l The detection methodology must possess known and quantified sensitivity,
specificity, repeatability and reproducibility.
l The target bacterial and host populations must be precisely defined.
l The sampling framework must be fully specified, indicating how the samples are
selected from the bacterial or host populations or the environment, including the
various levels of organisation within these populations or ecosystems and the
number of units from which samples are selected.
The distinction between clinical resistance (which is related to pathogen suscep-
tibility, PK, and the approved dosage regimen) versus epidemiological cut-off
values (which is purely a function of pathogen susceptibility) is fundamental to
how we consider resistance (Bywater et al. 2006 ; Simjee et al. 2008 ; Turnidge and
Paterson 2007 ). Accordingly, the definition of antibiotic resistance will vary as a
function of study objective. As such distinctions are frequently difficult to discern
and/or disentangle; there is a critical need to establish a common language and
common standards. This is particularly relevant when comparing resistance rates
within a global context. For example, similar to the previously mentioned EMEA
1999 report, resistance is defined both from a clinical and an epidemiological
perspective in the United States Food and Drug Administration (FDA) as it applies
to the term “resistance” in their (FDA/CVM Guidance#152). Within that guidance
document, the FDA states that “human exposure through the ingestion of antimi-
crobial resistant bacteria from animal-derived foods represents the most significant
pathway for human exposure to bacteria that have emerged or have been selected as
a consequence of antimicrobial drug use in animals”. From this, it would appear
that both clinical (human exposure through the ingestion of antimicrobial resistant
strains) and epidemiological (have emerged or have been selected as a consequence
of antimicrobial drug use in animals) interpretations have been applied within the
context of this guidance.
In addition to the challenge of defining “resistance”, we need to consider the
source of resistance data, which in greater part, comes from surveillance studies.
The World Health Organisation (WHO) published a document entitled, “Surveil-
lance Standards for Antimicrobial Resistance” (WHO 2001 ), which states that for a
variety of reasons, “data obtained from clinical sources are generally unrepresenta-
tive of the totality of disease within a population”. The report emphasised the
importance of understanding the relationship between the surveyed populations in
the context of the wider population. As part of this discussion it was accepted that
across a global environment, there are inconsistencies in the submission, analysis
and use of microbiological specimens. Furthermore, the report points to the limited
epidemiological relevance of the resulting conclusions unless the susceptibility data
are linked to disease incidence.
While this WHO report focused primarily on surveillance data generated within
human populations, the same principles apply to animal surveillance systems. As
corroborated by Franklin et al. ( 2001 ) in a report prepared by the Office Interna-
tional des Epizooties (OIE) ad hoc group of experts on antimicrobial resistance, “it
should be borne in mind when designing resistance monitoring and surveillance
236 M. Martinez and P. Silley