Content uploaded by Dahlia Remler
Author content
All content in this area was uploaded by Dahlia Remler
Content may be subject to copyright.
Information and Communications Technology in U.S.
Healthcare: Why is Adoption So Slow and Is Slower Better?
Michael C. Christensen, Dr.PH
Senior Research Fellow
International Center for Health Outcomes and Innovation Research
Columbia University
600 West 168th Street, 7th Floor
New York, NY 10032
Phone: +1 212 305 9100
Fax: +1 212 305 4256
E-mail: mcc153@columbia.edu
Dahlia Remler, Ph.D
Associate Professor
School of Public Affairs
Baruch College, CUNY
One Bernard Baruch Way, Box D-901
New York, NY 10010
Phone: +1 646-660-6725
Fax: +1 646-660-6701
E-mail: Dahlia_Remler@baruch.cuny.edu
and
Department of Economics
The Graduate Center, City University of New York
and
National Bureau of Economic Research
Abstract
Politicians across the political spectrum support greater investment in healthcare
information and communications technology (ICT) and expect it to significantly decrease
costs and improve health outcomes. We address three policy questions about adoption of
ICT in healthcare: First, why is there so little adoption? Second, what policies will
facilitate and accelerate adoption? Third, what is the best pace for adoption? We first
describe the unusual economics of ICT, particularly network externalities, and then
determine how it interacts with and is exacerbated by the unusual economics of
healthcare. High replacement costs and the need for technical compatibility are general
barriers to ICT adoption and often result in lock-in to adopted technologies. These effects
are compounded in healthcare because the markets for healthcare services, health
insurance and labor are inter-linked with each other and the government. Patient
heterogeneity further exacerbates these effects. Finally, ICT markets are often
characterized by natural monopolies, resulting in little product diversity that is ill-suited
to patient heterogeneity. The ongoing process for setting technical standards for health
ICT is critical, but needs to include all relevant stakeholders, including patient groups.
The process must be careful (slow), flexible, and allow for as much diversity as possible.
We find that waiting to adopt ICT is a surprisingly wise policy.
1
I. Introduction
Amidst all the disagreement about how to solve our healthcare problems,
politicians across the political spectrum agree on one thing: greater use of information
and communications technology (ICT) in healthcare. For example, Barack Obama’s
campaign web site proclaimed that his health plan would “[Lower] Costs Through
Investment in Electronic Health Information Technology Systems” (Obama 2008),
while John McCain in his campaign similarly stated that his administration would support
“Greater Use Of Information Technology To Reduce Costs” (McCain 2008). The
potential cost-savings—both direct and indirect through better health outcomes—are
considered substantial. The economic stimulus package recently passed by Congress
includes $19 billion for investments in health ICT (Kaisernetwork 2009).
While ICTs have dramatically changed the way we work, interact, and entertain
ourselves, healthcare lags far behind in adopting ICT (Shortliffe 2005). Visiting a new
doctor generally requires completing a whole new medical history on paper. Most
physicians still have their notes in paper charts. The results of many kinds of diagnostic
tests cannot be shared among physicians electronically. Only four percent of ambulatory
care physicians have a fully functional electronic records system, and only 13 percent
have a basic system (DesRoches et al 2008). Only 5% of all hospitals have computerized
physician order entry (Jha et al 2006).
ICT in healthcare does seem to offer great promise for improving patient
outcomes and reducing direct and indirect costs (Lehoux et al 2000; Danzon & Furukawa
2001; White House 2005). Clinical applications include remote diagnosis and surgery,
transmission of radiological images, interactive video visits, and continuous remote
2
analysis of self-monitored data. The most heralded application is the electronic medical
record (EMR), which can reduce diagnostic test replication and ensure that all physicians
and providers have up-to-date information for every patient encountered in clinical
practice. For chronic diseases, ICT has the potential to significantly improve health
outcomes over the long-term and thereby reduce direct and indirect costs given the much
greater opportunities for continuous monitoring and adjustment of treatment (Shea et al
2002; Christensen and Remler 2007).
While the lack of adoption of ICT in healthcare has been heavily criticized
(Kleincke 2000, Ortiz & Clancy 2003; BrailerA 2005), the reasons for the currently low
levels of adoption are not fully understood. Barriers to ICT adoption already discussed
include significant legal issues such as licensure, liability, malpractice, and
confidentiality (Kuszler 1999; Spielberg 1999; Stanberry 2000; Gottlieb et al 2005) as
well as financial barriers such as the lack of reimbursement (Cutler, Feldman, and
Horwitz 2005; Christensen and Remler 2007). However, the last point raises the
following question: if ICT in clinical care has such potential to improve health outcomes
and lower costs, why do insurers, including the government, then not reimburse for its
use?
We initially sought to answer two policy questions: First, given the enormous
interest in ICT for clinical care and the broad bipartisan agreement on its value, why is
there so little adoption? Second, what policies will facilitate and accelerate adoption of
ICT? Others have emphasized the need for technical standards (e.g., Walker et al 2005,
BrailerB 2005). We contend that the standard setting process should account for the
diverse interests in healthcare. This could be facilitated by a common operating platform
3
and communication standards that maximally accommodate diversity. During our
analysis, we discovered, to our surprise, that the currently low and slow adoption of ICT
in clinical care may be desirable for society. Therefore, we added a third policy question:
what is the best pace for ICT adoption? To address these questions, we examine how the
unusual economics of ICT and healthcare interact and exacerbate each another.
II. Information and Communications Technology Adoption
Competitive market analysis often shapes public policy, including healthcare
policy. Those well versed in policy know how market forces create incentives to adopt
valuable new technologies. Specifically, consumers in a market economy comparison
shop and “vote with their feet” pressuring producers for lower prices, higher quality and
more desirable products. Consequently, producers adopt any technology that enables
them to lower production costs. Producers also adopt technology to create new higher
quality products if the additional amount consumers are willing to pay exceeds new
production, adoption, implementation and transition costs. In the long-run, market forces
result in the best use of all inputs, including new technologies, through investment, entry,
and exit. Of course, those well versed in policy also know the traditional ways that
competitive market forces do not produce the best outcomes for society, such as when
monopolies prevail or externalities exist like pollution.
However, the same policy savvy individuals are less familiar with how adoption
of seemingly private ICT products does not fit the standard competitive market story.
