Epidemiologic Study

Epidemiologic Study Designs

Designs
Analytic
Analytic

Polly T.
Polly T. Chua-Chan,
Chua-Chan, MD,
MD, MHA,
MHA, DrPH,
DrPH, FPAFP
FPAFP
 Learning Outcomes
« At the end of the lecture, the learner should be
⚫

able to:
1. Describe
1. Describe the
the analytic
analytic study
study designs
designs in
in terms
terms of:
of:
a. identifying features
b. measures of disease occurrence and of
associations
c. advantages and disadvantages
2. Analyze
2. Analyze and
and correctly
correctly interpret
interpret data
data from
from an
an
analytic study
⚫ In epidemiology, we would like to determine
what factors cause a disease/ death?
⚫ Factors
- Exposure Factors
- Risk Factors
⚫ – event of interest (Disease/Death)
Outcome —

⚫ Exposure should preceed before the outcome

 Terminologies
⚫ Outcome: Health related event of interest
Outcome:
e.g. death, disease condition like AIDS
⚫ Exposed: Having, or being exposed to, a potential
cause or risk factor for disease or other outcome
e.g. being a smoker or having sex with
with someone
who
who has an STI
• Unexposed: Not having, not being exposed to, a
potential cause or risk factor for disease or
other outcome
• Cases: Individuals that are disease positive
• Controls: Individuals that are disease negative
 Analytic Studies
Analytic Studies

•» Test the hypothesis of relationship

between at minimum, two variables, one
independent variable
independent variable (exposure)
(exposure) and
and one
one
dependent variable (disease)
Smoking and
Smoking and Hypertension
Hypertension in
in a
a
Community
Community

» Hypothesis:
⚫ Hypothesis: Is
Is smoking
smoking related
related to
to
hypertension?
hypertension?

» Factor
⚫ Factor ?
?
-
− Smoking
Smoking
« Outcome
⚫ Outcome ?
7?
Hypertension
 Smoking and
Smoking and Hypertension
Hypertension in
in a
a
Community
Community
« Are there people who smoke and have
⚫

hypertension?

« Are
⚫ Are there
there people
people who
who do
do not
not smoke
smoke and
and have
have
hypertension?
hypertension?

« Are
⚫ Are there
there people
people who
who smoke
smoke but
but do
do not
not have
have
hypertension?
hypertension?
 Smoking and
Smoking and Hypertension
Hypertension in
in a
a
Community
Community
Risk Factor
Sc Disease Present
80
Ww
QL
(=
is C r
[4] is 5
|=
©
sof
selected
Population Sample
5
QL No Risk Factor
wr) Disease Present
wv
wv
oO
Pi
|
No Risk Factor
Mo Disease
Cross-sectional Study
« E and O are

\ point in time

Wwinvout
Without© outcome
 Uses of Cross-sectional Study

+ Estimate burden of disease (prevalence)

» Establish baseline data

» Determine association between coexisting

variables
Cross-sectional Study: Analysis
Measures of Disease Occurrence
» Prevalence in the population
» Prevalence among exposed
» Prevalence among unexposed

Measure of Association
» Prevalence ratio
Prevalence Proportion
number of existing cases of
Prevalence _ a certain disease at time point F
Proportion
population examined at time point

- Prevalence in the sample/population (P)

- Prevalence among exposed (Pg)
- Prevalence among unexposed (Pg)
Prevalence Ratio (PR)

Prevalence among exposed

PR =
Prevalence among unexposed
Prevalence
Ratio: Inferorelation
Null value:

» Exposure « Exposure Is » Exposure

variable not variable
negatively associated positively
associated with the associated
with the outcome with the
outcome outcome
 Prevalence Ratio
. IfIf the
⚫ the numerator
numerator is
is equal
equal to
to the
the denominator,
denominator,
the ratio = 1,1, no relationship. 11 is a null value.
. IfIf the
⚫ the prevalence
prevalence of
of the
the disease
disease is
is higher
higher among
among
the exposed than among the unexposed, the
1.
ratio is > 1.
•* The exposure variable is a risk factor to the
disease.
• If the prevalence of the disease among the
unexposed is higher than among the exposed, the
1, the exposure variable is protective of
ratio is < 1,
disease
the disease.
Cross-sectional Study: sues
Advantages | Disadvantages
v' Resource-efficient x Difficult to establish
v' Does not suffer from causality
attrition » Temporal ambiguity
between exposure and
v Generalizable outcome: “in came
= Often based on a first?
sample of the general
population
 Cohort Study
Cohort Study
•* This approach is suitable in determining the
influence of a particular risk factor in the
causation of an event, such as a particular
disease.
•» Begins
Begins with
with the
the identification/selection
identification/selection of
of a
a
group exposed to a risk factor, and a
comparable group not exposed.
•* Both groups are then observed if they will
develop the sickness.
 Cohort Study
Cohort Study

Cohort
- A group whose members share a distinct set of
characteristics, followed up within a specific period
of time
= the shared characteristics may be secondary to:
a) a common setting (a school cohort) and/or
b) 2a common experience (a birth cohort)

Cohort Effect refers to the secular change

of disease frequency in a group as influenced by
membership in that particular group.
« E Is measured at
present time and
the participants are
followed up to
measure O at a
future time
A study looked at whether exposure to
bisphenol A (BPA) early in life affects obesity
levels in
levels in children
children later
later in
in life.
life.

