Добавил:
Sekretar
kiopkiopkiop18@yandex.ru
t.me/Prokururor I Вовсе не секретарь, но почту проверяю
Опубликованный материал нарушает ваши авторские права? Сообщите нам.
Вуз:
Предмет:
Файл:Ординатура / Хирургия / Библиотека им академика М.И. Перельмана / Книга_5631_Библиотеки_им_академика_М_И_Перельмана
.pdf
8 Causality Assessment inPharmacovigilance
193
therapy including dosage adjustments, and helps
to identify interacting drugs [8].
3 Stages ofData Assessment
In assessing the causality, there are two important
issues that need to be addressed. First, is there a
convincing relationship between the drug and the
event? And second, did the drug cause the event?
During clinical research, as well as from the perspective of industry or regulatory agencies during
regular use of drugs, the assessment of reported
adverse event can happen in two stages. The rst
stage is the assessment oftheindividual case safety
report (ICSR) that is usually carried out soon after
the event is reported and is assessed on case-tocase basis, whereas the second stage is the interpretation of aggregated data and is adopted during the
benet–harm analysis and signal detection process
[7]. This kind of assessment is an integral part of
the work of regulatory authorities and marketing
authorisation holders and helps in identifying new
ADRs, therefore in getting a better knowledge of
medicines’ safety prole. For a healthcare professional facing an individual patient with an adverse
event, it can be difcult to make a proper assessment of causality. However, it is important to try
to conrm the suspicion of causality, as it could be
crucial for patient management to withdraw a suspected medicine/adjust the dose and start an appropriate treatmentin a timely manner.
4 Basic Criteria/Principles
ofCausality Assessment
In clinical practice, the methods adopted to assess
the causal relationship generally consider few
criteria in common [9–11]. These include the
following.
4.1 Complete Knowledge
ofClinical Event
The knowledge of reported adverse event should
encompass complete description of event rather
limiting to the specics of nature of event.
4.2 Patient’s Medical
andMedication History
The patient medical history should include chief
complaints, history of present illness, review of
systems, relevant past medication and vaccination history, as well as family, social, and allergy
history. Medication history should encompass
the complete dosing regimen details of all current
and recently prescribed medications, any overthe-counter (OTC) medications including herbal
or alternative medicines, and details on adherence to therapy.
4.3 Temporal Relationship (Time
toOnset)
Thetemporal relationship is thetime gap between
the administration ofthe suspect drug (start date
and duration of therapy) and the occurrence of
the adverse event (date of onset of event/initial
symptoms observed).
4.4 Biological Plausibility
andPossible Alternative
Causes
Biological plausibility is considered as an important criterion, and it should be explicable biologically based on the available facts. The
possible contributing factors that could as
wellbe attributed to an adverse event include (i)
medical condition—underlying concurrent medical conditions of patient (e.g. diabetes, asthma,
etc.) or past medical history and prior/ongoing
surgical procedures; (ii) other concurrent/concomitant medication use and past medication
history; (iii) social history such as alcohol use,
smoking, diet, etc.; and (iv) predisposing factors
such as age and intercurrent diseases (hepatic
and renal impairment).
4.5 Dechallenge
Dechallengerefers to the withdrawal/cessation or
reduction in dose of the suspected drug. If there is
194
M. Ramesh and A. Harugeri
a complete or partial resolution of adverse event
following withdrawal/cessation or reduction in
the dose (in the case of type A reaction) of the
suspected drug, it is called as positive dechallenge. On the other hand, if there is no change in
the outcome of an event and continued to exist in
the same intensity or worsened further despite the
cessation of the suspected drug, it is called as
negative dechallenge. Outcome of dechallenge
could be inuenced by corrective treatment for
the adverse event in addition to the dechallenge.
Hence, this potential inuence needs to be considered while deciding positive dechallenge.
