Epidemiology Lectures for UG

Gold Standard for Designs
for establishing causality
Randomize Clinical Trial
Dr Amita Kashyap
Professor P.S.M.

Learning Objectives
• To describe the important elements of
Randomized Trials.
• To define the purpose of Randomization and
of Masking (Blinding).
• To introduce design issues, including Stratified
Randomization, Crossovers, and Factorial
design.
• To illustrate the problems posed by
Noncompliance in randomized trials.

Purpose of Randomized trials
• For evaluating new drugs and other available
preventive or therapeutic measures
• For evaluating new diagnostic/ screening tests
• For assessing new programs for screening and
early detection
• To compare different approaches to
prevention, or new ways of organizing and
delivering health services.

Is it New!
• In 1537 Ambroise Paré; a surgeon,
treating soldiers
– The standard treatment for war wounds was
the application of boiling oil.
– The wounded were so numerous that, his oil
finished
– He instead applied a digestive made of yolks of
eggs, oil of roses and turpentine.
–Result !!!!!!!!!!!

A planned Historical trial
• 1747- A Scottish surgeon James Lind was
intrigued by the story of a sailor, who had
Scurvy; had been put ashore on an isolated
island. He subsisted on a diet of grasses and
then recovered from the scurvy.
• Lind conducted an experiment, he took 12 Pts
of scurvy on the board at Salisbury Ship.
• The cases were as similar as he could have the

• Two of these were given a quart of cider per day.
• Two others were given 25 gutts of elixir vitriol.
• Two others took two spoonfuls of vinegar.
• Two were put under a course of sea water.
• Two others had two oranges and one lemon/ day.
• Two others took the nutmeg.
• One of those who had taken oranges and lemons
was fit for duty after 6 days. The other was
appointed nurse to the rest of the sick.
• In 1795 (47 years later), the Admiralty made lemon
juice a required part of the standard diet of British
seamen and later changed this to lime juice.

Study Population
NEW TREATMENT CURRENT TREATMENT
IMPROVE DO NOT IMPROVE IMPROVE DO NOT IMPROVE
RANDOMLY ASSIGNED
Design of a Randomized Trial
We can always improve Results by omitting controls

• What intervention/s are to be compared
• Eligibility requirement for enrollment –
Inclusion & exclusion Criteria (for generalization)
• How many subjects in each Treatment Group
• Method and Technique of selection and
• Allocation to Treatment Groups
• What outcomes are clinically important and/
or are better predictors for desired outcomes
• How to measure outcomes – Objective criteria
(otherwise Blinding ) and same for both groups
Points to be addressed in RCT

Target Population (Population of Interest)
Inclusion Criteria
Method of Selection (Sampling Technique)
Defined Population
Target Sample (size)
Actual Sample
People to be included in study
Exclusion Criteria
Response Rate
Selection of Eligible Subjects

Comparison without using
Randomized Groups

1. Historical Control For Comparison –
• Suppose we have a therapy today that we
believe will be quite effective
• We would like to test it in a group of patients;
• For comparison, we will go back to the
records of patients with the same disease
who were treated before the new therapy
became available.
• This type of design seems inherently simple
and attractive BUT has its share of problems!

Issues in Historical Controls
1. Quality of data from medical records vs a very
meticulous system for data collection from the
patients currently being treated.
2. Change over calendar time – e.g. supportive
therapy, living conditions, nutrition, and lifestyles.
–It is useful when a disease is uniformly fatal and a new
drug becomes available, a decline in case-fatality/
morbidity that parallels use of the drug would
strongly support the conclusion that the new
drug is having an effect. e.g. discovery of insulin to
treat diabetes, of penicillin to treat serious infections.

