PHARMACOEPIDEMIOLOGY

Dr Priyanka
PG Pharmacology 2nd year
PROTOCOL
Introduction
History
Need for Pharmacoepidemiology (PE)
Aim & application
Measurement of outcomes
Concept of risk
Methods in Pharmacoepidemiology
Systems for studying drug effects in population
Advantage & disadvantage
Future of Pharmacoepidemilogy
Conclusion
 Pharmacoepidemiology
Derived from
Pharmacon drug
Epi among
Demos people
Logos study

Definition

It is the study of the use of & the effects of medicines in large

number of people
 Application of the principles of Epidemiology to the drug use & drug
effects

The bridge science between

Clinical Pharmacology focus of enquiry

Epidemiology methods of enquiry

Applied to the post marketing phase

Focus on Indicator of Result studied
Relationship between drug 1. Exposure outcome relationship 1. Causality

exposure & health outcomes in 2. Comparitive effectiveness 2. Quantification of benefit

defined population 3. Comparitive toxicity 3. Quantification of risk

HISTORY

1961 beginning of era of Pharmacoepidemiology (PE)

1970 Development of PE programs

1991 International Society of PE (ISPE)

2007 Active surveillance system

ORIGIN & EVOLUTION
Case reports associating maternal thalidomide intake & phocomelia
began the era of PE

Its nature was dramatic enough to draw attention on the detection,

prevention & management of ADRs

Led to establishment of spontaneous reporting & other surveillance

systems
Classical examples

1. Chloramphenicol Grey baby syndrome

2. DES Vaginal ca in adolescent offspring of mother who were on

DES during pregnancy

3. Isotretinoin fetal isotretinoin syndrome

4. Triazolam CNS disturbances

5. Fenoterol Death

6. Fluoxetin Suicidal ideation

7. OCPs Venous thromboembolism

 Drug withdrawn from market due to ADR
DRUG ADVERSE REACTION
Thalidomide Phocomelia
Clozapine Aganulocytosis
Terodiline Torsades de pointes
Phenformin Lactic acidosis
Nomifensine Haemolysis
Temafloxacin Haemolysis
Zimelidine Guillain-Barre syndrome
Phenylbutazone Stevens-Johnson syndrome, Blood dyscrasias
Ciloquinol Subacute myelo optic neuropathy
Zomepirac Anaphylactic shock
Benoxaprofen Liver toxicity, Phototoxicity
Practolol Fibrosis, Ocular toxicity
Dipyrone Agranulocytosis
Few drugs with serious ADRs are reserved for special situation

- Thalidomide

- Clozapine

Drugs with manageable toxicity & unique benefits may not be withdrawn

- Isotretinoin

- Anti cancer drug

 ISPE
International society of Pharmacoepidemiology

Formed in 1991

A non profit international professional organization

Provide forum for open exchange of scientific information & for

development of policy

Discuss, develop & disseminate information about PE methods

 Need for Pharmacoepidemiology
PE studies are of special significance due to few limitations of pre marketing studies :

1) Narrowly defined population do not include special population like children, elderly,
pregnant female or female of child bearing age

Enrolment of pregnant or planning to conceive female in RCT to assess the teratogenic

potential of a drug unethical

At the same time this information is vital

PE provide safe & efficient alternative approach

2) Small sample size usual sample size of phase 3 trail is insufficient to
detect less common ADR with frequency of < 1 / 1,000

Serious but uncommon ADR missed during pre marketing phase

Suprofen Acute flank pain

Troglitazone Hepatotoxicity

Statin Rhabdomyolysis

Cisapride QT prolongation
3) Short duration : Can miss
- events developing on chronic use
- events with long latency period

4) Restricted characteristics are studied ie limited inclusion criteria

-without concomittant disease or medication

5) Narrow set of indications only specific indication is pursued

Minoxidil hair growth stimulant action was not
known during pre marketing phase
6) Limited comparision group which is often limited to placebo

7) Differnce from real life condition research setting is different

Appetite suppressant drug PAH

COX -2 inhibitor Cardiovascular events

Troglitazone Hepatocellular damage

Rosiglitazone Heart attack & related death

RCT - can assess efficacy ( does the treatment work )

- but not appropriate for effectiveness (doesnt treatment benefit

outweigh its risk for those whom it is offered in clinical practice

PE models overcome these limitations :

Observational safe
Retrospective many sources of data
Can assess effectiveness
 AIM

