropriate

course of action. In 1987, in the same journal, Schwartz con-

ceded that the revolution had not occurred. In 1998, as we ap-

proach the next millenium, the early forecasts of a revolution

seem somehow quaint; predictions about the future of computers

in medical decision making have grown increasingly cautious.

The computer’s absence from this area has become all the more

stunning in light of its ubiquity elsewhere.

The rumbling on the horizon continues, however, and formi-

dable forces are gathering that may eventually push medical

decision-support tools over their tipping point. Changes in the

health care system increasingly demand cost-effective, evidence-

based, standardized, yet patient-specific medical decisions to be

made in real time, and policy makers are looking for technological

solutions that meet these needs and allow also for oversight of re-

source utilization and quality. Can computer-based tools deliver

on their promise?

Rationalizing Medical Work

helps frame some of the most im-

portant issues engendered by this question. In this book the

Dutch physician and cultural historian Marc Berg provides an

analysis of the brief history of medical decision support and its

apparent lack of success in affecting clinical practice. Berg takes

a broader view of the impact of these decision-support tools than

is typically found in medical journals; he examines the mutual

transformations both between the tools and the clinical problems

they attempt to formalize and between the tools and the sociopo-

litical context of the medical workplace.

After a brief history of the medical communities’ evolving

conceptualization of clinical practice as a scientific endeavor, he

analyzes the “discourses” of decision tool advocates and their crit-

ics. On the one hand, the advocates argue such tools are needed to

turn clinical practice into a true science. The critics, on the other

hand, claim the tools attempt to accomplish the impossible—

replace intuitive, skillful “knowing how” with impersonal, objec-

tive “knowing that.” By focusing in closer, he reveals a multiplicity

of subtly diverging views within the decision-support commu-

nity—“the different voices” of decision-support techniques. In

particular, he focuses on the often contentious debate between

the supporters of statistical tools (such as Kassirer and Pauker)

and the advocates of protocols that rely on clinical rather than

statistical logic (represented most eloquently and stridently by

Feinstein). There is a striking symmetry in the critiques and

countercritiques of these two camps: the advocates of clinically

based protocols argue that the statistical tools exclude critical

“soft” data and thus are built on a “decontentualized,” impover-

ished, codified base that sacrifices the rightness of clinical experi-

ence and observation for statistical convenience. The advocates of

statistical tools counter that clinically based protocols are rigid

and simplistic in structure, causing them to break down when

non-routine events occur, so that difficult decisions are left any-

way to experts. This debate illustrates the competing tension be-

tween the demands of encoding and decoding, between the need

to capture all the relevant clinical information and the need to ar-

rive at a clear result that can be applied meaningfully and flexibly

in a complicated clinical environment.

The second half of the book is devoted to a detailed analysis

of the mutual accommodations and assimilations between several

decision-support tools and medical practice, negotiated as the

tools became “localized” in space, scope, and rationale. Berg starts

with a reexamination of medical work as a perpetual exercise in

reconstruction, as an ongoing attempt to make a patient’s case

cohere. Each patient’s complaint must be transformed into a

manageable problem that matches a limited set of actions. By

Berg’s account, this constructivist process is distributed among

heterogeneous elements of a complex system—the physician, the

nurses, consultants, charts, laboratories, etc.; the management

of a patient emerges from the interaction of all these elements in

the system, not by a “top-down” process with the doctor directing

all care as it is conventionally portrayed. In this dynamic, highly

political jungle, the decision-support tool must stake out its terri-

tory, find its niche—a task made difficult if the tool is perceived

as an arrogant intruder in this highly evolved, finely tuned and

poised ecology.

Strikingly absent from this otherwise rich account is any

thoughtful consideration of the central and active role patients

play in their own management, not only by making decisions or

expressing preferences, but also by constructing the history—

what they reveal, distort, conceal—on which diagnosis rests. An

immigrant’s symptoms may be profoundly shaped by explanatory

constructs that are radically different from those of the referent

population. Patients made desperate by their circumstances may

complain of systems produced, volitionally or subconsciously, for

the purposes of securing some comfort, for attention, for rest, for

euphoriants, for sanctuary from abuse, for financial gain. Critical

elements of a history may be too shameful to report accurately.

Except in highly defined settings, history taking requires a skepti-

cal, compassionate, and culturally sensitive ear, an

experienced

ear, that is far beyond the capacity of any silicone-based diagnos-

tic tool. An understanding of the intensely human interaction

722

Book Review

JGIM

between doctor and patient is critical to finding the limited places

where computers can enhance medical decision making.

Although this book is aimed principally at scholars of “tech-

nology studies,” it has insights for anyone interested in the field

of medical decision making. Physicians and biostatisticians may

find Berg’s analysis of “discourses” somewhat fashionable, over-

elaborate, even vague. But he locates the lack of success of the

decision-support tools in these discourses, which often suppress

consideration of human factors and the broader, complex clinical

context in which the tools will need work. It is as if the tool build-

ers have focused their gaze past the messiness of clinical prob-

lems in order to discern their underlying, pristine mathematical

essence, only to have the messiness reemerge and seek its re-

venge. If advocates of decision-support techniques are to deliver

on their promises, they will have to listen closely to thoughtful

critics like Berg.—

D

AVID

K

ENT

, MD, CM

,

Robert Wood Johnson

Clinical Scholar, University of Michigan, Michigan Medical Center,

Ann Arbor.

REFLECTIONS

For V. H., Dying of Cervical Cancerat Cook County Hospital

It happened that I came to draw your bloodjust as transport arrived to take you to radiationand you needed to use the bedpan, so I waited,pleading with a wild-eyed, impatient man for five minutes of his timeso I wouldn’t have to push the cart with the crooked wheels alone.I emptied your pan, washed my hands, knelt, and tied the tourniquet,ignoring the angry, wild eyes as I stroked the veins on your handand prayed for one to rise. I missed—

pierced again—I felt the pain too, as if you were some voodoo doll,but you couldn’t know thatand you rolled your eyes as he took you away from me.

An instant too late, I realized I could have told youthat the sun rose as I walked to the hospitaland a dog barked in the distance as I opened the shaded glass door;that in the elevator, I stood shoulder-to-shoulder with a wisp of a manwhose cough is robbing him of life, pushed past him in my impatienceto find you this morning in a fitful sleep,your cheeks too rosy after the blood transfusion,relaxing your grip on the morphine pump for an hourto remind me you can leave this chaos behind.

M

ARGARET

G

OURLAY

Chicago, Ill.

r