The Westgard Rules

Quality in the Spotlight: New Dates 2021

Posted by Sten Westgard, MS

The Quality in the Spotlight meeting in Antwerp had been schedule for this May 11th and 12th.

This high level, interactive conference discusses the most trenchant quality issues in laboratory medical diagnostics.

However the COVID-19 epidemic had a different plan.

Now we are re-scheduled with the conference hotel that the conference will be moved to 19th and 20th April 2021.

Join us next year, for the Deep insights, the Limerick Contest, and the Westgard Award.

https://qualityinthespotlight.com/

Curated COVID-19 links: Questions about Quality of testing

Posted by Sten Westgard, MS

As hundreds of new methods rush to market, there are open questions about whether they are appropriate for care. Undoubtedly the established diagnostic manufacturers are bringing their best science to bear. Other companies, new to diagnostics might only be participating in a macabre gold rush.

MSN: Questions about accuracy of coronavirus tests sow worry. April 2, 2020

Washington Post: A 'negative' result doesn't mean you doesn't always mean you aren't infected. March 26, 2020.

NY Times: If you have coronavirus symptoms, assume you have the illness, even if you test negative., April 1, 2020

South China Morning Post: Spanish capital ditches 'unreliable' Chinese coronavirus test kits March 27, 2020

 

Curated COVID-19 links: How the US failed.

Posted by Sten Westgard, MS

Multiple newspapers have tackled this subject: How, why, and when did the US fail at handling the COVID-19 virus?

CNN: How the government delayed coronavirus testing. April 9, 2020

Washington Post: Inside the coronavirus testing failure: Alarm and dismay among the scientists who sought to help. April 3, 2020

NY Times: The Lost Month: How a Failure to Test Blinded the US to COVID-19. April 1, 2020

NY Times: Why the American Approach is Failing, March 24, 2020 

WSJ: America needed coronavirus tests. The government failed. March 19, 2020

 

A covid19 view from Connecticut

Posted by Sten Westgard, MS

Here's a glimpse of the covid19 situation in Connecticut this morning:

"As of Thursday morning, the state’s public health lab in Rocky Hill had completed just 90 of the 1,200 tests it has received from the federal Centers for Disease Control and Prevention.

"'That’s horrible. We are vastly, vastly, vastly under-testing at this point,' Dr. Howard P. Forman, director of the health care management program at Yale School of Public Health, said Wednesday. 'Without testing, you don’t know what prevalence of disease you have in the community.'"

Later in the article it becomes more stark.

"Because the state lab only had the staffing capacity to run 20 tests a day, a decision was made to only test those whose condition had deteriorated enough to be hospitalized....

"By next week, the state lab will have more staff trained so it can complete 60 tests daily. A cohort of nine employees will take turns working three shifts, 7 days a week. But that’s the maximum capacity the state lab will have going forward."

60 tests a day isn't going to help us - we probably need 1,000 tests a day.

When you "lean" out the lab staff, cut costs, reduce funding for public health labs, and labs in general, you set yourself up for these kind of failures. We reap what we sow.

https://www.wnpr.org/post/connecticut-has-run-only-90-covid-19-tests-thats-problem?utm_source=Connecticut%20Public&utm_campaign=d1c5f37cb3-NEWSLETTER_YOUR%20START_20200302_COPY_01&utm_medium=email&utm_term=0_685fda1a0e-d1c5f37cb3-115441717&mc_cid=d1c5f37cb3&mc_eid=07420def70

 

New Book: Quality Control for Qualitative Tests, Paulo Pereira, PhD

Posted by Sten Westgard, MS

We are delighted to congratulate Dr. Paulo Pereira on the publication of his new book,

Quality Control of Qualitative Tests for Medical Laboratories

Here is the Table of Contents

Chapter 1 – ISO compliance

Chapter 2 – Significant causes of uncertainty in qualitative tests

Chapter 3 – Measurement uncertainty and analytical total error in qualitative tests

Chapter 4 – Performance of binary classification tests

Chapter 5 – Agreement of binary classification tests

Chapter 6 – Computation of the cutoff for “in-house” and modified tests

Chapter 7 – Internal quality control and external quality assessment

It's the book we've wanted to write for decades. A strong serious treatment of QC for the testing fields that are particularly challenging for traditional statistical QC.

Q&A: 2020 Questions, Part 2

Posted by Sten Westgard, MS

A recent question came in from Russia:

"We calculate CVa as long-term analytical CV from our routine internal QC. And here we've faced with two internal opinions regarding the CVa to take into account while calculating Sigma:

1. We should take in our CVa final calculation only those points of long-term internal QC, which are due to our analytical system in-control conditions. So we take only those points, which don't have any violations of our rejecting QC rules (for example, 1-3s).
2. We should take in our CVa final calculation absolutely all points of our long-term internal QC, including those which have violations with our rejecting QC rules and which are not accepted in routine IQC."
   

