EMA 发布兽药 VGVP 信号管理模块修订草案并公开征求意见
Draft guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal Management - Revision 1
EMA 于 2026 年 9 月 18 日启动 VGVP 信号管理模块修订草案(EMA/522332/2021 Rev.1)的公开征求意见,截止 12 月 18 日。草案由 CVMP 于 9 月 10 日通过发布征求意见,内容覆盖信号检测、优先级排序、验证、评估、结果及风险最小化措施。该模块适用于欧盟各类审批程序授权的兽药及已登记的顺势疗法兽药,供上市许可持有人与主管当局履行药物警戒义务时使用。
材料列出兽药信号管理模块修订草案从信号检测到风险最小化措施的框架及征求意见安排,便于读者定位需核对的章节。
PDF 文字版;图形和原始排版请参阅官方 PDF。
第 1 页
1 18 September 2026
2 EMA/522332/2021 Rev.1
3 Committee for Veterinary Medicinal Products (CVMP)
4 Guideline on veterinary good pharmacovigilance practices
5 (VGVP) Module: Signal Management
6 Draft
Date coming into effect of first version 28 January 2022
Draft Revision 1 agreed by PhVWP-V 14 August 2026
Draft Revision 1 adopted by CVMP for release for consultation 10 September 2026
Start of public consultation 18 September 2026
End of consultation (deadline for comments) 18 December 2026
7
8
9
Comments should be provided using this template. The completed comments form should be sent to
vet-guidelines@ema.europa.eu
10
Keywords Veterinary pharmacovigilance; signal management; Regulation (EU)
2019/6; Union pharmacovigilance database
11
Official address Domenico Scarlattilaan 6 ● 1083 HS Amsterdam ● The Netherlands
Address for visits and deliveries Refer to www.ema.europa.eu/how-to-find-us
Send us a question Go to www.ema.europa.eu/contact Telephone +31 (0)88 781 6000 An agency of the European Union
© European Medicines Agency, 2026. Reproduction is authorised provided the source is acknowledged.
第 2 页
12 Guideline on veterinary good pharmacovigilance practices
13 (VGVP) Module: Signal Management
14 Table of contents
15 1. Introduction (background) ...................................................................... 6
16 1.1. Definitions ........................................................................................................... 7
17 2. Signal Management process .................................................................. 11
18 2.1. Signal detection ................................................................................................. 13
19 2.1.1. Signal triggers................................................................................................. 14
20 2.1.2. Methods for signal detection ............................................................................. 16
21 2.2. Signal prioritisation............................................................................................. 23
22 2.2.1. Factors to consider during prioritisation .............................................................. 23
23 2.2.2. Impact on future prioritisation during ‘routine monitoring’ .................................... 32
24 2.3. Signal validation ................................................................................................. 33
25 2.3.1. Methods for signal validation ............................................................................. 34
26 2.3.2. Outcome of signal validation ............................................................................. 40
27 2.3.3. Signal prioritisation after signal validation........................................................... 41
28 2.4. Signal Assessment.............................................................................................. 41
29 2.4.1. Broader search for data .................................................................................... 42
30 2.4.2. Assessment of case series ................................................................................ 46
31 2.4.3. Investigation of biological mechanism ................................................................ 56
32 2.4.4. Special types of signals .................................................................................... 58
33 2.4.5. Risk characterisation ........................................................................................ 60
34 2.4.6. Overall conclusion............................................................................................ 65
35 2.5. Outcomes .......................................................................................................... 66
36 2.5.1. Signal refuted ................................................................................................. 66
37 2.5.2. Signal under close monitoring ........................................................................... 67
38 2.5.3. Signal requiring risk minimisation measures ....................................................... 68
39 2.6. Risk minimisation measures................................................................................. 68
40 2.6.1. Amendment of the product information .............................................................. 68
41 2.6.2. Additional risk communication ........................................................................... 69
42 2.6.3. Batch recall, suspension, withdrawal or revocation of the marketing authorisation ... 70
43 2.6.4. Other risk minimisation measures ...................................................................... 70
44 3. Reporting to the EU network ................................................................. 72
45 3.1. Signal management reporting obligations .............................................................. 72
46 3.1.1. Annual statements ........................................................................................... 72
47 3.1.2. Signal submissions .......................................................................................... 74
48 3.1.3. Pharmacovigilance Alerts .................................................................................. 76
49 3.2. Other reporting obligations .................................................................................. 76
50 3.2.1. Reporting of sales data and calculation of dose factor........................................... 76
51 3.2.2. Notification on regulatory measures from non-EU regulatory agencies/Information
52 from other regulatory authorities worldwide ................................................................. 77
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 2/79
第 3 页
53 4. Targeted signal management by the competent authorities and the
54 Agency ...................................................................................................... 78
55 5. Documentation requirements related to the Signal Management Process
56 .................................................................................................................. 78
57
58
59
60
61
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 3/79
第 4 页
62 Tabulated list of abbreviations
63
64 Abbreviation Term
65 ABON Categories of the causality classification system for veterinary pharmacovigilance
66 AE Adverse Event
67 AER(s) Adverse Event Report(s)
68 ALP Alkaline phosphatase
69 ALT Alanine transaminase
70 AMR Antimicrobial Resistance
71 AR Assessment report
72 ATCvet code Standardised alphanumerical identifier used to classify veterinary medicinal products
73 according to the Anatomical Therapeutic Chemical classification system for
74 veterinary medicinal products
75 BPG Best Practice Guide
76 CAP(s) Centrally Authorised Product(s)
77 CI(s) Confidence Interval(s)
78 CIR Commission Implementing Regulation
79 CVMP Committee for Veterinary Medicinal Products
80 DaHPC(s) Direct Animal Healthcare Professional Communication(s)
81 DCP Decentralised Procedure
82 DLP Data Lock Point
83 DWH Data Warehouse
84 EEA European Economic Area
85 EMA European Medicines Agency
86 EU European Union
87 EVV EudraVigilance Veterinary
88 GABA Gamma-Aminobutyric Acid
89 GI Gastrointestinal
90 HCP Healthcare Professional
91 HLT High Level Term (VeDDRA hierarchy)
92 IC025 Information Component (lower bound of the 95% confidence interval)
93 IgE Immunoglobulin E
94 IRIS EMA’s online platform for submitting and managing regulatory and scientific
95 procedures for medicines throughout their lifecycle
96 IT Information technology
97 LEE Lack of Expected Efficacy
98 LLT Low Level Term (VeDDRA hierarchy)
99 MAH(s) Marketing Authorisation Holder(s)
100 MDR1 mutation Multidrug Resistance Gene Mutation 1 (also known as ABCB1 gene mutation)
101 MIT(s) Medically Important Term(s)
102 MRP Mutual Recognition Procedure
103 NAP National Authorisation Procedure
104 NCA National Competent Authority
105 NOS Not otherwise specified (suffix to VeDDRA LLTs)
106 NSAID Nonsteroidal Anti-Inflammatory Drug
107 PCR Polymerase Chain Reaction
108 PD Pharmacodynamics
109 PhV Pharmacovigilance
110 PhVWP-V Pharmacovigilance Working Party – Veterinary
111 PI Product information
112 PK Pharmacokinetics
113 PL Package leaflet
114 PMSS Post-Marketing Surveillance Study
115 PRR Proportional Reporting Ratio
116 PT Preferred Term (VeDDRA hierarchy)
117 QPPV Qualified Person for Pharmacovigilance
118 ROR Reporting Odds Ratio
119 SDR(s) Signal(s) of Disproportionate Reporting
120 SOC System Organ Class (VeDDRA hierarchy)
121 SPC Summary of Product Characteristics
122 SRP Subsequent Recognition Procedure
123 TSM Targeted signal management
124 TTO Time-to-Onset
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 4/79
第 5 页
125 UPD Union Product Database
126 UPhV-Database Union Pharmacovigilance Database
127 VeDDRA Veterinary Dictionary for Drug Regulatory Activities
128 VICH International Cooperation on Harmonisation of Technical Requirements for
129 Registration of Veterinary Medicinal Products
130 VGVP Veterinary Good Pharmacovigilance Practices
131 VMP Veterinary Medicinal Product
132 VSM Veterinary signal management
133 WHO–UMC World Health Organization – Uppsala Monitoring Centre
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 5/79
第 6 页
134 1. Introduction (background)
135 This module of the guideline on veterinary good pharmacovigilance practices (VGVP) brings together
136 general guidance for marketing authorisation holders (MAHs), national competent authorities and the
137 European Medicines Agency (the ‘Agency’) regarding signal management for veterinary medicinal
138 products (VMPs) authorised in the European Union (EU).
139 The objectives of this module are:
140 • To provide general guidance on scientific and quality aspects of signal management for VMPs;
141 • To describe the roles, responsibilities and procedural aspects of the EU signal management
142 process for VMPs.
143 This module is applicable to VMPs authorised in the EU irrespective of the authorisation procedure
144 (centralised or national procedure, including mutual recognition, decentralised and subsequent
145 recognition procedures) and registered homeopathic VMPs.
146 This guidance is applicable to both MAHs and competent authorities in the fulfilment of their
147 pharmacovigilance obligations and provides a regulatory framework against which pharmacovigilance
148 inspections are conducted, ensuring consistent implementation, oversight and verification of
149 compliance. This module should be read in conjunction with Regulation (EU) 2019/6 of the European
150 Parliament and of the Council of 11 December 2018 on veterinary medicinal products and Commission
151 Implementing Regulation (EU) 2021/1281, laying down rules for the application of Regulation (EU)
152 2019/6 of the European Parliament and of the Council as regards good pharmacovigilance practice and
153 on the format, content and summary of the pharmacovigilance system master file for veterinary
154 medicinal products (the Implementing Regulation).
155 The Veterinary Good Pharmacovigilance Practices (VGVP) Module on Signal Management has been
156 revised based on the collective experience gained from all stakeholders involved in the signal
157 management process. It aims to provide a structured and practical framework for the identification,
158 evaluation and management of safety signals related to VMPs. The use of artificial intelligence tools in
159 signal management will be covered in a separate guidance document.
160 This revision aligns with international standards and the Veterinary Union Pharmacovigilance Database
161 – Best Practice Guide (BPG) and supports a harmonised, transparent and scientifically sound approach
162 to signal management. It also integrates the legal framework established by Regulation (EU) 2019/6,
163 particularly Article 81(2), which outlines the obligation to take appropriate action when new
164 information affects the benefit-risk balance of a VMP.
165 A key concept in this module is the ‘effect on the benefit–risk balance’, defined as any new
166 information, data or scientific interpretation that may influence the established understanding of a
167 product’s safety or efficacy profile 1. When such information emerges, MAHs are required to evaluate its
168 potential impact on the benefit–risk balance and determine whether regulatory or other actions are
169 warranted.
170 This module outlines the key steps in signal management, from detection to decision-making and
171 provides guidance on how to document and communicate findings effectively. This revision reflects
172 current best practices and incorporates insights from real-world pharmacovigilance activities. It
173 supports a proactive, consistent and scientifically sound approach to signal management. The module
1 See also: Guideline on the evaluation of the benefit-risk balance of veterinary medicinal products
(EMA/CVMP/248499/2007 – Rev.1)
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 6/79
第 7 页
174 emphasises the importance of clinical judgement, qualitative data review and transparent decision-
175 making throughout the signal evaluation process.
176 The term ‘clinical signs’ is used in this document in line with general understanding/ common
177 terminology and in the context of this guideline also refers to symptoms reported in humans.
178 Examples are provided in grey boxes throughout the text to aid understanding. They are used only to
179 illustrate complex concepts and should never be considered an exhaustive list of possible scenarios.
180 1.1. Definitions
1 ABON system A causality assessment system previously used in European
veterinary pharmacovigilance to classify the likelihood of a causal
association between a veterinary medicinal product and an adverse
event.
2 Adverse event Adverse events are defined in Article 73(2) of Regulation (EU)
2019/6. The term adverse event implies that it is suspected. ‘Adverse
event’ may also be used in veterinary pharmacovigilance to describe
events included in the product information.
3 Amendment of the A type of regulatory action to the authorised product information to
product reflect updated knowledge, including additions, modifications or
information removal of existing information as authorised by the competent
authorities.
4 Annual Statement A submission made in IRIS at least annually. It is the practical
solution for MAHs to comply with the requirement set out in
Regulation (EU) 2019/6 Art. 81(2) to ‘record, at least annually, all
results and outcomes of the signal management process, including a
conclusion on the benefit-risk balance’.
5 Case series A group of individual case reports related to the same or similar event
and product, analysed collectively.
6 Causal association The likelihood that a veterinary medicinal product caused an adverse
event or contributed significantly to the occurrence of it.
7 Causality A structured evaluation of the likelihood that a veterinary medicinal
assessment product caused or contributed to an adverse event.
8 Close monitoring Enhanced pharmacovigilance activities applied to a signal for which
the available evidence remains insufficient to warrant implementation
of risk minimisation measures at this point in time, but which is
subject to continued assessment to determine the need for future
action and where new information is expected within a reasonable
time frame. Deliberate effort is made to seek additional information in
order to reach a conclusion. Submission of updates to a signal under
close monitoring is expected at least annually. See section 2.5.2.
9 Concurrent Medicinal products that are administered or used at the same time or
products in close temporal proximity and that may contribute to or influence
the occurrence of an adverse event. See also co-reported product.
10 Confounding A variable that may influence the observed association between a
factor product and an adverse event.
11 Co-reported Medicinal products that are reported in the same adverse event
products report, regardless of their temporal relationship or suspected
involvement. See also concurrent products.
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 7/79
第 8 页
12 Data lock point The point in time up to which data is included in a signal assessment
or the last date covered by an annual statement.
13 Dechallenge The withdrawal of a veterinary medicinal product, or in some cases a
reduction in dose, following the occurrence of the adverse event of
interest.
14 Disproportionality A method used to identify whether an adverse event is reported more
in adverse event frequently for a specific product or substance than would be
reporting expected, typically based on comparison with other products in a
database. Many different methods are available to determine
disproportionality in AE reporting.
15 Dose–response When a higher dose or prolonged treatment leads to an increased
relationship occurrence of the event, or to increased severity of the event.
16 Educational Targeted materials designed to inform and guide users or animal
material healthcare professionals to reduce risks and ensure the safe and
appropriate use of a veterinary medicinal product.
17 Follow-up A subsequent assessment of a signal under close monitoring, where
assessment of a newly available information is assessed in context of the cumulative
signal data.
18 Inflated When a bias falsely increase disproportionality, inflating it above the
disproportionality true level of disproportionality.
19 Lack of expected A situation in which a veterinary medicinal product does not achieve
efficacy (LEE) the intended therapeutic effect under the approved conditions of use,
as reflected by treatment failure or insufficient clinical response.
20 Masking of When a bias falsely lowers disproportionality, masking the true level
disproportionality of disproportionality.
21 Medically The MIT list comprises VeDDRA PTs that correspond to medical
Important Terms conditions which are commonly involved in suspected adverse events
(MIT) across different pharmacological or therapeutic classes, or which are
considered medically important independently of any established
causal association with a veterinary medicinal product, due to their
seriousness or potential impact on animal health or welfare, public
health, user safety or the environment.
22 Medically linked VeDDRA terms that are conceptually or clinically linked to the same
VeDDRA term adverse event or underlying condition, representing different
manifestations, stages, complications or descriptors of a shared
pathophysiological process.
23 New aspect of a Information relating to a known adverse event, such as changes in
known risk frequency, severity, outcome, affected population or conditions of
occurrence, which is not described in the currently approved product
information.
24 New risk An adverse event or risk associated with a veterinary medicinal
product which is not described in the currently approved product
information.
25 Non-validated A signal for which the signal validation process could not verify that
signal the available data was of sufficient quality and substance and/or
justify further investigation of the possibility of a causal association
between the product(s)/substance(s) and the suspected adverse
event(s) or where the events are considered adequately described in
the product information.
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 8/79
第 9 页
26 Pharmacovigilance Any suspected unexpected serious risk which may require urgent
Alert regulatory action and communication by the MAH, or the competent
authorities as referred to in Article 129, Article 130 and Article 134 of
Regulation (EU) 2019/6.
27 Rechallenge The reintroduction or dose increase of a veterinary medicinal product
after a dechallenge to assess whether the adverse event reoccurs or
worsens.
28 Refuted signal A validated signal for which, following further assessment, a causal
association is not supported by the available evidence at the time of
assessment.
29 Regulatory action An intervention or decision arising from signal assessment and the
evaluation of the benefit–risk balance, intended to address identified
or potential risks associated with a veterinary medicinal product and
to ensure its safe and effective use. A regulatory action can be
initiated by the MAH or the responsible competent authority.
30 Reporting An estimate of the frequency of an adverse event, based on number
incidence of reports in relation to estimated exposure (in accordance with the
guideline on calculation of dose factor).
31 Reporting odds A quantitative measure used in pharmacovigilance to detect
ratio (ROR) disproportionate reporting, comparing the odds of reporting a specific
adverse event for a given product with the corresponding odds for all
other products in the database.
32 Risk based An approach in which the methodology, extent and frequency of
approach pharmacovigilance activities are determined based on factors such as
product characteristics, time on the market, safety profile and
identified and potential risks and the need for additional information.
33 Risk The process of describing the nature, severity, frequency and
characterisation conditions of occurrence of a risk based on the totality of available
evidence.
34 Risk minimisation Interventions intended to reduce the occurrence, severity or impact of
measures adverse events associated with a veterinary medicinal product
through risk communication or other actions.
35 Routine The ongoing and systematic surveillance of safety information for a
pharmacovigilance veterinary medicinal product as part of standard pharmacovigilance
monitoring activities, aimed at detecting, evaluating and managing potential
risks.
36 Severity The intensity of an adverse event, reflecting the degree of deviation
from normal clinical status (e.g. mild, moderate or severe), without
implying its medical significance or outcome.
37 Signal Information that arises from one or multiple sources, including
observations and studies which suggests a potentially new causal
association, or a new aspect of a known causal association between
an intervention and an adverse event or a set of related adverse
events, that is judged likely to justify further investigation of possible
causality (Art. 1(c) of CIR (EU) 2021/1281).
‘New’ should be understood as compared to the currently approved
product information.
38 Signal assessment The process of evaluating a validated signal through a comprehensive
and structured review of all available data in order to assess the
likelihood of a causal association and to support decision-making on
appropriate action, if necessary.
39 Signal detection The process aimed at identifying new potential signals through
monitoring of post-authorisation safety data to identify patterns which
could suggest a new potential risk or a new aspect to a known risk.
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 9/79
第 10 页
40 Signal of Signals of disproportionate reporting (SDRs) refer to statistical
disproportionate associations between veterinary medicinal products and suspected
reporting (SDR) adverse events. It is important to separate the concepts of
‘disproportionate reporting’ in general from the specific concept of an
SDR. The difference is that the latter reaches a pre-defined ‘level of
disproportionality’. The criteria used to define an SDR are often
specific to the database used for signal detection and the method
used to calculate disproportionality.
41 Signal outcome The final conclusion of signal assessment reflecting the outcome and
the subsequent proposed actions or follow-up, if necessary.
42 Signal A continuous process aiming to evaluate the potential impact of a
prioritisation signal in comparison to other signals in order to identify relevance,
determine urgency and perform efficient signal management through
a risk-based approach.
43 Signal A validated signal for which signal assessment supports a likely causal
requiring risk association between the veterinary medicinal product and the adverse
minimisation event. The signal is considered to warrant risk minimisation
measures measures.
44 Signal trigger The source from which a potential signal originates, initiating further
evaluation through the signal management process.
45 Signal validation An initial screening of the data that triggered the potential signal,
with the aim of determining whether the signal is supported by
information of sufficient quality and substance to justify further
assessment. Therefore, signal validation is not intended to determine
causality.
46 Supportive case A case report that, based on assessment of relevant clinical and
temporal factors, provides evidence supporting a causal association
between an adverse event and a veterinary medicinal product.
47 Temporal When the reported event(s) occurred after the time of exposure to
association the veterinary medicinal product(s).
48 Time-to-onset The time interval between exposure to a veterinary medicinal product
(TTO) and the onset of an adverse event.
49 Validated signal A signal is considered validated when the signal validation process
has verified that the data triggering the signal is of sufficient quality
and substance to justify a need for further analysis and the event(s)
of the signal are not already adequately described in the product
information.
181
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 10/79
第 11 页
182 2. Signal Management process
183 A ‘signal’ is defined as information that arises from one or multiple sources, including observations and
184 studies which suggests a potentially new causal association, or a new aspect of a known causal
185 association between an intervention and an adverse event or a set of related adverse events, that is
186 judged likely to justify further investigation of possible causality 2.
187 The ‘signal management process’ is defined in Article 4(41) of Regulation (EU) 2019/6; in practice, this
188 is the collective term used to describe the steps of signal detection, signal prioritisation, signal
189 validation, signal assessment and recommendation for action (risk mitigation) as a whole and the
190 outcome should be documented 3. It is a structured process that should enable continuous monitoring
191 of the benefit-risk balance of a product 4 and therefore should be conducted with a sufficient frequency
192 to ensure timely identification, assessment and management of potential risks related to the use of
193 VMPs.
194 A risk-based approach shall be applied to determine the methodology, extent and frequency of the
195 signal management process 5. The frequency of monitoring should be proportionate to the identified
196 risk4. Use of a risk-based approach means that the following key factors should be taken into account:
197 type of product, length of time on the market, stability of the pharmacovigilance profile, identified and
198 potential risks and the need for additional information5. The rationale for the chosen approach shall be
199 documented3 (see also section 5).
200 Application of a risk-based approach is not only fundamental but also dynamic throughout the signal
201 management process. Rather than applying a fixed set of criteria, a risk-based approach involves
202 continuously evaluating the context and evolving nature of the signal. By integrating the above-
203 mentioned key factors, the process ensures that clinical relevance and potential impact on public
204 health guide prioritisation and decision-making. This flexible strategy allows for more efficient use of
205 resources, focusing efforts on signals that pose the greatest potential risk or require urgent attention.
