EPIDEMIOLOGIC STUDY

DESIGN

Presenter: Dr. Omaima El-Gibaly

Sources
 Jekel’s epidemiology, biostatistics, preventive medicine and public
health
 An Epidemiologic Approach to Reproductive Health. WHO. CDC
 Student handout of the department

4rth Year Medical Students - Class 2019/20

Learning objectives
By the end of this lecture the student should be able to:
 Classify the different types of epidemiologic study designs
 Differentiate between descriptive and analytic epidemiology
 Identify the main types of bias in epidemiologic research
 Selection
 Information
 Confounding
 Identify factors to consider in evaluating causation of disease

4rth Year Medical Students - Class 2019/20

Context matters: Why do we teach you study
design?
 Uses of epidemiology at glance
▪ Population or community health assessment; e.g. Assess
the health of the population or community they serve and
Why do we identify health care needs to set policy and design
programs.
teach you ▪ Individual decisions to avoid certain risk factors or
Uses of
behaviors; e.g. epidemiologists noted a variety of benefits
study design? Epidemiology
and risks associated with different methods of birth control
▪ Completing the clinical picture; e.g. epidemiologists have
documented the course of HIV infection, from the initial
exposure to the development of a wide variety of clinical
syndromes that (AIDS).
▪ Search for causes of disease. Much of epidemiologic
research is devoted to a search for causes, factors which
influence one’s risk of disease.
 To Acquire Epidemiologic Reasoning Skills

Steps of Epidemiologic Reasoning

Why do we Why do we
1- Define events or clinical cases
2-Count these events
teach you teach you
3-Describe cases by person, place and time to
determine the population at riskepidemiologic
study design? 5-Calculate rate of occurrences for these event among
exposed and non-exposed
study design?
6-Compare to identify differences in rates of occurrence
7-Draw inferences about the events that define the
health problem and agent or agents that cause it
8. Propose control measures and assess control
measures.
Figure 1.
Epidemiologic study
designs and
increasing strength
of evidence

4rth Year Medical Students - Class 2019/20

One step back – study design defined
 Q1. What is meant by study design?
 Study design refers to the manner )‫ ( أسلوب‬in which groups of
persons with a particular characteristics are assembled )‫ (تجميع‬and
compared )‫ (للمقارنة‬to evaluate the association between a risk factor
or exposure and a disease or outcome.

4rth Year Medical Students - Class 2019/20

Factor that differentiate experimental from non-experimental
designs
 Epidemiologic studies may be categorized into two major classes,
experimental and non-experimental
 The difference between these two classes is based on whether
the researcher has any control )‫ (تحكم‬over exposure-
the agent that may potentially cause the disease-that is being
evaluated.

4rth Year Medical Students - Class 2019/20

Example: read the following abstract
Lumbar instrumented fusion and cognitive intervention
and exercises in patients with chronic low back pain and
disc degeneration.

OBJECTIVES:
To compare the effectiveness of lumbar instrumented fusion
with cognitive intervention and exercises in patients with
chronic low back pain and disc degeneration.

4rth Year Medical Students - Class 2019/20

Example continues
PATIENTS AND METHODS:
Sixty-four patients aged 25-60 years with low back pain lasting
longer than 1 year and evidence of disc degeneration at L4-L5
and/or L5-S1 were randomized to either lumbar fusion with
posterior transpedicular screws and postoperative
physiotherapy, or cognitive intervention and exercises.

4rth Year Medical Students - Class 2019/20

Example continues
PATIENTS AND METHODS:
The cognitive intervention consisted of a lecture to give the
patient an understanding that ordinary physical activity would
not harm the disc and a recommendation to use the back and
bend it. This was reinforced by three daily physical exercise
sessions for 3 weeks. The main outcome measure was the
Oswestry Disability Index.

4rth Year Medical Students - Class 2019/20

 Study
Q1. Use the data from Population:
the abstract to fill in the
information in the boxes.
Group Group
Q2. Would you classify A: B:
this study as
experimental or
observational?

Q3. Why? Outcome Outcome

4rth Year Medical Students - Class 2019/20
Example 2:Comparison of oral Azithromycin with oral
Doxycycline in the treatment of acne vulgaris.

BACKGROUND:
Acne vulgaris is a common inflammatory disorder of the
skin that involves pilosebaceous units. This study was
conducted to compare the efficacy of oral azithromycin with
oral doxycycline in the treatment of acne vulgaris in our
population.

