week 12

Question 1

What are examples of adjustment methods based on stratification?

Answer 1

mantel-haenszel method, direct adjustment, indirect adjustment

Question 2

Is the cox proportional hazards regression based on stratification?

Answer 2

No

Question 3

What is the goal of standardization?

Answer 3

to remove the effects of confounding

Question 4

What are the 4 steps in direct standardization?

Answer 4

1) Calculate age-specific mortality rates for each group in each population. 2) Choose the standard (reference) population. 3) Calculate the age-adjusted rate in each population (expected). 4) Compare the age-standardized rates of the populations

Question 5

What are the three criteria of an instrumental variable?

Answer 5

1) It must affect the outcome only through exposure. 2) It must be causally associated with the exposure. 3) It must not be associated with any confounders (known or unknown) of the association between the exposure and the outcome

Question 6

An instrumental variable must affect the outcome through the exposure and a confounder. True or false?

Answer 6

False

Question 7

Why stratify?

Answer 7

goal is to understand the exposure-outcome relationship

Question 8

when to use indirect standardization?

Answer 8

stratum-specific rates are missing in one of the populations, or numbers in the strata are small - rates therefore unstable

Question 9

What is standardized mortality ratio (SMR)?

Answer 9

the relative measure of mortality when death is outcome (observed number of deaths over a specified time period/ expected number of deaths over a specified time period)

Question 10

What are the differences between direct standardization and indirect standardization?

Answer 10

Direct standardization is commonly used to adjust confounding when comparing vital statistics data across populations, while indirect standardization is commonly used to adjust for confounding in occupational cohorts to compare rates of mortality of workers to the general population. With direct, the stratum specific incidence rates are applied in both populations from the standard population. With indirect, the age distribution of the population of interest is applied to the external incidence rates of the comparison population.

Question 11

What is the Mantel-Haenszel odds ratio?

Answer 11

a measure of association that is adjusted for confounding

Question 12

How do you interpret Mantel-Haenszel adjusted odds ratio of 2.98?

Answer 12

After adjusting for sex, the odds of having CHD were 2.98 times higher among those who were obese compared to those who were not obese.

Question 13

How do you interpret a Mantel-Haenszel adjusted risk ratio of 1.83?

Answer 13

After adjusting for sex, the risk of CHD among those who were obese was 1.83 times the risk of CHD among those who were not obese.

Question 14

Is matching a type of stratification?

Answer 14

Yes, it could be seen as both an adjustment technique and a study design feature

Question 15

Why are stratification techniques limited?

Answer 15

data becomes sparse when there are too many strata, hard to interpret when there are many strata, only for categorical variables; we generally only stratify on 1 or 2 variables

Question 16

What is the solution to stratification limitations?

Answer 16

multivariable analysis

Question 17

What is multivariable analysis?

Answer 17

measures the relationship between two variables (exposure-disease), and can assess the effect of multiple variables while holding the effect of all other variables in the model constant; can more fully assess the presence of interaction; can use continuous variables

Question 18

What are some disadvantages of using multivariable regression model?

Answer 18

must be sure your model is appropriate for your data, more difficult to explain results

Question 19

How do you interpret a beta coefficient of 0.6 in a linear regression model, with cholesterol as the outcome.

Answer 19

For every year change, cholesterol level (outcome) changes by about 0.6, while adjusted for all other variables in the model

Question 20

What is the assumption for cox proportional hazards regression model?

Answer 20

hazards are proportional across time

Question 21

which regression model is most appropriate to use to analyze a study with a continuous outcome?

Answer 21

linear

Question 22

which regression model is most appropriate to use to analyze data from a cohort study that looked at an association between and exposure and time to an outcome?

Answer 22

cox proportional hazards regression

Question 23

which regression model is most appropriate to use to analyze a case-control study that used individual matching?

Answer 23

conditional logistic regression

Question 24

What regression model is appropriate for cross-sectional study?

Answer 24

linear regression

Question 25

What regression model is appropriate for cohort study design?

Answer 25

cox proportional hazards regression

Question 26

what regression model is appropriate for case-control or cross sectional?

Answer 26

logistic regression model

Question 27

Are instrumental variables affected by confounders?

Answer 27

no

Question 28

The instrumental variable is associated with the exposure or outcome?

Answer 28

exposure

Question 29

instrumental variables can be used for randomization purposes. true or false?

Answer 29

true

Question 30

What kind of RCT analysis provides an unbiased estimate of the relationship between treatment assignment and outcome?

Answer 30

intent to treat analysis

Question 31

What are propensity scores used for?

Answer 31

to more completely and efficiently control for confounding by reducing/addressing selection bias and non-randomization of exposure; mimic randomization by making exposed and unexposed groups as comparable as possible