4.6 Rechallenge
Rechallenge is the reintroduction of the suspected drug in a similar manner to the previous
exposure and is applicable in the positive dechallenge scenario. The re-occurrence of similar
adverse event following the reintroduction of
suspected drug refers to positive rechallenge
while failure to recur even after the reintroduction of suspected drug is referred to as negative
rechallenge. Well-proved positive rechallenge is
one of the strongest associations for causal relationship between the event and the medicine.In
real practice, only veryless often a rechallenge
isdone specically to establish causality,due to
ethical issues and concerns of causing harm in
the patient,especially in the case of type B ADRs.
Rechallenge is done in specic circumstances
(like when thereare no suitable safe and effective
treatment alternatives available) depending on
various factors including nature and severity of
initial presentation of the adverse event, under
strict medical supervision and appropriate
monitoring.
5 Approaches or Categories
ofCausality Assessment
There are three broad approaches or categories of
causality assessment adopted for establishing
causation in pharmacovigilance. They include:
• Expert/clinical judgement or global introspec-
tion method
• Algorithms or standardised assessment
method
• Probabilistic or Bayesian method
5.1 Expert/Clinical Judgement or
Global Introspection Method
In this approach, the causation of the adverse
drug reaction is established based on the clinical
judgement of the expert or panel of experts using
no standardised tool to arrive at conclusions in
assigning the causality category. Instead, an individual assesses the causation based on previous
knowledge and clinical experience in the eld of
pharmacovigilance. The expert(s)/clinician(s)
analyses all the available data on the reported
adverse event, following which an opinion is formulated on causality category based on the likelihood involvement of suspected drug in causing
the adverse event [6, 12].
The expert(s)/clinician(s) would carefully
evaluate each case to determine if the adverse
event was likely caused by the drug or occurred
due to other reasons. However, there are greater
chances of inter-individual variability as there
may be information bias between different
experts. Causality assessment using this method
is highly subjective and is based on the individual
assessor’s judgement and inuenced by several
other factors such as depth of clinical knowledge,
evidence, extent of available information, and
expertise and/or experience of the assessor.
Therefore, one assessor’s ‘probable’ may be
another assessor’s ‘possible’ for the given case
causality assessment [6]. Other important considerations such as pharmacologic plausibility,
dose–response, and timing may or may not be
included. These assessments done by experts are
generally expressed in terms of a qualitative
probability scale such as ‘denite’, ‘probable’,
‘possible’, ‘doubtful’, or ‘unrelated’ [9, 12–
15]. World Health Organization—Uppsala
Monitoring Centre (WHO-UMC) causality
assessment method is a widely used global intro-
8 Causality Assessment inPharmacovigilance
195
spection method toassess the causality of individual case safety reports required for the
submission to pharmacovigilance systems worldwide [16]. There is some degree of ambiguity as
to whether WHO-UMC causality assessment
method should be considered as global introspection or as an algorithm. More ideally, it could be
considered as a global introspection methodas in
theassessment using this scale,the assessor considers factors that might contribute to a causal
link between one or more drugs and an observed
ADR, weighs their importance, and decides the
probability of drug causation, but no specicchecklist or algorithm is given. It could be
considered as a global introspection method,
used for causality assessment, based on expert
judgement, though delimited by specic criteria.
5.1.1 Advantages
The major advantage of global introspection
method is its pivotal role in the identication and
rating of potential ADRsover the years. There is
also widespread support for continuing to
valueexpert clinical judgement [17–20].
5.1.2 Limitations
The main drawback of global introspectionis that
clinical judgement is characterised by inter- and
intra-rater dissimilarity, ambiguity, lack of reproducibility, and discernible prejudice [21–23].
5.2 Algorithms
Algorithms are problem-specic ow chart or
sets of specic questions with associated scores
that provide step-by-step guidance on how to
arrive at conclusions on causality of the reported
adverse event. These algorithms give a structured
and standardised method of assessment in a systematic approach in identifying the adverse drug
reaction or temporal sequence, previous drug/
adverse reaction history, dechallenge, and rechallenge [24–26].
