DEVELOPMENT OF A WEB-AVAILABLE EPIDEMIOLOGICAL

SURVEILLANCE SYSTEM INTEGRATING

GEOGRAPHIC INFORMATION

The Public Health Emergencies Support System at

the Portuguese General Directorate for Health

CCCCCCCCCCAndré Oliveira and Pedro Cabral CCCCCCCCCCCC

Instituto Superior de Estatística e Gestão de Informação (ISEGI)

Universidade Nova de Lisboa, 1070-312 Lisboa, Portugal

Keywords: Systems Integration, Internet, Geographic Information Systems, Epidemiological Surveillance.

Abstract: The application of geographic information tools in Public Health management already includes many areas

of study, one of which deals with the integration of Geographic Information Systems (GIS) in

epidemiological surveillance systems, with the objective of aiding Public Health officials in decision-

making. Some of these systems are already operational in several countries, acting in various spatial and

temporal scales, and with different levels of priority. The present article introduces the development of a

Public Health spatial data management infrastructure within the Portuguese General Directorate of Health,

baptized Public Health Emergencies Support System and essentially aimed at performing epidemiological

surveillance tasks. This is a multiplatform environment that brings together relational databases, geographic

information systems and web technology, making it possible to supply daily and weekly updated results to

health officials through the Internet. Satisfactory results were obtained with the implementation of SSESP,

since most of the planned infrastructure and functionalities are already operational. Some of the system’s

present handicaps and evolutionary perspectives are also discussed.

1 INTRODUCTION

There is an increasing conscientiousness about the

importance of detecting and interpreting spatial

distribution patterns of public health-related

phenomena, which is closely associated to an array

of new technologies recently made available to

researchers and professionals operating in this field

of expertise. Amongst these technologies,

Geographic Information Systems (GIS) occupies a

special place, and its contribution is already felt in

several public health application domains. These

applications range from understanding the

importance of environmental factors in the etiology

of certain diseases, to improving the efficiency of

delivering health care services to populations, where

they contribute to optimal location models for health

infrastructures, just to cite the most relevant.

These case studies and applications can be

included in the context of Health Geography, a

recent field of study (formally recognized in 1992).

This scientific area constitutes a methodological

platform, integrating and articulating knowledge

from the Earth, Social and Health Sciences, that

works as a holistic tool to study current problems at

different scales of analysis (Santana, 2005). Since

this is a broad area of studies, it can be further

divided in two main domains: Disease Geography

and Geography of Health Care Systems. According

to Boulos et al. (2001), Disease Geography deals

with the exploration, description and modeling of

the spatiotemporal incidence

of diseases and related

environmental phenomena, detection and analysis of

disease patterns and clusters

, the analysis of the

causes and the generation of new hypothesis. As for

Geography of Health Care Systems, it deals with the

planning, management and delivery of adequate

health services, providing, amongst other things,

adequate patient access to these services, after

determining the health needs of target communities

in terms of medical care as well as the service areas

238

Oliveira A. and Cabral P.

DEVELOPMENT OF A WEB-AVAILABLE EPIDEMIOLOGICAL SURVEILLANCE SYSTEM INTEGRATING GEOGRAPHIC INFORMATION - The Public Health Emergencies Support System

at the Portuguese General Directorate for Health.

DOI: 10.5220/0001822502380243

In Proceedings of the Fifth International Conference on Web Information Systems and Technologies (WEBIST 2009), page

ISBN: 978-989-8111-81-4

of health services, which include health promotion

and disease prevention activities.

