study design (post)المحاضرة دي من أهم المحاضرات اللي بييجي منها في الامتحان.pptx

Study design
By
DR. Shaimaa yaihya
Assistant professor of public health and
community medicine

Objectives of epidemiological studies:
• Measurement of disease incidence andor prevalence and
outcome (prognosis).
• Identification of etiologicrisk factors of diseases.
• Evaluation of therapeutic or preventive interventions.
• Evaluation of new approaches to health care delivery.

Methods of research design:
There are two basic approaches for epidemiological studies:
• Descriptive studies
• Analytic studies

Descriptive studies include:
• Case report study
• Case series study
• Ecological study
• Cross-sectional study-survey

Analytic studies include:
– Observational studies
• Cross-sectional study
• Case control study
• Cohort study
– Intervention studies
• Clinical trials (therapeutic or preventive)
• Community trials (mostly preventive)

Descriptive studies
• Descriptive epidemiology searches for pattern of diseases and
health related phenomena by examining characteristics of
person ,place &time.

Types of descriptive studies
• Case report:
Is a detailed description of disease occurrence in
a single person .it may suggest a new hypothesis
about the cause or mechanisms of disease.

• Case series:
It is a report on characteristics of a group of subjects who all
have particular disease or condition. Common features among
the group may be suggest hypothesis of disease causation
Example: pneumocystis carinii pneumonia and mucosal
candidiasis in previously healthy homosexual men: evidence of a
new acquired immunodeficiency(AIDS).

• Ecological study:
These are data on groups not individuals.it is possible
to measure association between exposures and
outcomes in groups and hypothesis generated from
such observation are proposed for more analytical
studies.

• Surveys:
Assess the prevalence of disease and risk factors at the same
point of time.
It is a "snapshot "of the disease and their potential risk factors
simultaneously in a define population.

Analytic studies include
Observational studies
1-Cross sectional study
Definition: - Are studies which give an answer to the current situation at
a given point in time.
Aim:
Cross-sectional study is mainly a descriptive study but may be analytic if
the relationship between disease and risk factor is done during analysis
and distribution of disease in relation to time, place and person.

Uses and Examples:
• Screening of the disease.
• Survey studies.
• Censuses.

Subject and method:
• The whole population if confined may be taken or representative
sample if the target population is large. No controls are used in
cross sectional study.
STUDY
POPULATION
Exposed,
Diseased
Unexposed,
Diseased
Unexposed,
No Disease
Gather Data on Exposure and Disease
Exposed,
No Disease

• Advantages:
• Measure population sample characteristics
• Determine disease prevalence.
• Can study multiple risk factors and multiple diseases at the
same time.
• Disadvantages: Show associations but do not indicate causal
relationships

• Case-Control study (Retrospective study):
In a retrospective study, persons who diagnosed as having a disease
(cases) are compared with persons who do not have the disease
(controls); with regard to their past exposure to the possible risk factor of
interest.

Pastexposure Cases Controls
Exposed a b
Notexposed c d
Total a+c b+d
The proportion of exposed among cases (a/a+c) and among controls
(b/b+d) are calculated and compared.
Advantages:
1-The number of subjects can be small.
2-Results can be obtained relatively quickly.
3-Low cost.

Disadvantages:
Selection bias: error in selection of cases and controls.
Information bias: - unavailable or inaccurate records.
Recall bias:- inaccurate recall of events in the distant past
Confounding bias: - there are variables that increase or decrease the
association between the disease and the factor under study.
Yields: - only an approximation of relative risk (odds ratio) but cannot
measure incidence (no population base).
Uses:
Particularly useful for etiologic study of rare diseases (high exposure)

- Odds Ratio (OR)
In retrospective study we do not know the incidence in either the
exposed or the non-exposed group. Hence we cannot calculate the
RR directly. In such cases we can obtain a good estimate of RR by
calculating the odds ratio, provided that these 3 assumptions:
• The incidence of the disease is low.
• The cases are representative to all cases with regard to exposure.
• Controls are representative of the reference population with
regard to exposure.

Inaprospectivestudy:RR= a/a+b
c/c+d
Ifincidenceislow,then(a)contributelittleto(a+b)and(c)contributelittleto (c+d).
Soexcluding(a)and(c)fromthedenominators= a/b = ax d
c/d b xc
Thatisoddsratio (OR)is calculated bymultiplyingthediagonals ofthe2×2table.