First and foremost, most ICT products - from fax machines to word-processing programs
to social networking sites - exhibit a particular kind of externality known as a network
4
effect (Katz and Shapiro 1985; Shy 2001; Shapiro and Varian 1999). A network effect
occurs when the value of a product depends on its use by other consumers, e.g. the more
people who use fax machines the more valuable the individual fax machine becomes.
Technologies with network effects may refer to both physical networks, such as
telephone networks, and virtual networks, such as compatible digital video software.
Because an individual does not consider the value they add for others when joining a
network there is an externality. Network externalities also give rise to positive feedback:
the more people who use a network, the more valuable it is, and the more people will
want to join that network (Katz and Shapiro 1994; Shapiro and Varian 1999).
Network externalities and positive feedback have several effects. First, in the
early stages of a technology launch, adoption starts slowly as each consumer awaits the
adoption by other consumers to a point when the technology is sufficiently valuable.
Second, while a technology might be valuable if enough consumers choose to buy it, it
may never get adopted if the size of the network effect is “big enough”. Third, however,
if the network effect is big enough, once adoption starts it accelerates and rapidly reaches
saturation—the so-called S-shaped adoption curve. Fourth, because adopters of a new
technology naturally seek the network they expect to offer the broadest and most valuable
connection, expectations of a network effect can lead to the realization of that network
effect. For instance, in the early days of the market for fax machines, the value of existing
fax machines increased as each new customer bought a fax machine and thereby joined
the network. Once people believed that enough consumers would eventually buy a fax
machine, adoption increased rapidly, and the value of the network increased dramatically.
5
Potential customers were now more tempted to buy a fax machine and join the network
as the value of the network had increased.
Network effects and positive feedback generally reduce product differentiation –
diversity in the kinds of products available. In particular, when one product does not
work with another product, the network effect and positive feedback can make a single
product dominant. Once a product has a significant market share, more people believe
that it will become the standard and buy it, driving other products out of the market. In
several historic examples, ICT manufacturers have competed for the same market and
only one of them has emerged as the winner (Katz and Shapiro 1994; Shapiro and Varian
1999). For example, the spreadsheet program Excel eventually pushed out Lotus 1-2-3
and other spreadsheet programs. This winner-take-all feature of ICT markets limits
diversity in the technologies available (Katz & Shapiro 1994). The more limited choices
available for ICT may limit adoption to some extent if accommodating diverse tastes is
important. As we will see later, accommodating diversity in healthcare is critical.
Adopting ICT is also more complicated than adopting other technologies in
several other ways. First, using new technology generally requires training, which
significantly increases short-term costs, particularly due to workers being pulled away
from productive tasks. Second, other large switching costs are the norm in the ICT
market (Shapiro and Varian 1999). Translation of information, for instance, is required
when new electronic storage or communication technologies replace old paper records, or
old electronic systems. Third, because, information is stored, manipulated, and
communicated using inter-linked technologies, i.e. a “system” of multiple pieces of
hardware and software, the various technologies must be interoperable—able to work
6
with one another. The need for interoperability can raise switching costs from training
and translation if switching one technology causes changes in how another technology is
used.
Fourth, switching costs may be so high that users are effectively locked-in to a
specific form of ICT, either at the system or vendor level, and new technologies may
never be adopted. Depending on the industry, the risk of system disruption or breakdown
makes the risk of using a new vendor or technology, especially an unproven one,
potentially huge. Such breakdown may imply irreversible damage to the company or
individual user. Because adopting a new technology cannot be easily reversed, adopters
of ICT need to think carefully about the added value of adopting the technology far into
the future. When there is uncertainty about the future requirements of the technology,
ICT users may be very reluctant to adopt a new technology. Fifth, switching costs are
non-linear. For instance, persuading five independent, yet highly inter-dependent, banks
to switch to a new ICT platform is more than five times as hard as getting one bank to
switch, yet all five of the banks need to switch as no single bank wants to be the first to
give up network externalities.
Finally, in order not to lose the value of existing ICT, the ability to integrate old
and new ICT (backwards compatibility) may be critical for adoption. For example, long
after the development of superior CD and jump drive storage technologies, many
computers continued to have floppy disc drive readers. Backwards compatibility is also
important as it lowers switching costs. Replacing CD players with MP3 players, for
instance, was not that costly because the music on CDs could be “ripped” for MP3
players. In contrast, the costs of switching from vinyl record players to CD players were
7
much greater because entire music collections were not compatible. However, backwards
compatibility can reduce the performance standard of the new technology below its
potential, e.g. a word processing program that accepts documents in their old format will
not run as quickly. The trade-off between better technology and backwards compatibility
is one example of the general conflict between innovative technologies and network
externalities and their associated switching costs (Shapiro & Varian 1999). The adoption
of even vastly superior technologies can be delayed for a long time, due to the need for
backwards compatibility.
Technical standards for data exchange can address many of the problems related
to ICT adoption. However, such standards do not necessarily emerge automatically in the
market place. Incompatible technologies often compete in a high-stakes winner-take-all
battle, where the market outcome can either be a truce with multiple producers, a duopoly
with only two competing producers, a monopoly, or a fight to death where no technology
survives commercially (Katz and Shapiro 1994; Shapiro and Varian 1999). AM stereo
radio provides an illustration (Shapiro and Varian 1999). In the early 1980s, the Federal
Communications Commission decided to let the market choose among four rival
technologies for the AM stereo radio. In the end, none were adopted due to the
substantial costs of the new AM compatible radios, the uncertainty about the winning
technology, and the limited incremental value to existing radios. The case illustrates that
ICT adoption may be especially difficult when multiple groups of adopters need to
coordinate and agree on a common technical standard.
To summarize: network effects and positive feedback can delay or inhibit ICT
adoption and limit the diversity of products available. Switching costs and the need for
8
interoperability inhibit adoption or even result in complete lock-in to suboptimal
technologies. Technical standards can address many of the barriers to ICT adoption, yet
history tells us that these do not necessarily emerge automatically in the market place.
III: Adoption of ICT in Healthcare
The unusual economics of ICT inhibit or slow down all ICT adoption, not just
healthcare ICT adoption. Yet the fantastic gains of ICT have outweighed those barriers in
most industries and aspects of public and private life. Why does healthcare ICT lag so far
behind?
