The Adoption and Diffusion of CT and MRI in the United States: A Comparative Analysis

Author(s): Alan L. Hillman and J. Sanford Schwartz

Source: Medical Care, Vol. 23, No. 11 (Nov., 1985), pp. 1283-1294
Published by: Lippincott Williams & Wilkins
Stable URL: http://www.jstor.org/stable/3765051
Accessed: 04-01-2016 11:31 UTC

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at http://www.jstor.org/page/
info/about/policies/terms.jsp
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content
in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship.
For more information about JSTOR, please contact support@jstor.org.

Lippincott Williams & Wilkins is collaborating with JSTOR to digitize, preserve and extend access to Medical Care.

http://www.jstor.org

This content downloaded from 152.118.24.10 on Mon, 04 Jan 2016 11:31:13 UTC
All use subject to JSTOR Terms and Conditions

CARE
MEDICAL
November 1985, Vol. 23, No. 11

The Adoption and Diffusion of CT and MRIin the United States

A ComparativeAnalysis
ALANL. HILLMAN,
MD, ANDJ. SANFORDSCHWARTZ,MD

This study examines and compares the rates and patterns of diffusion of
computerized tomography (CT) and magnetic resonance imaging (MRI)over
the first 4 years of their availability. Although early diffusion of CT was more
rapid than that of MRI, adoption of MRI in nonhospital settings equaled that
of CT. Analysis of attributesof the technologiesand attributesof the regulatory,
reimbursement,and market environments surrounding the early diffusion of
these technologies provides insight into their different diffusion patterns. In
particular,the technical and financial uncertainties surrounding MRI have inhibited its diffusion compared with that of CT. Medicare's DRG-based prospective reimbursement system and certificate-of-need (CON) regulation by
states have reduced overall MRIdiffusion and stimulated purchases of MRIby
nonhospital organizations.The FDA's premarketapproval(PMA)programhas
changed marketing strategies and influenced the diffusion of MRI to a lesser
degree. This analysis identifies problems in how the present health care system
evaluates and adopts new, expensive, diagnostic technologies and suggests
changes to make the system more responsive to present needs. Key words:technology; diffusion; CT;MRI. (Med Care 1985,23:1283-1294)

The increasing intensity of medical technology is one of the primary factors contributing to the burgeoning cost of health care
in the United States.1 One half of the annual
increase in the cost of a hospital day is due
to rising inputs of technologies and services.2
Unfortunately, there is evidence that the
adoption and diffusion of much medical
technology may not be optimal from either
a scientific or a social perspective.3 As medFrom the Section of General Medicine, Department
of Medicine, and the Leonard Davis Institute of Health
Economics, University of Pennsylvania.
Dr. Hillman is a Veterans Administration Fellow of
the Robert Wood Johnson Foundation Clinical Scholars
Program.
Address correspondence to: Alan Hillman, MD, RWJF
Clinical Scholars Program, 2L NEB School of Medicine/
S2, University of Pennsylvania, Philadelphia, PA 19104.

ical costs continue to rise and to account for

an increasing share of an already severely
constrained federal budget, system efficiency
becomes essential to forestall more severe
rationing of medical resources.4-6 Understanding the factors that influence the diffusion of medical innovation and examining
the impact of past health policy on that diffusion are prerequisites for developing public
policy that encourages more thoughtful
technology evaluation and adoption. Such
insight also can help facilitate the efficient
allocation of health care resources in the future.
The advent of two similar medical technologies within the past 12 years-computed
tomography (CT) and magnetic resonance
imaging (MRI)-offers policy analysts a
unique opportunity to compare the impact
1283

This content downloaded from 152.118.24.10 on Mon, 04 Jan 2016 11:31:13 UTC
All use subject to JSTOR Terms and Conditions

HILLMAN AND SCHWARTZ

of the different environments that surrounded their introduction. This study examines and compares the early diffusion
patterns of these technologies. Differences
in their patterns of diffusion are examined
in relation to the attributes of the technologies and the attributes of the environments
that surrounded their emergence. Although
there are important differences between
these two imaging devices that contribute to
their divergent patterns of diffusion, their
similarities permit insight into the impact of
specific policy initiatives that created the
unique regulatory, reimbursement, and
market environments surrounding each
technology. This analysis suggests areas to
be addressed by future policy concerning the
diffusion of medical technology.
Methods
Data regarding the diffusion of CT were
obtained from case studies published by the
U.S. Congressional Office of Technology
Assessment (OTA)7'8and from studies of CT
diffusion published by Baker9 and Banta.10
Data on the diffusion of MRI were obtained
from three sources between December 1984
and May 1985: (1) the February 1985 Magnetic Resonance Site Survey conducted by
the American College of Radiology (ACR);1
(2) telephone interviews with the marketing
departments of all MRI manufacturers that
are marketed in the United States; and (3)
telephone interviews with each U.S. MRI installation. These sources enabled us to compile a registry of MRI units that were operating or in the process of being installed by
December 31, 1984. For each MRI unit we
determined: (1) the status of its installation
and operation; (2) the type and strength of
the magnet; (3) the unit's manufacturer; (4)
the site of the unit (hospital-based versus
free-standing); (5) the academic affiliation of
the hospital-based units; and (6) the ownership status of each unit. Analysis of the
first five factors is reported in this article.