Hoepner, L., Whyatt,

Hoepner, Whyatt, R., Widen, E.,
E., Hassoun, A., Oberfield, S., Mueller, N.,
A., Oberfield, N., ... Rundle,
Rundle, A.
A. (2016).
Bisphenol A and Adiposity in an Inner-City Birth Cohort. Environmental Health Perspectives,
Perspectives,
124(10), 1644-1650.
124(10), 1644-1650. https://doi.org/10.1289/EHP205
https://doi.org/10.1289/EHP205
•» Factor?
Bisphenol A (BPA)
•* Outcome?
Adiposity
Adiposity

At the start, those with the outcome of

Interest should be excluded.
interest
A study
A study looked
looked at
at whether
whether exposure
exposure to
to bisphenol
bisphenol
A (BPA) early inIn life affects obesity levels in
children later in life.
Positive associations were found between
prenatal BPA concentrations in urine and
increased fat mass index, percent body fat,
and waist circumference at age seven.
L., Whyatt, R.,
Hoepner, L., R., Widen, E., Hassoun,
Hassoun, A., Oberfield, S.,
S., Mueller,
Mueller, N.,
N., ...
... Rundle, A. (2016).
(2016).
Cohort. Environmental Health Perspectives,
Bisphenol A and Adiposity in an Inner-City Birth Cohort. Perspectives,
124(10), 1644-1650.
124(10), 1644-1650. https://doi.org/10.1289/EHP205
https.//doi.org/10.1289/EHP205
 Relationship of
Relationship of Smoking
Smoking
to Lung
to Lung Cancer
Cancer

•* An 1000 males, age 15

An investigator may survey 1000 15
and over, and note down who smokes and who
does not.

•« He found out that 700 smoke and 300 did not.

•* He observed them for 1010 years, and at the end

of the period, he had the following results.
 Nid not Incidence per
Personyrs of Developed 1000 person
Characteristics ~~ observation Lung Ca. developed
years

Smoker 6650 150 550 22.56

Not smoker 4585 10 290 2.18

Total 11,235 160 840 14.24

•» 160
160 cases of lung cancer developed over a 10 10
11,235
year period of observation, equivalent to 11,235
person years for aa cumulative incidence rate of
14.24/1000 person years.
14.24/1000
Measures of Disease Occurrence
» Incidence proportions
~lncidence rates

Measures of Association
~ Risk ratio
~ Rate ratio
incidence Proportion
number of new cases of
Incidence
: -
a disease In a specific ime period
x -
Proportion population at risk of developing the
disease In the same time period
» Describes the average risk for developing a
disease condition
» More useful in determining etiologic/risk factors
of diseases more than prevalence proportion
Incidence Rate
number of new cases of
Incidence _ a disease in a specific time period =
Rate
total person-time (PT) at risk

» Describes the speed at which new cases occur

“Time at risk”
= period or duration at which each person is
vulnerable ("at nsk”) of developing the disease
Relative Risk (RR)

Incidence among exposed

RR =
Incidence among unexposed

Also known as:

» Risk Ratio — if comparing incidence proportions
+ Rate Ratio — If comparing incidence rates
 Personyrs of Developed Did not Incidence per
Characteristics observation Lung Ca. developed 1000 person
years

Smoker 6650 150 550 22.56

Not smoker 4585 10 290 2.18

Total 11,235 160 840 14.24

Various
Various of ways of comparing the cancer rates between the two groups
(smoker/non-smoker).
(smoker/non-smoker).
1. Risk
1. Risk Ratio
Ratio (RR)
(RR) -- divide
divide the
the incidence
incidence of
of cancer
cancer inin the
the smokers
smokers (22.56)
(22.56)
by the incidence in the non-smokers (2.18) and the quotient is 10 10 plus,
plus,
meaning the
meaning the incidence
incidence among
among smokers
smokers isis 10
10 times
times as
as great
great than
than
hon-smokers.
among non-smokers.
• RRRR is
is perhaps
perhaps the
the more
more reliable
reliable indicator
indicator that
that the
the association
association of
of
smoking and
smoking and lung
lung cancer
cancer is
is indeed
indeed aa causal
causal relationship.
relationship.
 Relative Risk: hlenorelation
Null value:
0 1.0 00

