Principles of Lab testing Flashcards
highest source of error
preanalytic- wrong tube, wrong patient
phases of testing
pre analytic
analytic
post analytic
source of error in analytic phase
imprecise instrument
source of error post analytic phase
results don’t make sense in clinical context
precision
consistent, repeatable results (cluster on target)
measures of central tendency
mean
median
measures of variability
standard deviation
coefficient of variation
meaning of coefficient of variation
low = precise
accuracy
closest to true value (near bullseye)
sensitivity
probability of a positive test given a disease
a/a+c
use of sensitivity
rule out a diagnosis- screening
specificity
probability of negative test given no disease
d/b+d
use of specificity
confirm a diagnosis
what changes with prevalence
PPV and NPV
what is constant
sensitivity and specificity
which s better for individual patients
PPV and NPV
prevalence
probability of having the disease (pretest probability)
PPV
probability of disease given positive result
a/a+b
NPV
probability of no disease given negative result
d/c+d
Bayes’s therom
P(D+|T+) = P(T+|D+) * P(D+)/P(T+)
truth table set up
top- Disease + disease -; a to b
side- test + test -; c to d
types of bias
availability
representativeness
anchoring
value-induced
availability
overestimation based on recent experience
representativeness
overestimation of rare disease based on typical picture
anchoring
failure to change theory based on new information
value-induced
overestimate based on association with bad outcomes
source of reference interval
95% confidence interval
determination of cut off values
trade off between sensitivity and specificity- use ROC curves
ROC curve
sensitivity vs 1-specificity
best with greatest AUC has greater discriminatory vale
LR+
sensitivity / 1-specificity
LR-
1-sensitivity / specificity