Module 8 & 9 Flashcards
Define “Confounding”
Confounding occurs when a variable is related to the exposure AND either causes or prevents the outcome
What is “Stratification”?
Stratification is a way to address confounding by separating out participants based on the confounding variable and analyzing them separately
What are the drawbacks of stratification?
It is difficult to stratify multiple data when multiple confounders are present. The process of stratification reduces sample size, thus decreasing power.
List three alternatives to stratification in order to address confounding
Randomization, Matching, Multivariable Statistical Adjustment
What is “Randomization”?
Randomly assigning individuals to exposure groups. Used only in interventional study designs
What is “Matching”?
Matching individuals in both test and control groups based on exposure to potential confounding variables. Used in “Case/Control” studies. Requires prior knowledge of confounding variables
What is Multivariable Statistical Adjustment?
A statistical method for consideration of confounding variables to utilize AFTER study participants are selected. Used in a variety of study types
What is Multivariable Regression
A complicated statistical adjustment used for addressing confounding variables in a study. The result of Multivariable Regression is the “adjusted estimate”; a statistical estimate of the effect of that exposure on the outcome.
What is the “adjusted estimate”?
The result of Multivariable Regression is the “adjusted estimate”; a statistical estimate of the effect of that exposure on the outcome.
List three type of logistical regression and what data-set they are typically used for
Multivariable Logistic Regression - used for CATEGORICAL variables Multiple Linear Regression - Used for CONTINUOUS variables Multivariable Cox Regression - Used for Time-to-Event data
What is the “correlation coefficient”, which type of statistical analysis is it associated with, and what information does it provide?
Multiple Linear Regression yields a “Correlation Coefficient.” (r). r = 0 = no correlation, r > 0 = positive correlation, r < 0 = negative correlation. Doesn’t relate to slope, only estimates how well X can predict Y. Sometimes referred to as R^2
What is the “Median Survival Time”?
On a Kaplan-Meier Survival Curve, it is the time on the X-axis where the KMSC crosses 0.5 on the survival function axis. (i.e., 50% probability of the outcome)
Describe “Censoring” and how it relates to the Kaplan-Meier Survival Curve.
Censoring is what gives the KMSC it’s unique step-shape. It occurs when participants achieve the event under study or are lost to followup.
List for key considerations for interpreting Klan-Meier survival curves.
What defines the beginning of follow-up? How long was follow-up conducted? Are their differences between cohorts that might lead to differences in censoring? Is there a difference in the Median Survival Times between groups?
What is the log-rank test?
A statistical test used to see if there are statistically significant differences between groups. Mentioned in connection with interpretations of Kaplan-Meier Survival curves and Median Survival Times