Interlinked Markets
Most healthcare services are covered through health insurance. Therefore, the
economic incentives for technology adoption by patients and providers depend on the
operations of the insurance market and the interactions between insurers and healthcare
providers (Weisbrod 1991). Further complicating matters, health insurance is typically
provided by employers, as part of labor compensation, or by government. Consequently,
there are, in principle, three interlinked markets in healthcare: (1) healthcare services, (2)
health insurance and (3) the labor market, as well as government’s role as insurer. Within
each of these markets there are, of course, many different types of healthcare providers,
insurers and employers creating a myriad of unique linkages across the three principle
markets. The result of all these complicated linkages is that the market forces from
patients to healthcare providers are far more indirect and blunted than are the market
forces from consumers to producers in most sectors. In order for patient preferences to
9
drive healthcare providers to adopt available technology, the market forces must be
transmitted through each stage and each stage brings its own transaction costs.
Consider the example of patients who want to ask simple questions of their
primary care physician (PCP) by e-mail. What options exist for such patients to obtain
these e-mail services? A patient who tries to persuade his PCP to adopt e-mail will have
little leverage, since he is not likely to change PCP for such a small issue as e-mail
communication. The patient-physician relationship encompasses so much more. What
about persuading the insurer to influence the PCP (through payment or other means)?
The patient has even less leverage with the insurer than the physician, since even more
services are bundled together in the health insurance package. Moreover, the patient may
have limited insurance options. How about trying to influence the employer to influence
the insurer to influence the PCP? In this instance the patient has the least leverage to
negotiate, because no employee will change job just for the option of e-mail
communication with his PCP.
In contrast, conventional market forces do apply to healthcare services not
covered by insurance but purchased directly out-of-pocket, and in fact, ICT adoption has
been much more rapid for these services. For example, psychotherapy is often not
covered by insurance (Zuvekas 2001) and there has been rapid growth in psychotherapy
via e-mail and other internet media. Similarly for long-term care not reimbursed by the
government (Medicare and Medicaid) “smart technologies” for communications and
monitoring of the elderly living alone have been adopted relatively rapidly (Wallace
2003).
10
What about the incentives of healthcare providers, insurers, employers, or the
government for adopting ICT? Any stakeholder will be financially motivated to pay for
ICT if its adoption lowers the use of other forms of healthcare sufficiently to result in
lower net costs of that stakeholder. Health care providers and insurers will also be
motivated to adopt and pay for ICT if it substantially improves health outcomes – but
only if those choosing and compensating the stakeholder recognize the outcomes.
Unfortunately, many of the potential benefits of ICT, both reduced costs and improved
outcomes, occur far in the future, when the covered individual will likely have another
insurer or provider. For instance, approximately one in five Americans switch health
plans every year (Community Tracking Survey 2007). Moreover, if ICT benefits
chronically ill patients (or other less healthy patients) more than other patients, an insurer
who supports ICT adoption might disproportionately attract less healthy and more
expensive enrollees. In that case, the usual selection forces would dissuade insurers from
supporting ICT adoption. Finally, because many existing provider payment forms are tied
to volume, rather than health outcomes, providers can reap few if any financial benefits
from ICT adoption. Different forms of payment to healthcare providers are needed to
provide financial incentives to adopt ICT (Miller et al 2005; Christensen and Remler
2007). Interestingly, healthcare systems with a single long-term insurer, such as the
Veterans Affairs (VA) or British National Health System (NHS), can internalize the long-
term benefits of ICT adoption and consequently they have been early adopters (Evans,
Nichol, and Perlin 2006; Greenhalgh et al 2008).
What about the specific incentives of employers and the government? Employers
may care about ICT’s value in enhancing patient care either because they want to provide
11
more generous compensation that employees appreciate or because ICT increases
employee productivity through better health. The government could value ICT if it
provides net savings to the government insurance programs or if voters support the
technology. For example, we have seen limited changes to Medicare payment policy for
the type of ICT that directly benefits healthcare providers in rural areas (Puskin 2001).
However, until recent Obama administration proposals, neither government nor
employers took any actions to push for broad ICT adoption in healthcare.
Of course, the interlinked markets in healthcare blunt the market forces for
adopting many innovations and technologies, not just ICT. Indeed, many innovations
desired by patients, such as scheduling systems that reduce patient waiting time, are not
implemented. Yet, healthcare is known for its rapid adoption of many new technologies.
Why the difference? Certainly, existing payment structures, and perhaps physician
preferences, drive rapid adoption of new diagnostic tests, drugs and procedures.
Primarily, however, the slow and limited adoption of ICT is driven by the interaction of
the distinct features of ICT and healthcare markets.
Interaction of Barriers to Adoption: ICT and Healthcare Market
The many interlinked markets in healthcare and patient heterogeneity
significantly intensify the general barriers to ICT adoption– low product differentiation,
high switching costs and technical compatibility. While the standard competitive market
only has consumers and producers, producers in the ICT market also encompass
manufacturers of add-on products, information goods producers, systems producers,
component producers and infrastructure operators. For healthcare ICT, consumers
include many kinds of healthcare providers; patients of different diseases, ages, education
12
levels, and disease severity; and insurers of all kinds. These consumers (or ICT users)
differ in their resources for surmounting the barriers to ICT adoption. They also have
greatly different needs for ICT, creating demand for product diversity and conflicting
with the naturally low product differentiation in ICT markets.
The switching costs of ICT are also likely to be particularly high in healthcare.
The EMR, for instance, is difficult to develop and adopt due to clinical complexity and
the need to serve many purposes (Lorenzi et al 2008). To implement a new electronic
medical record, existing data from paper records must be entered; an extremely time-
consuming task. If existing data are not entered immediately, then healthcare providers
may have to work with parallel paper and electronic records for an extended transition
period. Additionally, the value of integrating data across providers can only be gained if
many providers adopt compatible ICTs, yet healthcare providers who stand to gain very
differently from adoption will be unlikely to agree on the timing and type of technology
adoption.
The switching costs are also likely to be higher given the many different actors
involved in health care. Patients, providers, insurers, and producers of software, hardware
and communication infrastructure all depend on each other – directly or indirectly – in
the use of the same or associated technologies. If technological change or a change in
user requirements causes one component of the overall communication and data-sharing
infrastructure to be replaced, all agents face switching costs. In addition, patients, in
particular those with chronic diseases, who may benefit greatly from ICT adoption, are
older on average, raising requirements for user-friendliness. Their needs for ICT are also
disease-specific, necessitating different types of ICT. As a consequence, changes to the
13
operating platform of healthcare ICT could require changes to many disease-specific
types of patient software.
