DESAIN PENELITIAN

Sumarni
Faculty of Medicine & Public Health
Tadulako University
Proses Ilmiah

Problem

Hipotesis

Prediksi

Uji

Bukan
Kesimpulan
 HASIL PENELITIAN

Pertanyaan Penelitian
Tujuan
Hipotesis

Nilai
HASIL
= sebenarnya + Kesalahan
 Kesalahan

• Kesalahan sistematis/bias
• Kesalahan acak
• Kesalahan sistematis dihindari
dengan penggunaan metodologi
dan desain penelitian yang tepat
• Kesalahan acak dihindari dengan
penggunaanstatistik yang tepat
 KESEIMBANGAN ANTARA
VALIDITAS & FISIBILITAS

Desain
Metodologi
Statistik

Fisibilitas Validitas

Proposal penelitian
Alam nyata Alam impian

Problem Hipotesis

Pembuktian Prediksi

Hasil Analisis Data

Interpretasi

Kesimpulan

Hipotesis yang baru

DESAIN PENELITIAN Laboratorium

Eksperimental Binatang Percobaan

Perlakuan Manusia

Deskriptif
(Tidak ada
perbandingan kelompok)

Paparan alamiah
(Ada perbandingan Cross-sectional
kelompok)
Observasional Analitik Cohort
Case control
Causal comparative correlational
Metaanalisis
DESAIN PENELITIAN MENURUT WAKTU

E Prospektif O

Cross sectional
E O Alamiah Cohort
Retrospektif

Case-control E Eksperimental O
Perlakuan

Lampau Sekarang Y.a.d.

Case-Control Study
 Schematic diagram of
case control study
Time
Onset of study

Exposed
Cases
Unexposed

Exposed
Controls
Unexposed
Direction of inquiry
 Odds & Odds Ratio (OR)

Odds: probabilitas terjadinya suatu keadaan

dibandingkan dengan probabilitas tidak terjadi

Odds ratio:
• ukuran hubungan paparan dengan kejadian penyakit
• untuk mengetahui seberapa besar resiko suatu
paparan sehingga menyebabkan penyakit
dibandingkan bila tidak terkena paparan
 Case control study

Case Control

Exposed a b

Unexposed c d

Odds ratio = ad / bc Odds = prob of even/ 1- Prob of even

 Reye’s syndrome
• Occurs in children 2-16 yrs as a serious
complication of influenza B or less often –
influenza A or varicella zoster infection.
• It begins with 1-2 day of nausea and vomiting
followed by CNS symptoms, including changes
in mental status, that range from lethargy to
coma and encompass delirium and seizure.
Elevated serum transaminase level and elevated
amonia concentration, and hepatomegaly are
common
Aspirin use during viral illness among
children and subsequent risk of
developing Reye’s
Reye’s syndrome
Treatment + -

Aspirin use 28 35

No aspirin 2 25
use
Odds ratio = 28 x 25 / 35 x 2 = 10
 Interpretation of OR
• The odds of having disease in question are OR
times greater among those exposed the suspected
risk factor
• For rare disease ( e.g. chronic disease which have
prevalence < 10%) OR approximates RR
– That is, the risk of the disease is approximately OR times
greater among those exposed to the suspected risk
factor
– RR = [P(D+/E+)] : [P(D+/E-)] = [a/(a+b)] : [c/(c+d)]. For
rare disease a and c are small relative to b and d. Thus
a/b approximates a/(a+b) and c/d approximates c/(c+d)
 Interpretation of OR….. cont

• The larger the value of OR, the stronger the

assoociation.
• When the value of OR is close to 1, the disease
and exposure to the risk factor are unrelated
• Value of OR < 1 indicate a negative association
 OR and study design
• Since P(D+/E+), P(D+/E-), P(D-/E+) and
P(D-/E-) can be estimated from result of a
cohort or an experimental study, OR can
be calculated directly these study designs
are employed
 OR in a case-control study
• The conditional probability of disease
P(D+/E+), P(D+/E-), P(D-/E+) and P(D-/E-
) can not be calculated directly from the
result of a case control study
• Estimates of the conditional probability of
exposure can be obtained in a case-
control study.
 Oral Contraceptive Use &
Breast Cancer
• Problem: Is OC associated with the risk of breast cancer?
• Research Hypothesis: There is an association between OC use
and breast cancer
• Cases: All women 20-54 years old with newly diagnosed breast
cancer who reside in one of eight geographic regions in the
United States. The women were identified through population-
based tumor registries
• Controls: Women of the same ages selected at random from the
general population of the same region and the same period
when cases were diagnosed
Selection of Cases
Usually based on • Incident outcomes
outcome • Prevalent outcomes
 Selection of Controls
Should be selected according to
predetermined criteria to ensure the
absence of outcome
May be better to choose subject with
similar referral, surveillance, and other
factors capable of distorting the study
sample
 Exposure
The primary exposure of • Dose, duration, and
interest is prior period of exposure
exposure, not should be specified
contemporaneous one ahead of time
Need biologic model for
how exposure causes
outcome
 Case-Control Study

