Laboratory Quality Control: Why QC Passes but Results Are Wrong

Every laboratory has seen it: the controls are in range, the run is accepted, and a clinician calls to say a result makes no sense. Sometimes the clinician is right. Laboratory quality control is essential, but it was never designed to guarantee that every patient result is correct. Understanding its limits is the first step to closing the gaps.

Why can QC pass when patient results are wrong?

QC passes when the control material behaves as expected at the moment it is tested, which does not guarantee that every patient sample in the run behaved the same way. QC is a sampling exercise with a probability of error detection that is always below 100%.

Three limits explain most surprises:

• Timing. QC is run at set points. An error that starts after the control was tested, or a short intermittent fault, may affect patient results before the next QC event.
• Statistics. Every QC rule balances the probability of detecting a real error against the probability of false rejection. A loose rule such as a single 1-3s limit with two controls misses a meaningful share of small systematic shifts.
• Material. Control material is not patient blood. It may react differently to a new reagent lot or an interfering substance, so QC can stay stable while patient results move.

What do Westgard rules detect?

Westgard multirules combine several control limits so that a laboratory can detect both random and systematic error while keeping false rejections low. Each rule is sensitive to a different type of error.

The rules only work if the mean and SD are your own, established from your analyser and control lot, not copied from the manufacturer’s insert. A wide manufacturer range makes almost everything pass. The Westgard website explains the rules and their power functions in detail.

Which errors does QC commonly miss?

QC commonly misses errors that affect patient samples but not control material, errors outside the concentrations the controls cover, and errors that happen before or after the analysis.

Commutability and matrix effects

Many control materials are processed, stabilised or spiked. They are not always commutable, meaning they do not react to changes in the method the same way patient samples do. A reagent change can shift patient results while QC barely moves, or the reverse.

Reagent and calibrator lot changes

A new reagent or calibrator lot can introduce bias. If the laboratory checks only that QC is in range after the change, a shift in patient results can go unnoticed, especially near clinical decision limits.

Errors at the extremes of the measuring interval

Two control levels may sit in the normal and moderately abnormal range. Problems with linearity, dilution or very low concentrations near the limit of quantitation can affect patient results at the extremes without affecting the controls.

Carry-over

A very high sample followed by a low one can contaminate the second result. QC run in isolation will not show this.

Pre-analytical errors

Most laboratory errors arise in the pre- and post-analytical phases: wrong patient identification, wrong tube, haemolysis, delayed transport, incorrect storage or clotting. QC monitors only the analytical phase, so none of these appear on a Levey-Jennings chart.

Interference

Haemolysis, lipaemia, icterus, biotin, heterophile antibodies and some drugs interfere with particular methods in individual patients. QC material does not carry these substances.

Population-specific bias

Reference intervals and method performance verified in one population may not hold in another, and imprecision may differ at the concentrations most common in your patients. A method can pass QC while systematically misclassifying a local patient group.

How can a laboratory catch errors that QC misses?

A laboratory catches errors that QC misses by adding checks based on real patient data and by verifying changes before they reach patients.

1. Patient-based quality control. Monitor moving averages (or moving medians) of patient results for stable analytes such as sodium, calcium or MCV. A gradual shift in the average can reveal a bias long before QC fails. Exclude obvious outliers and set limits from your own historical data.
2. Reagent lot verification with patient samples. Before releasing a new reagent or calibrator lot, test a set of patient samples on the old and new lots, covering low, normal and high concentrations. Compare the differences with predefined acceptance limits, then switch.
3. Delta checks. Compare a patient’s result with their previous result. A change larger than biological variation explains is flagged for review, which catches sample mix-ups and contamination.
4. Review flags and repeat rates. Track how often results are flagged, diluted or repeated. A sudden rise is an early signal of a problem.
5. External quality assessment. Proficiency testing compares your results with peer laboratories on unknown samples and exposes bias your internal QC cannot see. Investigate every unacceptable result as a nonconformity.
6. Pre-analytical controls. Train phlebotomists and ward staff, use clear rejection criteria, and monitor indicators such as haemolysis and mislabelling rates.
7. Occurrence reporting. Make it easy for staff and clinicians to report suspicious results and near misses, then trend them. Repeated complaints about one analyte are data.

How do sigma metrics help you choose QC rules?

Sigma metrics tell you how much room your method has before errors become clinically important, so you can choose fewer rules for high-performing methods and stricter rules for weaker ones.

Calculate sigma as (allowable total error minus absolute bias) divided by CV, all in percent at a clinical decision level. Take bias from EQA or comparison studies and CV from your own QC data.

• Six sigma or higher: a simple 1-3s rule with two controls usually gives good error detection with few false rejections.
• Four to six sigma: a multirule such as 1-3s/2-2s/R-4s is generally appropriate.
• Below four sigma: use the full multirule, more controls and more frequent QC, and treat improving the method as a priority.

Sigma is only as good as the allowable error you choose and the data you feed it, so document your sources and review them when methods or lots change.

What is a risk-based QC plan (IQCP)?

A risk-based QC plan, called an Individualised Quality Control Plan (IQCP) in the United States, sets QC frequency and other controls for each test based on a documented risk assessment rather than a single default.

The approach, described in CLSI guidance on risk-based QC, works through the testing process step by step:

1. Map the full process, from test request to report.
2. Identify what can go wrong at each step, including operator, reagent, specimen, environment and instrument failures.
3. Estimate the severity and likelihood of each failure, often using a tool such as FMEA.
4. Decide which controls reduce each risk: built-in instrument checks, external QC, patient-based QC, training or competency assessment.
5. Write the QC plan, implement it and monitor its effectiveness through occurrences, complaints and EQA, then review it.

Risk-based QC plans are especially useful for point-of-care testing and for analysers with extensive internal checks, where daily liquid QC alone may not be the most effective control.

What does ISO 15189:2022 expect from laboratory quality control?

ISO 15189:2022 expects the laboratory to have a procedure for monitoring the validity of examination results, to use internal QC and external quality assessment, and to base its controls on risk.

Clause 7.3.7 (ensuring the validity of examination results) requires internal QC designed to detect clinically significant error, participation in EQA or suitable alternatives, and comparability of results across methods and equipment. The 2022 edition also places risk management at the centre of the standard (clauses 5.6 and 8.5), so assessors increasingly ask why you chose your QC rules and frequency, not just whether QC was run. A laboratory that can show sigma calculations, lot verification records, patient-based QC limits and a reviewed QC plan is in a much stronger position than one that can only show Levey-Jennings charts. For the wider system, see our guide to building a laboratory quality management system, or the ISO website for the standard itself.

Which training helps teams design better QC?

Teams design better QC when they understand risk assessment, error investigation and the quality system around QC, not just how to plot a chart.

• Risk Management in the Clinical Laboratory (3 days) covers FMEA, root cause analysis and CAPA, with day 3 dedicated to building an IQCP.
• Laboratory Occurrence Management (3 days) teaches you to capture, investigate and trend errors, near misses and nonconformities, including QC failures and suspicious results.
• Establishing and Maintaining a Laboratory QMS (5 days) places process control and QC within the 12 quality system essentials.

Courses run every quarter in person in Nairobi and live online, and each can be delivered in-house for your team. Participants receive a certificate of completion.