Group Title: BMC Health Services Research
Title: Concepts for risk-based surveillance in the field of veterinary medicine and veterinary public health : Review of current approaches
CITATION PDF VIEWER THUMBNAILS PAGE IMAGE ZOOMABLE
Full Citation
STANDARD VIEW MARC VIEW
Permanent Link: http://ufdc.ufl.edu/UF00100016/00001
 Material Information
Title: Concepts for risk-based surveillance in the field of veterinary medicine and veterinary public health : Review of current approaches
Physical Description: Book
Language: English
Creator: Stärk,Katharina
Regula, Gertraud
Hernandez, Jorgé
Knopf, Lea
Fuchs, Klemens
Morris, Roger
Davies, Peter
Publisher: BMC Health Services Research
Publication Date: 2006
 Notes
Abstract: BACKGROUND:Emerging animal and zoonotic diseases and increasing international trade have resulted in an increased demand for veterinary surveillance systems. However, human and financial resources available to support government veterinary services are becoming more and more limited in many countries world-wide. Intuitively, issues that present higher risks merit higher priority for surveillance resources as investments will yield higher benefit-cost ratios. The rapid rate of acceptance of this core concept of risk-based surveillance has outpaced the development of its theoretical and practical bases.DISCUSSION:The principal objectives of risk-based veterinary surveillance are to identify surveillance needs to protect the health of livestock and consumers, to set priorities, and to allocate resources effectively and efficiently. An important goal is to achieve a higher benefit-cost ratio with existing or reduced resources. We propose to define risk-based surveillance systems as those that apply risk assessment methods in different steps of traditional surveillance design for early detection and management of diseases or hazards. In risk-based designs, public health, economic and trade consequences of diseases play an important role in selection of diseases or hazards. Furthermore, certain strata of the population of interest have a higher probability to be sampled for detection of diseases or hazards. Evaluation of risk-based surveillance systems shall prove that the efficacy of risk-based systems is equal or higher than traditional systems; however, the efficiency (benefit-cost ratio) shall be higher in risk-based surveillance systems.SUMMARY:Risk-based surveillance considerations are useful to support both strategic and operational decision making. This article highlights applications of risk-based surveillance systems in the veterinary field including food safety. Examples are provided for risk-based hazard selection, risk-based selection of sampling strata as well as sample size calculation based on risk considerations.
General Note: Start page 20
General Note: M3: 10.1186/1472-6963-6-20
 Record Information
Bibliographic ID: UF00100016
Volume ID: VID00001
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: Open Access: http://www.biomedcentral.com/info/about/openaccess/
Resource Identifier: issn - 1472-6963
http://www.biomedcentral.com/1472-6963/6/20

Downloads

This item has the following downloads:

PDF ( PDF )


Full Text


0
BMC Health Services Research ioed Central



Debate

Concepts for risk-based surveillance in the field of veterinary
medicine and veterinary public health: Review of current
approaches
Katharina DC Stark*1, Gertraud Regulal, Jorge Hernandez2, Lea Knopf1,
Klemens Fuchs3, Roger S Morris4 and Peter Davies5


Address: 'Federal Veterinary Office, CH-3003 Bern, Switzerland, 2College of Veterinary Medicine, University of Florida, Gainsville, FL, USA,
3Institute of Applied Statistics, A-8010 Graz, Austria, 4EpiCentre, Massey University, Palmerston North, New Zealand and 5Department of
Veterinary Population Medicine, St. Paul, MN 55108, USA
Email: Katharina DC Stark* katharina.staerk@bvet.admin.ch; Gertraud Regula gertraud.regula@bvet.admin.ch;
Jorge Hernandez HernandezJ@mail.vetmed.ufl.edu; Lea Knopf- lea.knopf@bvet.admin.ch; Klemens Fuchs klemens.fuchs@joanneum.at;
Roger S Morris R.S.Morris@massey.ac.nz; Peter Davies davieool @umn.edu
* Corresponding author



Published: 28 February 2006 Received: 29 June 2005
BMC Health Services Research2006, 6:20 doi: 10.1 186/1472-6963-6-20 Accepted: 28 February 2006
This article is available from: http://www.biomedcentral.com/1472-6963/6/20
2006Stark 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.



Abstract
Background: Emerging animal and zoonotic diseases and increasing international trade have
resulted in an increased demand for veterinary surveillance systems. However, human and financial
resources available to support government veterinary services are becoming more and more
limited in many countries world-wide. Intuitively, issues that present higher risks merit higher
priority for surveillance resources as investments will yield higher benefit-cost ratios. The rapid
rate of acceptance of this core concept of risk-based surveillance has outpaced the development
of its theoretical and practical bases.
Discussion: The principal objectives of risk-based veterinary surveillance are to identify
surveillance needs to protect the health of livestock and consumers, to set priorities, and to
allocate resources effectively and efficiently. An important goal is to achieve a higher benefit-cost
ratio with existing or reduced resources. We propose to define risk-based surveillance systems as
those that apply risk assessment methods in different steps of traditional surveillance design for
early detection and management of diseases or hazards. In risk-based designs, public health,
economic and trade consequences of diseases play an important role in selection of diseases or
hazards. Furthermore, certain strata of the population of interest have a higher probability to be
sampled for detection of diseases or hazards. Evaluation of risk-based surveillance systems shall
prove that the efficacy of risk-based systems is equal or higher than traditional systems; however,
the efficiency (benefit-cost ratio) shall be higher in risk-based surveillance systems.
Summary: Risk-based surveillance considerations are useful to support both strategic and
operational decision making. This article highlights applications of risk-based surveillance systems
in the veterinary field including food safety. Examples are provided for risk-based hazard selection,
risk-based selection of sampling strata as well as sample size calculation based on risk
considerations.



