Goal of Quality Control

1
Goal of Quality Control

• The goal of Quality Control (QC) is to catch all
  significant errors without repeating tests
  unnecessarily
• ?A significant error is defined as a wrong answer
  that causes a change in the diagnosis or
  treatment of a patient or a failure in
  proficiency testing

Lab Pointer
Contributed by Melissa Bethel,
MT(ASCP) Esoterix Coagulation Aurora, Colorado
Balance optimal quality with cost effectiveness
2
Whats Wrong With 2SD Ranges?

• With 2 controls, there is 10 chance that the
  run will be rejected when there is NOTHING wrong
• With 3 controls per run, there is 15 chance of
  rejection when there is NOTHING wrong
• This wastes
• Time
• and causes frustration!!!

Not againmy control is out of 2 standard
deviations!
3
What to do?

• Analytes show varying biological variation and
  assays show varying accuracy precision.
  Therefore, in order to detect clinically
  significant errors, it is best to determine QC
  rules for an assay that are specifically based
  on 1) its Total Allowable Error (TEa) and 2) its
  specific performance.

Should I automatically repeat my controls?
Thats not true troubleshooting! You need to
develop QC ranges based on how your assay
performs and its total allowable error.
TEa allowable error based on CLIA requirements
for proficiency testing or determined from
individual and group biological variance
4
Solution Assay Specific QC Rules

• Determine TEa for the analyte from CLIA
  regulations or biological variation tables
• Using the Operational Process Specification
  (OPSpecs) charts obtainable from the Westgard web
  site at www.westgard.com, calculate which
  Westgard rules are optimal based on the precision
  accuracy of each analyte in relation to the
  permitted biological variation

• Validate assay to determine imprecision ( cv)
  and inaccuracy (bias)
• Establish mean and standard deviation using the
  historical coefficient of variation (cv) for the
  assay (not simply using a 2SD range established
  from 20-30 runs)

5
Example Historical CV

• Take monthly cv for each control level
• Average for 1 year
• Calculate the SD based on
• SD mean x historical cv
• Use the calculated SD to set the mean /- 2SD
  limit

Takes into account slight variation in
instrumentation and reagents over time.
6
Total Allowable Error

• For Proficiency Testing, CLIA acceptable limits
  for TEa are
• Prothrombin Time 15
• APTT 15
• Fibrinogen 20

7
Westgard OPSpecs Charts

• An Operational Process Specifications (OPSpecs)
  chart (at left) shows the relationship between
  the quality required for a test and the
  imprecision, inaccuracy, and QC that are
  necessary to assure quality is achieved in
  routine operation.
• Allowable inaccuracy on the y- axis versus
  allowable imprecision on the x-axis
• One or more lines representing the operating
  limits (allowable limits of imprecision and
  inaccuracy). These lines correspond to different
  Westgard control rules and different numbers of
  control measurements that would provide at least
  90 detection of medically important systematic
  errors.
• A top line, maximum limits for stable process,
  defines the limits of inaccuracy and imprecision
  for a method that is perfectly stable and would
  require no quality control.
• Normalized Operating Point indicates the
  imprecision and inaccuracy of an individual
  method (example, Prothrombin Time)
• X-coordinate method cv () divided by defined
  TEa (), then multiplied by 100 (1.04 cv
  divided by 15 TEa x 100 6.9)
• Y-coordinate method bias () divided by defined
  TEa (), ratio then multiplied by 100 (1 bias
  divided by 15 TEa x 100 6.7)

Prothrombin Time
The box to the right of the chart matches the
dots and dashes of the different operating limit
lines on the chart to the particular control
rules and numbers of control measurements (N)
being considered. The probablity for false
rejection (Pfr) and the number of runs (R) over
which the control rules are applied is also
shown. An R of 1 means the rules are applied to
the control data in a single run.
Prothrombin Time Example One run using two
controls offers 90 AQA (analytical quality
assurance for systematic error or the chance for
detecting medically important systematic errors)
and 0 probability for false rejection when using
the 13s Westgard Rule.