You are on page 1of 8

From each these studies one can show: whether any association exist between the suspected factor

or and the disease of the hypothesis if so, what is the strength of the association between the suspected factor and the disease under study

Case-Control study Study-Design It is a study between the two groups, one group of persons having a particular disease under study called "cases" and another group of persons called "controls" who are all comparable with cases in respect of age, sex, literacy level, occupation, marital status, socio-eonomic status but free from the disease under study. the control group is taken for the purpose of comparison of observations. Study is now made by obtaining information from each member of both the groups, about the exposure to the suspected factor made in the hypothesis. The information may be obtained either by interview method, or questionnaire method or by E-mail, etc. Suppose the hypothesis is, "Smoking 20 cigarettes per day over 20 years, result in lung cancer," or alcoholism results in cirrhosis of liver or oral pills results in cancer of the breast, etc. the study proceeds backwards to know the history of exposure to the suspected factor under study. Then the "Exposure rate" (or Frequency of exposure) is calculated in both the groups and compared. If the exposure rate is more among the cases than among the controls, an association is said to exist between the suspected risk factor and the disease (e.g. smoking and lung cancer; alcoholism and cirrhosis of liver). Steps in case-control study Selection of cases Selection of controls Matching Measurement of exposure among both groups Analysis

Selection of Cases There are two specifications to be satisfied, i.e. Diagnostic criteria and eligibility criteria.

a. Diagnostic criteria: This must be specified before the study is undertaken. For example, for the diagnosis of lung cancer, all cases should have malignant cells in sputum or similar radiological lesions. Once the criteria is established, it should not altered. b. Eligibility criteria: Usually employed criteria is that only New cases are eligible for the study and not the old ones or advanced cases. c. Size: The size of the sample of cases is estimated by the particular formula and selection is done by a particular sampling procedure/method (explained in statistics). d. Sources: The cases may be drawn either from the hospital or from the general population. All the new cases should be collected during a specified period of time. Selection of Controls This group should have all the features as those of cases, except that they do not have the disease under study. (explained already). a. Sources:They are drawn from the sources like hospital, relatives, neighborers or general population. The control group drawn from the hospital, may be having a different disease but does not matter. Ex: I a study of cancer cervix, the control group would be women admitted either for delivery or other gynecological problems. The control group drawn from the relatives has an advantage of sharing common attributes and easy accessibility. The control group can also be selected from the same locality or from the same industry or from the same school as that cases. b. Size of the control group: The sixe of the control group depends upon the size of the cases. If the size of the cases (i.e. study group) is small (say less than 50), then the size of the control should be double, triple or even four times that of the study group. Thus selection of cases and controls is crucial. Failure to select a comparable group can introduce Bias in the results. Matching It is a process of selection of controls in such a way that not only they resemble cases (study group) in all the attributes and free from the disease under study but also free from the Confounding factor . A confounding factor is a one which is not only associated with exposure under investigation, but also independently can act a risk-factor for the development of the disease under study. For example, in a study of role of alcohol in the etiology of esophageal cancer, smoking is a confounding factor. That means smoking, which is invariably associated with alcoholism can independently result in cancer of esophagus.

For another example, in study of role of oral contraceptive pills in the etiology of cancer breast if the women taking pills (cases) are of younger age group and if the control group belong to higher age group and not taking the pills, Age is a confounding factor because as the age advances, the women are otherwise at a risk of Ca-breast. So to remove the confounding factor, control group should also belong to the same age group as the cases. Matching Procedures There are two procedures. One is Group matching, wherein the control group is similar to study group (cases) in all the attributes and free from the disease and also the confounding factor, as already explained. The other is Paired matching wherein the control group is resembling (matching) the study group very closely. Ex: In twin studies of monozygotic twins, if one of the twins is a case, the other one is selected as control. Measurement of Exposure in Both the Groups This consists of collection of data among both cases and control groups about their exposure to the suspected cause with reference to duration and frequency of exposure. This is done in the following dimensions depending upon the hypothesis formulated. How many of both the groups had exposure to the suspected factor? Ex : In the study of lung cancer, how many of them were smoking? What was the duration of exposure?, i.e. since how long were they smoking? What is the frequency of exposure?, i.e. how many cigarettes per day? Such informations can be obtained either by questionnaires, interviews or from thr hospital records. Etc.

Analysis The data is analysed to find out: a. Whether any association exist between the disease and the suspected factor; b. If so, what is the strength of the association; a. To know the existence of the relation between the disease and the suspected factor, exposure rate (i.e. frequency of exposure) has to be estimated in both the groups and compared. Framework of Case-Control Study Suspected factor (H/o exposure to Risk factor) Present Absent Cases (Disease present) a c Controls (Disease abset) b d

Total

(a + c)

(b + d)

