14/08/57 1 Case-control study Patarawan Woratanarat, M.D., Ph.D. (Clin. Epid.) Department of Orthopaedics Faculty of Medicine Ramathibodi Hospital Objectives To understand A concept of case-control study Conduct a case-control study Selection of study population The principle of measurement Data collection Analysis
65
Embed
CASE CONTROL STUDY - Mahidol University · Case-control study Patarawan Woratanarat, M.D., Ph.D. (Clin. Epid.) Department of Orthopaedics Faculty of Medicine Ramathibodi Hospital
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
14/08/57
1
Case-control study
Patarawan Woratanarat, M.D., Ph.D. (Clin. Epid.)
Department of Orthopaedics
Faculty of Medicine Ramathibodi Hospital
Objectives
� To understand
� A concept of case-control study
� Conduct a case-control study
� Selection of study population
� The principle of measurement
� Data collection
� Analysis
14/08/57
2
Types of research
� Qualitative / quantitative
� Descriptive
� Exploratory/observational: case-control, cohort, cross-sectional study
� Experimental: RCT
4 groups
Think about your research question?
14/08/57
3
Study designs
� Guideline
� Therapy � RCT/Systematic review
� Diagnosis � Cross-section
� Screening � Cross-section
� Prognosis � Cohort
� Causation � Cohort, case-control
A concept of case-control study
14/08/57
4
Study designs
� Case-control study
Direction of the study
Population
People with disease
People withoutdisease
Exposed
Exposed
Not exposed
Not exposed
Study designs
� Case-control study
Direction of the study
Population
THR patientsWith DVT
THR patientWithout DVT
Spinal anesthesia
Spinal anesthesia
General anesthesia
General anesthesia
14/08/57
5
Case-control studies
� Advantages
� Valuable for rare conditions
� Short duration
� Inexpensive
� Small sample size
� Yield odds ratio
� Disadvantages
� Limit to one outcome
� Potential selection bias
� Measurement bias
� Survivor bias
� Do not establish a temporal sequence
� Do not yield absolute risk estimates
Conducting a case-control study
14/08/57
6
Research process
� Phase I: Identify the research question
� Phase II: Design the study
� Phase III: Methods
� Phase IV: Data analysis
� Phase V: Communication
Research question
� Hypothesis: � a statement in which an attempt is made to
generalize about the nature of the universe in which we live.
� To act as a guide in interpreting the wider meanings of a particular data set
� Research question� Identifies the issue to be addressed by the
research , it does not have to be stated in a testable form
14/08/57
7
Research question
� Research problem� Clinical experience, theory, literature
� Research question should be� Important
� Answerable
� Feasible
� Identify� Target population
� Variables
Research question
� Hypothesis: non directional
� Ho: There is no difference in the reduction of DVT in Thai patients who undergo elective total hip replacement under spinal anesthesia compared with general anesthesia
� Ha: There is a difference in the reduction of DVT in Thai patients who undergo elective total hip replacement under spinal anesthesia compared with general anesthesia
14/08/57
8
Research question
� Hypothesis: directional
� Ho: spinal anesthesia does not reduce risk of DVT in Thai patients who undergo elective total hip replacement from 10% to 3% when compared with general anesthesia.
� Ha: spinal anesthesia reduce risk of DVT in Thai patients who undergo elective total hip replacement from 10% to 3% when compared with general anesthesia.
Gr 2: Case - definition, inclusion & exclusion criteria
Gr 3: Control – definition, inclusion & exclusion criteria
Gr 4: Matching – yes/no and why?
14/08/57
10
Design the study
� Population and sample
� Target population/reference population
� Study population
Target population
Accessible population
Study population
Thailand
Bangkok
Ramathibodi Hospital
Sampling bias…….
