week 12 Flashcards
What are examples of adjustment methods based on stratification?
mantel-haenszel method, direct adjustment, indirect adjustment
Is the cox proportional hazards regression based on stratification?
No
What is the goal of standardization?
to remove the effects of confounding
What are the 4 steps in direct standardization?
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
What are the three criteria of an instrumental variable?
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
An instrumental variable must affect the outcome through the exposure and a confounder. True or false?
False
Why stratify?
goal is to understand the exposure-outcome relationship
when to use indirect standardization?
stratum-specific rates are missing in one of the populations, or numbers in the strata are small - rates therefore unstable
What is standardized mortality ratio (SMR)?
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)
What are the differences between direct standardization and indirect standardization?
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.
What is the Mantel-Haenszel odds ratio?
a measure of association that is adjusted for confounding
How do you interpret Mantel-Haenszel adjusted odds ratio of 2.98?
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.
How do you interpret a Mantel-Haenszel adjusted risk ratio of 1.83?
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.
Is matching a type of stratification?
Yes, it could be seen as both an adjustment technique and a study design feature
Why are stratification techniques limited?
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
What is the solution to stratification limitations?
multivariable analysis
What is multivariable analysis?
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
What are some disadvantages of using multivariable regression model?
must be sure your model is appropriate for your data, more difficult to explain results
How do you interpret a beta coefficient of 0.6 in a linear regression model, with cholesterol as the outcome.
For every year change, cholesterol level (outcome) changes by about 0.6, while adjusted for all other variables in the model
What is the assumption for cox proportional hazards regression model?
hazards are proportional across time
which regression model is most appropriate to use to analyze a study with a continuous outcome?
linear
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?
cox proportional hazards regression
which regression model is most appropriate to use to analyze a case-control study that used individual matching?
conditional logistic regression
What regression model is appropriate for cross-sectional study?
linear regression
What regression model is appropriate for cohort study design?
cox proportional hazards regression
what regression model is appropriate for case-control or cross sectional?
logistic regression model
Are instrumental variables affected by confounders?
no
The instrumental variable is associated with the exposure or outcome?
exposure
instrumental variables can be used for randomization purposes. true or false?
true
What kind of RCT analysis provides an unbiased estimate of the relationship between treatment assignment and outcome?
intent to treat analysis
What are propensity scores used for?
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
