Chapter 11 Exercises 

11.1 Descriptive epi, tuberculosis rates. Go to wonder.cdc.gov and search the site for current tuberculosis data that can be used to address these descriptive analyses:

(A) Time trend. Determine tuberculosis morbidity rates over the past 50 years. Plot the rates over time (Y axis = "rate per 100,000 persons per year"; X-axis = year). In plain terms, describe the observed trends and hypothesize things that might have influenced the reported trends (e.g., hypothesize why there was a small increase in the mid-1970s and again in the early-1990s). Search for information that corroborates your hypotheses.
(B) Age-specific rates. Identify and report TB morbidity rates by age. Plot age-specific TB rates as separate lines on a single graph. Discuss trends.
(C) Race-specific rates. Identify and report TB  morbidity rates by race. Plot these rates on a single graph. Then, summarize your findings.
(D) Foreign and native incidents. Identify and report TB morbidity rates by country of origin (US Born vs. Foreign born). What percentage of cases are currently in Foreign-born persons? What country provides the largest number of Foreign-born cases? Present your analysis in graphical form--perhaps in the form of a pie chart. 
(E) Summary. Summarize what you've learned.

11.2Injuries in farm worker. Read this abstract: McGwin et al. (Am J Epidemiol 2000;152:640-50)  (link is active). Then, answer these questions:

(A) What are the exposures in this study? What is the disease? 
(B) Explain why experimentation is impossible when addressing this issue.  
(C) Explain why this study is observational (i.e., nonexperimental). 
(D) Are data prospective or retrospective?
(E) What measures of association can be estimated from these data? 
(F) Summary data are shown below. Calculate the rates and RRs of injury in each group. Is there a risk associated with race? Is there a risk associated with being a worker? 

Group (number)	Cases	Person-years
Caucasian Owners (0)	67	2047
Af-Am. Owners (1)	27	821
Af-Am. Workers  (2)	37	359

11.3  Influenza vaccination and primary cardiac arrest. A case-control study examined the relation between influenza vaccination and primary cardiac arrest (PCA) (Siscovick et al. 2000). Incident cases of PCA (n = 315) were identified from paramedic reports. Community controls (n = 549) were identified using a random digit dialing technique. Spouses of cases and controls were interviewed to ascertain who had received influenza vaccination during the prior year. Data were:

(A) Explain why this is a case-control study and not a cohort study.
(B) Calculate the odds ratio associated with vaccination. Interpret the results. What does this suggest about the effects of influenza vaccination on PCA risk?
(C) The investigators relied on information from surrogates (spouses) to ascertain the exposure status of cases and controls. Although this was necessary for cases (deceased), it was not necessary for controls. Suggest are reason surrogates used to collect exposure information from the controls.

11.4  Case-control study of pancreatic cancer and meat consumption (Norell et al. 1986; Ahlbom & Norell, 1990, p. 52). Data from a case-control study of pancreatic cancer found these data for eating fried or grilled meats at least once per week. Calculate the odds ratio and discuss the results. 

11.5 Doll & Hill, 1950. An early case-control study of smoking and lung cancer found these data (Doll & Hill, 1950). Calculate the odds ratio and interpret the results. 

11.6 IUDs and infertility.  A study of infertility found prior use of IUDs in 89 of 283 infertile women. In contrast, 640 out of 3833 controls (fertile women) had used IUDs in the past (Cramer et al., 1985; Rosner, 1990, p. 381). Put the data in 2-by-2 table form, determine the odds ratio, and interpret your results.

11.7 OC studies from the 70s. A causal relation between oral contraceptive use and cardiovascular disease was first postulated in the late 1960s. Numerous studies have since confirmed the relationship. Abstracts from two early epidemiologic studies on this topic are shown below. Read these abstracts and answer the questions that follow.