Page 1 of 8
(page number not for citation purposes)







http://www.biomedcentral.com/1472-6963/6/20


Background
Emerging animal and zoonotic diseases and increasing
international trade have resulted in an increased demand
for veterinary surveillance systems, while human and
financial resources available to support government veter-
inary services are becoming more and more limited in
many countries world-wide. This constrains all areas of
activities of veterinary services, including monitoring and
surveillance programmes. One option to respond to this
situation is to ensure that the most relevant programmes
are maintained. Several government veterinary services,
including those of England and Wales (Anonymous,
2000) and New Zealand (Thornton, 2004) have therefore
undertaken evaluations of their surveillance programmes.
One recommendation of the review in England and Wales
was that a structured approach was needed to determine
priorities for surveillance, and that this approach should
be based on risk (Anonymous, 2000). Consequently, a
method to develop risk profiles for animal diseases is
being developed (Anonymous, 2003). A similar outcome
resulted from the evaluation of surveillance needs in New
Zealand, which stated that a 'risk-based' approach to iden-
tify priority diseases will provide the basis for resource
allocation (Thornton, 2004). These examples illustrate
the background and motivation behind the 'risk-based'
philosophy for veterinary surveillance. The core rationale
underpinning risk-based strategies is that issues that
present higher risks merit higher priority for surveillance
resources as these investments will yield higher benefit-
cost ratios. This axiomatic foundation has led to a rapid
rate of acceptance of the concept of risk-based surveillance
that has outpaced the development of its theoretical and
practical bases. The phrase 'risk-based surveillance' is
increasingly prevalent in government documents of many
countries, being applied across a range of contexts includ-
ing chemical residue monitoring, exotic disease surveil-
lance, or sampling for food safety assurance in general.

Inevitably, because the term 'risk' is defined differently in
various contexts, its use in relation to surveillance has
tended to be vague. The objectives of this article are to
review reports on risk-based surveillance, and to develop
a framework which may hopefully lead to more uniform
use and understanding of both the terminology and its
practical implementation.

Review and discussion
Terminology
In order to discuss 'risk-based' surveillance, it is essential
to clarify the meaning of risk. Risk (as a noun) has numer-
ous common language uses that influence its interpreta-
tion. A short list includes 1) the chance of something
going wrong; 2) any hazardous entity likely to cause
injury, damage, or loss; 3) in insurance, the probability,
amount, or type of possible loss incurred and covered by


an insurer; 4) in finance, the possibility of loss in an
investment or speculation.

Risk has been defined by medical epidemiologists as 'the
probability of disease developing in an individual in a
specified time interval' (Rothman and Greenland, 1998).
In contrast, in the field of risk analysis, risk is defined as
the synthesis of two components, namely the probability
of occurrence of an undesired event and the consequences
or costs of this event (International Animal Health Code).
The Society for Risk Analysis defines risk as 'the potential
for realization of unwanted, adverse consequences to
human life, health, property, or the environment; estima-
tion of risk is usually based on the expected value of the
probability of the event occurring times the consequence
of the event given that it has occurred.' Thus risk may be
viewed as the absolute value of expected losses. We would
argue that the epidemiological definition for risk (proba-
bility only) is less useful in the context of surveillance
than the risk analysis paradigm, because the probability of
occurrence of an adverse event alone is not sufficient to
drive allocation of resources, and expected consequences
of disease events must have substantial influence.

Furthermore, risk is being evaluated in the framework of
risk analysis consisting of risk management, risk assess-
ment and risk communication (International Animal
Health Code). Surveillance itself is an integral component
of both risk management (e.g., as a tool for early detection
of an event) and risk communication. The term 'risk-
based surveillance' could consequently imply that alloca-
tion of surveillance activities is guided by the probability
of events with or without consideration of the conse-
quences of the event; the management of the event; or the
process of communication (including perceptions) of the
event.

