Principles of Epidemiologic Research
What is alpha equal to? - correct answer ✔0.05 (one-sided test), 0.1 (two-
sided test)
What is beta equal to? - correct answer ✔0.1
What is power equal to? - correct answer ✔1-0.1
A cohort study of the relationship between smoking and coronary heart
disease (CHD) is planned. A sample of men will be selected at random from
the population and asked to complete a questionnaire. Subsequently they will
be monitored to record outcome or death. The investigator plan to follow the
cohort for 5 years, at which point they would like to be 90% sure of being able
to detect when the risk ratio for smoking is 1.4, using one-sided 5%
significance test. Previous evidence suggests that non-smokers have an
annual death rate from CHD of about 413 per 100,000 per year. Assuming
that equal numbers of smokers and non-smokers are to be sample, how many
should be sampled overall? - correct answer ✔61,896
Alpha = 0.05 (one-sided test ) is equivalent to 0.1 (two-sided test) For further
explanation, please see: https://stats.oarc.ucla.edu/other/mult-pkg/faq/pvalue-
htm/Links to an external site.
Beta = 0.1 and Power = 1 - 0.1 = 0.9
Expected RR = 1.4
Incidence among unexposed = 413/100,000 per year = 5 x 413/100,000 =
0.02065 over 5-years
If the expected RR = 1.4 and the incidence among unexposed over 5-years =
0.02065 then, expected incidence among exposed = 1.4 x 0.02065 = 0.02891
,A case-control study of the relationship between smoking and coronary heart
disease (CHD) is planned. A sample of men with newly diagnosed disease will
be compared for smoking status (smoker vs. non-smoker) with a sample of
controls. Assuming an equal number of cases and controls, how many
individuals are needed to detect an approximate odds ratio of 2.0 with 90%
power using a two-sided 5% test? Prior studies have estimated that 30% of
the male population are smokers. - correct answer ✔376
Alpha = 0.05 (two-sided)
Beta = 0.1; Power = 0.9 (Beta = 1 - .0.1)
Prevalence of exposure (smoking) among controls = 0.3
Expected Odds Ratio = 2.0
Ratio of cases to control = 1 (same number for each)
A case-control study of the relationship between smoking and coronary heart
disease (CHD) is planned. A sample of men with newly diagnosed disease will
be compared for smoking status (smoker vs. non-smoker) with a sample of
controls. Assuming that for every case and 4 controls are to be selected, what
is the appropriate sample size needed to be detect an approximate odds ratio
of 2.0 with 80% power using a two-sided test? Prior studies have estimated
that 30% of the male population are smokers. (zα/2=1.96; zβ=0.84) - correct
answer ✔89 cases
356 controls
445 total
Alpha = 0.05 (two-sided)
Beta = 0.2; Power = 0.8 (Beta = 1 - .0.2)
Prevalence of exposure (smoking) among controls = 0.3
, Expected Odds Ratio = 2.0
Ratio of cases to control = 4 (1 case: 4 controls)
Suppose we wish to undertake epidemiologic studies to assess the effect of
current oral contraceptive (OC) use in relation to risk of myocardial infraction
(MI) among women of childbearing age. From previous studies, we know
about 10% of such women in the United States currently use OCs and the
incidence of MI among non-OC users is 2.1 per 1000. We wish to detect a
relative risk of 1.8 with a conventional alpha level of 0.05 and beta level of
0.20, representing a power of 80% to detect an effect of this magnitude of the
effect if one truly exists. Calculate the sample size for each of the following
study designs with an equal number in each groups. (zα/2=1.96; zβ=0.84)
cohort study
- sample size:
- number in each group:
unmatched case-control study
- sample size:
- number in each group: - correct answer ✔Sample size: 32,606
# in each group: 16,303
Sample size: 814
# in each group: 407
Cohort Study
Alpha = 0.05 (two-sided)
Beta = 0.2; Power = 0.8 (Beta = 1 - .0.2)
Incidence of MI among non-OC users = 2.1/10000 = .0021
Expected Risk Ratio = 1.8
What is alpha equal to? - correct answer ✔0.05 (one-sided test), 0.1 (two-
sided test)
What is beta equal to? - correct answer ✔0.1
What is power equal to? - correct answer ✔1-0.1
A cohort study of the relationship between smoking and coronary heart
disease (CHD) is planned. A sample of men will be selected at random from
the population and asked to complete a questionnaire. Subsequently they will
be monitored to record outcome or death. The investigator plan to follow the
cohort for 5 years, at which point they would like to be 90% sure of being able
to detect when the risk ratio for smoking is 1.4, using one-sided 5%
significance test. Previous evidence suggests that non-smokers have an
annual death rate from CHD of about 413 per 100,000 per year. Assuming
that equal numbers of smokers and non-smokers are to be sample, how many
should be sampled overall? - correct answer ✔61,896
Alpha = 0.05 (one-sided test ) is equivalent to 0.1 (two-sided test) For further
explanation, please see: https://stats.oarc.ucla.edu/other/mult-pkg/faq/pvalue-
htm/Links to an external site.