* EXposure * Exposure « EXposure

variable is variable variable is
protective has no a risk
factor effect on factor
outcome
Relative Risk (RR)
* The relative risk calculated from a cohort study is
the incidence of disease in the exposure group
over the incidence of disease in the unexposed.
* RR>1 shows association, where the conclusion
can be read "those with the exposure were more
likely to develop disease."
* The RR is a more powerful effect measure than
the OR, as the OR is just an estimation of the RR.
2. Risk Difference - also called attributable risk
of the exposed (smokers).
- also called excess risk, or absolute risk.
•* This
This is
is done
done by
by subtracting
subtracting from
from the
the incidence
incidence of
of
cancer among smokers, that of the non-
smokers.
22.56 - 2.18 = 20.38 /1000 person years
•» Useful measure of the extent of the lung cancer
problem
problem
3. Attributable Fraction (exposed) or etiological
fraction
• by dividing the risk difference by the rate of occurrence
among the exposed population.
(20.38 ÷+ 22.56)
(20.38 22.56) xx 100
100 == 90%.
90%.
• ItIt measures
measures the the impact
impact on on the
the incidence
incidence among
among smokers
smokers ifif
the risk factor (smoking in this case) is eliminated.
• In
In this
this particular
particular case,
case, ifif we
we assume
assume thatthat smoking
smoking isis the
the
only determinant
only determinant of of cancer
cancer of of the
the lungs,
lungs, elimination
elimination of
of
smoking will
smoking result in
will result in 90%
90% reduction
reduction of of lung
lung cancer
cancer
incidence among smokers.
• Attributable fraction is
Attributable fraction is a
a useful
useful tool
tool in
in deciding
deciding priority.
priority.
• IfIf to
to control
control lung
lung cancer,
cancer, one one will have to
will have to choose
choose between
between
smoking control
smoking control and
and airair pollution
pollution control,
control, computation
computation of of
attributable fraction will will help.
 Person yrs of Developed Did not Incidence per
Characteristics observation Lung Ca. developed ee

Smoker 6650 150 550 22.56

Not smoker 4585 10 290 2.18
Total 11,235 160 840 14.24

4. Population Attributable Risk or attributable fraction (population)

• IfIf from
from the
the incidence
incidence of
of lung
lung cancer
cancer in in the
the total
total population
population of of
observation (14.24/1000),
observation (14.24/1000), the the incidence
incidence of of lung
lung cancer
cancer inin the
the
unexposed group
unexposed group (2.18/1000)
(2.18/1000) is is subtracted,
subtracted, and
and the
the difference
difference isis
divided by
divided by the
the incidence
incidence inIn the
the total
total population,
population, the
the result
result is
is what
what is
is
referred to as population attributable risk.
(14.24 –
(14.24 — 2.18)
2.18) ÷+ 14.24
14.24 xx 100
100 == 84.69%
84.69%
• The
The population
poputation attributable
attributable risk is 85%
risk is 85% which
which represents
represents the
the probable
probable
amount of
amount of reduction
reduction in the incidence
in the incidence rate
rate in
in the
the general
general population
population ifif
smoking will
smoking will be
be eliminated.
eliminated.
In summary, the various means of comparison are:

Let A = Incidence of Ca among smokers

B = Incidence of Ca among non-smokers
C = Incidence of Ca in the population
RR = Risk ratio or relative risk
RD = Risk difference
AFE = Attributable fraction (exposed)
AFP = Attributable fraction (population)

RR = 2 RD = A-B
B

AFE = — X 100 AFP = = 2X 100

 Cohort Study
Cohort Study

« Follow
⚫ Follow Up
Up Study
Study
⚫ Longitudinal Study
Longitudinal Study
. Incidence
⚫ Incidence Study
Study
 Types of Cohort
** Exposure →
— Outcome
WITHOUT OUTCOME

Start
R (Prospective/ concurrent)

(Retrospective, non-concurrent) R

R
(Ambispective)

R - Researcher
 Uses of
Uses of Cohort
Cohort Study
Study

» Calculate risk of developing disease

» Establish the temporal relationship between
study variables
» Expedient in examining multiple outcomes
+ Practical In studying rare exposures
 Cohort Study: sues
| Advantages | Disadvantages
v More likely to » Resource-intensive
determine causality + May need long-
(as compared to a follow-up
cross-sectional study) Expnilve

x Suffers from attrition

 Case-Control Study
Case-Control Study
⚫ A type of study that attempts to capture
the advantages of both the cross-sectional
study and the cohort study.

o It
⚫ It tries
tries to
to eliminate
eliminate temporal
temporal ambiguity
ambiguity of
of
the cross-sectional study while at the
same time shorten the duration of the
study.
 Case-Control Study
Case-Control Study
. We select the cases from a target
⚫

population.

« Cases
⚫ Cases are
are a
a group
group of
of individuals
individuals with
with the
the
outcome or disease
•* Then select another group of individuals
without the outcome or disease as
Controls
Controls
 Case-Control Study
Case-Control Study
« Go
⚫ Go back
back in
in time
time (retrospective)
(retrospective) to
to
determine exposure in In the cases and in
the control
« Compares the exposure status among the
⚫

cases and among the controls

« Trohoc study
⚫
Investigating outbreaks
Investigating outbreaks of
of food
food poisoning
poisoning

•» Get cases reported by the hospital, clinic, or concerned

citizens. The investigator will attempt to produce a
complete list of all those who were sick.