Most algorithms are in the form of a questionnaire that allows the user to gather adequate information and also assess the causal relationship
between the drug and the suspected reaction. In
algorithms-based approach, as causation is established using a well-structured, standardised, and
validated scale, the chance of difference of opinion between two assessors (inter-individual variability) is greatly minimised. Naranjo’s [27]
scale is widely accepted and used worldwide in
case causality assessment by personnel in clinical
practice, pharmacovigilance, and researchers. A
number of otheralgorithms have been published
including the Karch and Lasagna algorithm [21],
theJones algorithm [28], the Yale algorithm [29],
and the Begaud algorithm [30]. None of the existing algorithms are unanimously acknowledged as
a recognised tool or a gold standard method as the
evaluation of the same ADR reports by using different algorithms demonstrated signicant variations in the results [11, 27, 31].
5.2.1 Advantages
Algorithmic methods are widely used for their
simplistic and comprehensive approach, which
creates a logical approach to causality assessment and scientic validity. These methods have
poor sensitivity, but good specicity, and provide
better reproducibility between assessors as they
eliminate or at least reduce inter- and intra-rater
disagreement [17, 18, 31, 32].
5.2.2 Limitations
Algorithmic methods do not provide certainty of
case diagnosis. Also, they fail to ascertain the
causality consistently due to a lack of regard to
the confounding variables. Moreover, their
inability to verify and invalidate the causality and
to provide precise quantitative measurement of
the probability of a causal relationship are the
major drawbacks [18, 32].
5.3 Probabilistic or Bayesian
Approaches
Probabilistic, or Bayesian, approaches involve
the use of a statistical approach to assess the
causal relationship between the drug and the outcome. It involves the use of specic ndings in a
196
M. Ramesh and A. Harugeri
case to transform the prior estimate of probability
into a posterior estimate of probability of drug
causation. The prior probability is calculated
from epidemiological information, and the posterior probability combines this background information with the evidence in the individual case to
provide an estimate of causation [6, 9, 19, 32].
Examples are the Australian method [33],
the Bayesian Adverse Reactions Diagnostic
Instrument (BARDI), and MacBARDI [24, 34].
5.3.1 Advantage
The probabilistic/Bayesian methods are trustworthy in assessing the ADRsin routine practice or
automated evaluation of suspected case reports as
they are highly sensitive, and have positive predictive value, and provide an outcome as probabilities [35].
5.3.2 Limitations
The major drawbacks of probabilistic/Bayesian
methods are poor specicity and the fact that they
are complex in practical approach as they require
specically calculated information data, such as
specic ADR incidence, to reproduce the likelihood distribution [35].
6 Methods ofCausality
Assessment
Different methods ranging from simple questionnaires to comprehensive algorithms are available
to determine the causal relationship between a
suspected drug and adverse event. However, it is
essential/desirable that the causality of a reported
event is objectively assessed using a wellstructured, standardised, and validated scale so as
to ensure that the assessment is reliable and
reproducible. Despite many systems being developed for a structured and harmonised assessment
of causality, none of these systems have been
shown to produce a reliable and quantitative estimation of likelihood relationship. Nevertheless,
in clinical practice, the WHO probability scale
and Naranjo’s algorithm are widely used as they
offer convenience of ease of use and are less
time- consuming owing to their simple methodology. The following are the common methods
adopted for establishing causation in pharmacovigilance [36].
6.1 WHO-UMC Causality
Assessment Scale
The WHO-UMC causality assessment scale is a
practical tool for the assessment of case
reports. It basically considers the clinicalpharmacological aspects of the case history and
the quality of the documentation of the observation. This scale establishes the causal relationship between suspected drug and observed event
under any of the six categories and guides the
assessor in the selection of one category over
another [16]. The various causality categories of
WHO-UMC causality assessment scale are
listed in Table8.1.