2. Simultaneous Nonrandomized Controls
• Simultaneous (Concurrent) controls -has issues
of selecting controls!!.
Ex. – Anti emetic Trial on Sea
–Assign patients to Tt group by the day of the
month on which the patient is admitted to the
hospital: (The problem here is that the
assignment system is predictable)
Ex. – Anti coagulant Tt in Coronary Disease

Ex. - Selection Bias in Simultaneous Control
Results of a Trial of Bacillus Calmette-
Guérin Vaccination: II
TUBERCULOSIS
DEATHS
No. of Children Number Percent
Vaccinated 556 8 1.44
Un-Vaccinated 528 8 1.52
Results of a Trial of Bacillus Calmette-
Guérin Vaccination: I
TUBERCULOSIS
DEATHS
No. of Children Number Percent
Vaccinated 445 3 0.67
Un-Vaccinated 545 8 3.30

Steps to Randomly Assign
Study Subjects to Treatment
& Control Group

• Suppose subjects are to be assigned:
Therapy A and B -
–Every odd number to A and
–Every even number to B
• We close our eyes and put a finger anywhere
on Random table,
• And write down the number intersecting the
column and row – it is our starting point.
• Write down the direction we will move in
the table
Using Random Table – have written Protocol
Ctd…

00–04 05–09 10–14 15–19
Table of Random Numbers
Next patient assignment is not predictable;
It is important to spell out in writing - The approach for
Randomization, before randomization is actually done.
0-1 0-4
07
Starting
Point

What Randomization does..
• Randomization increases the likelihood
that the groups will be comparable not
only in terms of variables that we
recognize and can measure, but also in
terms of variables that we may not
recognize, may not be able to test and
measure now, with today’s knowledge
and technologies.

What Randomization does..
• Ensure Un-predictability of the assignment;
remove subjective biases of the investigators,
either overt or covert.
• Randomization is not a guarantee of
comparability because chance may play a
role in the process of random assignment.
• If there are enough participants, we hope that
randomization will increase the likelihood that
the groups will be comparable to each other in
regard to all factors that may affect prognosis.

Blinding (Masking)
We would not like the subjects to know which
group they are assigned to, especially when the
outcome is a subjective measure
• Single Blinding - using a placebo
• Double blinding - To mask Study Investigator or
data collectors in regard to which group a patient
is in – person putting therapy allocation cards in
envelop is different
• Triple blinding - When Statistician is also Blinded

Physicians’ Health Study: Side Effects
According to Treatment Group
Side Effect Aspirin Group
(%)
Placebo Group
(%)
P Value
GI symptoms
(except ulcer)
34.8 34.2 0.48
Upper GI tract
ulcers
1.5 1.3 0.08
Bleeding
problems
27.0 20.4 <0.00001

Effect of lack of comparability

An Example on How RCT is
planned and conducted

A Patient’s Profile
• An active 13-year old student of 8th Class complains of
increasing thirst, frequency of urination and fatigue for last
1 wk. Her pediatrician noted that-
• She is consuming >3lit. of liquid a day and urinating 8
times per day (has to wake up once or twice)
• Physical Examination - Her wt. dropped 2 kg since her last
visit 3 months back. Remainder of PE is unremarkable.
• Urine analysis – shows normal microscopy, the dipstick
examination is negative for blood, WBC and Bilirubin but is
4+ positive for glucose and 1+ positive for ketones
• Her blood test reveals 565 mg/ dl glucose and mildly
elevated blood urea
• Diagnosed Type I diabetes and trained in self insulin
therapy with dietary control and glucose monitoring

• During 1980s, physicians argued that strict control of
Blood Glucose levels reduce long term complications,
While others opined that physiologic abnormalities
associated with DM, are the cause of complications
• NOW– Should we give standard twice a day injection of
normal + long acting insulin to this girl OR
• Should we give NEW Intensive Therapy – 4 injections
of insulin per day with strict frequent blood glucose
monitoring – even though there are increased chance
of Hypoglycemic Coma with Intensive Therapy?
• To determine Safety and Efficacy of new therapies,
researchers must conduct Clinical Trials OR
• Learn from other’s research (Important is to be able to
Critically assess Clinical Trials )

Step 1. State your Research Question
• Is intensive therapy, including
frequent injections (defined- e.g. as 4
times a day) of insulin with frequent
monitoring (defined – before giving
insulin) of blood glucose levels,
“superior” to “standard therapy for
Diabetes Mellitus”