Signal generation

Quantification of risk &

benefit

Hypothesis testing

 Signal generation

Identification of ADR

Detection of new indications

- Minoxidil in Hypertension hair growth stimulant action

Risk / Benefit quantification

Risk quantification of ADRs - need large sample size (Rule of 3)

Quantification of beneficial effects - cure rates

 Hypothesis testing

Comparison groups required

Statistical methods (cohort & case control) are used to confirm

association between drug exposure & clinical event
( by chance / true)

Conclusion is based on the ability to reject null hypothesis

 APPLICATION
PE can provide both descriptive (qualitative) & analytic (quantitative) informations
which can be applied to -

Estimation of the risks of drug use

Use in patient counselling

Formulation of public health policy decisions

Formulation of therapeutic guidelines & discovery of new indications

Facilitation of pharmacoeconomic evaluations

 Estimation of risk of drug use

Commonest use

Case control, cohort and RCT.

Risk quantification provides information on benefit vs risk

Identification of risk situation

Dose related psychiatric disturbances of triazolam

Lowering of recommended dose / withdrawal
 Use in patient counselling
If significant risk of serious congenital malformation counsel for
termination

If lesser risk counsel for continuation of otherwise desired pregnancy

 Formulation of public health policy decision

Prescribers have inappropriate prescribing

( regulatory agencies )

impose restriction on specific drugs / practitioners)

Also provide much needed facts for evidence based medicine to

help forming rational health care policy
 Formulation of therapeutic guidelines & discovery of new indications

Provide effectiveness data for drugs in group of people which are usually not included in
phase 3 trail

Facilitation of Pharmacoeconomic evaluation

Measures effects of drugs on overall healthcare costs & resource consumption

Drugs serious adverse effects hospitalization monetary cost & resource

consumption (Important being avoidable)

Provides real life estimate of effectiveness, compliance & ADR

Provide foundation for further studies

 MEASUREMENT OF OUTCOMES
Included are studies on
Functional status
Symptom status
Patient satisfaction with various aspects of care
Quality of life studies

Therapeutic outcomes are classified as: Cure, improvement, no change,

deterioration OR success/ failure

Major indicators are morbidity and mortality

 Outcome measures
Prevalence
Cumulative incidence
Incidence rate

Drug use measures

Monetary units
Numbers of prescriptions
Units of drug dispensed
Defined daily doses (DDDs)
Prescribed daily doses

Diagnosis & therapy surveys

 Outcome measures

Prevalence Proportion of people affected with disease or exposed to

particular drug in a population at a given time

- A Cross sectional measure

- Varies between 0 & 1

- Can also be expressed as %

 Cumulative incidence Average risk of individual developing outcome in a
specified time interval

Number of new cases of disease / outcome that occurred in a population

during a specified time period
__________________________________________________________
initial no. of people in that population

A longitudinal measure

Expressed as % / in cases per 1000 or multiple there of (depending upon

rarity)
Incidence rate Ratio of new cases to the person time at risk

No. of new cases

_____________________
Person-year of exposure

Person time sum of all exposures * length of that exposure

 Drug use measure
Monetary units quantify amount being consumed by populations.

Indicates burden on society from drug use

Advantage: Convenient & conversion to common unit international comparision
Disadvantage quantity of drug actually consumed not known.
wide variation in price

Numbers of prescriptions used due to availability & ease of use of this

method.
Advantage can provide good estimate of no. of people exposed &
treatment episode for short term treatment

Disadvantage Quantities of drug dispensed & duration of treatment vary

greatly
 Units of drug dispensed(packs,tablets)
Easy to obtain

Compare usage trends within countries

Disadvantage- No information on quantity actually consumed by

patient

Limitation - Difficult to determine actual no. of patients exposed to

drug.
Defined daily doses (DDDs)
Developed to produce a standard measure of drug utilization

DDD = estimated average maintenance dose / day of a drug when used in its
major indication

Utilisation = DDD / 1000 inhabitants / day

Advantage :
- relatively easy & inexpensive
- allow drug comparison between drug utilization between different
geographic areas
- evaluation of trend over time
- provide rough estimate of prevalence of drug treated population for
chronic disease (DM)
- antibiotic utilization comparison can be related to antibiotic resistance
 Disadvantage
- Not a recommended dose, rather a technical unit for comparision
& many drugs have not yet been assigned with DDD

- Pediatric uses are not considered in calculations

- Difficult if wide variation in dose

> 1 major indication of a drug
Aspirin low dose to avoid cardiac arrest
moderate dose pain
high dose inflammation condition
 PRESCRIBED DAILY DOSE
Average daily dose of a drug that has actually been prescribed

Calculated from a sample of prescriptions

Provides refined estimate of number of person-days (yrs) of

exposure Eg risk calculation with antidepressant Newer less
toxic antidepressants were more common with suicidal cases

Validate DDD

Disadvantage doesnt indicate how many people are exposed

 III. Diagnostic & therapeutic survey
Survey data describing the prescription pattern of individual
clinicians & rates of disease encountered in practice.

