Measure Of Association And Bias/Confounding Flashcards
Measures of association are used in what types of studies?
Analytical studies
If there is the same amount of disease in both exposed and non-exposed group, that means the out come is (associated or non-associated)
Non- associated
What type of measure of association can we use in cross-sectional and case-control studies?
Odds ratio
What types of measure of association can be used in prospective cohort or clinical trials?
Relative risk (incidence) Attributable risk
What type of measure of association can be used in retrospective cohort and cross-sectional studies?
Prevelence ratio
What is the odds ratio?
= odds of exposure in diseased animals / odds of exposure in non-diseased animals
If the odds ratio = 1, what does this mean?
Exposure is the same between diseased and non-diseased => no association
If the odds ratio is > 1, what does this mean?
Exposure among diseased is greater than control => positive association
If odds ratio is < 1, what does this mean?
Exposure among diseased is less than control => negative association
What is relative risk?
Compares risk of exposure and non-exposed groups
RR= (risk in exposed group) / (risk in un-exposed group) RR= [a/(a+b)] / [c/(c+d)]
What is the amount/proportion of overall disease incidence in a population or group that can be attributed to a specific exposure?
Attributable risk
What is the difference between AR due to etiologic fraction and AR in the population
AR (etiologic)- how much disease is attributed to factor in study
AR (population) - how much disease in target population is attributed to factor in study
What is a measure used to compared the amount of disease that has occurred in the exposed and un-exposed population in retrospective studies?
Prevelence
Case control uses what measure of association ?
Odds ratio
Prospective cohorts uses what measure of association?
Relative risk
Retrospective cohorts uses what measure of association ?
Prevelence rate
Clinical trials use what measures of association?
Relative risk
__________ relates to the absence of systemic error in a study result
Validity
_______________ is the extent to which a measure of association from a study differs from the true measure of association in the source population
Bias
What does it mean for a study to have internal validity?
The study result is valid with respect to population under study
What does it mean for a study to have external validity?
The study result is valid to a wider population beyond the study and source population
What are sources of bias?
Selection bias -> error in selection of study subjects (sample is different from population)
Informational (misclassification) bias -> error in measurement
What is an unknown factor that will disrobed the relationship between he exposure and outcome?
Confounding
How can bias be reduced in a cross-sectional descriptive study?
Random sampling (a biased sample would have more or less disease than the true population)
How can bias be reduced in case-control studies?
Match controls with demographics, age, and breed
How can bias be reduced in a cohort study?
Matching to ensure exposed and non exposed have similar demographics
What are 3 types of selection bias?
Self selection bias -> based on volunteers
Healthy worker bias-> those working are in better health than pop
Diagnostic bias-> diagnosis influenced by vets knowledge, exposure, and expectations
An error is considered ______ if it is not related t the exposure or the disese
Non-differential
If a error in measurement is related to the exposure or outcome it is ____________
Differential
How can information bias be reduced?
Evaluation of accuracy of measuring tools
What are types of information bias?
Observer variation (interpretation)
Deficiency of tools and technical errors
Recall bias
Reporting bias
T/F: confounders influence both exposure and outcome by distorting measure of association.
True
What are the 3 conditions for a confounder?
Associated with exposure
Associated with outcome
Not in the causal pathway between exposure and outcome
If you realize there may be a condfounding factor affecting a study you already preformed, how can you reduce this effect?
Stratify data -> parturition results based on the confounding factor
Prior to a study how can confounding factors be reduced?
Match the study so confounder is equally represented in both groups
Restriction - do not enroll individual with confounding factor
Randomization
What type of error in a study is due to chance
Random error - fluctuations around a true value because of sampling or measuring
What type of error is due to bias?
Systematic error - error that is inherent to the study method being used -> results are skewed in a predictable and repeatable error
Random variation in a study is measured by what two tests?
Confidence interval -> range of values in which true value is found
P-value -> statistically significant
A large confidence interval will mean what about the sample size
Small sample size
T/F: if the 95% confidence interval for the odds ratio includes 1, the odds ratio is statistically significant?
FALSE
If the 95% confidence interval for the odds ratio DOES NOT include 1, it is statistically significant
Odds ratio > 1
Confidence interval includes 1
Is there an association?
No association, not statistically significant
Odds ratio <1
Confidence interval does not include 1
Is there an association?
Negative association
Statistically significant
A P-value is statistically significant if it is _______
<0.05
What is a null hypothesis?
Statement saying there is no association. (We want to reject the null hypotheses)
What are the types of error in statistical significance?
Type 1 (a) - False positive (reject null when it is NOT false)
Type II (B)- False negative (accept the null when it is FALSE)
If the explanatory and outcome variables are both categorical, what type of central tendency test do you use?
Chi-squared (for all two by two tables)
If the explanatory variable is categorical and the outcome variable is continuous, what test do you use for central tendency?
Students t test
If the explanatory and outcome variables are both continuous, what measure of central tendency is used?
Correlation
