Centers for Disease Control and Prevention Epidemiology Program Office Case Studies in Applied Epidemiology No.

731-703

Cigarette Smoking and Lung Cancer
Student's Guide
Learning Objectives After completing this case study, the participant should be able to:

G Discuss the elements of study design, and the advantages and disadvantages of case-control versus prospective cohort studies; G Discuss some of the biases that might have affected these studies; G Calculate a rate ratio, rate difference, odds ratio, and attributable risk percent; G Interpret each measure and describe each measure's main use; and G Review the criteria for causation.

This case study is based on the classic studies by Doll and Hill that demonstrated a relationship between smoking and lung cancer. Two case studies were developed by Clark Heath, Godfrey Oakley, David Erickson, and Howard Ory in 1973. The two case studies were combined into one and substantially revised and updated by Nancy Binkin and Richard Dicker in 1990. Current version updated by Richard Dicker with input from Julie Magri and the 2003 EIS Summer Course instructors.

U.S. DEPARTMENT OF HEALTH AND HUMAN SERVICES Public Health Service

and later more. Question 1: What makes the first study a case-control study? The first study in a case-control study because they first identified a group of individuals with the disease (lung cancer) and the controls were patients with other disorders without lung cancer. Information about lung cancer came from death certificates and other mortality data recorded during ensuing years. These patients were then interviewed concerning smoking habits. This case study deals first with the case-control study. The first was a case-control study begun in 1947 comparing the smoking habits of lung cancer patients with the smoking habits of other patients. then with the cohort study. Data for the case-control study were obtained from hospitalized patients in London and vicinity Page 2 over a 4-year period (April 1948 . numerous epidemiologic studies were undertaken between 1930 and 1960.CDC / EIS Summer Course 2003: Smoking and Lung Ca . The remainder of Part I deals with the case-control study. more accurate exposure recall. and easy record access. Two studies were conducted by Richard Doll and Austin Bradford Hill in Great Britain. were asked to notify the investigators of all patients admitted with a new diagnosis of lung cancer. Initially. as were controls selected from patients with other disorders (primarily non-malignant) who were hospitalized in the same hospitals at the same time. 20 hospitals. To test this apparent association. . unless death occurs. The second was a cohort study begun in 1951 recording causes of death among British physicians in relation to smoking habits. Information about present and past smoking habits was obtained by questionnaire.February 1952). less loss to follow up. Question 2: What makes the second study a cohort study? The second study a cohort study because exposure is ascertained from past records and we compare exposed to unexposed individuals. Data for the cohort study were obtained from the population of all physicians listed in the British Medical Register who resided in England and Wales as of October 1951.Student's Guide A causal relationship between cigarette smoking and lung cancer was first suspected in the 1920s on the basis of clinical observations. Question 3: Why might hospitals have been chosen as the setting for this study? Easy access to compliant patients.

thus affecting the evaluation of exposure Question 7: How representative of the general population without lung cancer are hospitalized patients without lung cancer? Since hospital based controls are ill. You may risk only treating patients with severe lung cancer. some patients may not be smokers because they have a disease that prevents them from being so. they may not accurately represent the exposure history in the population that produced the cases. friends Page 3 Question 5: What are the advantages of selecting controls from the same hospitals as cases? It would deal with the problem that certain characteristics are particular to a hospital.CDC / EIS Summer Course 2003: Smoking and Lung Ca . Question 8: How may these representativeness issues affect interpretation of the study's results? Since possibly only extreme cases of lung cancer may be assessed because we are choosing from patients in the hospital. For example. it may cause the disease-exposure association to seem to be less significant than it actually is. Question 6: How representative of all persons with lung cancer are hospitalized patients with lung cancer? Not all people with lung cancer are hospitalized. .Student's Guide Question 4: What other sources of cases and controls might have been used? General population: Neighborhood. and missing out on those with an early form of lung cancer.