Overall, there are three major sources of switching costs that must be addressed to
adopt a new ICT in healthcare. First, there are the upfront costs of purchasing the
technology. These include the costs of new durable hardware (such as mainframe
computers), the operating systems (to store and manipulate information) and any
complementary products (such as disease management software). Second, there is the
cost of information storage in databases and the cost of moving clinical data from one
database to another. Old medical records need to be stored for legal reasons. Hospitals
and insurers must transfer massive information encoded in specialized formats to new
systems. Third, there is the usual cost of the training involved in using the new ICT
(Schuster et al 2003; Bossen 2007). This training is often brand-specific and considerable
additional time will be required to become equally efficient with a new technology.
Examples of Difficulties in Healthcare ICT Adoption
A recently released report on early adopters of an EMR in the British National
Health System (NHS) makes clear the substantial switching costs involved (Greenhalgh
et al 2008). Regarding training, for instance, the report states that effective
implementation requires “[e]nough [training] for front-line staff at the right time in a real
working environment. The IT literacy of many NHS staff was low. Formal training…did
not always have a positive impact on the ability of the staff to actually use the
system…This…highlights the need for ongoing, local, on-the-job training” (section
1.19). With regard to data recording, the report states “GPs… worry about workload,
especially in the phase 2 of the upload in which selected aspects of patients’ medical
14
history will be added to the record” (section 1.15) and “the evaluation revealed a more
complex and less easily quantifiable picture [of workload]” (section 1.35) The report also
noted that switching and adoption costs were exacerbated by a lack of “spare human and
technical capacity that could be used to buffer the stress of innovation” (section 1.17).
The high switching costs encountered in the British NHS are likely to be even
higher in the US, given the many payers and decision makers in US healthcare. As
discussed, the switching costs of ICT adoption rise more than proportionately as new
users are added. In health care these non-linear switching costs are likely to be
particularly pronounced given its many interlinked markets. A network large enough to
overcome collective switching costs is unlikely to emerge automatically.
Because the decision to adopt a system-wide EMR was taken centrally, the
Veterans Affairs (VA) health care system was able to surmount the large switching costs
(including going from paper records in 150+ medical centers). The VA now has an
advanced system of electronic medical records and telemedicine applications linking
pharmacy, laboratory, and all other care (Evans et al 2006). The system is now partially
credited for the strikingly good outcomes achieved for diabetes and other chronic
diseases in the VA patient population (Sawin et al 2004). However, even when the
switching costs can be overcome, and compatible ICT extensively adopted, early
adoption of health ICT may still be problematic. An annual software upgrade in the VA
in August of 2008 resulted in faulty displays of medical records and consequently
incorrect doses of drugs, unimplemented treatment cessations and delays in treatments
(Associated Press 2009). Such problems make early ICT adoption problematic, as we
show in the next section.
15
A four-internist family practice group described their EMR adoption and the
problems they encountered, as well as the benefits, illustrating the barriers we describe
(Baron et al 2006). They summarized that “its financial impact is not clearly positive;
work flows were substantially disrupted; and the quality of the environment initially
deteriorated greatly.” Adoption costs included two full days of on-site training and a 50%
reduction in schedule for three days after going live, as well as the $140,000 purchased
costs of hardware, software, training and one year of support. Virus attacks, data service
interruptions and untimely technical support interrupted care. They described having to
“redesign every office system” and felt like they were “redesigning an airplane in flight.”
While they also described many of the long-term benefits predicted, it is little wonder
that most primary care doctors have avoided EMR adoption, given the many stresses on
primary care doctors.
Both the potential value of and barriers to ICT adoption are illustrated by the use
of software to store and analyze information about blood glucose for diabetes patients. If
coupled with information about medications, diet and exercise, such information
substantially helps diabetes patients maintain good health and avoid disabling or fatal
outcomes. Many blood glucose monitoring and data management systems exist (Diabetes
Forecast 2008) and have been around for many years (e.g., Diabetes Forecast Resource
Guide 2003). Diligent patients now benefit substantially from the information and
analysis provided by the software. While providers can review the patients´ self-collected
data in hard copy easily (and hence make use of it in their overall disease management),
it is much more difficult to receive the information electronically and analyze it. Every
meter manufacturer is potentially a different case, implying adoption of multiple software
16
programs. Moreover, despite the proliferation of blood sugar measurement software, we
are unaware of any attempt to integrate that data into an EMR or broader form of
healthcare ICT. Thus, even for diligent patients and physicians who collect, analyze and
communicate the information, the diversity of methods makes aggregation of information
almost impossible.
In summary, patient heterogeneity, inter-linked markets and higher switching
costs exacerbate the general barriers to ICT adoption– low product differentiation, high
switching costs and technical compatibility. So, network effects do not emerge
automatically but require intervention to ensure the required technical compatibility and
interoperability.
IV: The Value of Waiting
At any given point in time—now, five years ago, or next year – the potential for
ICT to reduce costs and improve outcomes would argue for adopting as quickly as
possible. In other words, in a static framework, society loses when potentially valuable
adoption opportunities are missed. However, a dynamic perspective that incorporates
uncertainty provides a less clear picture. As described, the future value of specific ICT
systems is highly uncertain, due to both rapid technological change and uncertainty
relating to adoption decisions by other healthcare providers and insurers. Moreover,
decisions to adopt ICT are irreversible, due to a variety of lock-in effects, as well as the
sunk investment costs. Once system-wide technology adoption has occurred in
healthcare, it is hard, if not impossible, to turn back to previous systems of
communication and information storage. Observe the resilience of the traditional paper
17
record. Though the combination of uncertainty and irreversibility of ICT investment
reduces the value of all ICT investments, the effect is greater for healthcare.
The real option value theory of investment demonstrates the value of waiting to
invest when investments are irreversible and uncertainty about future market conditions
exists (Dixit and Pindyck 1994; Luerhman 1998). This theory, used to value capital
investments in the corporate world, has also been applied in the social sector as well,
including ICT adoption in higher education (Oslington 2004). Briefly, if one invests now,
when the future value of the investment is uncertain, the return on the investment may be
either high or low depending on future market conditions. Investing now destroys the
option to invest at a later time point. However, if one waits until the uncertainty is
resolved, the value of the investment will be known and the optimal decision—whether to
invest or not and the specific type of investment—can be made. Depending on the
upfront investment costs and the level of uncertainty about the future market conditions,
the real option approach to investment may dictate that waiting to adopt will result in the
highest net present value.
Consider a healthcare provider’s decision to invest in a particular ICT without
knowing whether a much better technology will come along and/or before he knows what
specific type of ICT other providers will adopt. If the provider maintains his option to
invest at a later time point, he can invest when other providers have made clear the type
of ICT they consider acceptable and when superior technologies are developed. Option
value theory interacts with the theory of network externalities and positive feedback.
