international journal of medical informatics 79 (2010) 225–231
journal homepage: www.intl.elsevierhealth.com/journals/ijmi
Choos hc
techn
Rodolph
a
University
b
Université
c
INSERM - U729, Paris, France
article info
Article histor
Received 28
Received in
31 Decembe
Accepted 4
Keywords:
Healthcare i
technologie
Investments
Production f
Econometri
abstract
1. Int
It has beco
sion withou
hospitals [1
and Reinve
Technologi
cerning hig
∗
Correspon
Switzerland
E-mail a
1
Address
1386-5056/$
doi:10.1016/y:
September 2009
revised form
r 2009
January 2010
nformation
s
unction
cs
Objectives: Choosing and justifying the right amount of investment in healthcare information
technologies (HITECH or HIT) in hospitals is an ever increasing challenge. Our objectives are
to assess the financial impact of HIT on hospital outcome, and propose decision-helping
tools that could be used to rationalize the distribution of hospital finances.
Design: We used a production function and microeconomic tools on data of 21 Paris university
hospitals recorded from 1998 to 2006 to compute the elasticity coefficients of HIT versus non-
HIT capital and labor as regards to hospital financial outcome and optimize the distribution
of investments according to the productivity associated with each input.
Results: HIT inputs and non-HIT inputs both have a positive and significant impact on hospi-
tal production (elasticity coefficients respectively of 0.106 and 0.893; R
2
of 0.92). We forecast
2006 results from the 1998 to 2005 dataset with an accuracy of +0.61%. With the model used,
the best proportion of HIT investments was estimated to be 10.6% of total input and this
was predicted to lead to a total saving of 388 million Euros for the 2006 dataset.
Conclusion: Considering HIT investment from the point of view of a global portfolio and
applying econometric and microeconomic tools allow the required confidence level to be
attained for choosing the right amount of HIT investments. It could also allow hospitals
using these tools to make substantial savings, and help them forecast their choices for the
following year for better HITECH governance in the current stimulation context.
© 2010 Elsevier Ireland Ltd. All rights reserved.
roduction and objectives
me almost impossible to make a strategic deci-
t involving information technology (IT) in modern
]. For example, the new 2009 American Recovery
stment Act (ARRA) regarding Health Information
es (HITECH or HIT) gives strong incentives con-
h technology investments and especially electronic
ding author at: University Hospitals of Geneva, Service of Medical Informatics, 24, rue Micheli-Du-Crest, CH-1211 Genève 14,
. Tel.: +41 0 22 30 55 878.
ddresses: rodolphe.meyer@sim.hcuge.ch (R. Meyer), patrice.degoulet@egp.aphp.fr (P. Degoulet).
: Université Paris Descartes & HEGP, Laboratoire SPIM, 15 rue de l’école de Médecine, 75006 Paris, France.
health records (EHRs) in US hospitals [2]. However, HIT con-
tinues to increase expenditure on lines for which nearly all
decision makers believe that clear profitability has not been
demonstrated [3]. Difficulties in capturing the impact of IT
in national economies have first been expressed by Economy
Nobel Prize winner Robert Solow’s in a New-York Times 1987
interview: “You can see the computer age everywhere but in the
productivity statistics”. If there is widespread agreement about
the importance of health information systems (HIS), the per-
– see front matter © 2010 Elsevier Ireland Ltd. All rights reserved.
j.ijmedinf.2010.01.001ing the right amount of healt
ologies investments
e Meyer
a,c,∗
, Patrice Degoulet
b,c,1
Hospitals of Geneva, Geneva, Switzerland
Paris Descartes & HEGP, Paris, Franceare information

226 international journal of medical informatics 79 (2010) 225–231
ceived lack of financial benefits raises the recurrent problem
of justifying the associated investments and deciding on the
appropriate
that IT inv
ion of chief
chief execu
essential to
IT investm
Relation
complex an
difficult to g
available a
net presen
return-on-i
every inves
[4,11–13] m
and contin
mization) t
a particula
computeriz
support sys
tions syste
these meth
ing numbe
as they bec
racy requir
the metho
Furthermo
uate a futu
against inn
uary 2008 H
“they divert
beyond the i
accounting
identified a
added valu
of studies h
ing from th
a global po
and trying
a PACS acq
value contr
columnists
abandoned.
processes, d
cient custom
performance
Thus, m
techniques
benefits of
results of s
agers to ma
benefits an
A comp
vided by th
Econometr
cation of q
and elucida
can be exte
individual
ular investments in a global portfolio perspective as validated
in prior research on industrial business [5,22]. In a prelimi-
tudy
e an
ing c
perio
ed b
to t
lev
base
nd U
orre
udy
“crit
com
e. A
wee
efits.
