The
New Navigators: from Professionals
to
Patients
R.
Baud et
al.
(Eds.)
IDS
Press,
2003
A Dynamic Web Application within an
n-tier Architecture: a Multi-Source
Information System for End-Stage
Renal Disease
Mohamed Ben Said
a
,
Ana
Simonet
b
,
Didier Guillon
b
,
Christian
Jacquelinet
C
,
Franck
Gaspozb, Eric
Dufour
d
,
Claude
Mugnier
a
,
Jean
Philippe Jais
a
,
Michel Simonet
b
#,
Paul
Landais
a
#
aService de Biostatistique et d'informatique Medicale, Hopital Necker, Universite Paris
5,
France
b TIMC-IMAG, Universite
J.
Fourier, Grenoble, France
CEtablissementfranr;ais des Greffes, Paris, France
d
R
2i
Developpement, Caen, France
# both authors equally contributedto the work.
Abstract
A Multi-Source Information System (MSIS) has been designed for the Renal
Epidemiology and Information Network (REIN) dedicated to End-Stage Renal
Disease. Interoperability has been considered at
4 levels: semantics, network,
formats andcontents. An n-tier architecture has been chosen at the network level. It
is
made out
of
a universal client, a dynamic
Web
server connected
to
a production
database and to a data warehouse. The MSIS
is
patient-oriented, based on a
regional organization. Its implementation
in
the context
of
a regional
experimentation
is
presented with insights on the design and underlying
technologies.
The
n-tier architecture is a robust model and flexible enough to
aggregate multiple information sources and integrate modular developments. The
data warehouse
is
dedicated
to
support health care decision-making.
95
Key
words
Dynamic Web interface; n-tier architecture; Multi-Source Infonnation System; Data warehouse; Business
logic; end-stage renal disease
1. Introduction
Nearly 42,000 patients suffering from end-stage renal disease (ESRD) have been treated
this year in France. However, no coordinated infonnation is available. The Renal
Epidemiology and Infonnation Network (REIN) is dedicated
to
respond to this information
need [1]. The REIN program is organized at a national and at a regional level.
It
was
experimentally developed via a regional feasibility study in 4 regions: Limousin,
Languedoc-Roussillon, Lorraine and Rhone-Alpes. A national committee for guidance and
follow-up involves several organizations:
Societe de Nephrologie, Societe francophone de
dialyse, INSERM, Paris 5 University and
J.
Fourier Grenoble University, Institut de Veille
Sanitaire, Etablissement
fran9ais des Greffes (EfG), and representatives
of
patients'
associations. In a collaborative work, guidelines for the REIN program and for the MSIS-
REIN system have been specified [2].
Client
Client
a
Inf
o
rm
a
ti
o
n
System Tier
,..sH..
...
..
...
. ....
.....
.
Client
Ti
e
r
<
:6^e,
s/ si'%,e:'
>
9fi ^
z'?^^
;.
^
Wa ^
^
.,.y
a
^
,
,,•rT
-
r
^
.^
:, x,y
^
?aax
^
, z r
y
^
, sri
f itin. .
^
r
w.
"'^
r
,;
ä.tia,o
^
,rH.sa;
✓
.
^
nz
^
ru
^
w...
^
..
^
'wn.au.x.
^
.;. ,
;.
^
,c
^
;.
^
,wíc«.z.:.w,.x<i
^
w.an..xr,;w:
ui:á%v.s
^
.,w
^
:.«
^
i«
^
»ti.
.
^
.ta;ó
^
:i
Mi
dd
l
e
Ti
e
r
96
2.1 Medical and Health Information Systems on Internet
2. Material and Methods
Patients and organizational support
The Multi-Source Information System (MSIS-REIN) is currently being tested in two
administrative regions: Limousin and Languedoc-Roussillon. In the Limousin region
(710,939 inhabitants), the target population was easy to identify since the number of ESRD
treatment units was limited. There was no prior information system dedicated to ESRD
patients and the potential commitment of the professionals was important. Languedoc-
Roussillon, (2,295,648 inhabitants), is a larger administrative region with diversified health
care structures for treatment of ESRD. Local proprietary information systems dedicated to
ESRD did exist in some units of this region.
