case control study moderator : chetna maliye presenter reshma sougaijam
TRANSCRIPT
Case control study
Moderator : Chetna MaliyePresenter Reshma Sougaijam
Frame work
1. Introduction
2. Design and steps of case control study
3. Comparison of case control and cohort study
4. Advantage and disadvantage of case control study
5. Confounding and bias
Introduction
• Types of epidemiological studies:
• Case control studies:
A case control study is defined as an epidemiological
approach in which the researcher starts by picking up
cases who have already developed a particular disease or
outcome of interest , and a comparison group (controls)
of subjects who except for the fact that they have not
developed the particular disease, are otherwise similar to
the case.
Design and steps of case control study
Steps of conducting case control study
• Step 1- Specify the total population and actual (study
population)
• Step 2- Specify the measure study variables and
their scales of measurement
Outcome variable:
Exposure variable
Specify the scale of measurement of the exposure
variable
Potential confounding factor (PCF):
• Step 3- Calculate the sample size : The sample size may be
estimated by
• n= (Zα+Zβ)2(p₁q₁+p₂q₂)/(p₂-p₁)2
• where, Zα is the normal deviate corresponding to the level of the
significance to be used in the test,
• Zβ is the normal deviate corresponding to the two tailed probability
• p₁ and p are the proportions of exposed subjects in cases ₂and control respectively.• q = 1-₁ p and₁
• q = 1-₂ p₂
Example • To study the association of oral contraceptive use with breast cancer.
• Proportion of OC use among breast cancer patient is 40% and proportion of OC use in the population who do not have breast cancer is 20%. With 95% Confidence interval, and 80% power of the study, we have
• Zα=1.96, Zβ=1.65• p =40 ₁ p =20 ₂• q = 60₁ q =80 ₂
• Hence n= (Zα+Zβ)2(p₁q₁+p₂q₂)/(p₂-p₁)2
=(1.96+1.65)²(40x60+20x80)/(20-40)
=(3.23)² (4000)/(20)²
=10.43x4000/400 =417.2/4
=104
• Step 4- specify the selection criteria of
cases
Diagnostic criteria
State the inclusion or exclusion criteria
Sources of cases: Cases may be drawn from
a. Hospitals b. General population
Incident or prevalent cases
• Step 5- Specify the selection procedure of controls• Source of controls
1. Hospitals controls
2. General population
3. Relatives
4. Neighborhood controls• Exclusion / inclusion criteria• Number of control per case• Number of control group
• Matching: Matching is defined as the process of selecting the controls so that they are similar to the cases in certain characteristics, such as age, race, sex, socioeconomic status, and occupation
• Matching may be of 2 types
1. Group matching (frequency matching)
2. Individual matching
• Problem with matching:1. Practical problem with matching: if an attempt is made to
match according to too many characteristics, it may prove difficult or impossible to identify an appropriate control
2. Conceptual problem with matching: once we matched control to cases according to a given characteristics, we cannot study that characteristic.
• Step 6- specify the procedures of measurement and specially take care to ensure validity and reliability
• The basic measurement should have two essential
requirements. That is, it should be “valid” and “reliable”
1. The measurement process should be valid: the measurement which we are making and recording should correctly measure what we really intend to measure
2. Secondly the measurement process should have “Reliability”: this is the ability of a measurement process to give consistent results when repeated applications are made.
• Step 7- Analysis of data
•Calculate the odds ratio and its 95%
confidence interval.
•Control of confounding will require
stratified analysis using Mantel-Haenszel
technique or a multiple logistic regression.
Calculation of Odds Ratio in case control study
• Odds Ratio in a case control study is defined as the ratio of the Odds that the cases were exposed to the odds that the control were exposed
Example of calculating odds ratios in case control studies:
1. Unmatched case control study in which controls were not
matched to the cases.
Example : Let us assume the following: a case control study is
carried out in 10 case and 10 controls. N indicates non exposed
individual and E indicate exposed individual.
Thus 6 of cases were exposed and 3 of control were exposed
Case control
Exposed 6(a) 3(b)
Non exposed 4(c) 7(d)
10 10
The odds ratio in unmatched case control study
equals the ratio of the cross product.