Table 8.1 WHO-UMC causality categories
Causality category Criteria
Certain • Event or laboratory test abnormality, with plausible time relationship to drug intake
• Cannot be explained by disease or other drugs
• Response to withdrawal plausible (pharmacologically, pathologically)
• Event denitive pharmacologically or phenomenologically (i.e. an objective and
specic medical disorder or a recognised pharmacological phenomenon)
• Rechallenge satisfactory, if necessary
Probable/likely • Event or laboratory test abnormality, with reasonable time relationship to drug intake
• Unlikely to be attributed to disease or other drugs
• Response to withdrawal clinically reasonable
• Rechallenge not required
8 Causality Assessment inPharmacovigilance
Table 8.1 (continued)
Causality category Criteria
Possible
Unlikely • Event or laboratory test abnormality, with a time to drug intake that makes a
Conditional/
unclassied
Unassessable/
unclassiable
• Event or laboratory test abnormality, with reasonable time relationship to drug intake
• Could also be explained by disease or other drugs
• Information on drug withdrawal may be lacking or unclear
relationship improbable (but not impossible)
• Disease or other drugs provide plausible explanations
• Event or laboratory test abnormality
• More data for proper assessment needed
• Additional data under examination
• Report suggesting an adverse reaction
• Cannot be judged because information is insufcient or contradictory
• Data cannot be supplemented or veried
197
6.2 Naranjo’s Algorithm [27]
Naranjo’s ADR probability scale is a simple
questionnaire used to assess the causality of
ADRs in a variety of clinical situations. It consists of ten questions, and the answer to each of
the questions could be positive (yes), negative
(no), or unknown or inapplicable (do not know).
Each of the questions is assigned with score ranging from −1 to +2, and based on the total score,
the causality of reported adverse event is categorised into any of the four categories as ‘Denite’
(≥9), ‘Probable’ (5–8), ‘Possible’ (1–4), and
‘Doubtful’ (≥0) (Table8.2). Simplicity and wider
applicability are the advantages of this ADR
probability scale and is reliable and valid in
assessing the adverse drug-related events, while
some minor modications may be required in
special circumstances.
Table 8.2 Naranjo’s algorithm
Sl. No. Criteria Yes No
1 Are there previous conclusive
reports on this reaction?
2 Did the adverse event appear
after the suspected drug was
administered?
3 Did the adverse reaction
improve when the drug was
discontinued or a specic
antagonist was administered?
4 Did the adverse reaction
reappear when the drug was
readministered?
5 Are there alternative causes
(other than the drug) that
could on their own have
caused the reaction?
6 Did the reaction reappear
when a placebo was given?
7 Was the drug detected in the
blood (or other uids) in
concentrations known to be
toxic?
8 Was the reaction more severe
when the dose was increased,
or less severe when the dose
was decreased?
9 Did the patient have a similar
reaction to the same or
similar drugs in any previous
exposure?
10 Was the adverse event
conrmed by any objective
evidence?
+1 0 0
+2
+1 0 0
+2
−1
−1
+1 0 0
+1 0 0
+1 0 0
+1 0 0
−1
−1
+2 0
+1 0
Do not
know
0
0
198
M. Ramesh and A. Harugeri
7 Various Perspectives
7.1 Clinical Practice Perspective
Understandingand assessing the causal relationship between the observed adverse event and the
suspected drug/s will greatly inuence the determination of the possibility of the drug causing the
observed event and futurecourse of action including prescribing decisions. Clinicians should
always considerADRs as adifferential diagnosis. When all the other alternative explanations
for observed AE fail, the cause suspicion will fall
onthe drug. However, in many cases a decisive
causal relationship may not always be practically
possible, and a level of uncertainty will exist.
