Desain penelitian

Siswanto
 DESAIN PENELITIAN

PENELITIAN

STUDI OBSERVASI
(NO PENGENDALIAN studi intervensi
SAMBUNGAN) (PEMERIKSA MENENTUKAN WHO
TERPAPAR ATAU TIDAK TERKENA)

NO PERBANDINGAN PERBANDINGA
GROUP N GROUP

DESKRIPTIF ANALITIK

REVIEW PENGAWASAN SURVEY STUDI CROS STUDI STUDI

CASE SEC SIONAL KASUS KOHOR
CON trol
 SKEMA PENELITIAN SIKLUS

STUDI
DESKRIPTIF
ANALISIS HASIL,
SARANKAN LEBIH MODEL
LANJUT-DESKRIPTIF BANGUNAN
DAN BARU HIPOTESIS PENYUSUNAN
HIPOTESIS

UJI HIPOTESIS

STUDI ANALITIS STUDI EKSPERIMENTAL:

- X - SECTIONAL
- STUDI KASUS- - UJI KLINIS
CONTROL
- UJI COBA LAPANGAN
 STUDI DESKRIPTIF
• Information is collected only on those
individuals with a health problem or a
particular exposure. There is no comparison
group. Much useful information can be
derived from these studies but no definite
analysis of cause-effect association can be
made from these information
Characteristic :
1 . Hypothesis test -
2 .No comparison group
3 . Cause-effect association -
EXAMPLES :
• In epidemiology compiling and analyzing
data by time, place and person is desirable
for several reasons. 1. becomes intimately
familiar with the data 2. this provides a
detailed description of the health of
population 3. such analysis identifies the
population that are at greatest risk of
acquiring a particular disease
• TIME
Disease rates change over time.
Some of these changes occur regularly and
can be predicted.
Example : the seasonal increase of influenza
cases with the onset of cold weather is a
pattern that is familiar to everyone
Malaria by year, United States, 1930 - 1990
• Place
We describe a health event by place to gain insight
into the geographical extent of the problem.
For place, we may use place of residence,
birthplace, place of employment, hospital unit,
urban and rural etc, depending on which may be
related to the occurrence of the health event.
• Example : diseases that are passed from one
person to another spread more rapidly in
urban areas than in rural ones, because the
greater crowding in urban areas provides
more opportunities for susceptible people to
come into contact with some one who is
infected.
• Person
When we organize or analyze data by “person”
there are several person categories.
Inherent characteristics of people ( age, race, sex),
acquired characteristics ( immune, marital status),
their activities ( occupation, use of tobacco,
drugs), the conditions under which they live (
socioeconomic, access to medical care)
• Example : Sex
• For some disease, this sex-related difference is
because of genetic, hormonal, anatomic, or other
inherent differences between the sexes.
• Premenopausal women have a lower risk of heart
disease than man of the same age. This difference
is attributed to higher estrogen level in women.
 PERTUSSIS (WHOOPING COUGH)
INCIDENCE BY AGE GROUP,
UNITED STATES, 1989
50
REPORTED CASES PER

40
100.000 POP

30

20

10

0
< 1 TH 1-4 TH 5-9 TH 10-14 TH 15-19 TH 20+TH
AGE GROUP (YEARS)
 PERTUSSIS (WHOOPING COUGH)
INCIDENCE BY AGE GROUP,
UNITED STATES, 1989
60
REPORTED CASES PER

50
100.000 POP

40
30
20
10
0
0-4 TH 5-9 TH 10-14 TH 15-19 TH 20+ TH
AGE GROUP (YEARS)
New smear-positive case notification by age
and sex (2005)
 CROSS SECTIONAL STUDIES
The comparison is made between a group of persons
who has the disease and a group that does not have
the disease, but the characteristic and/ or exposure of
the two groups are observed in the same time
D+ D- TOTAL