A regional relay brings together decision makers, health professionals including a
nephrologist as the regional representative of the Multi-Source Information System (MSIS-
REIN), an epidemiologist who ensures the control of exhaustiveness and quality, and
patient representatives. Local agreements, federating the actors, have been signed in each
region. They have been documented in a "User Agreement Document" followed by
bilateral conventions between EfG and the partners.
MSIS main objectives and architecture
MSIS-REIN is dedicated to collect continuous and exhaustive records of all ESRD cases
and their clinical follow-up. It collates in a standardized representation a minimal patient
record elaborated by health professionals. MSIS aimed to fulfill the following
requirements: scalability, portability, reliability, accessibility, and cost effectiveness
oriented toward non-proprietary software. The use of standard references, the respect of
privacy, confidentiality and security [3] of patient information were required as well.
The architecture of MSIS-REIN, represented in figure 1, is based on an n-tier architecture
[4]. The client tier connects to a middle tier that is in relation with several databases. The
client tier uses a Web browser to communicate. The middle tier supports client services
through Web containers and business logic services through component containers. The
information system tier may access 3 types of database: the identification database, the
production database and the data warehouse. Business logic components in the middleware
support transactions toward the databases.
Figure 1— MSIS-REIN n-tiers architecture
2.1 Medical and Health Information Systems on Internet
97
Web server application
As part of the MSIS-REIN, a Web server application has been developed to interact with
the production database system. It consists of a collection of Web components: Java Server
PagesTM (JSP), Servlets and other resources (graphics, scripting programs or plug-ins)
organized in a directory structure. Web components interact within a Web container, which
corresponds to a runtime environment providing a context and a life cycle management [4].
Tomcat is the Web container used in MSIS-REIN. It is free and available as open source
software. JSPs have been used, an extensible Web technology that uses template data,
HTML or XML elements, custom elements, scripting languages and server-side Java
objects to return dynamic content to a client, usually a Web browser. Servlets have also
been used. A Serviet is a Java program that extends the functionality of a Web server,
generating dynamic content and interacting with a client using a request-response
paradigm. JSPs are compiled into Java Servlets at the server side. As the source file is
modified, JSPs are compiled and run "on the fly", while Servlets need to be compiled first
into Java classes. In MSIS-REIN, Forte For Java Community EditionTM is used to compile
Java programs. The Web application has been developed under WindowsTM and deployed
in a Linux environment.
Screen layout design and user interface
Once a connection with the MSIS-REIN server is established and the user authenticated, a
first screen with the principal navigational possibilities is displayed. It introduces the screen
structure, and the different functionalities represented in figure 2. Customized areas for the
region and for each ESRD treatment unit are illustrated by specific logos. A palette-menu
area is provided for browsing the main chapters. Patient information can be accessed within
2 mouse clicks whatever the user's position in the application [5, 6].
Region
lo
g
o
1
-
Navigation
- .1.st
level
Dial
ys
i
s u
nit
Current
locati
o
n
in the
a
pp
lication
2
Navigation
2nd:
level
3
-
Information content a
r
ea
Data
entry forms
P
ati
e
nt
information summary
Information content commands
Figure 2 - Screen layout structure
Production database
Designed to support day-to-day operations and updates, a production database system is
transaction-driven and requires a rapid response time and a high availability. It deals
generally with few records and is accessed by multiple users connected using a large
network. It is a domain-oriented application. MyS QLTM[7] was selected as database server
in the experimentation phase for its reliability in the context of Internet and its free
availability as an "Open Source" software, supported by a large community of developers.
98
2.1 Medical and Health Information Systems on Internet
Data warehouse
A data warehouse [8, 9] has been created which builds up unified views of patient
information, generated from distinct databases. Unlike the production database, the data
warehouse is a collection of data which are subject-oriented, integrated, time variant, non
volatile and organized to support analytical and decision processes. Data are not created "de
novo" by end-users, but derived periodically from the production database considering
cautiously privacy, integrity and confidentiality. The updating process identifies the
necessary information to transfer, aggregates and integrates data into a new state or in a
new version of an existing state. The update process has also been designed to yield the
extensions of the persistent views, to check the consistency of the integrated information, to
manage message acknowledgements and to update data warehouse dictionaries. When the
state of the data warehouse changes, new persistent views and data presentations are
generated and made available to the users.
