Signal Detection in Pharmacovigilance: Methods and Algorithms

Welcome
Signal Detection in Pharmacovigilance:
Methods & Algorithms
Name: David Lalrinmawia
Qualification: Pharm D
Student ID: 127/072023
10/18/2022
www.clinosol.com | follow us on social media
@clinosolresearch
1

Index
 Introduction
 Definition of Signal
 Types of Signals
 Definition of Signal Detection
 Goal of Signal Detection
 Signal Management Process Flowchart
 Sources of Data
 Traditional Pharmacovigilance Methods
 Data Mining Algorithms
 Conclusion
 References
10/18/2022
@clinosolresearch
2

Introduction
Definition of Signal
 According to WHO, signal is defined as “Reported information on a possible casual relationship between an
adverse event and a drug, the relationship between unknown or incompletely documented previously”.
 In CIOMS Reporting, Hauben and Aronson defined signal as “Information that arises from one or multiple
sources (including observations and experiments), which suggests a new potentially casual association, or a
new aspect of a known association, between an intervention and an event or set of related events, either
adverse or beneficial, that is judged to be of sufficient likelihood to justify verificatory action”.
Types of Signals
 Adverse Events – Reactions that pose a negative impact on the patient’s health, well being, quality of life or
the condition itself.
 Beneficial Events – Reactions that indicate a positive impact on the patient’s health, well being, quality of life
and condition.
10/18/2022
@clinosolresearch
3

Introduction
Definition of Signal Detection
 Signal detection in pharmacovigilance is the process of actively searching for and identifying safety signals
from a wide variety of data sources.
 It involves looking at the adverse reaction data for patterns that suggest new safety information and
specifically whether the new information changes the benefit to risk ratio associated with the use of a
pharmaceutical product.
 It is the most important objectives of pharmacovigilance.
 It is the core Uppsala Monitoring Centre (UMC) activity.
Goal of Signal Detection
 The goal of Signal Detection in PV is to detect unknown casual relationships or associations between
medicines and unexpected events.
10/18/2022
@clinosolresearch
4

10/18/2022
@clinosolresearch
5

Sources of data
Sources of clinical safety data described in the ICH E2D Guideline
10/18/2022
@clinosolresearch
6
Different sources Description of the source
Unsolicited sources Spontaneous reports, literature, internet.
Solicited sources Organized data collection systems like clinical trials,
registries, surveys of patients or health care
professionals.
Contractual agreements Inter company exchange of safety information.
Regulatory authority sources Individual case safety reports like Suspected
Unexpected Serious Adverse Reactions (SUSAR)
that originate from regulatory authorities.

Methods and Algorithms
1) Traditional pharmacovigilance methods
This method is important in the assessment of designated medical events (DME).
It mainly include:
Review of individual cases or case series in a pharmacovigilance database or in published medical or
scientific literature.
Aggregate analyses of case reports using absolute case counts, simple reporting rates or exposure
adjusted reporting rates.
It also include both qualitative (manual medical review of individual case or case series) and simple
quantitative approaches (frequency/reporting rates, sorting, cross-tabulation).
10/18/2022
@clinosolresearch
7

Once a signal is detected as a result of individual or aggregate analysis of spontaneous adverse events
reports
Investigated
Signal triage, clarification and early evaluation, and if required, formal evaluation using independent
data sets such as hypothesis testing research studies.
Such investigation should be conducted in an integrated, holistic fashion within the context of biological
plausibility and other available scientific evidence.
10/18/2022
@clinosolresearch
8

2) Data Mining Algorithms
Disproportionality Methods includes:
Relative Reporting Ratio (RRR)
Reporting Odds Ratio (ROR)
Proportional Reporting Ratio (PRR)
Bayesian Methods includes:
Multi-item Gamma Poisson Shrinker (MGPS)
Information Component (IC)
Bayesian Confidence Propagation Neural Network (BCPNN)
10/18/2022
@clinosolresearch
9

• Relative Reporting Ratio (RRR)
It is the ratio of how many ADRs under exposure were actually observed over the number of expected
events under the assumption that ADR and drug exposure were independent.
RRR is estimated as-
where, p(drug i) is probability of a target exposure being reported
p(ADR j) is probability of a target event being reported
p(drug i,ADR j) is joint probability of report on target event under exposure to the target drug
A RRR value of 1- Normal background noise, e.g., associations by chance.
An increase of RRR – Over proportional association between the drug and adverse event.
RRR value <1 – Negative association, e.g., the usage of drug protects the patient from the adverse event.
10/18/2022
@clinosolresearch
10