The term 'targeted' surveillance has been used to describe
concepts which, from an epidemiological perspective,
could constitute risk-based surveillance. This term was
first used with reference to surveillance programmes of
bovine spongiform encephalopathy (BSE) (Doherr et al.,
2001, 2002; Morignat et al., 2002). The concept of tar-
geted surveillance for BSE was to concentrate sampling for
screening tests on defined sub-populations of cattle that
were expected to have a higher prevalence of the disease
(Doherr et al., 2001). The sub-populations were defined
by age and by so-called 'exit routes', i.e. whether cattle
went to normal slaughter, emergency slaughter or were
cattle that died or were euthanised at the farm. Similarly,
targeted surveillance was defined more generally as sur-
veillance 'focusing the sampling on high-risk populations
in which specific, commonly known risk factors exist'
under the assumption that the event to be surveyed would
be more common in the targeted population than in the


Page 2 of 8
(page number not for citation purposes)


BMC Health Services Research 2006, 6:20







http://www.biomedcentral.com/1472-6963/6/20


general population (Salman et al., 2003). Assuming that
the epidemiological intelligence is correct (i.e. high-risk
groups can be predicted from risk factors), for a fixed sur-
veillance investment, targeted surveillance should yield
both higher sensitivity and higher positive predictive
value than surveillance conducted randomly across the
population. Martin and Cameron (2003) argue that the
assumption in traditional surveillance, that the probabil-
ity of disease is constant across all individuals in the refer-
ence population is not realistic. They propose the
evaluation of surveillance systems using scenario trees.
The nodes of their trees represent factors affecting the
probability of disease occurrence in sub-populations that
may be targeted by surveillance. Elsewhere, targeted sur-
veillance was defined as surveillance "to answer a specific
question about a defined disease or condition using
agreed mechanisms for detection" (Anonymous, 2003), a
definition which does not explicitly incorporate risk-
based concepts.

In the context of food safety, a 'risk-based food safety sys-
tem' has been defined by the SaFoodChain working group
(SaFoodChain website). This definition states that a risk-
based food safety system is 'a system that demonstrates to
consumers and other stakeholders that foods are being
produced under conditions which minimize adverse
health effects. This is achieved by using information on
the nature of human health hazards associated with par-
ticular food products, the likelihood of consumers being
exposed to these hazards, the consequences of exposure
and the capacity of the production and processing system
to mitigate risks arising from hazards which are above a
threshold of concern'. The motivation behind this
approach is to promote efficient risk management and to
facilitate risk communication, primarily in the context of
quality assurance.

Risk-based approaches in disciplines other than veterinary
medicine
Risk-based approaches are widely used in many fields
unrelated to animal health. Particularly, ecologists have
widely adopted risk-based methods. In the management
of environmental pollution risks, the so-called 'risk-based
corrective action' (RBCA, 'Rebecca') has become a stand-
ard (Standard guide for risk-based corrective action). The
formal definition of RBCA is: "A streamlined approach in
which exposure and risk assessment practices are inte-
grated with traditional components of the corrective
action process to ensure that appropriate and cost-effec-
tive remedies are selected, and that limited resources are
properly allocated" (EPA, 1995). In RBCA, contaminated
sites are classified (scored) according to exposure scenar-
ios and consequences to public health. Corrective actions
are then defined for the different site categories. In the cat-
egorisation, four tiers of risk assessment were described


where the level of detail and complexity of the risk assess-
ment increases from tier I to tier IV ("Use of risk-based
decision making").

Modelling approaches have also been used in environ-
mental sciences to predict likely sites of contamination or
spills with the objective to target intervention and to
obtain cost-effective surveillance (Smalley et al., 2000) or
for early warning purposes (Grayman and Males, 2002).
Furthermore, environmental scientists developed sam-
pling methods that take into account risk factors for con-
tamination, specifically in water. Such risk-based
sampling has been described for heavy metals in water
(Preston and Shackelford, 2002), and for watershed mon-
itoring (Foran et al., 2000). Risk-based sampling was also
applied using spatial risk factors (Ericson and Gonzalez,
2003).

In the management of both private and government busi-
nesses, risk-based auditing is a broadly known concept.
This method is primarily used for internal auditing and
qualitative scores are used to categorise activities with
respect to business risks (McNamee, 1997). Managers are
expected to monitor their management practices and
processes with special focus on major business risks, e.g.
market or financial risks. The motivation for this
approach appears to be similar to the motivation behind
risk-based surveillance, namely: "The identification, prior-
itization and sourcing of key organizational risks is critical
to ensuring that internal audit resources are allocated to
the areas that matter most". Risk-based auditing was also
described with relation to auditing of research grants and
auditing of subsidy payments.

Closer to the veterinary field, risk-based concepts are
applied in the surveillance of human diseases. In Italy,
special surveillance programmes were implemented for
workers who are exposed to specific occupational risks
(Franco et al., 2002). The risk categorisation of workers in
this example included both scientific evidence as well as
risk perception. The objective of this approach was early
detection of disease. Similar concepts are applied in
screening programmes focusing on certain age groups, e.g.
mammography in women or prostate cancer in men.

