Download presentation
Presentation is loading. Please wait.
1
Case control studies
2
LEARNING OBJECTIVES AT THE END OF THIS LECTURE LEARNER SHOULD BE
Able to know the various analytical studies, their types, and their indications? Able to choose the suitable analytical study for testing the hypothesis Able to analyze and test the causal hypothesis Able to measure the various biases
3
PERFORMANCE OBJECTIVES
By knowing the population risks, the student must be able to save that population by eliminating or decreasing the risk.
4
Study designs Study types Case-control cohort Quazi experimental
Descriptive (formulation hypothesis) Individual based Case studies Case series Population based co-relational Analytical (testing hypothesis) Observational cohort Case-control Cross-sectional interventional RCT’s (III) Quazi experimental
5
CASE-CONTROL STUDIES Two groups of persons:
Cases & controls and study is an analytical and observational Cases (group with the disease in question) Controls (group without the disease in question) Case control design is also called as retrospective or backward looking study as the study commences with the cases in which disease in question had already appeared.
6
CASE CONTROL DESIGN Proportion of the cause in cases is compared to that of controls. If it is significantly more in cases than in control group, causal association is suspected. This comparative design increases its efficiency.
7
Case Control Study Cases Controls Present Past Risk Factor Present
Risk Factor Absent Comparison Controls Risk Factor Present Risk Factor Absent Present Past 7
8
A+C B+D A B C D 2 x 2 Contingency Table for Cases and Controls
Exposure + Exposure - C D Total Cases A+C B+D Exposure Among Cases A/(A+C) Exposure Among Controls B/(B+D)
9
METHODOLOGY
10
STEPS OF CASE CONTROL STUDY
Cases selected Controls selected Enquiry (questionnaire) and records’ verification (check list) for the amount of exposure in both groups Comparison analysis and risk measurement If exposure is more in cases than in controls causal association is suspected .
11
. AVOID SELECTION BIAS, INFORMATION BIAS & MEASURMENT BIASES CASES
GENERAL POPULATION . HOSPITALS RELATIVES NEIGHBOURS AVOID SELECTION BIAS, INFORMATION BIAS & MEASURMENT BIASES CASES CONTROLS AVOID CONFOUNDING BIAS . MATCHING OF CASES WITH CONTROLS FIND EXTENT OF CAUSE IN CONTROLS b / b+d FIND EXTENT OF CAUSE IN CASES a/a+c
12
All cases diagnosed in the community
Sources of cases Sources of controls All cases diagnosed in the community All cases in a segment of population All cases diagnosed in all hospitals All cases diagnosed in single hospital Any of the above method Sample of general population Non cases in the same segment of population Sample of patients in all hospitals (Non cases) Sample of patients in same hospital (Non cases) Spouses, siblings, or associates of cases
13
CASES SELECTION Study begins with cases, i.e. The patients in whom the disease has already occurred. Cases should be representative of all patients with the disease Usually new cases (incident cases) will be chosen. Their past records, registers, case sheets etc were searched and analyzed for the information about the exposure and its duration In a similar fashion, the same information was also obtained from the controls and The risk in both groups was compared.
14
The new cases, which are similar clinically, histologically, pathologically and in their duration of exposure (stage) will be chosen to avoid any error and for better comparison. The past records, registers, case sheets etc are searched and analyzed for the information about the exposure and its duration
15
SELECTION OF CONTROLS Number of controls taken may be equal, twice or four times to that of cases. The ratio depends upon the availability of suitable matched controls, available time and finances. Equal number, if they are better matched, are enough to conduct the study. Control group or comparison group must be very carefully chosen otherwise validity of the study will be defective. Controls should be representative of all the people with out the disease Number of controls taken may be four times, twice or even equal to that of cases. The ratio depends upon the availability of suitable matched controls, available time and finances. Equal number, if they are better matched, are enough to conduct the study.
16
MATCHING Matching is a comparative technique of neutralizing all other variables present in cases and controls, except the variable (disease) under study, to eliminate the systematic errors (biases) while conducting the study. This improves the efficiency of comparison and also the validity of the study by avoiding errors. Cases are made identical with controls i.e. all the known variables of the cases like age, sex, occupation, social status etc are neutralized with those of controls except the factor (disease) under study. For example, cases and controls of same age, same sex, same occupation and living same village may be selected for better comparison, if feasible. “ Like should be compared with the like” principle is to be followed. When every thing remains common and same for both the groups and if the cause is found excessive in the cases than in controls, then it is presumed to cause the disease in the cases.
17
MATCHING This improves the efficiency of comparison and also the validity of the study by avoiding errors. For example, cases and controls of same age, same sex, same occupation and living same village may be selected for better comparison, if feasible. “Like should be compared with the like” principle is to be followed.
18
Type of Matching Matching may be of two types: Group matching
Individual matching
19
Group Matching (frequency matching):
Controls are selected in such a manner that the proportion of controls with a certain characteristic is identical to the proportion of cases with the same characteristic. Thus, if 25% of the cases are females, the controls will be selected so that 25% of that group is also females.
