EBM Flashcards
preferred measure of central tendency for continous data with symmetric distributions
mean
sensitive to extreme values and biased based on the presence of outliers
mean
preferred measure of central tendency for skewed distributions or distributions with outliers
median
measures of variability
2
- standard deviation
- variance
- difference between highest and lowest measurements
- sensitive to extreme observations
range
best representation of variation when data points are clustered around and dispersed from the mean
standard deviation
measures of central tendency
3
mean
median
mode
- utilized when researchers are trying to develop theory which necessitates taking an inductive approach
- participants are purposively selected and have a variety of lived experiences
grounded theory
used when examining a single phenomenon
phenomenology
involves making inferences
deductive reasoning
gets into the details of how language is beings used
discourse analysis
embedding oneself in the setting and observing behavior
ethnography
used to compare the means between two groups
t-test
used to test for differences in proportions between 2 groups
chi-square test
compares means of a continous variable between two or more groups
ANOVA
most appropriate method for determining efficacy when analyzing RCTs
ITT
includes every subject who was randomized ignoring non-compliance, protocol deviations, withdrawal, etc.
ITT
approach that actually measures the beneficial effects of treatent in patients who complete treatment
per-protocol analysis
when a small degree of non-compliance is observed, the analytic sets for the ITT and per-protocol analysis
contain nearly the same subjects, resulting in minimal differences between the two
case-control study use […] ratio
odds
cohort or RCTs use […] ratio
hazard
confidence interval below 1 indicates
data are consistent with a statistically significant protective effect
observational study with forward directionality
cohort
useful to study diseases with a long latency period
case-control
occurs when a third factor affects the magnitude of the relationship between the exposure and disease
effect modification
occurs when investigator inadvertently conveys their expectations to the subject, who then produces the expected result
Pygmallion effect
patients with severe disease are less likely to be studied because they are more likely to succumb to the disease before being studied
late look bias
best way to control confounding in a non-observational study
randomization
occurs when a third factor is either positively or negatively associated with both the exposure and outcome of interest
confounding
occurs when observer is aware of what exposure group the study subject has been assigned to
observer bias
occurs with earlier detection of a disorder which makes the prognosis seems better when in reality the disease was merely detected earlier
lead time bias
occurs when selected subjects are not representative of the population to be studied
selection bias
systematic bias in regard to classifying subjects
measurement bias
ability to use results from one study to draw conclusions about populations different than that used in the study
generalizability
occur when a study is conducted in very specific population and an attempt is made to apply findings beyond the represented population
problems with generalizability
the stricter the inclusion/exclusion criteria the less…
generalizable
can serve to improve the generalizability of specific study findings
meta-analysis
what measure of central tendency is least effected by outliers?
mode
stating there is an effect or difference when none exists
type I error
stating there is not an effect or difference when one exists
type II error
in order to minimize outcome switching and p-hacking..
only outcomes that are known or predicted to be associated with the patient’s response to treatment should be collected
small sample size with just enough data to answer a limited question
QI project
large samples of data with additional data that may be analyzed later if needed
research
large data sets of all available data related to patient outcomes
accountability
focused on improving care for patients
QI projects
compares outcomes of different groups of patients or healthcare systems
accountability
focused on gaining new knowledge
research
seeks to eliminate bias
research
adjusts for bias
accountability
accepts bias as inherent to the process
QI
alway inherent to the process or system being studied
common cause variation
non-random
results in an unpredictable system
special cause variation
due to irregular or unnatural causes
special cause variation
- random
- results in predictable system
- affects all outcomes in the process
common cause variation
the same […] value can reflect different distribution of data
mean
midpoint value within the distribution of data
median
may not shift with different distributions of data and is not necessarily affected by sample size
mean
- organized around the median
- simple bar graph
- doesn’t use statistical measures
run chart
- organized around the mean
- used control limits (3 stan. dev from mean)
process control chart
three standard deviations from the mean or greater
special cause variant
lie within 3 standard deviations above or below the mean
common cause variants
