Metrology Monday! #65 Automation for Time Savings and Quality

Last week I participated in our annual CEO Kaizen.? The term Kaizen is a Japanese term that means “change for the better” or “continuous improvement.”? Nearly 5% of our company participated in a Kaizen event in one way or another last week, and the finals were reviewed by our company president and the CEO of Fortive.?

I had the opportunity to lead a team on an event that addressed automation.? One of the great things about Kaizens are that you may go into the week with some preconceived notions, but through learning and evaluating data, the notion may be completely disproven.? This is where the magic happens, as some of the greatest inventions and discoveries have been made completely by accident.? Going into the Kaizen, we thought that the greatest value in automating processes that are traditionally done by hand (e.g. take measurement, write down or key in results) and finding ways to automate them would have great appeal to people who are trying to reduce technician time.? As we all know, good technicians are extremely valuable and hard to find. If we automate tasks so that they can continue to do other important activities, it is a big win.?

When we started talking to our customers about a proposed solution, we found out that what customers really valued was the reduction in errors in the data transfer process.? One customer told us that they have evaluated calibration certificates where the data came from manual measurements and found that 10-20% of the certificates that they sampled had data transcription errors in them.?

This inspired me to have another look at what ISO/IEC 17025 has to say about this.?

Section 7.11 addresses Control of Data and Information management.? In other words, this section lays out requirements for Laboratory Information Management Systems (LIMS).? When we revised the 2017 version, we borrowed from what was already in ISO/IEC 15189, Medical Laboratories, Requirements for quality and competence.? This document is for testing in a medical laboratory and is very similar to what we do in ISO/IEC 17025.? We fundamentally used the 15189 requirements and dropped them into the 2017 edition of 17025.? However, I was adamant that we keep one requirement from 17025, and it is in the 2017 edition in section 7.11.6.? This requirement, “Calculation and data transfers shall be checked in an appropriate and systematic manner” was a requirement that carries back to the original version of 17025 in 1999.? What it means is that whether it is one computer system talking to another, or if it is a person transferring hand-recorded data into a spreadsheet or other device, there must be a process in place to check that this transfer maintains data integrity.

Where I really learned this lesson was back in the days when I owned and operated a proficiency testing company for calibration laboratories.? I required that the customer provided a document associated with their measurement data for the proficiency test and I would send the customer back a Data Verification Form with all data typewritten into our system for them to evaluate, and attest to the accuracy via signature.? This did two things:? 1) it was a way to double-check that I appropriately interpreted and entered all the customer data, but more importantly 2) it made the customer agree that this was their ‘final answer.’

?It was not unusual to find during the evaluation of the customer proficiency that they made errors in transcribing data.? They could miss a digit. For example, if the customer was measuring resistance with an HP 3458A multimeter, I knew the resolution they would use.? I would see the customer submit a measurement result such as 99.9995 ohms when the actual meter reading was 99.99995 ohms (they dropped one of the nines in transcription).? The difference between what they really measured and what they reported is 4.5 parts per million, which could be enough of an error for them to fail the proficiency test.? Since I had witnessed this happen several times in real life, I pushed for us to keep 7.11.6 in the latest version of ISO/IEC 17025.

I really want to commend the customer for not only having a system in placed to ensure that data transfers are checked, but they also acted on it, revising their original calibration certificates, and performing the corrective actions necessary with their regulatory body, even though they were time consuming and expensive to do.? It really speaks to their integrity.

I was also an ISO/IEC 17025 assessor for nearly 10 years.? I can say that laboratory conformance this requirement is weak at best.? The term appropriate is makes evaluating conformance very tricky.? I would still defend the standard’s use of the term appropriate in this instance because appropriate checks can look very different for completely automated systems and systems that are very manual.? I find that completely automated systems usually have a high degree of accuracy in data transfers, but whenever there is a human involved, not so much.

This is where I wanted to put some questions to my audience.? What are the appropriate checks that your organization does in a systematic manner to ensure that data that is transferred manually at any point does not have errors?? I am sure that many of you have great practices and I hope you can share some of them with us.? I also suspect that many laboratories do little to nothing to check manual data transfers.? It is these people that I am writing the article for today.? We have a lot of things to do to keep our laboratories running, and it is easy to overlook some of the fine points.? Last week brought this back to my attention and I wanted to share this with my readers, because as a group, we are all committed to continuous improvement.? #MetrologyMonday #FlukeMetrology??