Selection of cases
� Definition
� Diseases, ICD-10
� Example: osteoporotic hip fracture definition
� Thai adults, age ≥≥≥≥ 51 years old whom are admitted in orthopedic wards with the first episode of osteoporotic hip fracture, ie. fracture of femoral neck, intertrochanter, subtrochanter sustained from low-velocity accident.
� (ICD-10, S72.0-72.9)
14/08/57
11
Selection of cases
� Probability samples (Random selection)
� Simple random sampling
� Systematic sampling
� Stratified random sampling
� Cluster sampling
� Non probability samples
� Convenience sampling
� Quota sampling
� Proposive sampling
� Snowball sampling (chain referral)
Sampling techniques
Selection of cases
� Whole population
� Hospital
� Incident cases
� Avoid prevalent cases (distort exposure)
� Example:
New case of spinal stenosisFloor activity
Chronic spinal stenosisFloor activity X
5 years
14/08/57
12
Selection of controls
� Definition: no outcome (case definition)
� Example:
� Thai adults who are neighborhoods ofcases aged 51 years and were not directrelatives of cases. No fracture offorearm, spine, and hip.
Selection of controls
� Sampling
� Site: the same as cases
� Hospital or community
� Has an opportunity to expose to the exposure
� Can be cases in the future
� Example:
� Controls of CACx: male?, child?
14/08/57
13
Population vs Hospitalbased case-control study
� Population based
� Can define source of population
� Cases and controls are from the same source
� Exposure in the controls represent real situations
� Hospital based
� Convenience
� Good cooperation
� Baseline characteristics are similar to cases
� Convenience for searching available exposure data
Examples
A case-control study
� Risk factors for Hip fracture
� Drugs vs. road traffic accident
14/08/57
14
Risk factors for hip fracture
� Frequency matching case-control study; 1:1 of cases :controls. Matched by sex and age + 5 years (not less than 51years old).
� Hospital controls: same hospital
� Community controls: neighborhoods� Search for address registry and national ID
� Pick up people who was in required age and lived within 1 km from case’s address
Risk factors for hip fracture
Total recent activity scores
Cases vs Hospital controls
Cases vs Community controls
OR (95%CI) P-value OR (95%CI) P-value
Inactive* 0.80
(0.51-1.25)
0.341 0.32
(0.20-0.50)
<0.0001
Active 0.53
(0.32-0.87)
0.012 0.20
(0.12-0.34)
<0.0001
14/08/57
15
DRUG vs. ROAD TRAFFIC INJURYDRUG vs. ROAD TRAFFIC INJURY
• Case verification by ER nurses• Informed consent
3. Alcohol breath testBlood for alcohol levelUrine collection
5. Case admissioninterview by ward nurseswithin 72 hours
4. Notification To ward & Research center
6. Specimen & questionnairepickup by Research center (Rama)
Mobile unit
1. Verify site from case RTI area
2. Search gas stations
3. Contact gas stations
4. Data collection
Controls
14/08/57
16
Selection of controls
Hospital-based?
� Recommendation� Unspecified disease (reflect real
exposure)
� New patients
� Low number of underlying diseases
� Avoid disease that correlated with the interesting exposure
� Example: Patients, aged 51 years, who are newly admitted (not 1st admission) in other wards in the same hospital and were not severely ill.
14/08/57
17
Matched??
� Advantages� More reliable for a
comparison between case and controls
� Need small samples
� More specific controls
� Discard confounding factors which were matched
� Disadvantages� Time and budget consuming
� Hard to find a specific control– discard matched case
� Unable to find a relationship between matched variables and outcome
� Residual difference if match for continuous or ordinal data
� Overmatching: cannot find the difference between cases and controls
Matched
� 1:1
� 1:2 – 1:4
� Decreased sample size of cases
Alpha Power Po OR Match N of cases
0.05 0.8 0.03 3 2
4
216
116
14/08/57
18
Frequency matching
� Match 2-3 variables
� Example
� Controls were matched to the cases according to sex and age + 5 years. And they were admitted to the same hospital within 90 days before or after the admission date of the cases.
