Research Design, Biostatistics, and Literature Evaluation
Describes the impact of a single predictor variable on the time-to-event between cohorts
Compares the survival times between two cohorts while controlling for a singular predictor variable
Survival curves are typically analyzed using the log-rank test.
| d. | Results typically reported as hazard ratio with 95% CI |
|---|
Cox proportional hazards model
Describes the impact of several predictor variables on the time-to-event
Compares the survival times between two cohorts while controlling for several predictor variables
Results typically reported as hazard ratio with 95% CI
Sensitivity: Probability of identifying a true positive from a group that is known to be positive
Avoiding false negatives
Example: Proportion of people with disease correctly identified as having the disease through
laboratory assay
Avoiding false positives
Positive predictive value (PPV): Probability of identifying a true positive from a group that may or may
not be positive
Avoiding false positives
Example: Proportion of people with positive laboratory assay result who actually have the disease
Negative predictive value (NPV): Probability of identifying a true negative from a group that may or
may not be negative
Avoiding false negatives
A
A + C
Sensitivity =
x 100
D
B + D
Specificity =
x 100
A
A + B
PPV =
x 100
D
C + D
NPV =
x 100
Has disease or outcome
Does not have disease or outcome
Positive
A
B
Negative
C
D
NPV = negative predictive value; PPV = positive predictive value.