Advantages Disadvantages
• Statistically more • Enhanced potential for
efficient when outcome sample distortion
are rare • Exposure ascertainment
• Quicker when outcome more prone to error and
are delayed bias
• Less costly
COHORT STUDY
 Cohort Study

• Cohort: a group of people that exposed to

particular exposure in a period of time

• Cohort study can be conducted

prospectively or retrospectively
 Cohort study

• Retrospective Cohort, both exposure and

outcome observed have been occurred
before study conducted

• Prospective Cohort study, subject of the

study may already been exposed or not
yet been exposed when a study started,
but the outcome not yet been available
 Retrospective and prospective
cohort study

Disease
Exposed No disease
Disease
Unexposed No disease
Disease
Exposed No disease
Disease
Unexposed No disease

Past Present Future

 Requirement

• Exposure criteria is clear and specific

– Intensity
– Duration
– Regularity
– Variability
• Duration of follow-up
• Outcome measurement: blind, objective
 Schematic diagram of cohort
study
Time
Onset of study

Disease
Exposed
No disease

Disease
Unexposed
No disease
Direction of inquiry
 Cohort study
Disease
+ -

Exposed a b

Unexposed c d

Relative risk (risk ratio) = a/(a+b) : c/(c+d)

 Relationship between 10 minute Apgar
scores and risk of death in the first year of
life among children with birth weights of at
least 2500 g

Death
+ -
Apgar score
0-3 42 80

Apgar score
4-6 43 302

Relative risk (risk ratio) = 42/(42+80) : 43/(43+302)

= 2.8
 Interpretation of RR
• The disease is RR times more likely to occur
among to exposed to the suspected risk
factor
• The larger the value of RR, the stronger the
association
• The value of RR close to 1 indicate that the
disease and exposure to the risk factor are
unrelated
• Value of RR < 1, indicate a negative
association
 RR and study design
• RR can be directly calculated only in a
cohort or experimental study
• Because incidence can not be estimated
from a case control study, RR can not be
calculated directly. Under some
circummctances, the RR in a case control
study can be estimated by OR
 Result of hypothetical cohort
study
Disease Disease Total
present absent
Exposed a b a+b

Non c d c+d
exposed
Total a+c b+d

a/(a+b) a/b
RR = OR =
c/(c+d) c/d
 Strength of the association
Relative Risk of disease if exposed to risk factor
risk (RR) =
Risk of disease if not exposed to risk factor

Odds that exposed individual will have

Odds disease
ration =
Odds that non-exposed individual will have
disease

Odds of disease = P(disease)/1 – P (disease)

= P (disease) / P (no disease)
Pemilihan Subjek kasus dan kontrol

• Keduanya diambil dari populasi yang sama

• Keduanya belum menderita penyakit yang akan diteliti

saat penelitian dimulai

• Karakteristik sampel kasus dan sampel kontrol sama

• Ada informasi yang ekual antara kedua grup

• Kedua grup harus dapat dihubungi dan dapat di follow

up
Contoh beberapa
penelitian
 CRP and BMI predict outcome in end-
stage respiratory failure,
Cano et al, 2004, Chest, 126: 540-546

• Menentukan faktor prognosis kematian dan

kesakitan utk penyakit pernafasan stadium akhir
• Penelitian Cohort (COPD: 42,8%, restrictive
disorder 36,3%, campuran 13,5%, bronkiektasis
7,4%)
• CRP, BMI, PaO2 dan kortikosteroid oral mrpk
prediktor independen kemampuan hidup
 Graf failure

< 0,5 g/hr: 13,3%

0,5-1 g/hr: 25,3%

> 1 g/hr: 45,8%

HR: 4,58
HR: 7,03
 Oscarson et al.,
menyimpulkan
• Kenaikan kadar TnT pasca operasi
memberi risiko 15 kali kematian pd th
pertama
• Pemeriksaan TnT untuk lansia (sering
terjadi silent myocardial ischemia) pd
masa perioperatif bermanfaat utk menilai
risiko kematian pd tahun pertama pasca
operasi
Anti p53
Ekspresi p53
Ekspresi keduanya
Seronegative myasthenia gravis:
disease severity and prognosis
(Romi et al., 2005, European Journal of
Neurology, 12: 413-418)
Anti reseptor asetilkolin berkorelasi dengan derajat
Myastenia Gravis
 CD2+ Vs CD2-
CD2+APL CD2-APL

Hitung lekosit 34,5 ±13,1/L 6,8 ±2,1/L

Karakteristik morfologis 50% 0%

Angka remisi sempurna 50% 87%

Kematian 66,7 13
(Overall Survival)
 The association between short sleep
duration and obesity in young adults: a
13-year prospective study

Hasler G ., et al., Sleep. 2004 Jun

15;27(4):661-6
The association between short sleep duration
and obesity ………..