• Reporting Odds Ratio (ROR)
It is the odds of a certain event occurring with your medicinal product, compared to the odds of the same
event occurring with all other medicinal products in the data base.
A signal is considered when the lower limit of the 95% confidence interval (CI) of the ROR is > 1. The 95%
CI gives an indication of the precision of the estimate of the ROR.
ROR formula-
10/18/2022
@clinosolresearch
11
Event of interest All other events Total
Product of interest a b a+b
All other products c d c+d
Total a+c b+d a+b+c+d

• Proportional Reporting Ratio (PRR)
It is used to get a measure of how common an adverse event for a particular drug is compared to how
common the event is in the overall database.
It is also used to measure the strength of the statistical association between a risk factor(specific drug) and a
condition(specific adverse event).
PRR formula-
PRR= [a/(a+b)] / [c/(c+d)]
10/18/2022
@clinosolresearch
12
Event Y All other events Total
Product X a b a+b
All other products c d c+d
Total a+c b+d a+b+c+d

• Multi-Gamma Poisson Shrinker (MGPS)
MGPS is based on the metaphor of the “market basket problem”, in which a database of “transactions”
(adverse event reports) is mined for the occurrence of interesting (unexpectedly frequent) item sets.
For example, these item sets can represent simple drug event pairings or more complex combinations of
drugs and events representing interactions and/or syndromes.
MGPS produces Empirical Bayesian Geometric Mean (EBGM) scores.
EBGM include Bayesian “shrinkage” and stratification to produce disproportionality scores toward the null,
especially when there are limited data and small number of cases.
• Bayesian Confidence Propagation Neural Network (BCPNN)
It is used in signal detection to search single drug – single ADR combinations.
It can also be used for detecting relation between group of similar drugs and particular ADR.
It is a computational framework based on a statistical neural network where learning and inference is done
using principle of Bayes law.
It will help in focusing clinical review on the potentially most important combinations of drugs and adverse
reactions.
10/18/2022
@clinosolresearch 13

• Information Component (IC)
It is a measure of strength of the quantitative dependency between specific drug and specific ADR.
It is a logarithmic measure of disproportionality used to evaluate strength of association between drug and
ADR.
IC formula-
where, p(x) – probability of specific drug ‘x’ being listed in the case report
p(y) - probability of specific ADR ‘y’ being listed in the case report
p(x,y) – probability of specific drug – ADR combination being listed in the case report
10/18/2022
@clinosolresearch
14

Conclusion
 Detection of signal will help in identifying early warning signs of Adverse Events.
 It will help in improving the patient outcome along with their quality of life.
 It will also help to identify and describe suspected harm to patients, caused by
their use of medicine.
 Lastly it will also help in enhancing drug development.
10/18/2022

References
• https://who-umc.org/media/164092/general-information-about-signal-published-in-who-pn.pdf
• https://cioms.ch/wp-content/uploads/2018/03/WG8-Signal-Detection.pdf
• https://trinetx.com/life-sciences/pharmacovigilance/pharmacovigilance-signal-
detection/#:~:text=Signal%20detection%20in%20pharmacovigilance%20is,wide%20variety%20of%20data%20sources
• https://www.simbecorion.com/types-of-signal-in-
pharmacovigilance/#:~:text=The%20most%20common%20form%20of,adverse%20or%20beneficial%20drug%20reactio
ns
• https://www.clinskill.com/signal-detection-in-
pharmacovigilance/#:~:text=Signal%20detection%20is%20one%20of,the%20quality%20of%20the%20information
• https://allaboutpharmacovigilance.org/pharmacovigilance-guidance-material/icsr-processing-aggregate-reporting-and-
signal-management/
10/18/2022

Thank You!
www.clinosol.com
(India | Canada)
9121151622/623/624
info@clinosol.com
10/18/2022
@clinosolresearch
17

Signal Detection in Pharmacovigilance: Methods and Algorithms

More Related Content

What's hot (20)

Similar to Signal Detection in Pharmacovigilance: Methods and Algorithms (20)

More from ClinosolIndia (20)

Recently uploaded (20)

Signal Detection in Pharmacovigilance: Methods and Algorithms