20
Individual Matching (matched pairs):
For each case, selected for the study, a control is selected who is similar to the case in terms of the specific variable or variables of concern. for example, if the case enrolled in a study is a 45-year-old educated female, control also will be a 45-year-old educated female.
21
LIMITATION OF MATCHING
By matching, we can match only the known confounding variables like age, sex, occupation etc, but not the unknown confounders playing a role in causation. This limitation must be borne in mind and search must be continued to find and neutralize the unknown variables also to improve the validity. Randomization solves this problem of neutralizing unknown variables. To find a matched pair is difficult
22
ENQUIRY ABOUT EXPOSURE
After the cases and controls are selected Now collect Information in both groups in a similar manner. Searching the available records with regard to the exposure to the suspected cause and its duration.
23
ANALYSIS FOR RISK MEASUREMENT
The proportion of the cause in the cases (a / a+c) In controls ( b / b+d ) are measured and compared. If the former is greater or more than the later, association found between that cause and the disease is considered as a causal one. If the former is greater than the later, association found between that cause and the disease is considered as a causal one.
24
EXAMPLE ORAL CONTRACEPTIVES AND BREAST CANCER
We choose breast cancer patients as cases Compare with normal people or other patients (other than breast cancer patients) as control group Look for use of OC in both groups Both the groups were made identical in all their other characteristics except for the presence of lung cancer, which will be present only in cases. This can be done by proper selection and by matching the known variables in both groups.
25
PROPORTION OF SMOKING IN CASES AND CONTROLS
O C PROPORTION OF SMOKING IN CASES AND CONTROLS cases controls Then, we measure the proportion of smoking in cases and controls and compare . . O Smoking OC OC Smoking If it is significantly more in cases than in control group, causal association is suspected. ARROWS SHOW THE EXTENT OF SMOKING AMONG CASES AND CONTROLS OC
26
2 X 2 table for case-control design
Exposure to OC Breast cancer present (cases) Breast cancer absent (controls) Positive a b Negative c d Total a + c b + d If the proportion of smoking is more in lung cancer patients (cases=a/a+c) than in controls (b/b+d), the lung cancer in the cases group is attributed to the exposure to smoking. If the proportion of OC is more in Breast cancer patients (cases=a/a+c) than in controls (b/b+d), the breast cancer in the cases group is attributed to the exposure of OC
27
Odds Ratio OR = 1 OR > 1 OR < 1
Odds comparison b/w cases and controls Odds of exposure for cases is more than odds of exposure for controls Odds of exposure for cases is equal to the odds of exposure for controls Odds of exposure for cases is less than odds of exposure for controls Exposure as a risk factor?? Exposure increases risk of disease (risk) Exposure is not a risk factor Exposure reduces risk of disease (protective effect)
28
ADVANTAGES AND DISADVANTAGES OF C-C STUDIES
only realistic study design for uncovering etiology in rare diseases. It is preferred even to the cohort and randomized trials in such a circumstance. important in understanding new diseases Reveals the study of several different etiological exposures useful if induction period is long relatively inexpensive and easy to carry out There is no need for follow up and there are no ethical problems. However, the main drawback is the frequent chances of biases occurring. This design, as it is purely an observational one and based more on subjective assessment, cannot confirm the causality and needs further testing by cohort and controlled trials. But it does not mean that it has to be followed always by other studies.
29
Disadvantages: 1. Susceptible to bias if not carefully designed 2. Especially susceptible to exposure misclassification 3. Especially susceptible to recall bias 4. Restricted to single outcome 5. Incidence rates can not be calculated 6. Cannot assess effects of matching variables 7. Selection of appropriate control group is difficult
30
BIASES Systematic errors, or deviation of results or inferences from the truth may arise at any point in the course of study or throughout due to chance. They can arise during selection process of cases and controls or while collecting information from both the groups or while conducting the study or while measuring and analyzing or due to confounding.
31
RECALL BIAS OR MEMORY BIAS
Is the inability on the part of an individual (case or control) to recollect things happened in the past accurately. Similarly, patient may give wrong information or exaggerate to please the investigator. patient who had suffered in the past from severe illness and pain may recollect all the things happened accurately but not the one who had only mild attack patient who had suffered in the past from severe illness and pain may recollect all the things happened accurately but not the one who had only mild attack.
32
INTERVIEWING BIAS Errors can occur while collecting data by interviewing, if the interviewing techniques are not standardized and applied in a similar fashion and for similar duration for all the cases and controls. If the technique differs from a case to another case or from a case to a control, the comparison and the validity of design will suffer. Questionnaire should be easy to apply with simple questions to both groups alike If the technique differs from a case to another case or from a case to a control, the comparison and the validity of design will suffer. Questionnaire should be easy to apply with simple, open-ended questions and without any leading questions and applied to all the cases and controls alike.
