Quality Assurance in Clinical Chemistry

Quality Assurance - Clinical Chemistry

• Quality assurance (QA) is the process of monitoring any activity linked to a lab result.
• QA includes errors that can occur at three stages:
  • Pre-analytical: errors before the sample reaches the lab or is analyzed. Examples include collecting a HbA1c sample in a SST tube.
  • Analytical: errors during the analysis of a sample. For instance, analyzer malfunction during testing.
  • Post-analytical: errors after the analysis of a sample.

Descriptive Statistics - Basic Terms

• Qualitative data: Subjective observations that can't be quantified. Examples include turbidity, color, and odor.
• Quantitative data: Measurable characteristics. Examples include glucose (mmol/L) and creatinine (µmol/L).
• Data: A collection of related observations used to draw conclusions about a population. For instance, reference intervals for healthy people.

Accuracy and Precision

• Accuracy: Closeness of a result to the true value.
• Precision: Reproducibility of a result.
• In labs, accurate and precise results are desired.

Key Terminology

• Population: All possible observations.
• Sample: Part of a population.
• Random sample: Sampling where no part of the population has preference.
• Parameter: Calculation using population information.

Calculated Statistical Values

• Statistics: Values calculated from random samplings of a population.
• Classified into two groups:
  • Measures of Central Tendency: Distribution of values around a central value (mean, median, mode).
  • Measures of Dispersion: How spread out the values are (range, variance, standard deviation).

Measures of Central Tendency - Mean

• The mathematical average of a set of values.
• Parametric mean (μ): Data based on all of a population.
• Statistical mean (x̄): Data based on a sample of a population.

Mean - Example

• Provided example data set to calculate the mean.

Measures of Central Tendency - Median

• Median is the middle number when values are arranged in sequential order.
• If the number of values is even, the average of the two middle numbers is the median.
• If the number of values is odd, the middle number is the median.

Median - Example

• Provided example data set to calculate the median.

Measures of Central Tendency - Mode

• Mode is the most frequently occurring number in a group of numbers.

Mode - Example

• Provided example data set to calculate the mode.

Measures of Dispersion - Range

• Range is the difference between the highest and lowest value.

Range - Example

• Example data set provided to calculate the range.

Measures of Dispersion - Variance

• Variance (s²) measures the precision of a group of numbers.
• High variance indicates wide ranges of values.
• Low variance indicates closely grouped values.
• The smaller the variance, the more precise the numbers.

Variance - Formula

• Formula for calculating variance is provided.

Variance - Example

• Example data set provided to demonstrate variance calculation steps.

Measures of Dispersion - Standard Deviation

• Standard Deviation (SD) is the most frequent measure of precision.
• Symbol is s(or SD).
• Standard deviation is the square root of the variance.
• Formula for calculating standard deviation is included.

Standard Deviation - Example

• Provided example problem which resulted in the standard deviation result.

Coefficient of Variation

• Used to compare precision of two or more data groups.
• Expressed as a percentage.
• Lower the CV, the greater the precision of a data set.

Coefficient of Variation - Example

• Example provided to show calculation method.

Probabilities Associated with Standard Deviation

• Statistically, analyzed 30 times, results will fall within a certain percentage range for 1 SD, 2 SD, 3 SD from the mean.
• 68.2% of the time results fall within ±1 SD.
• 95.5% of the time results fall within ±2 SD.
• 99.7% of the time results fall within ±3 SD.

Normal Distribution Curve- Examples

• Depiction of the bell curve including percentages and regions identified for 1, 2 and 3 SD.

Quality Control Analysis

• Acceptable quality control results fall within expected limits.
• If the result is outside accepted limits, no further patient results will be reported until the problem is corrected/resolved.

Types of Errors - Random and Systematic Errors

• Random Errors: Errors due to chance, extreme variations on consecutive days in lab results. Instrument, operator, reagent/environmental issues.
• Systematic Errors: Errors affecting all samples in the same direction. Reagent issues, calibration problems, etc.

Common QC Problems - Shifts

• QC results consistently distributed on one side of the mean for 6 or more days.
• Cause must be identified, samples re-analyzed for correction of problem.
• Examples: incorrectly prepared reagents, wrong timing/temperature, wrong wavelength.

Common QC Problems - Trends

• QC results consistently decrease or increase over 6 or more consecutive days.
• Cause must be identified, samples re-analyzed for correction of problem.
• Examples: reagent degradation, temperature issues, change to light source.

Levey-Jennings Charts

• Daily QC plotted on Levey-Jennings charts.
• Charts visualize shifts and trends.
• At least 2 levels of QC material should be analyzed daily for each test.

Levey-Jennings Charts - Example

• Example provided of a graph visualization and acceptance/rejection criteria of a test on a Levey-Jennings Chart.

Westgard Multirules

• A set of quality control rules used to interpret quality control results for acceptability in a lab.
• If any rule (except the warning rule) is violated the lab results are not acceptable.

Westgard 12s Rule

• One point exceeds +2SD.
• Acceptable random error in results is often indicated by this rule. (ex: air bubble in sample).

Westgard 13s Rule

• One point exceeds +3SD.
• Indicates unacceptable random error or beginning of systematic issues in results.

Westgard 22s Rule

• Two points above the mean and exceed +2SD.
• Indicates a systematic error within the test protocol

Westgard 4s Rule

• Four consecutive results beyond the +1SD line.
• Indicates potential systematic error in the test.

Westgard 10x Rule

• Ten consecutive results on the same side of the mean.
• Indicates a systematic error.

Establishing Statistical Quality Control Limits

• Determining acceptance range using multiple data points for a certain testing period to determine a mean and Standard Deviation.
• Most labs accept the mean ± 2SD as the quality control range or acceptable values. This will equate to a 95% confidence that results are within acceptable range. QC results outside this range trigger a review and resolution for the test.
• Methods of establishing acceptable ranges using 1, 2, and 3 SD.
• Example calculation providing results.