B8-045, B8-073 EBM Lectures Flashcards
screening tests should have high […]
sensitivity
diagnostic tests should have high […]
specificity
[…] test identifies a patient at risk for the condition
screening
[…] test determines presence of condition
diagnostic
calculation for senstivitiy
true positives/ disease positives
or TP/(FN + TP)
calculation for specificity
true negative/disease negatives
or TN/(FP+TN)
calculation for false positive rate
FP/(FP+TN)
or 1-specificity
calculation for PPV
TP/Test +
or (TP/TP+FP)
used to define a “cut off” value for a test
ROC curves
in ROC curves, the area under the curve can be used to compare
different tests
pretest probability and likelihood ratios can be used to determine
post-test likelihood of disease
calculation for LR positive
sensitivity/ (1-specificity)
calculation for LR negative
(1-sensitivity)/ specificity
if LR = 0, the test has […] value
no discriminatory
if LR+ is greater than or equal to 10, the test has […] value
high diagnostic value
if LR- is less than or equal to 0.1, the test has […] value
high diagnostic value
cross sectional studies are good at identifying [incidence/prevalence]
prevalence
what type of study design can be used to estimate exposure/disease burden?
cross-sectional
what type of study design can quickly be used to study several exposures/disease simultaneously?
cross-sectional
what type of study design can be used to compare rate of diseases between populations?
ecologic study
what type of study design is efficient for rare exposures?
observational cohort
what type of study design establishes temporal relationship between exposure and outcome?
observational cohort
what type of study design can be used to compare aggregate data across groups?
ecologic studies
relationships observed for groups are assumed to hold true for individuals
ecological fallacy
subjects alter behavior when they know they are being studied
Hawthorne effect
[…] bias: investigator’s evaluation is impacted by knowledge of exposure status
observer bias
[…] bias: exposure and control group are not treated equally
procedure bias
[…] bias: subjects provide inaccurate data
information bias
[…] bias: study population not representative of intended source poplulation
selection bias
the key indicator in evaluating effectiveness of a screening tool is
reduction in mortality
in areas of higher prevalence, PPV is […] and NPV is […]
in areas of higher prevalence, PPV is [higher] and NPV is [lower]
higher prevalence increases
pretest probability
a variable other than the one being studied influences the results
confounding error
a type of bias in which the patients in a study are not randomly assigned to a treatment group
selection bias
selection bias can be avoided by
randomization
[…] bias occurs when information is gathered at an inappropriate time in the study
late-look bias
(ex. sample of patients has less severe disease because those with more severe disease have died)
[…] bias occurs when the temporal characteristics of a disease are not considered
lead time bias
(important to consider in screening tests, can inflate survival time due to early detection)
[…] bias occurs when patients reports of the past are selective or exaggerated
recall
(common in case control studies)
how can recall bias be reduced?
decreasing time from exposure to follow up
[…] bias occurs when data outcomes are distorted by how data are gathered
measurement
(ex. Hawthorne effect)
how can measurement bias be reduced?
placebo groups
using objective and proven methods of data collection
the proportion of people who have a condition at one point in time
prevalence
how is prevalence calculated?
number of individuals with illness/total population
[…] disease prevention is preventing the actual occurrence of disease
primary
(ex. providing free condoms, vaccinations)
[…] disease prevention is aimed at early detection
secondary
(ex. screening tests, regular BP measurements)
[…] can help ensure appropriate external validity
randomization
