Questions: Quality Control and Quality Assurance in Analytical Labs

Question 1

A laboratory analyzes a control sample alongside every batch. The last eight consecutive control results have all fallen on the same side of the mean — but none have exceeded the ±2 standard deviation warning limit. According to Westgard rules, what should the analyst do?

• A. Accept the batch — all results are within warning limits so the system is in statistical control
• B. Reject the batch immediately — seven or more consecutive results on one side signals systematic drift requiring investigation
• C. Re-analyze only the control sample to confirm the pattern is real
• D. Document the pattern and continue until a result exceeds the action limit

Answer: (B)

Westgard rules include pattern-based triggers, not just threshold-based ones. A run of seven or more consecutive results on one side of the mean — even within ±2σ — indicates systematic drift (a shift in the method's baseline) rather than random variation. The probability of this occurring by chance is less than 1%, so it signals a real problem such as reagent degradation or instrument drift. Accepting the batch because 'no result exceeded the warning limit' misses this. This is the classic mistake of checking only individual rules while ignoring run patterns.

Question 2

A laboratory processes a real groundwater sample spiked with a known concentration of lead alongside unspiked samples. Recovery is 78%. What is the most accurate interpretation?

• A. The method has a contamination problem — the spike is contributing excess signal
• B. There is a systematic bias or matrix effect: the sample matrix is suppressing recovery relative to the calibration standard
• C. The instrument is out of calibration and must be recalibrated before results can be reported
• D. The spike result is within acceptable limits and indicates the blank is clean

Answer: (B)

Spiked sample recovery tests for matrix effects and systematic bias: does the analyte behave the same in a real sample matrix as in a calibration standard made in clean solvent? Recovery of 78% suggests matrix suppression — the sample matrix interferes with detection or extraction of the analyte, causing it to read low. This is distinct from contamination (which would inflate the result) or calibration error (which would affect all samples equally). Matrix-matched standards or standard addition can correct for this. Method blanks — not spikes — detect contamination.

Question 3

A single control result that falls within ±2 standard deviations of the established mean proves the batch is in statistical control and most results are acceptable.

• T. True
• F. False

Answer: False

False. A single within-limit result is necessary but not sufficient. Statistical control also requires the absence of non-random patterns: systematic trends, runs of consecutive results on one side of the mean, or gradual drift can all indicate a system drifting out of control even when no individual result has crossed a threshold. Westgard rules codify these pattern-based triggers specifically because single-result checks miss systematic problems that develop over successive batches.

Question 4

Certified reference materials (CRMs) serve a different QC function than spiked samples: CRMs provide an independent accuracy check for the method as a whole, while spikes primarily test for matrix effects within a specific sample.

• T. True
• F. False

Answer: True

True. CRMs have compositions established by authoritative bodies using multiple independent methods — they represent the gold standard for accuracy verification. If your method gives the right answer on a CRM, your entire analytical chain (reagents, calibration, instrument, procedure) is working. Spikes test whether a known amount of analyte added to a specific real sample is recovered correctly, which targets matrix effects and extraction efficiency but does not verify absolute accuracy in the same way. Both are needed: CRMs catch method-level errors, spikes catch sample-specific matrix problems.

Question 5

What is the fundamental distinction between quality assurance (QA) and quality control (QC), and why does a laboratory need both rather than one or the other?

Model answer: QA is the system of policies, procedures, documentation, and management practices that creates an environment where reliable data can consistently be produced — SOPs, training records, equipment maintenance logs, calibration documentation, and proficiency testing. QC is the set of specific technical checks performed during each analytical batch — control charts, blanks, duplicates, spikes, and reference materials — that verify the system is actually working correctly at the moment of measurement. You need both: QA ensures the right procedures exist and analysts are competent; QC catches failures in real time when something goes wrong despite the QA framework.

The distinction matters because each addresses a different failure mode. QA prevents failure by building a capable system; QC detects failure when it occurs. A lab with strong QC but no QA might catch errors but will struggle to prevent them systematically. A lab with strong QA but no QC has good policies but no feedback mechanism to know when those policies are failing in practice. ISO/IEC 17025 requires both, and traceability — the ability to follow a reported value back through the entire chain of QC checks and QA documentation — depends on them working together.