MEDICALCARE

Since the installation time for MRI appears

to be longer than for CT, we chose the conservative approach of including in this analysis MRI units that were still being installed,
even though the data available for early CT
diffusion included only fully operating units.
Units located in manufacturers' headquarters
were not counted. The unit of analysis was
the MRI unit. Thus sites with multiple MRI
units were counted more than once. Hospital-based units were defined as lying
within a hospital complex and having formal
organizational ties with it. Academic centers
were defined as hospitals having a primary
affiliation with a medical school or their own
residency training program in diagnostic
radiology. Initial availability of CT and
MRI was defined as the month in which the
first clinical human imaging prototype of
each technology was installed in the United
States Uune 1973 for CT and December 1980
for MRI).8'12
In addition, we collected data on the
number of units ordered and expected to be
ordered in 1985. Most of the manufacturers
provided estimates of these data, usually in
the form of ranges of expected sales (several
manufacturers were reluctant to offer such
information for competitive reasons). These
estimates were used to develop "optimistic"
and "pessimistic" predictions of MRI diffusion in 1985.
Results
Figure 1 compares the diffusion rates of
CT and MRI over their respective first 4 years
of clinical availability. The rate of diffusion
of MRI initially lagged well behind the pace
set by CT. At the end of the first 4 years of
CT availability (June 1973-May 1977), more
than 400 units were installed, and the installation rate was accelerating. In fact, the
rate of diffusion of CT between 1975 and
1978 actually is somewhat conservative, because manufacturers were unable to keep
pace with demand during that time period

1284
This content downloaded from 152.118.24.10 on Mon, 04 Jan 2016 11:31:13 UTC
All use subject to JSTOR Terms and Conditions

Vol. 23, No. 11

THE DIFFUSION OF CT AND MRI

(Fig. 1). For example, in 1975 manufacturers

had twice as many orders for CT units as
could be filled.8
In contrast, we found only 151 MRI units
that were either partially or completely installed during the first 4 years of the technology's availability (December 1980-December 1984). Of these 151 units, 102 were
operating by the end of 1984, 28 were in the
late phases of installation (site complete and/
or magnet in place), and 21 were in earlier
phases of installation. Only two units were
dismantled between December 1980 and
December 31, 1984. Furthermore, unless
manufacturers' most optimistic projections
come true for 1985, the rate of MRI diffusion
will continue to fall behind the pace set by
CT. Thus, although the early diffusion of

71
(MAY'80)

1300-

CT
1200.

1100.
1042

1000951
900-

800

2
700

0
8)

600

E
Z

500
475/
/ 400'OPTIMISTIC'

400

MRI.
*325

300.

200.

/
..'

202/

AVERAGE

.250'PESSIMISTIC'

100.
442

45/

- - -I
IOA47
/11
'73
I 1....n
1/0U

1981

10_
1974
I

11975

1976

'

1977

1978

1 1979

1980

'

I
1982

1983

1984

1985

YEAR

FIG. 1. The diffusion of CT and MRI since the introduction of the first clinical human imaging prototype
in the United States (CT, June 1973; MRI, December
1980). The CT curve refers to the x-axis labelled 6/73;
the MRI curve refers to the x-axis labelled 12/80. CT
data from Banta?1and OTA.7

400-

In
cl
C

0)
.0

E
z

300

325

(81%)

200

100
79
(52%)

(76
(19%)

72
(48%)
I

HOSP

AMB

CT at 4 Years
(May'77)

HOSP

AMB

MRI at 4 Years
(Dec '84)

FIG.2. Comparison of the number and percentage

of CT and MRI units by the type of organization purchasing the unit (hospital versus free-standing ambulatory organization) at the end of the first 4 years of
clinical availability for each technology. CT data from
OTA.8

both technologies followed the pattern of the

early portion of a sigmoid curve typical of
the diffusion of many new medical innovations,7 the slope of ascent of MRI was less
than that of CT.
There was a striking difference in the rates
of early purchase and installation of CT and
MRI units among health care organizations
and settings. Whereas only 19% of CT units
installed in the first 4 years of its availability
were located outside of a hospital,8 48% of
MRI units were owned by free-standing imaging organizations (FIOs) (Fig. 2). While the
acquisition of hospital-based MRI units
lagged far behind the purchase of CT units,
the number of FIO-based MRI units approximately equaled the number of outpatientbased CT units at comparable points in time
relative to their introduction. In 1984, FIObased MRI units accounted for 57% of all
purchases, a major increase over the 25%
placed in these ambulatory settings the year
before. Further, 85% of hospital-based MRI
units were purchased by academic centers.
Few have been purchased by community
hospitals.
While the rate and pattern of diffusion differed strikingly between CT and MRI, both
1285

This content downloaded from 152.118.24.10 on Mon, 04 Jan 2016 11:31:13 UTC
All use subject to JSTOR Terms and Conditions

HILLMAN AND SCHWARTZ

MEDICAL CARE

1000-

TAL

/90
800700

:
o
,

/
/4
1

0
600-

500'

400

30
200'

BODY

/
-

EAD

*/1984.

100
.sP
73

74

75

76

, .