Interpretation of the odds Ratio
 OR = 1 indicate that there is no association
between exposure and disease
development.
 OR > 1 indicates that the exposure is risky.
 OR < 1 indicate that the exposure is
protective
20

Cohort study
• A cohort study starts with 2 groups of people (cohort) free of the disease
under study, one group is exposed to a risk factor and the other one is not.
Both groups of cohort are followed over time to determine differences in
the rates of disease development (or rate of death from disease).
• A cohort is a group of persons who share a common attribute (feature or
experience) e.g. a birth cohort are person born in the same year in same
period of years.

• See the illustrated table to clarify the sequence of events in
the prospective study:
Currentexposure Disease No disease
Exposed a b
Notexposed c d
Total a+c b+d
The incidence of cases among exposed (a/a+b) and among those who
did not exposed (c/c+d) are calculated and compared.

Types of cohort study:-
– Concurrent prospective cohort study.
– Non- Concurrent prospective cohort study or historical study
or retrospective cohort study: - in this study, exposure has
been started in the past and documented in records. This type
of design is used to reduce the duration of the study. Exposure
is detected in the past and measurement of development or
non-development of disease is ascertained at the time when
study begins or followed longitudinally from time of start of
research for more time in the future.

Advantages:
• Lack of bias in ascertaining the exposure, since cohort is classified
as to the state of exposure before the disease develops.
• Permits calculation of incidence rates, thus both relative risk and
attributable risk can be calculated.
• Permit observation of development of additional diseases as by
product e.g effects of Chernobyl nuclear explosion.

Disadvantages:
• Require large number of subjects.
• Long follow up period.
• Attrition: - (loss of subjects from follow up) because of disinterest,
migration, or death from other causes or potential loss of staff and/or
funding.
• Potential change of exposure status of subjects over time.
• Changes in diagnostic criteria and methods over time with advances
in technology.
• Very costly.

Uses:
• Particularly useful to study outcome when exposure is rare,
but incidence among exposed is high.

Smoking Lung cancer Total
YES NO
Exposed 70 6930 7000
NO exposed 3 2997 3000
Total 73 9927 10000
Measuring of risk in epidemiological studies
Direct risk:
It is the incidence or the attack rate of the disease. It can be calculated for either:
Incidence of lung cancer among smokers: 70/7000 = 10 per 1000
Incidence of lung cancer among non-smokers: 3/3000 = 1 per 1000
Total group= 73/10000 =7.3 per 1000

Relative Risk (RR)
The relative risk measures the strength of association and is expressed as a ratio.
RR = 1 indicate that there is no association between exposure and disease
development.
RR > 1 indicates that the exposure is risky.
RR < 1 indicates that the exposure is protective.
incidence among exposed
RR= = 10/1
Incidence among non-exposed
(i.e.: lung cancer is 10 times more common among smokers than non-smokers).

Interpretation of the relative
risk
 If RR = 1 indicate that there is no
association between exposure and
disease development.
 A RR > 1 indicates that the exposure
is risky.
 RR < 1 indicates that the exposure is
protective.
30

Attributable Risk (AR)
Since some non-exposed individuals to certain factor develop the disease, attributable
risk (AR) is the magnitude or proportion of excess risk which can be attributed to
the exposure under study, and subsequently a measure of the magnitude of the
potential impact of its elimination.
 Disease incidence in exposed group (E) = incidence due to the exposure +
background incidence
 Disease incidence in non-exposed group (N) = incidence not due to the exposure
(background incidence)

Therefore; the excess incidence in the exposed group which is
attributable to the exposure = incidence in exposed group (E) -
incidence in non-exposed group (N)
AR for the exposed group= E - N
The proportion of total incidence in the exposed group
attributable to the exposure = (E) - (N) / (E) X 100
(AR = 10 – 1 / 10 X 100= 90 %)
(90% of the cases of lung cancer among smokers are attributed to
their habit of smoking).

Intervention studies
1. Clinical trial:
It is an experimental study that is designed to compare the
therapeutic benefit of two or more treatments (testing for new
treatment or vaccine).
• The purpose of clinical trial is to compare a new agent, drug or
vaccine with a traditional one with regard to it's:-
 Effectiveness.
 Safety (toxicity and side effects).
 Cost-effectiveness.

