Intervention studies
Example
0
0.5
1
1.5
2
2.5
3
3.5
%withpneumoniaover2yrs
No vaccination Vaccination
Basic features of intervention studies
 An intervention study involves an intentional
change in some aspect of environment or status
of the subjects of the investigation.
 Intervention studies differ from observational
studies in that the researcher seeks to compare
two or more groups that differ as a result of
deliberate action rather than natural or found
variation.
Observational vs. Intervention studies
What is the difference between observation and
intervention studies?
• Both can be used to compare differences between
groups
• Intervention studies involve exactly what they say;
there is an intervention in one of the groups (at
least). So, the research team does intervene
rather than just observe the study participants
Main principle of intervention studies
 A randomised controlled trial is a type of
experiment.
 In simple form, the participants are distributed on a
strictly random basis into two groups that do not differ
in number or quality: the control and intervention
groups.
 The control group receives no intervention or inactive
or PLACEBO intervention or the “routine” care.
 The intervention group receives one single uniform
intervention that we are testing.
 Groups should be equal apart from variable under test
 If the condition of the two groups differs at the end
of the trial, then it can only be the result of the
intervention.
Everything except the
intervention is (hoped to be)
the same in the two groupsDefined study
sample
Intervention
group
Control
group
Measure
outcome
Measure
outcome
Randomisation to two groups
Key issues in RCTs
 Careful entry criteria
 Assessment (Pre- & Post-intervention)
 Randomisation
 Allocation Concealment
 Blinding (Masking)
 Contamination
 Analysis – ITT
 Interpretation
Entry criteria
 Aim in any research is to draw conclusions
about a population from a sample
 Representative sample
BUT
 In RCT participants often have a disease
 Participants must be able and willing to take
part
 Compromise: exclusion criteria vs. ideal
sample
Assessment
OUTCOME
 Careful selection of outcome
 Careful measurement of outcome
◦ Reliability and responsiveness
◦ Baseline assessments may inform participants
TIMING
 Sufficiently long after intervention to allow it to
work
 Not so far after intervention that effect is lost
 Multiple follow ups
◦ Variation / change over time (short-term vs long-
term)
Randomisation
 … is allocation of the units of analysis to the
different experimental groups or conditions
according to chance, such that each unit has
an equal probability of selection into
each group
 Most powerful way of ensuring
characteristics not systematically allocated
to a particular group
 Can randomise in groups (clusters)
The aim of randomisation is to…
create groups that are comparable with
respect to known or unknown
confounding factors
There are two steps in the process
1. Generating an unpredictable allocation
sequence e.g. tossing a coin, using a
computer random number generator
2. Concealing the allocation sequence from
the investigators
Not always possible
Allocation concealment
 … is making sure that neither investigator
nor patient can predict group assignment
Adequate methods
Off-site randomisation e.g. needing a phone
call
Sequentially numbered, sealed, opaque
envelopes
Blinding
 If participants or researchers know whether
participant is receiving intervention then
there is risk of:
◦ Measurement error
◦ Different investigations & care study group etc.
◦ Acceptability bias (Researchers influence participants
behaviour)
 Different “levels” of blinding: can blind
participants, researchers and/or statisticians
or none
“Levels” of blinding in interventions
Single blinding: If participants cannot be blinded
to group allocation, as in behavioural trials (e.g.
dietary advice), allocation and assessment of the
outcomes should be concealed from researchers
Double blinding: neither researcher nor
participant knows which of the study groups the
participant has been allocated to (e.g. high quality
drug trials)
Triple blinding: as for the double blinding but
additionally the statistician is blinded
Open RCT: no blinding
Allocation concealment vs. Blinding:
is there any difference?
 Allocation concealment: neither investigator
nor participant can predict group assignment
 Blinding: neither investigator nor participant
knows which group the participant is assigned
to (double blinding); single blinding if only
investigator (or, in some cases, participant) is
blinded
 How can you achieve this?
◦ Allocation concealment: off-site randomisation
◦ Blinding is not always feasible
Contamination
 … occurs when the behaviour of the
participants in one group is influenced by
what happens in another group
 It is particularly a problem in evaluation of
health education and community
interventions.