•* From them and from other sources, he will also attempt

to produce a list of people who were in the same party
but who were
were not sick.
 Ate dishA Ate dishA

SICA NOT SICK

Did not eat dishA Did not eat dishA

•» Sick group and a non-sick group (a

reference or
reference or control
control group).
group).
•* For each group he will attempt to find out
what they ate. He will test for various dishes.
• Let us assume that dish A A was indeed the
culprit.
•* Assuming
Assuming dish A was the culprit.
•* There were 145145 people in the party, 80 of whom
got sick. Of the sick, 65 ate dish A. Of the 65 not
18 ate dish A,
sick, 18 A, 47 did not.

Sik Not sick

Ate dish A CB 18
Did not eat dish A 15 a7
Tota EE
 Sick Not sick
Ate dish A 65 18
Did not eat dish A 15 47
Total 80 65

•» The association between sickness and dish A A

can be measured by calculating the odds ratio.
65/15 ÷+ 18/47
18/47 = 47/18 x 65/15
= 3055/270
11.3
= 11.3
•* The figure 11.3
11.3 means that the odds of eating
dish Ais 11.3 times as great among the sick
A is 11.3
compared to the not sick.
•* Where
Where other
other dishes
dishes served
served in in the
the party
party give
give
much lower
much lower odds
odds ratio
ratio (obtained
(obtained by by computing
computing for
for
each dish
each dish as
as done
done for
for dish
dish A)
A) the
the conclusion
conclusion
would be that
would be that dish
dish A
A was
was the
the cause
cause of
of the
the
poisoning.
poisoning.

• This is an epidemiological hypothesis. While it

may be
may be enough
enough to to be
be a
a basis
basis for
for concrete
concrete
preventive action,
preventive action, further
further studies
studies such
such as
as
microbiological studies
microbiological studies of
of food
food samples,
samples,
examination of
examination of the
the kitchen,
kitchen, the
the premises,
premises, the
the
food handlers, the process of food preparation,
its source
its source and
and procurement
procurement willwill need
need to
to be
be
further investigated.
 « O Is measured at
present time and E
of the participants
In the past Is
estimated

Longitudinal
Retrospective
Observational
Case-Control vs Retrospective
Cohort
Cohort SILoly:

Prospeclive vs Reltrospeclive
« Dependent on the temporal relationship between the
initiation of the study and the occurrence of the oulcome

li) Darian of

CASE-CONTROL STUDY

FAST TIME FUTURE"

 Uses of Case-Control Study

» Determine association between variables

» Estimate relative risk

» Expedient in examining multiple exposures

« Practical In studying rare diseases
Case-Control Study: Analysis
Measure of Disease Occurrence
Pe None

Measure of Exposure Qceurrence

~ Exposure Odds

Measure of Association
Odds

probability of an event happening

Odds =
probability of an event not happening

probability of having the exposure

Exposure Odds = —
probability of not having the exposure
Exposure Odds
» Exposure Odds among Cases

probability of having the exposure among cases

probability of not having the exposure among cases

» Exposure Odds among Controls

‘probability of having the exposure among controls

probability of not having the exposure among controls
 Odds,
Odds Ratio (OR OR) OR=-( =
Null value:
0 1.0 “0d

» Exposure » Exposure » Exposure

variable is variable has variable is &
protective no effect on risk factor
factor outcome
Odds ratio (OR)
* The ratio of two odds. The odds ratio calculated from a
case-control study is the ratio of the odds of exposure
among the cases to the odds of exposure among the
controls.
* |f OR>1, then the conclusion is the "those with the
disease are more likely to have the exposure”
* |f OR=1 the exposure and disease are not associated.
* |f OR<1, it can be said that the exposure has a
protective effect against the disease
Case-Control Study: sues
Advantages Disadvantages |
v'Less resource- x Likely to suffer from
intensive (as problems with recall
compared to a cohort
study)
v' No attrition
 The effect
The effect ofof adding
adding pilates
pilates to
to a
a treatment
treatment
regimen of
regimen of NSAID
NSAID useuse for
for individuals
individuals with
with
chronic low
chronic low back
back pain.
pain.
Individuals who
Individuals who included
included the
the pilates
pilates method
method in
In
their therapy
their therapy took
took fewer
fewer NSAIDs
NSAIDs and
and experienced
experienced
statistically significant
statistically significant improvements
improvements in in pain,
pain,
function, and
function, and quality
quality of
of life
life ...

otti, L.,
Natour, J., Cazotti, L., Ribeiro, L., Baptista,
Baptista, A.,
A.,, & Jones, A. (2015). Pilates improves pain, function and quality of life in patients with chronic low back pain: a randomized controlled trial. Clinical Rehabilitation, 29(1), 59-68
29(1), 59-68.
https://doi.org/10.1177/0269215514538981
https://doi.org/10.1177/0269215514538981
 Experimental Study
Experimental Study
« Randomize
⚫ Randomize into
into Test
Test (Experimental)
(Experimental) group
group and
and
Control group
« Randomization
⚫ Randomization ensures
ensures that
that both
both groups'
groups’
characteristics will become homogenous (the
same), EXCEPT for the treatment