4rth Year Medical Students - Class 2019/20

METHODS:

A study was conducted in the dermatology department of hospital X in

2011. All male and female patients of age (14-30 years) having
moderate acne vulgaris on face only were included in the study.
Patients were randomly divided into two groups. Patients in first group
were given azithromycin 500 mg daily before meal for 4 consecutive
days monthly for 3 months while patients in second group were given
doxycycline 100 mg daily after meals for 3 months. Patients were
followed up after taking treatment for three months to see the response
of each drug. Response was graded as excellent, good, moderate,
mild and no response.
4rth Year Medical Students - Class 2019/20
 Study
Q1. Use the data from Population:
the abstract to fill in the
information in the boxes.
Group Group
Q2. Would you classify A: B:
this study as
experimental or
observational?

Q3. Why? Outcome Outcome

4rth Year Medical Students - Class 2019/20
Effect of combining mosquito repellent and insecticide treated net
on malaria prevalence in Southern Ethiopia

 A study was conducted in 16 rural villages with 1,235

households in southern Ethiopia between September and
December of 2008. The villages were randomly assigned to
intervention (mosquito repellent and LLINs, eight villages) and
control (LLINs alone, eight villages) groups.
 Households in the intervention villages received mosquito
repellent (i.e., Buzz-Off petroleum jelly, essential oil blend)
applied every evening. Households in the control group
received long-lasting insecticidal nets (LLINs) only.
4rth Year Medical Students - Class 2019/20
Effect of combining mosquito repellent and insecticide treated net
on malaria prevalence in Southern Ethiopia

 3,078 individuals in intervention and 3,004 in control

group were enrolled into the study. The baseline
survey was followed by two follow-up surveys, at one
month interval. The primary outcome was detection
of Plasmodium falciparum, Plasmodium vivax, or
both parasites, through microscopic examination of
blood slides. Analysis was by intention to treat.
4rth Year Medical Students - Class 2019/20
 Study
Q1. Use the data from Population:
the abstract to fill in the
information in the boxes.
Group Group
Q2. Would you classify A: B:
this study as
experimental or
observational?

Q3. Why? Outcome Outcome

4rth Year Medical Students - Class 2019/20
Non-experimental study designs
 The non-experimental studies comprise the
majority of all epidemiologic studies. Non-
experimental studies are further divided into two
groups: Descriptive [who?, where?, and when?]
studies and analytic studies [why?].
 Descriptive studies generate hypothesis, analytic
studies test hypothesis.

4rth Year Medical Students - Class 2019/20

Observational studies: Descriptive Studies
examples
Descriptive studies
There are no comparison groups.
 Types of descriptive studies:

 1. Case report

 2. Case series

 3. Descriptive cross sectional

 4. Descriptive longitudinal

4rth Year Medical Students - Class 2019/20

 Descriptive study example
The prevalence of modern contraceptive use and its
associated socio-economic factors in Ghana
Abstract
O BJECTI VE:

This study estimated the prevalence of modern contraceptive use (MCU) and
the sociodemographic factors associated with MCU among sexually active
men in Ghana. A survey of a representative sample of 3,373 sexually active
men in Ghana. Question was: what were the personal characteristics of the
men who use modern contraceptive methods in term of their education, age,
knowledge about contraceptive methods, knowledge of service delivery points,
how many children they have, etc.
4rth Year Medical Students - Class 2019/20
Another descriptive study example
 Increasing trend of prevalence of infertility in Beijing.
 METHODS: This cross-sectional study was conducted in Beijing in 2012. A
stratified cluster sampling method was used to select 12 448 couples of
whom the female partners were born between 1955 and 1985. All subjects
were interviewed face to face. Infertility was defined as the failure to
achieve a clinical pregnancy after 12 months of regular unprotected sexual
intercourse. Results: The prevalence of infertility was 4.2% (3.1% as
primary and 1.1% as secondary infertility). An increase in the prevalence
of infertility according to the age of the female partner was found:
1.3% infertility for married females born in the 1950s and 11.4% for married
females born in the 1980s.

4rth Year Medical Students - Class 2019/20

Concepts
that need Important
concepts and
to be
definitions
explained
Exposure & Outcome
 Exposure includes potential risk factors in the etiology of
disease as well as interventions or therapies in the treatment of
disease. Exposure must antedate the occurrence of disease.
 Similarly, outcome or disease includes any health problem on
the continuum from complete health to death

4rth Year Medical Students - Class 2019/20

Measures of effect
 Analytic epidemiology focuses on measures of effect to
quantify the magnitude of the association between the
exposure and the health problem under study.
 These measures of effect are most commonly expressed as
ratios of or differences between rates, probabilities, and
proportions.

