Lecture 12 10/17/24 Flashcards
What is analytical performance?
how well the test measures
What is diagnostic performance?
how well the test detects disease
What is pre-analytical error?
things that go wrong before analyte measurement
What are the problems that fall into pre-analytical error?
-patient preparation
-sample acquisition
-sample handling
-sample storage
What is analytical error?
things that go wrong during analyte measurement
What are the problems that fall into analytical error?
-instrument
-reagents
-instrument operator
What is post-analytical error?
things that go wrong after analyte measurement
What are the problems that fall into post-analytical error?
-data reporting
-data interpretation
What is random error/imprecision?
-variation of results during repeated measurement
-represented mathematically by coefficient of variation of repeatability data
What can be reflected by imprecision?
unpredictable events that cause random measurement errors
How is “true” analyte concentration determined at the manufacturing level?
series of measurements and calibrations is done with a reference measurement procedure and reference material representing “true” conc.
How is “true” analyte concentration determined at the clinic level?
referral lab results are often a proxy for “the truth” relative to in-clinic equipment
What is systematic error?
-difference between a measured result and some representation of “true” concentration/activity
-typically predictable
What is constant bias?
measured value is consistently above or below the “true” value throughout the range of measured conc/activity
What is proportional bias?
difference between the measured value and “true” value varies with analyte conc/activity
What is important to note about constant and proportional bias?
they are not mutually exclusive and can occur together
What are two examples of every day bias?
-malfunction that causes measurement error in a predictable, systematic way (problem to correct)
-two instruments are used to measure the same analyte in the same sample and give slightly different results (normal)
How does correlation differ from agreement?
correlation indicates an association and trending in the same direction, but not necessary giving the same result
How can bias between two instruments be eliminated?
-use instrument-specific reference intervals and medical decision thresholds to interpret data
-use the same instrument to follow trends over time
What is assayed quality control material?
substance of known concentration that is measured during quality control
Why do we run controls?
to document that the machine is measuring correctly and without analytical error
What is control data?
results of measuring quality control material
What are control limits?
numerical thresholds for deciding whether control data are acceptable or not
What is calibration?
adjusting how an instrument measures according to the assigned concentration of a calibrator solution
What are the characteristics of instrument flags?
-symbols printed with data
-give warnings or indicate problems
-should not be ignored
-can be resolved using instrument manual
What are practical tips for quality control?
-interpret control data at time of measurement
-archive control data and review over time
-do daily QC on CBC instruments; follow manufacturer instructions for CHEM instruments
-re-run controls before running patient samples when problems occur
How can pre-analytical error be prevented?
-proper patient preparation
-proper sample acquisition
-proper sample labeling, handling, storage, and transport
How can analytical error be prevented?
-follow manufacturer instructions
-perform QC and review control data
-do not use expired materials
-do not ignore instrument warning flags
How can post-analytical error be prevented?
-use appropriate reference intervals
-check data transfer to medical record
What are interferents?
-substances that cause analytical error due to interference in the test system
-pre-analytical factors that cause analytical error
How can hemolysis be caused by patient or sample prep. error?
-rough venipuncture
-poor sample handling
How can hemolysis be caused by patient disease?
intravascular hemolytic anemia
How can lipemia be caused by patient or sample prep. error?
did not fast patient
How can lipemia be caused by patient disease?
-endocrine disease
-idiopathic hyperlipidemia
-pancreatitis
How can icterus be caused by patient disease?
-hemolytic anemia
-cholestatic liver disease
What are the characteristics of population-based reference intervals?
-how we decide if results are “normal” or “abnormal”
-reflect analyte conc. in a clinically healthy adult pop.
What are the lower and upper reference limits?
statistical estimates calculated from all reference sample data that become more accurate with larger reference group size
What are tips for using reference intervals?
-reference intervals are method-specific
-RIs in textbooks may not apply to your chosen method
-patient values can be within the RI and a pathologic state can still be present
-some animals will have values outside of the RI but still be healthy
-values should be interpreted in addition to patient physical exam
What are the characteristics of medical decision limits?
-based on expert consensus
-help us decide whether a disease is present
-help stratify patients for disease classification, monitoring, and treatment
What are the steps of implementing a new lab test?
-characterize analytical performance
-implement QA/QC procedures to monitor analytical performance
-develop reference intervals or decision thresholds
-characterize diagnostic performance
What are the metrics used to evaluate diagnostic test performance?
-sensitivity
-specificity
-positive predictive value
-negative predictive value
