Use LEFT and RIGHT arrow keys to navigate between flashcards;
Use UP and DOWN arrow keys to flip the card;
H to show hint;
A reads text to speech;
47 Cards in this Set
- Front
- Back
How to measure incidence.
|
Cohort
|
|
define prevalence
|
number/proportion of cases in the population at a specific moment in time
|
|
define incidence
|
number of new cases that arise in the disease free population over a period of time
|
|
if mortality of a disease decreases, what happens to prevalence?
|
increased prevalence
|
|
Population is measured for prevalence of the disease at a specific point in time. What type of study?
|
cross-sectional study (aka prevalence study)
|
|
Study which compares one diagnostic test to the gold-standard?
|
cross-sectional study (aka prevalence study)
|
|
calculate sensitivity
|
true positives divided by total with disease (a/a+c)
|
|
calculate specificity
|
true negatives divided by total without disease (d/d+b)
|
|
How to remember how to get sensitivty and specificity off a 2x2 chart?
|
sE is before sP in alphabet, so sEnsitivity is left column and sPecificity is right column
|
|
SPIn
|
positive test in a specific test rules in the disease
|
|
SnNOut
|
negative test in a sensitive test rules out the disease
|
|
calculate Positive predictive value (PPV)
|
true positives divided by total positive results (a/a+b)
|
|
calculate Negative predicitve value (NPV)
|
true negatives divided by total negative results (d/d+c)
|
|
how to organize a 2x2 table
|
outcome/disease is always on top
|
|
when do you want a highly sensitive test?
|
when you want to screen for a disease to rule it out (usually more false positives)
|
|
low disease prevalence means what for PPV and NPV?
|
PPV goes down and NPV goes up
|
|
testing someone who has risk factors means what for PPV?
|
PPV goes up (higher pre-test odds)
|
|
calculate positive likelihood ratio (+LR); remember it's a ratio of likelihoods
|
true positive probability/false positive probability (sens/1-spec)
|
|
calculate negative likelihood ratio (-LR); remember it's a ratio of likelihoods
|
false negative probability/true negative probability (1-sens/spec)
|
|
how to remember how to calculate Likelihood Ratio? (where do you put the diseased population?)
|
truly diseased population always goes on top of the calculation: for +LR, true positives are on top, for -LR, false negatives are on top
|
|
when are exposures and outcomes in a prospective cohort study?
|
exposure in the future, outcome further in the future
|
|
when are exposures and outcomes in a retrospective cohort study?
|
exposure in the past, outcome in the more recent past
|
|
define the population of study in a cohort study
|
those who don't have the outcome but who could all potentially experience that outcome
|
|
define the population of study in a case control study
|
those who have the outcome presently
|
|
Cheapest type of study.
|
cross sectional study
|
|
when are exposures and outcomes in a case control study?
|
study group is defined by presence of outcome; exposures are measured (past or present)
|
|
Cohort study: what is measured?
|
cOhort measures Outcomes in a group with similar exposures
|
|
Case control study: what is measured?
|
casE control measures Exposures in a group with similar outcomes
|
|
Case Control studies cannot measure risk. Instead they measure what?
|
Odds Ratio: how much more likely it is for a person with outcome is to have an exposure than a person without an outcome
|
|
Define Absolute Risk (AR)
|
incidence of outcome
|
|
Define Relative Risk (RR)
|
incidence of outcome in exposed/incidence of outcome in unexposed
|
|
Define Attributable Risk (aka Absolute Risk Reduction or ARR)
|
incidence of outcome in exposed - incidence of outcome in unexposed
|
|
Define Relative Risk Reduction (RRR)
|
1-RR
|
|
Define Odds Ratio (OR)
|
odds of exposure in person with outcome/odds of exposure in person without outcome
|
|
Calculate Odds based on Probability
|
Odds = probability of event/1-probability of event
|
|
Calculate Probability based on Odds
|
Probability = Odds/1+Odds
|
|
Surgeon only operates on patients without significant comorbid conditions, then reports outcomes are better than other surgeons. What bias?
|
selection bias
|
|
Data gathered from 5 hospitals about stroke. One of the hospitals doesn't have MRI, so they use data from CT scans. What bias?
|
measurement bias
|
|
Study finds that construction workers who work 60 hours or more per week are more likely to have skin cancer. They conclude being over-worked causes skin cancer. What bias?
|
confounding bias
|
|
Patients with diarrhea are asked if they ate a specific food in the last week. Worry about what bias?
|
recall bias
|
|
New screening test claims to improve cancer survival by five years. Later study finds this cancer's average age of onset is earlier. What bias has this screening test created?
|
lead-time bias
|
|
Cancer prognosis is different in 2 hospitals. It is found that they use 2 different screeing test and one fails to catch the cancers with poor prognosis. What bias?
|
length bias
|
|
Type 1 (alpha) error vs Type 2 (beta) error: how to remember?
|
Type 1: rejects the null
Type 2: don't reject null (remember TWO is a DOUBLE negative) |
|
Define p value
|
probability that the differences found in the study occurred by chance
|
|
How to increase Power in a study.
|
increase number of subjects
|
|
How to decrease the Confidence Interval in a study?
|
increase number of subjects
|
|
Conclusion of insignificant difference based on confidence interval? (2)
|
it crosses 1 on plot of RR or Odds
it crosses 0 on plot of ARR |