Provide comprehensive assessment of drug use when combined with

national sales & prescription survey
 MEASUREMENT OF RISK : ATTRIBUTABLE RISK & RELATIVE RISK

Factor status Outcome Total

AE No AE
Exposed
A B A+B
Not exposed
C D C+D
Total
A+C B+D N
 Attributable Risk: Excess risk.
Difference between the risk in exposed group and the baseline risk in
an unexposed population.
AR=[A/(A+B)]-[C/(C+D)]

Relative Risk: Risk Ratio

Ratio of the risk rate in exposed group/risk rate in the nonexposed
group.
RR=[A/(A+B)]/[C/(C+D)]
 Time risk relationship

Risk varies with time in a number of different ways that are

dependent on the drug and/or the type of ADR it produces.

Time risk relationships demonstrate that exposure time must always

be considered and the risk should ideally be expressed as a function
of time
Pharmacoepidemiological methods
1) Qualitative Models:
Drug utilisation review(DUR)

2) Quantitative Models:
i. Case reports
ii. Case series
iii. Cross-sectional studies
iv. Cohort and Case-control studies

3) Meta-analysis model
1.
QUALITATIVE MODELS

DRUG UTILISATION REVIEW (DUR)

Assess how well the drugs are being used

Focuses on drugs which are

1. costly (3rd generation cephalosporin)
2. biotechnological agent
3. widespread use (cardiac drugs)
4. narrow margin of safety (aminoglycoside)
5. drugs prone to misuse / overuse antibiotic

Other focus - Serum concentration of drug / indicator of toxicity (serum

creatinine)
FRAMEWOK DUR.

Criteria of appropriateness (based on literature analysis & inputs from clinical experts)
are developed for -
i) indication (is the drug right for this patient ? )
ii) regimen (dose, interval, duration, route)
iii) monitoring (blood concentration of drug, serum enzyme, renal function)
iv) follow up (inappropriate regimen modified or not)

These criteria applied to the drug in question

Rate of appropriateness is determined

Feedback is provided as report, educational intervention, restriction on use
 QUANTITATIVE MODELS
1) CASE REPORT
i) weakest form of evidence for causation

ii) passive surveillance

iii) serve as an alert help clinicians to identify a potential problem

iv) may lead to hypothesis

v) strong correlation if
- disappearance of symptoms on discontinuation
- recurrence on rechallenge
- drug concentration is related to ADR
2) CASE SERIES
Group / cluster of reports that may be generated by a single clinician, group of
clinicians, hospital, pharmaceutical company or regulatory agencies ( most
important)

Helps in syndrome identification (if same malformation among exposed people)

- thalidomide induced phocomelia
- fetal alcohol syndrome
- GBS with zimelidine
Consistency in the syndrome strengthens the causal relationship

Occasionally false alarm Bendectin (doxylamine/pyridoxine)

original case series identified it as teratogen but later not confirmed
Major disadvantage lack of comparision group ( important for risk factor
identification
3) SURVEY OF DRUG USE
Single group cohort study

Includes sales data

number of prescriptions dispensed
total number of pills consumed

Limited value in PE (more important for marketing)

classical case - drug utilization pattern study could not find any relation between bendectin &
birth defects

Can be used to compare trends in different areas areas with less use of H2 receptor antagonist
higher incidence of gastric surgery

Refined forms (DDD, PDD) are more in use

4) CROSS SECTIONAL STUDY
Determine the proportion of use of a drug in a defined population
Snap shot
Can compare drug use between countries or different areas
Reason for difference in use can be determined

5) COHORT STUDY
One of the most powerful design for analytical purpose (hypothesis testing)
Framework
Cohort of people with event of interest is assembled

Exposure status is ascertained

Followed up over time to determine event development
 Advantage

i) Direct calculation of incidence rate & hence of RR, AR

ii) Information on a number of different outcomes for 1 drug
iii) Provides strong evidence (because exposure has occurred before outcome)