• STUDY OBJECTIVES: to test the hypothesis

that short sleep duration is associated with
obesity and weight gain during young adulthood.
• DESIGN: Prospective single-age cohort study of
young adults. Information was derived from 4
interviews when participants were ages 27, 29,
34, and 40 years.
• SETTING: Community setting.
• PARTICIPANTS: 496 young adults.
 The association between short sleep duration
and obesity …….

• RESULTS: This study showed an association

between short sleep duration and obesity (at
age 27 years, odds ratio: 7.4, 95% CI: 1.3-43.1)
and a negative association between sleep
duration and body mass index in young adults.
These associations persisted after controlling for
a variety of potentially confounding variables,
including family history of weight problems,
levels of physical activity, and demographic
variables. Associations between sleep duration
and obesity diminished after age 34 years.
 The association between short sleep duration
and obesity ……..

• There was a trend (P = .08) for average change

rate of weight gain to be negatively associated
with average change rate of sleep duration.
• CONCLUSIONS: Because sleep duration is a
potentially modifiable risk factor, these findings
might have important clinical implications for the
prevention and treatment of obesity.
 Smoking cessation in a
population-based cohort study
Garcia M .et al., Arch Bronconeumol.
2004 Aug;40(8):348-54
 Smoking cessation in a population-
based cohort study
• Objective: To study the incidence rates and the
determinants of smoking cessation in a
population-based cohort.
• Material and methods: We used data from the
Cornella Health Interview Survey Follow-up
Study. Subjects who declared they were daily
smokers at baseline (1994) and had complete
follow-up, with information on smoking status in
2002, entered into analysis. We calculated
incidence rates and the relative risks of
cessation (with 95% confidence intervals) using
the Cox model.
 Smoking cessation in a population-
based cohort study
• Results: Out of 353 daily smokers, 100 quit
smoking during the follow-up period (cumulative
incidence of 28.3%). The incidence rate of
cessation was higher among men (42.34 per
1000 person-years) than among women (24.97
per 1000 person-years), with a relative risk of
cessation of 1.69 (95% confidence interval, 1.02-
2.79) for men. Age and level of education were
associated with a higher relative risk of quitting
in men.
 Smoking cessation in a population-
based cohort study
• Conclusions: The main determinants for
smoking cessation are sociodemographic
(sex, age, and level of education).
 Factors predisposing to perinatal
death related to uterine rupture during
attempted vaginal birth after
caesarean section: retrospective
cohort study

Smith GC ., et al., BMJ. 2004 Jul 19

 Factors predisposing to perinatal death
related to uterine rupture ……….

• OBJECTIVE: To determine the factors

associated with an increased risk of perinatal
death related to uterine rupture during
attempted vaginal birth after caesarean
section.
• DESIGN: Population based retrospective
cohort study.
• SETTING: Data from the linked Scottish
Morbidity Record and Stillbirth and Infant
Death Survey of births in Scotland, 1985-98
 Factors predisposing to perinatal death
related to uterine rupture ……..

• PARTICIPANTS: All women with one previous

caesarean delivery who gave birth to a
singleton infant at term by a means other
than planned repeat caesarean section (n=35
854). Main outcome measures All intrapartum
uterine rupture and uterine rupture resulting
in perinatal death (that is, death of the fetus
or neonate).
 Factors predisposing to perinatal death
related to uterine rupture ……….

• RESULTS: The overall proportion of vaginal

births was 74.2% and of uterine rupture was
0.35%. The risk of intrapartum uterine rupture
was higher among women who had not
previously given birth vaginally (adjusted
odds ratio 2.5, 95% CI 1.6 to 3.9, P<0.001)
and those whose labour was induced with
prostaglandin (2.9, 2.0 to 4.3, P<0.001).
 Factors predisposing to perinatal death
related to uterine rupture……….