Beta = 0.1 and Power = 1 - 0.1 = 0.9
Expected RR = 1.4
Incidence among unexposed = 413/100,000 per year = 5 x 413/100,000 =
0.02065 over 5-years
If the expected RR = 1.4 and the incidence among unexposed over 5-years =
0.02065 then, expected incidence among exposed = 1.4 x 0.02065 = 0.02891
,A case-control study of the relationship between smoking and coronary heart
disease (CHD) is planned. A sample of men with newly diagnosed disease will
be compared for smoking status (smoker vs. non-smoker) with a sample of
controls. Assuming an equal number of cases and controls, how many
individuals are needed to detect an approximate odds ratio of 2.0 with 90%
power using a two-sided 5% test? Prior studies have estimated that 30% of
the male population are smokers. - correct answer ✔376
Alpha = 0.05 (two-sided)
Beta = 0.1; Power = 0.9 (Beta = 1 - .0.1)
Prevalence of exposure (smoking) among controls = 0.3
Expected Odds Ratio = 2.0
Ratio of cases to control = 1 (same number for each)
A case-control study of the relationship between smoking and coronary heart
disease (CHD) is planned. A sample of men with newly diagnosed disease will
be compared for smoking status (smoker vs. non-smoker) with a sample of
controls. Assuming that for every case and 4 controls are to be selected, what
is the appropriate sample size needed to be detect an approximate odds ratio
of 2.0 with 80% power using a two-sided test? Prior studies have estimated
that 30% of the male population are smokers. (zα/2=1.96; zβ=0.84) - correct
answer ✔89 cases
356 controls
445 total
Alpha = 0.05 (two-sided)
Beta = 0.2; Power = 0.8 (Beta = 1 - .0.2)
Prevalence of exposure (smoking) among controls = 0.3
, Expected Odds Ratio = 2.0
Ratio of cases to control = 4 (1 case: 4 controls)
Suppose we wish to undertake epidemiologic studies to assess the effect of
current oral contraceptive (OC) use in relation to risk of myocardial infraction
(MI) among women of childbearing age. From previous studies, we know
about 10% of such women in the United States currently use OCs and the
incidence of MI among non-OC users is 2.1 per 1000. We wish to detect a
relative risk of 1.8 with a conventional alpha level of 0.05 and beta level of
0.20, representing a power of 80% to detect an effect of this magnitude of the
effect if one truly exists. Calculate the sample size for each of the following
study designs with an equal number in each groups. (zα/2=1.96; zβ=0.84)
cohort study
- sample size:
- number in each group:
unmatched case-control study
- sample size:
- number in each group: - correct answer ✔Sample size: 32,606
# in each group: 16,303
Sample size: 814
# in each group: 407
Cohort Study
Alpha = 0.05 (two-sided)
Beta = 0.2; Power = 0.8 (Beta = 1 - .0.2)
Incidence of MI among non-OC users = 2.1/10000 = .0021
Expected Risk Ratio = 1.8