Disadvantage
i) Large sample size - difficult to recruit/monitor
high drop out rate over time
costly
ii) Longer follow up duration variables may change over time & the study
may not be anymore relevant

Retrospective Cohort study can overcome these limitation (high quality data is already
collected)
 Retrospective Cohort study
relevant sample is collected from past from a large pre existing data base

followed from that time to the present (with regard to the incidence of a given
outcome)

Disadvantage require a definitive hypothesis before starting the study (not

needed for cohort)
6) CASE CONTROL STUDY

This is also one of the most powerful design

Cases (with outcome of interest) & control (without the disease but otherwise similar) are identified

Exposure status is compared

Odds ratio is calculated (tell only about association NOT causality)

Classical eg
- vaginal carcinoma & DES exposure in pregnant female
- Reye syndrome & aspirin
- PUD & NSAIDs
- Venous TE with OCP
 Advantage
i) Delayed/rare events can be studied at lower cost
ii) Information on multiple risk factors associated with single outcome

Disadvantage
i) Causality cannot be established
ii) Susceptibilty to bias
- Recall bias: mothers with malformed child will have better recall (to overcome this, control
can be mothers with neonate with other malformation)
- Selection bias
- There can be difference between quality of information obtained
from study & comparision group
iii) Missing data hospital record may not record outcome
 Odds ratio
odds of developing the disease in exposed patient
odds of developing disease in the unexposed group

OR = AD/BC

- it gives estimation of relative risk (OR RR if sample size is large)

- attributable risk cannot be determined directly
 NESTED CASE CONTROL STUDY

Efficient variation of case control & cohort studies

Used when variables are expensive to measure & can be studied at

the end of the study

All cases & only a random sample of all cohort are chosen for study
from the same defined population

Eg - Serious coronary heart disease risk in relation to COX-2 selective

inhibitors ( chosen from same health plan) : rofecoxib had higher risk
than with celecoxib
 CASE-COHORT STUDY

Hybrid of cohort & case control study

As case control cases are identified & exposure status is determined

As cohort comparision group is taken as cohort of exposed individual & their outcome is
determined
Eg-
1st group female with Ca breast were given a questionnaire to determine exposure
2nd group cohort of exposed female selected from computarized data base with a linkage to
Ca registry
Data from both the groups were compared & a significant relationship was found
 case cohort study.

Advantage
rare phenomenon can be examined ( as case control)
more statistical power due to larger cohort used in comparison group

Disadvantage
Statistics is more difficult to calculate
Cohort Case control

Study group exposed people People with desired outcome

Incidence rate (hence RR,AR ) can be calculated OR can give direct estimation only for RR ( AR
directly indirect estimation)

Information on a number of different outcomes for 1 Information on multiple risk factors associated with
drug single outcome

Prospective as well as retrospective retrospective

Choice between case control & cohort study depends upon

- Question to be answered
- Frequency of the disease
- Exposure under study
 META-ANALYSIS MODEL
Comparison of results from a group of studies to arrive at a summary estimate of effectiveness
of a drug or risk of an adverse event
Analyze pooled data
Can be applied to RCT/ case control / cohort
Commonly used tools OR
Risk / rate difference AR

Application
- To resolve controversies (by comparing magnitude of difference between treatment with respect to
effectiveness or ADR)
- Results can be used as input into pharmacoeconomic analyses & to verify therapeutic
guidelines

Caution heterogeniety is seen more in observational study than that in RCT

Model Strength
Case report Signal generation
Case series Confirm association
Cross-sectional - Identify correlating factor
- Estimate prevalence
Longitudinal study - Detection of trend over time
- Can measure effects of an event
Non-exposure comparision Suggest causation
Case control study - Strength of association
- Least costly model
- Multiple exposure can be studied
Cohort study - Measure incidence & strength of - Costly
association
- Controls multiple factors
- Multiple outcomes can be studied
RCT - Establishes causation
- Controls all factors
- The only experimental model
Weakness
No causation
No causation
Confounding variables
Confounding variables
Confounding variables
- Causation not established
- Controls a single variables
- Recall bias
- Time consuming
- Generalisability
- Costly
- Focus on a single factor
Application of PE
I. Non interventional study
DUR
Prescription audit
Prescription event monitoring
Post marketing surveillance
Case control study
Cohort study
II. Interventional study
RCT
- Appropriateness of prescribing
- Quality of care
Identifying heavy users/prescribers
- Identification of potential ADR
- Measures incidence
Measure prevalence
- Measures prevalence
- Determines association
- Measures incidence
- Determine effectiveness
- Determining effectiveness
- Establish causation
Criteria for causal nature of association
Association should be consistent with mechanism of disease
Suspected cause should precede the disease
Association should be strong