Proposed concepts
In animal health surveillance, decisions at operational or
strategic levels can be distinguished. Strategic decisions
are needed regarding which issues warrant surveillance,
and operational decisions assure the most effective use of
resources based on epidemiological principles. One possi-
ble framework for terminology, in line with previous
authors (Doherr et a., 2001, 2002; Morignat et al., 2002;
Salman et al. 2003), would be for the term 'targeted sur-
veillance' to denote use of epidemiological risk factors to


Page 3 of 8
(page number not for citation purposes)


BMC Health Services Research 2006, 6:20








http://www.biomedcentral.com/1472-6963/6/20


Table I: Steps in the design of veterinary or food safety surveillance programmes, the possible application of risk assessment steps to
obtain risk-based surveillance programmes and the epidemiological contributions providing the basis for risk assessments


Surveillance design
steps

Selection of disease or
agent



Sampling
Selection of strata



Selection of units

Sample size


Risk assessment steps


Hazard identification,
hazard characterisation,
exposure assessment,
consequence assessment


Exposure assessment,
consequence assessment,
risk factors

Not applicable (random
selection)
Release assessment


Epidemiological
contributions


Case reporting, outbreak
investigations, systematic
review



Risk factor studies, models
for population attributable
risk, meta analyses



Random non-risk-based
surveys, cross-sectional
studies


Examples


References


Selection of diseases based
on economic significance
for producers, selection of
zoonotic agents based on
public health significance

Age strata, spatial strata
(regions), product types,
products from certain
producers


Repeated surveys,
confidence in disease
freedom after defined time
periods


Paige et al., 1999; Stark et
al., 2000; Breidenbach et
al., 2004; Brolisauer et al.,
2004


Doherr et al., 2001;
Morignat et al., 2002;
Breidenbach et al., 2004;
Brolisauer et al., 2004


Hadorn et al., 2002a


optimize population sampling at an operational level.
The term 'risk-based surveillance' could be reserved for the
context of strategic decision making that employs princi-
ples of risk analysis in relation to surveillance of animal
disease.

An alternative, that we recommend, is that risk-based sur-
veillance be defined as a more inclusive term in which
both the epidemiological and risk analysis perspectives
are integrated. Based on the definition of risk and the
review of the use of the term 'risk-based' in various fields,
we propose to define risk-based veterinary surveillance as
follows:

A surveillance programme in the design of which exposure and
risk assessment methods have been applied '. .rii. I with tradi-
tional design approaches in order to assure appropriate and
cost-effective data collection.

This relatively broad definition allows for the application
of risk assessment approaches at all steps of the design of
surveillance systems for early detection and management
of diseases or hazards of interest. Table 1 illustrates the
different steps during design where surveillance can be
based on risk assessment principles. These options for
risk-based design will be illustrated in the later sections of
this article.

Considering the existing applications of risk-based veteri-
nary surveillance programmes together with applications
of risk-based methods in other fields, the objectives of the
risk-based approach can be derived. The principal objec-
tives of risk-based veterinary surveillance are:


* to identify surveillance needs
livestock and consumers


to protect the health of


* to set priorities,

* to allocate resources effectively and efficiently. An
important goal is to achieve a higher benefit-cost ratio
with existing or reduced resources.

In risk-based systems, public health, economic, and trade
consequences of diseases play an important role in selec-
tion of diseases or hazards. Evaluation of risk-based sur-
veillance systems shall prove that the efficacy of the risk-
based approach is equal or higher than that of traditional
surveillance; however, the efficiency (cost-benefit) shall
be higher in risk-based systems.

The design of risk-based surveillance systems requires
prior, epidemiological knowledge on, e.g., the difference
in occurrence of disease between population strata or the
influence of risk factors. This type of information cannot
be generated by risk-based surveillance systems them-
selves, but needs to be obtained using traditional, quanti-
tative epidemiological approaches (Table 1).

Risk-based hazard selection
When surveillance programmes have broad objectives,
the list of candidate hazards to be considered can be
extensive and selection of hazards becomes critical. For
example, the European Union requests that trading part-
ners conduct monitoring of zoonotic pathogens in milk
and milk products (Directive 92/46/EEC). The lists of dif-
ferent milk products on the market and potential hazards



Page 4 of 8
(page number not for citation purposes)


BMC Health Services Research 2006, 6:20







http://www.biomedcentral.com/1472-6963/6/20


are very long. However, some agents cannot survive the
processing steps employed for specific milk products. For
example, Brulisauer et al. (2004) showed that surveillance
of zoonotic bacteria in hard cheeses was not cost-effective
because the probability of survival of these pathogens
through the production process was extremely small.
Other surveillance objectives with extensive lists of haz-
ards include surveillance for chemical residues in animal-
derived food and the monitoring of antimicrobial resist-
ance in bacteria of animal origin. In both instances, risk-
based hazard selection will be required to assure the most
important hazards are included. Paige et al. (1999)
described how risk profiles were used for substance selec-
tion in the national residue monitoring programme in the
U.S.A. Ledergerber (2005) used risk profiling and risk
ranking to prioritise combinations of bacteria and antimi-
crobial resistance which should be included in a classifica-
tion scheme to identify livestock herds that harbour
'important' antimicrobial resistance patterns. For exam-
ple, fluoroquinolone resistance in Salmonella and cepha-
losporine resistance in E. coli were ranked higher than
ampicilline resistance in E. coli or Salmonella. This rank-
ing was based on estimated health implications of these
resistance patterns when transferred to humans.

