7/28/2019 Observational Study Design
1/45
OBSERVATIONAL STUDY
DESIGN
Oleh :
Dr. Nurjannah, MPH
7/28/2019 Observational Study Design
2/45
Objectives
1. Describe each study design
2. State the advantages and disadvantagesof each study design
3. Understand the difference betweendescriptive and analytic studies
4. To be able to aply different study design
to the same research question5. Recognize each study design in medical
literature
7/28/2019 Observational Study Design
3/45
Type of Study
Epidemiological
Study
Observational
Study
Interventional
Study
Grup Individual
Descriptive or
Analitycal Descriptive Analytical
Ecological
Study
Cross
Sectional study
Cohort
Study
Case Control
Study
7/28/2019 Observational Study Design
4/45
Type of Studies
Observational Analytic Studies
Cross Sectional studies
Case Control studies
Cohrt Studies
7/28/2019 Observational Study Design
5/45
Cross Sectional
The simple form an observational
Both exposures and outcome are
measured at the same time
A snapshot of the health (or other)
experiences of the population at that
particular point in time
7/28/2019 Observational Study Design
6/45
Cross Sectional
Also called a survey (collection ofopinions) or pool (a questionnaireadministered to a sample of a people,
often about a single issue) Designed to determine what is
happening ? right now
Examines a characteristic or set ofcharacteristic in a set of subjects at onepoint in time ( prevalence)
7/28/2019 Observational Study Design
7/45
Advantages
Quick and inexpensiveFirst step for a cohort study
Able to yield prevalence estimates
Researcher has control over selection ofsubjects
Researcher a control over the mesurmentsused
Can study several factors or outcomes at theone time
Often provides early clues for hypothesis
generation & later more definitive study
7/28/2019 Observational Study Design
8/45
Disadvantages
Do not establish the true temporal sequence ofevents
They can only ascertain whether exposure is
associated with a given outcome; they cannotdetermine whether the exposure caused the
outcome
Potenstial bias in measuring exposurePotential sampling bias and/or survivor bias
They are not feasible for rare condition
Does not yield incidence or true relative risk
7/28/2019 Observational Study Design
9/45
Example Breast Feeding and
Obesity cross sectional study
Abstract :
Objective : to assess the impact of breastfeeding on the risk of obesity and risk of
being overweight in children attendingschool in Calima. In a sample of 80childen, early feeding, diet andlifestylefactors were assessedusingresponse to a questionnaire completed byparents
40
7/28/2019 Observational Study Design
10/45
Sample : 80 children aged 9 & 10, all Filipinos
Main Outcome Measure
Height and weight measurementBMI calculation : kg / height (m)2
Being overweight was defined as having a
body mass index above :male : 18.86 for 9 y/o
19.61 for 10 y/o
7/28/2019 Observational Study Design
11/45
Female : 19.2 for 9 y/o
20.2 for 10 y/o (NCHS) BMI for age
Questionnaire: Was your children breast
fed?, How long was your child exclusively
breast fed?
Other question : number of sibling, parents
ages and education, child health (LBW),
early feeding, frequency of eating selected
food
7/28/2019 Observational Study Design
12/45
Statistical Analyses
The prevalence of overweight and obese
children were calculated according to the
duration of breast feeding
X2 test (chi square) were used to compare
breastfed and nonbreastfed children and
their association to an obese / overweight
child.
7/28/2019 Observational Study Design
13/45
Result
The prevalences of obesity in children who
had never been breastfed was 4.5 % as
compared with 2.8% in breast fed children.
A clear dose response effect was identifiedfor the duration of breast feeding on the
prevalence of obesity
7/28/2019 Observational Study Design
14/45
The prevalence was 3,8 % for 2 month of
exclusive breast feeding, 2.3 % for 3-5 month,1.7 % for 6-12 month, and 0.8% for more than12 month
Breast feeding as a significant protective factor
against obesity development Similar relations were found with the prevalence
of being overweight
The protective effect of breast feeding was notattributable to differences in social class orlifestyle
7/28/2019 Observational Study Design
15/45
Conclusion
Prolonged breast feeding may helpdecrease the prevalence of obesity inchildhood
Since obese children have a high risk ofbecoming obese adults, such preventive
measures may eventually result in areductin in the prevalence ofcardiovascular diseases related to obesity
7/28/2019 Observational Study Design
16/45
Case Control Study
A case-control study is distinguished bythe fact that subject are selected on thebasis of whether or not they have the
disease (or other outcom) of interestCase (person/group with a given disease)are then compared to control (person/group without the given disease) in termsof their history of exposure to ahypothesized causal factor
7/28/2019 Observational Study Design
17/45
Case Control Study
Ascertain of exposure of background of
the two group and compare the proportion
Best suited for study of diseases where
medical care usually sought (hip fracture,
cancer) bacauses it make to easier to
identify cases
7/28/2019 Observational Study Design
18/45
Stucture
Exposure Outcome Design
Exposed a
Not exposed b
Exposed c
Not exposed d
Population
Case (peoplewith desease)
Control (people
without desease)
41
7/28/2019 Observational Study Design
19/45
Selection of cases
Ideally, investigator identifies & enrolls allincident cases in a defined population in aspcified time period
Selected cases from registries or hospital,clinics
Whenn all incident cases in population are
included, the study is a representative;otherwise there is potential for bias (e.g.referal bias
7/28/2019 Observational Study Design
20/45
Selection of Controls
Critical that exposure in the control is
representative of the exposure in the
population
Ideal controls would have same/similar
characteristics as the cases
Matching case control
