Week 6 (lecture 5)

Question 1

Critiques of biomedical literature studies include what?

Answer 1

Biases: Omission of data leads to bias because the study will generate an estimate that on average is fundamentally different than the population parameter that one is trying to estimate
Confounder
Omitted variable
Efficiency

Question 2

Define confounder

Answer 2

Any factor that prevents the researcher from directly and appropriately interpreting a statistical result within the practical context of the study
Could be an omitted variable, but it could also reflect a flaw in the study’s design

Question 3

Define omitted variable

Answer 3

Specific and observable factor, but is omitted from the analysis

Question 4

Define efficiency

Answer 4

If adjusting for the confounder leads to changes in any estimate of variation (usually a reduction in the standard deviation or standard error), then any results obtained without adjusting for confounding (usually exhibiting higher levels of variation) are said to be inefficient

Question 5

Errors of Omission and Commission are what?

Answer 5

Variables that change the relationship between the independent and dependent variable may generate biased or inefficient results

Question 6

Explain Errors of Omission and Commission

Answer 6

Variables that change the relationship between the independent and dependent variable may generate biased or inefficient results
1) Confounding effect  confounders
Variable that when accounted for in an analysis leads to a meaningfully different interpretation of the relationship between the primary independent variable and the dependent variable compared to when the confounding variable is ignored in or excluded from the analysis
2) Effect modification (moderator effect)  moderators
Alters the strength and/or direction of the relationship between the independent variable and the dependent variable
3) Mediating effect  mediators
The difference between a confounder and a mediator is that a confounder is not an intermediate variable in a causal pathway, whereas a mediator attempts to link the primary independent variable and the dependent variable

Question 7

What is linear regression?

Answer 7

Used when the dependent variable or outcome variable of interest is a continuous variable (interval or ratio data)

Question 8

Give an example of linear regression

Answer 8

Study looking at obesity causing DM II (↑HbA1c)
Obesity = waist-to-hip ratio
Question: How does a slight increase (or decrease) in waist-hip ratio, on average, increase (or decrease) patient A1C levels?
α = 0.05

Question 9

Give Coefficient of determination (R2) as an example of liner regression

Answer 9

1) R-squared (R2) – assesses variation in dependent variable; mathematically, using the magnitude of the trend line’s slope will predict the dependent variable
2) R2 ≈ 1 (better job of predicting the dependent variable)
Regression explains the dependent variable well
3) R2 ≈ 0 (worse job of explaining the dependent variable)
Regression explains very little variation in the dependent variable
4) R2 > = 0.66 is considered acceptable

Question 10

Give an exmaple of time-to-event analysis and how to do it

Answer 10

1) Can we predict how long it will take for the average person in the study to develop nephropathy in our diabetes study?
2) Does being diagnosed with diabetes reduce the length of time before the onset of nephropathy?
3) Kaplan-Meier method and the Cox proportional hazards regression model
-Common analysis in biomedical literature for time-to-event analysis
-Methods able to compute measurement with censored data
-Attrition, changing arms of study or no primary outcome observed

Question 11

Explain the Kaplan-Meier Method

Answer 11

Survival curve summarizes the probability of survival over time estimated from a sample
Acceptable to use log-rank tests for testing the null hypothesis
Limited if adjusting for predictors when the dependent variable consists of possibly censored time-until-event data
Does not quantify the effect of a predictor variable on survival time

Question 12

Cox Proportional Hazards Regression Model
1) When is it used?
2) Explain hazard ratio

Answer 12

1) Used to address the typical multivariable problems addressed by multiple linear and logistic regression (e.g., confounding, effect modification, dummy variables)
2) The hazard of an event at any point can be thought of as the risk of event occurrence at time (t)
As with linear regression and logistic regression, additional predictor variables can be added to address the possibility of confounding or effect modification