The identification and selection of hazards are key steps in
the design of surveillance systems, and they will be guided
by the objectives of the surveillance. In risk assessment,
the equivalent steps are hazard identification and hazard
characterisation (Table 1). Consequences of hazard expo-
sure should also be included in the characterisation of a
hazard. A limitation in this respect is the lack of informa-
tion on the nature and size of consequences. In the review
of surveillance programmes in England and Wales, it was
recommended to use a structured and transparent
approach for hazard selection (Anonymous, 2000). Haz-
ard selection was previously conducted with help of deci-
sion trees or flow charts in the context of serological
surveillance programmes for exotic diseases in pigs in
Denmark and Switzerland (Stark et al., 2000; Hadorn et
al., 2002b). The proposed decision tree included hazard
characteristics such as prevalence, performance of diag-
nostic tests, domestic economic implications and interna-
tional trade implications. Epidemiological data on case
frequencies due to specific hazards and results of outbreak
investigations provide important input to the hazard
identification step.

Risk-based selection of population strata
Once it has been decided which hazards should be
included in a surveillance programme, the next step is the
sampling design. In non-risk-based random sampling, all
units in a population have the same probability to be
selected (Cameron et al., 2003). Limitations of this
approach have been identified by several authors. Key


concerns are the economic viability of such surveillance
programmes when the prevalence of the hazard
approaches zero (Paisley, 2001; Doherr and Audige,
2001; Webb et al., 2001). The cornerstone of risk-based
sampling strategies is stratification of the target popula-
tion into categories that display heterogeneity in probabil-
ity of harbouring a hazard, or heterogeneity in the severity
of consequence if the hazard is present. The probability of
individual units within strata to be sampled can neverthe-
less be calculated, which makes this type of sampling a
special case of probability sampling and different from
non-probability or purposive sampling as described by
Cameron et al. (2003). Strata to be used for risk-based
sampling are derived from epidemiological studies assess-
ing the probability of occurrence of the hazard in the stra-
tum (e.g. herds having imported animals from abroad,
herds with specific risk factors) and/or the consequence of
the occurrence of the hazard in this stratum (e.g. poten-
tially infected animals in sale yards or markets). Prior
information on the probability of hazard occurrence in a
strata may originate from a range of quantitative epidemi-
ological studies, e.g., risk factor studies. The risk of a stra-
tum as part of a risk-based surveillance design can be
derived qualitatively (a risk score or risk category) or
quantitatively using risk assessment methods. Results of
meta analyses can support this step. Similar to this
approach, Thurmond (2003) described a sampling strat-
egy referred to as "proportional risk sampling". In propor-
tional risk sampling, the sample size in different strata
reflects the respective risk level.

Unit selection within a risk stratum is done at random and
all aspects of random sampling are applicable (Cameron
et al., 2003). Issues to be considered in non-risk-based
sampling such as test performance and clustering (Martin
et al., 1992; Cameron and Baldock, 1998a and 1998b,
Ziller et al., 2002) are equally relevant in risk-based sam-
pling.

Many sampling schemes in surveillance programmes that
use at first sight non-risk-based designs, turn out to use
some kind of risk-based strata. For example, samples for
laboratory testing are often collected from specified age
groups of animals. The reason for this lies in the epidemi-
ology of infectious diseases in populations. For example,
older animals are more likely to have antibodies against a
hazard (and therefore, to show positive serological
results) because they were exposed to a hazard if it was
present. This makes this age group a 'high-probability'
group according to our definition. More explicit stratum
definitions can be found in protocols for BSE surveillance
(see above; Doherr et al., 2001; Morignat et al., 2002).

A more complex approach to stratum definition which
takes probability and public health consequence into


Page 5 of 8
(page number not for citation purposes)


BMC Health Services Research 2006, 6:20







http://www.biomedcentral.com/1472-6963/6/20


account to monitor foreign substance levels in imported
food of animal origin was described by Breidenbach et al.
(2004). They developed a qualitative risk scoring system
that took into account the residue legislation and surveil-
lance programmes in the country of origin, the history of
residue detection in commodities of the country of origin,
the type of commodity (i.e. risk of accumulation of sub-
stances), and the public health significance of a substance.
Combining these factors, a cumulative, relative risk score
was calculated and all substance-commodity-country
combinations with the highest scores were included in the
programme.

Another way to define risk strata is the incorporation of
spatial information. If the underlying phenomena show
spatial dependency, point pattern analysis (Diggle, 1983),
lattice data (Besage, 1974) or geostatistical methodology
(Cressie, 1991) can be used to identify spatial risk factors.
Fuchs et al. (2004) applied spatial point pattern analysis
to study the resistance behaviour of Enterococcus to tetra-
cycline in bulk milk and to identify high density and low
density areas. To evaluate space-time interaction of sca-
bies in Styrian chamois, Fuchs and Deutz (2002) used
geostatistical methods to find critical distances. Wagner
and Fuchs (2005) designed a risk-based sampling plan on
the basis of spatial logistic regression for the BHV-1 mon-
itoring in a free region.

