Laboratory QA/QC

1. Laboratory QA/QC An Overview

2. Definitions (1) • Quality Assurance: QA is defined as the overall program that ensures the final results reported by the laboratory are correct. • QA is a broad plan for maintaining quality in all aspects of a program. • QA establishes the need for quality control.

3. Definitions (2) • Quality Control: The measures that must be included during each laboratory procedure to verify that the test is working properly.QC refers to routine technical activities with the purpose to control error. • QC can be considered as the “HOW” of the QA process. • QC is applicable to field, lab and office procedures (administration).

4. Definitions (3) • Quality Assessment - quality assessment (also known as proficiency testing) is a means to determine the quality of the results generated by the laboratory. Quality assessment is a challenge to the effectiveness of the QA and QC programs. • Quality Assessment may be external or internal.

5. Why QC? “QC aims at simply ensuring that the results generated by the test are correct. However, QA is concerned with much more. It checks whether the right test is carried out on the right sample, and that the right result and right interpretation is delivered to the right person at the right time”

6. Why QA? We need QA to: • Understand data reliability; • Quantify areas of analytical uncertainty; and • Standardize measurement to allow for repeatable and comparable data across time and place.

7. QA vs. QC • Quality Assurance (QA) • broad program plan • establishes the need for QC • Quality Controls (QC) • individual checks and balances • the “How to” of QA

8. Where QC is applicable? • Quality control is applicable in all aspects of a soil, plant and water sampling project including: • Field data collection and sampling • Laboratory analysis and processing • Data evaluation and assessment • Reporting and project documentation QC provides steps to ensure lab data will meet defined standards of quality with a standard level of confidence

9. QC in the Field • In most cases field QC (soil, water and plant sampling) is out of laboratory control; • QC is particularly critical in field data collection; • Often the most costly aspect of any project and the most limiting factor is field sampling; • Data is never reproducible under the exact same condition or setting; • Therefore, field sampling QA is also needed to assure that best possible (most reliable) set of data is obtained.

10. Choice of Sampling Unit -What Does a Sample Represent? Irrigation project 7.5 cm core Representing 10 ha

11. www.odc.gov/noeh/dls QC in the laboratory • Laboratory data analysis, data measurement, and data acquisition: • Chain of custody forms • Equipment calibration • Storage practices • Analytical methods • Holding times • Limit of detection (LODs), previously known as MDLs.

12. Variables affecting results quality • Educational background and training of personnel; • Condition of the samples; • Controls used in the test runs; • Reagents quality; • Maintenance status of equipment; • Interpretation of the results; • Recording of results; and • Reporting of results.

13. Errors in measurement • True value: This is an ideal concept which practically cannot be achieved. • Accepted true value: The value approximating the true value, the difference between the two values should be negligible (not statistically significant). • Error: The discrepancy between the result of a measurement and the true (or accepted true value).

14. Sources of error • Input data required: Such as standards used, calibration values, and values of physical constants; • Inherent characteristics of the quantity being measured; • Instruments used: Accuracy, repeatability; • Observer unreliability: Reading errors, blunders, equipment selection, analysis and computation errors;

15. Sources of error (cont.) • Environment: Any external influences affecting the measurement; and • Theory assumed: Validity of mathematical methods and approximations.

16. Random Error • An error that varies in an unpredictable manner, in magnitude and sign, when a large number of measurements of the same quantity are made under effectively identical conditions. • Random errors create a characteristic spread of results for any test method and cannot be accounted for by applying corrections. • Random errors are difficult to eliminate, but repetition reduces the influences of random errors.

17. Random Error (cont.) • Examples of random errors include: • errors in pipetting; • changes in incubation period; or • the time used for extraction/centrifuging. • Random errors can be minimized by training, supervision and adherence to standard operating procedures (SOPs).

18. Random Errors

19. Systematic Error • An error that, in the course of a number of measurements of the same value of a given quantity, remains constant when measurements are made under the same conditions, or varies according to a definite law when conditions change. • Systematic errors create a characteristic bias in the test results and can be accounted for by applying a correction. • Systematic errors may be induced by factors such as variations in incubation temperature, change in the reagent batch or modifications in testing methodology.

20. Systematic Errors

21. QC: Internal vs. External Measures • Internal Quality Control: • “Controllable” by those responsible for performing the laboratory analysis. • External Quality Control: • A “set of measures” established for and conducted by people outside the analytical laboratory (lab auditors, regional or national laboratories, accreditation process, etc).

22. Quality control (QC): Internal • Internal Quality Control: • Equipment calibration • Proper training and certification of practitioners • Proper sampling techniques • Proper data documentation

23. Internal Quality Control Samples IQC samples comprises either • In-house prepared aliquot of known values, or • International standards with values within significant ranges for the element to be measured.