Identifying the culprit drug(s) can be lifesaving or helpful in preventing the further drugrelated harm. Assessing causality is of
importance as it directly inuences decisionmaking by a clinician on future course of action
which could be either to continue or to withdrawthe likely drugs, need for any dose adjustment of the likely drugs, and need for any
specic/non-specic treatment for the ADR.A
causality assessment shouldbe performed based
on patient’s detailed clinical history, examination to look for temporal correlation between
suspected drug and drug reaction, biological
plausibility, and outcomes of dechallenge and
rechallenge, if carried out. At times, these assessments are highly subjective in nature and are
mainly based on clinical knowledge and experience of an individual clinician. Identifying the
culprit drug could be challenging at many times
when multiple drugs are suspected to be responsible for causing a drug reaction. The situation
becomes further challenging when the underlying disease is capable of producing the same
manifestation as caused by the drug. To a good
extent, the principles and methods of causality
assessment or causality assessment tools help
clinicians to identify the culprit drugs. Though
none of the multiple criteria or algorithms available is specic or complete, the causality assessment system proposed by the WHO-UMC and
the Naranjo probability scale are the generally
accepted and most widely used methods in clini-
cal practice as they offer a simple methodology
and they are time tested.
Formalised causality assessment offers some
objectivity and reduced inter-individual variations. Causality assessment tools can be used in
clinical practice, but an awareness of their potential limitations is required. Afew of the difculties in causality assessment noticed in clinical
practice are incomplete information about
thesuspectedADR, polypharmacy, variable clinical responses, poor understanding of biological
plausibility, other alternative causes, and lack of
healthcare professional training. Establishing
causality in patients with polypharmacy is difcult as dechallenge–rechallenge analysis may not
bepossible or ethicalfor every individual drug. It
should be borne in mind that clinical judgement
obviously is helpful in identifying the relationship likelihood in actual clinical practice and
there is no substitute for clinical observations and
experience of the healthcare professionals when
it comes to identifying the cause of an adverse
event.
In addition to its benets in determining the
relationship likelihood between the event and the
drug/s and subsequent course of action, causality
assessment is benecial for other reasons as well.
If case reports/case series based on drug reactions are being submitted for publication, completion of causality assessmentis an expectation.
When completing in-house documentation of
ADR reports in a healthcare institution, causality
assessment of the reported ADRs is useful to
quantify the type of ADRs based on causality as
well as for routine dissemination of the evaluated
ADR data. From the patient perspective, data on
the causality assessment will be helpful
whencommunicating the details of the observed
ADR to the individual patient. Additionally, causality assessment will be helpful in deciding the
need of providing an alert card for the patient
regarding the specic ADR, as well as documenting the ADR details in the patient medical records
to caution future prescribing. The relevance of
causality assessment in clinical practice signies
the importance of training healthcare professionals on the concept, basis, and process of causality
assessment [11, 16, 37].
8 Causality Assessment inPharmacovigilance
199
7.2 Industry Perspective
Reporting of safety information by the pharmaceutical companies for their proprietary molecules
is mandated by most of the regulatory authorities
worldwide. The major focus of regulators is to
scrutinise the products in clinical development
and post-marketed products to ensure medications with a favourable benet–harm ratio are
available in the market. This requires clinical trial
sponsors and marketing authorisation holders to
continuously evaluate their products for potential
adverse events, ensuring the benets outweigh
its harms. Therefore, pharmaceutical companies
musthave a robust pharmacovigilance system in
place to closely monitor the adverse events associated with their products both as a regulatory
requirement and product liability. Many multinational pharmaceutical companies have adopted
a structured way of capturing ADRs and their
subsequentcausality assessment, in comparison
to the relatively less developed systems in academicresearch and general healthcareprovision.
However, establishment of conrmed causal relationship between the drug and the adverse event
is a challenge and carries utmost importance in
the continuing scenario of emerging drug safety
concerns. Initially, investigators/sponsors were
conservative in their interpretation of ‘reasonable
possibility’ of association between the drug and
the adverse event. The spontaneous reports are
considered by pharma companies as reports with
implied causality by reporter.
Solicited reports are classied as study reports
and therefore should have an appropriate causality assessment by a healthcare professional or
pharmaceutical company (International Council
for Harmonisation [ICH] E2D, 2003). A method
of causality assessment is recommended to be
applied for assessing the causal role of medicinal
products in the solicited adverse events.
