Measures of Impact Flashcards
What are 2 ways to describe the frequency of disease in a population?
- incidence: rate, risk, CI
- prevalence: proportion
Frequency numerically describes ______
The movement of well to ill/force of disease (incidence)
Prevalence
The commonness of disease at a time point/window
________ describes those factors/exposures related to disease
Measures of association
Cohort study - summary
Monitor a cohort exposed to a factor, and another unexposed
- risk of disease in each cohort is calculated
- RR: numerical multiplier of risk in exposed, compared to unexposed
Case-control study - summary
One group of cases, the other non-cases
- odds of exposure to punitiive risk factors is calculated for cases and non-cases
- OR: numerical multiplier of odds that expressed association between cases and exposure
_____ is an estimate of RR
OR
- interpreted the same, represent the same exposure, disease relationships
Measures of impact
Describe the impact of exposures/disease at the population level
RR is reflective of the _____
Whole population
If you contain exposure, then you will _______
Reduce the amount of disease
- there will still be some incidence of disease without exposure
What happens if you decrease the % exposure?
Many fewer exposed
- much less D+
- changed number of cases and distribution of cases
- occurs even if relative risk remains the same!!
If you have a high number of unexposed, then decreasing the %exposed would ______
Not reduce the incidence of disease!
- there may be another exposure that is causing disease
How is RR misleading at the population level?
- most disease is not related to exposure
- there is “baseline” disease occurence in E-, and disease is not reliant on a single type of exposure (multifactorial disease)
- still, those who are exposed carry a much higher risk of disease
Why do we still use RR?
The relative exposure: disease relationship tends to hold at this level, regardless of baseline risk variation
- no matter what the disease incidence in the E- group is, the RR does not change!!
What are 4 measures that compliment the RR and describe disease at the population level?
- attributable risk (exposed, aka risk difference)
- population attributable risk
- attributable fraction (exposed)
- population attributable fraction
Attributable risk
Difference in risk (incidence) between exposed and unexposed groups
- ____ cases per 100 animals
- a/(a+b) - c/(c+d)
- expresses increased incidence of disease, over the baseline, in exposed animals
AR directly applies to _____
Exposed animals only
- for every 100 animals exposed, expect an additional ____ cases
- over baseline
Population attributable risk
Difference in risk (incidence) between the total population and unexposed group
- (D+/N) / (D+/E-) minus D+/E-
- ___ cases per 100 animals in the population
- expresses amount of disease in the total population attributable to exposure —> if exposure were eliminated, ___% of population would not be affected with D+
PAR applies directly to _____
Population
- in ______ animals, ___ cases could be eliminated if you eliminate exposure
- if you eliminate exposure, everybody is at the same risk of disease as the E-
Attributable proportion
Proportion of incidence in exposed that is attributed to the exposure
- what fraction (%) of disease in the exposed is really due to exposure?
- (RR-1)/RR
- among exposed, ___% of D+ is due to exposure
Disease in E+ =
Baseline + disease associated with exposure
Population attributable proportion
Proportion of incidence in population attributed to exposure
- what fraction (%) of disease in the population is due to exposure?
- the disease incidence that would be eliminated if the E+ group had incidence equal to the E- group
- (incidence in population - incidence in E-)/incidence in population
- ____% of disease in the population can be eliminated if you eliminate E+
AR - summary
Expresses the increased incidence of disease, over the baseline, in exposed animals
- exposed
PAR - summary
Expresses the amount of disease in the total population attributable to exposure
- total population
