Epi Midterm Flashcards
ANOVA
analysis of variance
best used for:
parametric, interval data, 3 or more groups
example:
effect of a medication on white blood cells in 3 different groups
students t-test
best used for:
parametric, nominal data, 2 groups
example:
effects of a diet in two different groups
Chi square
best used for:
nonparametric, nominal data, 2 groups
example:
assessing number of people who are exposed to a food type who acquire a food borne illness
used in analysis of contingency tables
Mann Whitney
best used for:
nonparametric, ordinal, 2 groups
used for clinical scores
Odds ratio
tells you the odds of developing a disease due to exposure
used in retrospective studies
Relative risk
rate of disease in exposed divided by the rate of disease in unexposed
used in prospective studies
Fishers exact
best used for:
nonparametric, nominal data, 2 groups, SMALL sample sizes
example:
acceptance to vet school between two small groups
wilcoxon rank sum test
best used for:
nonparametric, nominal data, 2 groups
example of a study in which multiple regression could be used?
changes in white blood cell count in dogs with a certain disease receiving either no treatment, drug A, or drug B
positive predictive value
probability that subjects with a positive screening test truly have the disease
negative predictive value
probability that subjects with a negative screening test truly don’t have the disease
adjusted rate
rate that is adjusted to eliminate the effects of a confounding variable
null hypothesis
Ho
there is no difference between the exposed group and the unexposed group
there is no association between variable A and variable B
incidence rate
the number of NEW cases of a disease in a population in a specified time period
hawthorne effect
participants alter behavior as a result of being in the study
healthy worker effect
workers are generally healthier and have lower disease rates than the general population as those who are disabled or physically ill cannot do the work
type I error
Alpha
rejecting the null hypothesis when it should be accepted
type II error
Beta
accepting the null hypothesis when it should be rejected
amplifying host
increases the chance of exposure
example:
infectious agents multiply in the host making infection more likely
second stage relative risk
two factors have a high relative risk
need to determine which is the most likely cause
can do so with a 2x2 contingency table
prevalence rate
total number of cases (both NEW and OLD) of a disease in a population in a specified time period
volunteer effect
form of selection bias
individuals that volunteer to participate in a study are different in some way from the population
p value
you are willing to be wrong (in repeated trials) aka to reject the null hypothesis when it should have been accepted 5% of the time
in other words: 95% of the time the observed difference in a population is real and not just due to chance
nominal
non quantitative values
examples:
Male, female
ordinal
represent a rank order
example:
clinical score of severity 1-4
interval
can interpret the degree of difference between values
example:
white blood cell count
cyclic disease pattern
periodic changes over several years
casses:
fluctuations in population immunity
attributable risk
risk difference
the incidence of disease that is directly related to exposure to the determinant
secular disease pattern
gradual change over a long period of time
causes:
pollution
management changes
slowly spreading agent
3 factors that would lead you to conclude that the association between a disease and a determinant was causal?
strength of association: large relative risk
temporality: factor occurs in the population before the disease
consistency: similar results from different studies on different populations
specificity: single cause
plausibility: makes sense biologically
coherence: does not conflict with present knowledge
experimental evidence: has been shown in natural experiments
incidence vs prevalence
incidence is the number of NEW cases
prevalence is the number of NEW and OLD cases
Difference between relative risk and odds ratio?
relative risk is the relation between disease exposure and non exposure- useful for prospective studies
odds ratio is similar but gives the chances of exposed individuals to develop disease - useful for retrospective studies
What can you use a scatter plot for?
Shows the relationship between two variables, helps detect outliers
best used with continuous variables
can analyze with a correlation coefficient
How can you determine if an outbreak is due to a single source or multiple sources?
second stage relative risk
When to use fisher’s exact vs chi square
Fishers exact should be used with small sample sizes (<5 in each group)
What measure of central tendency should be used for each type of data?
Nominal – mode
Ordinal - median
Interval - mean
Two pitfalls of epidemiologic studies
extrapolation: results that are obtained in the study population are extrapolated to the general population
information bias: inadequate medical records, lack of recall
population bias: cases may not be selected properly, example is cases admitted to a hospital for a disease (may be cases of the disease that are not admitted)
selection bias
relation between exposure and disease is different for participants and non-participants in the study
example: patients die before admission to the hospital
information bias
measurement error in assessment of exposure or disease
example:
loss of follow-up
incomplete medical records
prevarication bias
lying
example:
members of a religious group may lie about drinking
confounding
distortion of an effect of an exposure because it is mixed with the effects of an extraneous factor
What does “power” mean and how do you use it in study design?
probability of rejecting the null hypothesis when it should be rejected
1-beta
can be used to determine sample size
What are two things that influence a secular disease pattern?
new diagnostic tests
increased life expectancy
high prevalence in one area
How is the concept of target organ helpful in an outbreak investigation?
easier to direct the investigation towards a more specific area
will hopefully make it easier to detect the source of the outbreak
Advantages and disadvantages of prospective studies
Advantages:
establish incidence
true relative risk
assess more than one outcome
disadvantages:
expensive
small number of determinants
time delay in results
Advantages and disadvantages of retrospective studies
advantages:
inexpensive/quick
rare conditions
many determinants
disadvantages:
information may not be available
time sequence not known
types of disease transmission
direct: vector infects host
indirect: fomite, disease is acquired from something the vector touched
How can you determine estimate of incubation period from an outbreak curve?
time from the earliest part of the curve to the peak of the curve
Important characteristics of a diagnostic test?
highly sensitive
accurate
repeatable
skewness
measure of asymmetry of distribution curve
kurtosis
measure of the peakedness of the curve
endemic disease
occurs with predictable regularity
sporadic disease
occurs rarely and without regularity
epidemic disease
occurrence of a disease in a population in excess of what is normally expected
common source epidemic
all cases within one incubation period
example:
food borne illness
propagated epidemic
progressive epidemic
example:
infectious disease with animal-animal spread
diurnal disease pattern
changes that occur over a short period of time
bias
systematic error in data collection or evaluation that leads to an incorrect result