Spontaneous reports are considered suspected
adverse reactions, since they convey the suspicions of the primary sources, unless the reporters
specically state that they believe the events to be
unrelated or causal relationship can be excluded
[38]. Comments from the primary source regarding causality assessment are reported by pharma-
ceutical companies to regulatory authorities as
‘Reporter’s comments’ (ICH- E2B (R2) B.5.2).
Change in the causality assessment in the followup reports of ADRs is considered as signicant
new information. In pharmaceutical industry,
structured causality assessment methods for individual cases have been used primarilyto facilitate case processing rather than to reach denitive
conclusions. Events are considered ‘related’ or
‘not related’ for the purpose of case processing
and regulatory reporting. Pharma companies’
designation of relatedness for regulatory reporting purposes is independent of the investigator’s/
report’s assessment. Despite follow-up attempts,
usually complete information required for comprehensive causality assessment is barely
received by pharma companies particularly from
unsolicited sources.
To date, there are no internationally agreedupon standards or criteria by pharma companies for
assessing causality in ICSR. Nature of the event,
consistency with the known drug prole, physiological plausibility, temporal association, availability of sufcient and consistent information, specic
laboratory investigations, more probable/likely
alternative reasons, confounding medical conditions/medications, and dechallenge/rechallenge
results are the criteria typically considered by the
pharma companies for causality assessment.
Denitive decisions about causality (leading to risk
management plan/label changes, etc.) usually
incorporate evidence from various sources.
However, no single ‘one size ts all’ approach
exists as sometimes individual case reports provide
compelling evidence of causality [37–40].
7.3 Regulatory Perspective
Causality assessment is necessary for adverse
events reported in clinical trials. Most of the clinical trial protocols require investigator’s causality
assessment for both serious and non-serious
adverse events. The investigator’s assessment of
causality is not required for non-serious adverse
events by the regulations,except some circumstances like non-serious adverse events of special
interest. Serious adverse events (SAEs) that are
200
M. Ramesh and A. Harugeri
either expected or unexpected and related to the
suspected drug qualify for expedited reporting to
regulatory authorities. Causality assessment is
well described in the International Council for
Harmonisation (ICH) E2A Guideline and the
report of Council for International Organizations
of Medical Sciences (CIOMS) working group
VI.The adverse events do not imply any judgement about causality. Suspected adverse reaction
suggests a causal association. ADRs imply ‘reasonable possibility’ that there is evidence to suggest a causal relationship between the drug and
the adverse event as judged by either the reporter
or the sponsor. The United States Food and Drug
Administration (US FDA) considers causality
assessment as sponsor’s responsibility to decide
expedited reporting needs as they are better positioned for assessment than the individual clinical
trial investigators. However, for all serious
adverse events reported to the sponsor, as per 21
Code of Federal Regulations 312.64, clinical trial
investigators are required to provide a causality
assessment. In general, ‘reasonable possibility’
conveys that there are evidence or opinions to
suggest a causal relationship even if the relationship is unknown or unstated.
Causal association between the drug and the
adverse event in clinical trials is supported by
evidences like uncommon event and known to
occur with drug exposure, one or more uncommon events in the population exposed, and higher
frequency of occurrence in treatment group than
control group. The sponsor decides the causality
assessment scale to be used by investigators.
Many sponsors usually demand ‘Yes/No’ answer
to the question ‘Was there a reasonable possibility that the drug caused the adverse event?’ for
the causality assessment as CIOMS VI Working
Group recommends using a simple binary decision for causality assessment. Although many
sponsors use home-grown causality assessment
methods, the ‘Introspection’ is the sponsors’ causality assessment method of choice. ADRs associated with marketed drugs (unsolicited reports)
usually considered to imply causality for reporting purposes to regulatory authorities. For the
purposes of safety reporting to regulatory authorities, solicited reports associated with marketed
products are classied as study reports and therefore should have an appropriate causality assessment by a healthcare professional or a Marketing
Authorisation Holder (MAH). Currently, there
are no standard international causality assessment methods and nomenclature to describe the
attributability between drug exposure and adverse
event; hence, terms like certainly, denitely,
likely, probably, possibly, or likely related or not
related are in use [40–42].
