EpiScanGIS: an online geographic surveillance system for meningococcal disease.

ral
International Journal of Health
ssBioMed CentGeographics
Open AcceResearch
EpiScanGIS: an online geographic surveillance system for
meningococcal disease
Markus Reinhardt1,2,4, Johannes Elias2, Jürgen Albert1, Matthias Frosch2,
Dag Harmsen3,4 and Ulrich Vogel*2
Address: 1Computer Science II, University of Würzburg, Germany, 2Institute for Hygiene and Microbiology, University of Würzburg, Germany,
3Department of Periodontology, University of Münster, Germany and 4Ridom GmbH, Würzburg, Germany
Email: Markus Reinhardt - mreinhardt@hygiene.uni-wuerzburg.de; Johannes Elias - jelias@hygiene.uni-wuerzburg.de;
Jürgen Albert - albert@informatik.uni-wuerzburg.de; Matthias Frosch - mfrosch@hygiene.uni-wuerzburg.de;
Dag Harmsen - dharmsen@gmx.net; Ulrich Vogel* - uvogel@hygiene.uni-wuerzburg.de
* Corresponding author
Abstract
Background: Surveillance of infectious diseases increasingly relies on Geographic Information
Systems (GIS). The integration of pathogen fine typing data in dynamic systems and visualization of
spatio-temporal clusters are a technical challenge for system development.
Results: An online geographic information system (EpiScanGIS) based on open source
components has been launched in Germany in May 2006 for real time provision of meningococcal
typing data in conjunction with demographic information (age, incidence, population density).
Spatio-temporal clusters of disease detected by computer assisted cluster analysis (SaTScan™) are
visualized on maps. EpiScanGIS enables dynamic generation of animated maps. The system is based
on open source components; its architecture is open for other infectious agents and geographic
regions. EpiScanGIS is available at http://www.episcangis.org, and currently has 80 registered users,
mostly from the public health service in Germany. At present more than 2,900 cases of invasive
meningococcal disease are stored in the database (data as of June 3, 2008).
Conclusion: EpiScanGIS exemplifies GIS applications and early-warning systems in laboratory
surveillance of infectious diseases.
Background
The incidences of transmissible infectious diseases display
considerable fluctuation in time and space. Forecasts are
hampered by multiple influencing factors, such as viru-
lence of causative agents and their genetic variants, social
networks and travel, herd immunity, and climate changes.
Members of the epidemic intelligence community
sion makers regarding the implementation of control
measures. GIS visualize complex spatio-temporal events
and thus help to analyze data on geographic maps consist-
ing of several layers of information. Data can be shared
over local computer networks or via the internet. In con-
trast to traditional maps, GIS are updateable, and help to
appropriately target intervention and prevention pro-
grammes, especially in less developed countries [1,2].
Published: 1 July 2008
International Journal of Health Geographics 2008, 7:33 doi:10.1186/1476-072X-7-33
Received: 8 April 2008
Accepted: 1 July 2008
This article is available from: http://www.ij-healthgeographics.com/content/7/1/33
© 2008 Reinhardt et al; licensee BioMed Central Ltd.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0),
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.Page 1 of 7
(page number not for citation purposes)
increasingly rely on geographic information systems
(GIS) to assess outbreaks in real time and to counsel deci-

International Journal of Health Geographics 2008, 7:33 http://www.ij-healthgeographics.com/content/7/1/33
The WHO launched a public health mapping GIS pro-
gramme in 1993 [3]. In addition, the revised WHO Inter-
national Health Regulations (IHR 2005, Annex 1) set out
core capacities, e.g. use of early warning systems and most
efficient information technologies, to be implemented by
member states for improved detection and surveillance of
health threats including meningococcal disease. National
GIS programmes for surveillance of communicable dis-
eases can be accessed via the internet, such as the Swedish
SmiNet, which provides infection epidemiological data in
various formats including a GIS [4]. The German Surv-
Stat@RKI displays data on notifiable diseases in Germany
not only as tables, but also chloropleth maps can be gen-
erated [5]. GIS have also been established to monitor
infectious disease spread and target intervention strategies
in more refined geographic units such as military installa-
tions [6], hospitals [7], or cities [6].
A variety of specialized applications for scientific and pub-
lic health purposes have been introduced. Animated maps
showing geographic movements of influenza waves
throughout a country inform the public about current
trends and foster efforts to promote vaccination [8]. A web
GIS based on open source components has been launched
for Canadian West Nile virus surveillance, which makes
information on dead birds readily available to the private
and public sector [9]. Environmental information can be
integrated into GIS to study environmental triggers and
other risk factors for infectious diseases [10-17]. For this
purpose, satellite imagery and remote sensing have
become increasingly important [18,19]. All these projects
demonstrate the capacity of GIS to integrate, evaluate, and
visualize tremendous amounts of multi-purpose data. GIS
projects are also employing historical datasets to unravel
the transmission patterns of infectious diseases [20]. Fine
typing of pathogens, a method to define groups or even
clones within a given species, can provide an additional
layer of information, as exemplified for avian malaria
[21]. Scan statistics has been used to complement GIS
with regard to spatio-temporal cluster detection [22-24].