Risk-based sample size calculation
If repeated surveys to document disease freedom are con-
ducted, risk assessment methods can be useful to estimate
the likely change in the probability of disease presence
since the past survey. In using a risk-based approach, the
sample size calculation takes into account the level of con-
fidence that remains from previous survey. The necessary
sample size for an upcoming survey is estimated based on
the difference to the targeted level of confidence. This
approach was proposed by Cannon (2001) and was
applied in the context of surveillance to demonstrate free-
dom from disease by Hadorn et al. (2002a). In this appli-
cation, a risk assessment was conducted to estimate the
probability of hazard introduction (e.g. enzootic bovine
leucosis) during a one-year period between two consecu-
tive surveys. Introduction pathways considered were
importation of live animals, migratory wildlife or vectors.
It was shown that the required sample size for consecutive
surveys could be reduced by 60-80 % (Hadorn et al.,
2002a, Hadorn et al., 2002b). This reduction in sample
size was primarily due to a limited number of live animals
being imported from countries with a lower disease status.
The accuracy of the estimated sample size does, however,
depend on the completeness of the risk assessment. The
latter should not only include the probability of hazard
introduction but also the consequences of residual, non-
detected infection in the country. Additionally, the


"aging" of information originating from passed surveys
needs to be considered. An improved risk model is being
developed to address these issues (Knopf, personal com-
munication).

Analysis and integration of data from risk-based surveys
Data from risk-based surveillance can be analysed using
similar analytical techniques as in non-risk-based surveys
(see, for example, Dargatz and Hill, 1996; Audige and
Beckett, 1999; Suess et al., 2002), provided that risk
assessment approaches are applied to hazard selection
only and extrapolation of results to other hazards is not
required. However, when population strata have been
selected for sampling based on risk considerations, extrap-
olation of results to the general population is not possible
unless the risk difference between high-risk and low-risk
strata is known. This, however, may not be easy to assess.
Martin and Cameron (2003) propose the use of scenario
trees to compare the sensitivity of random surveillance
systems with systems conducted in high-disease-probabil-
ity populations. They define the sensitivity ratio as a meas-
ure to assess representativeness vs. the level of targeting in
a surveillance system.

Furthermore, in international trade, the question of
equivalence between sampling designs may become an
issue. Trade between countries of equivalent animal
health status does not pose a problem, but participating
countries need to document their health status. However,
when one country is using risk-based sampling and the
other is not, they may operate surveillance systems at dif-
ferent confidence levels of case detection. If risk-based sur-
veillance works as expected, a country using the risk-based
approach should be more likely to detect positive samples
than if using non-risk-based surveillance. This may lead to
trade restrictions and other negative consequences. How
can the results of one system be compared with the other?
If this problem is not resolved, it is likely to pose a major
barrier to the adoption of risk-based surveillance in rela-
tion to trade.

Another important issue is the integration of data from
different sources, risk-based and non-risk-based.
Although Cannon (2001) proposed an approach for use
of data from different sources, data from risk-based sur-
veillance designs were not considered. One data source
may be historical data. The issue of weighting of historic
data and the quantification of aging of data are unre-
solved (Thurmond, 2003). Graphical methods have also
been proposed to integrate data from different sources
(Hueston and Yoe, 2000), but reports of successful appli-
cations of this approach are lacking. In the context of BSE
surveillance, where data is being collected from different
risk-based population strata, the need of integration of
surveillance results became evident (Ducrot et al., 2003).


Page 6 of 8
(page number not for citation purposes)


BMC Health Services Research 2006, 6:20








http://www.biomedcentral.com/1472-6963/6/20


Summary
The need for efficient and cost-effective surveillance sys-
tems will induce the use of risk-based selection of hazards
and population strata, and risk-based sample size calcula-
tion. Risk-based surveillance systems offer a more efficient
approach for early detection and management of diseases.
However, these innovative methods can only be estab-
lished if there is international agreement on the method-
ology for risk-based surveillance, and the interpretation of
its results. There is an urgent need for the development
and evaluation of standardised, internationally accepted
methods for risk-based surveillance.


Competing interests
The authors) declare that they have no competing inter-
ests.


Authors' contributions
KS, GR, RM and LK developed the proposed concepts for
risk-based surveillance. KS and PD were involved in draft-
ing the article. KF contributed aspects on spatial risk. All
authors read, critically revised and approved the final
manuscript.


Acknowledgements
We thank all members of the SaFoodChain group for valuable discussions
on risk-based approaches to food safety.