FR + A B A+B

FR - C D C+D

TOTAL A+C B+D A+B+C+D

 CROSS-SECTIONAL-STUDY
TOTAL POPULATION
SAMPLING

STUDIED POPULATION

D+ COMPARE D-

STUDIED STUDIED
CHARACTERISTIC CHARACTERISTIC
Note :
• The difference between X sectional and
retrospective studies (case-control study) depends
upon the factor of time.
• In many instances, particularly where knowledge
of the pathogenesis and natural history of disease
is minimal, the difference between X-sectional and
retrospective studies is difficult and sometime
impossible to make
 Study Process
1. Statement of the research question or
hypothesis.
2. Research variable identification :
Operational definition of risk factor (independent
variable) and disease ( dependent variable)
3. Assessment of risk factor and disease :
Interview, physical examination, laboratory test,
special procedures , medical record.
4 Analysis of data
Prevalence ratio = A/(A+B):C/(C+D)
Statistically method depend on variable scale
 ADVANTAGES

1 Quick and easy to perform

2 Straight forward data analysis

3 Loss to follow up –
 DISADVANTAGES
1 Difficult to interpret association in terms of
cause and effect
2 Not suitable for the rare disease, since
sample size requirement will have to be
large.
 CASE-CONTROL
STUDY

THE STUDY MOVE BACKWARD FROM

DISEASE ( EFFECT) TO RISK FACTOR
(CAUSE).
PERSON WITH AND WITHOUT DISEASE
ARE IDENTIFIED AND THEN THE
PRESENCE OR ABSENCE OF PREVIOUS
EXPOSURE TO THE RISK FACTOR IS
DETERMINED
 scheme of case-control study

select cases select appropriate

controls

obtain information about previous

exposure to proposed risk of
factors each group

compare the frequency of

exposure between the two group
 STUDY DESIGN OF CASE-
CONTROL
Risk factor +

Risk factor - CASES

POPULATION

Risk factor +

CONTROLS
risk factor -
 STUDY PROCESS

1 Selection of cases
. Clearly define case definition
. Should be incidence cases
. Representative of total cases.
2 Selection of controls
. Should be representative of the
population from which the cases come
. Be sure that the risk factor under study is not
also related to disease among control group
 3. Sources of cases and controls
Cases : Hospital, Community,
Registration or surveillance system
Controls : Hospital, Community, relative of
cases, Neighbors
4. Assessment of exposure to risk factor
. Should be ascertain in the similar procedure between
cases and controls.
. Use record or documents as much as possible ( The goal
is to obtain as accurate information as possible about each
individuals exposure to the main risk factors)
. Clearly define exposure to risk factor
5. Try to minimize bias
. From selection of cases and controls :
Selection bias
. From data collection about risk factor
exposure : memory or information bias
. From data analysis : Confounding bias
 Method of data analysis
Odds ratio : Odds of cases : odds of controls

A/(A+C) B/(B+D)
=: =A/C:B/D=AD/BC
C/(A+C) D/(B+D)

D+ D-

FR + A B

FR - C D
 ADVANTAGES
1 Efficient for the study of rare diseases
2 Efficient for the study of chronic disease (diseases
with a long latency)
3 Tend to require a smaller sample size than other
designs.
4 Less expensive than other designs
5 Many risk factors can be studied simultaneously
 DISADVANTAGES
1 Risk of disease cannot be estimated directly
2 Not efficient for the study of rare exposures
3 More susceptible to selection bias
4 Information on exposure may be less
accurate than other design ( memory bias)
5 Can investigate only one disease or outcome
ODDS RATIO FOR SMOKERS AND NON
SMOKERS

DAILY AVERAGE LUNG CONTROLS O.R

CIGARETTES SMOKED CA

0 7 61 1.0
1-4 55 129 3.7
5-14 489 570 7.5
15-24 475 431 9.6
25-49 293 154 16.6
50 + 38 12 27.6

SUMBER : LILIENFELD
 COHORT STUDY

The study move forward from risk

factor (cause) to disease (effect).
Population exposed and not exposed
to a risk factor are identified and
then both population were followed to
determine the frequencies of health
problems.
 STUDY DESIGN OF COHORT

Disease +

Risk factors +
Disease -

population Disease +
Risk factors -
Disease -
 STUDY PROCESS
1 Selection of studied cohort
. Total population, divided by risk factor
exposure
. Special group who possessed certain
characteristics or exposure such as doctors
. Risk or exposure group, those who recieves
risk factor such as industrial workers.
. From survey for special group : DM,
hypertension
2. Selection of comparison group or control
group without risk factor from
. General population
. Sample population without risk factor
3. Make sure that both exposure and non
exposure group do not have the disease of
interest before the study begins
4. Data collection
Risk factors : records, medical
examination, measures of the
environment, questionnaire
Problem : effect -
changes in exposure
Information on outcome/effect :
periodic/non periodic medical
examination, surveillance of death
certificate.
Problem : loss to follow up
5.Method of data analysis
RELATIVE RISK = A/ (A+B) : C/ ( C+D )
D+ D-