« Manipulation?
⚫ Manipulation?
One group
One group given
given NSAID
NSAID alone
alone (Control)
(Control)
- One
− One group
group given
given NSAID
NSAID +
+ Pilates
Pilates (Experimental)
(Experimental)
 Experimental Study
Experimental Study
+ a cohort study, only with:
py O+ (" assignment
\O-/ manipulation of E
+ 4 randomization
random allocation

+ best study design for

controlling confounders

» provides strongest causal

inference
Prospective
Experimental
 Randomization

« a.k.a. Random allocation

= Process of assigning each participant into a group—
whether experimental or control—by chance
= Differentiate from Random Sampling

« A defining feature of a modern trial design

= Absence in an experimental study makes the study
guasr-expernmental in nature
 MNon-participants
(do not meet
selection criteria)

Participants
Figure 1:
Design of a Randomized Controlled Trial
Randomization

Treatment

11
Experimental Sivoy:

Classification
| Clinical Trial || community Trial |
» Intervention is allocated = Intervention is allocated
to individuals to an entire community

Therapeutic Trial | Prophylactic Trial

+ Intervention is a « Intervention is a
treatment agent preventive agent
| Parallel Design | Crossover Design |
+ a.k.a. "between- « a.k.a. “within-subjects”
subjects” design design
Experimental Study: Analysis
Measures of Disease Occurrence
» Incidence proportions
~Incidence rates

Measures of Association
~ Risk ratio
~ Rate ratio
Experimental Soy:
Treatment Effect
Treatment > | controL — | INTERVENTION
Effect
| contrOL

= The proportion of cases of disease that were

prevented by the intervention under study among
those who received it

« May also be given by:

Treatment Effect = 1- RR
 Clinical Trials

Experimental Event Rate (EER) — Number of events or

Disease (risk) in the experimental group.

EER=
a+b
 Clinical Trial

Control Event Rate (CER) - Number of events or Disease (risk) in the

control group

CER =
c+d
 Clinical Trial
Risk Ratio — ratio of the risk of the disease in the experimental group
(EER) and the risk in the control group (CER).

EER
CER
 Clinical Trial
Absolute Risk Reduction (ARR) — also called the Risk difference is the
difference in the event rates for the EER and CER.

ARR =CER-EER
Clinical Trial

CER -EER
RRR
=
CER
ARR
CER
 Clinical Trial
Numbers Needed to Treat (NNT) - the number of patients who would
have to receive the treatment for one of them to benefit.

Calculated as 1 divided by the absolute risk reduction.

« Ethical considerations
= potential threat to experimental group
= loss of potential benefit to control group

» Resource-intensive
= may need long follow-up
= @Xpensive
• Intervention
[Intervention done but randomization was not
possible?

•« Quasi-experimental
Quasi-experimental study
study
 Intention to Treat vs
Per Protocol Analysis
• |n
In interventional studies, a subset of participants often
do not conform to the protocol →= “protocol violations”
Types:
• One or more participants (for some reason) do not receive the
respective interventions to which they were randomized
• Inadvertently receive an intervention meant for the other trial
arm
• Receive a prohibited concomitant intervention
• Not available for assessment of the planned outcome either
reason.
because of loss to follow-up or for another reason.
•« During the analysis of the trial results, the
researcher is tempted to exclude such
“nonconforming” participants.

•« The
The motivation
motivation is
is not
not one
one of
of deceit,
deceit, but
but of
of
integrity, ensuring that comparisons are made
between those participants in each trial arm who
strictly adhered to the planned treatment so that
the true efficacy of one intervention over the
other can be assessed.
 Exclusion Poses Problems:
1. It violates the principle of randomization
1.
In a 2-arm study, randomization ensures
comparability of the two groups
l.e., balanced for known and unknown confounders or
i.e.,
prognostic factors, only as they were
were originally
randomized.
When some participants in either or both the
groups are excluded, the remaining participants in the
two groups can no longer be considered as balanced.
The problem becomes larger as the number of
exclusions increases.
 Exclusion Poses Problems:
At times,
2. At times, the
the noncompliance
noncompliance is
is related
related to
to a
a
particular intervention or to disease severity.
E.g. the inability to complete the scheduled treatment or
appearance of unacceptable side effects may be more frequent
In patients with severe disease.
in
These may occur more often in the active treatment arm
than in the placebo arm. Hence, exclusion of the participants
who do not complete the treatment or follow-up as planned
would lead to differential exclusion of patients with severe
disease in the treated group, with the residual group unlikely to
resemble the original group obtained at randomization. This may
is.
make the treatment look better than it actually is.
 Exclusion Poses Problems:
3. Exclusion of participants in one or both groups,
particularly if their number is large, may lead to a
significant reduction in sample size and hence in In
study power.

4. Exclusions can introduce a bias.

Often the decision to exclude a particular
participant is controlled, at least to some extent, by the
investigator, who may be tempted to exclude patients
who
who are not doing well
well in a particular arm
 Exclusion Poses Problems:
5. The purpose of a trial is to assess the proportion
of persons in a group who may be expected to
benefit from a particular treatment.
Those who
who do not complete treatment can of
course not be expected to benefit from it. The
proportion of responders among those who complete
treatment thus provides an exaggerated estimate of
treatment effect –— this does not accurately reflect the
beneficial effect that may be expected in clinical
practice among those who who are prescribed this
particular treatment.
 Intention to Treat vs
Per Protocol Analysis
•« To obviate (or minimize) these problems, it is
recommended that “intention-to-treat (ITT)
analysis” be used.

