Quality Assessment And Troubleshooting

Statistical Quality Control And Westgard Rules

Quality control (QC) confirms that a method performs within established limits before patient results are released. Each new QC lot is assayed ~20 times to establish a mean and standard deviation (SD), then plotted daily on a Levey-Jennings chart with lines at the mean and ±1SD, ±2SD, and ±3SD. Roughly 95.5% of in-control points fall within ±2SD and 99.7% within ±3SD, so points outside ±2SD are statistically unusual.

The Westgard multirules convert those limits into reject/accept decisions. Memorize which rule flags random error versus systematic error, because the troubleshooting answer depends on the error type.

A classic trap: 1-2s alone is a warning, not a rejection. If a stem shows a single Level 1 control at +2.4SD and asks the next step, the best answer is to evaluate the other rules, not to discard the run.

Shifts, Trends, And Root-Cause Troubleshooting

Patterns over time point to the cause:

• A shift is an abrupt move of consecutive points to one side of the mean (often a 10x or 2-2s violation). Likely causes: a new reagent lot, fresh calibration, new control lot, or a new technologist/instrument component.
• A trend is a gradual, progressive drift across many points. Likely causes: deteriorating reagent, an aging ion-selective electrode, a failing photometer lamp, or evaporation of control material.
• Random scatter with a sudden out-of-range single point (1-3s, R-4s) suggests bubbles, a clot, a pipetting error, or electrical noise.

Worked example: a glucose analyzer's QC was centered for two weeks, then both control levels jumped +2.5SD the morning after a reagent change and stayed there for six days. This abrupt, sustained offset is a shift caused by the new reagent lot, not a calibration drift—recalibrate against the new lot. By contrast, controls creeping upward 0.3SD per day over two weeks describe a trend consistent with reagent deterioration.

Quality assessment is broader than analytical QC. It covers the total testing process in three phases:

1. Preanalytical — patient prep, correct tube/additive, no hemolysis, correct order of draw, timely transport. Most laboratory errors occur here.
2. Analytical — QC, calibration, linearity, calibration verification.
3. Postanalytical — result verification, delta checks, critical (panic) value notification, and correct LIS reporting.

A delta check flags a result that differs implausibly from the patient's prior value (e.g., a hematocrit that jumps from 28% to 45% in hours), prompting a check for a mislabeled or switched specimen before reporting.

Choosing A QC Strategy And Reacting To Failures

Not every analyte needs the full six-rule set. Westgard recommends matching QC stringency to method performance: a high-sigma method (large allowable total error relative to its imprecision) can run a single, lenient rule such as 1-3s with minimal false rejections, while a low-sigma method needs the full multirule scheme with more control levels to catch error reliably. The exam may frame this as balancing error detection against the false-rejection rate — using only 1-2s rejects ~5% of good runs, wasting reagent and time, which is why 1-2s is demoted to a warning.

When a run is rejected, follow an ordered troubleshooting path rather than guessing:

1. Repeat the control once — a true random error (bubble, clot, mispipette) often does not recur.
2. If it recurs, open fresh control to rule out deteriorated or improperly stored QC material.
3. Check reagent (expiration, lot, on-board stability) and recalibrate if a shift coincided with a new lot.
4. Inspect the instrument — lamp energy, probe/aspiration, temperature, fluidics for carryover.
5. Do not report patient results generated during an out-of-control run; once resolved, repeat patient samples run since the last acceptable QC.

Patient-Based Indicators And Documentation

Beyond control material, laboratories monitor quality with patient data:

• Delta checks compare a patient's current and previous result; a large unexplained change suggests a specimen mix-up or analytical error.
• Moving averages (e.g., Bull's algorithm) track the running mean of red-cell indices (MCV, MCH, MCHC) because indices stay stable across a population; a drift signals calibration loss.
• Critical value policies require immediate notification with read-back and logging of the caregiver, time, and result.
• Absurd-value and limit checks catch transcription or interface errors (e.g., a potassium of 25 mmol/L).

Worked example: a CBC shows MCHC of 39 g/dL (well above the usual 32–36 ceiling). Rather than reporting it, recognize that an impossibly high MCHC is a classic flag for a preanalytical interference — lipemia, cold agglutinins, hemolysis, or a clotted/lipemic specimen — warranting smear review and specimen inspection. Documenting every out-of-control event, its root cause, and the corrective action closes the quality loop and is itself an inspectable requirement under accreditation.