2 Article 1(c) of the Commission Implementing Regulation (CIR) EU 2021/1281
3 Article 17(1) of CIR (EU) 2021/1281
4 Article 77(4) of Regulation (EU) 2019/6
5 Article 17(3) of CIR (EU) 2021/1281
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 11/79
第 12 页
206
207 Figure 1: Signal management process
208 Throughout all phases of signal management, clinical judgement plays a critical role. It is essential not
209 only in interpreting data but also in selecting what data may support a potential signal. The ability to
210 discern meaningful patterns and assess clinical relevance is key to ensuring that signals are neither
211 overlooked nor overestimated.
212 Signal Detection:
213 The detection phase aims to identify new potential signals. It focuses on identifying new or changed
214 risks through detection of abnormal patterns or monitoring of new information. Signal detection should
215 be objective and non-biased and based on previous knowledge of the product and the pharmacological
216 and biological characteristics of the active substance. It should ideally be based on a combination of
217 quantitative (e.g. disproportionality, number of cases, number of reacting animals) and qualitative
218 (e.g. review of case narratives, literature reviews) methods.
219 Signal Prioritisation:
220 Signal prioritisation is an integral part of the signal management process and may be performed
221 continuously throughout the entire process. Prioritisation aims to evaluate the potential impact of a
222 signal in comparison to other signals in order to identify the most relevant signals, determine their
223 urgency and perform efficient signal management through a risk-based approach.
224 Signal Validation:
225 Signal validation aims to determine if additional analysis of a potential signal is warranted. It focuses
226 on determining whether the potential signal is supported by data of sufficient quality and substance to
227 justify further assessment.
228 Signal Assessment:
229 Signal assessment aims to decide whether any action is necessary. Through a broader and deeper
230 analysis of all available data pertaining to the signal, focus is placed on understanding the causal
231 relationship between the potential risk and the VMP and further characterisation of the risk.
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 12/79
第 13 页
232 Outcomes and Risk minimisation measures:
233 This stage focuses on determining how to manage the potential risk and deciding on the optimal course
234 of action. It involves analysing the potential risk in relation to the benefits of the product,
235 recommending appropriate risk minimisation measures, if necessary, as well as documenting the
236 outcome of the signal management process.
237
238
239 Figure 2: Flow diagram of signal management process
240 2.1. Signal detection
241 Signal detection is the process of monitoring post-authorisation safety data for information that could
242 suggest a new potential risk or a new aspect to a known risk. Potential risks may be related to animal
243 or public health or to protection of the environment 6. The detection phase is focused on the
6 Article 4(41) of Regulation (EU) 2019/6
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 13/79
第 14 页
244 observation of new potential risks or new aspects and/or change in reporting patterns. The detection
245 phase generally does not require deep analysis of data.
246 Irrespective of the frequency of monitoring according to the chosen risk-based approach for a product,
247 MAHs are obliged to conduct a signal detection analysis for each of their active substances or products
248 in the Union pharmacovigilance database (UPhV-database) at least once a year 7. In practical terms,
249 the reference to UPhV-database means the EudraVigilance Veterinary Data Warehouse (EVV-DWH) in
250 this context. This analysis should be performed within 2 months of the annual due date (see also
251 section 3.1.1.2).
252 Signals may be detected from any source, using a variety of methods. This section will summarise the
253 most important considerations relevant for signal detection.
254 2.1.1. Signal triggers
255 The source of a detected signal is also called ‘the signal trigger’. Signals can arise from any source,
256 including e.g. from case reports, clinical trials, scientific literature or public/social media. Potential
257 signals should not be dismissed or validated based solely on their source. Regardless of the original
258 source of a potential signal, other data sources may be investigated for relevant evidence during
259 further assessment (see also section 2.4). Depending on the source of a potential signal, different
260 methods may be used to determine whether it should be treated as a signal (see also section 2.1.2).
261 Signal triggers for any signal progressing on to the validation stage should be recorded in a traceable
262 manner.
263 Below are examples of the more common signal triggers. The list should not be considered exhaustive.
264 2.1.1.1. Adverse event reports:
265 The most common signal trigger is adverse event (AE) reports. AE reports are the basis of adverse
266 event databases such as the UPhV-database (EVV-DWH) or the MAHs’ own safety databases. Many
267 MAHs maintain their own pharmacovigilance databases, which may be used for signal detection in
268 addition to the UPhV-database (EVV-DWH). AE reports in databases may be spontaneous or solicited
269 (e.g. from clinical trials) and may concern one or more suspected adverse events, lack of expected
270 efficacy, medication errors or a combination of these. The threshold for when AE reports trigger a
271 signal may vary and different methods can be applied to detect signals in databases (see section
272 2.1.2.1).
273 2.1.1.2. Scientific literature:
274 Monitoring of scientific literature can be used to detect potential signals arising from e.g. clinical trials,
275 non-interventional studies, literature reviews or meta-reviews, published case studies or case series or
276 potential signals pertaining to the active substance or to a class of substances (e.g. NSAIDs, beta-
277 lactam antibiotics etc.). Some of these types of literature may also be found as case reports in adverse
278 event databases.
279 2.1.1.3. Public/Social Media:
280 Monitoring of public and social media is usually part of the collection and recording of adverse events
281 as AERs in the UPhV-database (see also VGVP module on collection and recording of suspected adverse
7 Article 17(7) of CIR (EU) 2021/1281
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 14/79
第 15 页
282 events for VMPs 8). In such cases, information from these sources is included in the signal management
283 process as AERs.
284 However, there may be situations where the information detected is not suitable for creation of AERs
285 for submission into the UPhV-database but may still constitute a signal trigger (i.e. justify further
286 investigation). It may be relevant to progress a signal from such a source through the signal
287 management process, to either be able to ease public concern or to ensure proper risk mitigation
288 measures are in place. Swift and thorough management of signals from media sources may also
289 contribute to maintain trust from the public. However, such signals should, like all other signals, be
290 further supported by other evidence in order to progress through the different steps of the signal
291 management process.
292 2.1.1.4. Other sources
293 Potential signals may be triggered from other sources than the ones mentioned above. Such sources
294 include, but are not limited to:
295 • Requests for signal submissions from the CVMP or a national competent authority in the EEA.
296 • Alerts, communications or reports from other regulatory authorities (non-EEA).
297 • Communication from manufacturers or distributors (potential impact on quality).
298 • Veterinary toxicology centres or human poison control centres.
299 • Academic poster presentations, conferences or collaborations.
300 • Environmental monitoring (e.g. for antiparasitics with wildlife impact).
301 • Slaughterhouses (residues).
302 • Signals arising from other signals; when patterns or associations noted during assessment of
303 one signal leads to other signals. This occurs most often with signals that are new aspects of
304 known risks but may also occur in more complex situations for new risks (see example 1).
305 Similarly, signals of interactions between two drugs may also arise from noticing patterns of
306 co-reported drugs during assessment of broader signals.
307 • Direct communication from the public (e.g. a veterinarian, pet owner or farmer) providing
308 information of a nature not suited for a regular case report (e.g. questions, observation of
309 unspecific trends, complaints, etc.).
310 • Ongoing post-marketing clinical trials.
Example
1: A signal of an event, e.g. seizure, is suspected to be more serious in a subpopulation of the
target animal species which also have impaired renal function. This is suspected due to a
pattern of co-reporting of the event with kidney-related VeDDRA terms. The signal
assessment should therefore include an assessment of both the overall drug-event-
combination (all reports of any type of seizure in the entire population) and a specific
assessment of the drug-event-combination in the specific subpopulation (i.e. reports in
animals who also have renal impairment).
8 https://www.ema.europa.eu/en/documents/regulatory-procedural-guideline/guideline-veterinary-good-
pharmacovigilance-practices-vgvp-module-collection-recording-suspected-adverse-events-veterinary-medicinal-
products_en.pdf
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 15/79
第 16 页
311 2.1.2. Methods for signal detection
312 Signals can arise from many different sources and may concern an array of different topics; hence
313 different methods may be needed to ensure that all relevant signals are detected. Signal detection
314 benefits from the use of a combination of quantitative and qualitative methods.
315 • Quantitative methods are mostly used to detect signals in adverse event databases and
316 examples include; use of disproportionality measures, counting of reports, trend analyses and
317 other forms of mathematics/statistics applied to data. In some circumstances, quantitative
318 methods may also be applied in a limited manner to other signal sources. For example, trend
319 analyses of (social) media reports, scientific publications or other external information sources
320 may help identify potential emerging safety concerns requiring further evaluation.
321 • Qualitative methods are based on review and interpretation of non-numerical information, e.g.
322 screening of case narratives or abstracts of scientific literature. They are often applied during
323 detection of signals arising from other sources than adverse event databases. However,
324 qualitative methods may be applied during signal detection in databases as well, although
325 caution should be taken to not cross over to signal validation (see section 2.3). During signal
326 detection, qualitative review should focus on identifying potential signals and more detailed
327 evaluation should be left for the validation and assessment steps.
328 Following detection, it is important to prioritise signals before moving on to validation (see section
329 2.2).
330 Before initiating signal detection, prior knowledge about the following aspects is expected: product
331 information, the mechanism of action of the product/substance in question, the outcome of previous
332 signal management activities and any other safety-related activities or procedures and any other
333 relevant pharmacovigilance data of which the MAH can reasonably be expected to be aware.
334 2.1.2.1. Signal detection in adverse event databases
335 Signal detection in adverse event databases should generally be performed at the VeDDRA level of
336 preferred terms (PTs).
337 However, it may occasionally be appropriate to perform signal detection at a higher level (HLT or SOC)
338 (see example 2). Conversely, there are situations where it may be relevant to stratify into LLT level
339 because certain LLTs belonging to a PT are already listed in the product information (see example 3)
340 (see also sections 2.2.1.6 and 2.2.1.8 on signal prioritisation and section 2.3.1.2 on signal validation).
341 The VeDDRA system is structured around organ classes, but sometimes signals may concern
342 ‘syndromes’ or collections of clinical signs from different organ systems, or there may be more than
343 one way to code the same event, leading to a need for combination of several terms to catch all
344 relevant reports. In such situations, it may be necessary to combine several similar PTs or LLTs from
345 different SOCs into one signal to allow for detection of broader signals or account for differences in
346 coding of reports (see example 4).
347 Whether there is a need to broaden or narrow a particular search at the signal detection stage will
348 need to be determined based on the clinical picture on a case-by-case basis.
349 It is generally not advised to combine different species in the same signal. However, there may be
350 situations where this can be relevant (e.g. product use errors, prescription errors, etc.).
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 16/79
第 17 页
Examples
2: Since similar events may be coded differently, it may be relevant to consider a high level
group of VeDDRA PTs together, e.g. an event where the animal experienced local swelling
redness and pain at the injection site, may be coded with any combination of ‘injection site
erythema’, ‘injection site inflammation’, ‘injection site lesion’, ‘injection site mass NOS’,
injection site pain’ and/or several other PT’s, leading to a need for analysis of all ‘Injection
site reactions’ (HLT) together.
3: In a situation where the PI includes the LLT ‘enteritis’, it may be relevant to monitor other
LLTs belonging to the PT ‘enteritis’, either because they are more specific than what is
currently in the PI, e.g. ‘lymphoplasmacytic enteritis’, or more serious, e.g. ‘haemorrhagic
enteritis’.
Or, it may be relevant to monitor only specific LLTs out of the 102 LLTs belonging to the PT
‘Lack of efficacy’ due to properties of the product. For example, it may only be necessary to
monitor ‘Lack of efficacy – Parvovirus’ and the unspecified LLTs ‘Lack of efficacy NOS’ and
‘Partial lack of efficacy’ in case of a vaccine against parvovirus.
4: ‘Pancreatitis’ (digestive tract disorders SOC) may be combined with ‘Elevated pancreatic
enzymes’ (investigations SOC), or
‘Reluctant to move’ (systemic disorders SOC) may be combined with ‘Lameness’ and ‘Joint
pain’ (musculoskeletal disorders), or
PT’s from the SOC ‘Application site disorders’ may be relevant to combine with PTs from the
SOC ‘Ear and Labyrinth disorders’ for an ear product, or
PT’s ‘Anaemia’, ‘Haemolytic anaemia’, ‘Immune mediated haemolytic anaemia’ and ‘Aplastic
anaemia’ for a signal on anaemia to capture all relevant cases.
351 Number of cases or reacting animals
352 During signal detection in databases the first parameter to consider would be number of cases
353 reporting the different VeDDRA terms, both cumulatively from the date of authorisation and/or in a
354 specified time period (e.g. since the data lock point of the last annual statement). Special attention
355 should be paid to what is most commonly reported for the product and any observed changes in such
356 patterns over time. The number of reacting animals should also be taken into account. For reports in
357 species usually kept in herds, the number of reacting animals should be interpreted together with the
358 number of cases (representing independent herds), as a large number of reacting animals from a
359 single herd may provide different information from similar events reported across multiple independent
360 herds.
361 VeDDRA terms with a high cumulative number of reported cases or a large number of reacting animals
362 should be considered as potential signals. Such VeDDRA terms may not necessarily present with
363 disproportionate reporting due to masking effects (see next sub-section), yet their volume of reports
364 can still be clinically relevant and warrant further assessment.
365 Disproportionality in reporting
366 The concept of disproportionality in reporting is mostly used when analysing case reports in adverse
367 event databases such as the UPhV-database (EVV-DWH) or the MAHs’ own safety databases. When an
368 adverse event is described as having ‘disproportionate reporting’, or as being ‘disproportionately
369 reported’ in general, it means that the drug/product event combination is reported more often than
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 17/79
第 18 页
370 expected; i.e. the adverse event in question is reported more often in combination with the
371 product/drug in question compared to the average for all other products in the database.
372 The difference between the concept of ‘disproportionate reporting’ in general and the specific concept
373 of a ‘Signal of Disproportionate Reporting’ (SDR) is that the latter reaches a pre-defined ‘level of
374 disproportionality’. The criteria used to define an SDR are often specific to the database used for signal
375 detection and the method used to calculate disproportionality, which is why it is important to document
376 which database an SDR was detected in and the specific criteria/threshold values and method for
377 calculation of disproportionality it uses.
378 It should be noted that disproportionate reporting does not automatically imply causality. It is a
379 mathematical tool flagging up product-event pairs and disproportionate reporting can be seen for a
380 product-event pair with no causal association, while it may be absent for a product-event pair with a
381 well-established causal association (see also section about factors influencing the ROR below).
382 Detection of disproportionate reporting should therefore always be combined with qualitative review of
383 data during further assessment.
384 Methods of calculating disproportionality
385 When performing signal detection, disproportionality measures should generally be evaluated based on
386 cumulative data, but comparison of disproportionality between different time periods may be relevant
387 to identify trends and changes in disproportionality.
388 Disproportionality can be calculated using different methods such as Proportionate Reporting Rate
389 (PRR), Information Component (IC025) or Reporting Odds Ratio (ROR). The three mentioned methods
390 are all based on a 2x2 contingency table:
Reports of adverse event of interest Reports of other adverse events
Reports with drug/product of interest a b
Reports with other drugs/products c d
391 Table 1: 2x2 contingency table
392 Note that the contingency table only states the number of reports and therefore does not take into
393 account that reports may have multiple reacting animals. Therefore, in species with herd level reports,
394 it may be necessary to apply other quantitative measures during signal detection such as number of
395 reacting animals or use more qualitative methods such as case review.
396 The UPhV-database (EVV DWH) uses ROR for calculation of disproportionality. ROR can be calculated
397 from the contingency table using a simple formula:
𝑎𝑎/𝑐𝑐
398 𝑅𝑅𝑅𝑅𝑅𝑅 =
𝑏𝑏/𝑑𝑑
399 If ROR is precisely 1, the adverse event of interest is reported with the drug/product at a rate identical
400 to the average for all products in the database. Consequently, if ROR is greater than 1, there is
401 disproportionality in the reporting. However, the ROR alone is not sufficient to judge whether an
402 association between a product and an adverse event is statistically robust. This is why 95% confidence
403 intervals (CIs) are calculated and displayed as ROR(–) and ROR(+). A useful rule of thumb is based on
404 the lower bound:
405 If ROR(–) > 1, the association is considered statistically significant, meaning the reported event occurs
406 more frequently than expected.
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 18/79
第 19 页
407 Other measures of disproportionality may be used by MAHs during signal detection. MAHs should
408 justify and document their choice of method(s) and include a description of those methods, including
409 the underlying principles and any criteria and thresholds applied.
410 Criteria for SDRs in the UPhV-database (EVV DWH)
411 In EVV-DWH, an SDR is defined as a signal fulfilling all the following criteria:
412 • ROR ≥ 2
413 • ROR(-) ≥ 1
414 • Cumulative number of reports ≥3 for Medically Important Terms (MI terms, see section
415 2.2.1.3) or ≥5 for all other VeDDRA terms.
416 SDRs can be a useful tool to detect signals, but not all VeDDRA terms highlighted as SDRs are
417 necessarily signals; for example, terms highlighted as SDRs may already be listed in the product
418 information (see sections 2.2.1.6 and 2.2.1.8 on signal prioritisation and section 2.3.1.2 on how to
419 determine coverage by the PI during signal validation). In addition, there are factors potentially
420 affecting disproportionality which should be considered, as described below. Further signal validation
421 may also end in non-validation of an SDR (see section 2.3 on signal validation). See also ‘Veterinary
422 Union Pharmacovigilance Database – Best Practice Guide’ 9 and ‘Veterinary Pharmacovigilance EVVET -
423 Data Warehouse user manual’ 10.
424 Factors affecting disproportionality
425 Certain factors beyond true disproportionate reporting may affect disproportionality. When such factors
426 lead to an increase in the ROR, this is referred to as ‘artificial inflation of disproportionality’.
427 Conversely, when such factors lead to a decrease in the ROR, it is referred to as ‘masking of
428 disproportionality’. Many of these factors can affect the ROR in both directions depending on the
429 situation. It is therefore important to consider these factors on a case-by-case basis during signal
430 detection.
431 Such factors include, but are not limited to:
432 • Stimulated reporting: This may occur e.g. in cases of media attention on a certain product or
433 adverse event or immediately following publication of new warnings. Stimulated reporting may
434 artificially inflate the ROR.
435 • Increase in exposure (sales): Increased exposure may lead to increased disproportionality (i.e.
436 artificial inflation), although the reporting incidence remains stable.
437 • Duplicate reports: A high number of duplicate reports in a database may both artificially inflate
438 or deflate disproportionality. If there is a high number of duplicates only on the specific term of
439 interest that would lead to an artificial inflation of the ROR, while the reverse situation (high
440 number of duplicates for other adverse events or other products) could lead to masking. In the
441 UPhV-database (EVV DWH) duplicates are considered to most often lead to inflation of the
442 ROR.
443 • Co-prescription (bystander products): Drugs/products that are often prescribed with other
444 drugs with a more high-risk safety profile or to animals with many co-morbidities may
445 experience artificially inflated disproportionality measures.
9 https://www.ema.europa.eu/en/documents/other/veterinary-union-pharmacovigilance-database-best-practice-
guide_en.pdf
10 https://www.ema.europa.eu/en/documents/other/evvet-data-warehouse-user-manual_en.pdf
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 19/79
第 20 页
446 • Confounding by indication: When there is disproportionality of clinical signs which may be
447 related to the indication of the product/drug the ROR may be either artificially inflated or
448 masked. Artificial inflation occurs in a situation where signs of the underlying disease are
449 mistakenly reported as AEs, while masking occur in a situation where AEs are not reported as
450 they are falsely assumed to present signs of the underlying disease.
451 • Rarity and common/general clinical signs:
452 o Rarity may artificially inflate ROR in certain situations; for example, for rare product
453 types, disproportionality may be artificially inflated for certain VeDDRA terms which, by
454 nature, are linked only to a small subset of products e.g. implant site reactions occur
455 only with products that are implants. In such situations, it may make sense to also
456 determine disproportionality in relation to a smaller subset of similar products rather
457 than all products in the database. However, the robustness of calculations is
458 compromised when based on very small datasets.
459 o Conversely, clinical signs (VeDDRA terms) that are non-specific and may have many
460 causes, or signs often related to drugs or specific classes of drugs may be so widely
461 reported on all products/drugs in the database, that differences become difficult to
462 detect (masking). Hence, the ROR for an event which is commonly occurring for many
463 other products but only occurs very rarely for the product of interest, may be masked
464 by the large amounts of reports present for other products.
465 • Composition of the database: If a database includes only a limited number of products, or
466 mostly a few specific types of products (e.g. mostly antiparasitics, mostly vaccines etc.) this
467 may lead to both artificial inflation and masking of events, depending on the situation. This is
468 very important to consider when MAHs perform disproportionality analyses in their own safety
469 databases.
470 These and any other potential factors which may affect disproportionality should be considered when
471 appropriate during interpretation of disproportionality in signal detection.
472 Screening of case narratives
473 Screening of case narratives (qualitative review) may be necessary in some cases, to determine if
474 there is a signal or not. For example, for signals concerning large flocks/herds of poultry or schools of
475 fish or other production animals, there may be few cases concerning very large number of animals,
476 leading to a need for review of the case to determine if a signal is present. Another example may be in
477 cases of more complex signals involving new aspects of known risks or concerning more specific
478 subpopulations or other special circumstances of the signal that cannot readily be identified through
479 quantitative methods alone. Sometimes, such complex signals are triggered from screening or
480 reviewing of case narratives during validation or assessment of other signals (see example 5).
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 20/79
第 21 页
Example
5: When reports of similar events are coded under different VeDDRA terms they may not
generate a clear quantitative signal, e.g. situations of syndromes or newly arising disorders,
where no precise VeDDRA terms exist yet, where there is no consensus on nomenclature in
the veterinary field, or where general knowledge of the condition may be low, can lead to
reports of varied clinical signs related to the syndrome/disorder rather than the diagnosis
itself. Review of the case narratives reveals that the events are described similarly, despite
differences in coding. Identification of such a recurring pattern may trigger further
investigation of a potential signal.
481
482 Identification of potential quality and batch-related issues through signal detection
483 Signals related to one or more specific batches of a VMP are often detected from adverse event
484 databases by recognition of certain patterns of reporting. Examples of important patterns to monitor
485 for in adverse event databases are included below:
486 • Temporal or geographical clustering: A cluster of adverse event reports linked to the same
487 batch number within a short period or localised area is a key indicator of a potential
488 batch‑related issue (see examples 6-7).
