When Good Assays Fail QC
A laboratory validates a new immunoassay, runs it for a week, and immediately starts flagging QC failures. The instrument is functioning properly. The reagents are within date. The control materials are from a reputable manufacturer. So, what went wrong? In many cases, the answer has nothing to do with the assay or the QC material itself. The problem is how the QC ranges were established, or more precisely, how they weren’t.
Setting QC ranges using too few data points, borrowing ranges from a previous reagent lot, or defaulting to manufacturer package inserts without in-house verification are among the most common, and most preventable, sources of QC trouble in laboratories today. The result is a QC program that either flags too many false rejections, wasting time and reagents on runs that were perfectly acceptable, or sets limits so wide that real analytical shifts pass through undetected.
At UTAK, we have spent more than 50 years supporting laboratories through new assay implementations, reagent lot transitions, and QC troubleshooting. In our experience, the single most impactful step in building a defensible QC program is getting the initial range establishment right. Our technical support team fields questions on this topic constantly, which is what prompted this guide. What follows is a structured, practical methodology that laboratories can apply the next time they bring a new assay online.
Why Range Establishment Matters
Quality control ranges define the decision boundaries a laboratory uses to accept or reject an analytical run. Every time a QC result falls within the established limits, the laboratory gains confidence that patient or case results from that run are reliable. Every time a result falls outside, it triggers investigation and potential corrective action. The integrity of the entire QC program depends on how well those boundaries reflect the true, stable performance of the assay on a specific instrument in a specific laboratory.
Ranges set too wide mask real shifts in assay performance, allowing systematic errors to persist undetected across multiple runs. Ranges set too narrow generate false rejections that consume staff time, delay result reporting, and erode confidence in the QC program itself. Neither outcome serves the laboratory or its patients.
Accreditation bodies recognize this. CLIA regulations require laboratories to establish the number, type, and frequency of QC testing, and to define acceptable limits of performance. The College of American Pathologists (CAP) accreditation checklists include specific requirements for QC procedures and documentation [1]. ISO 15189:2022 expects laboratories to implement QC procedures that verify the attainment of intended quality of results [2]. In each case, the expectation is that QC ranges are defensible, documented, and based on data, not borrowed assumptions.
The Range Establishment Process
The Clinical and Laboratory Standards Institute (CLSI) guideline C24-Ed4, Statistical Quality Control for Quantitative Measurement Procedures: Principles and Definitions, provides the most widely referenced framework for QC planning and range establishment in medical laboratories [3]. The process outlined below draws on those principles and translates them into a step-by-step methodology.
Step 1: Gather Preliminary Data
Begin by collecting a minimum of 20 QC data points for each control level, measured under normal operating conditions. This means distributing measurements across multiple days, multiple operators (where applicable), and, if the laboratory uses more than one instrument for the assay, across instruments. The goal is to capture the normal variability the assay will exhibit in routine use, not the artificially tight performance of a single operator running the same instrument on a single morning.
Twenty data points represent the practical floor for generating a statistically stable mean and standard deviation. Fewer than 20 points increase the risk that a single outlier will distort the calculated limits. When possible, 30 or more data points collected over 10 to 20 operating days provide a stronger foundation, particularly for assays with inherently higher variability [3].
Step 2: Calculate Initial Target Values
From the preliminary dataset, calculate the mean and standard deviation (SD) for each control level. The mean becomes the target value, and the SD forms the basis for control limits.
A manufacturer’s certificate of analysis (COA) can serve as a useful starting reference, particularly when bringing up a new assay where in-house data is not yet available. However, the COA reflects performance across multiple instruments and laboratories during the manufacturer’s testing process. It does not reflect the specific performance characteristics of one laboratory’s instrument, reagent lot, calibration, and operating environment. In-house verification is essential.
Worked Example: Suppose a laboratory collects 20 measurements for a mid-range urine drug QC at a nominal concentration of 500 ng/mL. The values (in ng/mL) are: 492, 508, 497, 515, 503, 489, 511, 498, 505, 501, 494, 510, 507, 496, 502, 513, 499, 488, 506, and 509. The calculated mean is 502.2 ng/mL and the SD is 7.8 ng/mL. The ±2SD control limits are 486.6 to 517.8 ng/mL (warning), and the ±3SD limits are 478.8 to 525.6 ng/mL (rejection). These limits are specific to this assay, on this instrument, with this reagent lot. They should not be carried forward to a new lot or a different instrument without re-verification.