33
INFORMATION BIAS Case control design is principally an informative design, in the sense, information regarding cause is obtained from both the cases and controls and compared. Any subjective information obtained from cases or controls is vulnerable for bias and one must be very careful while collecting the information Usually, the information will be obtained both from cases and controls either by means of personal interviews or by verification and analysis of their case sheets and registers etc. While doing so, errors may occur if those methods of interviewing and analysis were not standardized and applied in a similar fashion for both the groups. Improving the methods of collecting the information can minimize information bias. By repeatedly training the interviewers with standardized, simple techniques will minimize this error.
34
MEASUREMENT BIAS Errors usually occur while measuring the exposure factor or the suspected cause. example, while measuring the exposure to smoking , all aspects of the smoking (quality, duration, quantity) should be similarly measured in both groups Measurement bias will creep into the study and spoils it, if it is not measured in a similar manner using similar technique or method both in cases and controls. For example, while measuring the exposure to smoking , cases and controls can be compared only when all aspects of the smoking ( its quality, duration, its quantity etc ) are similarly measured in both the groups.
35
Interviewer’s bias If the interviewer tends to probe for answers more while interviewing cases as compared to interviewing controls, this will lead to information bias. e.g. information about oral contraceptive and DVT.
36
CONFOUNDING BIAS The variable which is capable of causing the effect or disease directly on its own And also indirectly with the association of another factor. Allowing this variable into the study is confounding bias . As diseases are usually multifactorial in causation, several factors (confounders) will participate in a web fashion (indirectly, cumulatively, combinedly, complementing etc.) and they will be operating both in cases and controls. If they are not neutralized both in cases and controls and the confounding bias is allowed to continue, the meaning of comparison between cases and controls looses its significance, thereby the validity of the study HENCE, IT IS IMPERATIVE TO FIND OUT ALL THE CONFOUNDING VARIABLES PLAYING A ROLE IN CAUSAL MECHANISMS AND ELIMINATE OR NEUTRALIZE THEM BEFORE ACCEPTING CAUSAL ASSOCIATION. Hence, it is imperative to find out all the confounding variables playing a role in causal mechanisms and eliminate or neutralize them before accepting causal association.
37
Confounding bias Coffee drinking pancreatic cancer
Exposure outcome confounding variable
38
Confounding bias Exposure outcome confounding variable
39
Confounding bias Coffee drinking pancreatic cancer Cigarette smoking
Exposure outcome confounding variable
40
The confounding factor here could be cigarette smoking, people who drink coffee are more likely to smoke than people who do not drink coffee. It is well known that smoking is associated with CHD. It is thus possible that the relationship between coffee and CHD is actually the reflection of strong association between smoking and CHD. Smoking here confounds the observed association between coffee drinking and CHD.
41
EXAMPLES FOR CONFOUNDING BIAS
Age is a best-known confounder, as by its own increase, it can directly cause the disease and indirectly mingling with other factors related to age. Usually presence of confounders leads to indirect causal associations e.g. Goiter is seen mostly at high altitudes, but actually, the iodine deficiency at high altitudes is the cause of goiter there. similarly, alcoholism is suspected to be the cause of liver cancer but the smoking, which is usually associated with alcoholism may be the confounding variable causing the disease. Known confounders can be neutralized by matching both the groups. Random sampling in which there are equal chances for the confounders to be present in both the groups can minimize unknown ones.
42
Nested case control study
It is a case control study nested in a cohort It decreases the cost and controls the biases specially the recall bias Population COLLECT INITIAL DATA Urine, serum etc FOLLOW UP YEARS Develop disease DO NOT Develop disease CASES CONTROLS
43
EXAMPLES OF CASE-CONTRIL STUDIES
Doll’s study on smoking and lung cancer Thalidomide use by pregnant women and congenital defects in the offspring study Oral contraceptives and thrombo-embolism
44
Oral contraceptives and Thromboembolic disease
Period: 1968 – 1969 Cases: Women admitted in hospitals with venous thrombosis or pulmonary embolism without medical causes. Controls: Women admitted to the same hospital with other disease matched for age, marital status and parity. Conclusion: Users of oral contraceptives were 6 times likely as nonusers to develop thromboembolic disease.
45
Thalidomide Tragedy Study Period: 1958 – 1961 Cases: 46 mothers who delivered deformed babies. Controls: 300 mothers who delivered normal babies. conclusion: 41 out of 46 mothers of deformed babies had thalidomide during their early pregnancy.
46
SUMMARY Less expensive and quicker analytical study to test hypothesis immediately If done carefully and wisely by eliminating biases, it is really valuable for investigating rare diseases. However, the main drawback is the frequent chances of biases occurring due to subjective assessment.
47
HOME WORK Deaths occurred due to Reye’s syndrome in 1980’s. it is a syndrome which occurs when a child recovers from a viral infection such as chicken pox. A case control study was done to test the hypothesis that Reye’s syndrome occurs due to salicylate medication during the treatment of viral infection such as chicken pox. Out of 66 cases 43 were exposed and out of 120 controls 54 were exposed. 1. who are your cases and controls? How will you select? [4] 2. what is your exposure? [2] 3. what is the outcome of interest, calculate and interpret? [4]
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.