489 • Increased frequency of a specific adverse event associated with a single batch: An unusual
490 increase or clustering of reports of a particular adverse event associated with an individual
491 batch, compared to other batches or historic data, indicates a batch related issue (see
492 examples 8-9).
493 • Signs of contamination or sterility failure: An unusual pattern or clustering of adverse events
494 suggestive of contamination or loss of sterility in the suspected VMP may indicate a quality
495 defect or batch-related issue (see examples 8 and 10-11).
496 The above list should not be considered exhaustive and other methods may be applied in addition to
497 these for detection of quality and batch-related issues.
498
Examples
6: Multiple reports within 24–72 hours following administration of a specific batch.
7: Reports originating from the same region or distributor, all citing the same batch.
8: Higher frequency of anaphylactic reactions linked to a specific batch.
9: Sudden increase in ‘lack of efficacy’ reports linked to a specific vaccine batch.
10: Several reports, associated with the same batch, describe injection-site abscesses, swelling
and local infections occurring shortly after administration. Although the individual events
differ, the clustering of reports concerning a single batch may suggest contamination or loss
of sterility and warrant further investigation.
11: Injection‑site infections emerging across multiple animals after use (of the same batch).
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 21/79
第 22 页
499 2.1.2.2. Monitoring of scientific literature:
500 Monitoring scientific literature is part of the signal management process 11. Its purpose extends beyond
501 searching for adverse event reports and it should be considered a tool to keep up with current scientific
502 knowledge.
503 Monitoring of scientific literature usually involves performing searches for literature in databases and
504 specific journals if applicable and/or setting up automatic notifications on such platforms. The MAH
505 should establish the most relevant source(s) of published literature for each product. In order to
506 narrow down the search, relevant filters should be applied.
507 Monitoring of scientific literature is ideally performed on active substance level 12. Active substances
508 may have several different names and/or spellings, which should be taken into account when
509 performing literature searches; it is recommended to leave out names of salts (e.g. 'hydrochloride’). In
510 certain cases, it may make sense to also monitor closely related active substances (e.g. prednisone,
511 prednisolone and methylprednisolone), while for vaccines, it may be necessary to broaden the search
512 strategy to include terms related to the vaccine’s disease indication and relevant immunopathological
513 mechanisms, rather than relying exclusively on specific antigen names. Further filters may be applied,
514 but care should be taken to avoid narrowing the search too much.
515 Consideration should be given to scientific literature concerning potential environmental concerns
516 related to the active substance of the VMP or its metabolites as well as concerns regarding resistance
517 for antimicrobials and antiparasitic agents as these issues are difficult to detect through adverse event
518 reporting alone.
519 Once search results have been retrieved, an initial screening of abstracts and/or the whole article
520 should be performed in order to determine if the article in question is relevant to the product or the
521 active substance(s) of the product and contains information on safety or (lack of) efficacy which is
522 either new or may represent a new aspect of a known risk (e.g. frequency, outcome, further
523 characterisation etc). Thorough assessment of methods and validity of the articles is not necessary
524 during signal detection but may be applied during validation (see section 2.3.1.3).
525 2.1.2.3. Monitoring of other sources
526 Although pre-emptive monitoring is not possible for all potential alternative sources of signals (e.g.
527 spontaneous communication directly to the MAH), others (e.g. social media) are often monitored to
528 some extent as part of the collection and recording of adverse events. There may be situations, where
529 the information detected from other sources (see also sections 2.1.1.2 - 2.1.1.4) is not suitable for
530 creation of AERs to be entered into the UPhV-database, but may in itself constitute a signal trigger (i.e.
531 justify further investigation). Examples of additional sources which may be monitored are:
532 • Non-peer-reviewed literature sources; e.g. public media targeting veterinarians or other animal
533 professionals in the countries of authorisation, national media in the countries of authorisation,
534 etc.
535 • Social media; particularly the MAH’s own social media platforms.
536 • Relevant websites or online forums, e.g. websites of EMA, NCAs or the MAHs own website.
537 • Relevant veterinary or other scientific conferences.
11 Article 81(1) of Regulation (EU) 2019/6
12 Article 13(3) of CIR (EU) 2021/1281
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 22/79
第 23 页
538 Pharmacovigilance staff should be sensitised to treat such information, when becoming aware of it, as
539 a potential signal trigger. It is recommended to include this in internal process documents for e.g.
540 monitoring of social media. The reasoning for it to be treated as a signal trigger should be recorded as
541 well as the date it was decided that the known information should be looked into further (and hence
542 classified as a trigger).
543 2.2. Signal prioritisation
544 Signal management should utilise a risk-based approach proportionate to the identified risk5. It is
545 therefore important to have an established and documented process of signal prioritisation throughout
546 the process2.
547 Signal prioritisation should be a continuous process throughout all steps of the signal management
548 process and is often carried out in parallel with each step. In many cases, signals may move directly
549 from detection through the validation phase and sometimes even through the assessment phase.
550 However, where multiple potential signals are identified simultaneously, prioritisation may be
551 necessary to ensure that the signals with the greatest potential impact on safety are progressed to the
552 next stage. If a potential signal is not prioritised at this point in time, the potential impact on safety
553 should be carefully considered. Pharmacovigilance alerts should always be prioritised.
554 Signal prioritisation during and following signal detection is essential to decide:
555 1. Which signals to start a validation on first,
556 2. Their urgency, both in general and in relation to other signals,
557 3. The extent of continued monitoring of the issue in the future.
558 Rationale behind prioritisation of signals should be documented.
559 2.2.1. Factors to consider during prioritisation
560 The factors listed in this section can help determine impact on overall safety, as well as urgency of the
561 signal and should be considered during prioritisation. Not all factors listed in this section may be
562 relevant for all signals during all steps of the signal management process, e.g. due to lack of
563 knowledge, or because another factor already determined the conclusion on the prioritisation.
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 23/79
第 24 页
564
565 Figure 3: Decision tree for prioritisation during signal detection
566 Priority is determined by weighing impact on safety against the strength of evidence. All three exist on a
567 continuum. Safety should always be weighed highest.
568
569 2.2.1.1. Pharmacovigilance Alerts (PhV Alerts)
570 First and foremost, it is important to determine whether a potential signal could be a PhV Alert. A
571 Pharmacovigilance alert is defined as any suspected unexpected serious risk which may require urgent
572 regulatory action and communication by the MAH or the competent authorities as referred to in Article
573 129, Article 130 and Article 134 of Regulation (EU) 2019/6.
574 When a MAH becomes aware of a PhV Alert from any source, they should notify the competent
575 authority(ies) of the Member State(s) where the VMP is authorised or the Agency in case of centrally
576 authorised products. This should be done as soon as possible and no later than 3 working days after
577 establishing that a validated signal or a safety issue from any source constitutes a PhV Alert. See also
578 section 3.1.3.
579 In some cases, PhV Alerts may concern or involve a quality issue, or recalls of specific batches may be
580 necessary. However, a batch recall in itself, i.e. without evidence of a safety risk, is not considered a
581 PhV Alert.
582 See examples 12-16 of what may be considered PhV Alerts and examples 17-18 of what is not a PhV
583 Alert.
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 24/79
第 25 页
Examples of potential Pharmacovigilance Alerts
12: Major safety issues identified in the context of ongoing or newly completed studies, e.g. an
unexpected increase in rate of fatal or life-threatening adverse events. Where applicable,
separate (national) regulatory reporting requirements for studies should also be considered.
13: Major safety issues identified from any source, which may lead to considering a
contraindication, a restriction of use of the VMP or its withdrawal from the market.
14: Major safety-related regulatory measures outside the EU, e.g. a restriction of the use of the
VMP or its suspension.
15: Major negative impact on the environment possibly caused by the use of a VMP.
16: Increased level of residues in food-producing animals that may represent a safety concern
for humans. For example, detection of residues exceeding the MRL in cases where the
approved withdrawal period has been respected.
Examples that should not be considered a Pharmacovigilance Alert
17: Events that are associated with the use of a veterinary medicinal product in humans as part
of a suicidal attempt are not considered Pharmacovigilance Alerts. However, repeated
reports of suicidal attempts or emerging patterns indicating an unusual association between
a specific VMP and the adverse event may constitute a potential signal and should be
evaluated through the signal management process.
18: New safety information related to quality defects or suspected falsified medicinal products
which might influence the evaluation of the benefits and risks of the VMP. These should be
notified in accordance with quality defect processes.
584 2.2.1.2. Safety (for the animal, user or environment):
585 For all remaining potential signals, the main focus during prioritisation should always be placed on the
586 risk to safety (for the animal, user or the environment). Some signals are likely to affect the benefit-
587 risk balance of a product more than others, if a causal association is considered possible in the end.
588 This is most often correlated with the seriousness and severity (see section 2.2.1.4) of the clinical
589 sign/disorder, but other factors may also come into play, such as size of the affected population and
590 properties of the signal itself; e.g. number of cases and affected animals already reported, number of
591 fatal cases etc. (see example 19).
592 The Medically Important Terms list (MIT-list, see section 2.2.1.3) is a useful tool when assessing
593 seriousness and severity of a signal during the signal management process. However, the specifics of
594 the signals should always be considered (see example 20).
595 The expected extent of future risk minimisation measure(s) arising from the signal may be considered
596 for guidance, as this is usually related directly to patient safety (see example 21).
597 The remaining aspects in this section (2.2) can also be considered but should always be secondary to
598 overall safety in the decision-making when prioritising potential signals for validation and further
599 assessment.
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 25/79
第 26 页
Examples
19: A validated signal of acute renal failure which appears to occur in the entire target
population (i.e. not only in a subset of the population), that already has many reported
cases and a high number of fatal cases, would appear more urgent to assess for causality
compared to a validated signal of mild injection site inflammation occurring in a specific
breed, with no fatal cases. Both signals will need to be assessed, but the first is likely to
affect patient safety the most, if causality with the product is found to be confirmed.
20: The MI-term ‘Anorexia’ may cover signals of great severity and/or seriousness, but could
also concern less severe/serious cases of mildly lowered appetite and further assessment of
a validated signal of severe ‘Anorexia’ in a species that is sensitive to sequela from anorexia
should be prioritised higher than assessment of a validated signal of mildly lowered appetite
in a less sensitive species.
21: The following questions may be considered regarding potential future risk minimisation
measure(s):
If the events of the signal are concluded to be causally associated to the product, what
actions would that necessitate?
Would the product information need amendment and to what extent? Addition of adverse
reactions? Warnings or precautions? Additional monitoring advice? Relative or absolute
contraindications?
Could there be a need to disseminate immediate information through Direct Animal
Healthcare Professional Communication (DaHPC or ‘Dear Doctor’-letter)?
Or is it serious enough that a suspension or revocation of the marketing authorisation may
be considered?
600 2.2.1.3. VeDDRA Medically Important Terms (MIT)
601 The MIT list comprises VeDDRA PTs that correspond to medical conditions which are commonly
602 involved in suspected adverse events across different pharmacological or therapeutic classes, or which
603 are considered medically important independently of any established causal association with a
604 veterinary medicinal product, due to their seriousness or potential impact on animal health or welfare,
605 public health, user safety or the environment.
606 How to interpret and apply the MIT list in signal management:
607 Medically Important Terms (MITs) in veterinary pharmacovigilance are intended to function as a
608 prioritisation tool within signal detection and signal management activities. These events represent
609 medical concepts that require enhanced attention, prioritised assessment and timely regulatory follow-
610 up, in line with a proportionate pharmacovigilance approach.
611 In food-producing animals, MITs additionally include VeDDRA PTs that may be associated with
612 significant herd- or flock-level consequences (e.g. substantial productivity losses, reproductive
613 impairment or increased mortality). Where such effects reflect or accompany medically important
614 safety concerns, they justify prioritisation within signal detection, validation and assessment activities.
615 For events occurring in humans, all reports are regarded as medically important and should therefore
616 be handled as MITs for the purposes of prioritisation and signal management, irrespective of the
617 reported clinical term, seriousness classification or outcome.
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 26/79
第 27 页
618 All events that are reported as ‘parent-offspring’ events should be considered medically important and
619 particularly any congenital conditions reported in this setting.
620 Malignant neoplastic conditions and neoplastic conditions which otherwise have a serious impact on
621 animal health and welfare should be considered medically important on a case-by-case basis.
622 Detected signals involving MITs should generally be prioritised particularly in the following situations:
623 • Where they fulfil the criteria of a signal of disproportionate reporting (SDR) in DWH.
624 • Where an unusual or unexpected increase in reporting frequency or number of affected animals
625 is observed.
626 • Where there is evidence of increased severity or a change in outcome (compared to the
627 product information), such as an increase in fatal cases.
628 2.2.1.4. Severity and/or seriousness
629 Severity and seriousness are usually defined at case level. Here, severity refers to the intensity of an
630 adverse event, e.g. mild, moderate or severe, but severity does not express anything about the
631 medical significance of the event itself. Seriousness of an event is linked to outcome e.g. whether an
632 event was fatal, led to hospitalisation and/or long-term harm to the patient (see example 22).
633 While general criteria for seriousness can be found in the VICH GL24 (Pharmacovigilance of Veterinary
634 Medicinal Products: Management of Adverse Event Reports (AERs)), it is not possible to establish
635 general criteria for severity, as that is affected by the sign or disorder considered.
636 Both seriousness and severity may be used to prioritise signals. Seriousness can be evaluated by
637 screening for fatal cases, number of cases leading to full recovery and in some cases by clinical
638 judgement of the VeDDRA term itself (see example 23). The MI terms described in section 2.2.1.3
639 above may provide guidance in the evaluation of seriousness and severity.
640 Severity can sometimes be estimated by looking at reported LLTs (i.e. if it is the more or less severe
641 LLTs belonging to the PT that are mostly reported), or duration of the event, but many times a
642 screening of narratives is necessary to evaluate whether e.g. the PT ‘Anorexia’ covers a transient
643 partial loss of appetite or a more severe case of complete anorexia for a prolonged period of time.
644 Signals where the majority of the reported events are considered serious should be prioritised over
645 signals concerning less serious events, although rarity of the event should also be considered here (see
646 section 2.2.1.5).
647 Similarly, signals reporting more severe events should be prioritised higher compared to signals
648 reporting mild versions of the same event. Severity will, however, always need to be weighed against
649 seriousness and number of reports/reacting animals and potentially other factors, when comparing
650 different signals.
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 27/79
第 28 页
Examples
22: For example, pruritus may be mild, moderate or severe, but a severe case of pruritus may
still not always be a serious case – if the animal recovered on its own with no long-term
sequelae. Conversely, severe heart failure is usually linked to serious outcome and even
‘mild’ heart failure may be considered serious as it is likely to have a long-term negative
effect on the patient. Furthermore, a severe hypersensitivity reaction is usually serious,
while a mild hypersensitivity reaction is usually not. Hence severity and seriousness are
sometimes linked and sometimes not.
23: For example, ‘heart failure’ can be assumed to be more serious in general, compared to e.g.
‘injection site pain’.
651 2.2.1.5. Number of reported events and effect on special populations
652 Signals with higher numbers of reports and/or reacting animals or signals with increasing numbers
653 should have a higher priority. However, this should, as a minimum, be weighed against seriousness
654 and severity and other factors may come into consideration as well.
655 Signals affecting certain special populations e.g. pregnant animals, very young animals, or animals of
656 food-producing species that are close to slaughter may generally be prioritised, albeit with
657 consideration to potential off-label use (see section 2.2.1.11). Conversely, signals affecting a very
658 small subpopulation of animals, e.g. animals with a specific uncommon co-morbidity, may be
659 prioritised lower if there are other, more urgent signals to perform validation on.
660 2.2.1.6. Novelty
661 Novelty of the information may also be considered in the sense that it will often be more important to
662 progress with potential signals which may represent completely new information (e.g. a new adverse
663 event for the product) rather than signals concerning further characterisation of an already known
664 adverse event (new aspect of a known adverse event). The evaluation of novelty should, however, be
665 secondary to the evaluation of patient safety e.g. a potential signal that changes an already known
666 event from ‘mild and transient’ to ‘potentially life-threatening and/or chronic’ may affect patient safety
667 more than a new adverse event that in and of itself is mild and non-serious.
668 2.2.1.7. Preventability
669 One of the main aims of the entire signal management process is to mitigate the risks of treatment
670 with VMPs as much as possible. Any knowledge on preventability of the clinical signs/disorders or the
671 most severe outcomes can be taken into consideration during prioritisation. If a serious adverse effect
672 or outcome is known to be easily preventable, the information is of high value, as the benefit of
673 treatment will remain, while the risk is mitigated. However, preventability should not be regarded as a
674 decisive criterion for prioritisation and should be considered alongside other factors relevant to the
675 potential impact of the signal.
676 2.2.1.8. Potential signals concerning VeDDRA terms listed in the SPC
677 Known adverse events of a product are generally listed in section 3.6 of the SPC, while additional
678 information contributing to the description of adverse events as well as advice regarding risk mitigation
679 may also be found in sections 3.3 and 3.5. Information relating to efficacy and lack thereof is usually
680 included in section 3.4. It should be noted that adverse events are included in the SPC at VeDDRA LLT
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 28/79
第 29 页
681 level, while signals are usually detected at PT level. Therefore, when a signal is detected at PT level, it
682 may be necessary to review the reported LLTs of cases to determine whether the reported events are
683 already adequately covered by the SPC or whether they represent a new or insufficiently characterised
684 aspect of the risk.
685 1. Monitoring of adverse events that are already listed with the exact wording in SPC section 3.6
686 should generally not be prioritised, with the exception of certain specific situations and factors,
687 where potential signals on listed events should undergo further prioritisation as normal:
688 a) Signals concerning increased frequency, both sudden and gradual.
689 b) Signals for VeDDRA PTs that include an LLT listed in the SPC, but also include LLTs
690 considered more severe/serious in nature, or more specific, compared to the listed LLT.
691 In these cases, monitoring of such LLTs should still be part of routine monitoring (see
692 section 2.1.2.1 and 2.3.1.2 for further guidance).
693 c) Signals concerning a change in outcome towards more a severe outcome compared to
694 what is described in the SPC, e.g. a rise in fatal cases.
695 d) Signals further characterising the already known risk where the additional information
696 may impact the understanding, management or communication of that risk.
697 2. Even if certain VeDDRA terms are not explicitly included in section 3.6 of the SPC it may be the
698 case that the suspected adverse event is already adequately covered by the text included in
699 the product information. If so, they should be handled as described in point 1 above.
700 Further guidance on how to determine whether a potential signal is adequately covered by the SPC is
701 provided in section 2.3.1.2.
702 2.2.1.9. Other previous knowledge on subject
703 Previous knowledge on the subject of the signal includes knowledge from previous signal validations
704 (irrespective of outcome), previous signal assessments, knowledge from previously assessed clinical
705 trials or any other procedure on the product in question including the initial authorisation procedure.
706 For example, a signal validation performed recently, which led to non-validation of the signal, may lead
707 to not prioritising the signal when detected again if the number of new cases received since then is
708 low. Conversely, in a situation where new data has been received and a causal association was highly
709 suspected in a recent signal assessment, but the evidence was not sufficient at the time of the
710 assessment, such previous knowledge may lead to prioritisation of the signal.
711 Knowledge from other products with the same active substance or with an active substance in the
712 same drug class may also be considered. For example, if the event is listed in the PI of another product
713 with the same substance, it may lead to increased suspicion of a potential causal association and
714 prioritisation of the signal.
715 2.2.1.10. The signal trigger
716 The signal trigger may in certain circumstances influence prioritisation. For example, in cases of
717 intense media interest, it may be of interest to the public and the MAH themselves to prioritise such a
718 signal, despite other factors. In other cases, a signal triggered from social media discussions with no
719 corroborating evidence from other sources, e.g. adverse event reports, product complaints or
720 literature, may not be prioritised. For example, discussion about lack of efficacy circulating on social
721 media, while there are no actual adverse event reports on lack of efficacy received in the period of
722 concern.
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 29/79
第 30 页
723 In other cases, the strength of evidence of a source may factor into the decision e.g. a signal triggered
724 from a newly published double-blinded clinical trial indicating a new risk may be prioritised higher than
725 if the same signal was identified in an adverse event database.
726 2.2.1.11. Special situations
727 Batch-related issues
728 If a batch-related issue is detected from the signal management process (i.e. is a potential cause of
729 reported adverse events or lack of efficacy), it should be considered a potential/validated signal.
730 Batch-related issues which are not related to a detected signal (e.g. if the faulty batches never reached
731 end users or there were no reports related to the affected batches), do not need to be considered as
732 signals and should follow the process related to quality defects 13. If there is suspicion of a batch-
733 related issue involved in a signal, it should be considered whether the issue could be a PhV Alert (see
734 sections 2.2.1.1 and 3.1.3).
735 Batch‑related issues, such as those arising from manufacturing deviations, quality defects, stability
736 failures or supply chain problems, are generally considered high‑priority due to their potential to affect
737 large numbers of animals within a short timeframe and their preventable nature. The criteria explained
738 in section 2.2.1.4 regarding seriousness and severity of the events should be taken into consideration
739 during prioritisation of batch-related issues.
740 Signals involving off-label use, overdose or medication errors
741 The obligation to perform pharmacovigilance extends to include use outside the terms of the marketing
742 authorisation (i.e. off-label use, overdose or medication errors) 14.
743 The following signals in non-target species should always be prioritised:
744 1. Signals concerning user safety, i.e. occurring in humans, should always be prioritised and all
745 VeDDRA terms are considered MI-terms in humans.
746 2. Signals concerning serious consequences of accidental exposure (a medication error) in non-
747 target species.
748 3. Signals concerning serious and severe adverse events in non-target species, where the
749 expected risk minimisation measures following a conclusion of a possible causal association
750 would involve a need to at least inform the public (see examples 24-25).
751 Signals involving some extent of use outside the terms of the marketing authorisation should generally
752 be prioritised in a similar way to all other signals and the evaluation of the role of non-authorised use
753 will usually be performed at a later stage.