•* The principle of ITT analysis is that all

all
participants should be analyzed in the group to
which they had been randomized, i.e. as if they
had received the intervention which they were
supposed to receive, irrespective of the
treatment actually received.
 Intention To Treat (ITT)
• ITT - analysis which “Includes all randomized
patients in the groups to which they were
randomly assigned,
randomly assigned, regardless
regardless of
of their
their
adherence with the entry criteria, regardless of
the treatment they actually received, and
regardless of subsequent withdrawal from
treatment or deviation from the protocol.” (Fisher
1990)
1990)
 Intention To Treat (ITT)
• The use of ITT analysis ensures maintenance of
comparability between groups as obtained through
randomization, maintains sample size, and eliminates
bias.
• |n
In addition, results obtained in such analysis more
closely represent clinical practice, dealing with
“effectiveness” of the intervention rather than “efficacy.”
In view of these advantages, ITT is today considered
as a de facto standard for analysis of clinical trials,
though a minority school of thought believes that this
approach is too conservative.
 Per-Protocol (PP) Analysis
• |n
In contrast, per-protocol (PP) analysis refers to
inclusion in the analysis of only those patients who
who
strictly adhered to the protocol.
•« The PP analysis provides an estimate of the true
efficacy of an intervention, i.e., among those who
completed the treatment as planned.
• However, as discussed above, its results do not
represent the real life situation and it is likely to show
an exaggerated treatment effect.
Ranganathan, P and Aggarwa;,
Ranganathan, 144-146
Aggarwa;, R. Perpect Clin Res 2016 Jul-Sept: 7(3): 144-146
•« The CONSORT guidelines for reporting of “parallel
group randomized controlled trials” recommend that
both ITT and PP analyses should be reported for all
planned outcomes to allow readers to interpret the
effect of an intervention. ((CONSORT 2010 statement
CONSORT 2010 statement updated
updated
guidelines for
guidelines for reporting
reporting parallel
parallel group
group randomised
randomised trials
trials Schulz
Schulz KF,
KF, Altman
Altman
DG, Moher
DG, Moher D,
D, CONSORT
CONSORT Group)
Group)

•» There are some special situations. E.g. in noninferiority

trials, the use of PP analysis is considered particularly
important.
 high

confidence

ow @
Sources: MDJ
Sources: MDJ Peters
Peters et
et al.
al. (2015),
(2015), Levac
Levac et
et al.
al. (2010),
(2010), Godfrey
Godfrey C,
C, Khalil
Khalil H,
H, et
et al
al
 Systematic Review
Systematic Review
•« “a review of the evidence on a clearly formulated
guestion that uses systematic and explicit methods to
question
identify, select and critically appraise relevant primary
research, and to extract and analyze data from the
studies that are included in the review.” The methods
used must be reproducible and transparent.
Undertaking Systematic Reviews of
of Research on Effectiveness.
Effectiveness. CRD’s Guidance for
or Commissioning Reviews.
those Carrying Out or Reviews. CRD Report Number
Number 4 (2nd Edition).
Edition).
NHS Centre
NHS Centre for
for Reviews
Reviews and
and Dissemination,
Dissemination, University
University of
of York.
York. March
March 2001.
2001.

•« A comprehensive review of all relevant studies on a

particular clinical or health-related topic/question.
 Systematic Review
• The scope of the review is identified in advance (eg
sub-questions and/or sub‐group
review question and sub‐questions sub-group
analysis to be undertaken)
• Comprehensive search to find all relevant studies
(published/unpublished/on-going studies)
• Use of explicit criteria to include / exclude studies
• Application
Application of established standards to critically
appraise study quality
• Explicit methods of extracting and synthesizing study
findings (qualitative or quantitative)
• May include a meta-analysis (quantitative synthesis)
*optional
 Systematic Review
Systematic Review
Uses:
• |dentifies,
Identifies, appraises and synthesizes all available
research that is relevant to a particular review question
•« Collates all that is known on a given topic and identifies
the basis of that knowledge
•« Comprehensive report using explicit processes so that
rationale, assumptions and methods are open to
scrutiny by external parties
•« Can be replicated / updated
 Systematic Review
Systematic Review
Limitations:
•« Systematic reviews with narrowly defined review
questions provide specific answers to specific
questions
• Alternative questions that have not been answered
usually need to be reconstructed by the reader

Source: Cochrane. Background to Systematic Reviews

Source:
 Advantages
« Exhaustive
⚫ Exhaustive review
review of
of the
the current
current literature
literature and
and
other sources (unpublished studies, ongoing
research)
research)

« Less
⚫ Less costly
costly to
to review
review prior
prior studies
studies than
than to
to
create a new study