References
I. Anoymous: Veterinary surveillance in England and Wales: A
review. In Report Ministry of Agriculture, Fisheries and Food, Lon-
don; 2000.
2. Anoymous: Partnership, priorities and professionalism: A
strategy for enhancing veterinary surveillance in the UK. In
Report Department for Environment, Food and Rural Affairs, London;
2003.
3. Audig6 L, Beckett S: A quantitative assessment of the validity of
animal-health surveys using stochastic modelling. Prey Vet
Med 1999, 38:259-276.
4. BesageJE: Spatial interaction and the spatial analysis of lattice
systems. journal of the Royal Statistical Society 8 1974, 36:192-225.
5. Breidenbach E, Sievi M, Weber U, Heiz H, Stark KDC: Fundamental
principles for risk-based planning of random sampling to
trace hazard in imported meat. Proceedings 5th World Congress
Foodborne Infections and Intoxications; 7- I 1 June 2004; Berlin 2004 in
press.
6. Brulisauer F, Berger T, Klein B, Danuser J: Risk based surveillance
of milk and dairy products. Proceedings International ConferenceVet-
erinary Public Health and Food Safety Towards a Risk based Chain Con-
trol: 22-23 October 2004; Rome. Istituto Zooprofilattico Sperimentale
delle Regioni Lazio e Toscana 2004:50-51.
7. Cameron AR, Baldock FC: Two-stage sampling in surveys to
substantiate freedom from disease. Prey Vet Med 1998,
34:19-30.
8. Cameron AR, Baldock FC: A new probability formula for sur-
veys to substantiate freedom from disease. Prey Vet Med 1998,
34:1-17.
9. Cameron A, Gardner I, Doherr MG, Wagner B: Sampling consid-
erations in surveys and monitoring and surveillance systems.
In Animal disease surveillance and survey systems methods and applica-
tions Edited by: Salman MD. Ames: Iowa State Press; 2003:47-66.
10. Cannon RM: Sense and sensitivity-designing surveys based on
an imperfect test. Prev Vet Med 2001, 49:141-163.
II. Cressie NAC: Statistics for Spatial Data New York, Chichester.
Toronto: John Wiley & Sons; 1991.


12. Dargatz DA, Hill GW: Analysis of survey data. Prev Vet Med 1996,
28:225-237.
13. Diggle PJ: Point pattern analysis London: Academic Press; 1983.
14. Doherr MG, Heim D, Fatzer R, Cohen CH, Vandevelde M, Zur-
briggen A: Targeted screening of high-risk cattle populations
for BSE to augment mandatory reporting of clinical sus-
pects. Prev Vet Med 2001, 51:3-16.
15. Doherr MG, Hett AR, Cohen CH, Fatzer R, Rufenacht J, Zurbriggen
A, Heim D: Trends in prevalence of BSE in Switzerland based
on fallen stock and slaughter surveillance. Vet Rec 2002,
150:347-348.
16. Doherr MG, Audig6 L: Monitoring and surveillance for rare
health-related events: a review from the veterinary perspec-
tive. Philos Trans R Soc Lond B Biol Sci 2001, 356:1097-1106.
17. Ducrot C, Roy P, Morignat E, Baron T, Calavas D: How the surveil-
lance system may bias the results of analytical epidemiolog-
ical studies on BSE: prevalence among dairy versus beef
suckler cattle breeds in France. Vet Res 2003, 32:185-192.
18. EPA: Use of risk-based decision-making in UST Corrective
Action Programs. Report 1995 [http://www.epa.gov]. U.S. Environ-
mental Protection Agency, Office of Solid Waste and Emergency
Response (OSWER) Directive 9610.17. Environmental Protection
Agency
19. Ericson JE, Gonzalez EJ: Hierarchical sampling of multiple
strata: an innovative technique in exposure characteriza-
tion. Environ Res 2003, 92:221-231.
20. Foran J, Brosnan T, Connor M, Delfino J, DePinto J, Dickson K, Hum-
phrey H, Novotny V, Smith R, Sobsey M, Stehman S: A Framework
for Comprehensive, Integrated, Watershed Monitoring in
New York City. Environ Monit Assess 2000, 62:147-167.
21. Franco G, Cella MT, Tuccillo E, Ferrari F, Minisci E, Fusetti L: From
risk-based health surveillance to health promotion: an evi-
dence-based experience in a health care setting. Int J Occup
Med Environ Health 2002, 15:1 17-120.
22. Fuchs K, Deutz A, Wagner P, Kofer J: Spatial point pattern anal-
ysis to study the resistance behaviour of enterococcus to tet-
racycline in bulk milk. In Proceedings of the SVEPM: 24 26 March
2004 Edited by: Reid SWJ, Menzies FD, Russell AM. Martigny, Swit-
zerland; 2004:134-142.
23. Fuchs K, Deutz A: Use of variograms to detect critical spatial
distances for the Knox's test. Prev Vet Med 2002, 45:37-45.
24. Grayman WM, Males RM: Risk-based modeling of early warning
systems for pollution accidents. Water Sci Technol 2002,
46:41-49.
25. Hadorn DC, RufenachtJ, Hauser R, Stark KDC: Risk-based design
of repeated surveys for the documentation of freedom from
non-highly contagious diseases. Prey Vet Med 2002, 56:179-192.
26. Hadorn DC, Hauser R, Stark KDC: Epidemiologische Grundla-
gen und Resultate der Stichprobenuntersuchung 2001 in der
schweizerischen Schweinepopulation. Schweiz Arch Tierheilk
2002, 144:532-541.
27. Hueston WD, Yoe CE: Estimating the overall power of com-
plex surveillance systems. In Proceedings of the 9th International
Symposium on Veterinary Epidemiology and Economics: 6-11 August,
2000 Edited by: Salman MD, Morley PS, Ruch-Gallie R. Breckenridge,
Colorado; 2000:758-760.
28. International Animal Health Code [http://www.oie.int]. Office
International des Epizooties, Paris
29. Ledergerber U: Foundation of a herd classification scheme by
risk profiling and risk ranking of antimicrobial resistance in
the food chain. Report 2005 [http://www.dfvf.dk/
Default.asp?lD=9726]. International EpiLab, Copenhagen
30. Martin SW, Shoukri M, Thorburn MA: Evaluating the health sta-
tus of herds based on tests applied to individuals. Prev Vet Med
1992, 14:33-43.
31. Martin PAJ, Cameron A: Documenting freedom from avian
influenza. In Report International EpiLab, Copenhagen; 2003.
32. McNamee D: Risk-based auditing for internal auditors 1997 [http://
www.mc2consulting.com]. Mc2 Management Consulting
33. Morignat E, Ducrot C, Roy R, Baron T, Vinard JL, Biacabe AG, Madec
JY, Bencsik A, Debeer S, Eliazsewicz M, Calavas D: Targeted sur-
veillance to assess the prevalence of BSE in high-risk popula-
tions in western France and the associated risk factors. Vet
Rec 2002, 151:73-77.





