CC REVIEW

• Quality


  The degree of excellence of something
• Statistics


  The method used by laboratorians to monitor data values and determine if errors are present as these will be affecting the quality of the test

  
• Types of statistics used in clinical laboratory


  • Descriptive statistics
  • Inferential statistics
• Descriptive statistics


  • Data values are plotted in graphical representations to visualize patterns
  • Descriptive statistics include analysis of patterns described by their center, spread and shape to evaluate test quality

  
• Measures of center


  • Mean
  • Median
  • Mode
• Mean


  The average of the data values
• Median


  The middle point of the data
• Mode


  The most frequently occurring value
• Measures of spread


  • Range
  • Standard deviation
  • Coefficient of variation
• Range


  The difference between the highest and lowest data value
• Standard deviation


  The distribution of all data around the mean
• Coefficient of variation


  Expresses standard deviation as a percentage of the mean
• Shape of data distribution


  • Normal distribution (Gaussian curve)
  • Skewed distribution
  • Bimodal distribution
• Inferential statistics


  Involves drawing conclusions about the means or standard deviations of two sets or groups of data
• Calculating standard deviation


  1. Determine variance first

  2. Take square root of variance
• 1. T-test

  Measures significant difference between means of two groups
• 1. F-test

  Measures significant difference between standard deviations of two groups
• COM (Comparison of Methods)


  Involves measuring a sample using the test method and a reference method to determine the relationship between them

  
• Linear regression


  Describes the relationship between two variables mathematically with the formula y = mx + b
• Correlation coefficient


  Indicates the strength of the relationship between two methods, ranging from -1 to +1
• Method evaluation


  1. Method selection

  2. Validation

  3. Implementation

  4. Routine testing
• Imprecision


  Dispersion of repeated measurements about the mean due to random error

  
• Inaccuracy


  Difference between a measured value and its true value due to systematic error
• Allowable error


  The maximum error allowable which will not cause a negative effect on the performance of the test
• Types of error


  • Random error
  • Systematic error (proportional and constant)
• Quality control


  Includes detection of analytic error that could affect the quality of the test and lead to erroneous patient results
• Quality control terms


  • Accuracy
  • Precision (reproducibility and repeatability)
  • Controls
  • Control limits
  • Standards
  • Reliability
• Performing quality control


  1. Assay controls the same way as patient samples

  2. Compare determined control value to expected value

  3. Values outside control limits require further analysis

  
• Types of quality control


  • Intralab QC (internal)
  • Interlab QC (external)
• Histogram


  A graphical representation used to locate the source of error

  
• Histogram analysis techniques


  • CUSUM (Cumulative Sum)
  • Youden plot (Twin plot)
  • Levey-Jennings chart
• Histogram


  Sheet of rectangular coordinate of graphing paper where data for analysis are plotted to locate the source of error

  
• CUSUM (Cumulative Sum)


  1. Calculating the actual difference between the individual values and the expected mean value

  2. Determining the sum of this difference to check cumulative effect of all control observations

  
• If CUSUM is used, Levey-jennings is also done
• Youden plot (Twin plot)


  Compares the results acquired to a high or low control serum
• Levey-Jennings


  Most commonly used histogram, used to display and detect trends, shifts and dispersion
• TREND


  • Displayed when the control value direction moves progressively in one direction (progressive increase or decrease), indicating deterioration of reagent or control
• SHIFT


  • Displayed when abrupt changes occur, indicating sudden malfunction of the instrument
• DISPERSION


  • Occurs when random error or lack of precision increases, referring to the increased frequency of both high and low outliers
• Westgard Multirules


  • 1:2s - One value exceeds the ± 2SD
  • 1:3s - One value exceeds the ± 3SD
  • R:4s - One value exceeds the +2SD and another exceeds - 2SD (the two values are distant by 4 SD)
  • 2:2s - Two consecutive values exceed either the +2SD or -2SD
  • 4:1s - Four consecutive values exceed either the +1SD or -1SD
  • 10x - Ten consecutive values fall on one side or the other of the mean