Information fluxes and networking
MSIS-REIN is oriented towards interoperability at three levels: local, corresponding to the
ESRD treatment units, regional and national. Information is transferred through a secured
extranet enabling confidentiality and data availability. Interoperability at the networking
level refers to integrating heterogeneous hardware, software and computer environments
taking into account security requirements and communication protocols. MSIS-REIN takes
advantage of the capacity of multiple information systems to exchange and analyse
information. MSIS-REIN integrates four paradigms: semantics, networking, formatting,
and contents as presented earlier.
A rapid connection such as Asymmetrical Digital Subscriber Line (ADSL) connection or
faster like Renater, the French inter-university network, is recommended to use MSIS-
REIN. Data is encrypted and transferred via a secure connection, which requires Web
browsers, 6.0 or newer for Internet Explorer and 6.2 or newer for Netscape. Older Web
browser versions do not support the encryption level used in MSIS-REIN.
3. Results
The client interface was successfully tested. Data entry forms are highly structured and
separated into blocks, requiring either scrolling or section links in order to "jump" to
different sections, as represented in figure 3. A number of controls aim at making data entry
easier and at making sure that the information, which will be sent to the server, is
consistent. Some controls check data format, field length and the range of possible values.
Other controls check contextual and logical consistency within subsets of information such
as co-morbidities, methods of treatment or dates. To avoid free text entry variations and
typing errors, requests are sent to online dictionaries and the user selects thesaurus items.
This facility was first developed for medical diagnoses entry. It refers to the thesaurus of
Nephrology or to the International Classification of Diseases 10
th
version (ICD 10)
thesaurus. It has been extended to administrative information such as city of residency or
birth. A checklist reminds the user of the minimal requested information to create or update
a patient record. Additional controls delimit the way to submit data to the server.
Confidentiality issues and functional specifications have influenced the design of the
production database. A conceptual model was implemented to document patient clinical
events such as: admission in ESRD treatment unit, follow-up, change in the treatment
modalities, and transfer from one unit to another or death. Once validated and consolidated,
patient information is processed in such a way that anonymous data is generated and
periodically exported to the data warehouse.
In Limousin, data collection is exhaustive, including patients treated for the first time this
dottier
n.
aroi ti `
1iii
2.1 Medical and Health Information Systems on Internet
ne
e
;
,
........ . .,.,..., ................ ....
feNt f.tE1
_._.. ...
.:
...
:.:......
99
Car7surt4r "roddiGatifrn du
dossier d p
r
Wit€;`
t
ttrtr
&
darts
re
COMM
1stgtá
t^^
3aQp2
Wtrttrtiaaaa nurnere dtx O5 ytër
den
e i+elre Centre
Nam
Rari?.ranyciiqtie
(3
premiers
tarartäres} :
t~
^
et?ri n7áti
^
! :
^
Fir
ena
ik:
r
----
s
e.
:
C
toreU!)n
FOrrrtnin
age
naise,ance:
^
ziVnontàaaa
Cómrnurae d
e
nalysan : ?
Pays
de
na!ssanca :
^
.
Commune
d
e residence
: ?
Liwats <
[su
^
us
^
.rk
- .±....,.: T^'H
:6M.
:
•.}d.:'
Idendfitsdixt
ilü5'
^
:2ik
'
i
^
n"::
^
?R2Yftt
R
=F
R NCE
:y.t^:;á•:^^^ß'
^
y
^^:
S
^
^
g^^ ^
cj
á^^^^^
,`%;
:
ó.
w..
^
'•;
^
;-?,,.
^cx^:^.
c^
äii
^
,u
^
'.'
^
á•
intsie t
o
itt
at
r :
?` ä=
^
02NL. rs plu
+2!otui@ro
V.
.
4
ee lai
,
uAL!
rsphr01,
;
40.ae
.
1:,,Pers
e^^;
ït
^
rttd
^^
en COWS, Wind
.»
F
Figure 3 — Example of MSIS-REIN screen: patient data entry form.
year (2002). Average entry duration per patient record is about 5 minutes. This
experimentation has also made it possible to simulate data transfer between the production
database and the data warehouse, to develop queries interfaces and persistent views, and to
generate dynamic responses with geographical representations.
The use of MSIS-REIN in Languedoc-Roussillon has showed its ease of use and efficiency
to respond to local needs of information management. The regional experimentation
enabled us to refine the Web application and facilitate the information flow between the
MSIS-REIN components. A particular improvement was reached for the entry form at the
client side.