Deeply related with these areas is another more

specific investigation and application domain known

as Public Health Surveillance. Lawson (2006)

defines this activity as the systematic and continuous

retrieval of health data for analysis and

interpretation, for planning, implementation and

evaluation of public health practices, as well as the

dissemination of these data to public health

authorities in appropriate time, so that it can

ultimately be used in prevention and control. This

activity is operated through dedicated systems,

currently known as Epidemiological Surveillance

Systems, which main task is to survey the eventual

outburst of epidemics

. There are already multiple

systems of this kind in present days, and they act in

multiple spatial and temporal scales, and with

different priority degrees. Amongst several other

examples is the Integrated System for Public Health

Monitoring of the West Nile Virus (Gosselin et al.,

2006) operating in the province of Quebec, Canada

and EpiGIScan (Reinhardt et al., 2008), dedicated to

real-time surveillance of Meningococcal Meningitis

in Germany.

The eminent threat of an influenza pandemic

potentially caused by an eventual mutation of the

H5N1 virus, making it transmissible between

humans, has triggered the development of strategies

and information systems to deal with this situation.

In this context, the development of epidemiological

surveillance systems integrating several types of

technologies such as relational database systems,

geographic information tools and web information

distribution technologies has recently known

increased interest and investment. In fact, bringing

these technologies together makes it possible to

combine the power of massive health data

processing with the ability of detecting spatial

patterns of disease distribution and also

communicate decision-making critical information

to public health officials in due time. These

capabilities are decisive in our days, since

globalization has a controversial effect on world

population health; on one hand, the fast pace of the

global movement of people and goods are the recipe

for rapid spread of diseases and vectors

, increasing

pandemic

risk; on the other hand, information and

communication technologies are allowing health

authorities to improve analysis methods and tools,

transmit valuable information and, creating networks

that can signal alerts and allow health officials to

coordinate efforts on unprecedented geographic and

temporal scales.

This situation was the driving factor for the

development of an epidemiological surveillance

system and public health data management

infrastructure, physically located within the

Portuguese General Directorate for Health (DGS),

baptized Public Health Emergencies Support System

(SSESP).

This article aims to introduce the system, in its

implementation, architecture, functionalities and

future perspectives. Thus, the second section

introduces SSESP development phases, starting with

planning and moving from conceptual and logical

modeling to physical implementation, including

present system architecture and capabilities. The

third and final section discuss system´s strengths and

limitations, projected but not yet implemented

functionalities and draws on future prospects and

recommendations

2 SYSTEM DEVELOPMENT

The initial step in putting the system together was to

perform a state of the art analysis of the

epidemiological surveillance carried on within DGS.

During this phase, public health officials,

epidemiologists and information systems

professionals were consulted, previously existing

systems were identified and observed, and a cost-

benefit analysis was performed, in an effort to

identify the necessary functionalities to be

implemented, which, in turn, would influence the

physical architecture of the system. As the outcome

of this first phase, functional specifications were

listed, which implied essentially to build upon some

of the existing capabilities and also to prepare some

totally new ones.

In fact, prior to the development described here,

a semi-automatic epidemiological surveillance

system, Hospital and Health Centers Emergency

Information System (SIU) was already operating in

DGS. It consisted of an Oracle 9i database, into

which data regarding episodes of emergency

services demand in public hospitals (SONHO

system) and health centers (SINUS system) of the

Portuguese mainland, was automatically transferred

on a daily basis and stored. This database was, in

turn, connected via Open Database Connectivity

(ODBC) to a Microsoft Access 2003 database, were

data processing was performed, manually started on

a daily basis by a person in charge. This person was

also in charge of using Microsoft Excel 2003 to

DEVELOPMENT OF A WEB-AVAILABLE EPIDEMIOLOGICAL SURVEILLANCE SYSTEM INTEGRATING

GEOGRAPHIC INFORMATION - The Public Health Emergencies Support System at the Portuguese General Directorate

for Health

239

produce charts and tables containing surveillance

information and, finally, Microsoft Word 2003 and a

File Transfer protocol (FTP) connection to update a

dedicated website with these products, where they

were displayed and made available to health

officials. This previous system was considered

inadequate to current needs, and it was thus decided

that: (1) the whole apparatus of daily manual

processing of data should be replaced by an entirely

automatic one; (2) the current supporting

infrastructure should combine storage and

simultaneous processing of alphanumeric and

geographic data; (3) all daily alphanumeric data

processing tasks should be executed within a robust

relational database management system; (4) the

existing website should be replaced by a new one

containing all the previous products, plus

geographical display capabilities. It should also bear

more usability (Nielsen and Loranger, 2006) to the

final user and be updated on a daily basis.