While network externalities in general may delay or completely prevent adoption of
technologies, the uncertainty and irreversibility of investment compound the effect. The
18
value of delaying investment also increases with the cost of adopting the wrong
technology. Due to network externalities and positive feedback, the cost of adopting the
wrong ICT is generally higher than other kinds of technology. In healthcare, the value of
delaying investments in ICT is likely to be so much greater than in other sectors because
the costs of adopting the wrong type of ICT will be so much higher.
Why is the penalty for the wrong ICT adoption so much greater in healthcare?
First, relative to other industries (such as banking and insurance), the consequences of
technical errors in healthcare are likely to be larger in magnitude, more salient, more
attention-getting, and engender stronger emotions. The publicity surrounding the VA
system software bugs in 2008 illustrates the attention-getting and salience of errors. Most
importantly, some health care errors result in death or permanent disability, completely
irreversible states. Therefore, technical problems due to poor interoperability or data
storage can result in severe consequences for both early adopters and society at large.
Second, the larger switching costs in healthcare, due to the large number of actors
and the coordination problems among them, makes the financial losses from adopting the
wrong ICT so much greater in healthcare. Individual providers and insurers could face
substantial financial losses if their particular ICT system is not compatible with the future
standards and requirements for healthcare ICT.
Third, the federal regulation protecting the confidentiality of health information
introduced by Health Insurance Portability and Accountability Act of 2003 implies that
being stranded with an incompatible technology or losing information due to system
failure and breakdown is simply not an option (U.S. Department of Health and Human
Services 2003). Many states have introduced legislation that sets even higher standards
19
for how patient data should be protected and the extent to which it can be shared across
providers (Gottlieb et al 2005). The effect of these federal and state regulations is an
increase in the cost of adopting the wrong technology.
The uncertainty in returns of ICT investments and the irreversibility of these
investments may imply that there are real advantages to approaching ICT adoption
carefully and waiting for the right technology to come along before system-level adoption
takes place. Again, the recently released report on the early adopter of the British NHS
EMR and the very recent experience in the VA system provide evidence in this regard.
For instance, the British NHS had originally planned to have an EMR universally in use
by 2009, yet they have repeatedly been plagued by many of the problems our theory
predicts: lack of interoperability, errors in translation of paper patient files (and resulting
slowdowns to avoid such errors), difficulties training staff and extensive time needed to
train staff (Greenhalgh et al 2008). The EMR is currently considered “an immature
technology which staff have described as ‘clunky’ and which currently interfaces poorly
with other ICT systems. Many staff have given up using it ‘until it works better’ (section
1.14).
In the US, the EMR has been adopted by approximately 5% of all U.S. hospitals
for computerized physician order entry (Jha et al 2006). The fact that no widespread
adoption of the EMR took place ten years ago when the first versions were introduced
may actually, from a societal perspective, have been the right decision. The technical
standards available at the time did not provide for the kind of interoperability needed
today for an optimal use of the technology. If half of all U.S. hospitals had adopted the
EMR back then they may have been locked–in to a technology that could not be used for
20
all the clinical applications needed today. Perhaps the history of telemedicine in the US
provides the best evidence on the value of waiting. The technologies used in the early
demonstration projects at Nebraska Psychiatric Institute and Massachusetts General
Hospital/Logan International Airport Medical Station (Brown 1995), while novel and
advanced at their time, would clearly be outdated in today’s healthcare environment. Had
nationwide adoption of these technologies taken place in the 1960/70s, the healthcare
system would have been locked-in to technologies vastly inferior to the technologies
available today. Given the large collective switching costs healthcare providers need to
be careful about the technologies they lock into and going slowly may in fact not only be
a good idea but also the optimal approach in healthcare.
V: Technology Standards in Healthcare
Some of the reasons for delaying adoption, in particular the uncertainty about
other providers’ adoption decisions, can be alleviated with common technical standards.
Such standards facilitate and accelerate adoption and improve the choice of ICT for a
number of reasons (Farrell and Saloner 1985; Hammond 2005; Pedersen and Fomin
2005). First of all, standards enhance compatibility, which ensures that the ICT works for
all stakeholders, vastly expanding the network, and its value. Secondly, technical
standards also reduce the risk that ICT users end up stranded with an incompatible
technology that cannot be used in the future (Katz and Shapiro 1994). Finally, technical
standards reduce switching costs and thereby lock-in to specific technologies. By
reducing switching costs, standards make future switching to superior, yet-to-be-invented
21
technologies, possible and cheaper. However, standards cannot reduce the value of
waiting to see what superior technologies are invented.
How much standardization should there be? Should there be a standard for every
single device, every single form of software and every possible interaction between
them? Endless and rigid standards may inhibit innovation and the use of certain types of
ICT. The socially optimal extent of standardization in healthcare depends on ICT user
heterogeneity. The more heterogeneous users’ ICT needs, the more important is
flexibility in ICT design to meet those needs. Since heterogeneity in healthcare – both
across medical conditions and among patients with a given medical condition – is much
greater than among consumers or producers in other sectors, product differentiation is
particularly desirable. Standardization can have negative consequences if the standards
do not accommodate this heterogeneity. A tradeoff therefore exists between seeking more
standardization to allow for greater adoption and less standardization to allow for greater
product differentiation.
ICT has many diverse possible applications in chronic disease management, for
instance, and there are therefore many different forms of ICT for which standards could
be required. For example, medical devices can capture patients´ physiological data;
software can transmit such data to an EMR, and the data can be presented and analyzed
in different forms, depending on the audience. In the case of diabetes, blood glucose data
could be integrated with data on food consumption, medication and exercise for
transmission to the EMR. Standards would be needed for the patient’s device, the EMR
and the analysis software, to ensure widespread interoperability. The ideal ICT standard
will vary across chronic diseases. The technology that will be best for diabetes care is
22
unlikely to be best for the management of heart disease. Therefore, it is highly unclear
how to select a technical standard to work with all ICT applications. A standard EMR, for
instance, adopted in all hospitals without regard to the specific infrastructure, medical
specialties and patient populations treated could do more harm than good.