FR + A B

FR - C D
 ADVANTAGES
• Direct calculation of relative risk
• May yield information on the incidence of disease
• Clear temporal relationship between exposure and
disease
• Particularly efficient for study of rare exposure
• Can examine multiple effect of a single exposure
• Minimizes bias
• Strongest observational design for
establishing cause and effect relationship
 DISADVANTAGES
• Time consuming
• Often requires a large sample size
• Expensive
• Not efficient for the study of rare diseases
• Lost to follow-up
• Changes in exposure
• Ethic
 A cohort study of infant feeding practices in
city, suburban and rural areas in Zhejiang
Province, PR China
Liqian Qiu†1,2, Yun Zhao†2, Colin W Binns*2, Andy H Lee†2 and
Xing Xie†1
Address: 1Women's Hospital, School of Medicine, Zhejiang University,
PR China and 2School of Public Health, Curtin University, WA,
Australia
Published: 3 March 2008
International Breastfeeding Journal 2008, 3:4
 CLINICAL TRIALS

STUDY PROCESS
1. Statement of the research questions
Is intensive therapy, including more
frequent insulin injection and blood
glucose monitoring “superior” to standard
therapy for diabetes mellitus ?
• The parameter that is measured to answer the most
important question of the clinical trial is the
primary end point ( to assess treatment efficacy,
more than one end point can be measured)
. Quality of life
. Survival
. Complication
3. RANDOMIZATION
With randomization the determination of
treatment group assignment is based on
probability alone and is not influenced by
the physician or patient preference
 INTERVENTION
GROUP EFFECT

POPULATION
(SUBJECT) R

CONTROL EFFECT
GROUP

SCHEME OF CLINICAL TRIAL

 EFFECT

INTERVENTION INTERVENTION
GROUP EFFECT GROUP

SUBJECT R

CONTROL
EFFECT CONTROL
GROUP
GROUP

EFFECT

SCHEME OF CROSS-OVER CLINICAL TRIAL

4. Blinded Fashion
“ Blinding” means that the treatment
assignment is not known to certain persons.
Single blind study : the treatment assignment
is unknown to the patients or physician
Double blind study : the treatment assignment
is unknown either to the patients or to their
physicians.
5. Data collection ( see Cohort)
6. Method of data analysis ( see Cohort)
7. Ethical
 ADVANTAGES
• Randomization tend to balance prognostic
factors across study group
• Detailed information can be collected
• Dose level can be predetermined by the
investigator
• Blinding of participant can reduce distortion
in assessment of outcome
 DISADVANTAGES
• Subject exclusions may limit ability to
generalize finding to other patient ( external
validity)
• Ethical concerns may arise

• Financial cost
 EKSPERIMENTAL DESIGN

I. PRE-EXPERIMENTAL DESIGN
1. The one - shot - case study
XO
2. The one - group pretest - posttest D.
O1  X  O2
3. The static group comparison
X  O1
-  O2
II. TRUE EXPERIMENTAL DESIGN
1. Prestest - post-test with control group
“Population” O1  X  O2
R
(Subject) O1  O2
2. The posttest - only control group
X  O1
Population
R
-  O2
3. “Randomized Solomon Four Group”

O1  X  O2
O1  O2
R

X  O2

O2
III. QUASI EXPERIMENTAL DESIGN
1. Time Series Design
O1  O2  O3 X  O 4  O5  O6

2. Control Time Series Design

Ex group O1  O2  O3 X  O 4  O5
Control group O1  O2  O3  O 4  O5
3. Non Randomized Control Group Pretest-
Post Test Design
Ex group O1  X  O 2

Control group O1  O2
 The Effect of Vitamin D as Supplementary Treatment
in Patients with Moderately Advanced Pulmonary
Tuberculous Lesion

Aim: to compare the vitamin D group of pulmonary

tuberculosis patients with a placebo group in terms of
clinical improvement, nutritional status, sputum conversion,
and radiological improvement.