77

78

Year
FIG.3. Diffusionof body, head, and total CT scan-

ners by year. Data from Banta.'o

imaging technologies underwenIt similar

changes in the distribution of type?s of units
installed. Initially, all CT units we !relimited
to imaging of the head. However, within 4

TOTAL

150-

125

SUPERCONDUCTING

f0

100

75

0.

E
z

50'

rRESISTIVE
PERMANENT

81

82

83

84

Year
FIG. 4. Diffusion of MRI units, total, and type of
magnet, by year.

years, body scanners had overtaken head

scanners in volume (Fig. 3). While MRI has
been able to image the complete body almost
from its clinical introduction, there was a
marked change in the diffusion rate of the
different types of magnets sold over the first
years of MRI's existence (Fig. 4). The earliest MRI units were resistive magnets. While
the diffusion rate of this type of unit has re-

mained unchanged since 1981, its market

share had dropped to 16% at the end of
During this same time period, permanent magnets captured an 8% and growing market share. In contrast, the diffusion

rateof superconductingmagnetsaccelerated
sharply in mid1983. This type of unit accounted
for 76%
of all
all units
units installed
installed as
as of
of
76% of
counted for
December 31, 1984 and 86% of all magnets
installed during 1984.
Discussion
New, expensive, equipment-embodied,
diagnostic imaging technolgies such as MRI
and CT often are adopted rapidly in medicine.13-15 Medical students, trainees, and
practitioners are socialized to believe that
increased specialization and additional technology are desirable.16"8This orientation has
been reinforced by pressures to practice de-

fensive medicine.19
Rapid adoption of expensive innovations
such as MRI also stems from institutional
factors.20'21Large hospitals, teaching and research institutions, hospitals with highly
trained physician staffs, and urban hospitals
each are associated with early adoption of
new technologies,2'22-24 possibly because
they have greater resources with which to
meet capital and operating expense requirements.8 The prestige and high visibility of

new technologyis especiallyattractiveto ac-

ademic institutions that see themselves as

health care leaders. These factors, however,
do not explain the observed differences in
the diffusion of CT and MRI. To understand
these differences, we must examine charac-

1286
This content downloaded from 152.118.24.10 on Mon, 04 Jan 2016 11:31:13 UTC
All use subject to JSTOR Terms and Conditions

Vol. 23, No. 11

THE DIFFUSION OF CT AND MRI

teristics of the technologies themselves and

analyze the environments surrounding their
emergence. The diffusion of innovative
medical technologies such as CT and MRI is
a complex process, influenced by a variety
of factors related to the technologies themselves and to the environments from which
they emerge. While insufficient empirical
data exist to permit precise determination of
the impact of individual factors on the diffusion of CT and MRI, inferences can be
made about the probable impact of the central factors surrounding the diffusion of CT
and MRI. These factors are summarized in
Tables 1 and 2.
Attributes of the Technology
Attributes of a technology are important
determinants of its diffusion.25-28 CT and
MRI share many technologic attributes that
tend to stimulate the adoption process. When
introduced, each promised improved diagnostic capability and increased safety compared with existing technologies. Although
expensive, both CT and MRI are integrated
easily into a hospital's organizational structure, an important determinant of technology
adoption.27'28 However,

there are important

differences in the technologic attributes of

these two imaging modalities that may contribute to the lagging diffusion of MRI compared with CT.
MRI is a revolutionary development in diagnostic imaging that, in its present clinical
form, uses magnetic fields to image the density and distribution of the body's hydrogen
nuclei.2931Its advantages over other imaging
modalities include absence of radiation, lack
of required contrast injection, and minimal
patient discomfort. Although current applications are limited to imaging, other important potential uses are being explored.32
These factors should act to stimulate MRI's
diffusion. However, adoption of MRI may
be suffering because MRI has been introduced into an environment already replete

TABLE1.

Factors Affecting the Early Diffusion

of CT and MRI
Impact on Diffusion of

Factor
Attributes of the
technology
Technical uncertainty
Marginal clinical
advantage
High cost
Perceived profitability
Attributes of the
environment
Reimbursement
policy
Regulatory
CON
PMA
Market competition

CT

MRI

111

tTT
I
TT

T
11
11

TT(cost-based)

11(DRGs)

0
NA
0

1
0
?

For each factor, its assessed impact on the diffusion

of CT and MRI is indicated by the number and orientation of the arrows. Upgoing arrows indicate factors
that stimulate diffusion; downgoing arrows indicate
factors that inhibit diffusion. The number of arrows (one
to three) indicates the strength of the influence.
0, indicates neutral factors; ?, unknown effect; NA,
not applicable.

with CT capability. Whereas MRI might have

represented a more distinct marginal advance had it been introduced before CT, its
marginal efficacy over that of CT and other
diagnostic modalities has not been convincingly established,3334 with the possible exception of selected central nervous system
problems.35
MRI also is hampered by more technologic
uncertainty than was CT at a similar stage
of development. There is great uncertainty
over the relative advantages and disadvan-

TABLE2.

