Measuring Uncertainty in Food Testing Laboratories

Every measurement in a food testing laboratory carries some degree of doubt. When a laboratory reports that a food sample contains 2.1 μg/L of lead, the actual value might be slightly higher or lower. This inherent variability is called measurement uncertainty, and understanding it is critical for making informed decisions about food safety, regulatory compliance, and product quality.

Measurement uncertainty describes the dispersion of values that could reasonably represent the true amount of what you’re measuring. It’s not an error or a mistake-it’s a fundamental characteristic of every analytical measurement. For food testing laboratories, properly estimating and reporting this uncertainty ensures that test results are reliable, comparable across different laboratories, and legally defensible.

What exactly is measurement uncertainty? 🔗

Think of measurement uncertainty as a confidence interval around a test result. When laboratories report results, they typically provide both a measured value and an uncertainty range. For example, a result might appear as “Ochratoxin A = 2.07 μg/kg ± 0.47 μg/kg,” where ±0.47 represents the expanded uncertainty at approximately 95% confidence.

The reported value is the laboratory’s best estimate, but the uncertainty tells you the realistic range where the true value likely falls. In this example, the actual ochratoxin content could reasonably be anywhere between 1.60 and 2.54 μg/kg.

Why food testing laboratories must estimate uncertainty 🔗

Since 1999, ISO/IEC 17025-the international standard for testing and calibration laboratories-has required laboratories to estimate measurement uncertainty for their testing procedures. This requirement exists because uncertainty affects how we interpret results, especially when comparing them against regulatory limits or specifications.

Without knowing the uncertainty, you cannot make truly informed decisions about compliance. Improper calculations of contamination levels can mean products are either unnecessarily discarded or, worse, sold with potential health risks. For food safety testing, understanding uncertainty ensures that decisions about product release or rejection are based on solid statistical ground.

The ISO 17025:2017 requirements 🔗

The 2017 revision of ISO/IEC 17025 strengthened requirements around measurement uncertainty. Laboratories must identify all significant contributors to uncertainty, evaluate their impact, and report uncertainty when it’s relevant to the validity of results or affects compliance with specifications. Importantly, laboratories must now agree with clients in advance on how they will account for uncertainty when making compliance decisions.

Major sources of uncertainty in food testing 🔗

Multiple factors contribute to measurement uncertainty throughout the testing process. Understanding these sources helps laboratories minimize uncertainty and helps clients interpret results more accurately.

Environmental conditions 🔗

Temperature fluctuations, humidity levels, and atmospheric pressure can all affect measurement results. Chemical reactions may proceed at different rates under varying temperatures, while moisture in the air can affect weighing procedures. Laboratories control and monitor these conditions, but small variations remain a source of uncertainty.

Instrument performance and calibration 🔗

No analytical instrument is perfectly precise. Even calibrated equipment has inherent limitations based on its design, age, and maintenance status. Additionally, the reference standards used for calibration carry their own uncertainty, which compounds through the measurement chain.

Sampling and sample preparation 🔗

One of the largest contributors to uncertainty often occurs before the sample even reaches the instrument. Studies show that sampling and sample preparation can account for 60-90% of total variance in results, far exceeding the contribution from the analytical measurement itself. How you collect, homogenize, and process samples dramatically affects the final result.

Operator variability 🔗

Different analysts may perform the same test slightly differently. Timing variations, interpretation of color changes, or technique in pipetting all introduce small differences. Well-trained staff following standard procedures minimize but cannot eliminate this source of uncertainty.

Method limitations 🔗

Every analytical method has inherent limitations in its precision, accuracy, and detection capabilities. The method’s performance characteristics-established through validation studies-define its expected uncertainty range for specific applications.

How laboratories calculate and report uncertainty 🔗

Laboratories use statistical approaches to estimate measurement uncertainty. The process typically involves identifying all significant uncertainty sources, quantifying each contribution, and combining them using established mathematical methods.

The standard uncertainty represents about 68% of the possible spread of results in a normal distribution. However, laboratories usually report the expanded uncertainty, multiplying the standard uncertainty by a coverage factor (typically 2) to provide approximately 95% confidence. This means that if you repeated the measurement many times, about 95% of results would fall within the stated uncertainty range.

Different reporting formats 🔗

Laboratories may report uncertainty in different ways: as standard deviation, standard uncertainty, or expanded measurement uncertainty. It’s essential to understand which format your laboratory uses, as they represent different levels of confidence. Always check the accompanying statement that explains the coverage factor and confidence level.

Uncertainty and compliance decisions 🔗

Measurement uncertainty becomes particularly important when assessing whether a product meets regulatory limits or specifications. Consider a pesticide limit of 5 mg/kg. If your test result is 4.8 mg/kg with an uncertainty of ±0.5 mg/kg, the true value could be anywhere from 4.3 to 5.3 mg/kg-potentially above the limit.

Decision-makers must consider whether results that span a limit should be treated as compliant or non-compliant. Regulatory authorities typically give the benefit of doubt to producers when uncertainty spans a limit, while producers should take a more conservative approach in their own compliance assessments to ensure they stay well within legal requirements.

Minimizing uncertainty in practice 🔗

While uncertainty cannot be eliminated, laboratories can take steps to minimize it. Strict internal controls at all stages of testing, regular equipment calibration, proper staff training, and participation in proficiency testing programs all help reduce uncertainty. Using validated methods appropriate for the matrix and analyte, maintaining consistent environmental conditions, and following meticulous sample handling procedures also contribute to lower uncertainty.

For microbiological testing, which presents unique challenges due to the living nature of the measurand, specialized approaches outlined in ISO 19036 help laboratories estimate uncertainty appropriately.

Why smaller uncertainty doesn’t always mean better 🔗

It’s tempting to assume that a laboratory reporting smaller uncertainty provides better results. However, this isn’t always true. Different laboratories may calculate uncertainty using different methods, and a smaller reported uncertainty might simply reflect less thorough evaluation rather than superior performance.

Before relying on uncertainty estimates, verify that the laboratory has included all relevant uncertainty components in their calculation. A laboratory that accounts for more sources of variability may report larger-but more honest-uncertainty values.

What do you think? How might understanding measurement uncertainty change the way you evaluate laboratory results for your products? When setting internal specifications for your food products, should you build in a safety margin that accounts for measurement uncertainty?

References

1. https://www.campdenbri.co.uk/white-papers/uncertainty-of-measurement.php
2. https://www.ifst.org/resources/information-statements/how-choose-and-instruct-laboratory-chemical-food-analysis
3. https://www.lablynx.com/resources/faqs/what-standards-and-regulations-affect-a-food-safety-laboratory/
4. https://www.qse-academy.com/measurement-uncertainty-in-testing-and-calibration/
5. https://link.springer.com/article/10.1007/s00769-010-0721-6
6. https://www.isobudgets.com/4-ways-to-calculate-uncertainty-in-microbiology-labs/