Page 7 of 8
(page number not for citation purposes)


BMC Health Services Research 2006, 6:20








http://www.biomedcentral.com/1472-6963/6/20


34. Paige JC, Chaudry MH, Pell FM: Federal surveillance of veteri-
nary drugs and chemical residues (with recent data). Veteri-
nary Clinics of North America: Food Anim Pract 1999, I 5:45-60.
35. Paisley LG: Economic aspects of disease monitoring with spe-
cial reference to bovine paratuberculosis. Acta Vet Scand
200 1:17-25.
36. Preston BL, Shackelford J: Risk-Based Analysis of Environmental
Monitoring Data: Application to Heavy Metals in North
Carolina Surface Waters. Environ Manage 2002, 30:279-293.
37. Risk Analysis Glossary [http://www.sra.org/
resources glossary p-r.php]
38. Rothman KJ, Greenland S: Modern epidemiology 2nd edition. Philadel-
phia: Lippincott Williams & Wilkins; 1998.
39. SaFoodChain the food safety consortium [http://
www.safoodchain.org]
40. Salman MD, Stark KDC, Zepeda C: Quality assurance applied to
animal disease surveillance systems. Rev Sci Tech OIE 2003,
22:689-696.
41. SmalleyJB, Minsker BS, Goldberg DE: Risk-based in situ bioreme-
diation design using a noisy genetic algorithm. Water Ressour
Res 2000, 36:3043-3052.
42. Standard guide for risk-based corrective action E2081-00
[http://www.astm.org].
43. Stark KDC, Mortensen S, Olsen AM, Barfod K, Botner A, Lavritsen
DT, Strandbygard B: Designing serological surveillance pro-
grammes to document freedom from disease with special
reference to exotic viral diseases of pigs in Denmark. Rev Sci
Tech OIE 2000, 19:715-724.
44. Suess EA, Gardner IA, Johnson WO: Hierarchical Bayesian
model for prevalence inferences and determination of a
country's status for an animal pathogen. Prev Vet Med 2002,
55:155-171.
45. Thornton R: Ambitious domesticated animal surveillance
review initiated. Biosecurity Issue 2004, 5 1:6-7.
46. Thurmond MC: Conceptual foundations for infectious disease
surveillance. j Vet Diagn Invest 2003, 15:501-514.
47. Use of risk-based decision making [http://www.epa.gov/
swerustl /directiv/od961017.htm#Whatis]
48. Wagner P, Fuchs K: Sampling plan for the BHV-I monitoring in
a free region (in German). Proceedings of the 5th International Sym-
posium on BHV I-, BVD- and Paratuberculosis Eradication: 9-11 March
2005; Stendal, Germany 2005 in press.
49. Webb CR, Wilesmith JW, Simmons MM, Hoinville LJ: A stochastic
model to estimate the prevalence of scrapie in Great Britain
using the results of an abattoir-based survey. Prev Vet Med
2001, 51:269-287.
50. Ziller M, Selhorst T, Teuffert J, Kramer M, Schluter H: Analysis of
sampling strategies to substantiate freedom from disease in
large areas. Prev Vet Med 2002, 52:333-343.

Pre-publication history
The pre-publication history for this paper can be accessed
here:


httn://www.biomedcentral.com/1472-6963/6/20/nrenub


Page 8 of 8
(page number not for citation purposes)


Publish with BioMed Central and every
scientist can read your work free of charge
"BioMed Central will be the most significant development for
disseminating the results of biomedical research in our lifetime."
Sir Paul Nurse, Cancer Research UK
Your research papers will be:
available free of charge to the entire biomedical community
peer reviewed and published immediately upon acceptance
cited in PubMed and archived on PubMed Central
yours you keep the copyright
Submit your manuscript here: BioMedcentral
http://www.biomedcentral.com/info/publishing adv.asp


BMC Health Services Research 2006, 6:20




University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - - mvs