Analysis
 Simple - randomisation reduces risk of
confounding
 Compare groups at baseline to see if they are
broadly similar (no need for statistical tests)
 Compare groups at follow-up to estimate effect
(statistical tests)
 Must consider all, not only those completed the
RCT
 Intention to Treat (ITT) analysis
◦ Analyse all according to baseline allocation
◦ Even if participants didn‘t receive intervention
◦ Minimise loss to follow-up or substitute data
Interpretation of findings
 In any controlled trial, there is a
comparison group that controls for (takes
account of) background risk and extraneous
variation in the outcome of interest and
allows the effect of the intervention to be
quantified.This can be done via:
 Incidence risk/rate: ratio or difference
between intervention and control groups
 Changes in prevalence: ratio or
difference between the intervention and
control groups
Challenges in RCTs
General points
 Complex to set up and administer
 Must account for many variables
 Some require long follow up
 Difficult with long-term interventions
 Difficult for uncommon health gains or
uncommon adverse effects
 Potential loss to follow up
Generalisability (External validity)
 Extent to which findings applicable to wider
population
 High degree of control within RCT may
restrict generalisability
 Specific population may not represent
general population (e.g. willing volunteers)
Advantages of RCTs
 Experimental: groups treated similarly
except intervention
 Randomisation: characteristics
similarly distributed
 Blinding
patients
investigators
statisticians
 ITT analysis Prevents attrition bias
Tells us that
difference at the
end is only due to
intervention
Gold-standard epidemiological study design to
assess effectiveness of interventions
Special types of RCT
 Factorial design
◦ Tests cumulative effect of two or more
interventions
 Cross-over design
◦ All participants receive all treatments & act as own
controls
 Community-basedTrials
◦ Intervention assigned to groups (randomise schools
to treatment and control groups)
 Complex interventions
◦ Many components, difficult to disentangle their
effects
Factorial design
Tests cumulative effect of two or more interventions
Group A Group B
Intervention X
Intervention Y
Group C Group D
Intervention Y
Follow up AssessmentFollow up Assessment
Cross-over trial
All participants receive all treatments & act as own
controls
Intervention
Washoutperiod
Follow up 2
Group A
Group B
Intervention
Baseline Follow up 1
Complex intervention
“comprises a number of separate elements
which seem essential to the proper
functioning of the intervention although
the active ingredient of the intervention
that is effective is difficult to specify”
MRC definition (2000)
Intervention studies: less powerful options
 Are there any other studies (apart from RCTs) that
may be classified as interventions?
 Quasi-experimental trial
Non-random allocation to intervention and control groups
Patient or physician preference or other (convenient) way
Concerns about internal validity (non-comparable groups)
 Intervention (non-controlled) studies
Single group with assessments before and after
intervention
V weak design as can’t attribute to intervention
Summary
 Intervention studies are experiments
 RCTs are the gold-standard design for
assessing the effectiveness of interventions
 Simple concept but many key features - need
to carry out properly
 Randomisation is the most important, but
others (generalisability/entry criteria,
assessment, blinding, allocation concealment,
analysis-ITT) also matter
 Not always applicable – PH interventions are
usually more complex than a clear-cut simple
experiment
Types of comparisons in different types of studies
Study design Type of comparison
Ecological
studies
Comparing disease frequency between
populations
Cross-sectional
studies
Comparing disease frequency between persons
with and without characteristic of interest
Cohort studies Comparing disease incidence between exposed
and unexposed persons
Case-control
studies
Comparing frequency of (past) exposure
between cases and healthy controls
Interventional
studies
Comparing incidence of events in persons
exposed to the intervention of interest and in
control group
hierarchy of major study designs
systematic review of RCTs
RCT
cohort
case control
interventional
observational
validity
ecological
Epidemio
logical
studies
Observa
tional
studies
Interventi
on
studies
Data from
groups
Data from
individuals
DESCRIP
TIVE (a)
ANALYTIC
ECOLOGI
CAL
STUDY (b)
ANALYTIC
DESCRIP
TIVE
CROSS-
SECTIONAL
STUDY (c)
CROSS-
SECTIONAL
STUDY (c)
COHORT
STUDY (d)
CASE-
CONTROL
STUDY (e)
Data fro
groups
Data from
individuals
COMMUNI
TY TRIAL
(f)
CLINICAL
TRIAL, INDI
VIDUAL
TRIAL (g)
Applications of different observational and analytical study designs
Ecological Cross
sectional
Case
control
Cohort
Investigation of rare disease ++++ - +++++ Investigation
of rare exposures ++ - - +++++
Examining multiple outcomes + ++ - +++++
Studying multiple exposures ++ ++ ++++ +++
Measurement of time
relationships between expo and
outcome
+ - + +++++
Direct measurement of incidence - - + +++++
Investigation of long latent period - - +++ +++