« Less
⚫ Less time
time required
required than
than conducting
conducting a
a new
new study
study
 Advantages
« Results
⚫ Results can
can be
be generalized
generalized and
and extrapolated
extrapolated
Into the general population more broadly than
into
individual studies

« More reliable and accurate than individual

⚫

studies

« Considered an evidence-based resource

⚫
 Disadvantages
Disadvantages
» Very
⚫ Very time-consuming
time-consuming

» May not be easy to combine studies

⚫
 Design pitfalls
Design pitfalls to
to look
look out
out for
for
⚫ Studies included in systematic reviews may
be of varying study designs, but should
collectively be studying the same outcome.
⚫ Is each study included in the review studying the same
variables?
⚫ Some reviews may group and analyze
studies by variables such as age and
gender; factors that were not allocated to
participants.
⚫ Do the analyses in the systematic review fit the
variables being studied in the original studies?
 Meta-Analysis

• “* the combination of data from several

Independent primary studies that address the
independent
same question to produce a single estimate like
the effect of treatment or risk factor. It is the
statistical analysis of a large collection of
analysis and results from individual studies for
the purpose of integrating the findings.
GV. Primary,
Glass GV. Primary, secondary,
secondary, and meta-analysis of
of research.
research. Educ Res.
Res. 1976;5:3–8.
1976,5:3-8.
 Meta-Analysis

•+ Meta-analysis provides a logical framework to a

research review
research review where
where similar
similar measures
measures from
from
comparable studies are listed systematically and
the available effect measures are combined
wherever possible.
Meta-analysis and the integration of research in special education.
Kavale KA,
Kavale KA, Glass
Glass GV
GV J
J Learn
Learn Disabil.
Disabil. 1981
1981 Nov;
Nov; 14(9):531-8.
14(9):531-8.
 Meta-Analysis
–
Fundamental rationale —
•* it reduces the quantity of data by summarizing data
from multiple resources and helps to plan research as
well
well as to frame guidelines.
• It also helps to make efficient use of existing data,
ensuring generalizability, helping to check consistency
of relationships, explaining data inconsistency, and
quantifies the data.
•* It|t helps to improve the precision in estimating the risk
by using explicit methods.
 Meta-Analysis
« A
⚫ A subset
subset of
of systematic
systematic reviews;
reviews;

« A
⚫ A method
method for
for systematically
systematically combining
combining pertinent
pertinent
qualitative and quantitative study data from
several selected studies to develop a single
conclusion that has greater
greater statistical power.

⚫ This conclusion
This conclusion is
is statistically
statistically stronger
stronger than
than the
the
analysis of any single study, due to increased
numbers of subjects, greater diversity among
subjects, or accumulated effects and results.
 Uses of
Uses of Meta-Analysis:
Meta-Analysis:
⚫ To establish statistical significance with
studies that have conflicting results
⚫ To develop a more correct estimate of
effect magnitude
⚫ To provide a more complex analysis of
harms, safety data, and benefits
⚫ To examine subgroups with individual
numbers that
numbers that are
are not
not statistically
statistically
significant
significant
 Advantages
Advantages

⚫ Greater statistical power

⚫ Confirmatory data
Confirmatory data analysis
analysis
⚫ Greater ability
Greater ability to
to extrapolate
extrapolate to
to general
general
population affected
⚫ Considered an evidence-based
resource
 Disadvantages
⚫ Difficult and
Difficult and time
time consuming
consuming to
to identify
identify
appropriate studies
⚫ Not all studies provide adequate data
for inclusion and analysis
⚫ Requires advanced statistical
techniques
⚫ Heterogeneity of study populations
 Design pitfalls
Design pitfalls to
to look
look out
out for
for
« The
⚫ The studies
studies pooled
pooled for
for review
review should
should be
be
similar in type (i.e. all randomized controlled
trials).
« Are
⚫ the studies
Are the studies being
being reviewed
reviewed all
all the
the same
same type
type of
of
of different types?
study or are they a mixture of
« The
⚫ The analysis
analysis should
should include
include published
published and
and
unpublished results to avoid publication
bias.
« Does the meta-analysis include any appropriate
⚫

relevant studies that may have had negative

outcomes?
 Systematic Review
Systematic Review vs
vs
Meta-Analysis
•« Systematic review will refer to the entire process
of collecting, reviewing, and presenting all
available evidence

•+ Meta-analysis
Meta-analysis will
will refer
refer to
to the
the statistical
statistical
technique involved in extracting and combining
data to produce a summary result.
An Introduction to meta-analysis,
An introduction meta-analysis, Cochrane Collaboration open learning material for
reviewers, Version
reviewers, Version 1.1,
1.1, November.
November. 2002.
2002.
 Quality of
Quality Evidence
of Evidence

