PART 2

Please read the statements below and indicate whether they are true (T) or false (F).

	Statement	T or F
1	This statement is a cumulative rate: "Risk of being diagnosed with invasive breast cancer is about1 in 8 women (13%)in their lifetime and 1 in 39 women (3%) will die from breast cancer in the US."
2	This statement is an incidence rate: "Therateof newcasesof femalebreast cancerwas 129.1per 100,000women per yearin the US."
3	Odds is the probability of developing the disease divided by the probability of not developing the disease of interest
4	Odds is the ratio of the probability of occurrence of an event to the probability of nonoccurrence
5	Relative risk indicates the strength of the association between exposure and disease, and thus could be used to explore the causal relationship of the exposure and the disease of interest
6	In a prospective follow-up study, a risk odds ratio can be estimated
7	In a case control study, an exposure odds ratio can be estimated
8	Exposure odds ratio can be used to estimate the risk odds ratio if the data has no selection bias, no information bias and confounding is controlled or controllable
9	Prevalence ratio can be used to estimate risk ratio if exposure is not a prognostic factor (that is, the duration of the disease is the same for exposed cases and unexposed cases) and the disease does not affect the exposure status of the study subjects
10	OR always overestimates the RR
11	AR is a quantity used to estimate the number of cases of the disease among the exposed that can be eliminated if the exposure were eliminated
12	The following statement is an AR: "105 of the 113 incident cases of lung cancer among 1,000 US smokers in year 2020 are attributable to smoking."
13	The following statement is an AR: "Among smokers in the US, cigarette smoking is responsible for about90%of lung cancers."
14	PAR is a quantity used to quantify the risk of disease in the total populationthat can be attributed to the exposure
15	The following statement is PAR: "In the US, cigarette smoking is responsible for about60%of lung cancers."
16	A causal relationship exists between the exposure and the disease of interest when you report AR and/or PAR
17	AR is always larger than PAR because AR is calculated by comparing a complete exposed group with a complete non-exposed group; while PAR is calculated by comparing an incomplete exposed population with a complete non-exposed group
18	Confounding is a validity issue. It must be prevented and/or controlled in data analysis
19	For a third variable to be a confounder, it must be associated with the exposure of interest and associated with the disease of interest and the variable is not an intermediate variable of the causal pathway between the exposure and the disease of interest
20	A limitation associated with restriction to control confounding is its inability to study the independent effect and interaction with other factors
21	Formal randomization with large sample size can make the comparison groups have the same rate of a disease in the absence of treatment
22	Randomization to control for confounding needs the right comparison groups in data analysis
23	In a randomized trial, you must compare groups as they were originally randomized. Any other comparison would defeat the purpose of randomization
24	The advantage of stratification over restriction to control for confounding is that stratification can evaluate the independent effect of the stratification variable and its interaction with other factors
25	Standardization is a technique used to remove the confounding effect and make an overall estimate by combining stratum-specific estimates using the same weighting system for the comparison groups, such as ORMH in a case control study
26	Matching in epidemiology studies means to intentionally force the comparison group to have the same or similar distribution as the index group with respect to certain potential confounders
27	Matching in a prospective cohort study can eliminate confounding from the matched variable
28	Matching in a case control study can introduce a negative confounding if the matched variable is associated with the exposure of interest
29	The negative confounding introduced by matching in a case control study is controllable by matched analysis
30	Overmatching describes the situation in which matching in fact reduces the study efficiency rather than promotes the study efficiency as expected. For example, the matching variable is strongly associated with the exposure but not associated with the disease
31	Overmatching increases the number of concordant pairs and these pairs do not contribute to estimation and significance testing, and thus reduces the study efficiency
32	Bias is introduced into the study by the investigators (unintentionally or intentionally) in the process of recruiting study subjects or collecting exposure information
33	Bias is a validity issue and must be prevented
34	Selection bias is a type of systematic bias intentionally or unintentionally introduced into the study due to invalid subject selection when the study population is organized
35	In case-control study, if different criteria relating to exposure have been used for selection cases and controls, selection bias would be introduced
36	In a prospective cohort study, use of different criteria to select or exclude subjects is usually not the source for selection bias but loss-to-follow-up and withdrawals are considered to be a type of selection bias
37	In a randomized trial, formal randomization is needed to prevent selection bias
38	High refusal rate for subjects to participate or high rate of loss to follow-up casts doubt about the validity of the study due to potential for selection bias
39	Different criteria used to obtain information from the actual study population for exposure and disease could introduce information bias
40	The consequence of information bias is to misclassifying people in terms of exposure or disease status, and thus information bias is also called misclassification bias
41	Nondifferential misclassification bias means the probability of misclassification of disease (or exposure) does not differ between the comparison populations
42	Differential misclassification bias is the worst and must be prevented since the probability of misclassification of disease or exposure differs between the comparison groups
43	Sensitivity is the ability of a test to correctly identify those with the disease (true positive rate)
44	Specificity is the ability of the test to correctly identify those without the disease (true negative rate)
45	A moderator (effect modifier) is a variable that affects (changes or influences) thestrength of the association betweenthe exposure and the outcome of interest
46	Amediatoris a variable that explains the relationshipbetweenthe exposure and the disease of interest; it links the exposure to the disease of interest
47	Effect modification is a change in the magnitude or direction of the association between exposure and disease according to the level of a third variable
48	Effect modification is a quantitative measure of the association, not a qualitative issue
49	Restriction could eliminate confounding while eliminates the ability to evaluate the effect modification
50	Effect modification is something that needs to be reported while confounding is something that needs to be controlled
51	Stratification reveals effect modification but eliminates confounding if the confounder is a discrete variable
52	To judge effect modification is to see if stratum-specific ORs are equal; to judge confounding is to compare if the crude OR equals the adjusted OR
53	Usually, formal statistical significance testing is needed to assess if there is effect modification
54	Additive and multiplicative models can be used to statistically evaluate effect modification, but statistical evaluation of effect modification is totally model dependent
55	An additive model for evaluating effect modification means that the joint absolute effect of two risk factors is equal to the sum of their independent absolute effects
56	A multiplicative model for evaluating effect modification means that the relative risk for the disease following joint exposure to both risk factors is equal to the product of the relative risk for exposure to either of the risk factors alone
57	In causal inference, temporality (that is, the exposure precedes the outcome) is the only absolutely essential criterion