754 In situations where there is knowledge of widespread use outside the terms of the marketing
755 authorisation for specific purposes, e.g. in a specific non-target species or for a specific off-label
756 indication in the target species, safety signals occurring in these circumstances should give rise to
757 further analyses of the signal(s) and considerations of whether it may be relevant to implement risk
758 minimisation measures in an attempt to reduce the off-label use to reduce the risk. For example,
759 safety signals where the majority of cases report use against a contraindication listed in section 3.3
760 may often not require prioritisation immediately. However, continued reporting of adverse events in
13 https://www.ema.europa.eu/en/human-regulatory-overview/post-authorisation/compliance-post-authorisation/quality-
defects-recalls
14 Article 11 of CIR (EU) 2021/1281
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 30/79
第 31 页
761 such cases should lead to concern that risk mitigation through a contraindication is not sufficiently
762 effective and thus may require further action from the MAH (see section 2.6 and example 26).
763 Signals specifically concerning previously unknown risks connected to overdose in the target species
764 can usually be prioritised below signals concerning use within the terms of the marketing authorisation.
765 However, as the MAH also has an obligation to keep the section of the PI concerning overdose up to
766 date with newest scientific knowledge, such signals cannot be dismissed with the argument of
767 overdose alone.
768 Signals concerning medication errors should be prioritised according to the risk the errors pose, as well
769 as preventability of the errors observed.
770 Signals receiving media attention
771 In some circumstances, signals that could cause/are causing media attention and/or public concerns
772 may deserve special attention. Prioritising further analysis of such signals may ensure a faster
773 conclusion (no matter the final outcome of the signal) and increase trust from the public.
774 Antimicrobial or antiparasitic resistance
775 Signals which may be a sign of antimicrobial/antiparasitic resistance should generally be prioritised.
776 Environmental risks
777 Signals may directly concern a risk to the environment. However, even for signals which do not
778 primarily concern the environment, it may be relevant to consider whether there could be
779 consequences for the environment. This is especially true for signals involving adverse events in an
780 animal or species other than the treated one, e.g. accidental exposure. Please also refer to section
781 2.3.1.3 on validation of environmental signals.
782 Signals with the primary topic of an environmental risk may arise from adverse event reporting,
783 scientific literature or other sources and should generally be prioritised. An additional environmental
784 aspect to a signal primarily focused on an adverse event would similarly increase priority.
Examples
24: A signal suggests that a serious adverse event may occur in a non-target species following
exposure to a veterinary medicinal product. Early recognition and discontinuation of
treatment appear to improve the clinical outcome. If confirmed, communication to
veterinarians and animal owners may be necessary to ensure prompt recognition and
intervention.
25: Several reports describe severe neurological adverse events in a non-target companion
animal species following accidental exposure to a product intended for another species. If a
causal association is confirmed, a precaution may be added to SPC section 3.5 and the
corresponding section of the PL under the sub-heading ‘other precautions’ in accordance with
the QRD template for VMPs.
26: A product contraindicates use in animals below a certain age, but serious adverse events
continue to be reported predominantly in that age group. This may indicate that the existing
risk minimisation measure (the contraindication) is insufficient and additional measures are
therefore implemented, e.g. strengthened wording of the contraindication, additional
warning added to the package or other measures as described in section 2.6.
785
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 31/79
第 32 页
786 2.2.2. Impact on future prioritisation during ‘routine monitoring’
787 If a potential signal is not prioritised for immediate progression to validation, it will usually return to
788 ‘routine monitoring’ and should be considered again during subsequent signal detection activities. Such
789 signals may be re-detected and prioritised at a later time.
790 If a validated signal is not prioritised for immediate assessment, it should be subject to enhanced
791 monitoring until the signal assessment has been completed. Any new information received during this
792 time should lead to a new assessment of urgency.
793 Non-validated and refuted signals (see sections 2.3.2.2 and 2.5.1 respectively) usually return to
794 ‘routine monitoring’.
795 However, factors determined during prioritisation, prior signal validations or assessments may
796 influence future prioritisation during routine monitoring for both potential signals for which progression
797 to validation were previously deferred, previously non-validated signals and previously refuted signals
798 (see also sections 2.3.2.2 and 2.5.1, see examples 27 and 28 and figure 4):
Examples
27: A specific VeDDRA term has been determined to be biologically implausible during signal
detection and/or prioritisation. Future (routine) monitoring may have lower priority.
28: A potential signal of low seriousness and severity was not prioritised for immediate
progression to validation due to presence of other signals determined to have a greater
impact on overall safety. However, a causal association could be possible based on
mechanism of action. The issue should then be given higher priority if re-detected during
routine monitoring.
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 32/79
第 33 页
799
800 Figure 4: Impact on future prioritisation of previously detected signals
801 Previous status: Refers to the outcome of previous prioritisation validation or assessment of the same signal,
802 depending on how far the signal came in the signal management process.
803 Previously detected signals, which led to risk minimisation measures, are not covered by this figure.
804
805 2.3. Signal validation
806 Once a signal has been identified through the signal detection process and has been prioritised for
807 validation, the next step in signal management is to conduct the signal validation. Signal validation is
808 an initial screening of the data that triggered the potential signal, with the aim of determining whether
809 the signal is supported by information of sufficient quality and substance to justify further assessment.
810 Therefore, signal validation is not intended to determine causality. At this stage of the signal
811 management process a causal association is only suspected and further assessment is needed to
812 conclude on causality.
813 When to perform a signal validation on a detected signal is determined through signal prioritisation
814 (see section 2.2).
815 The outcome of signal validation is either to ‘validate’ the signal, or ‘non-validate’ the signal. All
816 validated signals should be further assessed (see section 2.4) and entered into IRIS 15 Veterinary Signal
817 Management (VSM), irrespective of the eventual conclusion (see sections 2.5 and 3.1).
15 https://iris.ema.europa.eu/ - EMA’s online platform for submitting and managing regulatory and scientific procedures for
medicines throughout their lifecycle
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 33/79
第 34 页
818 It is important to have an established protocol for signal validation. However, different steps might be
819 required for different signals depending on their origin. As a minimum, a documented process for
820 validation of signals originating from adverse event databases should be established and made
821 available for inspection.
822 All signal validations performed should be documented and recorded in a traceable manner and the
823 documentation should be sufficient to demonstrate the rationale and outcome of the validation during
824 inspections as well as what data the validation process was based on. See also section 5.
825 2.3.1. Methods for signal validation
826 The methods used to validate a signal may differ depending on the origin and other properties of the
827 signal. Signal validation is usually performed on the initial data present when the signal was detected
828 (the signal trigger). The process of validation is thus not meant to include searching for additional data
829 from other sources, except for in specific circumstances (see sections 2.3.1.1 and 2.3.1.3).
830 2.3.1.1. Initial considerations
831 During signal validation, clinical judgement should always be applied and signals which are biologically
832 implausible should be non-validated. Hence, the product should be able to reasonably cause the event
833 (see examples 29 and 30).
834 However, caution should be exercised as to not dismiss signals due to confirmation bias; i.e. signals
835 should not be non-validated simply because they are not considered to fit the known safety profile. Not
836 everything is discovered in clinical trials and it is important to stay open-minded and unbiased to
837 possible associations.
838 Additionally, a critical assessment of whether the reported events are similar enough to constitute a
839 signal may be necessary in some situations, while in others it may be relevant to consider whether the
840 initial signal needs to be expanded to e.g. cover several similar or medically linked VeDDRA terms in
841 the validation. Certain VeDDRA terms are very broad or non-specific and may encompass a variety of
842 disorders, while others overlap and may be used interchangeably by reporters leading to a need for a
843 combined validation (see examples 31 and 32).
Examples
29: A topical product with a well-established pharmacokinetic profile, known to have no or very
limited systemic uptake, is unlikely to be causally associated with a signal of ‘fracture’.
30: A product not used in the ears would be unlikely to directly cause ‘tympanic rupture’.
31: In case of a detected signal on a VeDDRA PT such as ‘carcinoma NOS’ the cases may all be
reporting the same or at least similar types of carcinomas, or they may be reporting
completely different carcinoma types, which would affect the decision on whether to validate
the signal.
32: In case of a detected SDR on ‘gastrointestinal ulceration’ it may be relevant to also include
any additional cases reporting ‘intestinal perforation’ and ‘gastric perforation’.
844 2.3.1.2. Data screening
845 An initial verification of the data (most often case reports) should be performed. The verification
846 should, as a minimum, ensure that:
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 34/79
第 35 页
847 • There is a temporal association,
848 • The adverse event is not already adequately described in the current product information,
849 • The signal is not based on duplicate reports (which would artificially inflate the apparent
850 number of reports and any potential disproportionality).
851 Temporal association:
852 Temporal association is usually discussed using the term ‘time-to-onset’ (TTO). TTO is usually
853 expressed as the time period (in minutes/hours/days/weeks etc.) from initiation of treatment with the
854 product and until the first clinical signs of the event occur.
855 Presence of a temporal association means that the reported event(s) occurred after the time of the
856 first exposure to the VMP(s). Cases where the events occurred prior to the time of exposure do not
857 have a temporal association (also sometimes expressed as having a ‘negative TTO’). Such cases would
858 never be considered supportive of the signal, regardless of any other factor considered in future
859 causality assessment (see also section 2.4.2).
860 In addition to determining whether there is a temporal association, exclusion of cases with a very
861 unreasonable TTO can be considered, even when a temporal association exists. It should be
862 emphasised that the goal of the validation stage is not to evaluate causality and therefore also not to
863 evaluate the finer details of TTO in each case. However, there may be certain situations where TTO is
864 unrealistically long (e.g. anaphylactic shock occurs many months following a single administration of
865 an inhalant anaesthetic) or unrealistically short (e.g. cancer is diagnosed the day after the first
866 administration). The pharmacokinetic profile of the product needs to be taken into account when
867 evaluating these situations - some products are known to have a very short half-life and are usually
868 given as single administration only, while other products are used for longer periods of time or are
869 specifically designed to elicit a long-term effect or have special delivery systems prolonging the effect.
870 Description in the product information:
871 Known adverse events of a product are generally listed in section 3.6 of the SPC and section 7 of the
872 PL, while additional information contributing to the description of adverse events as well as advice
873 regarding risk mitigation may be found in sections 3.3 and 3.5 of the SPC and the corresponding
874 sections of the PL. Information relating to efficacy and lack thereof is usually included in section 3.4 of
875 the SPC and the corresponding section of the PL.
876 Even if certain VeDDRA terms are not explicitly included in the PI it may be the case that the
877 suspected adverse event is already covered by the text included in the product information.
878 It should be noted that adverse events are included in the PI at VeDDRA LLT level, while signals are
879 usually detected at PT level, therefore it may sometimes be necessary to check the reported LLTs of
880 the cases belonging to a given PT (see also section 2.2.1.8).
881 Furthermore, the product information for a generic or hybrid product should be kept in accordance with
882 the PI of the reference product 16. Therefore, potential signals which are already included in the PI of
883 the reference product should be non-validated. Instead, the generic MAH should submit a variation to
884 the terms of the marketing authorisation without undue delay, in order to rectify any inconsistencies.
885 In cases of doubt as to whether the suspected adverse event(s) is covered by the PI, focus should be
886 on the following:
16 Article 18(6) of Regulation (EU) 2019/6
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 35/79
第 36 页
887 • Are the events of the potential signal more serious or severe in nature compared to the terms
888 already included in the PI? (see also section 2.2.1.4)
889 If the events of the signal are considered more serious compared to the terms included in the PI, the
890 events of the signal should not be considered covered by the current PI (see example 33).
891 Since one PT may include both more and less serious LLTs, situations may arise where a signal is
892 detected on a PT where one LLT is already listed. In such cases it would be necessary to check the
893 reported LLTs of the cases (see example 34).
894 • Are the events of the potential signal more specific compared to the terms already included in
895 the PI?
896 If the events of the signal are more specific compared to the currently approved PI, this may be a
897 signal that would further characterise a known risk (i.e. a new aspect) or the signal may represent a
898 separate subgroup and should therefore not be considered covered by the PI in most cases (see
899 examples 35-36).
900 In these situations, the listed events should not be considered a reason to non-validate the signal on
901 their own. However, the clinical relevance of the signal should be evaluated and the relation to already
902 listed events may play a role in the prioritisation of signals (see section 2.2.1.6).
903 Since one PT may include both more and less specific LLTs, situations may arise where a signal is
904 detected on a PT where one LLT is already listed. In such cases it would be necessary to check the
905 reported LLTs of the cases (see examples 37-38).
906 There may be certain exceptions to this rule, e.g. if the PI already includes adverse events that
907 encompass other events in a way where additional specification may not provide useful information
908 (see example 39).
909 • Does the potential signal constitute a new aspect of the already included term?
910 For example, is it a signal of potentially increased frequency or severity? Or potentially increased
911 frequency or severity in a subset of the target population (e.g. animals with certain comorbidities,
912 certain breeds, certain age groups, pregnant females etc.)? Would it constitute a change in duration or
913 outcome compared to what is currently described in the PI? To answer questions regarding frequency,
914 it would be relevant to check the rate of reporting against the estimated exposure of the VMP
915 (incidence of reported adverse events).
916 • Are the events of the potential signal always linked to the terms included in the PI?
917 Some terms may be linked in some cases and not in others. For such situations, it may be needed to
918 review how many of the cases concern events related to the already listed event (see example 40).
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 36/79
第 37 页
Examples
33: In a case where ‘Elevated creatinine’ is included in the PI, but the validation is performed on
a signal of ‘Acute kidney injury’, the signal would not be considered covered by the current
PI.
34: In a situation where LLT ‘Elevated temperature’ is included in the PI and a signal has been
detected on the PT ‘Hyperthermia’, which the listed event belongs to. However, the cases
mostly report the LLT ‘Fever’. Since ‘Elevated temperature’ may include temperature
increases below the threshold of fever, a signal where the LLT reported is ‘Fever’ would be
considered more serious and more specific.
35: If the currently approved PI lists ‘Injection site reaction’ and the signal undergoing validation
concerns ‘injection site necrosis’ this could be further specification of the risk.
36: If the PI lists ‘Neoplasia’ and the signal concerns ‘injection site neoplasm’ that would further
specify the risk.
37: In a situation where the LLT ‘Colitis’ is already listed in the PI and the cases turn out to
mostly report the specific illness of ‘Ulcerative colitis’ this would be considered to either
further characterise the risk, or it may represent a new risk.
38: Similarly, if the PI lists ‘cardiac enlargement’ however the reported cases of the signal
mostly code ‘hypertrophic dilated cardiomyopathy’ this would be considered further
characterisation (specification) of the risk, as hypertrophic dilated cardiomyopathy is a
specific type of cardiac enlargement.
39: If the signal undergoing validation concerns ‘injection site oedema’, but the PI already lists
‘injection site inflammation’ in section 3.6, the added information of including ‘injection site
oedema’ may be considered negligible, since oedema is a well-known cardinal sign of
inflammation, despite oedema being more specific.
However, in the reverse situation, where ‘injection site oedema’ was the listed term,
‘injection site inflammation’ should not be considered automatically covered by the PI, as
more clinical signs than oedema are usually connected to inflammation (e.g. warmth and
erythema).
40: If ‘Hepatic failure’ is listed in section 3.6 of the PI, validation of a signal of ‘Jaundice’ may
focus solely on cases where non-hepatic causes of jaundice are suspected.
919 Duplicate reports:
920 In cases of doubt as to whether reports are in fact duplicates of each other or not, it is better to
921 consider the reports as unique.
922 It may also be relevant to consider the source of the reports; in situations where all reports originate
923 from the same reporter, it should be assessed whether there may be factors related to the reporter
924 which are more or at least equally likely to have caused the events (e.g. geographical, concurrent
925 treatments, prior disease history, handling of the products or animals, etc.). Signals concerning
926 medication errors, where all reports originate from the same reporter, can usually be non-validated.
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 37/79
第 38 页
927 2.3.1.3. Special types of signals
928 Certain types of signals may need to be assessed differently and therefore other criteria may be
929 relevant during validation as well.
930 As mentioned in section 2.1.1, signals may arise from other sources than analysis of adverse event
931 databases. Often, such signals are detected using qualitative methods and, in some cases, they may
932 be considered validated based on the initial screening during signal detection.
933 If a signal has arisen from more than one source e.g. both a new published article and data from an
934 adverse event database, both sources should be verified using appropriate methods and considered in
935 the final decision on whether to validate the signal or not.
936 Signals arising from monitoring of scientific literature:
937 Validation of a signal arising from scientific literature may encompass an evaluation of the strength of
938 the article(s) or study/studies that triggered the signal but does not necessarily need to encompass
939 searching for additional data. Furthermore, presence of AE reports in adverse event databases
940 concerning the same term may be used to validate the signal where relevant.
941 Signals arising from media (including social media):
942 Signals arising from media attention or social media comprise a challenge during the validation
943 process, as the triggering information may not always be verifiable in nature. In such cases the
944 validation step may include considerations mentioned in the above sections 2.3.1.1 and 2.3.1.2,
945 combined with targeted signal detection in adverse event databases and scientific literature to see if
946 the signal can be verified through other sources.
947 Medication errors:
948 For signals (in animals) involving medication errors it is particularly important to consider whether the
949 reports originate from different (unrelated) sources. Signals of medication errors which can all be
950 traced back to the same source may be non-validated, if they are considered related to the specific
951 source rather than representing a general problem with the product. Furthermore, for reports of
952 medication errors it is particularly relevant to check whether the reported medication or product use
953 error is related to the investigated product, or to another product. Additionally, whether the medication
954 error has led to safety concerns and the severity and seriousness of any such concerns should be taken
955 into consideration, as well as the number of reports and reacting animals. Co-reported safety issues
956 may be checked against the PI.
957 Considering that many of the VeDDRA terms related to medication errors are rather broad, it may also
958 be important to identify patterns in the reported events; are they reporting similar types of errors or
959 completely different errors? In situations where no trend or pattern can be identified, the signal can be
960 non-validated.
961 Signals involving overdose:
962 When referring to overdose in pharmacovigilance, it is usually in reference to use of greater dosages
963 than what is approved (e.g. 20 mg/kg rather than the approved 10 mg/kg). However, depending on
964 the pharmacodynamic and pharmacokinetic properties of a drug (e.g. if accumulation is expected), it
965 may also be relevant to consider prolonged use as overdose in some situations.
966 It is necessary to evaluate at what level overdose becomes relevant to the assessment of the reported
967 events. For some products, a minor overdose may be irrelevant and should thus be evaluated on the
968 same ‘terms’ as non-overdose cases, while for others the therapeutic index is narrower and even a
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 38/79
第 39 页
969 minor overdose is considered significant for the risk profile. Patterns of reported events may also differ
970 between different levels of overdose.
971 The MAH should also keep the product information up to date regarding clinical signs of overdose and
972 systematic exclusion of overdose cases at the signal detection stage is therefore not recommended.
973 Suspicion of batch-related risks:
974 For signals suspected to be tied to batch-related issues, case reports can be checked for presence of
975 batch-numbers as well as which precise batch-numbers apply. However, the basic principles of signal
976 validation should still be considered. In cases where batch-related risks are coupled to safety signals, it
977 is necessary to evaluate whether it could be a PhV alert.
978 Lack of expected efficacy (LEE):
979 When validating a potential signal of lack of efficacy, the data screening will need to be modified for
980 the purpose. If the signal originates from an adverse event database (i.e. is based on AE reports), the
981 cases may be screened for a temporal association in a similar manner as described above, but it will be
982 highly important to know the expected time-to-onset of expected effect as well as duration of the
983 expected effect to determine if there is a temporal association with potential lack of efficacy. In
984 addition to screening for temporal association, duplicate reports and whether lack of efficacy is already
985 described in the PI, it is usually important to assess which expected effect is reported as lacking and
986 whether this is actually an indication of the product. Furthermore, it may be assessed whether the
987 product was handled and used in accordance with the product information and whether the deviations
988 are likely to affect efficacy. In cases where LEE is already described in the product information, it may
989 be of relevance to check whether frequency has increased or whether there are new aspects to the
990 already described lack of efficacy.
991 Signals concerning a risk to the environment:
992 When validating a signal where the primary focus is a risk to the environment, it is important to
993 consider whether the signal can be traced back to use of the product or active substance in animals,
994 either in a specific or unspecific way. If a potential environmental risk cannot be led back to use of the
995 product in animals, it is not considered a signal (see examples 41-43).
996 Environmental signals should otherwise be validated using the same methods as described in sections
997 2.3.1.1 and 2.3.1.2.
Examples
41: Published scientific literature describing a toxic effect on the environment after experimental
exposure (e.g. pouring out the active substance with the purpose of studying the effect)
would not be considered a signal.
42: Adverse event reports describing death of fish and plant life in a local pond after treated
dogs bathed in the water would be considered a signal.
43: Detection of the active substance of a product in natural water sources; This is considered a
signal if the active substance is only commercially available as product(s) for use in animals,
since there would not be other reasonable sources than the product. However, if the
substance is also available from other sources, e.g. human medicinal use, pesticides, feed
additives etc., it would not be considered a signal as the presence in the environment may
be from these sources.
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 39/79
第 40 页
998 2.3.2. Outcome of signal validation
999 When finalising signal validation, the signal should be concluded as either validated or non-validated.
1000 2.3.2.1. Validated signal
1001 A signal is considered validated when the signal validation process has verified that the data triggering
1002 the signal is of sufficient quality and substance to justify a need for further analysis and the event(s) of
1003 the signal are not already adequately described in the product information.
1004 Validation of a signal does not imply that a causal association with the VMP has been established, only
1005 that the observations meet the criteria for a plausible hypothesis, with sufficient data to justify its
1006 progression to further analysis/assessment.
1007 Validated signals will need to be investigated further to assess causality with the product/substance
1008 (see section 2.4). All validated signals should be submitted to authorities through IRIS-VSM following
1009 assessment. The timeline of signal submission in IRIS depends on the outcome of the assessment (see
1010 section 3.1.2.2).
1011 2.3.2.2. Non-Validated signal
1012 A non-validated signal is a signal for which the signal validation process could not verify that the
1013 available data was of sufficient quality and substance to justify further investigation of the possibility of
1014 a causal association between the product(s)/substance(s) and the suspected adverse event(s) or
1015 where the events are considered adequately described in the product information (see section 2.3.1.2).