Incentives Toward Nonhospital

Placement of MRI

1. Regulatory initiatives limited to inpatients (e.g.,

CON, DRGs)
2. Federal Tax Code
3. "Megacorporate" medicine
4. Technology limited to stable, cooperative patients

1287
This content downloaded from 152.118.24.10 on Mon, 04 Jan 2016 11:31:13 UTC
All use subject to JSTOR Terms and Conditions

HILLMAN AND SCHWARTZ

tages of the different types and sizes of MRI

magnets, each of which, in turn, is associated
with widely varying clinical claims, purchase
prices, and operating costs. Moreover, small
changes in MRI pulsing sequences can result
in significant alterations in image appearance, and optimal pulsing sequences for organs, diseases, or symptoms are not yet defined. Movement artifact limits the application of MRI at present to potentially
important organs, such as the lungs and the
heart. These technical uncertainties, coupled
with experience stemming from the rapid
obsolescence of early CT units,9'36is delaying
MRI acquisition until experience with the
technology accumulates, technical standards
are agreed upon, the pace of technologic
change stabilizes,37 and aspects of the environment, such as reimbursement, are
sorted out.
Cost differences between MRI and CT
units also help explain why the adoption of
MRI is lagging behind that of CT. When
first introduced, CT head scanners cost
and body scanners
$300,000-$400,000
$400,000-$500,000 per unit.8 Siting costs
were not excessive. In contrast, the purchase
price for MRI equipment ranges from $1
million for a 0.15-tesla resistive magnet to
$2 million for a 1.5-tesla superconducting
magnet (tesla refers to the strength of the
magnet).38 Site preparation costs can add up
to $300,000-$600,000.38 Thus total capital
expenses for a MRI unit can range from $1.3
to $2.6 million, depending on the specifications of the particularunit selected. Estimates
of annual operating expenses differ by up to
hundreds of thousands of dollars in various
analyses, depending on the assumptions
made. However, two conclusions are apparent: (1) resistive and permanent magnets are
roughly equivalent in cost and substantially
less expensive to purchase, site, and operate
than superconducting systems; and (2) all of
the systems are very expensive-approximately 50-100% more expensive in real
dollars than were early CT scanning systems.

MEDICALCARE

Profitability is another important factor

affecting the decision to adopt a new technology.25'26While innovative new technologies such as CT and MRI always pose some
concerns regarding profitability, the costbased retrospective reimbursement system in
place when CT entered the market permitted
most institutions with CT scanners to achieve
a profit from their units early in the diffusion
process.8'0'39'40In contrast, the potential
profitability of MRI is unclear because of the
uncertainty surrounding third-party reimbursement policy (see below) and the technology itself. Profitability of MRI systems
will be extremely sensitive to the volume of
patient throughput.38'41'42Technologic improvements in hardware and software may
increase throughput and ultimately may reduce the charges necessary to achieve breakeven performance.38'41'42Most important,
prospective reimbursement will reduce
MRI's profitability by limiting the ability to
recover capital and operating costs.
Environmental Factors
The environments in which CT and MRI
emerged were similar in two important respects: both arose in periods of concern about
the high level of medical care costs and both
were marketed before their efficacy was established adequately.l0'2 However, other
components of their environments differed
in important ways that help to explain the
observed patterns of diffusion.
Reimbursement Policy
Reimbursement policy is a major determinant of profitability that, in turn, is an important determinant of the rate of technology
diffusion.43'44CT and MRI entered the marketplace under very different third-party
reimbursement systems. CT developed at a
time when the reimbursement system for
hospital services was determined retrospectively, based on the costs or charges for per-

1288
This content downloaded from 152.118.24.10 on Mon, 04 Jan 2016 11:31:13 UTC
All use subject to JSTOR Terms and Conditions

Vol. 23, No. 11

THE DIFFUSION OF CT AND MRI

formed services. Such a system has been

shown to stimulate the diffusion of new
technologies, including that of CT.7 Even
though CT represented the first medical
technology for which Medicare reimbursement was withheld pending demonstration
of clinical efficacy, this policy did not overcome the significant stimulus to diffusion
generated by cost-based reimbursement.7
In contrast, MRI arose in an environment
in which reimbursement for hospital services
is dominated by Medicare's diagnosis-related
group (DRG)-based prospective payment
system. Under DRGs, the payment a hospital
receives for a patient's care is determined
mainly by the patient's diagnosis. It is not
influenced by whether or not specific diagnostic or therapeutic services are provided.
This payment mechanism alters the incentives for acquiring and using new technolthus shifting technologies from
ogies,25'45-52
being revenue producers to cost producers.
Concern that the DRG system will be extended to include outpatient care, physicians'
fees, and other payors extends its influence
beyond the inpatient Medicare setting. Many
private insurers have adopted a conservative
"wait-and-see" approach with respect to
reimbursement for MRI. The DRG-based
prospective reimbursement system affects
the adoption and diffusion of a new medical
technology such as MRI in two ways: (1) by
its capital cost reimbursement policy and (2)
by the degree and frequency with which
rates for specific DRG categories in which
the costs of care are influenced strongly by
the technology are adjusted.45The as yet undetermined DRG policy on hospital reimbursement for capital costs especially inhibits
the rate of adoption and diffusion of MRI by
generating significant financial uncertainty.
While these factors have slowed the diffusion of MRI, specific attributes of the system alter its impact on adoption decisions.25
The additional reimbursement teaching
hospitals have received for the costs of student and trainee education may reduce the