The diversity needed for healthcare ICT could be addressed by a common
operating platform with communication standards that allow disease-specific
modifications. Because new diagnostic tests and new ways of analyzing data will
continue to be developed, the capacity to add new modules, new variables and data
structures is critical. The means of communicating across new modules should also be
flexible. Such flexibility and expandability will make ICT potentially valuable to a
variety of different providers, facilitating network expansion and value. The practicalities
of how much of the operating platform that should be universal and how much should be
idiosyncratic is clearly an open question. The need to enable disease-specific technical
modifications implies that full ICT compatibility will not be possible. An open non-
proprietary technical platform, like open source software, that allows others to write for it
and develop specific ICT systems will go along way for addressing the need for product
differentiation.
Connecting for Health has a promising initiative for standard setting (Halamka et
al 2005). The organization proposes a minimum set of standards that aims to protect
privacy, patient control and data security, while allowing timely access to data across
information networks. These standards will facilitate interoperability of standard
interfaces and transactions at local, regional and national levels and address secure
transport of data, provide the essential components required for the infrastructure,
23
including secure connectivity, reliable authentication, and formats for health data. By
focusing on a minimum set of standards that provides for interoperability, the approach
recognizes the heterogeneity in technology needs and applications that is crucial in
healthcare ICT.
To reach agreement on specific standards for data recording and exchange formal
standard setting is needed. The process needs rules and procedures for reaching
consensus on technical decisions. Avoiding dominance by any single group of
stakeholders is critical. While the formal process is often criticized for being slow,
cumbersome, and political (David & Shurmer 1996), it has historically been absolutely
paramount to launching new technologies (Shapiro and Varian 1999).
In 2005, the Office of the National Coordinator for Health Information
Technology created the Health Information Technology Standards Panel (HITSP) to
harmonize existing health IT standards and the Certification Commission for Health
Information Technology (CCHIT) to certify vendor products in reference to these
standards. The working procedure of the HITSP is described as “transparent” and
“consensus-based” and the panel is considered a “multi-stakeholder coordinating body
where private and public sector interests agree to cooperatively address the healthcare
information technology needs of the nation” (HITSP website, 2009). Primarily, the panel
work to identify missing or incomplete standards and recommend standards that enable
interoperability. The first set of interoperability standards was recognized by the federal
government in January 2008.
As of November 2008 the panel encompassed more than 450 organizations. While
the panel contains IT vendors, organizations representing “general consumer interests”,
24
healthcare providers, public health agencies, government agencies, and standard
developing organizations, its composition is of concern for several reasons. First, patient
organizations are almost entirely absent— a breast cancer organization being the sole
example. Thus, the chronically ill have no direct voice. Specific disease perspectives are
needed to ensure interoperability with all the relevant physical biometric devices and
usability by patients. Therefore, all major patient advocacy groups for chronic diseases
(e.g. the American Diabetes Association and the American Heart Association) should be
involved. Of course, all major types of providers (e.g., represented by the American
Academy of Family Physicians and the American Hospital Association) and major health
insurers (e.g., represented by American Health Insurance Plans and Centers for Medicare
and Medicaid Services) are needed, and they are now members. Involvement of all these
stakeholders will make the process slower and more cumbersome, making healthcare less
than it could be in the short-run. But in the long-run, their involvement will result in a
better solution.
A second problem with the panel’s composition is that more than 50% of the
organizations are designated as IT vendors. Vendors (individually and collectively) are
likely to pursue those standards that give them proprietary advantages. Third, the vendors
themselves are dominated by pre-internet vendors (e.g., GE, Siemens, and McKesson)
while representatives of emergent or potentially disruptive technologies based on the
internet and web-based applications have no chair positions in the various technical
committees (Kibbe and McLaughlin 2008). Such pre-internet vendors may not make
sufficient use of the internet and other new technologies now widely available and
adopted. On the other hand, vendors who have created healthcare ICT, including their
25
communication standards have technical knowledge that others lack, based on experience
with technical limitations and barriers. As a result of this expertise, they should play a
prominent role in the process.
Many observers of the standard setting process are disappointed, because after
several years of standards documentation and resolution of several standards “disputes”,
use and implementation of these standards is still far away. “Not a single data element
has been exchanged in the real world healthcare systems using standards this process has
developed or deployed,” one noted (Karp 2007). As we illustrated, a slow process could
be better than a rapid process. Nonetheless, the current process could be improved.
Critics such as Diamond and Shirky (2008) offer suggestions that fit with our own
analysis. First, they find the process focused on specifying too much too soon, stating that
“only critical requirements for any given increment of improvement” should be adopted.
Second, they note that “standards are not really standards unless they are widely adopted,
and this step cannot easily be mandated.” The standards-setting process should be more
sensitive to what users (providers or patients) want to adopt, rather than trying to dictate
what users should adopt. The standards for the internet were simple, minimal and created
incrementally.
The government has a key role in standard setting, but that is not its only role. The
government was critical in facilitating the internet, which was initially developed and
financed by the Department of Defense. The internet also illustrates how the government
can develop a clear technical platform needed in order to develop new forms of ICT. In
general, by financing, endorsing and adopting technologies, the government can help
achieve a critical mass of ICT users, subsequently leading to rapid uptake and widespread
26
use (Rosenfeld et al 2005; Middleton 2005). While the importance of influential players
in ensuring technology adoption has been debated (Chakravorti 2003; Berry & Keller
2003), for healthcare ICT the role of the government is likely to be critical. First, the
interlinked healthcare markets and resulting mismatch in financial incentives for ICT
adoption is a strong argument for government involvement. Second, the government is
both a major payer and regulator of healthcare and its ICT choices will inevitably be
quite influential. It is however essential that the government’s interests as an insurer
(Medicare and Medicaid) and even provider (e.g., county hospitals) are not allowed to
overwhelm the broader societal interests. For instance, if Medicare decided to endorse
one specific EMR to fit the Medicare population but not other patient groups, then this
could be a huge obstacle for more widespread adoption of the EMR.
VI. Conclusions
Politicians across the political spectrum support substantial funding for ICT in
healthcare, particularly the EMR, and believe that it will lower healthcare costs and
improve health outcomes. This agreement is particularly striking given the general
disagreement about healthcare policy. Presently, the proposed stimulus package includes
19 billion dollars for healthcare ICT and it is not controversial (Kaisernetwork 2009). The
gains in improved outcomes and efficiency seem self-evident to most citizens.
Government should continue its support for this process. To realize the potential of health
ICT, technical compatibility and interoperability standards are critical.