Meningococcal disease is caused by the Gram negative
bacterium Neisseria meningitidis, which poses a substantial
threat to childhood health throughout the world [25].
Case fatality rate reaches 10% even in countries with
excellent health care, there is a high rate of sequelae, and
transmission patterns and outbreak sizes are rather unpre-
dictable. Unexpected increases of reported cases of disease
have been observed in several countries, giving rise to tre-
mendous efforts for vaccine development or distribution
[26]. A variety of clonal lineages of meningococci occur
world-wide with varying distribution [27]. Whereas these
lineages are defined by multilocus sequence typing or for-
DNA-sequence typing of variable regions of genes encod-
ing immuno-dominant antigens [29,30]. We recently
used fine typing data to identify clusters of meningococcal
disease using automated scan statistics [31,32]. The com-
bination of reliable DNA-sequence typing with scan statis-
tics provides an unbiased approach for cluster detection
and early warning of public health agencies.
Here we report the development and implementation of a
meningococcal disease GIS, which visualizes the dynam-
ics of the spread of meningococcal disease including
results of scan statistics.
Results
An epidemiological GIS was developed which provides
timely access to laboratory surveillance data on meningo-
coccal disease in Germany, and which supplements statu-
tory meningococcal disease surveillance by the Federal
Health Authorities. All technical components of EpiS-
canGIS are described in detail in the Materials and Meth-
ods section.
EpiScanGIS comprises a data- and a mapserver and a pub-
licly available web server. Data transmission, database
content, and features of visualization were designed to
offer the highest possible level of data protection. Only
registered users may generate enlarged maps of single Fed-
eral States with enhanced resolution, and are allowed to
access disease cluster information. (Currently, 80 users
mainly from the public health sector have been regis-
tered). The architecture of EpiScanGIS is depicted in Fig. 1.
The German reference laboratory for meningococci
(NRZM) provides comprehensive laboratory surveillance
for meningococcal disease in Germany [33]. Data of the
NRZM stored in a local database are synchronized with
EpiScanGIS on a weekly basis. Date of sampling, date of
receipt, geographic coordinates as deduced from the
postal code, age of patient, and fine type of the meningo-
coccus are automatically transferred. An interactive inter-
net application allows users to generate maps dynamically
for answering queries adjusted to an arbitrary period, age
group, county, serogroup or fine type. The system is able
to show different map layers, which display population
density, yearly incidence, cases grouped by serogroup or
fine type, and the results of cluster analysis, in any possi-
ble combination. Additional information on specific
coordinates is extractable by pointing on the respective
map position.
As a tool for laboratory surveillance of infectious diseases,
EpiScanGIS incorporates a defined set of bacterial fine
typing data. Consequently, case information is onlyPage 2 of 7
(page number not for citation purposes)
merly multilocus enzyme electrophoresis [28], fine typing
of meningococci for outbreak detection is achieved by
included, if a complete fine typing dataset is available.
DNA-sequence based typing information was chosen,

International Journal of Health Geographics 2008, 7:33 http://www.ij-healthgeographics.com/content/7/1/33
because of its accuracy and reproducibility [29]. Cur-
rently, more than 2,900 cases have been compiled in the
database, with a total of 612 fine types (as defined by a
unique combination of capsular serogroup, outer mem-
brane protein PorA type and outer membrane protein
FetA type (Figure 2). Figure 3A depicts a typical query for
a specific fine type (B:P1.7-2,4:F1-5 with B, serogroup;
P1., PorA type; F, FetA type) in comparison to any other
serogroup B strain.
In EpiScanGIS, pathogen specific information is embed-
ded in generic attributes. The database does not store a
fine type character as a fixed column with each case
record, but connects the latter with a generic attribute. The
case record itself only contains a minimal demographic
dataset. Specific attributes are generated on the fly upon
data import. Thus, EpiScanGIS is an open platform, which
may be adopted for any other infectious agent.
EpiScanGIS is connected to an automated cluster detec-
tion system. We chose SaTScan for this purpose [34-36],
which we validated recently [31]. SaTScan is employed on
weekly basis; possible spatio-temporal clusters of menin-
gococcal cases caused by identical fine types are visualized
on the maps. EpiScanGIS extracts relevant information
tion of cases occurred. Figure 3B demonstrates the visual-
ization of a cluster. As described recently, cluster analysis
is initiated at the centroids of the 429 rural and urban dis-
tricts, Germany's administrative units.