Odds ratio=ad/bc=6x7/4x3=42/12=3.5
Calculating odds ratio in matched-pairs case-control study
•The case control pairs that had the same exposure
experience are termed concordant pairs, and those with
different exposure experience are termed discordant pairs.
Calculation of odds ratio in such a matched- pair study is based on the discordant pairs only
The odds ratio for matched pairs is therefore the ratio of discordant pairs
Odds ratio=b/c
Example: A case control study in which each case is matched with a control, resulting in 10 case-control pairs
Odds ratio=4/1=4
Difference between case control and cohort studiesCase control study Cohort study
Proceed from effect to cause Proceed from cause to effect
Start with the disease Starts with people expose to risk factors or suspected cause
Test whether the suspected cause occurs more frequently in those with the disease than among those without the disease
Test whether disease occurs more frequently in those expose, than those not similarly exposed
Usually the first approach to the testing of a hypothesis, but also useful for exploratory studies
Reserved for testing of precisely formulated hypothesis
Involves fewer numbers of subjects. Involves larger number of subjects
Yields relatively quick results Long follow up period often needed, involved delayed follow up.
Difference between case control and cohort studiesCase control study Cohort study
Suitable for the study of rare disease Inappropriate when the disease or exposure under investigation is rare
Yields only estimates of odds ratio Yields incidence rate , Relative Risk, Attributable Risk
Cannot yield information about the diseases other than that selected for study
Can yield information about more than one disease outcome
Temporal association is never proven Temporal association is proven
Recall bias is a potential problem Recall bias is not an issue
Relatively inexpensive Expensive
Advantage and disadvantages of different observational study designs
Advantages and disadvantages of case control studies
• Advantages :
Inexpensive, requires only a few subjects gives quick
results
Well suited for outcome which is rare
Helps in examining multiple etiologic factors- once we
have the case of the disease, we can take history of all the
factors that we feel may be risk factors
No attrition problem, because case control study do not
require follow up of individuals into the future.
No risk to the subject
Advantages and disadvantages of case control studies
• Disadvantages
Not a good method for studying rare exposure
Does not give ant idea of “incidence” or “prevalence”; it
only gives us a measure of Odds Ratio
Prone to various forms of selection and information bias
Temporal relationship is not proven
Bias and confounding• Bias:. Bias has been defined as any systematic error in the
design, conduct or analysis of a study that result in a mistaken estimate of an exposure’s effect on the risk of disease.
• Types of bias:
1. Selection bias: selection bias is an error in selecting a study group or groups within the study and can have a major impact on the internal validity of the study and the legitimacy of the conclusion.
2. Information bias (measurement bias): ): Information bias is a systematic error that arises because of incorrect information while making measurements on one or more variables in the study.
• Types of selection bias:1. Berksons’ bias (hospital selective
admission):
2. Incidence-prevalence bias (Survivorship bias, Neyman’s bias
•Types of information bias:
1. Recall bias:
2. Reporting bias
3. Observer’s (interviewer’s bias):
Confounding:
• A confounding variable is defined as one which explains away the observed association between an exposure and an outcome variable.
• Example : In a study to see the association of hypertension with coronary heart disease, age may be a confounding factor.
• This is because, age is a known risk factor for
CHD and age is associated with hypertension.
Control of confounding
The methods commonly used to control confounding in the design of an epidemiological study are:
Randomization
Restriction
Matching
At the analysis stage, confounding can be controlled by
Stratification
• Statistical modeling or multivariate analysis
References
• R.Beaglehole, R.Bonita, T. Kjellstorm. Basic epidemiology. World Health Organization, Geneva: AITBS Publisher; 2006
• Bhalwar R et al. Text book of Public Health and Community Medicine 1st ed. Pune :Department of Community Medicine Armed Forces Medical College; 2009
• Leon G. Epidemiology 3rd ed. Philadelphia: Elsevier Saunders; 2004
• Mac Mahon B, Trichopoulos D. Epidemiology Principles and Methods 2nd ed. New York: Little, Brown and Company;1996
• Park K. Text Book of Preventive And Social Medicine 21st ed. India: M/s Banarsidas Bhanot;2011