After the definition of the system guidelines, the

next stride was to define the specifications of the

information to be produced. It was decided that the

system should produce and display daily demand

numbers, separated per three main groups (hospital

emergency services, health centers emergency

services and both). Each group should be further

separated into: Portuguese mainland and per

District; all ages and also per age groups (0-14, 15-

24, 25-44, 45-64 and more 65 years old). The

information should be presented using line charts,

tables and thematic maps to monitor daily evolution

Figure 2: Information filtering options.

Figure 1: System Architecture.

WEBIST 2009 - 5th International Conference on Web Information Systems and Technologies

240

and spatial distribution of the demand. It was further

decided that epidemiological alert levels were to be

analyzed on a daily and weekly basis. Thus, weekly

alert levels should be displayed using line graphs; as

for daily alert levels, the system should be capable

of sending alert messages automatically through

electronic mail, after previously comparing demand

levels with previously established reference

thresholds. From this point on, the system´s

architecture was modeled, as well as each

component separately, and it was also decided which

types of data should be included. As the final result

of this process, an integrated environment was

obtained, as depicted in figure 1. The essential

entrance point to the system is a first server

(collection server) incorporating an Oracle 9i

database that daily accesses the various local servers

at hospitals and health centers automatically and

collects stored demand episodes.

This is a Sun SPARC II server operating on Sun

Solaris 8, were all code and mechanisms regulating

access to hospitals and health centers of the SONHO

e SINUS systems have been programmed. The core

of the architecture consists in two servers, identical

in hardware and operative system (Intel Dual Xeon

processor, 10 Gigabytes RAM memory, 300

Gigabytes hard disk storage capacity, operative

system Microsoft Windows 2003 Server R2 32 bits).

One of these is used as database server, containing

an Oracle 10g (10.2.0.1) database, which does the

massive part of the data processing, since demand

data is automatically transferred here, after being

gathered by the collection server, and then processed

in a daily sequence of scheduled tasks. The second

server (applications server) is the webserver, since it

includes all software applications to generate and

manage the website, and also contains the

geographic information system. The following

software applications are installed in it:

- ESRI ArcGIS Server 9.2: supports the WebGIS

component of SSESP; in other words, supports the

displaying of geographic contents online.

- ESRI ArcGIS 9.2: supports the preparation of

thematic maps and spatial analysis;

- Microsoft Visual Studio Professional 2005:

programming environment for creating the website;

- Microsoft Internet Information Services (IIS) 6.0:

Webserver software.

The website is the front-end of the system, and it

contains all final products except alerts sent via e-

mail. It was programmed in C# language and can be

accessed through DGS website, where there are two

levels of access, public and restricted to public

health officials. Database tables containing

surveillance information are permanently connected

to the website that displays the final products; these

are automatically updated once daily data processing

is complete. All products refer to the events of the

previous day.

After the first page of the website, the second

page contains user-available surveillance

information, and is divided in two functional parts.

The upper part is used to select the desired product,

and the lower part displays the result. As it can be

seen on figure 2, the upper part contains several

options groups that filter the information to be

displayed. From left to right, it is possible to choose

between hospitals, health centers or both; between

district or national results; all ages or age groups;

and finally deciding if the information should be

visualized as line charts, tables containing raw

demand values or a map, which is displayed in both

public and restricted areas. This is a choroplet map

that illustrates demand at the district level, using an

interval scale of 5 classes displayed as graduated

colors (figure 3).

Figure 3: Daily demand map.

