Analytical Performance Flashcards
(25 cards)
define analytical and diagnostic performance
Analytical: assesses how well an instrument/method measures analyte of interest
Diagnostic: assesses how well a test can differentiate between a healthy/unhealthy patient
what 4 things can help assess analytical and diagnostic performance
precision, accuracy, specificity, sensitivity
why is there need to monitor analytical performance in clinical labs
- ensures reliability of data (quality control) to assess stability of lab tests
- meet reqs of regulatory bodies
- validation (characteristics of assay), calibration (maintain accuracy), comparison ( compare between diff labs)
what is an inter-assay
repeating assessment on a diff day to ensure precision
what is a precision profile
in depth analysis of precision over a period
what is SD
a numerical value that indicates the extent of deviation for a defined data set- can help define how much variation occurs from the mean
what does a low/high SD value mean
the data points are close to the mean= low
the data points are spread form the mean = high
what is the formula for SD
see notes
what is the clinical significance of SD
- can assess analytical + diagnostic performance
- ref ranges in lab are usually established as the mean +- 2SD
what is the coefficient of variation
- scales the standard deviation by size of the mean
- makes it possible to compare across variables measured on diff scales
- allows comparison regardless of magnitude of analyte conc
wjat is the formula for CV%
SD/mean x 100
what is with-in-run precision
intra assay - one analyst
obtains the closeness of agreement between results of successive measurements obtained under identical conditions
i.e samples representing low, medium and high conc values are assayed in replicates of n=20 on the same day
what is between run precision
inter assay- maye have more than one analyst
i.e QC material representing low med and high conc valyes are assayed in replicates of around n=20 in diff analytical runs over a number of days
write a note on precision profile
is an extensive assessment of precision
- examine method variation over a number of analyte concs, days and optionally over 1/2 runs per day
- make a plot of %CV v Conc
- this helps make working ref ranges
- ensure that imprecision is below establised level
what are the requirements for precision profile
at least 3 replicates must be observed for each run + each run must have the same number of replicates
analyte conc must be known
how can accuracy be quantified
measuring inaccuracy or %error
inaccuracy = nominal value - experimental value
% inaccuracy = nominal value - exp value / nominal value x 100
define accuracy and precision
accuracy = proximity of results to true value
precision = repeatability of measurement
what is recovery study (matrix spiking)
tests ability of an assay to measure a known amount of analyte from a sample matrix
to do this, add known cocn of analyte (A)to sample matrix and measure recovered conc (C)
to account for matrix background generate a baseline measurement (B) i.e measure in absence of A
conc recovered = A-B
% recovery = (C/A) x 100
define specificity
measures proportion of negatives that are correctly identified
low specificity = high chance of false pos
can be determined by measuring cross-reactivity with molecules that are chemically similar i.e antigen/antibody reactions
what is formula for specificity
number of true negs/ number of tru negs + number of false pos
what is sensitivity
relates ability to identify pos results correctly
higher the sensitivity = lower chance of false neg result
what is the limit of detetion
the lowest quantity of a substance that can be distinguished from absence of that substance with degree of confidence
i,e a sensitive assay can measure low levels of the analyte
why may method comparison be needed
-purchase of a new analyser, new method introduced, more efficient/economical adaptation to existing method
what sort of data can be used to compare methods?
data from both methods representing the same clinical samples
independent data- diff clinical samples that represent similar study populations
