Skip to main content

Laboratory Operations

Quality Control

Let’s zoom in on the absolute heart of analytical quality, the process that gives us the confidence to report a patient result: Quality Control, or QC. If the entire quality system is the car, then QC is the engine’s diagnostic computer, constantly checking to make sure every component is running within its specified limits. It’s our proactive, real-time “dress rehearsal” before the main performance of testing patient samples. You would never want to board a plane if the pilot hadn’t first run through a comprehensive pre-flight checklist. In the same way, we never report patient results without first ensuring our analytical system has passed its own rigorous pre-flight check—and that is QC

The fundamental purpose of QC is to monitor the precision and accuracy of our methods. We are answering one critical question over and over again: “Is my testing system performing today exactly as it was designed to, and exactly as it did yesterday?” A “yes” answer means we can trust our patient results. A “no” answer means we stop everything and become detectives

Tools of the Trade: Control Materials

To perform QC, we use control materials. These are patient-like substances (e.g., a serum or plasma base) that have been carefully prepared by a manufacturer to contain a specific, known concentration of the analyte we want to measure. The key features of control materials are:

  • Known Values: The manufacturer provides an “assay sheet” that lists the expected mean value and acceptable range for our specific instrument and method. This is our target
  • Multiple Levels: We don’t just check one value. We typically use at least two, and often three, levels of controls. For example, a “low,” a “normal,” and a “high” control. This ensures our method is accurate across the entire reportable range. A test that is accurate for normal glucose levels might not be accurate for critically high levels, and using multiple levels of QC will detect this
  • Stability: These materials are manufactured to be stable for long periods, allowing us to use the same lot number for weeks or months. This long-term use is crucial for detecting slow, subtle changes in our testing systems

Process and the Chart: Levey-Jennings

The QC process itself is straightforward. At scheduled times—such as the beginning of a day, after a reagent change, or following instrument maintenance—we run our control materials just like they were patient samples. We then take the value our instrument obtains and plot it on a special graph called a Levey-Jennings (LJ) chart

Imagine a simple graph. The horizontal X-axis represents the date or run number. The vertical Y-axis represents the concentration of the analyte. We draw a solid line across the middle representing the control’s target mean, and then we draw dashed lines above and below it representing the acceptable limits, which are defined by standard deviations (SD). We draw lines at +/- 1 SD, +/- 2 SD, and +/- 3 SD from the mean. Each time we run the control, we plot the result as a single dot on this chart

When a testing system is in control, the plotted points will be distributed randomly around the mean, with most points falling within +/- 1 SD and very few falling between +/- 2 SD and +/- 3 SD. This is our visual confirmation that everything is working as expected

Rules of Engagement: Westgard Rules and Troubleshooting

What happens when the dots on our LJ chart don’t look random? What if they start to drift upwards or suddenly jump to a new level? This is where we put on our detective hats and use a formal set of statistical rules called Westgard Rules to interpret the patterns and identify the type of error we’re seeing. There are two main types of error:

  • Random Error: This is an error that happens unpredictably and affects only one or two data points. It’s a “one-off” fluke. Common causes include a small bubble in a reagent line, a bit of fibrin in the sample probe, or a power fluctuation. The Westgard rule 1_3s (one control point falls outside of +/- 3 SD) is a classic random error flag that requires immediate rejection of the run
  • Systematic Error: This is a much more insidious error. It is a persistent bias in the system that affects all results in the same direction. It shows up on an LJ chart not as a single outlier, but as a shift (a sudden jump in the data to a new level) or a trend (a slow, steady drift of the data in one direction). Common causes include a deteriorating reagent, a failing light source in the instrument, or a gradual loss of calibration

Westgard rules are the triggers that tell us when a pattern is no longer due to chance. For example, the rule 2_2s (two consecutive control points fall on the same side of the mean and are outside of +/- 2 SD) is a clear warning of a systematic problem. The rule 10_x (ten consecutive control points fall on the same side of the mean) indicates a subtle but definite shift has occurred

When QC Fails: The Troubleshooting Cascade

The moment a Westgard rule is violated, a hard stop is put in place. No patient results can be released. We then follow a logical troubleshooting sequence:

  1. Re-run the Control The first step is always the simplest. Was it just a fluke? Open a fresh vial of control material and run it again. This solves a surprising number of random errors
  2. Inspect the System If the QC fails again, look for the obvious. Are the reagents low? Are they expired? Do you see bubbles in the lines? Is the instrument displaying any error flags?
  3. Recalibrate If a systematic error (shift or trend) is indicated, the instrument’s calibration may have drifted. Performing a fresh calibration is often the next step to re-establish the correct baseline
  4. Perform Maintenance If calibration doesn’t fix the issue, a physical component may be failing. This may involve cleaning probes, changing tubing, or replacing a lamp
  5. Contact Technical Support If all else fails, it’s time to bring in the experts from the instrument’s manufacturer
  6. Document Everything Every step taken, from the initial failure to the final resolution, must be meticulously documented in a quality control log. If you didn’t write it down, it never happened

Only after the problem has been identified and corrected, and a new set of QC runs passes successfully, can we resume testing and release patient results. This entire process is the bedrock upon which the reliability and trustworthiness of the clinical laboratory is built

Key Terms

  • Quality Control (QC): The process of analyzing patient-like materials with known values (controls) to monitor the accuracy and precision of an analytical method and ensure the system is working correctly
  • Control Material: A substance, often with a similar matrix to patient specimens, containing a known concentration of an analyte used to verify the performance of a test system
  • Levey-Jennings (LJ) Chart: A graph used to plot QC values over time, with the mean and standard deviation limits, to visually assess for trends, shifts, and errors
  • Standard Deviation (SD): A measure of the variation or dispersion of a set of data points. In QC, it defines the acceptable limits around the target mean
  • Random Error: An unpredictable, one-off error that occurs without a real pattern, often due to chance occurrences like a bubble in a reagent or a small voltage spike
  • Systematic Error: A persistent error or bias in an analytical system that affects all results in the same way, appearing as a “shift” or a “trend” on a Levey-Jennings chart
  • Westgard Rules: A set of statistical rules applied to QC data to help detect both random and systematic errors and determine if an analytical run is in-control or out-of-control