impact of DRG reimbursement on these

hospitals and may explain, in part, why most
MRI units purchased by hospitals have been
placed in these institutions. Also, the conditions under which many of these university
hospitals purchased their systems (discounts
by manufacturers and manufacturer rebates
for research and consultation services) reduced their financial risk. Likewise, since
DRGs do not at present cover ambulatory
services, the trend toward ambulatory siting
of MRI units in FIOs is a predictable, though
unintended, effect of DRG reimbursement.
MRI is particularly well-suited to such "unbundling" since it is noninvasive, is most
appropriate for stable, cooperative patients,
and is contraindicated in patients who are
unable to lie still or who require metallic
monitoring or life-support systems (Table 2).
Siting of medical technology outside of hospitals shifts costs and risks from hospitals to
radiologists and other investors who continue to receive cost-based reimbursement.
Since hospitals receive the same level of DRG
reimbursement for inpatient admissions regardless of where the test is done, hospital
costs might be lower, but total Medicare costs
(inpatient plus outpatient) might be higher.
Although Medicare currently does not reimburse for outpatient MRI studies, several
private insurers have started to reimburse on
a limited basis. The expectation that others,
including Medicare and Blue Shield, will
follow suit in the future enhances the desirability of siting MRI outside of hospitals. Of
course, the advent of DRGs for outpatient
services would dilute this incentive.
The Federal Tax Code is another financial
incentive stimulating the trend toward nonhospital, entrepreneurial purchases of MRI
scanners. Investors can obtain a large investment credit, rapid depreciation, and
other tax benefits by leasing these units to
nonprofit hospitals. Under such tax-oriented
leasing, hospitals derive the benefit of lower
lease payments and avoid the need to obtain
certificate-of-need (CON) approval, while
1289

This content downloaded from 152.118.24.10 on Mon, 04 Jan 2016 11:31:13 UTC
All use subject to JSTOR Terms and Conditions

MEDICAL
CARE

HILLMAN AND SCHWARTZ

still being able to offer their patients the desired service.52

Regulation
A second major environmental influence
that helps to explain the different patterns
of diffusion of CT and MRI is governmental
regulation. Certificate-of-need (CON) programs to review hospitals' capital expenditures were established in 1974, contemporaneously with the early diffusion of CT. Although several studies have shown that
CON regulation can influence the nature and
extent of technology adoption,53 the earliest
years of CT development largely escaped
effective CON regulation. By 1978, only 35
states had CON mechanisms and, although
requirements were strengthened in 1978,54
CON programs in most states still cover only
purchases of technology by hospitals. This
loophole has allowed private physicians,
investor groups, and others to establish CT
facilities in free-standing ambulatory settings, including mobile facilities.55 Not only
was CON ineffective in the first period of
CT diffusion, but the rapid diffusion of CT
in 1975 may have been due, in part, to anticipatory behavior by hospitals hoping to
acquire the device before full CON review
took hold.9'0
By the time MRI emerged in 1981, the
CON program already was established, although a deregulation trend at the federal
level (as evidenced through the Omnibus
Budget Reconciliation Act of 1981 and the
Health Planning Block Grant of 1982) loosened CON requirements to some degree.
Through April 1984 (the most recent data
available), CON review agencies had received 168 applications for permission to
purchase a MRI system, of which 65 applications were approved, 27 were disapproved, 3 were exempted, and 73 were deferred or pending.56 As of April 1984, 17
states and the District of Columbia had developed formal guidelines to arrive at MRI

decisions; 16 states were in the process of

developing guidelines; and 17 states had no
guidelines at all.56 Only 6 of the 10 federal
health planning regions had approved any
MRI systems. Thus, despite the deregulation
trend, the CON process has discouraged the
diffusion and adoption of MRI hospitalbased scanners in many states.
Since CON review still is not required by
outpatient facilities or physician offices in 43
states, MRI systems can be obtained in these
jurisdictions through the ambulatory route
if approval of hospital requests is difficult or
impossible to obtain. Indeed, 48% of MRI
scanners installed as of December 1984 were
located outside of a hospital (Fig. 2). This is
a very high figure, especially for such an expensive device, and it indicates that CON
legislation probably is acting in concert with
other factors (Table 2) to divert MRI from
the hospital to the ambulatory setting.
MarketFactors
Market factors also may help explain
differences in the patterns of early diffusion
between CT and MRI. The health care environment is becoming more competitive
and entrepreneurial.57'58
Competition among
hospitals for patients and physicians may
encourage adoption of medical technology,36'59and one of the effects of the Medicare DRG system has been to increase the
competitive pressures on hospitals. Some
hospitals see advanced technology as a
means by which to attract patients and
maintain high occupancy rates. These pressures may be counterbalancing obstacles to
the adoption of MRI to some extent. Corporate structure and for-profit orientation
increasingly are infiltrating the delivery of
medicine, a trend that has been termed
"megacorporate" health care,60and proprietary interests have had a central role in ushering MRI into the outpatient setting.
The FDA premarket approval process
(PMA), established in 1976, also has influ-

1290
This content downloaded from 152.118.24.10 on Mon, 04 Jan 2016 11:31:13 UTC
All use subject to JSTOR Terms and Conditions

