Measures Of Association Flashcards
What is the difference between sensitivity and specificity?
Sensitivity = ability to correctly identify those who actually have the disease (true positive)
Specificity = ability to correctly identify those who actually do NOT have the disease (true negative)
What is the difference between incidence rate and prevalence?
Incidence rate = # NEW cases of disease in a given time period
Prevalence = TOTAL # of cases of disease in a given time period
(Ie. If prevalence is the overall % pts with disease, incidence rate is inflow % of new pts with disease)
What is the difference between relative risk and attributable risk?
Relative risk = the ratio of the risk of disease in exposed individuals to the risk of the disease in unexposed individuals
Attributable risk = the proportion of disease incidence or risk that can be attributed to a specific exposure
TLDR: RR = focuses on the strength of the association between exposure and outcome, AR = focuses on the proportion of diseases in the exposed group that can be attributed to the exposure.
Define attack rate
of people exposed who got disease
—————————————————-
Total # of people exposed
measure of morbidity
Define mortality rate
The # of deaths from a cause in a specific population in a specific time frame
Define standardization, what is the difference between direct and indirect standardization?
Set of techniques used to remove the effects of intersubject variability when comparing ≥2 populations
Direct: Outcomes if the populations had the same age distributions as a standardized population
Indirect: compares rate of disease observed in populations to EXPECTED rate in standard population
Define “population-attributable risk”
The attributable risk for an ENTIRE population regardless of whether or not everyone was exposed
AKA = what is the impact of removing the exposure on the total population?
Ie. PAR% = 12% then making all people stop [exposure] would eliminate 12% of all [disease] in the population
Define number needed to treat (NNT)
Number of patients needed to treat in order to produce one BENEFICIAL outcome
Aka = lower number is better
Ie. Need to treat X people with drug to cure 1 person
Define number needed to harm (NNH)
Number of patients who receive treatment before one ADVERSE EFFECT occurs
Aka = higher number is better
Ie. Treated X people with drug before 1 person got [AE]
What is the equation for calculating SENSITIVITY?
(True positive)
———————————————
(True positive) + (False Negative) [aka all positives]
What is the equation for calculating SPECIFICITY?
(True negative)
———————————————
(True negative) + (False positive) [aka all negatives]
What is morbidity and what terms can describe it?
Morbidity = rate of disease in a population
Described by:
- cumulative incidence proportion
- Incidence rate
- Prevalence
- Attack rate
What is relative risk/risk ratio?
Ratio of the risk of disease in exposed individuals to the risk of the disease in unexposed individuals
- RR < 1: negative association/possible protective
- RR > 1: positive association/possible causal exposure
Compare RR vs OR
RR: compares subjects who got disease w/ exposure to total exposures vs subjects who got disease w/o exposure to total non-exposures
OR: compares all “expected” outcomes vs all “unexpected” outcomes (expected being exposure = disease, non-exposure = no disease)
What is the RR calculation?
[(exposed&disease)/(total exposed)]
—————————————————————
[(non-exposed&disease)/total non-exposed)]
What is the OR calculation?
(Exposed&disease) x (non-exposed&no disease)
——————————————————————
(Exposed&no disease) x (non-exposed&disease)
What is the general equation for attributable risk?
AR = risk exposed - risk unexposed
Ie. If AR% = 85%, then 85% of [disease] is caused by [exposure]
What is a hazard ratio?
The rate at which an unfavorable event occurs, used in survival analyses
How should confidence intervals be interpreted? Do we want WIDE or NARROW confidence intervals?
Describes the reliability of study to reflect the entire population
- Confidence Interval is range of accurate:
WIDE = low precision, less credibility
NARROW = higher precision, more credibility
If range INCLUDES 0, then NOT statistically significant
