Biostats/Epi Flashcards
What is sensitivity?
Indicates how well a test can SCREEN for a disease
A higher sensitivity means taht negative results are less likely to be FNs and more likely to be TNs, thus better able to rule disease out (SNOUT)
Important for screening tests
What is specificity?
Indicates how well a test can CONFIRM the diagnosis
High specificity means positive results are less likely to be FPs and more likely to be TPs thus positive results better able to rule the disease in (SPIN);
Important for confirmatory tests
What is accuracy?
true positives + true negatives/ (all results)
Depends on sensitivity and specificity as well as prevalence of condition in population
Increases as total area under the ROC curve increases
P values and confidence intervals
P value is the probability of observing a given (or more extreme) result due to chance alone, assuming the null hypothesis is true. A result is generally considered statistically significant when p<0.05.
For study results to be stat sig, the confidence interval must NOT contain the null value.
What is susceptibility bias (selection bias)?
When the treatment regimen selected for a pt depends on the severity of the pt’s condition, a form of selection bias known as susceptibility bias (confounding by indication) can result.
To avoid selection bias, pts are randomly assigned to treatments to minimize potential confounding variables and there is intention to treat analysis.
Lead-time bias.
Happens when two disease interventions are compared and one diagnoses the disease earlier than the other w/o an effect on the outcome (survival).
This would make it appear that the intervention prolonged survival when it really just diagnosed the disease sooner.
Measurement bias.
Occurs from poor data collection with inaccurate results.
Observer bias.
Occurs when the observer is influenced by prior knowledge or details of the study in a way that affects the results.
Avoided by blinding observers from knowing tx assignments and by measuring objective outcomes (mortality) that are less likely to be skewed by observers.
Recall bias.
Occurs when a study participant’s answer to a question is affected by prior exposures.
More common in retrospective studies than prospective studies (RCTs).
Precision.
The measure of random error.
The tighter the confidence interval, the more precise the result.
Increasing the sample size increases precision.
What is confounding?
Refers to the bias that results when the exposure-disease relationship is mixed with the effect of the extraneous factors (i.e., confounders)
Confounders influence both the exposure and outcome.
Methods to deal with confounding: matching of cases and controls based on the confounding factor, or stratification of the study population based on the confounding factor.
What is selection bias?
Results from the manner in which people are selected for the study, or from the selective losses from follow-up.
P value and confidence interval.
Power of a study: represents ability to detect a different b/n 2 groups (exposed versus unexposed) when there truly is a difference
Increasing the sampel size increases the power of a study and narrows the CI surround the point estimate
CI express ss and are interrelated w/ p values.
What is the null hypothesis?
Statement of no relationship/no association between the exposure and the outcome.
To state the null hypothesis correctly, the study design should be considered.
Cohort study
A cohort study design is best for determining the incidence of a disease.
Comparing the incidence of the disease in 2 populations with and without a given risk factor allows for calculation of relative risk.
Can also calculate relative rate and median survival.
Case control study
Subjects w/ the disease of interest (cases) are compared to an otherwise similar group of disease free subjects (controls)
Informaton is then collected about exposure to risk factors.
CC study is retrospective and meant to determine assoc. b/n risk factors and disease occurrence.
An odds ratio can be calculated in a cc study but the incidence of a disease CANNOT be calculated.
Ecological study
The unit of observation is a population
Disease rates and exposures are measured in 2 (or more) populations and the assoc. b/n disease rates and exposure is determined.
However, results about assoc/ at the population level may not translate to the individual level.
Study cannot be used to determine incidence.
Randomization
Successful randomizatioin in a clinical trial allows a study to eliminate bias in treatment assignments.
An ideal randomizatioin process minimizes selection bias, results in near-equal treatment and control group sizes, and achieves a low probability of confounding variables.