Vol. 23, No. 11

THE DIFFUSION OF CT AND MRI

enced market strategies. This process requires

that the FDA approve new medical devices
that are not substantially equivalent to existing devices (class III designation).55 Under
an investigational device exemption (IDE),
manufacturers of class III devices must collect data on the device's safety (risks and
complications) and technical effectiveness (its
ability to do or measure what it claims) before they can market the device for profit.
While MRI is the first new class III diagnostic
technology to arise under the FDA PMA
program, empirical data to date suggest that
this program has done little to inhibit the
development or adoption of MRI systems.12
Despite the claims of some manufacturers,
the data clearly indicate that there has been
widespread placement of MRI units by
manufacturers under IDE.
However, the IDE and PMA processes require that manufacturers establish a research
and investigative relationship with medical
providers to gather the clinical data necessary
to satisfy the FDA's class III requirements12
and to continue to refine the evolving technology. Similarly, hospitals increasingly are
entering into commercial and risk-sharing
relationships with external organizations that
can operate devices such as MRI and CT exempt from CON and DRG purview. Thus
industry and purchasers now share several
common interests and incentives.
The PMA program could influence MRI
manufacturers by conferring competitive
market advantages to those manufacturers
who receive early premarket approval.12This
competitive environment also may affect the
MRI industry by accelerating the standardization of MRI systems and reducing the
number of manufacturers as purchasers shop
comparatively for the most cost-effective
products.45 The changing distribution of
magnet types already, in part, may be evidence of this phenomenon. The MRI industry will be forced to spend a larger fraction
of research and development funds than
otherwise would be spent on evaluating and

advertising the cost-effectiveness of their

units. (Alternatively, some manufacturers
may be attempting to segment the MRI market into a high field-strength research and
spectroscopy market and a lower fieldstrength diagnostic one.) Moreover, purchasers will seek to protect their investments
by opting for units that offer a reduced risk
of early obsolescence.
Conclusions
In an optimal medical care system, new
technologies and innovations would be
adopted rapidly once their safety and efficacy
are established and once favorable cost-effectiveness ratios are anticipated. The technologies would be purchased and sited in
the most efficient and appropriate settings
and would be available equally to everyone
in need. Payment would reflect the actual
costs of appropriate and efficient medical
care at all times, regardless of which technologies are used and whether they are costsaving or cost-increasing.
This analysis of the early diffusion of CT
and MRI identified several areas where the
present system deviates from the ideal. Most
experts believe that available data regarding
the clinical efficacy of MRI do not warrant
widespread adoption outside of research
settings.34 Indeed, MRI's early diffusion
lagged behind that of CT partly because of
the technical uncertainty surrounding MRI.
However, the adoption pattern of MRI is
fortuitous to the degree that uncertainty associated with DRG reimbursement policy
and CON regulation contributed to its slower
diffusion. Delayed adoption secondary to
uncertainty of future regulations is not an
indication of an appropriately functioning
system. In fact, MRI appears to have diffused
more widely at present than can be justified
on clinical merits.
For regulatory and reimbursement policies
to operate appropriately and for appropriate
clinical decisions to be made by providers,
1291

This content downloaded from 152.118.24.10 on Mon, 04 Jan 2016 11:31:13 UTC
All use subject to JSTOR Terms and Conditions

HILLMAN AND SCHWARTZ

timely and accurate information on the

safety, efficacy, and cost-effectiveness of
new technologies must be made available.
This is particularly challenging when rapidly
changing technologies ("moving targets")
are involved. The present
such as MRI'2'61'62
these requirements
does
not
meet
system
of
several
the
existence
technologydespite
assessment agencies and groups.63'64The activities of these agencies and organizations
remain both redundant and poorly integrated. Most are structured to synthesize and
integrate existing information on safety, efficacy, and cost; few resources exist to support the development, collection, and analysis of primary data. The limited support for
collecting data to compare MRI with other
diagnostic modalities is small relative to the
data needs and dollar amounts spent on the
purchase, siting, and operation of these units.
The opportunity for providers to "game"
the system is enhanced in the present uncoordinated regulatory system, with the frequent net result of increased costs in excess
of commensurate benefits. The better the
coordination among CON regulation, thirdparty reimbursement policy, FDA premarket
approval, and federal tax policy, the greater
the synergy and cohesion among these programs. The less coordinated these programs,
the greater the opportunity for providers to
exploit loopholes by which to circumvent
system controls and thereby undermine policy objectives. The present trend toward
purchase and operation of MRI by FIOs is,
in part, an example of such opportunities and
the resultant behavior and, in part, a response to tax and reimbursement incentives.
Finally, the impact of specific regulatory
and reimbursement policies on the adoption
and diffusion of new medical technologies
such as MRI must be monitored and analyzed. The voluntary registry maintained by
the American College of Radiology provides
an opportunity to track the diffusion of this
technology, document changes in market