Image: EBM Pyramid and EBM Page Generator,

Image: Generator, copyright 2006 Trustees of
Dartmouth College and Yale University. All Rights Reserved. Produced by Jan
Glover, David Izzo, Karen Odato and Lei Wang.
Glover,
⚫ Randomized Controlled Trials (RCTs) are the gold
standard for demonstrating causality between the use
of a specific medicine and intended and unintended
effects under ideal conditions.
⚫ –
Limitations of RCT —
⚫ Limited to evaluating specific interventions one by one
⚫ When selecting the patient sample according to strict
inclusion and exclusion criteria, they have reduced
external validity (or generalization), which limits the
transfer of their results.
« The
⚫ The highly
highly selective
selective populations
populations examined
examined within
within the
the
setting of RCTs are often not comparable with the
more heterogeneous populations in clinical practice.
« Heterogeneous
⚫ Heterogeneous population
population in
in the
the real
real world setting
world setting
(patients with varying genetic make-ups, who present
with different comorbidities or already receive different
medications for other morbidities).
. “Efficacy-effectiveness gap” (Eichleretal)
⚫ (Eichleretal) -- disparity
disparity
of findings on the therapeutic efficacy of medicines
from tightly controlled RCT settings and the
effectiveness of medicines in the real world.
 Real World Data
•« Real world evidence (RWE) means evidence
obtained from real world data (RWD), which
are observational studies based on real
data obtained from daily clinical practice.

•« RWD could be defined as studies that collect

data relevant to human health that do not come
from conventional randomized clinical trials.
 Real World Data
•« They document the real care that patients
receive in the clinic and include a variety of
cases (for example, patients suffering from
several diseases at once) without limiting strict
inclusion and
inclusion and exclusion
exclusion criteria.
criteria.

•* They can generate long-term data on the

effectiveness and safety of health interventions,
in addition to providing useful information for
economic health analyses.
 Real World Data
•« RWD and RWE provide the external validity that
RCTs lack.

Advantages over RCTs:

Advantages over RCTs:
• lower cost, larger sample size, and greater
representativeness than
representativeness than other
other research
research designs,
designs,
and high external validity.
 Real World Data
« Various
⚫ Various sources:
sources: patient
patient registries,
registries,
administrative claims, and social media
channels
« Application:
⚫ Application: Different
Different real
real environments
environments
providing insights into drug safety, in health and
financial terms, and its long-term effects.
« Real-world
⚫ Real-world data
data can
can improve
improve decision-making,
decision-making,
pre-authorization, and reimbursement of new
drugs and treatments and benefit medical
research and patient outcomes.
 References
«
⚫ Notes, Dr.
UP Notes, Dr. Paul Adrian V.
V. Pinlac
+ Aschengrau A and Seage GR (2008). Essentials
of Epidemiology in
Public Health, 2nd ed. Massachusetts.
« Friis RH and Sellers TA (1996). Epidemiology for Publ
ic Health
Practice. Maryland.
- Gordis L (2014). Epidemiology, 5th ed. Canada.
« Hennekens CH and Buring JE (1987). Descriptive Stud
ies.
In: Epidemiology in Medicine.
« Kelsey JL, Thompson WD, Evans AS (1986). Met
hods in
Observational Epidemioclogy. New York.
« Rothman, KJ (2002). Epidemiology: An Introduction
. New York.
+ Webb P and Bain C (2011). Essential Epidemiology:
An Introduction
for Students and Health Professionals, 2"d ed. New
York.
 Traditional Literature
Review/Narrative Review
• Describes and appraises previous work but does not
describe specific methods by which the reviewed
studies were identified, selected and evaluated
• Uses:
• Overviews, discussions, critiques of previous work and
the current gaps in knowledge
• Often used as rationale for new research
• To scope the types of interventions available to include
In a review
in
 Traditional Literature
Review/Narrative Review

Limitations:
•« The
The writers
writers assumptions
assumptions and
and agenda
agenda often
often
unknown
•« Biases that occur in selecting and assessing the
literature are unknown
•« Cannot be replicated
 Scoping Reviews
Scoping Reviews
• "Scoping reviews have great utility for
synthesizing research evidence and are often
used to categorize or group existing literature inIn
a given field in terms of its nature, features, and
volume.”
volume."
• "preliminary assessment of potential size and
scope of available research literature. Aims to
identify nature and extent of research evidence
(usually including ongoing research).”- Grant
and Booth (2009)
 Scoping Reviews
Scoping Reviews vs
vs
Mapping Review
•« They are two different review types.

•« Scoping reviews - more topic based

•« Mapping reviews - more question based
 Scoping Reviews
Scoping Reviews
Scoping Reviews are best designed for:
•« "When a body of literature has not yet been comprehensively
complex, or heterogeneous nature
reviewed, or exhibits a large, complex,
not amenable to a more precise systematic review."
•« Label body of literature with relevance to time,
time, location (e.g.
(e.g. peer-reviewed or grey
country or context), source (e.g.
literature), and origin (e.g. healthcare discipline or academic
literature),
field)
•« Clarify working definitions and conceptual boundaries of a topic
or field
•« Identify
|dentify gaps in existing literature/research
Sources: MDJ
Sources: MDJ Peters
Peters et
et al.
al. (2015),
(2015), Levac
Levac et
et al.
al. (2010),
(2010), Godfrey
Godfrey C,
C, Khalil
Khalil H,
H, et
et al
al