7/28/2019 Observational Study Design
21/45
Population-Based Controls
The best control group is a random
sample of individuals from same source
population (as the cases) who have not
the developed the disease
Population-based controls are the best
way ensure that the distribution of
exposure among the controls isrepresentative
7/28/2019 Observational Study Design
22/45
Hospitals Control
Hospital control are the most frequently usedsource
Hospital control may not be representative of
exposure rates in the target populationThe use of other ill as person control will
provide a valid result only if their illness is
unralated the exposure in question
7/28/2019 Observational Study Design
23/45
Benefit of Using Hospital Controls
Convenient
Cheap
NumerusAvoids non-response
When a population-based case registry is
not available, hospital control betterrepresent the subpopulation from wich the
cases arose
7/28/2019 Observational Study Design
24/45
Other Controls
Neighborhood controls are somewhat
matched on SES & environmental
exposure but may overmatch & be
expensive
Friends & relatives also cause problems
with overmatching with habits,
environment and ccupation & aregenerally a poor choice for controls
7/28/2019 Observational Study Design
25/45
Use of Multiple Controls
Case to control ratio used is usually 1:1; if
large number and cost is the same for
both group
If a study has a small number of cases,
increasing the number of controls increass
power of study
7/28/2019 Observational Study Design
26/45
Advantages
Relatively shorter time and inexpensiveGood for desease with long latency
Valuable for studying rare or uncommon
conditions
Multiple etiologic factors evaluated for single
deseases
Well suited for studying disease with longinduction period
A relatively small number of subjects are
required
7/28/2019 Observational Study Design
27/45
Disadvantages
Inefficient if the exposure is very rare
They are limited to one outcome variable
Incidence rates or absolute risk estimatescannot be directly derived from them
Do not establish the temporal sequence of
event; in some situations the temporal
relationship between exposure and
desease may thus be difficult to establish
42
7/28/2019 Observational Study Design
28/45
Disadvantages
Prone to bias (selection of cases andcontrols recall, misclassification)
Difficult to determine representativeness of
case and controlsUnless study is population based cantmeasure incidence of desease
Bad for rare exposure (despite a largenumber of case, may still end up with fewexpossed cases)
7/28/2019 Observational Study Design
29/45
Cohort Studies
An observational research design which
begin when a groupof people (a cohort)
initially free of desease, are classified
according to a given exposure, and thenfollowed up over time
7/28/2019 Observational Study Design
30/45
Cohort Stucture
Exposure Outcome Design
Disease a
No disease b
Disease c
Not disease d
Population
(People withoutdisease)
Exposed
Not
exposed
7/28/2019 Observational Study Design
31/45
Type of Cohort
1. Prospective
Fixed
Open or Dynamic
2. Retrospective
7/28/2019 Observational Study Design
32/45
Prospective Cohort
Fixed Cohort Study
When the exposure groups in a cohort
study are defined at the onset of the study
without movement of individuals between
exposure groups, the exposure group are
reffered to as fixed cohorts (occupational.
War)
7/28/2019 Observational Study Design
33/45
Structure
Prospective fixed cohort study
Exposure Disease
?
?
7/28/2019 Observational Study Design
34/45
Prospective Cohort
Open or Dynamic Cohort Study
The other type of prospective cohort study
is the open or dynamic cohort study where
in individuals can be unexpose at one time
and unexposed at another time. The
person time analysis can take this into
account in calculating incidence densities
7/28/2019 Observational Study Design
35/45
Structure
Prospective dynamic (open) cohort study
Exposure Disease
?
?
7/28/2019 Observational Study Design
36/45
Retrospective Cohort
The point of initial exposure occurred
some time in the past and the experience
of the population is followed up to the
present time
43
R t ti C h t St t
7/28/2019 Observational Study Design
37/45
Retrospective Cohort Structure
Prospective fixed cohort study
Exposure Disease
?
?= Present Basic on which group are selected at
= Absent beginning of study
? = To detemined
7/28/2019 Observational Study Design
38/45
Advantages
Provide strong information about thecausation of disease
Provide the measurement of the risk of
developing diseaseExposure can be measured without bias,because at the same time the outcomes
sre known; known confounder can bemeasured (especially in a prospectivestudy)
7/28/2019 Observational Study Design
39/45
Advantages
Can be used to examine multipleoutcomes
A range of factor that may influence the
outcome (e.g. smoking) can be measuredSuitable for examining the effect of rareexposures because this group can be
preferentially recruited at the baselineAllows the incidence of the disease to beestablished
Di d
7/28/2019 Observational Study Design
40/45
Disadvantages
Costly and time consumingMay be difficult to accurately define andmeasure exposure in some circumstances
Losses the follow up are not uncommon andmay introduce serious bias
Information bias may very in its effect over thecourse of data collection due to sometime
subtle drifting of the quality of data collectionUse of the retrospective design is only possibleif historical data of adequate quality are
available
Rancangan Penelitian
7/28/2019 Observational Study Design
41/45
Rancangan Penelitian
Retrospective
p
o
p
u
l
a
s
i
Efek (+)Retrospektif
Efek (-)Retrospektif
F. Resiko (+)
F. Resiko (-)
F. Resiko (-)
F. Resiko (+)
7/28/2019 Observational Study Design
42/45
Rancangan Penelitian Cross Sectional
Populasi /Sampel
Faktor Resiko (-)Faktor Resiko (+)
Efek (-)Efek (+)Efek (-)Efek (+)
7/28/2019 Observational Study Design
43/45
TERIMA KASIH
7/28/2019 Observational Study Design
44/45
7/28/2019 Observational Study Design
45/45