• What interventions are to be compared
• What outcomes are clinically important and/
or are better predictors for desired outcomes
• How to measure outcomes – Objective criteria
(otherwise Blinding ) and same for both groups
• How many subjects in each Treatment Group
• Eligibility requirement for enrollment –
Inclusion & exclusion Criteria (for generalization)
• Method and Technique of selection and
• Allocation to Treatment Groups
Points to be addressed in RCT

1. Percentage of patients surviving at a specified
time following treatment initiation
2. Patient’s ability to maintain an Active Lifestyle
3. Risk of experiencing one of the vascular events
4. Blood glucose level at specific points in time
5. Long term blood glucose level assessment –
HgbA1C
For assessing “superiority” of intensive therapy
in DM the possible outcomes could be -

Researcher in this Trial chose Diabetic Retinopathy
during 5 yr & HgbA1C as the Primary outcome bcz
1. Vascular disease is predominant factor for
morbidity & mortality from the disease
2. Vascular changes in one of the organs correlates
with vascular damage at other sites
3. The eye provides unique portal via which vascular
damage can be seen non-invasively
4. The measurement of progression of retinopathy
could be standardized & Objectively evaluated
without knowing the group of Patients.

Consider following questions while deciding
the Critical Level of Difference in outcome
• What would be Clinically important difference
for physician in treating such patients
• What difference would be meaningful for
patient who may suffer the consequences of
treatment
• What difference would justify use of more
effective treatment in spite of greater expense
or side effects

If  = 110 mmHg, 1 = 4 mmHg
95.4 % of times Pop. Mean BP
will be btw 102 to 118 mmHg
This is Confidence Interval
Standard Normal Curve
Area = 1 (for Probability estimate)
Mean = Mode = Median = 0
Standard Deviation = 1
Lets Learn Normal Distribution
To Know Clinically Important
Difference
Area beyond
 2 is
Significance Level
Standard Deviate Z = x -  /

How to interpret Z-Scores-
• A z-score < 0 represents an element < the mean.
• A z-score > 0 represents an element > the mean.
• A z-score = 0 represents an element = the mean.
• A z-score = 1 represents an element that is 1 SD greater
than the mean; a z-score = 2, 2 SDs greater than the
mean; etc.
• A z-score equal to -1 represents an element that is 1
standard deviation less than the mean and so on…
• If the Sample Size is large, - about 68% of the elements
have a z-score between -1 and + 1;
• About 95% will have a z-score between -2 and 2; and
• About 99% will have a z-score between -3 and 3

-1.96 +1.96
Acceptance area
Rejection Area
 = 0.025
Acceptance area
Rejection Area
 = 0.05
Single Tail Probability
Superiority Trial
Both Tail Probability
..
..
.. ..

Area btw   2  = 0.4772 + 0.4772 = 0.9544 = 95.4%

SE = /n
95% CI

n
+1.96
-1.96 
n
! Sample 5

Importance of Clinically Significant Difference

Step 2. State Hypothesis, How?
• Null hypothesis – there is no difference between the
Tt. Grps. regarding development of Diabetic Retinopathy
H : A = B
– A & B are proportions of persons developing Diabetic
Retinopathy in Standard & Intensive Therapy respectively
• Alternate Hypothesis – states that the two Tts. will
differ with respect to the development of Diabetic
Retinopathy
Ha : A  B
– A & B are proportions of persons developing Diabetic
Retinopathy in Standard & Intensive Therapy respectively

Alternate
Hypothesis
Null Hypothesis

Step 3. Determine sample Size –
Remember..
• Sample is determined in relation to the Primary
Research Outcome – (development of
retinopathy)
– Clinically important Magnitude of Difference in the
Primary Outcome
• Accepted level of Type I (Significance Level)
• Accepted level of Type II errors (Power of the
Study)

Trial
Results
Truth
Treatments Differ Treatments do not
Differ
Treatments
Differ
A
Correct
(True-Positive)
B
Type I -  Error
(False-Positive)
Treatments
do not Differ
C
Type II -  Error
(False-Negative)
D
Correct
(True-Negative)
Trial Results vs “Truth”

