Statistics and QC part 2 Flashcards
this is the agreement among replicate measurements; can be only determined with multiple measurements; reproducibility
precision
this is the agreement between the measured quantity and its true value; can be determined by only one measurement; closest to true value
accuracy
precision is determined by calculating the ________ and coefficient of variation statistics.
standard deviation
what are three guidelines used to establish limits for accuracy and precision in the clinical lab?
desirable limits for accuracy and precision must be defined 1 individually for each test or type of analysis performed 2 at each level of medical significance and 3 with regard to the flow of work in the lab
the difference between the measured value and the true value and assesses uncertainty in measurement
error
the amount of analytical imprecision and systematic error permitted for a single test result; also a point at which the results from the assay may no longer by useful
total error
unpredictable error including precision, imprecision, reproducibility, and repeatability
random error
what can cause random error?
1 instability of the instrument
2 variations in the temp
3 variations in the reagents and calibrators
4 variability in handling techniques, such as pipetting, mixing, and timing
5 variability in operators
this describes error that is consistently high or low and is predictable; caused by factors like deteriorating reagents or an instrument that has lost calibration; introduce bias
systematic error
What are causes of constant systematic error?
1 presence of interfering substances in all samples or reagents
2 presence of interfering substances in the reaction between the analyte and reagent
3 presence of interfering substances that destroy or inhibit the reagent
What are causes of proportional systematic error?
1 SD changes from reagent lot to reagent lot
2 SD changes from calibration to calibration
What is the calculation for SDI? What is it used for?
(Lab mean-Group Mean)/Group SD = SDI
USed to determine systematic error or bias
A target SDI is _____. An SDI of _____ or less is considered acceptable. SDI of _____ or greater is generally considered to be unacceptable performance and remedial action is required.
0.0; 1.25; 2.0
This quality control is used in real (actual) time to ensure day to day consistency of measurements.
Internal QC
This QC refers to a process of retrospectively comparing results from different laboratories through the use of an external agency; major purpose is to assess accuracy by comparing results with other labs with similar methods.
External QC
What is the target range?
95.5% confidence limits of the data or +- 2 SD
What are the # of data points, # of runs, and # of days required by NCCLS for a new lot of QC?
at least 20 data points from 20 separate analytical runs over no more than 30 days
What are the # of data points, # of runs, and # of days required by NCCLS to establish a target range for a new QC system or manufacturer?
40 data points from at least 20 analytical runs collected over no more than 30 days
this is an abrupt change in the mean that becomes continuous; it is evidence of systematic error and must continue for 5-7 data points
shift
What can cause a shift?
1 sudden failure or change in the light source
2 change in reagent formulation
3 change of reagent lot
4 major instrument maintenance
5 sudden change in incubation temperature
6 change in room temp or humidity
7 failure in the sampling or reagent dispense system
8 inaccurate calibration/recalibration
this is a gradual change in the mean that proceeds in one direction; it is an example of systematic error; continues for 5-7 data points
trend
What are possible causes for trends?
1 deterioration of the instrument light source
2 gradual accumulation of debris in sample/reagent tubing
3 gradual accumulation of debris on electrode surfaces
4 aging of reagents
5 gradual deterioration of control materials
6 gradual deterioration of incubation chamber temp
7 gradual deterioration of light filter integrity
8 gradual deterioration of calibration
What is the calculation for percent recovery? and what is an acceptable range?
Experimental Result/Theoretical Result x 100 with a range of 95%-105% being acceptable
Proportion of individuals with that disease who test positively with that test
Diagnostic sensitivity
Proportion of individuals with out the disease who test negatively for the disease
Diagnostic specificity
Number of diseases individuals with a positive test
True positive
Number of individuals without the disease with a negative test
True negative
Predictive value of a positive test (PV): fraction of positive tests that are true positives…formula?
TP/ (TP+FP) x100
Predictive value of a negative test: fraction of neg tests that are true negatives…formula?
TN/ (FN+TP) x100
Efficiency of a test is the fraction of all test results that are either true positives or true negatives…formula
(TP+TN) / (TP + TN + FP + FN) x100