Values represented on maps are a ratio, being the

crude demand values divided per 100.000

inhabitants, a common epidemiological measuring

unit (Beaglehole et al., 1993). These authors also

point out that the main factors in epidemiological

analysis are time (regarding the time frame into

which events take place), space (in the geographic

sense) and the affected persons (their number and

characteristics). Therefore, monitoring daily events

is insufficient, if not combined with the analysis of

the temporal evolution of demand, which is assured

in SSESP using line charts. The public part of the

website contains line charts displaying the evolution

of demand on a national scale (hospitals + health

centers) for the current year, as well as lines

representing evolution in the years of 2003 and

2005, which are considered a valid referential. The

DEVELOPMENT OF A WEB-AVAILABLE EPIDEMIOLOGICAL SURVEILLANCE SYSTEM INTEGRATING

GEOGRAPHIC INFORMATION - The Public Health Emergencies Support System at the Portuguese General Directorate

for Health

241

restricted area presents line charts bearing the same

information at district level.

However, to be efficient, epidemiological

surveillance needs to employ mechanisms enabling

it of detecting outbreaks

as early as possible, so that

investigation and control actions can be taken

timely. One critical question is the choice of demand

thresholds beyond which an alarm situation should

be triggered. To do this, SSESP uses the endemic

corridors methodology of Bortman (1999), which

plots four areas on a chart. These are classified,

according to the gravity of the situation, as “success

zone”, “security zone”, “alert zone” and “epidemic

zone”. A line, representing weekly demand, is

plotted over these areas. Based upon this method,

SSESP supplies two products: a national weekly

endemic corridor (figure 4), and the automatic

emission of alerts through e-mail, evaluated on a

daily basis with a daily corridor which is not

graphically represented. These two products are

meant to be used together: every time an alert is sent

the evolution of the weekly line in figure 4 should be

continuously observed, since a consistent increase in

its height and angle can be an epidemic.

Figure 4: Weekly endemic corridor.

Finally, the tables including crude demand

values include values regarding current and previous

year. This is a complementary form of analysis,

since sometimes the analysis of the remaining media

is insufficient. Additionally, this tabular data can

also be used to support other types of statistical

studies in a different environment.

3 CONCLUSIONS

This is a system under development that has already

been beneficial to public health decision support in

several occasions. Most of the planned architecture

is completed and operational being the main

implementation objectives attained.

This infrastructure presents several advantages.

The use of a user-friendly website that does not

require any technical knowledge of databases or GIS

applications enables seamless dissemination of vital

information to final users. It also features low cost to

final users, since no application has to be installed

locally and only a computer and Internet connection

are necessary. Although its main use is surveillance,

it is able to support epidemiological background

studies, since it constitutes a Data Warehouse to all

collected surveillance data.

Some additional features are planned. One of

these would be to enable users to remotely introduce

epidemiological data directly into the system, such

as the point location of a disease outbreak,

contaminated well or a new health service. The

systems’ geographical database includes total digital

aerial orthophotographic coverage of the Portuguese

mainland, as well as road network cartography,

administrative divisions, point location of hospitals

and health centers, amongst others. These several

datasets can be combined in a map window that

helps the user locating the desired event, a dialog

box allows him to classify the event and finally a

simple mouse click on the location transfers map

coordinates and associated attributes to a table on

the database for later use. Another planned feature is

to implement automatic alert production at the

district scale, since the present national coverage is

sometimes insufficient. Plus, alert messages should

also be sent via short message service (SMS), in

addition to the present use of e-mail. Another related

feature would be to produce a map depicting the

daily alert level, since no illustration of the daily

alert level is presently produced.

Besides some communication problems and

system architecture details that can easily be solved,

the most crucial limitation lies in the data it receives.