There are four possibilities –
A. Trial Results are same as the “Truth” in population
1. The Trial truly finds difference in treatment outcome
(Cell A in the Table)
2. The trial truly agree that there is no difference in
treatment outcome (Cell D in the Table)
A. Trial results differ to the “Truth” in population
3. The Trial wrongly finds “difference in outcome while
there is no difference” (Cell B in the Table)
– Type I or  error
4. The Trial wrongly finds “no difference in outcome while
in actuality there is a difference” (Cell C in the Table)
– Type II or  error

Power = 1- 
Population 1 Population 2
FP
FN
Cut off Bl Sugar Level = 107.5
If reduced to 104 ?
104

Factors that Affect Sample Size
Factor Effect on
Sample Size
Decrease acceptable Type I error Increase
Decrease acceptable Type II error Increase
Decrease Variability of Outcome Variable Decrease
Decrease expected difference in outcome
between the Groups
Increase

Sample Size Determination
Challenges
• Trade off between:-
–Size and Cost and Ethical issues
–Significance level and Power of study
• Human Error
• Coding in case of multicentric study
• Generalization

Step 4. Randomize
Stratified Randomization

Enrollment Criteria for Subjects
Variable Patient Characteristics
Age 13 to 39 yrs
Diagnosis Insulin dependent diabetes mellitus
Duration of
Disease
1 – 5 yrs
Past Medical
Condition
Absence of Hypertension,
Hypercholesterolemia, and Diabetic
Complications – No retinopathy
Urine Albumin
excretion
Less than 40 mg per 24 Hrs
Glycosylated
Hemoglobin (%)
Between 6-10

Baseline Characteristics of
Enrolled Subjects
Characteristic Standard Therapy
n=378
Intensive Therapy
n= 348
Age yrs 268 27 7
% Adolescents 19 16
Male 54 49
Duration of IDDM 2.61.4 2.61.4
Dose of Insulin
U/kg body wt/ day
0.620.26 0.620.25
Mean Bl Glucose
(mg/dl)
22980 23486
Glycosylated Hb% 8.81.7 8.81.6
Body wt (% of Ideal) 10314 10313

Design of a Planned Crossover trial

Design of a planned crossover trial
Each patient serves as his own control, holding constant the
variation between individuals in many characteristics that could
potentially affect a comparison of the effectiveness of two agents

Caution to be taken for Crossover Design
• Carryover effect: - no residual carryover from first
therapy. There must be enough of a “washout
period” to be sure that none of therapy A, or its
effects, remains before starting therapy B.
• Order in which the therapies are given may elicit
psychological responses. Patients may react
differently to the first therapy as a result of the
enthusiasm that is often accorded a new study.
• Finally, the planned crossover design is clearly not
possible if the new therapy is surgical or if the new
therapy cures the disease.

Noncompliance
• Dropouts from the study
– built checks on potential noncompliance into the
study.
• Drop-ins
• The net effect of noncompliance on the study
results - reduce any observed differences

Factorial Design
Economically use the same study population for testing
Two drugs – if the anticipated outcomes for the two drugs
are different, and their modes of action are independent.
Treatment A
+ -
Treatment B + Both A and B (cell a) B only (cell b)
- A only (cell c) Neither A nor B (cell d)
Evaluate the effects of treatment A by comparing the
results in cells a + c to the results in cells b + d
AND the results for treatment B could be evaluated by
comparing the effects in cells a + b to those in cells c + d

(cell a + cell b)
(cell c + cell d)
Factorial design-
(A) The effects of treatment A (Yellow cells) versus no treatment A.
(B) The effects of treatment B (Gray cells) versus no treatment B.

Study Population 22,071
Aspirin 11,037 Placebo 11,034
Carotene
5,517
Placebo
5,520
Carotene
5,520
Placebo
5,514
RANDOMLY ASSIGNED
Factorial design used in a study
of aspirin and beta carotene.
RANDOMLY ASSIGNED RANDOMLY ASSIGNED

Factorial design
(A) The effects of aspirin (Yellow cells) versus no aspirin.
(B) The effects of beta carotene (purple cells) versus no beta carotene.

