Info Skills Week 9 Terms/Objectives

Confounding

(factor) a factor in a study that is associated with both the exposure (treatment) and the outcome. (effect) the distortion of the relationship between an independent variable and a dependent variable due to another variable called a confounder.

Kaplan-Meier method

A method used for estimating survival functions from a sample; thus, a Kaplan-Meier survival curve summarizes the probability of survival over time estimated from a sample.

Linear regression

A regression model commonly used to describe and evaluate the relationship between a single continuous dependent variable and one or more independent variables. Follows y=mx+b trend. Standard measure of model fit is R2 (coefficient of determination). Test of the overall model is performed using the F-statistic.

Log-Rank test

A statistical test to compare the overall survival experience of two or more groups with respect to the study outcome; thus, it can be used to test the null hypothesis that two or more survival curves are equivalent.

Survival analysis

Survival analysis is generally defined as a set of methods for analyzing data where the outcome variable is the time until the occurrence of an event of interest. The event can be death, occurrence of a disease, marriage, divorce, etc.The time to event or survival time can be measured in days, weeks, years, etc. For example, if the event of interest is heart attack, then the survival time can be the time in years until a person develops a heart attack.

Effect modification

The finding that the relationship between the independent variable of interest and the dependent variable depends on the values of a third variable, called an effect modifier.

Correlation coefficient

a measure of how two variables are linearly related. It provides information about the strength and direction of the relationship between two continuous variables. The correlation coefficient varies from -1 to +1; either being a perfect positive relationship, a perfect negative relationship, or no linear relationship (0). Null hypothesis: r = 0.

Multivariable analysis

also called multiple regression, multivariable adjustment, or multivariable modeling; an extension of the bivariate (or simple) linear regression model (which involves only one independent variable) to the situation in which more than one independent variable is considered (in both cases, there is a single continuous dependent variable)

Cox regression

also called the proportional hazards model; type of survival analysis; regression model commonly used to describe and evaluate the relationship between a dependent variable that represents the occurrence and timing of an event with the possibility of censored data (for example, time-until-event data with censoring) and one or more independent variables

Hazard ratio

is conceptually identical to a rate ratio (a ratio of two rates). The interpretation of the HR is similar to that of an odds ratio; an HR=1 suggests no relationship between the predictor and the timing of event occurrence. Can be derived from the Cox regression. Hazard is the risk of an event occurrence at a particular time.

Logistic regression

statistical method used when an outcome variable consists of only 2 categories (binary); used for predicting a nominal dependent variable based on one or more predictor variables; used in estimating the parameters of a qualitative response model; measures the relationship between a categorical dependent variable and one or more independent variables that are usually continuous. Test of the overall model is performed using a chi-square statistic. T-statistics are not used to test the significance, but instead values reported by statistical software, known as the Wald test. Odds ratio is also commonly reported.

Odds ratio

the odds of exposure for cases and controls. Odds Ratio = (Odds of being exposed among cases)/(Odds of being exposed among controls).

Stratified analysis

use this method to identify confounding; if the stratified estimates of association differ from unadjusted estimates by more than 10%, then assume there is confounding; a stratum is a specific characteristic; a stratified analysis is an analysis done on each subgroup to control for the different characteristics