• Both factors were also associated with an

increased risk of perinatal death due to
uterine rupture. Delivery in a hospital with
<3000 births a year did not increase the
overall risk of uterine rupture (1.1, 0.8 to 1.5,
P=0.67). However, the risk of perinatal death
due to uterine rupture was significantly higher
in hospitals with <3000 births a year (one per
1300 births) than in hospitals with >/=3000
births a year (one per 4700; 3.4, 1.0 to 14.3,
P=0.04).
 Factors predisposing to perinatal death
related to uterine rupture………

• CONCLUSION: Women who have not

previously given birth vaginally and those
whose labour is induced with prostaglandin
are at increased risk of uterine rupture when
attempting vaginal birth after caesarean
section. The risk of consequent death of the
infant is higher in units with lower annual
numbers of births.
 Advantages
• Direct calculation of risk ratio (relative risk)
• May yield information on the incidence of disease
• Clear temporal relationship exposure and disease
• Particularly efficient for study of rare exposures
• Can yield information on multiple exposures
• Can yield information on multiple outcomes of a
particular 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 study of rare diseases
• Losses to follow-up may diminish
validity
• Changes over time in diagnostic
methods may lead to biased results
CROSS SECTIONAL
 Penelitian potong lintang

• Deskriptif
– Mengukur prevalensi suatu penyakit pada suatu populasi

• Analitik
– Mengukur hubungan antar faktor resiko dengan kejadian suatu penyakit
 Struktur Penelitian

• Mirip cohort, tetapi pengukuran

dilakukan sekali
• Tidak ada follow-up
• Cocok untuk menjelaskan variabel-
variabel secara deskriptif dan juga pola
distribusinya
• Dapat juga untuk menguji hubungan
 Penelitian potong lintang

• Pertanyaan penelitian
• Kriteria populasi penelitian
(target/accessible population)
• Teknik pengambilan sampel
 Contoh

Chlamydia

+ -
+ 10 40 50
Kontrasepsi oral
50
- 5 45

15 85 100

Prevalensi = 15/100
Prevalensi relatif 10/5 = 2
 Prevalensi
Adalah jumlah kasus yang terjadi dibandingkan
dengan jumlah subjek yang diteliti

Bermanfaat bagi:
• Perencana program kesehatan
• Klinisi
Associations in Clinical Medicine

• To define associations between disease

and predisposing or causal factor

• To discover a cause-and-effect
relationship that will result in effective
treatment and prevention strategies
 Risk factor

• A condition, physical characteristic, or

behaviour that increase the probability that
a currently healthy individual will develop a
particular disease
 Type of risk factors

• Environmental risk factors

• Behavioral, or life-habit, risk factors
• Social risk factors
• Genetic risk factors
 Exposure
• An individual who has contact with or who
manifest the risk prior to becoming ill
• Exposure may occur at a single point in time
• More commonly, exposure is chronic (e.g.,
hypertension).
• Measure of chronic exposure include current
dose, total cumulative dose, years of exposure,
and years since the first exposure
 Relationship of risk factors to disease

• A risk factor may be a causal factor of the

disease in question or merely a marker for the
increased probability of disease

• For example, while poor prenatal care and drug

use constitute causal factors for neonatal
mortality, socioeconomic status would be
considered a marker for neonatal mortality
 Kelebihan penelitian
potong lintang
• Tanpa menunggu
• Memberi gambaran tentang prevalensi
atau faktor resiko
• Networks of causal links
• Dapat dimasukkan dalam penelitian cohort
atau perlakuan
 Kelemahan penelitian
potong lintang

• Sulit menetapkan hubungan kausal

• Kurang praktis untuk penelitian
terhadap kasus yang jarang
 Serial survey

• Merupakan seri penelitian-penelitian

potong lintang pada satu populasi
• Untuk melihat pola perubahan dalam
periode waktu tertentu
• Bukan cohort
 Study Design

Design Ranking
Strongest
Experimental

Prospective cohort

Retropective cohort

Case control

Cross-sectional Weakest
 Experimental vs observational studies

• In experimental studies, the investigator

controls the assignment of study subjects to
experimental and control groups and actively
manipulates one variable (independent
variable) while observing the resultant in another
(dependent variable)

• Example, study on caffein effect on systolic

blood presure
Effect of caffein --- sistolic BP
Placebo SBP +
SBP -

R
Low dose caffein SBP +
SBP -
High dose caffein SBP +
SBP -
6 months
 Hill’s criteria for causality
Used to evaluate the plausibility of putative cause-and-
effect relationship:
1. Study design
2. Strength of the association
3. Consistency
4. Correct temporal relationship
5. Dose-response relationship between exposure and
severity of the outcome
6. Plausibility
7. Specificity
8. Analogy