MEDICALCARE

trends, and accumulate information on purchasers. However, its voluntary nature and
the absence of data on when units are ordered reduce its potential usefulness. Our
survey of actual MRI sites identified several
errors and inaccuracies in the ACR data set.
It is clear that a voluntary registry is not sufficient to track the diffusion of MRI with the
timeliness and degree of detail required by
policymakers. A compulsory registry, administered either by a private organization
(e.g., ACR) or a governmental agency (e.g.,
FDA, NIH) with adequate funding for staffing to permit more extensive data collection
and validation is required.
If the medical system is to profit from experiences with diffusion of new technologies,
the present technology management system
must be altered. Efficient and optimal technology diffusion policy requires better collection of primary information on both clinical and regulatory issues. Most important is
the pressing need to support earlier, more
rigorous scientific evaluations of safety and
clinical efficacy. Similarly, concurrent primary data are needed to analyze the impact
of policy initiatives on providers' adoption
decisions and manufacturers' development
and marketing decisions. While these requirements are not new, today's environment of increased competition coupled with
vestiges of regulation magnifies the need for
a cohesive system. Perhaps health care goals
would be better served by dismantling CON
programs, the impact of which may be to
distort diffusion patterns in our newly competitive system. When the next generation
of imaging technology is developed, the
health care system should not be faced with
the same uncertainty and inefficient resource
allocation that exist today.
Acknowledgments
The authors are indebted to Drs. Bernard Bloom,
Randall Cebul, John Eisenberg, and William Kissick for
their helpful suggestions and to Ms. Amy Laub for her
administrative assistance.

1292
This content downloaded from 152.118.24.10 on Mon, 04 Jan 2016 11:31:13 UTC
All use subject to JSTOR Terms and Conditions

THE DIFFUSION OF CT AND MRI

Vol. 23, No. 11

References
1. Altman SH, Blendon R, eds. Medical technology:
the culprit behind health care costs? Hyattsville, MD:
National Center for Health Services Research and Bureau of Health Planning, 1979. (DHEW publication no.
(PHS) 79-3216)
2. Russell LB. Technology in hospitals. Washington,
D.C.: The Brookings Institution, 1979.
3. Medical technology and the health care system: a
study of the diffusion of equipment embodied technology. Washington, D.C.: National Academy of Sciences,
1979.
4. Fuchs VR. The rationing of medical care. N Engl
J Med 1984;311:1572.
5. Thurow LC. Learning to say "no." N Engl J Med
1984;311:1568.
6. Schwartz WB, Aaron HJ. Rationing hospital care:
lessons from Britain. N Engl J Med 1984;310:52.
7. U.S. Congress Office of Technology Assessment.
Policy implications of the computed tomography (CT)
scanner: an update. Washington, D.C.: U.S. Government
Printing Office, 1981.
8. U.S. Congress Office of Technology Assessment.
Policy implications of the computed tomography (CT)
scanner. Washington, D.C.: U.S. Government Printing
Office, 1978.
9. BakerSR. The diffusion of high technology medical
innovation: the computed tomography scanner example.
Soc Sci Med 1979;13:155.
10. Banta HD. The diffusion of the computed tomography (CT) scanner in the United States. Int J Health
Serv 1980;10:251.
11. American College of Radiology. MR Site List.
Washington, D.C., February, 1985.
12. Steinberg EP, Cohen AB. U.S. Congress Office
of Technology Assessment. Nuclear magnetic resonance
imaging technology: a clinical, industrial, and policy
analysis. Washington, D.C.: U.S. Government Printing
Office, 1984. (Technology case study 27)
13. Merton RK. Social theory and social structure.
New York: The Free Press, 1949.
14. Rogers EM. Diffusion of innovations. New York:
The Free Press, 1983.
15. Coleman JS, Katz E, Menzel H. Medical innovations: a diffusion study. Indianapolis: Bobbs-Merrill,
1966.
16. Schroeder SA, Schliftman A. The influence of
medical school on selection of career specialties. Med
Ann DC 1973;42:339.
17. Harris CM. Formation of professional attitudes
in medical studies. Br J Med Educ 1974;8:241.
18. Jonas S. Medical mystery. New York: W. W.
Norton, 1978.
19. Banta HD, Behney CJ, Willems JS, eds. Toward
rational technology in medicine: considerations for