Table 1. Odds of smoking among cases = a/c = 1350/7 = 192. Great Britain. Proportion smoked.995 and the prevalence of exposure in undiseased was 0. Odds of smoking among controls = b/d = 1296/61 = 21.350 7 1. 1948-1952. or inability to speak English. An additional group of patients were interviewed but later excluded when initial lung cancer Page 4 diagnosis proved mistaken. cases: 1350/1357 = 0.357 Question 9: From this table. discharge.955. About 15% of these persons were not interviewed because of death.CDC / EIS Summer Course 2003: Smoking and Lung Ca .700 patients with lung cancer.Student's Guide Over 1. .465 cases (1.357 Controls 1.86 Question 11b: Calculate the odds of smoking among the controls.995 Proportion smoked. calculate the proportion of cases and controls who smoked.24 Question 12: OR = ad/bc = 9 Calculate the ratio of these odds. were eligible for the case-control study.955 Question 10: What do you infer from these proportions? The prevalence of exposure in diseased were 0. Question 11a: Calculate the odds of smoking among the cases. severity of illness.296 61 1.357 males and 108 females). all under age 75. The following table shows the relationship between cigarette smoking and lung cancer among male cases and controls.24 = 9 They are the same. lung cancer cases and matched controls with other diseases. The final study group included 1. Smoking status before onset of the present illness. Cases Cigarette smoker Non-smoker Total 1. controls: 1296 / 1357 = 0.86 / 21. How does this compare with the cross-product ratio? Or cases / OR controls = 192. The prevalence of smoking in those who had lung cancer were higher.

296 1. Great Britain.27 9. Question 15: Interpret these results.Student's Guide Question 13: Page 5 What do you infer from the odds ratio about the relationship between smoking and lung cancer? The exposure (smoking) is positively related to the disease (lung cancer) Table 2 shows the frequency distribution of male cases and controls by average number of cigarettes smoked per day.9 9. the more positively associated to lung cancer the individual becomes. Table 2. lung cancer cases and matched controls with other diseases. Most recent amount of cigarettes smoked daily before onset of the present illness.357 Odds Ratio referent 6. cigarette smoking is positively associated with lung cancer. comparing each smoking category to nonsmokers. The more number of cigarettes smoked daily.5 16.350 1. . In general.357 # Controls 61 706 408 182 1. Daily number of cigarettes 0 1-14 15-24 25+ All smokers Total # Cases 7 565 445 340 1.07 Question 14: Compute the odds ratio by category of daily cigarette consumption. 1948-1952.CDC / EIS Summer Course 2003: Smoking and Lung Ca .

it is a a highly stressful job. Question 16: What are the other possible explanations for the apparent association? Those who are heavy smokers may also be prone to engaging in cancer prone activities such as heavy drinking or in an occupation that causes stress. their method of smoking. Question 17: How might the response rate of 68% affect the study's results? Perhaps only a certain type of physician would want to answer a questionnaire about cigarette smoking. Nonsmokers were defined as persons who had never consistently smoked as much as one cigarette a day for as long as one year. Data for the cohort study were obtained from the population of all physicians listed in the British Medical Register who resided in England and Wales as of October 1951. if they had stopped smoking.600 physicians. The next section of this case study deals with the cohort study.637 (68%) physicians.Student's Guide Although the study demonstrates a clear association between smoking and lung cancer. Usable responses to the questionnaire were received from 40. It may be a touchy subject for physicians.CDC / EIS Summer Course 2003: Smoking and Lung Ca . The questionnaire asked the physicians to classify themselves into one of three categories: 1) current smoker. .192 were females. or 3) nonsmoker.445 were males and 6. Or that the busiest physicians (thus the most prone to smoking) would be too busy to answer the questionnaire. to 59. for example. how long it had been since they last smoked. so the driver would want to smoke. of whom 34. 2) ex-smoker. Smokers and ex-smokers were asked the amount they smoked. but lung cancer could also be due to the exposure to air pollutants. In the case of a truck driver. and. Questionnaires were mailed in October 1951. the age they started to smoke. Page 6 cause-and-effect is not the only explanation.