24. Quality control (QC): External • External quality control: • Performance audits • Split sample analysis • Replicate (duplicate) sample analysis

25. QA/QC: Data objective and key concepts • Successful data collection and analysis is dependant upon “The PARCC Parameters”: • Precision • Accuracy • Representativeness • Completeness • Comparability The key concepts of QA/QC are the “PARCC” Parameters – the WHY of the QA

26. Key concepts of QA/QC: • Precision - • degree of agreement there is between repeated measurements of the same characteristic • can be biased – meaning there is a consistent error in the results • Accuracy - • measures how close data results are to a true or expected value – does not allow for bias

27. accuracy = (average value) – (true value) precision represents repeatability bias represents amount of error low bias and high precision = statistical accuracy Key concepts of QA/QC: Accuracy http://www.epa.gov/owow/monitoring/volunteer/qappexec.html

28. http://pubs.usgs.gov/fs/fs-0058-99 Key concept of QA/QC: Representativeness • Representativeness - • extent to which measurements actually represent the true environmental condition or population at the time a sample was collected. • Representative data should result in repeatable data  Does this represent this?? 

29. Key concepts of QA/QC: Comparability • Comparability - • the extent to which data can be compared between sample locations or periods of time within a project, or between projects  Will similar data from these sites be Comparable ?? 

30. Quality Assurance (QA) broad program plan establishes the need for QC Quality Controls (QC) standardized tests and methods the “HOW” of QA Review: QA vs. QC

31. Internal QC program for soil and water sample testing An internal quality control program depends on the use of internal quality control (IQC) samples, and using statistical analysis methods for interpretation.

32. Shewhart Control Charts A Shewhart Control Chart depend on the use of IQC samples and is developed in the following manner: • Put up the IQC specimen for at least 20 or more sample runs and record down the readings; • Calculate the mean (x) and standard deviations (Sd); • Make a plot with the sample run on the x-axis, and concentration readings on the y axis.

33. Shewhart Control Charts (cont.) • Draw the following lines across the y-axis: mean, -3, -2, -1, 1, 2, and 3 Sd; • Plot concentration reading obtained for the IQC specimen for subsequent sample runs. • Major events such as changes in the reagent batch and/or instruments used should also be recorded on the chart.

34. What is Shewhart Control Chart? A Shewhart control chart consists of: • Points representing a statistic (e.g., a mean, range, or proportion) of measurements of a quality characteristic in samples analyzed at different times [the data]; • The mean of this statistic using all the samples is calculated (e.g., the mean of the means, mean of the ranges, or mean of the proportions);

35. What is Shewhart Control Chart? (cont.) • A center line is drawn at the value of the mean of the statistic; • The standard error (e.g., standard deviation) of the statistic is also calculated using all the samples; and • Upper and lower control limits (sometimes called "natural process limits"), indicating the threshold at which the process output is considered statistically 'unlikely' are drawn typically at 3 Sd from the center line.

36. Westgard rules • The formulation of Westgard rules were based on statistical methods. Westgard rules are commonly used to analyze data in Shewhart Control charts. • Westgard rules are used to define specific performance limits for a particular analysis and can be use to detect both random and systematic errors.

37. Westgard rules • There are six commonly used Westgard rules of which three are warning rules and the other three are mandatory rules. • The violation of warning rules should trigger a review of test procedures, reagent performance and equipment calibration. • The violation of mandatory rules should result in the rejection of the obtained results.

38. Shewhart Chart +3 sd Sample reading +2 sd +1 sd Target value -1 sd -2 sd -3 sd Sample run

39. Warning rules • Warning 12SD : It is violated if the IQC value exceeds the mean by 2SD. It is an event likely to occur normally in less than 5% of cases. • Warning 22SD : It detects systematic errors and is violated when two consecutive IQC values exceed the mean on the same side of the mean by 2SD. • Warning 41SD: It is violated if four consecutive IQC values exceed the same limit (mean  1SD) and this may indicate the need to perform instrument maintenance or reagent calibration.

40. Mandatory rules • Mandatory 13SD : It is violated when the IQC value exceeds the mean by 3SD. The test is regarded as out of control. • Mandatory R4SD : It is only applied when the IQC is tested in duplicate. This rule is violated when the difference in Sdbetween the duplicates exceeds 4Sd. • Mandatory 10x : This rule is violated when the last 10 consecutive IQC values are on the same side of the mean or target value.

41. Westgard Rules: 1 3SD +3 sd Sample reading +2 sd +1 sd Target value -1 sd -2 sd -3 sd Assay Run

42. Westgard Rules: 10X +3 sd Sample reading +2 sd +1 sd Target value -1 sd -2 sd -3 sd Assay Run

43. Follow-up action in the event of a violation There are three options as to the action to be taken in the event of a violation of a Westgard rule: • Accept the test run in its entirety: This usually applies when only a warning rule is violated. • Reject the whole test run: This applies only when a mandatory rule is violated. • Enlarge the grey zone and thus re-test range for that particular test run: This option can be considered in the event of a violation of either a warning or mandatory rule.