8 Application ofArticial
Intelligence andMachine
Learning inCausality
Assessment
With the increasing number of reports, it has
become ever more necessary to implement
computer- based approaches to support the efforts
of human expert causality assessment of ADRs.
An automated solution for causality assessment
is yet to be found as the assessment workow is
not fully standardised to a level required for computation [43]. The US FDA requires pharmaceutical companies to have procedures for receipt,
assessment, and reporting of post-marketing
adverse drug experiences [44]. Efforts to standardise the workow for purposes of computation need expert judgement and exibility.
Articial intelligence (AI) approaches expected
to be applied for causality assessment need the
understanding of both the individual tasks and
how they are further assembled into a cognitive
framework for causality assessment. Causality
assessment by human expert processes involves
use of information that is both internal and external to the case report. Drug exposure, ADR outcome, temporality, and alternative explanations
are some of the important elements that need to
be considered by cognitive framework for causality assessment [45].
The MOdied NARanjo Causality Scale for
ICSRs (MONARCSi) is an early effort in developing machine-based learning tool for causality
assessment. The MONARCSi approach uses
matrix and weights determined by aggregating
the importance of presence or absence of a spe-
8 Causality Assessment inPharmacovigilance
201
cic drug-event feature. The nal scores
obtained for a specic drug–event pair are then
logistically transformed to estimate the probability or condence level in the ‘relatedness’ or
‘unrelatedness’ for drug–event causality. Based
on this logistic probability level, the model then
assigns causality as related or unrelated [46].
DeepCausality provided an AI-powered promising solution for causality assessment from
free text by integrating transformers, named
entity recognition, and Do-calculus into a unied framework to empirically estimate the
causal factors for suspected endpoints embedded in the free text [47].
Identication of low-value reports is important in applying AI to causality assessment of
ICSRs, since computerised algorithms can
achieve better causality assessment results (in
cases containing enough information) to make an
informed causality assessment. This can be
achieved by training the algorithm on reports
classied as either ‘assessable’ or ‘unassessable’
[43]. Machine learning capabilities have been
explored to predict ICSRs that are most useful for
causality assessment. Natural language processing extraction and visualisation of clinical data
have been used to support case series analyses for
causality assessment [48].
It is believed that pharmacovigilance was
lagging in adopting machine learning–causal
assessment integrated models [49]. AI may
become increasingly important for improving
the efciency and scientic value of ICSRs
including reducing the inconsistencies in assessment processes and improving overall decisionmaking [45]. Using AI-powered language
models to conduct causal inference, similar to
human experts, is still at the infant stage. AI for
causality assessment will have to be aligned
with the emerging best practices for the eld to
reach a state of maturity, as causality assessment still remains a challenge [50]. Regulatory
authorities have not yet endorsed any specic
approach forapplying AI to ADR case processing and evaluation. Research on enhancing data
quality and data representation will be an importantstep in successful implementation of AI in
causality assessment. The likelihood of AI sys-
tems that will reach a level of full automation
performance in the near term has been assessed
as low [45]. However, incorporating causal
inference paradigms to address currently prominent machine learning issues in causality assessment could be a new future. Learning from AI
applications in the aviation industry, it may be
that AI supporting, rather than supplanting,
human expertise in causality assessment be
benecial.
9 Applications ofCausality
Assessment
Continuous specication of the safety prole of
a drug product is an essential component
throughout its life cycle, from the drug discovery, topost- marketing and until the drug is out of
the market. Thus, the need forcausality assessment continues and has wider applications across
various elds including academicresearch, clinical practice, sponsored clinical research,
national pharmacovigilance programmes, and
regulatorydecision-making. It is an integral part
of clinical management as early detection of
adverse event and identication of culprit drug
canaidhealthcare professionals to make therapeutic decisions to improve patient safety.
