Path: Use of the Lab II Flashcards
How are results of laboratory tests interpreted?
reference interval or decision level is compared to patient’s test value
if accuracy of test and prevalence of dz known, can calculate predictive value of decision level
What is a reference interval?
relates to values for an analyte determined on a defined population of healthy individuals - “normal” constitutes middle 95% of bell - requires min of 40 subjects to establish
method dependent!
What is the lower limit of detection?
the lower limit of reference interval, most assays cannot capture concentration precisely equal to 0
What would an ideal reference interval for an analyte be like?
individual specific - can compare when ill to previous values when healthy, difficult
population-based = most cost effective and rational, must be free of any dz, especially one that will affect analyte being measured
remember that analytes can be gender or age specific
What are things to remember when choosing the most appropriate decision limit (cutoff)?
when pt populations clearly distinct, test can be used for screening and diagnosis
when populations overlap, FP and FN can result
use the ROC curve
What is the ROC curve?
plot of test sensitivity as Y coordinate and 1-specificity for false positive rate (FPR) as X coordinate
each point = operating point - generated using different cut-offs
area under curve is combined measure of test sensitivity and specificity at that operating point
cut off point should be point most located to top-left corner of the curve, offers best test performance
How are ROC curves calculated?
dividing cumulative sum of number of individuals w dz at each biomarker concentration by total number of such individuals
cumulative sum of those without dz at each biomarker concentration divided by total number
How can area under the ROC curve be used when choosing a particular test?
larger AUC means superior performance of the test
values b/w 0-1, closer to 1 is better
chance diagonal = 50/50 chance of particular disorder on test
independent of dz prevalence
may sacrifice sensitivity for specificity or vice versa depending on desired FPR
What is the concept of individuality?
expressed as ratio of within subject variability (CVi) to between subjects variability (CVg) = CVi/CVg
What is the meaning of the critical difference (CD)?
physicians take serial tests to monitor response to therapy - when is the magnitude of the change in values for analyte considered significant
What can affect quantitative values for analytes?
imprecision and intra-individual variation due to normal physiologic mechanisms
How is the degree if imprecision measured?
%CV (imprecision) = (SD/mean) x 100, SD and mean of different values
magnitude of analyte variation is CVa, magnitude of biological variability is CVb
What is the relative change value or critical difference?
change beyond that due to analytical variability and biological variability
How is the effectiveness of a test judged?
not by how quickly it is available, but what action is taken based on result
What are different types of test selection criteria?
single or group of tests to confirm clinical suspicion (variable)
reflex and reflexive testing
protocol driven