acqu
s pa
ial o
dy o
the
al ou
etr
asure
non
tion
es fo
Ma
Pro
cono
ente
ditur
, lead
ensi
tion
n lin
ele
ts o
shed
geth
ougl
step
arn o
stud
ly co
en a
6, N
n b
al (A
logy
sed a
˛
L
ˇamount of money to spend on IT. The result is
estments will inevitably be too low in the opin-
information officers (CIO) and always too high for
tive officers (CEO). It has as a consequence become
quantify accurately the added value of appropriate
ents in the healthcare sector [4].
ships between IT and hospital activities are indeed
d the economic impact of new IT investments is
rasp. To perform this task, the most common tools
re accounting methods like cost benefits analysis,
t value and internal rate of return [3,5–10]. These
nvestment (ROI) methods can be applied to almost
tment project of any kind. Cost-benefit analyses
ainly emphasize the indirect earnings (e.g., quality
uity of care, users’ satisfaction, and process opti-
ransformed into a monetary value of implementing
r clinical information system component (e.g.,
ed physician order entry (CPOE), clinical decision
tems (CDSS) or picture archiving and communica-
ms (PACS)) [1,3,14–18]. To achieve the best results
ods have to accumulate an exponentially increas-
r of variables which might result in them failing
ome overly complex. In most cases, as the accu-
ed increases, the amount of effort needed to feed
d rather than working the project also increases.
re, all these financial methods, when used to eval-
re investment, tend to be systematically biased
ovation [5,19]. As Christensen said in the Jan-
arvard Business Review about their exclusive use
resources away from investments whose payoff lies
mmediate horizon” [20]. In addition, all studies using
ROI reach their limits by focusing only on specific,
nd targeted types of benefits, neglecting the overall
e of the project on a strategic level. Only a handful
ave attempted to measure overall earnings result-
e integration of the different HIS components into
rtfolio or strategic approach [13,6,21,22]. Isolating
to measure the value added by a single project, like
uisition and deployment, is akin to assessing the
ibuted by the cheese to a pizza. As Computer world
pointed out “the idea that there are IT projects must be
There are only projects targeted at improving business
eveloping new products or services, delivering more effi-
er service or improving some other aspect of business
” [23].
any decision makers rely only on classic financial
that do not necessarily capture all the business
their IT investment [24] and the contradictory
ome of these studies frequently lead hospital man-
ke decisions solely on the basis of expected indirect
d/or empirical evidence.
lementary approach to respond to this issue is pro-
e set of tools emerging from econometric research.
ics is concerned with the development and appli-
uantitative and statistical methods to the study
tion of economic principles [25,26]. These methods
nded from the macroeconomic level to the level of
businesses to analyze the overall impact of partic-
nary s
positiv
(includ
8-year
expect
related
gration
study,
thousa
ment c
This st
until a
ship be
positiv
lag bet
of ben
the IT
[22].
Thi
financ
nal stu
assess
hospit
econom
to me
IT and
propor
incom
2.
2.1.
The e
repres
expen
ciency
field, s
utiliza
functio
to the
produc
establi
put to
Paul D
major
ing ye
initial
ticular
has be
In 195
functio
residu
techno
expres
Y = AKof 17 French acute-care hospitals, we observed a
d significant relationship between IT investment
apital and labor) and hospital productivity over an
d (1998–2005) [21]. The results also showed that the
enefits from the investments made were directly
he integration level of the HIS: the higher the inte-
el, the greater the benefits. Another econometric
d on a much larger but heterogeneous set of two
S hospitals, showed that higher levels of IT invest-
late with improved hospital cost performance [22].
also showed that IT acquisitions are cost-additive
ical mass” is achieved, at which point the relation-
es neutral for a period of time but ultimately turns
nother interesting result is that there is a natural
n technology implementation and the emergence
Cost reductions can be made in the same year as
isition, but generally it took 2–5 years to break even
per explores the relationships between hospital
utcomes and IT and non-IT inputs in a longitudi-
f 21 university hospitals. The objectives are (1) to
respective links between IT and non-IT inputs and
tcomes; (2) to assess the predictive capacity of an
ic model in an homogenous group of structures; (3)
the technical substitution relationship between
-IT investments; and (4) to compute the optimized
of IT inputs versus non-IT inputs to get the best
r an hospital.
terials and methods
duction functions
mic approach of capital efficiency is generally
d by the ratio of production divided by capital
e [26]. A large ratio indicates better capital effi-
ing to a greater output [27–29]. In the econometric
tive data on this efficiency are provided by the
of a production or cost function. A production
ks the growth and productivity of an enterprise
ments, or production factors, used to generate
r services [30]. A mathematical relationship is
between the production (output) and the factors
er to obtain it (inputs). The American economist
as and the mathematician Richard Cobb made a
forward by proposing a non-linear function link-
r output (Y), capital (K) and labor (L) [30,31]. The
ies with this function undertaken in 1930 par-
ncerned the industrial sector, and since then it
pplied in all economic sectors seeking efficiency.
obel Prize winner Robert Solow enhanced the
y introducing a new factor known as the Solow
) that is traditionally regarded as a marker of
level [32]. The Solow production function can be
s
(1)