No data is received concerning the diagnosis

associated to each emergency service demand

episode, only the motive (e.g. disease, accident),

which is still too general. This is a critical limitation

to the analysis capabilities, and there are already

several epidemiological surveillance systems

worldwide that make use of this kind of information,

such as GeoMedStat (Li et al., 2006), which uses the

International Classification of Diseases, ninth

revision (ICD-9) to encode each demand episode,

with very interesting results in outbreak surveillance

and detection. Other critical question is the fact that

SSESP does not receive data from all services;

several hospital emergency services are still outside

it. This is a crippling factor, since data from all

emergency services in the country should be

WEBIST 2009 - 5th International Conference on Web Information Systems and Technologies

242

received, in order to provide the full picture of the

daily health situation in Portugal, not a partial one.

The final picture is optimistic, though. Although

a certain array of problems subsist, much was

already achieved, and there is certainly a will to go

forward in system development, which implies a

constant effort to correct present insufficiencies and

adding new functionalities.

ACKNOWLEDGEMENTS

The authors would also kindly like to thank SSESP

project team members Mário Carreira (Coordinator),

Carla Cardoso (Statistician), Tatiana Plantier and

Ricardo Henriques (WebGIS developers).

REFERENCES

Beaglehole, R., Bonita, R. & Kjellström, T. (1993) Basic

Epidemiology, World Health Organization.

Bortman, M. (1999) Elaboracion de corredores o canales

endémicos mediante planillas de cálculo. Revista

Panamericana de Salud Pública, 5, 1-8.

Boulos, M., Roudsari, A. & Carson, E. (2001) Health

Geomatics: an enabling suite of technologies in health

and healthcare. Journal of Biomedical Informatics, 34,

195-219.

Colin, P. (2005) Dictionary of Medical Terms London, A

& C Black.

Cromley, E. & Mclafferty, S. (2002) GIS and Public

Health New York, The Guilford Press.

Epa (2008) Thesaurus of Terms Used in Microbial Risk

Assessment Environmental Protection Agency.

Gosselin, P., Lebel, G., Rivest, S. & Douville.Fradet, M.

(2005) The Integrated System for Public Health

Monitoring of West Nile Virus (ISPHM-WNV): a

real-time GIS for surveillance and decision-making.

International Journal of Health Geographics, 4:21.

Lawson, A. (2006) Statistical methods in spatial

epidemiology, West Sussex, John Wiley & Sons.

Li, H., Faruque, F., Williams, W. & Finley, R. (2006)

Real-Time Syndromic Surveillance. ArcUser

Magazine, 9 (1):17-19.

Nielsen, J. & Loranger, H. (2006) Prioritizing Web

Usability, Berkeley, New Riders.

Reinhardt, M., Elias, J., Albert, J., Frosch, M., Harmsen,

D. & Vogel, U. (2008) EpiScanGIS: an online

geographic surveillance system for meningococcal

disease. International Journal of Health Geographics,

7:33.

Santana, P. (2005) Geografias da saúde e do

desenvolvimento, evolução e tendências em Portugal

Coimbra, Edições Almedina.

Incidence corresponds to the number of episodes of a specific

disease happening in a population in a specific time period

(Beaglehole et al. 1993).

Atypical concentrations of public health events in time and

space (Cromley and McLafferty, 2002).

An Epidemic is the occurrence of an abnormally high number of

cases in a specific area or population, considering the area or the

time frame (Beaglehole et al. 1993).

Insect or other animal carrying a disease susceptible of being

transmitted to human beings (Colin, 2005).

An epidemic that occurs globally, or over a very wide area,

spreading through international borders, generally affecting a

large number of persons (EPA, 2008).

The sudden occurrence of several episodes of a specific disease

(Colin, 2005).

DEVELOPMENT OF A WEB-AVAILABLE EPIDEMIOLOGICAL SURVEILLANCE SYSTEM INTEGRATING

GEOGRAPHIC INFORMATION - The Public Health Emergencies Support System at the Portuguese General Directorate

for Health

243