1016 Just as validation of a signal does not imply establishment of a causal association, nor does non-
1017 validation of a signal mean that a causal association has been ruled out.
1018 Furthermore, a non-validated signal should not be confused with a refuted signal (see also section
1019 2.5.1). A non-validated signal has not been assessed fully and the question of causality has not been
1020 considered, while a refuted signal is a validated signal, where further assessment found the evidence
1021 insufficient to support a causal association.
1022 Outcome and rationale behind non-validation of signals should also be documented in a sufficient
1023 manner to allow for inspection 17 i.e. the decision to non-validate a signal and why this conclusion was
1024 reached should be documented by the MAH and this documentation should be made available to
1025 inspectors.
1026 It should be emphasised that a previous non-validation of a signal does not imply that the signal
1027 should never be investigated again. Following non-validation, monitoring of the issue should usually
1028 return to routine pharmacovigilance practices although there may be situations where future
1029 monitoring can be prioritised lower e.g. if a causal association was determined to be biologically
1030 implausible. Thresholds for when the signal is considered detected again may be established on a case-
1031 by-case basis and would need to consider both latency until redetection, seriousness and severity of
1032 the event as well as the amount of new data available and trends in reporting.
1033 Future validation processes may or may not lead to validation of a previously non-validated signal and
1034 previous non-validation of a signal should not automatically determine the outcome of any future
1035 validation. However, prior knowledge from validations may be taken into account when prioritising
1036 signals for validation (see also sections 2.2.1.9 and 2.2.2).
17 Article 17(6) of CIR (EU) 2021/1281
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 40/79
第 41 页
1037 2.3.3. Signal prioritisation after signal validation
1038 All validated signals will need to go through a full assessment and subsequently be submitted to IRIS-
1039 VSM. In most cases, it will likely make sense to move directly from the validation phase to the
1040 assessment phase. However, where multiple validated signals exist for the same product, it may be
1041 necessary to prioritise which signals to assess first. The general principles, factors to consider and
1042 impact on future monitoring explained in section 2.2. may be applied after validation as needed. If
1043 assessment of a validated signal is deprioritised, the potential impact on safety should be carefully
1044 considered and documented with the validation together with the rationale for deprioritisation.
1045 2.4. Signal Assessment
1046 Signal assessment is the comprehensive process of examining all relevant sources of information with
1047 the aim of determining whether there is a need for action. In order to achieve this, signal assessment
1048 should focus on assessing the causal relationship between a VMP and the AEs of interest for a validated
1049 signal. Therefore, signal assessment focuses on establishing whether the validated signal is likely to
1050 reflect a new causal association or a new aspect of a known causal association with the VMP, as well as
1051 further characterising the risk in order to implement appropriate action, if necessary. This assessment
1052 is based on a cumulative review of all available information, including, but not limited to, case reports,
1053 scientific literature, pharmacological data, clinical information and epidemiological findings.
1054 Clinical judgement is central to signal assessment and requires expert evaluation of the potential
1055 biological mechanism, the clinical and environmental context of the signal and the consistency of
1056 reported events. Within this framework, review of individual case narratives in the context of a case
1057 series is essential, as narratives provide additional critical information that cannot be captured through
1058 coded data or quantitative analyses alone.
1059 The outcome of a signal assessment may ultimately lead to a recommendation for actions, such as
1060 enhanced monitoring, a need for further investigation or implementation of risk minimisation
1061 measures, including updates to the product information (see also sections 2.5 and 2.6).
1062 Signal evaluation should therefore be conducted using a structured and scientifically sound set of
1063 methods applied in a proportionate manner and adapted to the characteristics of the specific signal
1064 under assessment and the available data. These methods aim to support a consistent and transparent
1065 assessment of whether the validated signal represents a new risk or new aspect of a known causal
1066 association of the VMP.
1067 The key elements to consider during signal assessment are outlined below. The list is not exhaustive
1068 and additional methods may be applied where considered relevant.
1069
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 41/79
第 42 页
1070
1071 Figure 5: Signal assessment
1072 2.4.1. Broader search for data
1073 One of the cornerstones of the signal evaluation phase is the conduct of a comprehensive and
1074 systematic search for evidence from all relevant sources, with the objective of ensuring that the safety
1075 topic under assessment is thoroughly explored and that all available evidence is taken into
1076 consideration and put into the correct context. Beyond the analysis of spontaneous reports and
1077 quantitative signal detection outputs, a broad search for additional data is essential to conclude on the
1078 outcome of the signal.
1079 The following section therefore outlines the key data sources that should be considered during signal
1080 assessment. Searching for further information through these sources helps ensure that conclusions are
1081 based on all the available evidence.
1082 2.4.1.1. Adverse event reports
1083 If the signal was not initially triggered by adverse event reports, a cumulative search for cases should
1084 be performed in relevant databases. These cases should be included for assessment of causality (see
1085 also section 2.4.2).
1086 However, even if the signal was triggered from AERs, it may be relevant to perform additional broader
1087 searches for data to ensure all relevant evidence is captured (see also section 2.1.2.1).
1088 It should be assessed whether conceptually related VeDDRA terms are reported, which may represent
1089 clinical sequelae, complications or different stages/descriptors of the same event. A predefined
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 42/79
第 43 页
1090 grouping of VeDDRA terms can be applied to capture the full clinical spectrum and avoid
1091 double-counting and display both grouped and individual-term analyses (see examples 44-45).
Examples
44: Related terms reflecting organ-specific toxicity:
For a suspected hepatic adverse reaction, reports may include VeDDRA terms such as
elevated liver enzymes, jaundice, hepatic failure, hepatic necrosis or hepatic
encephalopathy. These terms may represent different clinical stages, severity levels or
sequelae of the same underlying hepatic injury and should therefore be considered together
using a predefined grouping, while still allowing presentation of individual term analyses.
45: Related terms reflecting a recognised clinical syndrome:
For known syndromes, conceptually linked clinical signs should be assessed collectively. For
example, in a Limping-syndrome, reports of limping, pyrexia, lethargy and joint swelling
may be recorded under separate VeDDRA terms but represent manifestations of a single
clinical entity. Grouping such associated terms helps capture the full syndrome presentation
and avoids underestimation of the signal due to fragmented reporting.
1092 2.4.1.2. Experimental, non-clinical data and post-marketing surveillance studies
1093 Toxicology, pharmacokinetic (PK), pharmacodynamic (PD) and target organ data can provide evidence
1094 of a potential biological mechanism or identify susceptible organ systems or biochemical pathways.
1095 Data sources may be from the pre-authorisation phase of the product itself or new pre-clinical studies,
1096 or from scientific literature concerning the product, active substance(s) or similar substances published
1097 before or after authorisation of the product in the same or a different species (including humans).
1098 In addition, post-authorisation data generated under field conditions, including post-marketing
1099 surveillance studies (PMSS), may provide relevant complementary evidence. The definition of PMSS
1100 can be found in the VGVP module: Glossary 18.
1101 2.4.1.3. Clinical trial findings
1102 When pre- or post-authorisation clinical trial data exist, they can help characterise the adverse event
1103 regarding frequency, time-to-onset, dose administered and controlled-use conditions. Although
1104 veterinary clinical trials often have limited sample sizes, they may corroborate patterns seen in post-
1105 marketing pharmacovigilance.
1106 However, attention should be given to the fact that sample size of clinical trials affects the ability of the
1107 trial to detect very rare, rare and sometimes also uncommon events. Therefore, absence of an adverse
1108 event during clinical trials is not in itself sufficient reason to refute a signal but may point to the event
1109 occurring less commonly. Again, data sources may be from the pre-authorisation phase of the product
1110 itself or may be from scientific literature concerning the product, active substance(s) or similar
1111 substances published before or after authorisation of the product in the same or a different species
1112 (including humans).
18 https://www.ema.europa.eu/en/documents/regulatory-procedural-guideline/guideline-veterinary-good-
pharmacovigilance-practices-vgvp-module-glossary_en.pdf
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 43/79
第 44 页
1113 2.4.1.4. Other published literature (including data on similar substances or within the same
1114 product class)
1115 Scientific literature should be reviewed to identify evidence of the same or similar adverse events in
1116 the same species, in other species or in products that share the same active substance(s), excipients
1117 (if an excipient is suspected in the mechanism of the signal), mechanisms of action or pharmacological
1118 class. Exploring related VeDDRA terms can help capture different clinical manifestations or stages of
1119 the same pathophysiological process.
1120 2.4.1.5. Epidemiological data
1121 Information on the background prevalence of the reaction or disorder might provide contextual
1122 information to assess whether observed reporting incidences may exceed those expected. Background
1123 incidence may be available for the overall reference population or for relevant subpopulations (e.g.
1124 breed, underlying conditions, management practices).
1125 This factor may be particularly relevant for events with a long time-to-onset (e.g. neoplasia) or
1126 multifactorial aetiology (e.g. atopy, inflammatory bowel disease).
1127 However, important limitations must be considered. Comparisons between background incidence
1128 estimates and reporting incidences derived from spontaneous reports are inherently challenging due to
1129 underreporting, reporting biases and differences in study design and population definitions.
1130 Spontaneous reporting systems are not designed to estimate true incidence.
1131 Consequently, background incidence data should not be used in isolation to support or refute a causal
1132 association. Epidemiological information should be interpreted cautiously and qualitatively, in
1133 conjunction with other lines of evidence, including case-level data, biological plausibility and clinical
1134 pattern analysis (see examples 46-48).
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 44/79
第 45 页
Examples
46: Long latency events with high background incidence
For adverse events that develop slowly and have a relatively high background incidence in
the target population, epidemiological context is essential to interpret a potential signal.
For example, a signal concerning mast cell tumours in dogs requires careful consideration of
background incidence, as mast cell tumours are among the most common cutaneous
neoplasms in this species and neoplastic processes typically have a long TTO. Knowledge of
the expected incidence in the general dog population is therefore important to assess
whether the number and pattern of reported cases following exposure to the product exceed
what would be anticipated by chance alone and to avoid false attribution of causality.
47: Multifactorial diseases with fluctuating clinical course
Background incidence data are particularly informative for adverse events or conditions with
multifactorial aetiology and variable clinical expression, such as atopy, inflammatory bowel
disease or chronic renal disease. In these cases, episodic worsening may coincide temporally
with treatment initiation without being causally related. Population‑level data on expected
prevalence and incidence help contextualise observed reporting patterns and support
differentiation between disease progression and a treatment‑related effect.
48: Situations where background incidence alone is insufficient
Abortions in sheep have a known background incidence and are commonly associated with
infectious, nutritional or management-related causes. The background incidence may vary
considerably depending on region, flock management practices and breed. When evaluating
a signal of abortion potentially associated with a VMP, the overall abortion rate observed in
exposed flocks may not exceed the expected background level. However, epidemiological
evaluation should consider whether exposure is associated with:
• Abortions occurring shortly after treatment (plausible and shortened TTO) and/or
• Temporal clustering of abortion cases following product administration.
In such situations, the product may not increase the overall incidence of abortion but may
act as a trigger, shifting the timing of events. Therefore, background incidence alone may be
insufficient to assess the clinical relevance of the signal.
1135 2.4.1.6. Information from other regulatory authorities worldwide
1136 Information originating from non-EU regulatory authorities may provide valuable complementary
1137 evidence during signal assessment. Signals detected, risk communications issued, or regulatory
1138 measures taken by authorities outside the EU can serve as indicators of potential risks or provide
1139 important contextual information, particularly when EU case data are limited or still emerging.
1140 Such information may include relevant pharmacovigilance data as well as regulatory measures adopted
1141 by competent authorities in third countries (e.g. restrictions of use, updates to product information,
1142 suspensions or withdrawals), particularly where these are not yet reflected in the EU product
1143 information and may have an impact on the benefit-risk balance.
1144 The information from these sources should be evaluated critically, taking into account differences in
1145 authorised indications, formulations, target species, patterns of use, reporting systems and regulatory
1146 frameworks. While actions taken by non-EU authorities do not automatically imply a causal association
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 45/79
第 46 页
1147 under EU conditions of use, they may be supportive of the signal, or they may highlight potential class
1148 effects, or help identify subpopulations or conditions of use which may lead to increased risk.
1149 This information may be particularly relevant for serious, unexpected, or rare adverse events, for
1150 products or active substances used globally, or where similar safety concerns emerge independently
1151 across regions (see examples 49-50).
Examples
49: A safety communication issued by a non-EU authority warning of rare but serious
neurological adverse events associated with a specific pharmacological class may include
information in support of a similar signal detected in EU pharmacovigilance data.
50: Regulatory restrictions or label changes implemented in another region following reports of
environmental harm linked to an antiparasitic product (e.g. effects on non-target
invertebrates) usually indicates that an analysis in that country has found a correlation.
Available information may be valuable for assessment of a similar signal in the EU. It may
also provide relevant context when assessing an environmental signal under EU conditions,
particularly where exposure patterns or use scenarios are comparable.
1152 2.4.2. Assessment of case series
1153 The completeness, clarity, internal consistency and reliability of each case are important determinants
1154 of the ability to assess causality. High quality, well-documented cases provide a more robust and
1155 reliable evidence base and are critical for drawing scientifically sound conclusions and determining the
1156 appropriate signal outcome.
1157
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 46/79
第 47 页
1158
1159 Figure 6: Assessment of case series
1160 2.4.2.1. Case selection and search methodology
1161 For any analysis of case series conducted in the context of signal assessment, the methods used for
1162 case identification and selection should be clearly defined and transparently reported. This includes the
1163 databases searched, search strategies (e.g. terms or groupings applied), inclusion and exclusion
1164 criteria and any applied filters (e.g. seriousness, species, time period or level of documentation).
1165 Where focused analyses are conducted as part of the signal assessment (e.g. for a specific adverse
1166 event or signal of interest), it is essential that the approach to case selection is reproducible and
1167 sufficiently detailed to allow independent evaluation. Lack of clarity in case selection or search
1168 methodology may limit interpretability and introduce bias into the assessment.
1169 In situations where a complete review of all individual reports may not be feasible, selection criteria
1170 maybe useful to reduce the number of cases to be reviewed. The use of percentages may support
1171 interpretation, and the criteria should always be clearly stated.
1172 2.4.2.2. Key components of causality assessment at signal level
1173 The detailed review and assessment of adverse event reports (AERs), including case narratives, is a
1174 key component of signal assessment. This step aims to evaluate the potential causal association
1175 between exposure to a VMP and the AEs of interest, thereby contributing to the identification and
1176 characterisation of potential safety risks and supporting timely, proportionate regulatory
1177 decision-making.
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 47/79
第 48 页
1178 The assessment of individual cases in the context of signal assessment should not be interpreted as a
1179 requirement to assign a formal causality category to all reported adverse event reports. A distinction
1180 should be made between assigning a formal overall causality assessment of an AER (which may involve
1181 multiple events and/or multiple products) and the targeted assessment of a defined case series
1182 addressing a specific event (or group of related events) included in the signal and its potential causal
1183 association with a specific product/substance. The purpose of the latter is to determine the extent to
1184 which the available cases support or weaken the product-event association under investigation.
1185 As most AERs are based on clinical suspicion, certainty regarding causality is rarely achievable and not
1186 required for decision making. The structured evaluation of individual cases should provide a
1187 transparent and reproducible framework to assess causal plausibility, distinguishing potentially
1188 product-related events from coincidental or background occurrences.
1189 This assessment also supports the prioritisation of signals for further benefit–risk evaluation and the
1190 consideration of appropriate risk minimisation measures, including potential updates to product
1191 information. Relevant contextual factors, such as co-reporting of lack of expected efficacy and use
1192 outside the terms of the marketing authorisation (off-label use), may also be considered.
1193 All relevant cases should be assessed using a consistent approach and, where appropriate, categorised
1194 according to the strength of evidence supporting a causal association. The total number of supportive
1195 cases should be determined and their strength as evidence of an overall causal association should be
1196 clearly evaluated.
1197 When assessing whether cases support a causal association, the following elements should be
1198 considered:
1199 Quality and completeness of case data
1200 Adverse event reports, including case narratives and structured data fields, should include sufficient
1201 detail to support evaluation of causality and clinical plausibility. In addition to the minimum criteria,
1202 ideally cases should include: a description/diagnosis of the AE, the temporal relationship to VMP
1203 exposure (which may be captured in structured fields such as time-to-onset), relevant animal
1204 characteristics and clinical background, co-reported treatments, outcomes and available follow-up. The
1205 information should also enable identification of potential confounders and alternative aetiologies.
1206 Review of case narratives remains essential to provide clinical context, risk characterisation and
1207 interpretative detail that may not be captured in coded or structured data. However, ideal cases are
1208 not the norm, and flexibility should be applied even when all desired information is not present.
1209 Cases with incomplete, unclear or inconsistent information limit interpretability and may introduce
1210 uncertainty or bias into the assessment. Such poorly documented cases should therefore be
1211 interpreted with caution and weighed accordingly in the overall evaluation, in conjunction with other
1212 lines of evidence.
1213 Where the overall information available from the entire case series is insufficient to assess causality or
1214 clinical plausibility, additional data or clarification should be sought whenever feasible, applying a
1215 risk‑based and proportionate approach. Where relevant, this may include the identification of specific
1216 information needed (e.g. additional diagnostic investigations or clinical details) to support improved
1217 assessment in future cases.
1218 Time-to-onset (TTO)
1219 It should be assessed whether the interval between exposure to the VMP and the onset of the adverse
1220 event is compatible with a biologically plausible mechanism of action. The expected TTO depends on
1221 the pharmacological properties of the product and the nature of the adverse event. Both very short
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 48/79
第 49 页
1222 and delayed onsets may be plausible, provided they are consistent with the pharmacokinetics and
1223 pharmacodynamics of the product and/or known or hypothesised mechanisms, if available.
1224 A TTO that is poorly compatible with the expected pharmacology or pathophysiology may weaken the
1225 causal relationship for the case in question, whereas a consistent and reproducible TTO across cases
1226 can strengthen the signal. Characterising typical TTO over the entire case-series is also relevant for
1227 characterisation of the risk (see examples 51-53).
Examples
51: Compatible TTO (acute pharmacodynamic effect):
In horses treated with an α2‑agonist, profound sedation and bradycardia occurring within 5
minutes of administration is highly compatible with the known mechanism of action and
supports a signal of bradycardia.
51: Compatible TTO (delayed toxicity):
In cats, elevations of Alanine aminotransferase (ALT) and Alkaline phosphatase (ALP)
occurring 3–4 weeks after initiation of an azole antifungal are compatible with subacute or
idiosyncratic hepatotoxicity.
53: Non‑compatible TTO:
The occurrence of an adverse event on the same day as first administration for an event
known to require prolonged exposure (e.g. cataract formation, malignant neoplasia or
chronic organ damage), represents a time‑to‑onset that is not biologically plausible and
argues against a causal relationship.
1228
1229 Dechallenge
1230 A dechallenge refers to the withdrawal of a VMP, or in some cases a reduction in dose, following the
1231 occurrence of the adverse event of interest. Dechallenge may be undertaken intentionally to observe
1232 the effect of treatment withdrawal or may occur incidentally as part of routine clinical management.
1233 A dechallenge is considered positive when the adverse event or clinical signs improve after
1234 discontinuation or dose reduction. When assessing a dechallenge, it is important to consider potential
1235 confounding factors that may influence recovery, such as concurrent treatments, supportive care or
1236 the expected natural course and duration of the event. The time to improvement following withdrawal
1237 should also be evaluated to determine whether it is consistent with the elimination time of the product
1238 and expected end of clinical effect.
1239 A positive dechallenge is generally considered to significantly strengthen the causal association for any
1240 individual case and the presence of one or more cases with positive dechallenge(s) is considered
1241 supportive evidence in favour of the signal (see example 54).
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 49/79
第 50 页
Example
54: A signal is identified in dogs presenting with intense pruritus during treatment with an
ectoparasiticide administered via a collar. Across several cases, a consistent pattern is
observed: pruritus develops shortly after collar application, in the absence of concurrent
medications or other identifiable triggers (e.g. no evidence of flea or other ectoparasite
infestation, allergic disease or environmental factors).
Following removal of the collar, clinical signs improve markedly within a few days without
additional therapeutic intervention. This clear temporal relationship between exposure (collar
application) and resolution of clinical signs upon withdrawal (collar removal) constitutes a
positive dechallenge. Considering the known pharmacological activity of the ectoparasiticide,
this observation supports a causal association between collar use and pruritus.
1242 Rechallenge
1243 A rechallenge occurs when a product or active substance is reintroduced, or the dose is increased,
1244 after a previous dechallenge. A rechallenge is considered positive if the adverse event reoccurs or
1245 worsens following re‑exposure.
1246 Reintroduction of treatment suspected to cause adverse events may be unethical or present
1247 unacceptable risk to the animal, particularly when the suspected adverse events are serious. When a
1248 rechallenge is available, the assessment should consider the same factors as for dechallenge, including
1249 timing of recurrence, consistency of clinical signs and potential alternative explanations.
1250 Cases with positive rechallenge are considered very strong evidence in favour of a causal association
1251 (see example 55).
Example
55: (Same dogs with pruritus from an ectoparasiticide from example 54): In a few of the cases
described in example 54, following complete resolution of pruritus after removal of the
ectoparasiticide collar (positive dechallenge), the same product is re‑applied at a later time.
Intense pruritus reappears shortly after re‑exposure, consistent with the previous clinical
presentation and in the absence of alternative explanations.
This recurrence of clinical signs upon re‑administration constitutes a positive rechallenge and
provides very strong support for a causal relationship between the ectoparasiticide collar and
the observed adverse reaction.
1252 Alternative causes
1253 When assessing a signal, it is essential to systematically evaluate whether the reported adverse event
1254 may be more likely explained by causes other than exposure to the VMP. Consideration of alternative
1255 explanations helps avoid false attribution of causality and supports a balanced, scientifically robust
1256 assessment.
1257 Potential alternative causes may include underlying or concurrent disease, infectious agents,
1258 management or husbandry factors, diagnostic or therapeutic procedures, environmental exposures or
1259 the natural course of the condition being treated.
1260 The assessment of alternative causes should take into account:
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 50/79
第 51 页
1261 • Clinical context: e.g. presence of pre-existing conditions, comorbidities, age-related
1262 susceptibility or disease progression that could reasonably explain the event fully or partially.