� The important factors related to hip fracture are physical activity, Chinese race, and CVA.
� Physical activity and CVA also related to both IT & FN. Chinese race and sedative drugs are associated with FN whereas impaired walking ability is associated with IT.
Recommendation
� ICD register for evaluation and monitoring hip fracture incidence in Thailand.
� It is time to prevent hip fracture by exercise, prevent and give good care for CVA
� Verify calcium as a protective factor by prospective study with log diary.(CEA, CBA)
1. Case verification by ER nurses2. Informed consent
3. Alcohol breath testBlood for alcohol levelUrine sample collection
5. Case admissioninterview by ward nurseswithin 72 hours
4. Notification To ward & Research center
6. Specimen & questionnairepickup by Research center (Rama)
CaseCase
Mobile unit
1. Verify site from case RTI area
2. Search gas stations
3. Contact gas stations
4. Data collection
Controls
14/08/57
56
Materials & Methods
� Data monitoring: site visit
� Data quality assurance: call subjects
� Data editing
� Double data entry: EpiInfo
Materials & Methods
� Data analysis
� Mean + SD, percentage
� Conditional logistic regression
� Univariate analysis
� Multivariate analysis (backward stepwise)
� PAR calculation (using data from survey study)
� Stata 9.0 (StataCorp, Texas)
14/08/57
57
Sample size
P0 OR Case:Control ratio
Subject (N)
0.02
(Probability of hypnotics)
3 4 Case (250)
Control (1000)
0.02
(Probability of hypnotics)
2.5 4 Case (400)
Control (1600)
Results
200 CASES850 CONTROLS
14/08/57
58
Diagram 1 Distribution of injury (cases)
41
106 5
59
21
5
0
10
20
30
40
50
60
Brain/head Chest Abdomen Spine Limbs Face Skin
Injury (%)
Table Table Table Table 1 1 1 1 Characteristics of cases and controlsCharacteristics of cases and controlsCharacteristics of cases and controlsCharacteristics of cases and controlsCharacteristics Cases
N = 200 (%)Controls
N = 850 (%)P-value
Age (years), mean (SD) 30.18(11.8) 35.45(10.6) <0.001*
Gender MaleFemale
189(94.5)11(5.5)
803(94.5)47(5.5)
-
Type of vehicles
Car/van/truck/bus 22(11.0) 91(10.7) 0.803
Motorcycle 178(89.0) 759(89.3)
Type of driving
General 161(80.5) 588(69.2) < 0.001*
Commercial 39 (19.5) 262 (30.8)
Experience of driving (years)
< 4 99(49.8) 176(20.7) < 0.001*
5-10 62(31.16) 353(41.5) 0.149
11-15 10(5.0) 96(11.3) 0.553
> 15 28(14.1) 225(26.5)
14/08/57
59
Table Table Table Table 1 1 1 1 Characteristics of cases and controls (contCharacteristics of cases and controls (contCharacteristics of cases and controls (contCharacteristics of cases and controls (cont’ ))))
Characteristics Cases N = 200 (%)
ControlsN = 850
(%)
P-value
Timing of driving
Night time/dawn/dust 52(26.0) 66(7.8) <0.001*
Daytime 148(74.0) 784(92.2)
Duration of driving (minutes)
> 90 66(33.9) 189(22.7) <0.001*
41-90 49(25.1) 199(23.9) 0.030*
21-40 43(22.1) 223(26.7) 0.465
<20 37(19.0) 223(26.7)
Protective gear + head light
No 76(38.0) 184(21.7) <0.001*
Yes 124(62.0) 666(78.4)
Table 2 Single variable conditional logistic regression of one-month recall of drug and substances use
� Psychoactive drug use increased risk of RTI. It also contributed to RTI by 24%.
� It calls for legislative measures and/or publicity campaign to modify use of psychoactive drugs in addition to current measures for drink driving control.