# General Laboratory: Common Reasons for Out-of-Range QC Results Online Training

## Identify common reasons for out-of-range QC results, the principles of quality control and solutions to address common causes for out-of-range QC results. This clinical laboratory training qualifies for continuing education units (CEU).

Welcome to the General Laboratory: Common Reasons for Out-of-Range QC Results Online Training course. This course will focus on causes for out of range QC results that arise during the testing of QC samples. Select Next to continue. Course created by: Nils B. Person, Ph.D., FACB Senior Scientist Global Product Education Siemens Healthcare Diagnostics Graduate degree, State University of New York, Buffalo Upon successful completion of this course, you will be able to: Identify common reasons for out of range QC results Describe the principles behind how QC works Apply solutions for common causes of out of range QC Select Next to continue. All "out of Range" QC Results Using +/- 2 SD limits Targets not set effectively QC sample System None of these indicate any problem with the analytical system! Select Next to continue. "All results outside of a 2 SD range indicate a problem." Select Next to continue. “QC” = testing samples of known concentration to detect change in method performance Today’s QC results are checked against expected values to determine if there has been a change in performance Cannot use a single “target” value for QC samples since there is normal, routine variation in results Use simple statistical model to predict the expected pattern of future results Select Next to continue. Select the play arrow to begin the video. When complete, select Next to continue. Frequency: number of points at each concentration Select Next to continue. Frequency (Number of points at each concentration) Mean: central value around which all others fall — the most typical value Standard Deviation (SD): shows how widely the replicate results spread — measure of precision Ca, mg/dl If 95% of good results fall within +/- 2 SD, then 5% of good results must fall outside 2 SD. With 2 SD limits, by definition, 5% of the results for each QC sample run will be outside the defined range, even when nothing is wrong. 50 years ago: single control sample tested together with a small batch of patient samples manual assay with poor precision Using 1 QC sample: Using 2 QC samples: Using 3 QC samples: Using 2 QC samples each for 25 tests: 1 good QC result > 2 SD in 5% of runs 1 good QC result > 2 SD in 9% of runs 1 good QC result > 2 SD in 14% of runs 1 good QC result > 2 SD in 92% of runs! QC Protocol: 2 Levels of QC per assay 25 assays on system 2 QC runs a day 7 days per week 2 x 7 x 52 weeks = 728 QC runs On 92% of the 728 runs 1 of the 50 QC results exceeds 2 SD or 670 “out of range” QC due to normal analytical variation and nothing is wrong! Select Next to continue. “I’ve been using 2 SD ranges for QC for years and I don’t see that many out of range results. What gives?” What do you use for your target SD or range? Package insert SD or range Peer group SD Have you ever “adjusted” the SD because the range seemed “too tight”? Then you are not using 2 SD ranges The values used for target SD are larger than the actual SD Has a much lower false positive rate Example: QC material package insert range LEVEL 1 - 14151 Glucose - Hexokinase mg/dL 86.0 72.2 - 100 Actual performance: n = 73 Mean = 80.5 SD = 1.41 CV = 1.8% Package insert range is +/- 9 SD Compare to previous days Compare to results from other QC samples Look for patterns Remember: most of the time, a single QC result between 2 SD and 3 SD does not represent a real problem “Package insert values should be used to set QC target ranges to make sure results are accurate” Analytical processes are subject to inherent variation Replicate results for a single sample follow a Gaussian distribution All QC rules use these assumptions about QC data Expected results will be evenly distributed on both sides of the mean Expected results can be predicted using mean and SD - 2SD 2 SD - 1 SD 1 SD x Target mean Target mean If the QC target mean does not reflect the actual mean of the instrument, this basic assumption will not be met… Statistical QC will not work effectively. QC material package insert Kit Level Units Mean SD 2SD Range Progesterone 4 5 6 mg/ml 1.46 3.0 21 0.178 0.30 1.8 1.10 - 1.82 2.4 - 3.6 17.4 - 25 Actual: Mean:; 2.67 0.25 Range: 2.2 – 3.2 Mean: Calculate preliminary mean from replicates run on system Best: 20+ replicates over 20+ days Alternative: 6 - 10 replicates in a single run will do Use QC material package insert, peer group, or manufacturer’s expected mean as reference to validate actual mean not as the target mean How do i know the mean from my instruments is ok? Clinical judgement is acceptable Follow QC material manufacturer's recommendations Is your mean within +/- 1 SD of the package insert mean? Initial QC targets: Mean - Control Material: Lot: 20 Level: 6 QC Data: 2.22 2.26 2.28 2.29 2.47 2.25 2.25 2.30 2.19 2.24 Mean = 2.275 n = 10 Analyte Level Units Mean SD Free T4 4 5 6 ng/dL 0.8 1.3 2.2 0.08 0.10 0.17 Pl mean +/- 1SD: 2.03 - 2.37 Can also use peer group mean as a reference value to validate the target mean 2.275 Note of caution: Usual reference points used to validate a new mean... Package insert mean Peer group mean Manufacturer's experienced mean May no longer be valid as reference points if there has been a change in QC material performance Most common reason: New reagent lot SD New Lot of QC material: Best: estimate using at least 20 replicates across multiple days or ... Alternative: If means are comparable, use CV from prior lot to estimate QC manufacturer's expected SD Peer group SD (CV) CV (old lot) x Mean (new lot) = Est. SD (new lot) Control Material: Lot: 20 Level: 6 QC Data: 2.22 2.26 2.28 2.29 2.47 2.25 2.25 2.30 2.19 2.24 Mean = 2.275 SD = 0.075 n = 10 Initial QC targets Mean: 2.275 SD: Control Material: Lot: 20 Level: 6 QC Data: 2.22 2.26 2.28 2.29 2.47 2.25 2.25 2.30 2.19 2.24 Mean = 2.275 SD = 0.075 n = 10 Prior Lot Mean: 2.48 SD: 0.12 CV: 5.0 % Estimated SD: 2.275 X 5.0 % = 0.114 Free T4 4 5 6 ng/dL 0.8 1.3 2.2 0.08 0.10 0.17 SD from 10 replicates: 0.075 SD based on prior lot: 0.114 SD from package insert: 0.17 SD from peer group: 0.15 Prior Lot Mean: 2.48 SD: 0.12 CV: 5.0 % Initial QC targets Mean: 2.275 SD: Prior Lot Mean: 2.48 SD: 0.12 CV: 5.0 % Initial QC targets Mean: 2.275 SD: 0.114 Initial QC targets Mean: 2.275 SD: 0.17 SD from 10 replicates: 0.075 SD based on prior lot: 0.114 SD from package insert: 0.17 SD from peer group: 0.15 For statistical QC to be effective, values for target mean and SD must be based on data collected from the analytical system being monitored Package insert or other third party values should not be used except as a reference to validate instrument specific values When using 2 SD control ranges, it is important to remember that, generally, single QC results between 2 SD and 3 SD do not indicate a system problem Package insert mean used as QC target Brooks, Z.C., Performance Driven Quality Control, AACC Press, Washington, DC, 2001 Clinical and Laboratory Standards Institute (CLSI). Statistical Quality Control for Quantitative Measurement Procedures: Principles and Definitions; Approved Guideline-Third Edition; C24-A3, Wayne, PA, 2006 Westgard, J.O. et al., Basic QC Practices: Training in Statistical Quality Control for Health Laboratories. Wisconsin: WesTgardÒ QC. 1998. www.westgard.com Select Next to continue.