1263 • Temporal relationships: Whether the timing of onset is more consistent with an alternative
1264 cause than with exposure to the product.
1265 • Population patterns: e.g. occurrence of similar clinical signs in animals not exposed to the
1266 product, suggesting a background, management‑related or infectious origin.
1267 • Procedures and environment: e.g. recent interventions (e.g. surgery, injection, transport),
1268 housing conditions, feed changes, environmental stressors or toxins.
1269 Alternative causes should be evaluated in the context of all available information and weighed against
1270 evidence supporting a product‑related effect (see examples 56-57).
Examples
56: Diarrhoea is reported in calves following administration of an antibiotic. However, PCR
testing confirms rotavirus infection and similar diarrhoeal clinical signs are observed in
multiple untreated calves on the same farm. The temporal and epidemiological pattern is
therefore more consistent with an infectious background cause than with a drug-related
adverse event, making a causal association with the product unlikely.
57: An acute anaphylactic reaction is reported shortly after administration of multiple vaccines
during the same visit. The presence of more than one concurrently administered product
represents a relevant alternative explanation and prevents attribution of causality to a single
vaccine. Nevertheless, anaphylaxis is a well-defined, highly specific clinical syndrome with a
strong association with medicinal product exposure and few plausible non-drug causes. In
this context, the causal association with vaccination remains possible, even though
attribution to an individual product is not feasible.
1271 Co-reported treatments and potential drug–drug interactions
1272 The presence of co-reported treatments and their potential to contribute to, modify or fully explain the
1273 reported adverse event should be assessed. This assessment should include consideration of possible
1274 drug–drug interactions, including pharmacokinetic and pharmacodynamic interactions, additive or
1275 synergistic effects, as well as confounding by indication or by underlying disease requiring co-
1276 treatment.
1277 Particular attention should be paid to whether a co-reported medicinal product has a known association
1278 with the reported event, as documented in its product information or in the published literature and
1279 whether the time-to-onset is plausible and consistent with the reported event. In such situations,
1280 attribution of the event to the product under investigation may be weakened, especially when the
1281 alternative explanation offers a more plausible or well-established causal relationship (see examples
1282 58-60).
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 51/79
第 52 页
Examples
58: Drug–drug interaction (additive or synergistic effects)
Bradycardia is observed during sedation in a dog receiving an α2-adrenergic agonist in
combination with acepromazine and an opioid. All administered products are known to exert
cardiovascular depressant effects, and the timing and clinical presentation are consistent
with an additive or synergistic pharmacodynamic interaction. Unless overdose of a specific
product is identified, the event is more plausibly explained by combined drug effects rather
than by a single causal agent.
59: Concurrent treatment as the more likely culprit
A signal of gastrointestinal ulceration is reported in dogs receiving long-term phenobarbital
therapy. However, review of case narratives identifies recent concurrent administration of a
non-steroidal anti-inflammatory drug (NSAID) in the period immediately preceding onset of
clinical signs. NSAIDs are well recognised to be associated with gastrointestinal ulceration
and bleeding, with a plausible TTO, whereas phenobarbital is not commonly linked to this
adverse event. In this context, the concurrent NSAID represents a more plausible causal
explanation, weakening attribution to phenobarbital.
60: Equal plausibility of concurrent treatments
A dog receives an NSAID and an antibiotic concurrently and subsequently develops
diarrhoea and anorexia. Both medicinal products are known to be associated with
gastrointestinal adverse events and the TTO is compatible with exposure to either treatment.
In the absence of dechallenge or other distinguishing clinical features, both products
represent plausible contributors and attribution of causality to a single product is not
possible. The causal association with each product should therefore be considered possible.
1283 Adverse events specifically known to be drug-related
1284 Assess whether the reported adverse event represents a highly specific clinical syndrome or a
1285 well‑defined medical condition that is known to be strongly associated with drug exposure and has a
1286 limited number of alternative causes. Adverse events with a characteristic and recognisable clinical
1287 presentation may have a high predictive value for a drug‑related effect and can therefore strongly
1288 support causality, even in the absence of multiple cases.
1289 This criterion focuses on the nature and specificity of the event itself, rather than on its frequency or
1290 TTO. Events that are considered medically significant conditions often fall into this category due to
1291 their seriousness and strong association with medicinal products.
1292 Examples of specific syndromes known to be drug-related include immediate anaphylaxis, toxic
1293 epidermal necrolysis, agranulocytosis or acute hepatic failure, where alternative explanations are
1294 limited and a drug-related mechanism is well recognised (see also example 61).
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 52/79
第 53 页
Example
61: A cat develops a progressive, locally invasive fibrosarcoma at a previous injection site.
Injection-site sarcoma in cats is a well-recognised, specific pathological entity strongly
associated with prior parenteral administration of VMPs, particularly vaccines or other
injectable substances. Although TTO may be prolonged, the characteristic tumour type,
anatomical localisation and clinical progression form a highly specific clinical picture with
limited alternative explanations. The specificity of this malignancy therefore provides strong
support for a causal association with previous injection exposure, even in the absence of
clustering in time or multiple comparable cases.
1295 Consistency across cases
1296 Assess whether similar patterns are observed across multiple cases, including consistency in TTO,
1297 clinical presentation, severity, outcome or affected subpopulations. Reproducible patterns strengthen
1298 the signal and support a causal association, particularly when they are observed independently and are
1299 not readily explained by alternative causes.
1300 Use of predefined VeDDRA groupings is recommended to capture the full clinical spectrum of an event,
1301 including related clinical signs, stages or manifestations and to avoid fragmentation or double counting
1302 of conceptually linked terms (see example 62).
Example
62: A series of reports describes dogs developing urticaria followed by facial oedema within 30
minutes of injection. Similar time-to-onset and clinical progression are observed across
cases, with small breeds disproportionately represented. The consistency of clinical features,
timing and affected subpopulation supports a coherent hypersensitivity signal.
1303 Dose-response relationship (if applicable)
1304 Assess whether there appears to be a relationship between dose, duration of exposure, or cumulative
1305 exposure and the occurrence, severity or duration of the adverse event. The key question to address is
1306 whether higher doses, longer treatment duration or use outside the authorised conditions are
1307 associated with an increased likelihood, earlier onset or greater severity of the event. Identification of
1308 such patterns may support pharmacological plausibility and strengthen the causal assessment.
1309 However, it is important to recognise that not all adverse events are dose dependent. Idiosyncratic
1310 reactions, immunologically mediated events or hypersensitivity reactions may occur independently of
1311 dose or exposure duration. Therefore, absence of a clear dose–response relationship does not exclude
1312 a causal association and should be interpreted in the context of the known pharmacology of the
1313 product and the nature of the event (see examples 63-65).
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 53/79
第 54 页
Examples
63: A cat develops polyuria/polydipsia and elevated liver enzymes following administration of a
high dose of a corticosteroid over a prolonged period. The severity and persistence of clinical
signs increase with continued exposure, consistent with a dose and duration dependent
adverse event and supporting a causal relationship.
64: A cat tolerates long-term treatment with low-dose corticosteroids without adverse events.
Following an increase in dose due to disease flare-up, the cat develops clinical signs
consistent with iatrogenic Cushing’s syndrome. Across several cases, more severe clinical
signs are observed in animals receiving higher doses compared with those treated at lower
doses, supporting a dose-dependent relationship.
65: Several dogs develop acute hypersensitivity reactions following administration of an
ectoparasiticide. The reactions are reported after standard authorised doses and there is no
evidence that higher doses increase either the frequency or severity of the reactions. As
hypersensitivity reactions are typically immune-mediated and not dose-dependent, the
absence of a dose–response relationship does not argue against a causal association.
1314 2.4.2.3. Standardised systems for causality assessment
1315 All case reports included in signal evaluation should be categorised for causality using consistent and
1316 predefined categories and supported by a clear and explicit rationale. The assessment should
1317 document the applied criteria as well as the key elements supporting or arguing against a causal
1318 association, such as temporal relationship, dechallenge/rechallenge, biological plausibility and the
1319 presence of alternative causes.
1320 Use of standardised causality assessment systems ensures transparent, reproducible and harmonised
1321 evaluations, facilitates prioritisation of signals for further follow‑up and supports proportionate and
1322 consistent benefit–risk decision‑making at both MAH and regulatory authority level.
1323 Although standardised systems can support these objectives, there is no longer a regulatory
1324 requirement in the EU to assign a formal overall causality category to each adverse event report.
1325 Nevertheless, a structured and transparent evaluation of causal plausibility remains essential for the
1326 identification and selection of cases supporting a signal, as well as for the overall signal assessment.
1327 This approach reflects the distinction between assigning an overall causality classification to a single
1328 case and evaluating the causal plausibility of a specific medicinal product–event association across a
1329 series of cases within the context of signal management.
1330 Several causality assessment systems and criteria have been described in pharmacovigilance,
1331 however, the only system specifically adapted for veterinary pharmacovigilance is the ABON system,
1332 which is widely used and accepted within the EU veterinary regulatory framework (see examples 66-
1333 69).
1334
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 54/79
第 55 页
Examples
66: ABON System
The ABON system provides a structured approach to classify the likelihood of a causal
association between a VMP and a reported adverse event. It is specifically adapted for use in
veterinary pharmacovigilance and supports consistent and transparent assessment of
individual cases.
The ABON causality categories are defined as follows:
A – Probable: A reasonable temporal relationship to product administration is present; the
event is consistent with known pharmacological or biological properties of the product;
alternative causes are unlikely; and, where available, dechallenge information is supportive.
B – Possible: A temporal relationship exists; however, alternative explanations such as
underlying disease or concurrent treatments cannot be excluded.
O1 – Insufficiently documented (assessment pending): A causal association cannot be
assessed because key information is missing or still under investigation. Additional data may
reasonably be expected to allow further evaluation.
O – Unclassifiable / insufficient data: Available information is inadequate, contradictory or
incomplete, preventing meaningful assessment and no further data are expected to become
available.
N – Unlikely: The event shows no reasonable temporal relationship to product administration
or is more plausibly explained by alternative causes.
The ABON system is based on evaluation of six core factors:
1. Temporal relationship between exposure and event
2. Pharmacological or biological plausibility
3. Clinical and/or pathological findings
4. Prior knowledge, including published literature and product information
5. Consideration of alternative causes
6. Quality and reliability of the available data
By systematically applying these criteria, the ABON system supports harmonised
interpretation of individual cases and provides a common framework and terminology for
causality assessment in veterinary pharmacovigilance.
Several systems, optimised for use in human pharmacovigilance also exist, for example:
67: WHO–UMC Causality Assessment System
The WHO–UMC system classifies the likelihood of a causal association using standardised
categories (Certain, Probable/Likely, Possible, Unlikely, Conditional/Unclassified,
Unassessable).
68: Naranjo Algorithm
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 55/79
第 56 页
The Naranjo algorithm is a questionnaire-based scoring system designed to provide a
structured estimate of causality (Definite, Probable, Possible, Doubtful) based on factors
such as temporal relationship, dechallenge/rechallenge, alternative causes and prior
knowledge.
69: Bradford Hill Criteria
The Bradford Hill criteria describe a set of general epidemiological considerations (e.g.
strength and consistency of association, temporality, biological gradient, plausibility and
coherence) to support inference of causality between an exposure and an outcome.
References:
https://www.who.int/docs/default-source/medicines/pharmacovigilance/whocausality-
assessment.pdf
Naranjo CA et al. A method for estimating the probability of adverse drug reactions.
Clin Pharmacol Ther. Clin Pharmacol Ther. 1981 Aug;30(2):239-45. doi:
10.1038/clpt.1981.154.
Hill AB. The environment and disease: association or causation? Proc R Soc Med. 1965
May;58(5):295–300. doi: 10.1177/003591576505800503
1335 2.4.3. Investigation of biological mechanism
1336 The investigation of a biological mechanism is a critical component in the assessment of a
1337 pharmacovigilance signal. It aims to determine whether there is a scientifically plausible explanation
1338 for the potential causal relationship between the suspected adverse event and the VMP. Establishing
1339 such a mechanism strengthens the case for a causal association and supports regulatory decision-
1340 making.
1341 A structured assessment of biological plausibility typically begins with the key question:
1342 • Is there a reasonable biological mechanism that could explain the adverse event?
1343 If the answer is yes, this biological rationale reinforces the signal, increases its credibility and may
1344 guide further actions (e.g. targeted data collection, risk minimisation measures).
1345 If no plausible mechanism is identified, it is crucial to distinguish between a genuine absence of
1346 biological plausibility, suggesting that a causal association is unlikely, or insufficient data or incomplete
1347 scientific understanding, meaning that the lack of a mechanism should not be interpreted as disproving
1348 the association.
1349 Elements to consider when investigating whether there is a potential biological mechanism are
1350 summarised in this section. The list should not be considered exhaustive.
1351 2.4.3.1. Pharmacodynamics (PD)
1352 It should be evaluated whether the intended mode of action or secondary pharmacological effects of
1353 the product could logically lead to the observed event (see example 70).
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 56/79
第 57 页
Example
70: A prostaglandin analogue used for oestrus synchronisation is known to also cause smooth
muscle contraction. Contraction of smooth muscle in airways would be a plausible
mechanism for an adverse event such as broncho-constriction in susceptible animals.
1354 2.4.3.2. Pharmacokinetics (PK)
1355 The following aspects should be considered: systemic exposure, absorption, distribution, metabolism
1356 and excretion of the product, including the formation of active or toxic metabolites, potential
1357 accumulation with repeated dosing and pharmacokinetic interactions with concurrent treatments (see
1358 examples 71-73).
Examples
71: A product that is predominantly excreted via the kidneys, either unchanged or as
metabolites, may be more plausibly associated with renal adverse events than a product
primarily cleared through hepatic metabolism and biliary excretion. Such considerations are
particularly relevant when evaluating signals related to renal toxicity, changes in renal
function parameters or exacerbation of pre‑existing renal disease.
72: A signal of hepatic encephalopathy is identified for a VMP that is extensively metabolised in
the liver and cases appear to occur primarily in treated animals with pre-existing hepatic
disease. Impaired hepatic metabolism and clearance may result in accumulation of the
parent compound or active metabolites, leading to increased systemic exposure and
neurological signs. Therefore, the pharmacokinetic reliance on hepatic function supports a
plausible mechanism for the observed adverse events.
73: A drug known to cross the blood-brain barrier (in all or certain individuals) is more likely to
cause seizures or other neurological AEs.
1359 2.4.3.3. Toxicology and Non-Clinical Findings
1360 Prior or newly identified knowledge from non-clinical studies, including toxicology, safety pharmacology
1361 and target-organ findings, should be considered when investigating the potential biological mechanism
1362 behind a signal. Such data may originate from the product itself, related substances, or experience in
1363 other animal species or in humans and can provide important insight into relevant biochemical
1364 pathways or susceptible organs (see example 74).
Example
74: Mild liver enzyme elevations observed in repeat‑dose toxicology studies in dogs, or evidence
of hepatotoxicity reported in another species or in humans, can support the plausibility of
hepatobiliary adverse events reported post‑authorisation in the target species.
1365 2.4.3.4. Specific susceptibility of subpopulations
1366 Certain subpopulations may present specific physiological, genetic or pathological characteristics that
1367 increase their vulnerability to adverse events. Such susceptibilities may be related to species-specific
1368 or breed-specific traits, explicit genetic factors (e.g. polymorphisms affecting metabolism or target
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 57/79
第 58 页
1369 receptors), age, reproductive status, or the presence of underlying diseases or concurrent conditions.
1370 Identification of these factors can support biological plausibility and help refine the risk characterisation
1371 of a signal. Where a susceptible subpopulation is identified and considered clinically relevant, this
1372 information may contribute to subsequent risk minimisation measures or amendments to the product
1373 information (see examples 75-78).
Examples
75: Genetic or breed-related susceptibility:
Dogs of collie-type breeds carrying the MDR1 gene mutation have increased susceptibility to
ivermectin-induced neurotoxicity due to impaired P-glycoprotein function at the blood–brain
barrier.
76: Comorbidity-related susceptibility:
Epileptic dogs may be more susceptible to seizures when exposed to medicinal products with
the potential to lower the seizure threshold, such as certain ectoparasiticides.
77: Age-related susceptibility:
Very young puppies, with immature hepatic metabolic capacity, may be at increased risk of
adverse effects when treated with non-steroidal anti-inflammatory drugs (NSAIDs).
78: Species-specific sensitivity:
Cattle exhibit markedly higher sensitivity to alpha-2 adrenergic agonists compared to dogs,
resulting in a greater risk of overdose and pronounced pharmacodynamic effects at relatively
low doses.
1374 2.4.3.5. Literature and Post-Marketing Experience
1375 Published scientific literature and accumulated post‑marketing experience should be considered
1376 specifically to support the assessment of biological plausibility of a signal. This includes relevant
1377 evidence from case reports, mechanistic or pharmacological studies and clinical or field observations
1378 that help interpret the observed association.
1379 Particular attention should be given to evidence of known class effects, experience with the same
1380 active substance or information reflected in the product information of currently authorised products.
1381 Such data can provide important mechanistic context or indicate consistent patterns of adverse events,
1382 even when the precise underlying mechanism is not fully established (see example 79).
Example
79: If adverse gastrointestinal events are well recognised for other VMPs containing the same
active substance, or for products belonging to the same pharmacological class, this supports
the plausibility of a substance or class‑related mechanism.
1383 2.4.4. Special types of signals
1384 Some signals require specific consideration during assessment of causality due to their nature or
1385 inherent characteristic. These include signals related to lack of efficacy, medication errors, user safety,
1386 residues and public health and environmental risks. While the general principles of signal management
1387 apply, the assessment approach may need to be tailored to adequately address the specific risks
1388 involved and their broader implications.
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 58/79
第 59 页
1389 2.4.4.1. Signals concerning lack of efficacy and antimicrobial resistance
1390 Signals of lack of expected efficacy (LEE) may arise from individual reports of treatment failure or from
1391 emerging patterns across multiple cases. Assessment should distinguish between issues related to
1392 conditions of use (e.g. incorrect dosing, non‑compliance, inappropriate indication, management
1393 factors) and those potentially reflecting reduced effectiveness of the product.
1394 Causality assessment should focus on whether the reported lack of effect is reasonable in light of the
1395 authorised indication(s). For products with multiple indications (e.g. treatment of different conditions
1396 or both clinical and microbiological claims), consistency across reports regarding the disputed
1397 indication should be examined.
1398 In the context of antimicrobials and antiparasitics, LEE signals may be indicative of emerging
1399 resistance, including antimicrobial resistance (AMR). Evaluation should consider evidence of reduced
1400 susceptibility of target organisms, regional or species-specific patterns and consistency with known or
1401 plausible resistance mechanisms.
1402 For vaccines, LEE signals may also reflect the emergence or spread of new or variant strains against
1403 which the currently authorised vaccine provides insufficient cross‑protection. In such cases, the
1404 assessment should consider the antigenic match, epidemiological trends and available evidence on
1405 circulating strains, as well as the precise indication of the VMP and statements in the PI regarding
1406 demonstrated cross‑protection against other or specific strains.
1407 Such signals may warrant closer monitoring, further investigation and coordination with resistance
1408 surveillance systems, antimicrobial stewardship activities and, where relevant, epidemiological and
1409 vaccine strain monitoring programmes.
1410 2.4.4.2. Signals concerning medication errors
1411 Signals related to medication errors may involve incorrect dosing, wrong route or species, misuse,
1412 accidental exposure of target or non‑target species, or issues related to product presentation, labelling
1413 or administration devices.
1414 Assessment should focus on identifying:
1415 • The type and frequency of the error;
1416 • Whether a pattern or recurrent cause can be identified;
1417 • Whether the error is reasonably foreseeable and preventable;
1418 • The potential consequences for animal health, user safety, public health or the environment.
1419 Medication errors may be considered preventable when they are reasonably foreseeable and linked to
1420 product design, labelling, instructions for use or real‑world use conditions.
1421 In cases of intentional misuse, the assessment should focus on the extent of occurrence, preventability
1422 and, where identifiable, the underlying reasons for the misuse (see also Guidance notes on the use of
1423 VeDDRA terminology for reporting suspected adverse events in humans and animals).
1424 2.4.4.3. Signals concerning user safety
1425 Signals affecting the people who are handling and administering the VMP, i.e. users (e.g. veterinarians,
1426 farmers, animal owners) require particular attention. All VeDDRA terms are considered medically
1427 important for humans.
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 59/79
第 60 页
1428 The focus of the assessment is often similar to an assessment of medication errors, where focus is
1429 placed on identifying patterns leading up to the risk and the potential preventability of the risk and
1430 available treatment options.
1431 For example, needlestick injuries associated with mineral oil adjuvanted vaccines can result in severe
1432 human tissue injury and long-term morbidity. Such signals may warrant strong warnings in product
1433 information, enhanced training, changes in handling recommendations or implementation of safer
1434 administration practices.
1435 2.4.4.4. Signals concerning Residues/Public health
1436 Signals related to residues in food producing animals may indicate potential issues with withdrawal
1437 periods, compliance with authorised use conditions or residue depletion assumptions. Evaluation
1438 should consider whether reported findings suggest that existing withdrawal periods may be insufficient
1439 under certain conditions of use, animal categories or management practices.
1440 Such signals are of particular importance due to their potential implications for consumer safety and
1441 may require targeted investigations, review of residue data or regulatory action to protect public
1442 health.
1443 2.4.4.5. Signals concerning environmental risks
1444 Environmental signals relate to unintended adverse effects on non-target organisms or ecosystems
1445 following use of VMPs (see also sections 2.2.1.11 and 2.3.1.3). These may include, for example,
1446 adverse effects on dung fauna or water-environments associated with certain antiparasitic products.
1447 Assessment should focus on analysis of plausible routes of environmental exposure, assessment of
1448 preventability and consistency with known ecotoxicological properties of the active substance and
1449 potential real-world consequences for the environment. Where appropriate, the need for risk‑mitigation
1450 measures or restrictions of use should be considered.
