7.2 Quality Control & Statistical Quality Assurance

Quality Control and Statistical Quality Assurance

Quality Control (QC) is the process of monitoring analytical phases of testing to ensure that results are reliable before releasing them to patient charts. Statistical quality assurance involves analyzing QC data to detect errors, determine their nature (random vs. systematic), and take corrective action. A robust QC program requires statistical analysis, charting, rule evaluation, and proactive troubleshooting.

Basic Statistical Concepts and Calculations

Statistical analysis is the foundation of quality assurance. Technologists must be able to perform and interpret basic statistical calculations.

• Mean ($\bar{x}$): The arithmetic average of a set of values. It is a measure of central tendency. A change in the mean indicates a systematic error.

  • Calculation Example: A laboratory runs a glucose control 5 times. The values are 98, 102, 99, 101, and 100 mg/dL. The mean is $(98+102+99+101+100) / 5 = 500 / 5 = 100$ mg/dL.

• Standard Deviation (SD or $s$): A measure of the dispersion or spread of values around the mean. An increase in SD indicates a decrease in precision, representing a random error. The formula involves finding the square root of the variance.

• Coefficient of Variation (CV%): The standard deviation expressed as a percentage of the mean. It normalizes precision relative to the concentration, allowing comparison of precision across different assays or different control levels. A lower CV% indicates better precision. The target CV% varies by analyte but is typically less than 5% for most chemistry assays.

  • Calculation Example: An assay for cholesterol has a mean of 200 mg/dL and a standard deviation of 8 mg/dL. The CV% is $(8 / 200) \times 100 = 0.04 \times 100 = 4.0%$. If a second cholesterol assay has a mean of 100 mg/dL and an SD of 5 mg/dL, its CV% is $(5 / 100) \times 100 = 5.0%$. The first assay is more precise relative to its mean.

Diagnostic Calculations: Anion Gap and Osmolal Gap

While not strictly QC, these calculations serve as internal physiological checks for laboratory results, often highlighting analytical errors or specific clinical conditions.

• Anion Gap (AG): Represents unmeasured anions in the serum. It is used primarily to evaluate metabolic acidosis. An abnormally low or negative anion gap can indicate a laboratory error (e.g., pseudohyperchloremia due to bromide toxicity or a failing chloride electrode).

  • Formula: $AG = Na^+ - (Cl^- + HCO_3^-)$
  • Calculation Example: Sodium = 140 mEq/L, Chloride = 105 mEq/L, Bicarbonate = 25 mEq/L. $AG = 140 - (105 + 25) = 140 - 130 = 10$ mEq/L. (Reference range is typically 7-16 mEq/L without potassium).

• Osmolal Gap: The difference between the measured serum osmolality (via osmometer) and the calculated osmolality. An elevated gap indicates the presence of unmeasured osmotically active substances (e.g., ethanol, methanol, ethylene glycol).

  • Calculated Osmolality Formula: $2 \times Na^+ + (Glucose / 18) + (BUN / 2.8)$
  • Calculation Example: Sodium = 140 mEq/L, Glucose = 90 mg/dL, BUN = 14 mg/dL. Calculated Osm = $2(140) + (90 / 18) + (14 / 2.8) = 280 + 5 + 5 = 290$ mOsm/kg. If the measured osmolality is 310 mOsm/kg, the osmolal gap is $310 - 290 = 20$ mOsm/kg (Normal gap is usually $< 10$).

Levey-Jennings Charts

A Levey-Jennings chart is a graphical method for displaying QC data over time. The x-axis represents time (days or runs), and the y-axis represents the control value. Horizontal lines are drawn for the mean, $\pm 1 SD$, $\pm 2 SD$, and $\pm 3 SD$. According to Gaussian distribution:

• 68.2% of values fall within $\pm 1 SD$
• 95.5% of values fall within $\pm 2 SD$
• 99.7% of values fall within $\pm 3 SD$

Changes in Levey-Jennings plots are categorized as:

• Trend: A gradual, continuous movement of control values in one direction over six or more consecutive runs. Often caused by systematic errors like gradual deterioration of reagents, aging of the light source, or gradual accumulation of debris in tubing.
• Shift: An abrupt change in the mean that becomes continuous. The values distribute themselves around a new mean. Often caused by sudden systematic errors like changing to a new lot of reagents, major instrument maintenance, or sudden failure of a component.

Westgard Multirule QC Interpretation

Dr. James Westgard developed a set of statistical rules applied to QC data to improve error detection and reduce false rejections. These rules are applied when control values fall outside established limits.

Application of Westgard Rules and Troubleshooting

When a QC value is plotted, the 1-2s rule is evaluated first. If no value exceeds $2 SD$, the run is accepted. If a value exceeds $2 SD$, the other rules (1-3s, 2-2s, R-4s, 4-1s, 10x) are checked. If any of these rejection rules are violated, the run is rejected, patient results are held, and the problem must be investigated and resolved. If only the 1-2s rule is broken and no other rules apply, the run can be accepted, as it is statistically expected that roughly 1 in 20 (5%) of valid results will fall between $2 SD$ and $3 SD$.

Troubleshooting Out-of-Control Runs: When a rejection rule is violated, the technologist must perform systematic troubleshooting rather than blindly repeating the control until it passes.

1. Identify the Error Type: Is it a random error (e.g., 1-3s, R-4s) or a systematic error (e.g., 2-2s, 4-1s, 10x, shift, trend)?
2. Random Error Troubleshooting: Check for bubbles in reagents, improper mixing of the control, power fluctuations, or pipette variability. Repeating the control once is often acceptable for isolated random errors.
3. Systematic Error Troubleshooting: Check reagent expiration dates, reagent lot numbers, calibration status, and recent instrument maintenance. If a shift occurred immediately after changing a reagent lot, the new lot may require recalibration. If a trend is observed, the instrument may need cleaning or a lamp replacement.
4. Corrective Action: Implement the fix (e.g., recalibrate, change reagent, perform maintenance).
5. Verification: Rerun the control. If it passes, the run is accepted, and patient samples tested during the out-of-control period must be reassayed.

Proficiency Testing (PT)

In addition to internal QC, laboratories must participate in Proficiency Testing (external quality assessment). Blind samples are sent from a regulatory agency (like CAP or API) to the laboratory. The lab tests the samples as if they were patient specimens and submits the results. The agency compares the lab's results against peer groups using the same methodology to ensure accuracy across institutions. Failure to maintain acceptable PT performance can result in the loss of CLIA certification for that analyte.