RR Association
2 Weak
(2-4) Moderate
4 Strong

Association should be reproducible consistent with different methods of study

Dose response relationship higher risk in person with greater exposure to risk
factor
 System for studying drug effect in population
1. Spontaneous reporting
2. Prescription event monitoring
3. Post marketing surveillance
4. Record linkage system

SPONTANEOUS REPORTING
To record ADR
Focus new drugs & serious ADR
Clinicians are encouraged to report any & all types of ADRs
Provide early warning signal alert clinician & regulators
 WHO establishes a system for collection of ADR in 1968- signals possible
new ADR
cisapride associated tachycardia
drug induced pancreatitis

Advantage
Inexpensive
Simple to operate
Do not interfere with clinical practice

Disadvantage
Difficult to establish causation
Information obtained may be incomplete, unverified, inadequate
Voluntary under reporting true extent cant be ascertained
- selective reporting - bias
 PRESCRIPTION EVENT MONITORING (PEM)
Developed in UK at drug safety research unit to study large cohort of drug users
A method of generating & testing hypothesis about ADR in defined population of users

Information on first 5000-18000 prescriptions for a drug is obtained from national health service

Prescribers are contacted with questionnaire to determine subsequent event/clinical outcome

Estimation of incidence of various events

Comparison can be made between before & after drug use period

Eg erythromycin estolate induced jaundice

Advantage provide estimate of incidence of events over a defined follow up period
 POST MARKETING SURVEILLANCE
Originally designed to detect & quantify ADR of drugs once they enter the market
Usually by pharmaceutical manufacture
Performed as a single group follow up (cohort) study
Usually a cohort of 5000-10,000 is followed over months to years

Disadvantage: wrong interpretation due to lack of comparision group

cimetidine (early post marketing phase)

initially association was found to be with gastric carcinoma

further analysis revealed that it was prescribed to treat what were later found to be
early s/o gastric carcinoma
 RECORD LINKAGE SYSTEM
Provides accessible information on thousands of patients

Decreases the time needed for examining relation between drug exposure &
outcome

Can be used with all 3 studies cohort, case control, case cohort

Are equivalent to data from large clinical trial

Cost effective alternative to clinical megatrails
 Limitations
Sometimes incorrect results leading to incorrect conclusion

Problem in establishing actual exposure to drug or duration of use

Few data base can have inadequate number of patient to allow for examination
of rare events

Data base designed with the sole purpose of claim reimbursement will have
minimal patient information

Validation process must be built in to each study to verify the accuracy of data
1) UK GP research database- data on 4.4 million patient
including medical history & drug history
terodiline arrhythmia
suicide with antidepressants
liver disease with NSAIDs
2) Medicaid developed to pay for the healthcare in US
Advantage millions of patients rare disease can be studied
easy accessibility
confidentiality maintained
Disadvantage generalization to non Medicaid patients is a problem
people may leave the system incomplete record
potential for entry error (common to all database)
3) Motherisk programme - At hospital for sick children in Toronto
- Provide counselling service for drug exposure during
exposure
- Outcome after the delivery is compared between
exposed mother & matched control

4) Sweden register with information on drug exposure in all pregnant females

5)Saskatchewan a Canadian province with computarized database for all of its

residents
- relationship between agonist inhaler use & death was studied
 Advantage of PE
Can be used in many cases where other models cannot be used
(pregnancy, pediatric, elderly, concomitant drug & disease)

Generate drug effectiveness data

Act as signal generator - can alert clinician, manufacturer & regulator

of potential problem

Allow comparison of drug use pattern between areas excessive

prescribing / abuse can be identified

PE models are highly flexible - can use data from a variety of sources
 Disadvantage of PE
Difficult to establish causation

Bias & confounding factor

Selection bias- related to recruitment of study participants
Information bias related to accuracy of data collected
Confounding correlative factors that can make causation by the drug (being investigated)
unclear
Confounding by indication patients are selected because of severity of disease & result in
higher rate of outcome than that in the comparison group

Cohort study larger sample size, expensive, long duration

Retrospective & retrolective study recall bias