health policy. Springer Series on Health Care and Society. Vol. 5. New York: Springer, 1981.
20. Cyert RM, March JG. A behavioral theory of the
firm. Englewood-Cliffs, NJ: Prentice-Hall, 1962.
21. Mohr LB. Determinants of innovation in organizations. Am Pol Sci Rev 1969;63:114.
22. Kaluzny AD, Glaser JH, Gentry JT,et al. Diffusion
of innovative health care services in the United States:
a study of hospitals. Med Care 1970;8:474.
23. Russell LB, Burke CS. Technological diffusion on
the hospital sector. Washington, D.C.: National Planning
Association, 1975.
24. Rapoport J. Diffusion of technological innovation
among non-profit firms: a case study of radioisotopes
in U.S. hospitals. J Econ Bus 1978;30:108.
25. Romeo AA, Wagner JL, Lee RH. Prospective
reimbursement and the diffusion of new technology in
hospitals. J Health Econ 1984;3:1.
26. Globerman S. The adoption of computer technology in hospitals. J Behav Econ 1982;11:67.
27. Becker MH. Sociometric location and innovativeness: reformulation and extension of the diffusion
model. Am Soc Rev 1970;35:267.
28. Roos NP, Schermerhorn JR, Roos LL. Hospital
performance: analyzing power and goals. J Health Soc
Behav 1974;15:78.
29. Budinger TF, Lauterbur PC. Nuclear magnetic
resonance technology for medical studies. Science
1984;226:288.
30. Pykett IL. NMR imaging in medicine. Sci Am
1982;247:78.
31. Bradbury EM, Radda GK, Allen PS. Nuclear
magnetic resonance techniques in medicine. Ann Intern
Med 1983;98:514.
32. Hillman BJ,Winkler JD, Phelps CE, et al. Adoption and diffusion of a new imaging technology: a magnetic resonance imaging prospective. Am J Roentgenol
1984;143:913.
33. BakerHL, BerquistTH, KispertDB, et al. Magnetic
resonance imaging in a routine clinical setting. Mayo
Clin Proc 1985;60:75.
34. Schroeder SA. Magnetic resonance imaging:
present costs and potential gains. Ann Intern Med
1985;102:551.
35. Bradley WG, Waluch V, Yadley RA, et al. Comparison of CT and MR in 400 patients with suspected
disease of the brain and cervical spinal cord. Radiology
1984;152:695.
36. Utterback JM. Innovations in industry and the
diffusion of technology. Science 1974;183:620.
37. Aberathy DL, GriffinDT. NMR dilemma: should
a hospital be a technology "leader" or "follower"? Mod
Health Care 1983;13(Aug):60.
38. Freedman GS, Stephens WH, Fisher B. Economic
considerations in MRI. Appl Rad 1984;13(May/June):
55.
39. Wortzman G, Holgate RC, Morgan PP. Evaluation of cost-effectiveness. Radiology 1975;117:75.

1293
This content downloaded from 152.118.24.10 on Mon, 04 Jan 2016 11:31:13 UTC
All use subject to JSTOR Terms and Conditions

HILLMANAND SCHWARTZ
40. Evens RG. The economicsof computedtomography: comparisonwith other health care costs. Radiology 1980;136:509.
41. Evens RG. Economiccosts of nuclearmagnetic
resonanceimaging.J ComputAssistTomog1984;2:200.
42. BradleyWG. Practicaleconomicconsiderations
of clinicalnuclearmagneticresonance.JAMA1984;251:
1302.
43. MansfieldE. Speed of responseof firmsin new
techniques.Q J Econ 1973;77:290.
44. GreerAL. Advancesin the study of diffusionof
innovationin health careorganizations.MilbankMem
FundQ 1977;55:505.
45. U.S. CongressOfficeof TechnologyAssessment.
Diagnosticrelated groups (DRGs) and the Medicare
program:implicationsfor medical technology.Washington, D.C.:U.S. GovernmentPrintingOffice, 1983.
46. Anderson GF, SteinbergEP. To buy or not to
buy:technologyacquisitionunderprospectivepayment.
N EnglJ Med 1984;311:182.
47. WagnerJL, KriegerM. The price of progress?
Medicaltechnologyand health care costs. J Contemp
Bus 1980;9:19.
48. Sloan FA. Regulationand the risingcost of hospital care.Rev EconStat 1981;58:479.
49. CromwellJ,KanakJR.The effectsof prospective
reimbursement
programson hospitalsand servicesharing. HealthCareFin Rev 1982;4:67.
50. KidderD, SullivanD. Hospitalpayrollcost, productivity, and employment under prospectivereimbursement.HealthCareFin Rev 1982;4:89.
51. DetskyAS, StaceySR, BombardierC. The effectiveness of a regulatorystrategyin containinghospital

MEDICALCARE

costs:the Ontarioexperience1967-1981. N EnglJ Med

1983;309:151.
52. FrankelAN. The financingof NMRequipment.
Appl Rad 1984;13(uly/Aug):55.
53. SalkeverDS. BiceTW.The impactof certificateof-need controlson hospitalinvestment.MilbankMem
FundQ 1976;53:185.
54. Nationalguidelinesforhealthplanning.In:Federal Register1978;13040.
55. U.S. CongressOfficeof TechnologyAssessment.
Federal policies and the medical devices industry.
Washington,D.C.: U.S. GovernmentPrintingOffice,
1984.
56. U.S. Departmentof Healthand HumanServices,
Officeof HealthPlanning.Summaryof nuclearmagnetic
resonance regulations/guidelines/standardsand criteria/program positions by states within regions.
Washington,D.C., 1984.
57. GinzburgE. Themonetarizationof medicalcare.
N EnglJMed 1984;310:1162.
in medi58. CunninghamRM. Entrepreneurialism
cine. N EnglJ Med 1983;309:1313.
59. WarnerKE.The need for some innovativeconcepts of innovation:an examinationof researchon the
diffusionof innovation.Pol Sci 1974;5:433.
60. Freedman SA. Megacorporatehealth care: a
choice for the future.N EnglJ Med 1985;312:579.
61. PlattR. Costcontainment:anotherview. N Engl
J Med 1983;309:726.
62. BantaD. Computedtomography:cost containment misdirected.Am J PublicHealth 1980;70:215.
63. MarwickC. Legislationexpands federalrole in
medicaltechnologyassessment.JAMA1984;252:3235.
64. IglehartJK.Healthpolicyreport:anotherchance
fortechnologyassessment.N EnglJMed 1983;307:509.

1294
This content downloaded from 152.118.24.10 on Mon, 04 Jan 2016 11:31:13 UTC
All use subject to JSTOR Terms and Conditions