1451 2.4.5. Risk characterisation
1452 Risk characterisation aims to describe the risk clearly and comprehensively, based on the totality of
1453 available evidence, in order to support benefit–risk evaluation and regulatory decision‑making. It
1454 focuses on defining the nature, seriousness, severity and frequency of the event(s).
1455 The outcome of the risk characterisation should clearly describe:
1456 • What the risk is, including its clinical definition, seriousness and medical importance, as well as
1457 the observed range of severity and associated uncertainty;
1458 • Who is at risk, taking into account species, breed, genotype, age, physiological status and
1459 relevant comorbidities;
1460 • When and under which conditions it occurs, including time-to-onset, on‑label or off‑label use,
1461 route of administration, dose and conditions of use;
1462 • How often the event occurs, where feasible.
1463 2.4.5.1. Defining the risk precisely
1464 In order to define the risk more precisely, the collective experience of the event(s) from all available
1465 sources should be summarised. The summary should clearly describe what occurred and in which
1466 animals, humans or environmental compartments.
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 60/79
第 61 页
1467 Where available, the summary should include:
1468 • Average time-to-onset, preferably together with the observed range (from–to) and/or
1469 distribution of time-to-onset should be reported where available, to capture variability and
1470 support a more precise characterisation of the temporal pattern of the event;
1471 • Clinical course;
1472 • Reversibility of the event(s), including the outcome of any dechallenge;
1473 • Overall final outcome (recovered, recovered with sequelae, not recovered/outcome unknown,
1474 fatal);
1475 • Any reported long‑term sequelae.
1476 Based on this summary, consideration should be given to identifying the VeDDRA term(s) that most
1477 precisely describe the event(s).
1478 An assessment of overall seriousness, severity and medical importance should then be performed. This
1479 includes classification of whether the event(s) meet seriousness criteria (e.g. death, life-threatening,
1480 significant disability, congenital anomaly or other medically important events), in accordance with
1481 internationally harmonised definitions (VICH GL24) (see also section 2.2.1.4). The range of severity,
1482 from mild to severe and where possible the proportion of severe cases among all assessed cases
1483 should also be characterised. The list of medically important terms may also be used at the step of
1484 signal assessment to assist the assessment of medical importance.
1485 Following causality assessment, the number and/or percentage of supportive cases should be known. A
1486 reporting incidence may be provided, calculated from the cumulative number of spontaneous reports
1487 and the estimated cumulative exposure. Reporting incidence is subject to limitations, including under-
1488 reporting, case ascertainment bias, stimulated reporting and uncertainty in exposure denominators.
1489 Therefore, other data sources, such as clinical trials or post-marketing surveillance studies, should be
1490 considered where available. Reporting incidence may contribute to signal characterisation but is neither
1491 supportive nor dismissive of a signal in itself.
1492 See examples 80-82 on how to summarise information.
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 61/79
第 62 页
Examples
80: Based on EVVet data, the MAH safety database and literature, 18 cases of acute reactions
were identified in adult cattle following administration of an inactivated vaccine. Clinical
signs typically occurred within 5–30 minutes and included dyspnoea, collapse and
convulsions. Most affected animals improved following appropriate symptomatic treatment
and recovered completely. The final outcome was recovery in 13 animals and death in 5; no
long-term sequelae were reported. The most appropriate VeDDRA terms were Anaphylactic
reaction and Acute hypersensitivity. The events met VICH GL24 seriousness criteria (life-
threatening and fatal), with approximately 30% of cases classified as severe. Following
causality assessment, 14 cases were considered supportive. A reporting incidence was
estimated and interpreted with caution.
81: A total of 245 spontaneous reports described vomiting and diarrhoea in dogs treated with an
NSAID at authorised doses. The average time‑to‑onset was 2–5 days after treatment
initiation. Clinical signs were generally mild and reversible, resolving shortly after treatment
discontinuation. Most animals recovered fully and no long‑term sequelae were reported. The
VeDDRA terms Vomiting and Diarrhoea were considered most appropriate. The events did
not meet seriousness criteria and severity was predominantly mild to moderate, with a low
proportion of severe cases. Following causality assessment, 180 cases were considered
supportive. A reporting incidence was calculated as a descriptive element of signal
characterisation.
82: Thirty‑two reports described injection‑site reactions in cats following administration of a
long‑acting injectable product. The average time‑to‑onset was 7–21 days. The clinical course
included local swelling and, in some cases, abscess formation. Reversibility was variable and
surgical intervention was required in several cases. Most animals recovered, although
persistent local masses were reported as long‑term sequelae. The most relevant VeDDRA
terms were Injection site reaction and Injection site mass. While most events were
non‑serious, some were considered medically important. Following causality assessment, 20
cases were considered supportive and a reporting incidence was provided with the usual
limitations.
1493 Frequency estimates
1494 Following causality assessment, the number and/or percentage of supportive cases should be known.
1495 Where feasible, an estimate of event frequency may be provided to support signal characterisation.
1496 Frequency may be estimated either from the number of supportive cases identified through causality
1497 assessment or as a reporting incidence calculated from the cumulative number of spontaneous reports
1498 in relation to the estimated cumulative exposure (see Guideline on the calculation of dose factor to be
1499 submitted to the Union Product Database (UPD)41).
1500 Reporting incidence and other frequency estimates are subject to important limitations inherent to
1501 spontaneous reporting systems, including under‑reporting, reporting bias (including stimulated
1502 reporting following communications or media attention), case ascertainment bias and uncertainty in
1503 exposure denominators. Frequency estimates should therefore be interpreted with caution and
1504 considered hypothesis‑generating only. Where available, additional data sources, such as clinical trials,
1505 post‑authorisation surveillance or epidemiological studies, should be taken into account to support
1506 interpretation. Estimates should be reviewed as more robust data become available.
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 62/79
第 63 页
1507 Irrespective of the method used, frequency estimates cannot be used to conclude on the likelihood of a
1508 causal association. They represent one element of risk characterisation and may become relevant
1509 where the outcome of the signal assessment is an update of the product information (see example 83).
Example
83: There were forty cases reported cumulatively. Twenty‑five of these were considered
supportive following causality assessment. Frequency may be estimated from the 25
supportive cases alone, or a reporting incidence may be calculated from the total number of
40 reported cases.
Based on estimated cumulative exposure, the reporting incidence suggests that the
frequency is rare while the frequency estimate from supportive cases alone suggests that
the event is very rare. However, given the likelihood of under‑reporting and uncertainty in
exposure estimates, estimates should be interpreted cautiously and used primarily to
support signal characterisation rather than to quantify true incidence. If one of the two
frequencies are to be used for a future product information update, the worst-case-scenario
is generally preferable.
1510 2.4.5.2. Identifying special circumstances of the risk
1511 The risk characterisation should identify and describe any special circumstances under which the risk is
1512 more likely to occur or may be exacerbated. These circumstances may relate to specific
1513 subpopulations, conditions of use, or product-related factors and are critical for refining risk awareness
1514 and informing appropriate risk minimisation measures. Additionally, any circumstances identified which
1515 reduce the risk of the event(s) either with respect to occurrence or outcome, should also be described.
1516 Restriction to specific subpopulations
1517 Certain risks may be confined to, or more pronounced in, particular subgroups due to physiological,
1518 genetic or pathological characteristics. These may include:
1519 • Species, breeds or genotypes: For example, dogs of herding breeds carrying the MDR1
1520 (ABCB1) gene mutation are predisposed to neurotoxicity when exposed to certain macrocyclic
1521 lactones or loperamide, necessitating tailored warnings, dose restrictions or avoidance at
1522 higher doses.
1523 • Age or physiological status: Increased susceptibility may be observed in neonates, geriatric
1524 animals, pregnant or lactating animals, or specific production classes (e.g. high-yield dairy
1525 cattle).
1526 • Comorbidities and concurrent medication: Underlying renal or hepatic impairment,
1527 dehydration, or the use of interacting medicinal products (e.g. concurrent NSAIDs and
1528 corticosteroids) may increase the likelihood or severity of adverse outcomes.
1529 Conditions of use
1530 The circumstances under which a product is used may significantly influence the risk profile and should
1531 therefore be carefully evaluated. Relevant factors may include:
1532 • Use outside authorised conditions: Off-label use, overdose, incorrect route or frequency of
1533 administration, misuse, medication errors, or issues related to product quality or specific
1534 batches.
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 63/79
第 64 页
1535 • Environmental or management factors: external conditions such as temperature extremes,
1536 transport or handling stress, housing conditions or, in aquaculture, water quality parameters,
1537 which may contribute to or exacerbate adverse effects.
1538 Product and formulation-related factors
1539 Specific characteristics of the product or its formulation may influence the occurrence or nature of the
1540 risk, including:
1541 • Formulation and excipients: Certain excipients, solvents or adjuvants may contribute to
1542 hypersensitivity reactions or exacerbate toxicity.
1543 • Presentation and delivery devices: Issues related to applicators, syringes or other delivery
1544 systems may facilitate incorrect dosing, misuse or accidental exposure.
1545 See examples 84-86.
Examples
84: Genotype-restricted neurotoxicity (macrocyclic lactones/loperamide in MDR1-mutant dogs):
Use of certain macrocyclic lactones or loperamide is associated with acute neurological
adverse events (e.g. ataxia, tremors, seizures) in dogs carrying the MDR1 (ABCB1) gene
mutation, predominantly affecting herding breeds. Events may be severe but are uncommon
and mainly reported following off-label or high-dose use or in the presence of interacting
co-medications. When products are used according to authorised conditions and
breed-specific precautions are followed, the likelihood of occurrence is low. Risk minimisation
includes strengthened SPC/PL warnings, avoidance or dose adjustment in susceptible breeds
and targeted monitoring.
85: Several reports described increased incidence and severity of adverse cardiovascular and
respiratory events in pigs treated with a sedative during periods of high ambient
temperature. Most cases occurred during summer months and were associated with
treatment administered shortly before transport under conditions of heat and handling
stress. Clinical signs included hyperthermia, respiratory distress and collapse, with rapid
onset after administration. When the same product was used under comparable conditions
but at moderate temperatures and without transport‑related stress, adverse events were
infrequently reported and generally non‑serious. This suggests that high environmental
temperature combined with handling stress may exacerbate the risk and severity of adverse
outcomes.
86: Several reports described increased acute gill irritation and respiratory distress in salmon
following bath treatment with an antiparasitic medicinal product formulated as a
concentrated aqueous solution. Clinical signs occurred during or shortly after exposure and
included increased opercular movement, surfacing behaviour and, in some cases, mortality.
The events were more frequently reported when the product was used at the upper end of
the authorised concentration range and in sea water with elevated salinity. Comparable
adverse events were rarely reported for alternative formulations of the same active
substance administered via in‑feed treatment. This pattern suggests that formulation‑related
characteristics affecting local exposure at the gill level, in combination with water
parameters, may exacerbate the risk.
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 64/79
第 65 页
1546
1547 Figure 7: Risk characterisation checklist
1548 2.4.6. Overall conclusion
1549 The overall conclusion of a signal assessment should be comprised of both a risk statement and an
1550 overall evaluation of the causal association.
1551 Risk statement
1552 The outcome of risk characterisation should be summarised in a clear and concise risk statement that
1553 captures the key elements necessary to support regulatory decision‑making and risk communication.
1554 The risk statement should explicitly address:
1555 • Nature of the risk: A precise clinical description of the adverse event, including seriousness and
1556 medical importance and an assessment of how frequently it occurs and how severe it appears
1557 to be acknowledging any uncertainty associated with the available data.
1558 • Population at risk: Identification of the animal, user or environmental subpopulations at
1559 increased risk, including relevant species, breeds, genotypes, age groups, physiological states
1560 or comorbidities.
1561 • Circumstances of occurrence: Description of when and under which conditions the risk occurs,
1562 including authorised versus off‑label use, route of administration, dose or duration of treatment
1563 and relevant environmental or management factors.
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 65/79
第 66 页
1564 • Broader implications: Consideration of potential implications for user safety, public health or
1565 the environment, where applicable.
1566 • Likelihood under normal conditions of use: An overall assessment of the likelihood of the risk
1567 occurring when the product is used in accordance with the authorised conditions.
1568 • Consequences and preventability of the risk: Outcome and consequences of the risk, including
1569 any identified circumstances which may reduce the risk (occurrence or severity), or ways to
1570 treat the consequences.
1571 Causality statement
1572 The overall conclusion on causality should be derived from a structured and balanced assessment of
1573 the likelihood of a causal association between the VMP and the reported adverse event(s), integrating
1574 the key findings from signal assessment and characterisation and reflecting the weight of the available
1575 evidence.
1576 The conclusion should summarise the strength of evidence supporting or refuting the signal, including
1577 biological and pharmacological plausibility and the principal causality elements considered (e.g.
1578 temporal relationship, dechallenge and rechallenge where available, evaluation of alternative
1579 explanations and consistency across cases).
1580 Key uncertainties and limitations should be explicitly acknowledged, including small numbers of cases,
1581 incomplete documentation, reporting bias, confounding by underlying disease or concurrent medication
1582 and uncertainty in exposure and frequency estimates, together with their impact on interpretation of
1583 frequency and severity. Limitations of studies and other literature may also be summarised.
1584 Overall, the conclusion should set out a reasoned, evidence‑based assessment of the likelihood of a
1585 causal association, providing a transparent and proportionate basis for choosing the correct outcome of
1586 the signal assessment.
1587 2.5. Outcomes
1588 The outcome of the signal assessment and the justification thereof, shall be recorded internally and the
1589 rationale shall be kept ready for inspection17. The justification should explain the reasoning behind the
1590 outcome, not simply confirm that data were reviewed.
1591 All validated and assessed signals, including refuted signals and signals under close monitoring, should
1592 be submitted to IRIS-VSM in accordance with the Best Practice Guide9.
1593 Where the regulatory outcome of the signal assessment differs from the MAH’s initial proposal, it is
1594 expected that the MAH takes due account of the position (e.g. when conducting follow-up activities), in
1595 line with its pharmacovigilance obligations under Regulation (EU) 2019/6.
1596 A signal assessment should culminate in one of the following outcomes: the signal is refuted, the signal
1597 is placed under close monitoring, or the signal is determined to require risk minimisation measures.
1598 2.5.1. Signal refuted
1599 A refuted signal is a validated signal for which, following further assessment, a causal association is not
1600 supported by the available evidence at the time of assessment. The subject of the signal should return
1601 to routine pharmacovigilance activities. However, a refuted signal may be considered again as a
1602 potential signal, and it may be re-detected and re-submitted if new relevant information becomes
1603 available (see section 2.2.2).
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 66/79
第 67 页
1604 Refuted signals should not be confused with non-validated signals (see section 2.3.2.2). Non-validated
1605 signals have not undergone full signal assessment and therefore the question of causality has not been
1606 considered.
1607 Where signal assessment has been completed and the available evidence was not sufficient to support
1608 a causal association, the rationale should be clearly documented. See also section 2.4 regarding factors
1609 to consider.
1610 2.5.2. Signal under close monitoring
1611 Where signal assessment indicates a potential causal association between the VMP and the adverse
1612 event, but the available evidence is currently insufficient to confirm or refute this association, the
1613 signal is not closed and the VMP is placed under close monitoring. At the time of assessment, the
1614 signal does not justify immediate risk minimisation measures.
1615 Signals under close monitoring require increased pharmacovigilance activities, which may include
1616 increased monitoring frequency, a focus on specific searches, targeted follow-up of incoming adverse
1617 event reports, initiation of a PMSS or survey etc. The scope and nature of monitoring should be
1618 determined by the data required to conclude the signal assessment, including focused review of
1619 relevant incoming information and consideration of cumulative as well as period-specific data. The
1620 signal should be re-assessed as additional evidence becomes available.
1621 MAHs are required to submit follow-up assessments of their findings to IRIS-VSM at the latest at the
1622 next annual statement submission date. Shorter reporting intervals may be set by the MAH or in
1623 certain circumstances by regulators. If shorter reporting intervals are set by the MAH, they are
1624 encouraged to notify the relevant regulators by including this information in the signal assessment
1625 report. The signal assessment report should clearly state the additional data which has been
1626 considered since the last submission as well as a reasoning for the proposed outcome.
1627 Signals under close monitoring would normally be expected to progress to a conclusion (signal refuted
1628 or signal requiring risk minimisation measures) within two years. Where close monitoring needs to
1629 extend beyond this period, the continued need for close monitoring should be justified, taking into
1630 account whether the data required to conclude the signal assessment can reasonably be expected to
1631 become available. Cessation of close monitoring may be proposed by submitting a new signal (e.g.
1632 with the proposed outcome of ‘refuted’ or ‘amendment of the product information’), supported by a
1633 detailed justification. The justification provided within the signal submission will be assessed by
1634 regulators and the MAH will be informed of the recommended outcome (see also section 3.1.2.3).
1635 For signals related to outcomes with long latency or slow accumulation of evidence (e.g. neoplasia or
1636 rare adverse events in specific sub-populations), longer periods of close monitoring may be
1637 appropriate, provided that this is supported by a clear scientific rationale and an active monitoring
1638 strategy. If uncertainty remains despite close monitoring, further data generation (e.g. post-marketing
1639 surveillance studies) should be considered.
1640 2.5.2.1. Post marketing surveillance studies
1641 In accordance with Article 15(1) of Commission Implementing Regulation (EU) 2021/1281 post-
1642 marketing surveillance studies may be conducted by MAHs on their own initiative or shall be conducted
1643 by MAHs on request of a competent authority or the Agency in accordance with Article 76(3) and (4) of
1644 Regulation (EU) 2019/6.
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 67/79
第 68 页
1645 2.5.3. Signal requiring risk minimisation measures
1646 A signal requiring risk minimisation measures is a validated signal for which signal assessment
1647 supports a likely causal association between the VMP and the adverse event. The signal is considered
1648 to warrant risk minimisation measures.
1649 Where a signal requires risk minimisation measures, appropriate measures should be identified and
1650 implemented in a timely manner. The choice of risk minimisation measures should be proportionate
1651 (see also section 2.6).
1652
Outcome of assessment At the time of assessment Next step
Signal refuted Causality not supported Routine pharmacovigilance
activities
Signal under close monitoring Potential causal association; Additional activities (in addition
evidence inconclusive to routine monitoring),
Periodic update with signal re-
assessment
Signal requiring risk Causality supported Implementing risk minimisation
minimisation measures measures, monitor
effectiveness,
Routine pharmacovigilance
1653
1654 Table 2: Summary of signal assessment outcomes and subsequent actions
1655 2.6. Risk minimisation measures
1656 Risk minimisation measures are interventions intended to reduce the occurrence, severity and / or
1657 impact of adverse events associated with veterinary medicines. Their aim is to support veterinary
1658 healthcare professionals and animal owners in addressing the risks associated with veterinary
1659 medicines.
1660 The key routine risk minimisation tools available for authorised veterinary medicines are the summary
1661 of product characteristics (SPC) and the package leaflet (PL). These documents include a description of
1662 the risk of use of the veterinary medicine and intended actions for risk minimisation (e.g.
1663 contraindications for use). Following signal assessment it may be concluded that risk minimisation
1664 measures are required to help manage a specific risk that has been identified.
1665 If it is concluded that a causal association with a new or changed risk is considered likely (see also
1666 sections 2.4 and 2.5) then the need for any of the risk minimisation measures described in this section
1667 and/or any other measures should be considered, including a variation to the terms of the marketing
1668 authorisation. It is recommended to follow up on the effect of risk minimisation measures on the
1669 identified risk. Additionally, MAHs may be requested to specifically evaluate the impact.
1670 2.6.1. Amendment of the product information
1671 Following signal assessment changes to the product information may be required by either the
1672 addition, amendment or removal of currently authorised text e.g. in the SPC:
1673 • Section 3.2 e.g. removal or restriction of an indication.
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 68/79
第 69 页
1674 • Section 3.3 e.g. a new contraindication.
1675 • Section 3.5 e.g. a new special precaution for use in the target species, user safety or
1676 environmental risks.
1677 • Section 3.6 e.g. a new adverse event or change of frequency.
1678 • Section 3.7 e.g. addition of a warning that the product cannot be used during pregnancy or
1679 lactation.
1680 • Section 3.8 e.g. limitations on concurrent use with other medicinal products.
1681 • Section 3.9 e.g. a change to instructions for use.
1682 • Section 3.10 e.g. addition of clinical signs of overdose.
1683 • Section 3.11 e.g. special restrictions for use.
1684 • Section 10 e.g. a change to the prescription classification.
1685 The Regulatory outcome of the signal in IRIS will be published as ‘Amendment of the product
1686 information’ and, in accordance with Article 77(10) of Regulation (EU) 2019/6, the applicant should
1687 submit either a G.I.19 or G.I.4 variation requiring assessment. This variation will be assessed by the
1688 relevant authorities as per the established variation procedures.
1689 When submitting the variation, the MAH should refer to all relevant guidance on variations requiring
1690 assessment. Depending on the update, it may be necessary for MAHs to publish communication to end
1691 users informing of the change to the product information. EMA/the NCAs may also publish
1692 communication regarding changes to product information if considered necessary.
1693 As stated in section 2.3.1.2 if the product information update is related to harmonisation with the
1694 reference product, then this is not considered part of the signal management process and should be
1695 addressed via the relevant variation procedure.
1696 2.6.2. Additional risk communication
1697 Following signal assessment, if there is a need for immediate or targeted risk communication, a Direct
1698 Animal Healthcare Professional Communication (DaHPC) can be circulated to veterinarians and other
1699 animal healthcare professionals to inform them of important new safety information about a veterinary
1700 medicine and any actions they should take. This could be to advise that:
1701 • A new serious adverse event has been discovered which requires direct communication with
1702 animal healthcare professionals.
1703 • Another important change has occurred e.g. a restriction of an indication, a new
1704 contraindication, suspension, withdrawal or revocation of a product for safety reasons.
1705 • The product isn't being used as intended e.g. where post-marketing surveillance has
1706 highlighted a need for increased awareness regarding specific characteristics of the product.
1707 DaHPCs should contain specific advice, recommendations or information regarding a veterinary
1708 pharmacovigilance concern, in particular which may require certain action to be taken or changes in
1709 relation to the administration of a VMP.
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 69/79
第 70 页
1710 For more detailed information on the preparation, content, approval and dissemination of DaHPCs
1711 please refer to the VGVP module on veterinary pharmacovigilance communication 19.