Step 3: Set Control Limits
The standard model uses ±2SD as warning limits and ±3SD as rejection limits. This framework has been a cornerstone of laboratory QC since Levey and Jennings adapted Shewhart’s industrial control charts for clinical use in 1950, and it remains the most widely applied approach [4,5].
However, not all assays behave the same. A high-precision chemistry analyzer may produce QC results with a coefficient of variation (CV) under 3%, while an immunoassay for a low-concentration analyte might exhibit CVs of 10% or higher. Applying identical limit-setting strategies to both assays without considering their performance characteristics leads to either unnecessary rejections on the precise assay or undetected errors on the variable one. Before finalizing control limits, evaluate whether the assay’s observed imprecision supports the chosen limits, and document the rationale.
Step 4: Select QC Rules
Westgard multirule QC procedures, first introduced in 1981, remain the most widely adopted framework for interpreting QC data in clinical and forensic laboratories [4]. Rather than relying on a single control rule, multirule procedures combine several rules to improve error detection while minimizing false rejections.
The core rules most laboratories should understand include the following. The 1-2s rule serves as a warning: if a single QC result exceeds ±2SD, it triggers further inspection but does not by itself reject the run. The 1-3s rule is a rejection rule: a single result beyond ±3SD indicates a likely random error. The 2-2s rule flags systematic error when two consecutive results on the same side exceed ±2SD. The R-4s rule detects random error when the range between two control results within a run exceeds 4SD. The 4-1s rule identifies a trend when four consecutive results fall on the same side of the mean beyond ±1SD. And the 10x rule catches a sustained shift when ten consecutive results fall on the same side of the mean [4,6].
The rules applied should match the clinical or forensic risk profile of the assay. For high-risk assays where a false result could directly affect patient care decisions or a medicolegal outcome, apply the full multirule set. For lower-risk screening assays with well-characterized performance, simpler rule sets may be appropriate, but the rationale should be documented either way. CLSI C24-Ed4 emphasizes this risk-based approach, recommending that laboratories select QC rules based on the probability of detecting medically important errors relative to the quality required for the test’s intended clinical use [3,7].
Common Challenges and How to Address Them
Lot-to-Lot Transitions
A new QC lot will not perform identically to the previous lot. Matrix composition, analyte concentration, and stability characteristics all vary between manufacturing lots. Carrying forward the old lot’s ranges to a new lot without verification is one of the most common mistakes in QC management, and it often goes undetected until a pattern of unexplained failures or, worse, an undetected shift emerges.
The recommended approach is parallel testing: run the outgoing and incoming lots simultaneously for a defined overlap period, typically 10 to 20 runs. Collect sufficient data on the new lot to calculate its own mean and SD, then establish new ranges. Some laboratories use the outgoing lot’s ranges as provisional limits during the overlap period, which is reasonable as long as the new lot’s permanent ranges are established from its own data.
Ongoing Range Maintenance
QC ranges are not permanent. As cumulative data grows, the laboratory should review and, where justified, update target values and control limits at defined intervals, typically every six months or when a significant change occurs (new reagent lot, instrument maintenance, calibration update). Cumulative data often supports tightening limits as the dataset stabilizes, which improves the sensitivity of the QC program over time.
Be alert to “range creep,” where small, unnoticed shifts in the mean accumulate over months until the QC program no longer catches real problems. Periodic review, comparing current means and SDs against the original baseline, is the best defense against this gradual erosion of QC sensitivity.
Troubleshooting Persistent Failures
When a new assay consistently fails QC after range establishment, the instinct is often to widen the ranges. Resist that instinct. Persistent failures are a signal, not a nuisance, and widening limits to accommodate them defeats the purpose of the QC program.
Instead, investigate upstream. Common root causes include reagent instability, calibration drift, sample handling variability, and environmental factors such as temperature fluctuations. Systematic troubleshooting, working through potential causes methodically before adjusting ranges, protects the integrity of the QC program and often uncovers issues that would eventually affect patient results.
QC Range Establishment Checklist
- Minimum 20 data points collected across multiple days, operators, and instruments
- Mean and SD calculated from in-house data, not solely from manufacturer inserts
- Control limits set at ±2SD (warning) and ±3SD (rejection) as baseline, adjusted for assay characteristics
- Westgard rules selected based on assay risk and clinical or forensic impact
- Lot-to-lot transition procedure documented with parallel testing protocol
- Periodic range review schedule established (e.g., every six months)
- Troubleshooting protocol in place before adjusting ranges to accommodate failures
Frequently Asked Questions
How many data points do I need before setting QC ranges?