4rth Year Medical Students - Class 2019/20

Study Design
 Study design refers to the manner in which groups of persons with a
particular characteristics are assembled and compared to evaluate
the association between a risk factor or exposure and a disease or
outcome.
 The ideal study design from the scientific point of view is an
experiment.

4rth Year Medical Students - Class 2019/20

Experimental designs: Clinical trials and
randomized clinical trials
 In health research , experiments are most often used to
evaluate a new clinical therapy. This design is referred to as
a clinical trial.
 Clinical trial design may involve random assignment of
study subjects to either treatment or placebo groups.

4rth Year Medical Students - Class 2019/20

Experimental designs
 Under simple random assignment, each eligible study subject would
have an equal chance to receive the treatment or the placebo.
 Randomization minimizes the opportunity of bias and also provides
the theoretical for use of statistical models and hypothesis testing.

4rth Year Medical Students - Class 2019/20

Experimental designs
 The distinguishing feature of an experiment is that the scientist
determines the exposure status of each participant and then
observes the occurrence of a particular event.
 The investigator must also typically establish inclusion criteria
(eligibility criteria)

4rth Year Medical Students - Class 2019/20

Advantage & Limitations of experimental study designs
 Advantages of experimental  Limitation of experimental
studies /interventional studies

 If properly conducted, it provides  The design and implementation

strong evidence of the effect of may be complex, take long time
the intervention on health status and are expensive.
 Ensures temporal relationship  In some situations random
between exposure and outcome allocation of subjects to
 Ensures comparability of the intervention (treatment group)
comparison groups in RCT and and control group is not possible.
minimizes the possibility of bias.
 Ethical challenges
4rth Year Medical Students - Class 2019/20
Non-experimental designs
 From the ethical point of view, experimental studies of risk
factors are unacceptable. Thus, the majority of
epidemiologic studies are non-experimental.
 To promote non-biased data collection, standardized data
collection forms should be used for all groups and the
observer should not know to which group the study subject
belongs (blinded). If neither the participant nor the observer
knows (double blinded)

4rth Year Medical Students - Class 2019/20

Non-experimental designs
 Using non-experimental designs, we select persons on the basis of
either exposure status or disease status.
 Selection based on exposure is called cohort or follow-up design.
 In cohort designs the epidemiologist follows up a cohort to ascertain
the outcome of interest.

4rth Year Medical Students - Class 2019/20

Cohort studies defined
 A cohort study design based on enrollment of currently exposure
status, prospective cohort study.
 In contrast, historical records of exposure among a group of
persons may be used to select individuals for follow-up. This is used
mainly in occupational medicine and is called historical cohort
study

4rth Year Medical Students - Class 2019/20

 Start
People Free
from the disease follow-up measure outcome

Exposed Disease
Design of a case present/not
cohort study
Disease
Not present/not
Exposed
Advantages of cohort study design
 It permits the direct estimation of the rate of health problem and the relative
risk associated with the exposure of interest. It also provides estimates of the
attributable risk, thus indicating the absolute magnitude of the disease
attributable to a risk factor.
 Calculate disease occurrence i.e. incidence rate.
 It provides a clear temporal sequence of exposure and disease, because the
presence or absence of risk factor is recorded before disease occurrence.
(antecedent consequent relationship)
 It provides excellent opportunity to study rare exposures.
 It permits the assessment of multiple outcomes (risk and benefits) that may be
related to a specific measure. It permits discovering other diseases related to
the same risk factor.
 Less chance for bias4rth Year Medical Students - Class 2019/20
Disadvantages of cohort study design
 Not suitable for rare diseases where large numbers of subjects are required.
 Long-term follow up may be necessary when the latency period for the
outcome of interest is long.
 The most serious problem is attrition or loss of people from the sample during
the course of the study. This may affect the study .results There may be attrition
among the investigators.
 It is expensive and time consuming
 The exposure status, which is present at enrollment into the study, may change
during the conduct of the study. Over long period many environmental,
individual or intervention changes may occur and confuse the issue of
association.
4rth Year Medical Students - Class 2019/20
Case-control study designs
 When selection is based on current disease status
rather than exposure, the epidemiologist collects
information on exposure history from diseased
(cases) and non-diseased study subjects (controls).
 This design is called case-control, or retrospective,
or case-referent study.

