Upgrade to remove ads
Only $35.99/year

Biostats First Aid

Terms in this set (34)

Cross-Sectional Study
collects data to determine frequency of disease aka prevalence
Case Control
compares people with disease to people without disease

match cases to controls to fix for confounding
Crossover Study
participants act as their own controls
Cohort Study
Compares group with a given exposure to a group without exposure

prospective - exposure determined first, follow participants for outcome

retrospective - outcome determined first, exposure determined retroactively
Cutoff values
2x2 table
Sensitivity
proportion of people with disease who test positive
Specitivity
proportion of people without disease who test negative
Likelihood Ratio
PPV
proportion of people who test positive who have disease
NPV
proportion of people who test negative who do not have disease
Selection Bias
error in assigning subjects to study group. 3 common types

Berkson, Healthy Worker Effect, Attrition bias
Berkson Bias
type of selection bias

study population from hospital is less healthy than general populatoin
Healthy Worker Effect
type of selection bias

study population is healthier than general population
Attrition Bias
type of selection bias

loss to follow up only happens in one group
Recall Bias
problem in retrospective studies where participants aren't aware of disease
Measurment Bias
information gathered in distorted way
Procedure bias
subjects in different groups are treated differently
Observer Expectancy Bias
Investigator's decision is affected by knowledge of exposure status

Can be hawthorne or pygmalion/rosenthal effect
Hawthorne Effect
subjects change their behavior as a result of them being aware of being studied
Pygmalion/Rosenthal Effect
If observer expects treatment group to show recovery, observer will be more likely to document positive outcomes for this group
Lead Time Bias
early detection confused with increased survival
Statistical Hypotheses Table
Ho aka Null
hypothesis of no difference or relationship between disease and risk factor
H1 aka Alternative
Hypothesis of some difference or relationship between disease and risk factor
Type 1 error
stating a difference exists when nothing exists in reality

alpha is probability of making this error
Type 2 error
beta is probability of stating that there is no difference when a difference exists in reality

1 - beta aka power is probability of correctly rejecting null hypothesis when it is false or ability to detect a difference if one exists
Quantifying Risk
Odds Ratio vs. Relative Risk
odds - groups with disease over groups without disease

risk - exposed group over unexposed group
Attributable risk
difference in risk of having disease between those who are exposed and unexposed to a risk factor
Relative Risk Reduction
proportion of risk reduction attributable to an intervention

1 - relative risk
Absolute risk reduction
difference in risk attributable to intervention
Number needed to treat
1/absolute risk reduction
Number needed to harm
1/attributable risk
Sets with similar terms