Fewer questions. More confident decisions.

Part One

Do you rely on test results from analytical testing laboratories? Do you release products based on these results? In manufacturing, do you use test results to monitor and control your production process? Are raw materials released to production based on test results?

Do you find yourself waiting for laboratory test results? Do you have products you cannot sell because test results did not meet specification? Is the production process held up or not controlled correctly because of late, or worse, incorrect test results? Do you run out of raw materials at the worst possible time?

There are even more questions! Are you doing too much or too little testing? Do you wonder if you are testing the right thing? Do you question the specifications for your product? Is the timing of your testing coordinated with your production schedule?

Many people in a wide range of industries have been agonizing over these questions. Fortunately, we can now start to confidently answer them.

A New Way Forward

Test results from analytical laboratories are critical for today’s business. Analytical laboratories are implementing changes to ensure that the test results are fit for their intended use.

Changes are being made to international standards and through regulations and laws. The international standard, ISO/IEC 17025:2017, General Requirements for the Competence of Calibration and Testing Laboratories, is based on these changes and improvements. It ensures your needs and requirements are clearly understood by all.

International regulations such as United States Pharmacopeia (USP) general chapters and FDA guidances are including these new tools. Laws are written to ensure health and safety of our food and environment are including the new language.

New, improved systems and tools are being implemented in analytical chemistry. Let’s take a look at some of them.

Fit for Intended Use

How can someone ensure the laboratory’s test results accurately inform the decisions that are made based on the data?  How can you, the customer, tell the laboratory about your testing needs? The tools to support this communication are measurement uncertainty, decision rules, target measurement uncertainty, and probability.

These tools are integral to the new ISO 17025:2017 standard. They are part of the United States Pharmacopeia’s Lifecycle approach to analytical procedures and are included in FDA guidances.

Measurement Uncertainty

To understand and use these tools, a basic understanding of measurement uncertainty is needed. The measurement uncertainty defines the interval around the test result in which the true value is expected to be.

If the test result alone, without the measurement uncertainty, is used to make a decision, there is no knowledge about the possible location of the true value. One is making a decision without understanding the probability of being wrong. When the test result and the associated measurement uncertainty are used, the decision maker has greater confidence of where the true value lies. If the interval is too large for the decision being made, the decision maker can communicate the need for a more reliable test result.

Figure 1 illustrates the interval defined by the measurement uncertainty. In this example, the value needs to be above the specification for the product to be accepted. The interval is calculated as the test result ± the measurement uncertainty. Method 1 has a larger measurement uncertainty than method 2. Hence, the interval around the test result for method 1 is larger. In this example, the method 1 interval crosses the lower specification. We cannot be sufficiently sure the true value is above the specification. The probability of being wrong is not acceptable. For method 2, the interval is above the specification. You can be confident the probability of the true value being above the specification is acceptable; you can confidently accept the product.

Decision Rules

To communicate the product being tested, the specification and the acceptable probability of being wrong, one uses the decision rule. The decision rule allows the customer to define and communicate what they need from the laboratory. It uses language and information that both the laboratory and the customer understand.

A decision rule clearly states the use of the test result. The rule is written by the customer; the person or group that will use the result to make a decision. The decision maker considers the risk so they can include in the decision rule the acceptable probability of being wrong. This ensures the risks, both to the manufacturer and the customer, are understood and agreed to. The decision rule communicates to the laboratory the quality needed of the test result.

To better explain what a decision rule is, let’s look at an example:

The finished product will conform with the specification of 90 to 110 mg/g if the probability of being wrong is less than 5%. Otherwise, the product will be nonconforming.

This decision rule communicates to the laboratory what is being tested – the finished product; the specification – 90 to 110 mg/g; and the acceptable probability of being wrong – 5%.

The laboratory uses this information to select or develop a test method that will produce test results that meet the requirements. The laboratory uses the specification and 5% probability to calculate the target measurement uncertainty which guides the selection, development, and use of the appropriate test method for the finished product. The laboratory ensures the measurement uncertainty is continually met, with a 5% or less probability of being wrong.

The decision rule and its use are illustrated in Figure 2.

Fewer Questions

Measurement uncertainty and decision rules are two of the new tools being used by companies to ensure the voice of the customer is heard. Not only will these tools bring increased clarity and confidence to customers and the laboratory, their use will help break down silos by facilitating communication between functions, enabling and ensuring understanding by all involved.

A future blog will outline the use and benefits of two additional tools recently introduced: target measurement uncertainty and probability. In the meantime, check out our Quality webpage for more information.

Author

Jane Weitzel has been working in analytical chemistry for over 40 years for pharmaceutical and mining companies.  She is currently a consultant specializing in laboratory management systems and ISO/IEC 17025, an auditor, and an educator. Jane is a member of the USP’s Expert Committee on Statistics and the Expert Panel on Method Validation and Verification For the 2015 – 2020 cycle. Currently, Jane is a Senior Quality Expert with Zinata Inc.