Epidemiology module of practical skills for mbbs student

Evaluation of Study Designs

Study 1

Study 2

Study 3

Study 4

Study 5

Study 6

Is this study design experimental?

Is it a cross sectional study?

Is this a prospective cohort study?

Is this a retrospective cohort study?

Is this a case-control study?

Explain why you selected this study design:

Study 1

Study 2

Study 3

Study 4

Study 5

Study 6

Practical exercise

Case-control and Cohort Studies

A causal relationship between cigarette smoking and lung cancer was first suspected by clinicians in the 1920s on the basis of clinical observations. To test this apparent association Richard Doll and Austin Bradford Hill carried out two major studies. . The first was a case-control study, begun in 1947, in which the smoking habits of lung cancer patients were compared with the smoking habits of other patients. The second was a cohort study, begun in 1951, which related causes of death among doctors to their previously recorded smoking habits.

Part I: Case-control study

Data were obtained from patients in hospitals in and near London, over a 4-year period. Initially, 20 hospitals, and later more, were asked to notify the investigators of all patients admitted with a diagnosis of lung cancer. These patients and controls selected from in-patients with other disorders (primarily non-malignant) in the same hospitals at the same time were then interviewed about their smoking habits.

A total of 1,465 cases of lung cancer, all under age 75, were included in the study. These patients were culled from a larger number of which about 15% were not interviewed because of death, discharge, severity of illness, or inability to speak English.

Diagnoses of lung cancer in nearly all cases were based on biopsy, autopsy, sputum cytology wih or without bronchoscopic or radiographic evidence.

Interviews were conducted by four full-time social workers. Each worker, after interviewing a patient with lung cancer, interviewed a control patient of the same gender and the same 5-year age group. Table 1 compares cases and controls in terms of age and gender.

Table 1. Age and gender of cases and controls

	Cases		Controls
Age group	Men	Women	Men	Women
25-34	17	3	17	3
35-44	116	15	116	15
45-54	493	38	493	38
55-65	545	34	545	34
65-84	186	18	186	18
Total	1,357	108	1,357	108

Table 2. Number and per cent of male cases and controls by amount smoked

Number of daily cigarettes	Cases		Controls
	Number	%	Number	%
0	7	0,5	61	4,5
1-4	49	3,6	91	6,7
5-14	516	38,0	615	45,3
15-24	445	32,8	408	30,1
25-49	299	22,0	162	11,9
50+	41	3,0	20	1,5
Total	1,357	99,9	1,357	100,0
All smokers	1,350	99,4	1,296	95,5

The study showed a clear association between smoking and lung cancer(Table 2), but differences between case and control patients, particularly in terms of questionnaire response, might make the association spurious. Did lung cancer patients, knowing they had lung cancer, tend to exaggerate their smoking habits? Did controls, knowing smoking is ‘bad for you’ tend to under-report their smoking? Did interviewers, through prior knowledge of the hypothesis tend to exaggerate the smoking habits of lung cancer patients and minimize the smoking of controls? Possible answers to these questions came from patients initially interviewed as presumed lung cancer cases but later found not to have the disease. Table 3 compares the smoking habits of men aged 45-74 in this and other diagnostic groups.

Table 3. Percentage distribution of daily smoking habits for selected disease groups (standardized to age distribution of population of England and Wales)

Average daily number of cigarettes
Disease group	0	1-4	5-14	15-24	25+	Number of patients
Incorrectly presumed to have cancer	5,3	9,9	35,5	37,8	11,4	202
Confirmed lung cancer	0,3	4,6	35,9	35,0	24,3	1,224
Other lung diseases	1,9	9,9	38,3	38,7	11,2	301
Other cancers	4,6	9,4	47,2	26,0	12,8	473
Other diseases	5,6	9,0	44,8	26,9	13,7	875

Questions re Case-control Study

1. 
   How representative of all persons with lung cancer are cases admitted to hospital likely to be?
   

2. 
   How representative of all persons without lung cancer are other patients in hospital likely to be?
   

3. 
   What biases may have been introduced by failure to interview 15% of the cases?
   

4. 
   Why did the investigators require that control patients be of the same gender and age group as case patients?
   

5. 
   What do the data of table 2 suggest about the relationship between cigarette smoking and lung cancer?
   

6. 
   Compute the odds ratio for lung cancer occurring in all smokers compared with non-smokers. What does this represent? What does it suggest with regard to the etiological role of cigarette smoking in lung cancer?
   

7. 
   If the ‘control’ patients were asthma patients, would this change your opinion about the etiological role of cigarette smoking in lung cancer based on the results of this study?
   

8. 
   What does table 3 tell us? Why was age standardization of the data required?