Exposure rate (Frequency of exposure) among cases = = Exposure rate among controls = = If exposure rate among cases ( x 100

x 100 x 100

x 100 ) is greater than that of controls ( ),

that means there is the existence of the association between the disease and th suspected factor (i.e. Cancer lung and smoking). b. To know (measure) the strength of the association between the disease and the suspected factor, an indicator called Odds ratio (i.e. cross-product ratio) has to be calculated. It is the ratio of chance of exposure among cases to the chance of exposure among controls. Using the above framework, Odds ratio = Higher the value of odds ratio, greater is the strength of the association between the disease and the suspected factor. To ascertain whether the relation/association between the disease (lung cancer) and suspected factor (smoking) is statistically significant or not, a test of significance called Chi square is applied, to calculate p-value. p-value indicates whether the probability of association between the disease and the suspected factor has occurred by chance or by real fact. If the p-value is 0.005 or less than that the relation is considered as significant. Smaller the p-value, greater is statistical significance, p-value lesser than 0.001 is highly significant. Thus lesser the p-value, higher the significance that means the probability of occurrence of the disease by chance is ruled out. In otherwords the relation between the disease (lung cancer) and suspected factor (smoking) is a fact. Only if it is a fact, then it can be applied (projected) to the population at large stating that cancer lung results from smoking. (Explained in detail under Biostatistics). Example: On interrogation, 40 out of 50 cases of lung cancer and 60 out of 150 controls gave the history that they were smoking cigarettes. Does smoking predispose to lung cancer. The study tests the hypothesis Smoking Predisposes to Lung Cancer Exposure Cases Controls

Yes No Total

40 (a) 10 (c) 50 (a+c)

60 (b) 90 (d) 150 (b+d)

Exposure rate among cases =

x 100 =

x 100

= 80 percent Exposure rate controls = x 100 = = 40 percent Since the exposure rate is more among cases than among controls (80% > 40%) the relation between smoking and lung cancer exist. Odds ratio = = = =6

That means smokers are 6 times at a greater risk of getting lung cancer. (This indicates the strength of the association). Cohort Study This is also an another type of analytical study undertaken to support (accept) or refute the association between the suspected factor and the disease of the etiological hypothesis. Cohort study is so called because it is done on Cohorts. A Cohort is a group of persons possessing common characteristics. Ex: Those born during particular time constitute Birth Cohort, those married during a particular period constitute Marriage cohort, those to a common drug or a vaccine are called Exposure cohort, etc. Study Design This is also a study involving two groups of persons, one group called Study cohort, (Exposure cohort) who share common characteristics (like age, sex, occupation income, literacy level, marital status, etc) and are exposed to the suspected etiological factor and another group called Control cohort, who also have the same attributes as that of study cohort but are not exposed to the particular factor and are used for comparison of observations. Both the groups are then observed (followed) over a period time, under the same identical conditions, to determine the Incidence (frequency) of disease in both the groups and compared. Thus the study proceeds forwards from cause to the effect, unlike case-control study which proceeds backwards from effect to the cause. The synonyms for cohort study are Follow-up study, Forward looking study, Prospective study, Longitudinal study, and Incidence study.

Incidence rate of the disease

x 1000

among study cohort (Frequency)

Incidence rate of disease

x 1000

among control cohort (Frequency)

If the incidence of the disease is more in the study (exposed) cohort than the control cohort, it suggests that there is an association between the cause and the effect. Types of Cohort Studies There ar three types namely: a. Prospective cohort study b. Retrospective cohort study c. Combination of prospective and retrospective cohort study. Prospective Cohort Study This is the commonest and most scientific method of cohort studies. It begins in the present and continues in the future and then terminates. It is also called as Current cohort study. Famous example are Doll and Hills study on smoking and lung cancer, US Public Health Services Framingham Heart Study and study of oral contraceptives and health by the Royal College of General Practitioners. In this type, the exposure and the study for outcome occurrence start simultaneously. Retrospective Cohort Study (Historical Cohort Study) In this type the investigator goes back in time for about 15 to 20 years, to select a study group from the existing records. Thus exposure has occurred sometime back in the past but the disease (outcome) has not occurred and traced forward upto the present time for the occurrence of the

outcome (disease). It is also called as Non-current prospective study. The control cohort is also selected from the records and followed through the records for the outcome but they are not exposed to the factor. These studies are not only economical but also produce results more quickly than prospective cohort studies. Schematic Representation of Case-Control Study and Cohort Studies

Past

Present

Future

Case-control study (Retrospective study) Cases Study from effect to cause for the presence or absence of factor Prospective (Cohort) Study Controls Study cohort Study from cause to effect for the presence or absence of the disease

Retrospective Cohort Study

Controls cohorts

Study cohort Followed through the records upto present time Controls cohorts Combination of retrospective and prospective cohort study Study cohort Followed through the records upto present time and followed Controls into future and Prospective Cohort Studies Combination of Retrospective cohorts

In this type, cohort and control groups are identified from the past records and are followed-up prospectively upto present time and continued to follow up into future. Steps in Cohort Study a. b. c. d. Selection of study cohorts Selection of control cohorts Followed-up Analysis

Selection of Study Cohorts (or Exposure Cohorts) The estimated sample size (usually large in number) having common characteristic features, is obtained either from the general population as marriage cohort, birth cohort, exposure cohort, etc or from a special homogeneous group of population such as doctors, lawyers, nurses, industrial workers, etc. who are not only easily accessible but also can be followed-up easily. (Reference population is a one to which the results are applicable). Selection of Control Cohorts (or Unexposure Cohorts) This group is selected from the reference population. All of them have the same characteristic features such as age, sex, occupation, literacy level, socioeconomic status, etc. but not exposed to risk factor.

You might also like