Rank abundance curve demonstrating the frequency of cases (y-axis) belonging to one of 612 distinct fine typ s, which arestored in the EpiScanGIS databaseFigure 2
Rank abundance curve demonstrating the frequency
of cases (y-axis) belonging to one of 612 distinct fine
types, which are stored in the EpiScanGIS database.
The most frequent fine type is B:P1.7-2,4:F1-5 (n = 328 in
April 2008); 391 of 612 fine types occurred only once
1
10
100
1000
1 24 47 70 93 116 139 162 185 208 231 254 277 300 323 346 369 392 415 438 461 484 507 530 553 576 599
Finetypes ordered by frequency of occurrence (n=612)
N
um
be
r o
f c
as
es
p
er
fi
ne
ty
pe
(n
)
Structure of EpiScanGIS, an online geographical information system for meningococcal disease surveillance in GermanyFigure 1
Structure of EpiScanGIS, an online geographical information system for meningococcal disease surveillance in
Germany. EpiScanGIS generates a Flash-based Rich Internet Application, which is delivered to the user upon connection. The
graphical user interface (GUI) is initialized dynamically depending on the user's security role. The GUI forwards all interactions
to the server-side controller and receives and displays the appropriate map and information. Technical details of the applica-
tion are described in the Materials and Methods section.
www.episcangis.orgStart
SatScan™
module
- Retrieve current data
- Perform weekly scans
- Store new clusters
User Webserver Data- and Mapserver NRZM
UMN MapServer
PostgreSQL
database
Local
database
Case data
Patients age
Coordinates
Finetype
Receipt date
Sample date1010
Create dynamic
client application
depending on
security role
Automated
weekly
export
Interaction
Authentication
101
Maps
QueriesPage 3 of 7
(page number not for citation purposes)
such as number of cases, fine type, county, p-value, and
diameter of the circle, in which the abnormal accumula-
(December 2001 through April 2008).

International Journal of Health Geographics 2008, 7:33 http://www.ij-healthgeographics.com/content/7/1/33
EpiScanGIS enables real-time online animation that can
be used to show time-lapse videos of arbitrary maps with
flexible time-period adjustment. Such animations are
helpful to visualize spatial distribution of cases caused by
a given fine type over time.
Discussion
Meningococcal disease is monitored by national and
international surveillance systems in many parts of the
world. Surveillance is necessary to assess the disease bur-
den and to appropriately allocate resources for preventive
measures, such as vaccine development. Fine typing pro-
cedures are important for monitoring of clonal spread and
distribution, outbreak monitoring, and for the assessment
of vaccine coverage against this highly variable organism
[37].
A variety of characteristics make meningococcal disease an
ideal candidate for GIS applications. Firstly, meningococ-
cal disease is transmitted mostly within the community
via close contacts between humans. Nosocomial trans-
mission is an exceptional observation [32], in contrast to
since a considerable proportion of cases are health care
associated. The social networks, within which meningo-
coccal disease is transmitted, are at least partially under-
stood. Household transmission is of major importance,
given the fact that risk of secondary infection is highest in
this setting [38,39]. Institutional transmission is much
less common, yet can occur in schools [40], and military
camps [41]. Travel associated meningococcal disease has
been reported [42,43], and certainly, detection of travel
associated outbreaks is demanding because the geo-
graphic location of pathogen acquisition frequently
remains obscure. But again, travel associated disease is
rare in comparison to the large number of cases acquired
within other social networks. Secondly, invasive menin-
gococcal disease is rare. Because of the severity of disease,
the high rate of hospitalization, and the relative ease of
disease recognition, underreporting is comparatively low,
and laboratory confirmation is sought for in the majority
of cases. In contrast, other infections such as pneumococ-
cal disease or MRSA-related infections occur at a high fre-
quency with a plethora of disease entities, and in many
countries there is no mandatory notification of cases.
Examples for the use of EpiScanGISFigure 3
Examples for the use of EpiScanGIS. (A) For the years 2004 and 2005, a query was made for a very frequent meningococ-
cal serogroup B fine type (B:P1.7-2,4:F1-5), whose distribution (circles) is compared with all serogroup B cases (triangles) in the
database. Note that serogroup B cases due to this fine type are mostly found in the Western part of Germany. (B) Section of
the EpiScanGIS screen display: a spatio-temporal cluster is depicted by a circle surrounding the location of cases due to a cer-
tain fine type (B:P1.7-2,4:F1-5), which are in close spatio-temporal proximity. A description field provides additional informa-
tion on the cluster. Temporal information has been obscured for the purpose of publication.Page 4 of 7
(page number not for citation purposes)
e.g. Methicillin-resistant Staphylococcus aureus (MRSA), for
which geographic mapping might be much more difficult,
Thirdly, a consensus on molecular typing of meningococ-
cal disease has recently been reached in Europe [29]. Thus,

International Journal of Health Geographics 2008, 7:33 http://www.ij-healthgeographics.com/content/7/1/33
an extension of EpiScanGIS to other countries with a func-
tioning laboratory surveillance of meningococcal disease
put in place will be possible without changing typing
attributes.