Step 5. Analyze RCT Results
• Loss of Patients to Follow-up is likely
(“Intention to Treat Analysis”)
Standard
Therapy
Eligible
Subjects
Intensive
Therapy
Randomize
Received
Other Tt
Received
No Other Tt
Received
Other Tt
Received
No Other Tt
No
Retinopathy
Retinopathy
No
Retinopathy
Retinopathy
Time
Onset of Study
Direction of Inquiry

A well-reported Clinical-Trial has enough
Primary Data to enable the reader :-
1. To compare the main outcome measure
between the two treatment groups
2. To perform basic Statistical Tests to
determine whether it is justified to rule out
chance variation as a cause of difference
between the compared groups

Useful Methods of Comparisons
A. The Comparison of Risk in the groups –
risk of developing retinopathy
B. The Comparison of time of an event i.e.
Survival Analysis (e.g. Survival without
development of Retinopathy here)
C. The Comparison of two Means –e.g.
Mean Glucose Levels

A. The Comparison of Risk in the Groups
1. Percentage Rate Reduction (Attributable
Risk) = IR (standard) - IR (Intensive)
• The incidence rate (IR) of Retinopathy was
4.7 per 100 person years for Standard Therapy
and 1.2 per 100 person years for Intensive
Therapy
IR (standard)
X100
= 4.7-1.2 /4.7 X100 = 74%
74% retinopathy that occurred in Standard Therapy
could have been avoided if Intensive Therapy was
used Continued………

2. Confidence Interval – to determine Precision
of any ‘Point Estimate’, CI is a useful method.
The CI calculated in this Trial was 60-83%
(since this does not include ‘0’, the
difference in Percentage Rate Reduction is
Significant at an alpha level 0.05)
this means, if the Trial were repeated often,
95% of the times the Percentage Rate
Reduction would fall between 60 and 83%
Continued………

3. Relative Risk (Rate Ratio) -
Rate Ratio = Incidence Rate (Intensive)
Incidence Rate (Standard)
1.2/ 100 Patient Years
4.7/ 100 Patient Years
= = 0.26
That means, the Rate of developing Retinopathy is
About 1/4th of that of the Standard Therapy Group
95% CI was calculate as 0.17 – 0.40 (since it does not include
‘1’ difference in rate of developing retinopathy between two
Therapies is significant at an alpha level of 0.05)

• Graphic depiction of Time to Event Data
• Provide information on Rapidity with which
Event occurs
• Make use of data from patients who are
Followed for Varying Length of Time
• Can estimate Median Survival Time (time to
retinopathy development) and
• The percentage of without retinopathy at any
time along the curve
B. The Comparison of time of an
event i.e. Survival Analysis

• At the end of 9 yrs of follow-up 14% patients in
Intensive Therapy and 55% in Standard Therapy
Group Developed Retinopathy
• Relative Risk of Retinopathy in intensive to
standard therapy at 5 yrs is :-
6/15 = 0.4 (60% less than that in Standard therapy)
• The Median Time (point at which half of the
group develop retinopathy) – it was 8.5 yrs for
Standard Therapy while for the Intensive
Therapy Group has not been reached after 9 yrs
of follow-up

C. The Comparison of two Means
• Patients blood glucose levels has been tested
at different interval
• To test if difference in Bl. Glu. Levels is
Significantly differently managed bt two
therapies ‘t’ test is used
X1 – X2
Sp1/n1+1/n2
t =
X1 and X2 are mean bl.glu
N1 and n2 are sample size
Sp is an estimate of
pooled variance of the two means

Conclusion
• The randomized trial is generally considered
the gold standard of study designs.
• Many of the components of the randomized
trial that are designed to shield the study from
any preconceptions and biases of the
investigator and from other biases that might
inadvertently be introduced.

STUDY QUESTIONS AND APPROPRIATE DESIGNS
Type of Question Appropriate Study Design
Burden of illness
- Prevalence Cross Sectional Survey
- Incidence Longitudinal survey, cohort
Treatment Efficacy Randomized Controlled study
Diagnostic Test Evaluation Randomized Controlled study
Cost Effectiveness Randomized Controlled study
Establishing Association, Case Control Study,
Identifying Risk & Prognosis Cohort study,
and causation RCT

Epidemiology Lectures for UG

More Related Content

What's hot (20)

Similar to Epidemiology Lectures for UG (20)

More from amitakashyap1 (20)

Recently uploaded (20)

Epidemiology Lectures for UG