Study 1 (Mann et al., 1975): Mann, J. I., Vessey, M. P., Thorogood, M., & Doll, S. R. Myocardial infarction in young women with special reference to oral contraceptive practice. Br Med J, 2(5965), 241-245; 1975. Sixty-three women discharged from hospital with a diagnosis of myocardial infarction [heart attack] and 189 control patients were studied. All were under 45 years of age at the time of admission. Current oral contraceptive use, heavy cigarette smoking, treated hypertension and diabetes, pre-eclamptic toxaemia, and obesity were all reported by, and type II hyperlipoproteinaemia was found more often in, patients with myocardial infarction than their controls. The relationship between myocardial infarction and oral contraceptives could not be explained in terms of an association between the use of these preparations and the other factors. The combined effect of the risk factors was clearly synergistic.

Study 2 (Royal College, 1977): Royal College of General Practitioners' Oral Contraception Study. Mortality among Oral-Contraceptive Users. Lancet Oct 8;2(8041):727 - 31; 1977.
In a large prospective study carried out in the United Kingdom, the death-rate from diseases of the circulatory system in women who had used oral contraceptives was five times that of [women] who had never used them; and the death-rate in those who had taken the pill continuously for 5 years or more was ten times that of the [non-users]. The excess deaths in oral-contraceptive users were due to a wide range of vascular conditions. The total mortality-rate in women who had ever used the pill was increased by 40%, and this was due to an increase in deaths from circulatory diseases of 1 per 5000 ever-users per year. The excess was substantially greater than the death-rate from complications of pregnancy in the [non-users], and was double the death-rate from accidents. The excess mortality-rate increased with age, cigarette smoking, and duration of oral contraceptive use.

(A) What makes the first study a case-control study?
(B) What makes the second study a cohort study?
(C) The hospital was chosen as the source of cases and controls in Mann et al. (1977). Hospitals allow for access to a "captive audience," accurate diagnoses, and all sorts of medical information. In theory, the cases and controls should be a random sample of the catchment are of the hospital. What potential problems do you see in using hospitalized controls for this study?  
(D) The abstract for Royal College (1977) fails to report the actual mortality rates in the various group but notes that the "total mortality-rate in women who had ever used the pill was increased by 40%." Based on this statement, what is the rate ratio associated with oral contraceptive use? 
(E) The Royal College study (1977) also states that the increase "was due to an increase in deaths from circulatory disease of 1 per 5000 ever-users per year." Does this statistic represent a  rate ratio, rate difference, or attributable fraction? Explain.
(F) Comparison of case-control and cohort designs:
    (1) Which study allowed for a smaller total sample size?
    (2) Assuming the cohort study was not retrospective, which study took less time to complete?
    (3) Can case-control studies estimate the incidence of disease? Explain.
    (4) Which study, Mann et al. or the Royal College of Practitioners, was probably less prone to recall bias?
    (5) Which was less prone to selection bias?
    (6) Which was less prone to loss to follow-up?
    (7) Why do you think the case-control study done before the cohort study?

11.8 Breast Cancer in Canada (descriptive epi). Complete the "Breast Cancer in Canada" case study posted on http://www.math.mcmaster.ca/peter/sora/case_studies_99/breastcancer.html .

11.9 Descriptive epi project. Descriptive epidemiologic studies  may use routinely collected statistics to the clarify patterns of disease by person, place, and time factors. This exercise will give you a little experience using such sources. Go the the CDC Wonder Home page http://wonder.cdc.gov/. Select a cancer outcome and explore its data source. Using these data sources, make the following comparisons: 

(A) Regional comparisons. Does the risk of dying from the outcome differ by state or region (after adjusting for age and other population differences)?
(B) Temporal trends. What are the trends over time? In which regions are the rates increasing? Where are they decreasing?
(C) Age-specific trends. Compare age-specific trends. Are risks increasing for some age groups but not for others?
(D) Age-and cohort effects. Attempt to discriminate between trends due to changes in risk factors and changes in diagnostic practices and registration procedures. Also, look for birth cohort effects. 

Key

Last update: 12/07/2008