Standards, however, are not the only factors affecting adoption. Financial
incentives, i.e. subsidies, are paramount. In earlier work, we described how the form of
27
payment (payment design) to providers will strongly influence whether and how ICT is
used in healthcare (Christensen and Remler 2007). Insurers currently have little reason to
pay providers for ICT adoption, because the short-term benefits (financial and/or health
improvement) are small. While the long-term benefits of ICT likely justify the upfront
adoption costs, private insurers cannot benefit because most individuals will have a
different insurer well before the long term benefits materialize. In the US, only Medicare
and the VA presently have a financial interest in the long-term perspective. Thus, the
long-term and broader societal goals do justify government support. In addition to
financial matters, data security and confidentiality concerns must also be addressed, as
has been widely recognized (e.g., Diamond and Shirky 2008).
The present standard setting panel, HITSP, has generally the right framework but
could be improved in several ways. First, patient organizations have almost no role.
Making healthcare information technology accessible and usable for patients, particularly
chronically ill patients who need to manage their own disease for their residual lifespan
should be a greater priority. Second, vendors, particularly early, pre-internet vendors play
too large a role. For both financial reasons and technical familiarity, they could
implement lock-in restrictions that are incapable of adapting to more modern innovations.
As Diamond and Shirky noted, “it is better to share important but un-codified information
between Doctor A and Doctor B so that an informed clinical decision can be made, than
to have perfectly formatted data that never leaves Doctor A’s office.” Standards for
moving text could be done before deciding on data structures for the EMR.
If interoperable and compatible ICT in healthcare is widely adopted and it allows
for data aggregation and integration across regions and diseases, medical knowledge is
28
likely to substantially expand. Data mining and other methods of data analysis on such
large databases could provide valuable information to guide medicine, public health and
health services. The ultimate goal of ICT adoption should not just be a computerized
health care system but one that can be adapted to realize this potential.
Finally, our analysis suggests that the process should not be forced too quickly
and that waiting this long for ICT may not have been such a bad thing. A good standard
setting process will be slow and cumbersome. While the government may need to use its
regulatory powers and its powers as Medicare payer, to speed up the process, this should
not be done too quickly. The value of waiting to adopt ICT in healthcare is greater than in
other areas, because the costs of adopting the wrong type of ICT are so much higher: the
risks and irreversible consequences of technical errors and the consequences of lock-in
into a suboptimal technology.
29
30
Acknowledgements:
This paper has evolved greatly, taking very different forms over time. We thank those
who read it at various points for useful comments and suggestions: Lawrence D. Brown,
Sherry Glied, Sanders Korenman, Thomas Main, Michael Sparer (the editor for this
version), Mark Schlesinger (the editor for an earlier version of this paper), Joshua Graff
Zivin, participants in seminars at the University of Maryland Baltimore County and
Baruch College, City University of New York, and the referees. We also thank several
physicians and researchers in healthcare services for conversations providing a general
background for the article. Finally, Christensen would like to thank the Danish Research
Agency, Danish Ministry of Science, Technology and Innovation, for the funding
provided to conduct the research for this paper.
References
Associated Press. Lawmaker to investigate software glitches at VA. January 15, 2009
Baron RJ, Fabens EL, Schiffman M,Wolf E. 2006. “Electronic Health Records: Just
Around the corner? Or over the Cliff?” Annals of Internal Medicine 143(3): 222-226.
Berry J, Keller E. The Influentials: One American in Ten Tells the Other Nine How to
Vote, Where to Eat, and What to Buy. 2003. Free Press.
Bossen C. Test the artefact--develop the organization. The implementation of an
electronic medication plan. Int J Med Inform. 2007; 76(1):13-21.
Brailer DJA. Economic Perspective on Health Information Technology. Business
Economics July 2005; 6-11.
Brailer DJB. Interoperability: The Key To The Future Healthcare System. Health Affairs
2005; 24, Supplement 1: W5-19 to W5-21.
Brown N. A Brief History of Telemedicine. Telemedicine Information Exchange, May
30, 1995.
Chakravorti B. The Slow Pace of Fast Change: Brining Innovations to Market in a
Connected World. Harvard Business School Press. 2003.
Christensen MC, Remler DK. Information and Communication Technology in Chronic
Disease Care: What Are the Implications for Payment? Medical Care Research and
Review 2007; 64 (2): 123-147.
31
Cutler DM, Feldman NE, Horwitz JR. U.S. Adoption Of Computerized Physician Order
Entry Systems. Health Affairs 2005; 24(6): 1654-1663.
Danzon PM, Furukawa MF. Healthcare: Competition and Productivity in The Economic
Payoff from the Internet Revolution, Robert E. Litan and Alice M. Rivlin, eds.
Washington DC: Brookings. 2001.
David PA, Shurmer M. Formal Standards-Setting for Global Telecommunications and
Information Services. Telecommunication Policy 1996; 20(10): 789-815.
DesRoches CM, Camplbell EG, Rao SR et al. Electronic Health Records in Ambulatory
Care--- A National Survey of Physicians. New England Journal of Medicine 2008 359:5-
60.
Diabetes Forecast. 2003. “Pairing your Meter with Software”. January, p. 84.
Diabetes Forecast 2008 Resource Guide.2008. Available at
http://www.diabetes.org/diabetes-forecast/resource-guide.jsp Last accessed January 18,
2008.
Diamond CC, Shirky C. Health Information Technology: A Few Years of Magical
Thinking? Health Affairs 2008 DOI 10137/hlfaff.27.5w383.
Dixit AK, Pindyck R. Investment under Uncertainty. Princeton: Princeton University
Press. 1994.
32
Evans DC, Nichol PW, Perlin JB. Effect of the Implementation of an Enterprise-wide
Electronic Health Record on Productivity in the Veterans Health Administration. Health
Economics, Policy and Law 2006;1: 163-169.
Farrell J, Saloner G. Standardization, Compatibility, and Innovation. RAND Journal of
Economics 1985; 16: 70-83.
Gottlieb LK, Stone EM, Stone D, Dunbrack LA, Calladine J. Regulatory and Policy
Barriers to Effective Clinical Data Exchange: Lessons Learned From MedsInfo-ED.
Health Affairs 2005; 24(5): 1197-1203.
Greenhalgh T, Stramer K, Bratan E, Russell J, Mohammad Y, Wood G, Hinder S.
Summary Care Record Early Adopter Programme: An independent evaluation by
University College London. London: University College London. 2008.
Halamka J, Overhage M, Ricciardi L, Rishel W, Shirky C, Diamond C. Exchanging
Health Information: Local Distribution, National Coordination. Health Affairs 2005; 24
(5): 1170-1179.
