8. An Unbiased Look At Bias April 3, 2014

*|MC:SUBJECT|*
Does your Q.C. ensure accuracy?  Really?
View this email in your browser
An Unbiased Look at Bias

The need for the laboratory to assess and monitor bias is obvious.  If results are reported too far above or below their true value, patients will be misclassified and not receive the right treatment.  There are many varying and confusing definitions of bias, accuracy, inaccuracy and trueness.  For this discussion, let's use this: Bias is the variation between the average measured value (the mean) and the True Value. 

Bias changes whenever the mean changes.  That's virtually every time there is a change in reagent lots, calibrators or instrument components.  That change may be for the better or for the worse, acceptable or unacceptable. 
Webinar Thurday April 3, 2014  2:30 PM EST
Discussion Topics:
  1. What is the difference between accuracy, bias and trueness?
  2. Which of these would you choose if asked to prove that method accuracy is clinically acceptable today?  Select just one, or a combination.
    1. A recent proficiency  survey with acceptable results
    2. A ‘Pass’ on a Trueness experiment performed as part of method validation
    3. A recent Interlab Peer  Report with an SDI less than 2.0
    4. Current stable Daily QC with no QC rule failures
    5. Other evidence or report:
  3. How much matrix effect can be tolerated before QC results become meaningless?
  4. What is the role of patient samples for Q.C.?
 Explore what really happens with "Prof Zoe's Q-See Simulator"
Join the fun!  Join the panel!  Send a message
Register for Webinars
Apply the DIMS Test to this:
Levey-Jennings 1-2s quality control was developed 60 years ago, when the norm was to use patient samples.   Does It Make Sense to test artificial control samples without verifying that they behave exactly as patient samples?   Using surrogate controls fundamentally differentiates lab Q.C. from industrial processes,  yet labs use the same Q.C. concepts as industry.  Has this been adequately studied?
Surve. Does This Make Sense
Do normally-distributed QC samples reflect changes in a log-normal patient populations?

If an HIV Q.C. sample at a S/CO level of 3.0 drops by two SD, do current low positives move down by the same number of SD, the same percent or the same S/CO values?  Should we convert the log-normal results to normal to project changes?  Has this been studied and documented well enough?  Is there an alternative that should be considered?
"The Great Bias Study of 1994"
The same problems we identified twenty years ago still exist today. 
Why have analytical processes advanced so  much while Q.C. has changed so little?  




Time permitting, the panel will discuss case studies from this poster, such as:
A pooled patient sample was tested for Albumin in seven clinical laboratories. How would you decide if each laboratory is reporting clinically-acceptable results? Would you suspect unacceptable accuracy, precision, linearity, specificity, sensitivity, or reference intervals?  Are all these methods OK?

Test

Sample

Average

Lab A

Lab B

Lab C

Lab D

Lab E

Lab F

Lab G

Albumin

Patient Serum

 

 

 

 

 

 

 

 

g/L

High Pool

45.7

45.0

45.0

45.0

46.0

47.0

47.0

45.0

g/L

Low Pool

37.7

36.0

37.0

36.0

39.0

39.0

40.0

37.0

 

Ref Range Low

 

60

50

44

51

51

51

50

 

Ref Range High

 

40

35

34

39

38

30

35


Link to the original poster.   Join the discussions Thursday, 2:30 PM EST
Register for Webinars
Mark your calendar!

An Unbiased Look at Bias

3-Apr

When is a QC problem really a quality problem
True Confessions of a Service Rep 
…Problem Solving; Re-analyse patients; Recovery efforts

17-Apr

Evaluating and Monitoring Quality 
4 Key Numbers + 3 Easy Decisions
A Q.C. lesson that should never be taught

1-May

The cost of Failure to Control Quality
…  Implementation and Impact of Optimized Q.C.

15-May

EP 23  Risk Based Quality
…  Illustrated with OptimiZer Reports of Case Studies

29-May

The root cause of ineffective Q.C.; the solution; the challenge

12-Jun

Stats-Free Quality Control

26-Jun

Open Dialogue

10-Jul

How to get from here from there

17-Jul

     
Take The Q.C. Chart Challenge!
Thanks for caring for quality!         Zoe Brooks
Please forward this to friends and colleagues.
Zoe Brooks     April 1, 2014
Please note.  I am not being critical of any individual person, institution or organization.  I want to express my most sincere respect, admiration and appreciation to  Dr. Westgard and so many others.  This process does not contradict  today's  published  recommended, principles and processes; it just implements them differently. 
Last modified: Wednesday, 16 April 2014, 10:47 AM