Student's Guide The next section of this case study is limited to the analysis of male physician respondents. autopsy.14 19. The following table shows numbers of lung cancer deaths by daily number of cigarettes smoked at the time of the 1951 questionnaire (for male physicians who were nonsmokers and current smokers only).100 102. Diagnoses of lung cancer were based on the best evidence available.800 38. in 4 of the 157 cases this diagnosis could not be documented. bronchoscopy.2 Deaths from lung cancer 3 22 54 57 133 136 Personyears at risk 42.30 0.27 Rate Ratio referent 8. Of 4. Rate difference tells us the attributable risk.57 13. physical examination.600 38.87 Question 18: Compute lung cancer mortality rates. Table 3.400 Mortality rate per 1000 person-years 0. or death certificate. 35 years of age or older.32 2. or sputum cytology (combined with bronchoscopy or X-ray evidence).597 deaths in the cohort over the 10-year period. 1951-1961. 157 were reported to have been caused by lung cancer.07 0.000 person-years) of lung cancer deaths by number of cigarettes smoked per day. For each death attributed to lung cancer.42 1. about 70% were from biopsy.900 25. Daily number of cigarettes smoked 0 1-14 15-24 25+ All smokers Total Rate difference per 1000 person-years referent 0. Person-years of observation ("person-years at risk") are given for each smoking category. leaving 153 confirmed deaths from lung cancer.CDC / EIS Summer Course 2003: Smoking and Lung Ca . or X-ray alone. and rate differences for each smoking category.39 2. The number of cigarettes smoked was available for 136 of the persons who died from lung cancer. and only 1% were from just case history. rate ratios.5 1. What do each of these measures mean? Mortality rates tell us the number of people who die from lung cancer in a total population at risk Rate Ratio tells us that if the ratio is greater than 1.86 21. . the risk of disease was __ times the risk of disease in those who were not exposed. 29% Page 7 were from cytology. All certificates indicating that the decedent was a physician were abstracted.600 145.94 18. The occurrence of lung cancer in physicians responding to the questionnaire was documented over a 10-year period (November 1951 through October 1961) from death certificates filed with the Registrar General of the United Kingdom and from lists of physician deaths provided by the British Medical Association.43 1. then in the exposed.23 0.57 1. Doll and Hill physician cohort study. Great Britain. Number and rate (per 1. medical records were reviewed to confirm the diagnosis.

000 person-years Smokers Lung cancer Cardiovascular disease 1.Incidence of unexposed) / Incidence of exposed = (1. how many deaths from lung cancer would have been averted? 1.95 x 100% = 95% Question 20: If no one had smoked.30 x 100% = 0. 1951-1961. Great Britain.30 9.Student's Guide Question 19: What proportion of lung cancer deaths among all smokers can be attributed to smoking? What is this proportion called? Page 8 ARE% = (Incidence of exposed .Incidence of unexposed = The cohort study also provided mortality rates for cardiovascular disease among smokers and nonsmokers. rate ratios. .CDC / EIS Summer Course 2003: Smoking and Lung Ca .07) / 1.0.94 8.23 deaths per 1000 would have been avoided of no one had smoked.32 All 0.0.5 1.07 7. Mortality rate per 1. Mortality rates (per 1. and excess deaths from lung cancer and cardiovascular disease by smoking status.000 person-years 1.51 Non-smokers 0.30 . Doll and Hill physician cohort study.000 person-years).87 Rate ratio 18. ARE = Incidence of exposed .23 2.07 = 1. Table 4. The following table presents lung cancer mortality data and comparable cardiovascular disease mortality data.3 Excess deaths per 1.19 Attributable risk percent among smokers 95% 23% Question 21: Which cause of death has a stronger association with smoking? Why? Lung cancer has a stronger cause of death associated with smoking than CHD because it has a higher rate ratio.30 .

0.CDC / EIS Summer Course 2003: Smoking and Lung Ca .Student's Guide Page 9 In calculating the attributable risk percent. A similar measure. How do they compare? How do they differ from the attributable risk percent? PAR% LC = (0.55 per 1000 of CHD deaths in the population is attributed to smoking.7.7. Question 23: How many lung cancer deaths per 1. The attributable risk percent of 95% for smoking may be interpreted as the proportion of lung cancer deaths among smokers that could have been prevented if they had not smoked.94 . PAR CHD = 8.32 = 1.07 = 0.6% PAR% CHD = (8.87 per 1000 of lung cancer deaths in the population is attributed to smoking.0. The population attributable risk percent is often used in assessing the cost-effectiveness and costbenefit of community-based intervention programs. the excess lung cancer deaths attributable to smoking is expressed as a percentage of all lung cancer mortality among all smokers.87 . One formula for the population attributable risk percent is: PAR% = (Incidence in entire population ! Incidence in unexposed) / Incidence in entire population Question 22: Calculate the population attributable risk percent for lung cancer mortality and for cardiovascular disease mortality.32) / 8.07) / 0.87 x 100% = 17.94 .87 . the population attributable risk percent expresses the excess lung cancer deaths attributable to smoking as a percentage of all lung cancer mortality among the entire population. .94 x 100% = 92. the population attributable risk percent for a given exposure can be interpreted as the proportion of cases in the entire population that would be prevented if the exposure had not occurred. From a prevention perspective.47 The percent of cases in the population that have lung cancer attributed by smoking is higher than the percent of cases in the populated that have CHD.000 persons per year are attributable to smoking among the entire population? How many cardiovascular disease deaths? PAR LC = 0.