Hammond EW. The Making and Adoption of Health Data Standards. Health Affairs
2005; 24 (5): 1205-1213.
Jha AK, Ferris TG, Donelan K, DesRoches C, Shield A, Rosenbaum S, Blumenthal D.
How Common Are Electronic Health Records in the United States? A Summary of the
Evidence. Health Affairs 2006, Web Exclusive, p: W496 – W507.
33
Kaisernetwork. 2009.
http://www.kaisernetwork.org/DAILY_REPORTS/rep_index.cfm?DR_ID=56534, Last
accessed February 8, 2009.
Karp S. 2007. Testimony available at
http://www.chcf.org/documents/healthit/KarpITAdoptionIOM.pdf, Last accessed March
13, 2009.
Katz M, Shapiro C. Network Externalities, Competition, and Compatibility. American
Economic Review 1985; 75(3): 424-440.
Katz ML, Shapiro C. Systems Competition and Network Effects. Journal of Economic
Perspectives 1994; 8(2): 93-116.
Kibbe DC, McLaughlin CP. The Alternative Route: Hanging out the Unmentionables For
Better Decision Making in Health Information Technology. Health Affairs 2008 DOI
10.137/hlfaff.27.5w396. .
Kleinke JD. Vaporware.com: The Failed Promise of the Healthcare Internet. Health
Affairs 2000; 19(6): 57-69.
Kuszler PC. Telemedicine and Integrated Healthcare Delivery: Compounding
Malpractice Liability. American Journal of Law & Medicine 1999; 25: 297-326.
Lehoux P, Battista RN, Lance J-M. Telehealth: Passing Fad or Lasting Benefits?
Canadian Journal of Public Health 2000; 91(4): 277-280.
34
Lorenzi NM, Novak LL, Weiss JB, Gadd CS, Unertl KM. Crossing the implementation
chasm: a proposal for bold action. J Am Med Inform Assoc. 2008; 15(3):290-296.
Luehrman TA. Strategy As A Portfolio of Real Options. Harvard Business Review,
September-October 1998: 89-99.
McCain J. 2008. http://www.johnmccain.com/Informing/Issues/19ba2f1c-c03f-4ac2-
8cd5-5cf2edb527cf.htm, last accessed July 24, 2008.
Middleton B. Achieving U.S. Health Information Technology Adoption: The Need For A
Third Hand. Health Affairs 2005; 24(5): 1269-1272.
Miller RH, West C, Brown TM, Sim I, Ganchoff C. The Value Of Electronic Health
Records in Solo Or Small Group Practices. Health Affairs 2005; 24 (5): 1127-1137.
Obama BH. 2008. (http://www.barackobama.com/issues/healthcare/), last accessed July
24, 2008.
Ortiz E, Clancy CM. Use of Information Technology to Improve the Quality of
Healthcare in the United States. Health Services Research 2003; 38(2): 11-22.
Oslington P. The Impact of Uncertainty and Irreversibility on Investments in Online
Training. Distance Education 2004; 25(2): 233-242.
Pedersen MK, Fomin VV. The Economics of Standards and Standardization in
Information and Communication Technologies. Open Standards and Their Early
Adoption. Open Standards Research Report. 2005. Available on-line at:
http://ir.lib.cbs.dk/download/ISBN/x656517335.pdf. Accessed March 15, 2007.
35
Puskin DS. Telemedince: Follow the Money. Online Journal of Issues in Nursing. 2001.
Available at: http://nursingworld.org/ojin/topic16/tpc16_1.htm. Accessed March 2,
2007.
Rosenfeld S, Bernasek C, Mendelson D. Medicare´s Next Voyage: Encouraging
Physicians to Adopt Health Information Technology. Health Affairs 2005; 24(5): 1138-
1158.
Sawin, C.T., D.J. Walder, D.S. Bross, and L.M. Pogach. 2004. Diabetes Process and
Outcome Measures in the Department of Veterans Affairs. Diabetes Care, Vol. 27 Suppl
2 p. B90.
Schuster DM, Hall SE, Couse CB, Swayngim DS, Kohatsu KY. Involving users in the
implementation of an imaging order entry system. J Am Med Inform Assoc.
2003;10(4):315-321.
Shapiro C, Varian HR. Information Rules. Boston, Mass. Harvard Business School Press.
1999.
Shea S, Starren J, Weinstock R, Knudson P, Teresi J, Holmes D, Palmas W, Field L,
Goland R, Tuck C, Hripcsak G, Capps L. Columbia University’s Informatics and
Telemedicine (IDEATel) Project: Rationale and Design. Journal of the American
Medical Informatics Association 2002; 9(1): 49-62.
Shortliffe EH. Strategic Action in Health Information Technology: Why The Obvious
Has Taken So Long. Health Affairs 2005; 24(5): 1222-1231.
36
Shy O. The Economics of Network Industries. Cambridge: University of Cambridge
Press. 2001.
Spielberg AR. Online Without a Net: Physician-Patient Communication by Electronic
Mail. American Journal of Law & Medicine 1999; 25: 267-295.
Stanberry B. Telemedicine: Barriers and Opportunities in the 21st Century. Journal of
Internal Medicine 2000; 247: 615-628.
Walker J, Pan E, Johnston D, Adler-Milstein J, Bates DW, Middleton B. The Value of
Healthcare Information Exchange and Interoperability. Health Affairs 2005; 24,
Supplement 1: W5-10 to W5-18.
Wallace DJ.Gadget Knows Grandma Inside Out. New York Times, October 23, 2003,
http://www.nytimes.com/2003/10/23/technology/circuits/23elde.html, Accessed March 2,
2007.
Weisbrod BA. The Healthcare Quadrilemma: An Essay on Technological Change,
Insurance, and Cost Containment. Journal of Economic Literature 1991; 29 (2): 523-552.
White House, Office of the Press Secretary. Fact Sheet: Improving Care and Saving
Lives Through Health IT. 2005. Available on-line at:
http://www.whitehouse.gov/news/releases/2005/01/20050127-2.html. Accessed
March 2, 2007.
37
38
U.S. Department of Health and Human Services. Protecting the Privacy of Patient´s
Health Information. 2003. Available at:
http://www.hhs.gov/news/facts/privacy.html. Accessed March 8, 2007.
Zuvekas SH. Trends In Mental Health Services Use And Spending, 1987-1996; While
access to mental health services has greatly increased, high out-of-pocket costs still deter
many from seeking treatment. Health Affairs 2001; 20(2): 214-225.