A minimum of 20 data points per control level is the widely accepted starting point, consistent with CLSI C24-Ed4 guidance [3]. Fewer than 20 increases the risk that outliers will distort the mean and SD. For assays with higher inherent variability, 30 or more data points collected over 10 to 20 operating days provide a more stable foundation. The key is that data should be collected under routine conditions, across multiple days and operators, not concentrated in a single testing session.
Can I use the manufacturer’s target values instead of establishing my own?
Manufacturer target values, typically provided on the certificate of analysis, are a useful starting reference, especially during the first few days of a new assay. However, they reflect performance across multiple instruments and laboratories during the manufacturer’s testing process and do not account for the specific characteristics of your instrument, reagent lot, calibration, or operating environment. In-house verification is essential, and most accreditation standards expect laboratories to establish or verify target values using their own data.
How often should QC ranges be reviewed and updated?
Most laboratories review ranges at defined intervals, typically every six months, or when a significant change occurs, such as a new reagent lot, instrument maintenance, or calibration update. Cumulative data often supports tightening limits as the dataset stabilizes. The important thing is to have a documented schedule and to follow it consistently, as this is what inspectors expect to see during accreditation assessments.
What is the difference between Westgard rules and simple ±2SD limits?
A simple ±2SD limit (the 1-2s rule) uses a single criterion to evaluate each QC result, which has the advantage of simplicity but produces a high false rejection rate, approximately 9% with two control measurements per run [4]. Westgard multirule procedures combine several rules, each sensitive to different types of error (random, systematic, trending), which significantly improves error detection while reducing false rejections. The trade-off is slightly more complexity in interpretation, though most modern laboratory information systems apply Westgard rules automatically.
How do I handle QC range establishment when I have limited sample volume?
When QC materials are expensive or available in limited quantities, laboratories may need to balance statistical rigor with practical constraints. One strategy is to use the manufacturer’s COA values as provisional limits while accumulating in-house data, then transition to in-house ranges once 20 or more data points are available. Another approach is to extend the data collection period rather than increasing the number of runs per day, which still captures day-to-day variability without consuming excess material.
Building a QC Program That Holds Up
Range establishment is not a one-time task. It is the foundation of an ongoing QC program that evolves with each new reagent lot, each instrument change, and each assay added to the laboratory’s menu. The laboratories that do this well share a few characteristics: they collect enough data before committing to ranges, they document their methodology, they review ranges on a defined schedule, and they resist the temptation to widen limits when the real problem lies elsewhere.
Our technical team has worked through these challenges across hundreds of laboratory implementations over 50 years. Whether you are bringing up a new assay, navigating a difficult lot transition, or re-evaluating a QC program that has been generating more questions than answers, sometimes a conversation with people who have seen these problems in many different contexts can shortcut weeks of troubleshooting.
Reach out at welovecontrol@utak.com or call 888.882.5522.
References
[1] College of American Pathologists. CAP Accreditation Checklists – 2025 Edition. Available at: https://www.cap.org/laboratory-improvement/accreditation/accreditation-checklists
[2] International Organization for Standardization. ISO 15189:2022 – Medical laboratories: Requirements for quality and competence. Geneva: ISO; 2022.
[3] Clinical and Laboratory Standards Institute. C24-Ed4: Statistical Quality Control for Quantitative Measurement Procedures: Principles and Definitions. 4th ed. Wayne, PA: CLSI; 2016. See also: Parvin CA. What’s New in Laboratory Statistical Quality Control Guidance? The Journal of Applied Laboratory Medicine. 2017;1(5):581–584. https://doi.org/10.1373/jalm.2016.022269
[4] Westgard JO, Barry PL, Hunt MR, Groth T. A multi-rule Shewhart chart for quality control in clinical chemistry. Clinical Chemistry. 1981;27(3):493–501. https://doi.org/10.1093/clinchem/27.3.493 For current guidance on Westgard rules and QC planning tools, see: https://www.westgard.com/westgard-rules.html
[5] Levey S, Jennings ER. The use of control charts in the clinical laboratory. American Journal of Clinical Pathology. 1950;20(11):1059–1066.
[6] Westgard JO. Basic QC Practices. 4th ed. Madison, WI: Westgard QC; 2016.
[7] Westgard JO. A Total Quality-Control Plan with Right-Sized Statistical Quality-Control. Clinics in Laboratory Medicine. 2017;37(1):137–150. https://doi.org/10.1016/j.cll.2016.09.011
