Quality Control Definitions PDF

Definitions… There are several common terms used when analysing laboratory performance Quality Control...

Definitions… There are several common terms used when analysing laboratory performance Quality Control Accuracy Precision Specificity Definitions Sensitivity Lab. Management & Quality Control Dr. Mohamed Abdelfatah B.Sc., M.Sc., Ph.D. Clinical Chemistry Accuracy? The agreement between your value and the ‘true’ value Accuracy? How correct your result is Determined absolutely by direct comparison to a reference value More commonly assessed by using an assayed control serum, with accurate values assigned by the manufacturer The closer your result to the target value, the greater your accuracy Precision? Precision? The reproducibility of your results (i.e. the agreement between replicate measurements) The closer your results are to each other, for the same analyte, in the same serum, the better your precision The reproducibility of your There are 2 ways in which precision is assessed: results Within run performance (intra-assay precision) Between run performance (inter-assay precision) Accuracy & Precision - Accuracy & Precision - Example 1 Example 2 Accurate and precise The ideal situation Imprecise but accurate Repeat results are close to one another Results are widely spread, giving Mean is close to the ‘true’ value poor precision Lab can have confidence in single test results The mean is close to the ‘true’ value, giving apparently good No need to continually repeat tests accuracy An unacceptable situation Labs cannot waste resources on repeat runs to get an acceptable level of accuracy  Accuracy & Precision - Solving precision and accuracy problems Example 3 Poor accuracy relatively easy to solve Precise but inaccurate! Often a calibration problem Results are close together, Poor precision more difficult giving good precision A variety of causes Mean is not close to the Poor quality reagents ‘true’ value, giving poor Badly maintained instruments accuracy Inadequate training Specificity? Specificity? The ability of a method to measure Consequences of a lack of specificity solely the component of interest Falsely elevated values may occur A lack of specificity will affect Structurally similar hormones FSH, LH, TSH & hCG all have an identical alpha-subunit accuracy Drugs (both therapeutic drugs and Drugs of Abuse) The test is measuring Falsely low values may also occur components other than the Bromocresol Purple (BCP) method with bovine albumin analyte of interest The test does not measure the analyte 100% Bovine QC serum cannot be used with this method Sensitivity? Analytical & Diagnostic The ability to detect small quantities of a measured component Sensitivity Specificity Will affect both precision and accuracy at the bottom end of the clinical range Analytical Analytical How is sensitivity established? the ability of a diagnostic test to detect very The ability of a method to detect analyte with By determining at what point an assay’s precision reaches an unacceptable small amounts of the analyte less interference level Diagnostic Diagnostic Ability of a test to detect a given disease or Ability of a test to correctly identify the absence of condition. a given disease or condition. A proportion od individuals with a disease A proportion of individuals without a disease who who tested positive tested negative Ability to exclude false positive Ability to exclude false negatives Determination of Diagnostic Specificity and Sensitivity  Normal Distribution (Gaussian Curve) Normal distribution Values fall randomly about a mean value The mean (x) is the average Statistics… Standard Deviation of the set of values Equal numbers of results lie above and below the Coefficient of Variation mean Standard Deviation (SD) SD is defined as the square root of the sum of the Precision? squares of the single value deviations from the mean, divided by the number of the values minus one (and is quoted in the same unit of measurement) How disperse the values are How is precision quantified statistically? By measuring the Standard Deviation (SD) of the set of results  (xi - x) 2 SD = ( ) n-1  The average What does deviation for the set Mean result (x) = 100 mmol/L of results Standard Standard Deviation Standard deviation (SD) = 1.0 mmol/L Deviation The lower the SD, example Number of results (n) = 100 tell us? the better the precision Mean ±1 SD Mean ± 2 SD By the laws of statistical probability, By the laws of statistical 95% of all results should fall within ± probability, 68% of all 2 SDs of the mean results should fall within ± 1 SD of the mean i.e. 19 out of 20 In this example, 68% of In this example, 95% of results fall results fall within the range within the range 98 – 102 mmol/l 99 – 101 mmol/l Statistically, it is acceptable for 5% of results to fall outside this range Which is more Precise? Coefficient of Variation Potassium SD = 0.1 mmol/L A %CV takes into consideration the magnitude of the overall result Sodium SD = 2.0 mmol/L SD It’s impossible to say from this information alone! CV = x 100% What other information is required? The magnitude of the results Mean (x) The %CV expresses the SD as a percentage of the mean The lower the %CV, the better the precision Example: Why use a %CV for analysis of results? Potassium (mean = 5.0 mmol/l) %CV = (0.1 / 5.0) x 100% = 2.0% It allows comparison of precision on different sets of data, with Sodium (mean = 140 mmol/l) different magnitudes %CV = (2.0 / 140) x 100% = 1.4% Sodium has the better CV, and in this case is performing better than potassium