The odds ratios and rate ratios from the two studies by numbers of cigarettes smoked are given in the table below.30 0. Comparison of measures of association from Doll and Hill’s 1948-1952 case-control study and Doll and Hill’s 1951-1961 physician cohort study. The cohort study suggests a greater association than the case-control study.18 0.49 0. indicating that risk in developing lung cancer in smokers is greater than non-smokers. .0 2. Quitting smoking is a great way to prevent lung cancer.5 16.07 Rate Ratio 18. Great Britain. Doll and Hill physician cohort study.0 9. They began their cohort study in 1951.8 32. 1951-1961.4 18.5 Odds ratio from case-control study 1. Great Britain.1 Question 25: Compare the results of the two studies.CDC / EIS Summer Course 2003: Smoking and Lung Ca . Comment on the similarities and differences in the computed measures of association.000 person-years) of lung cancer deaths for current smokers and exsmokers by years since quitting. Daily number of Cigarettes smoked 0 1-14 15-24 25+ All smokers Rate ratio from cohort study 1.6 2. Table 5. and continues to drop as the years quit increase. Rate ratios and odds ratios from both studies are greater than 1.19 0. Doll and Hill began their case-control study in 1947.0 (ref) 8.Student's Guide The following table shows the relationship between smoking and lung cancer mortality in Page 10 terms of the effects of stopping smoking.6 7. Number and rate (per 1.5 9.7 1. years since quitting: <5 years 5-9 years 10-19 years 20+ years Nonsmokers Lung cancer deaths 133 5 7 3 2 3 Rate per 1000 person-years 1.1 19.0 (ref) Question 24: What do these data imply for the practice of public health and preventive medicine? Since the data clearly shows that even after 5 years quitting cigarette smoking.67 0.3 9. the risk of developing lung cancer drops dramatically.0 (ref) 7. As noted at the beginning of this case study. by number of cigarettes smoked daily. Cigarette smoking status Current smokers For ex-smokers. Table 6.

Student's Guide Question 26: Answer 26 Case-control Sample size Costs Study time Rare disease Rare exposure Multiple exposures Multiple outcomes Progression. It is more efficient and less expensive this way. so it makes sense to find diseased cases and compare with the exposure.CDC / EIS Summer Course 2003: Smoking and Lung Ca . spectrum of illness Disease rates Recall bias Loss to follow-up Selection bias Advantage Page 11 What are the advantages and disadvantages of case-control vs. cohort studies? Cohort Disadvantage Advantage Disadvantage Disadvantage Advantage Advantage Disadvantage Question 27: Which type of study (cohort or case-control) would you have done first? Why? Why do a second study? Why do the other type of study? I would have done the case-control study first. This is because lung cancer is generally a rare disease. Question 28: Which of the following criteria for causality are met by the evidence presented from these two studies? Answer 28 Strong association Consistency among studies Exposure precedes disease Dose-response effect Biologic plausibility x x x x x YES NO . You would do the other study to make better measurements and to control for confounding factors.

Smoking and health. 2:739-748. 1964. Hill AB. Brit Med J 1950. 8. 2. U. 21-29. damned lies. and 400. 5. 2:1271-1286.Student's Guide Page 12 REFERENCES 1. PHS Publication No. Doll R.CDC / EIS Summer Course 2003: Smoking and Lung Ca . S. 1:1399-1410. A study of the aetiology of carcinoma of the lung. Public Health Service. Smoking and carcinoma of the lung. Regulation 1998. Education. Marimont RB. 1103. Hill AB. . Brit Med J 1952. 1:1451-1455.000 smoking-related deaths. 3. 6. US Department of Health. PHS. Mortality in relation to smoking: 10 years' observation of British doctors. CDC. Doll R. Report of the Advisory Committee to the Surgeon General of the Public Health Service. Doll R. and Welfare. Lies. Doll R. Doll R. Brit Med J 1964. Hill AB. Hill AB. 4. 1460-1467. The mortality of doctors in relation to their smoking habits. Brit Med J 1954. Hill AB. 2:1071-1081. 7. Hill AB. Brit Med J 1956.58:295-300. The environment and disease: association or causation? Proc R Soc Med 1965. Levy RA. Lung cancer and other causes of death in relation to smoking.

Sign up to vote on this title
UsefulNot useful