1712 Other communication initiatives could include short information videos on EMA/NCA/MAH websites,
1713 publication of articles raising awareness of an identified risk in veterinary journals or information on
1714 the correct use of a product to minimise risk for the animal or the user.
1715 2.6.3. Batch recall, suspension, withdrawal or revocation of the marketing
1716 authorisation
1717 On occasion, the risk identified during signal assessment may require the recall of particular batches of
1718 product(s). This recall can be to wholesale, veterinary or other end user level (e.g. farmers, companion
1719 animal owners, etc). On rare occasions, in accordance with Article 129 of Regulation (EU) 2019/6, the
1720 suspension of the sale of any of the product may be required for a certain period of time until the
1721 identified risk has been mitigated.
1722 On very rare occasions, if the newly identified or changed risk(s) is deemed to outweigh the benefit of
1723 the product then the marketing authorisation may be revoked, meaning the VMP is withdrawn from the
1724 market and cannot be marketed anymore without submission of a new marketing authorisation
1725 application 20.
1726 2.6.4. Other risk minimisation measures
1727 In addition to the risk minimisation measures listed above other less frequently used measures are
1728 also available if required.
1729 2.6.4.1. Educational material
1730 Another option for minimising an identified risk and assisting with measures for prevention could be to
1731 provide educational material.
1732 Examples may include:
1733 • Printed leaflets for distribution in veterinary practices or other animal healthcare settings.
1734 • In-person training for veterinarians/other animal healthcare professionals by representatives of
1735 the MAH.
1736 • MAH/NCA social media content.
1737 • Internal training of sales representatives/product managers or other relevant staff.
1738 • Creation of a user manual or update of an existing user manual.
1739 See examples 87-89.
19 https://www.ema.europa.eu/en/documents/regulatory-procedural-guideline/guideline-veterinary-good-
pharmacovigilance-practices-vgvp-module-veterinary-pharmacovigilance-communication_en.pdf
20 Article 130 of Regulation (EU) 2019/6
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 70/79
第 71 页
Examples
87: A signal of a medication error concerning product use leading to increased risk to users: An
educational video or pamphlet is made to support the correct use and administration of the
product to avoid accidents.
88: A signal concerning user safety with a very toxic substance (e.g. products used for
chemotherapy): A video or pamphlet is made explaining the importance and showing the
correct use of personal protective equipment.
89: A signal where adequate monitoring may prevent the AE: A pamphlet is created for the vet
and a ’monitoring’ schedule/diary for the owner to help them keep track of veterinary visits
and explain the importance of them.
1740
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 71/79
第 72 页
1741 3. Reporting to the EU network
Signal management reportings Other reportings
Annual Signals PhV alerts Sales data & Regulatory
statements dose factor measures
(section 3.1.2) (section 3.1.3)
worldwide
(section 3.1.1) (section 3.2.1)
(section 3.2.2)
Frequency and Annually Action Within 3 days Annually (end Within 30 days
timing required: of detection of February)
(due date), or
within 30 days
higher
of assessment
frequency, if
applicable
no action: at
the latest by
due date
Where IRIS-VSM IRIS-VSM IRIS-VSM UPD IRIS-VSM
Relevant links IRIS portal IRIS portal IRIS portal UPD portal IRIS portal
Due date list BPG Dose factor
guidance
AR template
1742 Table 3: Reporting to the EU network
1743 3.1. Signal management reporting obligations
1744 MAHs are responsible 21 for continuously monitoring the benefit-risk balance of their VMPs, conducting
1745 the signal management process in compliance with the pharmacovigilance guidelines and recording
1746 their conclusions in the UPhV-database (IRIS-VSM).
1747 There are two main reporting responsibilities for MAHs; the annual conclusion on the benefit-risk
1748 balance (annual statement), which needs to be submitted at least annually for all VMPs with a valid
1749 marketing authorisation and the submission of all validated and assessed signals as applicable.
1750 Signal management submissions and annual statement submissions are also accessible by the public
1751 on the IRIS platform (public register) 22.
1752 3.1.1. Annual statements
1753 As laid down in the regulations, MAHs shall at least annually record a conclusion on the benefit-risk
1754 balance for each of their products in the UPhV-database and confirm that the signal management
1755 process has been conducted 23. This should be done irrespective of whether any signals were detected
1756 throughout the year. Furthermore, it applies to all VMPs with a valid marketing authorisation,
1757 regardless of the marketing status. Where MAHs are responsible for the same or a similar veterinary
21 Article 77 of Regulation (EU) 2019/6 and Article 18 of CIR (EU) 2021/1281
22 https://iris.ema.europa.eu/publicregistersignal/
23 Article 81(2) of Regulation (EU) 2019/6 and Article 19 of CIR (EU) 2021/1281
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 72/79
第 73 页
1758 medicinal product 24, authorised in different Member States through different authorisation procedures,
1759 the signal management process may be performed at active substance level for all the products
1760 combined.
1761 At the time of submission of the annual statement, the MAH will have to confirm that all validated
1762 signals have been submitted following assessment accordingly in the UPhV-database (IRIS-VSM).
1763 All MAHs shall conduct at least one signal detection analysis per year for each of their active
1764 substances or products in the UPhV-database7. This signal detection analysis should be performed
1765 within 2 months before the annual due date (see section 2.1).
1766 3.1.1.1. Marketing authorisation holder’s responsibilities
1767 The responsibility for the MAH to submit an annual statement in accordance with Regulation (EU)
1768 2019/6 is independent of the marketing status. Even if an authorised VMP is not marketed after
1769 authorisation, the annual statement should be submitted annually.
1770 In order to limit administrative burden, it is permitted to group same or similar products 25 from the
1771 same MAH and submit a grouped annual statement. However, it is not permitted to group products
1772 that are not the same or similar e.g. pharmaceuticals and immunologicals.
1773 If the marketing authorisation of a VMP is suspended this does not affect the requirement to submit an
1774 annual statement, since this situation corresponds to a temporary marketing cessation which could be
1775 lifted. The usual requirements in terms of annual statement submission apply. In this case the option
1776 in the annual statement ‘The marketing authorisation is currently suspended for reasons related to
1777 safety’ should be chosen.
1778 In all other cases, the MAH is asked to confirm that the benefit-risk balance remains favourable. This
1779 option should also be chosen in case a variation is ongoing concerning the safety of the VMP.
1780 For VMPs for which the marketing authorisation has been withdrawn, revoked or surrendered, there is
1781 no need to submit annual statements after cessation of validity of the marketing authorisation.
1782 3.1.1.2. Timelines for annual statement submissions (recommended due dates)
1783 Annual statements should be submitted in accordance with the list of recommended due dates 26 for
1784 VMPs including homeopathic medicinal products.
1785 The due dates are allocated to ATCvet codes and spread throughout the year with the aim to even out
1786 the submission of annual statements.
1787 MAHs with VMPs with an ATCvet code that is not already included in the list of due dates are advised to
1788 send a proposal for a due date to EMA via the ‘Send a question to EMA online form’ 27. The proposed
1789 new due date may take into account similar ATCvet codes with their assigned due dates. The MAH will
1790 be informed about the allocated date in due course.
24 Article 17(2) of Commission Implementing Regulation (EU) 2021/1281
25 https://www.ema.europa.eu/en/documents/scientific-guideline/vich-gl24-guideline-pharmacovigilance-veterinary-
medicinal-products-management-adverse-event-reports-aers_en.pdf
26 https://www.ema.europa.eu/en/documents/other/recommended-due-dates-submitting-annual-statements-centrally-non-
centrally-authorised-products-caps-non-caps-homeopathic-medicinal-products_en.xlsx
27 https://www.ema.europa.eu/en/about-us/contacts-european-medicines-agency/send-question-european-medicines-
agency
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 73/79
第 74 页
1791 At least once a year the list of recommended due dates will be updated with the new ATCvet codes as
1792 applicable and published on the EMA website 28.
1793 MAHs are reminded that the conclusion on the benefit-risk balance shall be recorded at least
1794 annually23. ‘Annually’ should be understood as within a 12 months’ time period and it is strongly
1795 recommended to submit them according to the due dates in order to facilitate the predictability of the
1796 annual statement submissions and to even out the workload of all involved stakeholders.
1797 It should be emphasised that the annual submission of all results and outcomes of the signal
1798 management process, including a conclusion on the benefit-risk balance, is the default option.
1799 However, some VMPs can have a higher frequency as a requirement of the marketing authorisation23
1800 or recommended by competent authorities/the Agency. In such cases, the annual statement should be
1801 submitted in accordance with the requested frequency.
1802 3.1.1.3. EU network/NCAs’ responsibilities
1803 Although the main responsibility for the assessment of the benefit-risk balance lies with the MAH, the
1804 regulators may perform an analysis following a risk-based approach to determine whether the MAH has
1805 detected and submitted all relevant signals in accordance with section 3.1.2.
1806 Any identified discrepancies and recommended risk minimisation measures (see section 2.6) will be
1807 communicated to the MAH.
1808 Furthermore, in some Member States, additional national legal requirements may exist for national
1809 competent authorities to perform safety surveillance/pharmacovigilance especially for nationally
1810 authorised VMPs, which may translate into specific requests to MAHs.
1811 3.1.2. Signal submissions
1812 3.1.2.1. Marketing authorisation holder’s responsibilities
1813 It is the responsibility of the MAH to continuously monitor the benefit-risk balance of their VMPs.
1814 The MAH should notify any change to the benefit-risk balance or a new risk to the competent
1815 authorities or to the Agency as applicable without delay and no later than within 30 days and take the
1816 necessary action 29.
1817 The notification of a signal should be done through the IRIS-VSM platform 30 for all validated signals
1818 once the assessment has been finalised, regardless of the MAH’s conclusions. Detailed guidance on the
1819 functionality is available on the IRIS platform page 31. In addition, the Veterinary Union
1820 Pharmacovigilance Database - Best Practice GuideError! Bookmark not defined. provides further practical
1821 guidance for signal submissions.
1822 The MAHs are advised to send separate signals for VeDDRA terms that are not medically linked, as well
1823 as for each target species. Same or similar products from the same MAH as well as medically linked
1824 VeDDRA terms (even if they are included in different SOCs) can be grouped in one signal submission.
1825 However, separate signals should be submitted for different proposed outcomes.
28 https://www.ema.europa.eu/en/veterinary-regulatory-overview/post-authorisation-veterinary-
medicines/pharmacovigilance-veterinary-medicines/signal-management-veterinary-medicines
29 Articles 81(2) and 77(10) of Regulation (EU) 2019/6
30 https://iris.ema.europa.eu/
31 https://iris.ema.europa.eu/homeguidance/
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 74/79
第 75 页
1826 The signal submission should be accompanied by an assessment report 32 outlining the
1827 pharmacovigilance data and their evaluation, as well as the conclusions of the MAH. In certain cases,
1828 several related signals are assessed in one assessment report. In such a case the same assessment
1829 report can be uploaded for several signal submissions.
1830 Once a signal has been submitted via the IRIS-VSM platform it can, in general, not be amended.
1831 However, in case of a mistake, the MAH can request a correction via the EMA IT service desk 33.
1832 Changes to a submitted signal should not be done after the regulatory assessment has been concluded
1833 and the outcome has been published in IRIS, unless there is a specific request for a correction by the
1834 responsible competent authority.
1835 For further information on the possible conclusions of the signal evaluation, please see section 2.5.
1836 In general, when the outcome of a signal leads to an amendment of the product information, it is
1837 expected that a signal with the MAHs conclusion ‘Amendment of the product information’ is submitted
1838 in IRIS.
1839 However, in case the amendment of the product information has been requested by the responsible
1840 competent authority/ the CVMP following assessment of an already existing signal, then a variation can
1841 be directly submitted.
1842 For VMPs for which the marketing authorisation has been withdrawn, revoked or surrendered it is
1843 strongly recommended that the MAH continues to perform pharmacovigilance activities, taking into
1844 consideration the recommendations in the VGVP module on collection and recording of suspected
1845 adverse events for veterinary medicinal products8.
1846 3.1.2.2. Timelines for signal management submissions
1847 For validated signals for which the MAH identified a change to the benefit-risk balance or a new risk
1848 (e.g. the MAH proposes regulatory or other actions in accordance with section 2.5.3), the legal
1849 notification time 34 is no later than 30 days after finalisation of the signal assessment. As the benefit-
1850 risk balance of a product needs to be monitored by the MAH continuously, the signal management
1851 submission can therefore occur at any time throughout the year.
1852 When submitting a signal, it is recommended that the data lock point (DLP) in the assessment report is
1853 not older than 3 months.
1854 For assessed signals for which the MAH proposes a refutation or close monitoring, the corresponding
1855 signals can be submitted by the time of the submission of the annual statement (see also section
1856 3.1.1).
1857 Pharmacovigilance alerts (see also sections 2.2.1.1 and 3.1.3) should be notified as soon as possible
1858 and no later than 3 working days following their identification by the MAH.
1859 3.1.2.3. EU network/NCAs’ responsibilities
1860 The competent authorities and the Agency evaluate the results and outcomes of the signal
1861 management process, consider options for risk management and recommend appropriate measures
1862 referred to in Articles 129, 130 and 134 concerning marketing authorisations 35.
32 https://www.ema.europa.eu/en/documents/template-form/veterinary-signal-assessment-report-marketing-
authorisation-holders_en.docx
33 https://support.ema.europa.eu/esc
34 Article 81(2) of Regulation (EU) 2019/6
35 Article 79 of Regulation (EU) 2019/6
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 75/79
第 76 页
1863 All submitted signals by the MAH will be evaluated by an assigned responsible competent authority and
1864 the proposed regulatory outcome agreed with other competent authorities where the affected VMPs are
1865 authorised or adopted by the CVMP as applicable.
1866 In the situation where the regulatory outcome is in agreement with the proposed outcome by the MAH,
1867 the outcome will be published on the IRIS platform. In case of differences to the MAH’s outcome, these
1868 will be outlined in an assessment report and communicated to the MAH before publication of the
1869 regulatory outcome on the IRIS platform.
1870 3.1.3. Pharmacovigilance Alerts
1871 Pharmacovigilance (PhV) alerts (see section 2.2.1.1.) should be reported to the relevant competent
1872 authority(ies) of the Member State(s) where the VMP is authorised and the Agency without delay and
1873 no later than 3 working days after its identification.
1874 The notification of a PhV alert should be done through the IRIS-VSM platform30 specifying the type of
1875 signal as ‘Pharmacovigilance alert’.
1876 When notifying a PhV alert, the MAH should describe the safety issue, the source(s) of information,
1877 actions taken or any planned actions with timelines and should provide any relevant documentation
1878 available at the time of initial notification. Any further information relevant to the issue should be
1879 provided to the Agency and relevant national competent authorities as soon as it becomes available.
1880 Upon being notified of a PhV alert, the national competent authorities or the Agency as appropriate will
1881 promptly assess the urgency and potential impact of the issue and agree on appropriate next steps and
1882 the potential regulatory procedure to address the matter raised. This may involve consultation with the
1883 incident review network, if warranted (see incident management plan for medicines for veterinary
1884 use 36). The MAH should collaborate with the Agency and national competent authorities in the
1885 assessment of pharmacovigilance alerts.
1886 3.2. Other reporting obligations
1887 3.2.1. Reporting of sales data and calculation of dose factor
1888 The Union Product Database shall contain the annual volume of sales and information on the
1889 availability for each VMP 37. It is the responsibility of the MAH to record in the UPD the annual volume
1890 of sales for each of its VMPs 38.
1891 The sales data should be recorded in the UPD by the end of February of each calendar year 39, 40.
1892 In addition to sales data, the MAH is also obliged to calculate and submit a dose factor; please see
1893 guideline on the calculation of dose factor 41. Documentation should be maintained within the MAH’s
1894 pharmacovigilance system.
36
https://www.ema.europa.eu/en/veterinary-regulatory-overview/post-authorisation-veterinary-
medicines/pharmacovigilance-veterinary-medicines/incident-management-plan-veterinary-medicines
37 Article 55(2d) of Regulation (EU) 2019/6
38 Article 58(12) of Regulation (EU) 2019/6
39 UPD - Questions and Answers
40 https://www.ema.europa.eu/en/documents/other/eu-implementation-guide-ig-veterinary-medicines-product-data-union-
product-database-chapter-7-submission-other-post-authorisation-data_en.pdf
41 https://www.ema.europa.eu/en/documents/scientific-guideline/guideline-calculation-dose-factor-be-submitted-union-
product-database-upd_en.pdf
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 76/79
第 77 页
1895 As described in Article 75(3) from Regulation (EU) 2019/6, the annual incidence data from suspected
1896 adverse events for each VMP by animal species and type of suspected adverse event should be made
1897 available for access to the general public in the UPhV-database (ADR website).
1898 3.2.2. Notification on regulatory measures from non-EU regulatory
1899 agencies/Information from other regulatory authorities worldwide
1900 Article 78(1)(k) of Regulation (EU) 2019/6 tasks the QPPV to communicate any regulatory measure
1901 taken by a competent authority in a third country and that is related to pharmacovigilance data to the
1902 competent authorities and to the Agency within 21 days of receipt of such information.
1903 These notifications should be submitted via the IRIS-VSM platform30 as signal type ‘Submission of
1904 regulatory measure acc. to Art.78 1(k) of the Regulation 2019/6’.
1905
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 77/79
第 78 页
1906 4. Targeted signal management by the competent authorities
1907 and the Agency
1908 As described in Article 81(3) from Regulation (EU) 2019/6, the competent authorities and the Agency
1909 have the option to perform a targeted signal management process for a given VMP or group of VMPs.
1910 Regulators can start a Targeted Signal Management (TSM) procedure at any time throughout the year.
1911 The start of a TSM will be published on the EMA website 42 and concerned MAHs will be informed
1912 accordingly. MAHs should collaborate with the Agency and national competent authorities in any TSM
1913 initiated by the competent authorities and provide any requested information in a timely manner.
1914 More detailed guidance on the targeted signal management procedure will be published separately.
1915 5. Documentation requirements related to the Signal
1916 Management Process
1917 Signal management is considered a critical process and shall therefore be well documented. MAHs shall
1918 ensure that their processes, procedures and policies to fulfil their obligations related to the signal
1919 management process and its related tasks are functioning properly and efficiently and are
1920 appropriately documented as outlined in EU Regulation 2019/6 and CIR (EU) 2021/1281. A quality
1921 management system shall be applied throughout. Please also refer to VGVP module:
1922 Pharmacovigilance systems, their quality management systems and pharmacovigilance system master
1923 files 43.
1924 It is important that the roles of all staff members involved are clearly defined and the individuals
1925 tasked with these responsibilities have the necessary qualifications and expertise to do so.
1926 Documentation regarding qualifications and training of relevant staff members should be kept up to
1927 date and available. It is expected that any IT-System used for and in conjunction with the signal
1928 management process and adverse event analysis is validated, approved and deemed as 'fit for
1929 purpose' by the QPPV and IT personnel. The relevant documentation for this is to be kept available for
1930 pharmacovigilance inspections.
1931 When the MAH opts to use its own database for signal detection and analysis, detailed description of
1932 the data collection process, the data-tables and statistical measures available and routinely used shall
1933 be made available on request or at the time of pharmacovigilance inspections.
1934 Upon inspection it is required that the entire process is not only described theoretically but also backed
1935 up with the appropriate documentation of its application. This includes the rationale for the method
1936 and periodicity of signal management activities and the individual steps applied throughout. Through a
1937 tracking system, all organisations should keep an audit trail of all signal management activities,
1938 allowing traceability (i.e. recording of dates and confirmation of timeliness) and process control of the
1939 details of all steps of signal management, including analyses, decisions and rationale.
1940 In summary, as a minimum, MAHs are expected to have procedures describing in details the overall
1941 signal management process and, for each step, how signal management is effectively carried out, i.e.
1942 for signal detection, signal prioritisation, signal validation, signal assessment, handling of potential
1943 pharmacovigilance alerts, submission of outcomes (signals and outcome statement to the EU network
42 https://www.ema.europa.eu/en/veterinary-regulatory-overview/post-authorisation-veterinary-
medicines/pharmacovigilance-veterinary-medicines/signal-management-veterinary-medicines#targeted-signal-
management-by-regulators-14572
43 https://www.ema.europa.eu/en/documents/regulatory-procedural-guideline/guideline-veterinary-good-
pharmacovigilance-practices-vgvp-module-pharmacovigilance-systems-their-quality-management-systems-and-
pharmacovigilance-system-master-files_en.pdf
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 78/79
第 79 页
1944 and relevant NCA). The level of details expected in procedures should be sufficient to allow,
1945 theoretically, any new employee to perform the corresponding tasks, should they have the relevant
1946 expertise.
1947 • Overall signal management process.
1948 • Procedure for signal detection (both in the UPhV-Database and MAH databases, if applicable).
1949 • Overall policy and procedure for the remaining individual steps of the signal management
1950 process, namely:
1951 o Signal prioritisation,
1952 o Signal validation,
1953 o Signal assessment,
1954 o Handling of potential Pharmacovigilance Alerts,
1955 o Submission of the outcomes of the signal management process (signals and annual
1956 statements) to the EU Network and relevant NCAs.
1957 Documentation of the application of above-mentioned procedures including their outcomes should be
1958 kept available. This includes, but is not limited to;
1959 • Established risk-based monitoring criteria for each product or group of products (how often and
1960 by which methods is monitoring performed for each product/product group and why?).
1961 • Compliance with established monitoring criteria.
1962 • Signal validations performed, their outcome and rationale.
1963 • Signal assessments performed.
1964 • Reporting of the outcome of the signal management process (signals and annual statements)
1965 to the EU Network and relevant NCAs.
1966 Documents and pharmacovigilance data relating to individual authorised veterinary medicinal products
1967 shall be retained as long as the product is authorised and for 5 years after the marketing authorisation
1968 ceases to be valid. 44
44 CIR (EU) 2021/1281 Article 5(3)
Guideline on veterinary good pharmacovigilance practices (VGVP) Module: Signal
Management
EMA/522332/2021 Rev.1 Page 79/79
来源:EMA 官方医药监管信息 · ema.europa.eu