4rth Year Medical Students - Class 2019/20

 Start Go back in time (past)

Cases Risk factor

Design of a present/not

case control Risk factor

Controls
study present/not
Case-control studies
 When studying etiology of disease, investigators who use a case-
control study design should recruit persons with newly diagnosed
disease (incident cases) rather than persons with history of disease
(prevalent cases)
 The study protocol must specify how persons with the incident
disease will be ascertained and how they will be distinguished from
persons with previous disease.

4rth Year Medical Students - Class 2019/20

Advantages of case control studies
 Most practical for studying health problems that occur
infrequently
 It is useful for studying health problems with a long latent
interval
 Inexpensive and quick
 Evaluate multiple risk factors (or protective factors) on the
health problems under study.
 It requires a smaller sample than other studies.
 There is no problem of attrition.
4rth Year Medical Students - Class 2019/20
Disadvantages of cases-control studies
 It is not possible to calculate disease occurrence
(incidence) so does not measure risk directly
 Improper selection of cases and control could give
wrong conclusions (selection bias).
 Exposure data are collected from records or by recall
after the disease has occurred. Records may be
incomplete, and recall of past events is subjected to
human error and the possibility of recall bias.
4rth Year Medical Students - Class 2019/20
Disadvantages of cases-control studies
 If the health problem is relatively common in the
population (i.e. >5-10%), the odds ratio is not a
reliable estimate of the relative risk.
 It could not determine the other possible health effects
of an exposure, i.e. it concerns with only one outcome.
 Temporality is serious problem, if it is not possible to
determine whether the risk led to the disease or vice
versa.
4rth Year Medical Students - Class 2019/20
Cross-sectional or prevalence studies

4rth Year Medical Students - Class 2019/20

Cross-sectional or prevalence studies
 The main difference between cross-sectional and
case-control and cohort studies is the exposure and
outcome are measured at the same point in time (no
temporality)

4rth Year Medical Students - Class 2019/20

Advantages of cross-sectional studies
 If it is based on a representative sample from the
population, thus making generalization of the results
from the sample to the general population.
 Take short time to be completed
 Inexpensive
 It's used for administrative purposes.
 Suitable for acute risk (single point risk) e.g. accidents
and acute infections
4rth Year Medical Students - Class 2019/20
Disadvantages of cross-sectional studies

 Both the outcome and exposure are ascertained

at the same time. The temporal relation between
the two is not clear (e.g. in case of arthritis,
limitation of activity and subsequent obesity or the
extra weight load on joints resulted in arthritis).
 The prevalence not incidence is the focus of the
study.
4rth Year Medical Students - Class 2019/20
 Basic Formulas for 2 x 2 Tables

2 x 2 Table for Case-Control or Cohort

Cumulative Incidence Data
Exposure Status
Outcome Exposed Unexposed
Present
a b m1
Absent c d m0
n1 n0 a+b+c+d=t

4rth Year Medical Students - Class 2019/20

Measures of Effect
The cumulative incidence relative risk (CIR) = [a/n1]/[b/n0]

For a randomized clinical trial or a cohort study is the ratio of the risk
of developing the disease in the exposed group relative to the risk
in the unexposed group.

4rth Year Medical Students - Class 2019/20

Measures of Effect

Odds Ratio = a*d / b*c

The odds ratio for a case-control study is the odds of the disease
in the exposed group relative to the odds of the disease in the
unexposed group.

4rth Year Medical Students - Class 2019/20

Definition of Bias [Systematic Error]

Bias is a systematic deviation between the true value

of a statistic and the value estimated by the
epidemiologic study.

4rth Year Medical Students - Class 2019/20

Categories of Bias
Generally, bias are typically grouped into one of 3 groups:
 Selection Bias ( in the comparison groups)
 Example: lost to follow-up in cohort studies and
hospital admission bias in case control studies
 Information bias ( in the data collection )
 Recall bias
 Confounding bias
4rth Year Medical Students - Class 2019/20
Assessment of various aspects of the association

1. Proper temporal sequence

2. Consistency of association
3. Strength of the association
4. Specificity of the association
5. Degree of exposure to a factor (dose-response
relationship)

4rth Year Medical Students - Class 2019/20

Causal Association
6. Biological plausibility of the association
7. Randomized trials provides the strongest evidence of causation.
Of the seven criteria only the first is absolute
requirement of causation

4rth Year Medical Students - Class 2019/20