The provided GIS solution would not have been possible
without the use of open source software. Using as much
existing tools as possible not only reduces the time of
development, but also increases software stability and
security. Advanced open source projects are characterized
by having many developers and frequent releases,
whereby errors are detected and fixed very fast. Further-
more, components can be used free of charge. Given the
fact that the infectious disease burden in developing
countries with low economic resources is especially high,
open source based systems for epidemiological surveil-
lance may be introduced to assist in prompt identification
of high risk regions and outbreaks. Provided that samples
are linked to geocoding data, EpiScanGIS could easily be
implemented in the African meningitis belt for surveil-
lance projects and outbreak management. Not surpris-
ingly, the use of open source components has found its
way into health geographics [44].
The majority of samples submitted to the reference labo-
ratory are provided by microbiological laboratories,
which send bacterial strains on a voluntary basis. The
results of the reference laboratory are reported back,
which is essential to involve the peripheral laboratories.
However, apart from the capsular serogroup, which might
support individual patient management, the fine typing
data themselves are of no specific interest to the sending
institution. EpiScanGIS for the sender of specimens now
offers the opportunity to query the database and extract
data on the geographical distribution of the reported fine
type, helping the sender to better understand the mission
and benefits of laboratory surveillance. Thus, EpiScanGIS
will help to stabilize the relation between sending labora-
tories and the reference laboratory, and thereby it will
maintain a high coverage of cases by the laboratory sur-
veillance system. Another issue is of similar importance in
this context: weekly written reports of the reference labo-
ratory to the public health offices on newly detected clus-
ters of meningococcal disease need to be supplemented
with geographical maps depicting the location of the clus-
ters and data on the general distribution of the fine type
under investigation. All of this can be accomplished using
EpiScanGIS, and therefore the system has become an
essential part of the reporting workflow.
EpiScanGIS still has limitations that will be subject to
future modifications. EpiScanGIS only accepts fully typed
cases of meningococcal disease. It would be desirable to
Koch-Institute in collaboration with the NRZM currently
develops algorithms to match the datasets of the RKI (stat-
utory notification system) and of the NRZM, which are
not fully overlapping. Inclusion of validated data from the
statutory notification system not present in the NRZM
dataset would be highly desirable to fully represent the
epidemiology of meningococcal disease. At present, in
EpiScanGIS incidence calculations are not possible for
specific fine types, but only for capsular serogroups. A fur-
ther adaptation of the system in this sense would facilitate
comparisons between fine types of interest. Finally, the
space-time scan statistics used and visualized by EpiS-
canGIS could be improved in future versions, i.e. by initi-
ating the spatial scanning window at a more refined
position on the map rather than at the centroid of the
administrative unit, which was decided for to reduce the
computing time, despite of the fact that the coordinates of
the cases are available through the 5-digit postal code.
Furthermore, changing to a flexible shaped scan statistic,
as described recently [45,46], might be beneficial.
In summary, automated computer assisted cluster detec-
tion of meningococcal disease has recently been intro-
duced into the German reference laboratory for
meningococci [31]. The system is now used prospectively
to identify living clusters of disease caused by a single fine
type. Data are reported to the local health offices, the Fed-
eral State offices, and the Robert Koch-Institute. To our
experience cluster reporting fosters the dialogue between
neighbouring public health offices, and initiates exchange
of information between various levels of public health
administration.
Conclusion
EpiScanGIS has become a valuable tool for real time lab-
oratory surveillance of meningococcal disease in Ger-
many with the potential to serve as an early warning
system. It serves as a model for future GIS applications in
infectious disease control. EpiScanGIS has the potential to
be extended to an international level and used for other
infectious diseases. Depending on the application, further
levels of information might be included, such as climate
data, socio-demographic data, and risk factors attributa-
ble.
Methods
System architecture
EpiScanGIS was developed using Java 2 Standard Edition
[47]. A PostgreSQL server [48] provides the relational
database backend. Apache and Jakarta Tomcat web server
[49] deliver the website. MapServer [50] and PostGIS [51]
are responsible for the GIS functionality. The web module
is built using the Struts application framework [52]. OurPage 5 of 7
(page number not for citation purposes)
further include cases in which e.g. for various reasons only
the serogroup could be determined by the laboratory, but
not PorA- and FetA antigen types. Furthermore, the Robert
Flash-based Rich Internet Application (RIA) employs the
OpenLaszlo platform.