Causality assessment is legally required by
health authorities including US FDA, European
Medicines Agency, and other health authorities
for expediting reporting of all serious adverse
events (SAEs). Furthermore, causality assessmentis needed to determine whether the adverse
event should be considered in the signal detection process, benet–harm determination, and
whether it should be listed in the reference safety
information (Investigator Brochure, Company
Core Safety Datasheet) and label (includingPackage Inserts and the Summary of Product
Characteristics). Also, as an essential component
of pharmacovigilance contributing to better
patient care, establishing causation of the
reported event is important prior to dissemination of safety information amongst healthcare
professionals, investigators, patients, and other
stakeholders [7, 42].
202
M. Ramesh and A. Harugeri
10 Challenges inCausality
Assessment
Causality in pharmacovigilance is a complex process and should be viewed considering both
patient-related and medication-related factors. It is
often highly subjective based upon clinician’s
judgement that is inuenced by several factors
including the assessor’s knowledge of the event,
experience, availability of time, and amount of
information. These factors lead to higher chances
of inter- and intra-individual variability in assigning causality. ADRsalso frequently mimic disease
symptomswith no distinct characteristic features
that can differentiate the iatrogenic cause from disease process. Eliminating all possible alternative
causes is often key to assigning causality. However,
in a routine clinical practice, this can be a challenge and maynot also be possible considering the
time and economic constraints within healthcare
systems. Also, many of the symptoms attributed to
ADRs occur in healthy individuals taking no medications. For example, headache can occur in any
individual due to various reasons, making it more
difcult to attribute to a medication. In addition,
incomplete information of ADR, polypharmacy,
variable clinical responses, poor understanding of
biological plausibility, and lack of training among
assessors pose additional challenges in causalityassessment [6, 7, 9].
What causality assessment aids in:[7]
• Decreasing the disagreement between
assessors
• Classifying uncertainty
• Marking individual case reports
• Improving scientic basis of evaluation
What causality assessment lacks:
• Giving accurate quantitative measurements of
the causation
• Distinguishing between valid and invalid
cases
• Conrming the relationship between drug and
event
• Quantication of the contribution of the drug
in the development of the adverse event
• Changing uncertainty to certainty
11 Summary andConclusions
Causality assessment is an essential and challenging aspect of pharmacovigilance. It is important in clinical practice to identify the culprit
drugs as the safety of the patients is paramount.
Usingcausality assessment tools may help clinicians in identifying the culprit drugs, understanding the probability that the suspected drug has
caused the event observed, and determining
future course of action includinginformed future
prescribing. There are multiple criteria or algorithms available for establishing a causal relationship in case of suspected ADR.However, a
standardised causality assessment method that
offers reliable and reproducible measures of the
relationship likelihood in suspected cases of
ADR seems to be unfeasible, since no single
method has achieved this to date, withno single
method universally accepted. Differences in
ADR causality assessment methods, and the
unavoidable subjectivity of judgements, will
likely impactreproducibility for most published
methods. Therefore, there is a need to develop a
high-quality causality assessment tool that can
establish asuitable diagnostic criterion for ADRs,
with universal acceptance.
12 Case Studies
12.1 Case Study 1: Causality
Assessment Using WHO-UMC
Causality Assessment Scale
Mr. K is a 38-year-old man, weighing 50kg, private employer, and married. He was diagnosed
with pulmonary tuberculosis(TB) 20days ago at
a tertiary care hospital. His lab investigations at
the time of diagnosis were as follows.
Sputum Acid Fast Bacillus (AFB): Positive (2
samples); Haemoglobin (Hb): 8.5 g%; WBC:
3800 cells/mm3; platelets: 2,10,000 cells/mm3;
liver tests: aspartate aminotransferase (AST):
45 IU/L (0–40 IU/L); alanine aminotransferase
(ALT): 40 IU/L (0–40 IU/L); alkaline phosphatase (ALP]: 160IU/L (150–280IU/L); albumin:
3.5mg/dL (3–5mg/dL); chest X ray: indicative
of tuberculosis, presence of granuloma. Mr. K is
Соседние файлы в папке Библиотека им академика М.И. Перельмана