The clinical trial is a prospective study with two differences:-
Prospective study Clinicaltrial
1. Subjectsselectthemselves for
exposureornon-exposure to
the factor
2. No blinding is applied
1. Randomization:- The investigator
randomly determines who will be
exposed (treated) or not
2. Blinding is applied

Blinding should be applied whenever possible since lack of blinding
could influence perception of outcome and reduce confidence in
the study result.
Blinding may be:
Subject under investigation ………..single blind
Subject and data collector ………….double blind
Subject + investigator+ data analysts ………..triple blind
Blindness can be achieved by using a similar shape or colour and
taste…or using placebo.

Problems encountered include:
• Ethical issues
• Non participation, attrition or non-compliance.
• Correctly defining inclusion and exclusion
criteria.
• Sample size.
• Expenses.
• Long period of follow up.

Phases of Clinical Trials
• Phase 1: 15-30 people
– What dosage is safe?
– How should treatment be given?
– How does treatment affect the body?
• Phase 2: Less than 100 people
– Does treatment do what it is supposed to?
– How does treatment affect the body?

• Phase 3: From 100 to thousands of people
– Compare new treatment with current standard
• Phase 4: From hundreds to thousands of
people
– Usually takes place after drug is approved
– Used to further evaluate long-term safety and
effectiveness of new treatment.

• Community trial: are usually preventive in
nature (sometime termed preventive trial
where the intervention is preventive measure,
e.g vaccine and chemo-prophylactic
drugs .community trials have been used to
evaluated new vaccines.

Bias
Definition: Systematic error that appears in a
study and leads to distortion of its results.

Types of bias:-
1) Selection bias:- errors of selection of the sample or method of data
interpretation.
Examples:-
 Source of the sample, population or hospital based.
 Type of the sample, random or non-random.
 Size of the sample, may be too small.
 Method of sampling, e.g. selection of the sample from telephone directory.
 Inaccurate definition of cases and controls in case control study.
 Errors in selection of the study design.
 Errors of selection of statistical test.

2) Information (misclassification) bias:
• Recall bias: - bias in recall of the disease episodes or
information about it.
• Interviewer bias: - errors in questionnaire or non-
training of the interviewer.
• Observer bias: - either inter-observer or intra-observer
variation.

3) Confounding bias: confounding factors are some risk
factors in the selected population affecting the disease
under the study, and may increase or decrease the
association between the risk factor and the disease under
study.E.g. the relationship between parity and uterine
carcinoma there may be other distorting (confounding
factors in the studied population that affect the incidence
of uterine carcinoma, for example age, age of marriage
and family history.

How to overcome bias?
• Good selection of the sample and method of statistical analysis.
• Careful standardization of the procedures.
• Intensive training of observers or interviewers.
• Periodic check on work.
• Using two or more observers making independent observations.
• Overcome confounding variables

• Confounding bias can be overcome by:
1)In the study design either by matching or blinding.
2) In data analysis either by stratification or adjustment.
Matching:- means that controls should be identical as cases in all
variables except the variable under study. Potential variables are
related to:-
• Person: age, sex, race, marital status, education.
• Place, urban or rural residence.
• Time: year, season or may be time of the day.

• Matching may be for:
A) The whole group: i.e. the characteristics of the whole group are similar
to those of the study group, except for the variable under study.
b) Individuals: e.g. matched pairs (matched triplets) i.e. for each
individual case, there should be a comparable control individual.
Individual matching is difficult, particularly in large sample studies.
Individual control is usually selected from the immediate living circle of
the case, e.g. spouse, sibling, relative or neighbor. By nature, such
selection implies that the case and control are similar, but not identical.

• Stratification: refers to break down of data presentation by
stratum, category or group e.g. low versus medium versus high
income, urban versus rural, age 0-4, 5-9,………etc. such
stratification of each of the study and control groups helps to
estimate and compare the risk for each category or stratum.
• Adjustment: the use of statistical procedures to remove the
effect of differences of composition of the study and control
groups to yield standardized rates. Adjustment is commonly done
for age, but it may be necessary for other variables.

study design (post)المحاضرة دي من أهم المحاضرات اللي بييجي منها في الامتحان.pptx

study design (post)المحاضرة دي من أهم المحاضرات اللي بييجي منها في الامتحان.pptx

More Related Content

Similar to study design (post)المحاضرة دي من أهم المحاضرات اللي بييجي منها في الامتحان.pptx

More from amralmohammady27

Recently uploaded

study design (post)المحاضرة دي من أهم المحاضرات اللي بييجي منها في الامتحان.pptx