4. Discussion
For the users, the objectives of an efficient networked information system are to avoid
redundancy, to reduce data entry tasks and, in return, to obtain structured results. Users
work on different platforms and use different software. Our focus during the design and
experimentation of MSIS-REIN was to adapt to existing environments and habits and to
introduce as few changes as possible. The paradigm of interoperability has been privileged
in particular on the networking part. The overall architecture is based on an n-tier
architecture, which separates production from analytical systems and integrates a
component organization approach. This n-tier architecture offers new possibilities to
manage the information flow. On the end-user side, a universal Web browser allows users
to interact with the system. Data is encrypted and transferred via a secure connection. The
advantage of this solution is its portability. The middleware provides the system with
necessary dynamics to transfer information to the databases and allows user interaction. If
required, adding databases to the information system tier will not introduce major changes
in the general architecture. MSIS-REIN has been designed to support a greater number of
regions and ESRD treatment units, without major work overload.
The attention of the professionals in Limousin is drawn to the quality control and
consolidation of the information entered in the system. Using MSIS-REIN in Languedoc-
100
2.1 Medical and Health Information Systems on Internet
Roussillon, where ESRD dedicated information systems are partially used, is often looked
at as a work overload rather than a potentially helpful process. Such difficulties raise the
issue of interfacing MSIS-REIN with existing systems, and a generic solution based on
publishing an XML data model used in MSIS-REIN has been proposed.
The implementation of the identification server in the next release will help to guarantee the
uniqueness of patient representation in MSIS-REIN. Automatic reminders will assist the
professionals in consolidating information and updating annual follow-up information. The
data warehouse functionalities are also bound to grow to respond to the needs of health care
professionals and decision makers.
The success of an information network such as MSIS-REIN is highly dependent on human
factors. Beside technology improvements, it is necessary to identify resistance to changes
and to deploy didactic skills to help users in understanding and adapting to this new
interactive information network.
5. Conclusion
The n-tier architecture is a robust model and flexible enough to aggregate multiple
information sources and integrate modular programming developments. The data
warehouse is the core of the data that supports health care decision-making. The MSIS-
REIN model developed for ESRD might be exported to other domains of medicine,
particularly for the follow-up of other chronic diseases.
6. Acknowledgments
The nephrologists in charge of the ESRD units of Limousin and Languedoc-Roussillon are acknowledged for
their fruitful cooperation and comments. JP Necker and X Ferreira are acknowledged for their skillful help
and Ms D Bernhard for rereading the manuscript. This work was supported by a grant of the Ministère de
l'Education Nationale, de la Recherche et de la Technologie N°99B0625 and by University of Paris 5.
7. References
[1]
Landais
P. L'insuffisance rénale
terminale
en France : offre de
soms
et prévention.
Presse
Med.2002;
31 176-85.
[2]
Landais
P, Simonet A, Guillon D, Jacquelinet C, Ben Said M, Mugnier C, Simonet M. SIMS@REIN : Un
Système d'Information Multi-Sources pour l'Insuffisance Rénale
Terminale.
CR Biol 2002;325:515-528.
[3]
Rindfleisch TC. Privacy, Information Technology, and Health Care.
Communications of the ACM
August
1997; 40; 93-100
[4]
Kassem N, Enterprise team. Designing Enterprise Applications with the JavaTM 2 Platform, Enterprise
Edition.
Addison-Wesley
New York 2000, pp 341.
[5]
Marcus A. Graphic Design for Electronic Documents and User Interfaces. ACM Press. New York. 1992.
PP
266.
[6]
Nielsen J. Designing Web Usability . New Riders Publ. Indianapolis. 2000. pp 419.
[7]
Dubois P. MySQL. Campuserve : Paris, 2000, pp 750.
[8] Jones K. An Introduction to Data Warehousing: What Are the Implications for the Network?
Int. J.
Network Mgmt
1998; 8:42-56.
[9]
Kimball R. The Data warehouse Toolkit: Practical Techniques for Building Dimensional Data
warehouses.
John Wiley.
1996; pp 416.
8. Address for correspondence
Dr Paul
Landais,
Service de Biostatistique et d'Informatique Médicale, Hópital Necker-Enfants Malades,
Université Paris 5. 149, rue de
Sèvres,
75743 Paris cedex 15. e-mail :
landais
@necker.fr