The answer, after the jump

Coming to Kuwait, March 23rd

Posted by Sten Westgard, MS

Coming to Kuwait March 23rd!

I'll be delivering two lectures:

• Breaking all the Rules: the evolution and optimization of QC with "Westgard Rules"
• Seven things you need to know about Six Sigma Metrics: maximizing QC efficiency in the laboratory

Hope to see you there...

http://kuwaitlabweek.com/

What's New on Westgard Web: January 2020

Posted by Sten Westgard, MS

Here's what's new at Westgard Web this month:

Congratulations, Cedars-Sinai Medical Center Core Lab

Cedars-Sinai Medical Center Core Lab becomes the latest laboratory in the USA to join the Sigma Verification Program (VP).

The first lab in Los Angeles to verify their Sigma metric performance, the largest set of methods submitted by a single laboratory to date

More, after the flip...

Q&A: 2020 Questions, part 1

Posted by Sten Westgard, MS

One of our resolutions this year is to answer more questions and share more with the readers. So here's the first installment of our 2020 Questions. This question comes to us from South Africa.

• Regarding which target mean to use.

Q:We use manufacturer target mean values for our IQC measurements, by default. We often change to peer group means if we see a long-standing bias, which is good, but we hardly ever use our own mean value in the Levey-Jennings plot. As far as I understand, this is incorrect and we should ideally use our own cumulative mean whenever sufficient data is available. Is that true?

A: Ideally, build your charts with your own cumulative mean (see recommendation from CLSI C24). Using the package insert mean / target mean is an acceptable place to start, but if your data is significantly different than that mean, you will start generating errors on the charts where there are none.

• Regarding what IQC tells us.

Q: As I understand it, the IQC should be used to monitor imprecision and, perhaps, to pick up any trend in one or other direction away from our acknowledged bias (user mean)....In our lab, we comment on both bias and imprecision of our IQC results and people seem to believe that this bias is a bias compared to an accurate value for that material, whereas I think it is simply an additional bias compounded on our acknowledged method bias....As I understand, the 4s or 10s rules are not intended to pick up a stable “bias”, but rather a trend which may lead to increased (or decreased) bias on our method in the long run.

A: IQC tells us about both imprecision and inaccuracy. Some of the "Westgard Rules" are best at detecting random errors (example, 1:3s and R:4s). Other rules are better at detecting emerging systematic errors (2:2s, 4:1s, 10:x, etc). Certainly EQA/PT is more focused on the long-term and methodological biases that might exist. Ideally, we would like to know our bias from a reference method, reference material, which would represent a "true bias." Practical and financial obstacles usually prevent us from getting that. The bias that we see emerging in QC might be a temporary problem with calibration, instrument machinery, even reagent drift, all of these are important to catch.

• Regarding how to interpret flagged results.

Q: Because our choice of Westgard rules is suboptimal, many tests flag with clinically insignificant deviations. In the past, as clinically trained personnel, it fell to us to determine the clinical significance of the flagged result and make a comment. We would say “not clinically significant” if the deviation from the target value would not affect patient management. We have now been told that this comment is too vague and to rather check the TEa for the analyte and to make a comment “within TEa” if this is the case. I am concerned that this is not a valid application of TEa because the result is (theoretically) only telling us of the deviation of the current result away from our own running mean, but doesn’t include our method’s bias. For example, if an IQC measurement deviates from the target mean by 5% and the TEa for that analyte at that level is 8%, we would say “within TEa.” Whereas, in fact, if our method has a 5% bias in the same direction, then the TE is 10% and not within TEa. Is this correct? (I have used simple addition because I figured that I am adding two absolute error values to get the total error of one IQC measurement away from an accurate value.)

A: Selecting the right rules helps you detect the right errors. A traditional approach from years back was to apply "Westgard Rules" on every test, all the rules for all the controls. We no longer recommend that unless you are in a lab that does everything manually. If you have the tools, we encourage labs to determine the Sigma metric of their tests, which then allows you to design your QC: select only the rules necessary to detect a medically important error, only run the number of controls necessary to detect a medically important error, only run QC when it is helpful to detect a medically important error. In that way, you avoid applying rules that will catch "errors" that are not clinically significant.

For example, with an assay that performs at the Six Sigma level, the full "Westgard Rules" are not necessary. A simple 1:3s rule will suffice, and any violations of the other rules in that test don't have to be flagged or trouble-shooted. When applying this approach, we have been able to help some labs reduce their QC outliers by as much as 85%.

This is a bit more scientific than using a "not clinically relevant" text comment. It's also better than applying the TEa to the size of the error at any given moment.

Muchas Gracias, Lima, Simed, Intercal

Posted by Sten Westgard, MS

Last month I was fortunate enough to attend the INTERCAL meeting in Lima, Peru, hosted by SIMED.