A load of cooked chicken strips arrives with the supplier’s certificate: aerobic plate count 60 000 cfu/g. Your own lab tests the same delivery and reports 180 000. The specification says 100 000 maximum, so you reject the load, the supplier objects, and by Monday two managers are arguing about whose laboratory is incompetent. Quite possibly neither is. Measurement uncertainty in food microbiology explains the gap.
In short
- A plate count is an estimate built from a small number of colonies and several dilutions, never an exact figure.
- The expanded uncertainty on a plate count is commonly around plus or minus 0.5 log, roughly threefold either way. Your lab can give you its own figure. Results closer together than that are effectively the same.
- Sampling errors are usually larger than laboratory errors, and no uncertainty figure on a certificate includes them.
- Use a laboratory accredited to ISO/IEC 17025 for the specific test, and write specifications that state the method, the sampling plan and how disputes are settled.
Uncertainty of measurement for microbiological methods is one of the lessons in module 4 of the Advanced Food Microbiology Course from ASC Food Safety Training (8 hours, online, R1 750). The dispute above is a good way into it.
Why do two labs get different counts from the same sample?
Two labs get different counts because they never test the same sample. Each lab takes its own 10 g portion from a food in which bacteria are clumped and unevenly spread, dilutes it, and counts a few dozen colonies on a plate. Every one of those steps adds variation.
Two competent labs will routinely differ by a factor of two or three.
A portion is weighed and blended, and a clump of twenty cells that stays together grows as one colony. The blend is diluted in tenfold steps, each with a small pipetting error that carries forward. One millilitre goes onto a plate, and which cells happen to be in it is a matter of chance. Then a person counts, and two analysts will not always agree on a crowded plate.
How big is measurement uncertainty in a food microbiology count?
For routine plate counts on food, laboratories commonly report an expanded uncertainty in the region of plus or minus 0.5 log10 cfu/g. Treat that as typical. Every accredited lab estimates its own figure for each method, often following ISO 19036, and it can be smaller or considerably larger.
Half a log sounds small until you convert it. It is a factor of just over three in each direction, so a reported 100 000 cfu/g is consistent with a true value anywhere from about 32 000 to about 320 000.
Worked example
Specification: aerobic plate count not more than 100 000 cfu/g (5.0 log). Assume both labs quote an expanded uncertainty of 0.5 log.
Supplier’s lab: 60 000 cfu/g = 4.78 log. Range 4.28 to 5.28 log, or about 19 000 to 190 000 cfu/g.
Your lab: 180 000 cfu/g = 5.26 log. Range 4.76 to 5.76 log, or about 57 000 to 575 000 cfu/g.
The results are 0.48 log apart and the ranges overlap from 4.76 to 5.28 log, so the labs do not disagree. The 5.0 log limit sits inside both ranges. Neither certificate proves that the load passes or that it fails.
The honest reading is that this product runs at about its limit, which is a conversation about the supplier’s process and cold chain, not a fight about certificates.
What is the countable range on a plate, and why does it matter?
A plate gives a reliable count only within a window, set by the method, that commonly runs from a lower limit of somewhere between 10 and 30 colonies to an upper limit of 250 or 300. Below the window, chance dominates. Above it, colonies crowd, merge and suppress one another, and the count reads low.
The lower end explains why low counts are so jumpy. Random scatter in a colony count scales roughly with the square root of the count. At 100 colonies that scatter is about 10 colonies, or 10 percent. At 9 colonies it is 3, a third of the count, before any other error is added. A result of 90 cfu/g built from nine colonies is a loose estimate, and labs often flag such figures as estimated.
Do not build tight specifications near the detection limit, where results bounce between <10 and 40 for no reason you can control. And tell the lab what range you expect, so that they plate the right dilutions. Low results are covered in how to read a food microbiology lab report.
Applying this to your own certificates takes practice, and a rejected load is an expensive place to get it. The advanced course in food microbiology testing and interpretation works through detection, enumeration and identification methods first, so that the uncertainty lesson lands on something you already understand.
Start the Advanced Food Microbiology Course, R1 750
Which sampling mistakes make a lab result meaningless?
A sampling mistake makes a lab result meaningless when it adds organisms to the sample, lets them grow before testing, or makes the sample unrepresentative of the lot. The lab’s uncertainty figure covers what happens after the sample is opened on its bench. Everything before that, from the scoop to the hours in a bakkie without ice bricks, is yours.
| Sampling error | What it does to the result | The fix |
|---|---|---|
| Unsterile scoop, knife or container, or a bare hand inside the bag | Adds organisms from the tool or the sampler. | Sterile tools and containers, sanitised or gloved hands, container open for as short a time as possible. |
| Warm transport or a long delay | Growth in transit. A 4 log product arrives as 5 log. | Insulated box with ice bricks, chilled and never frozen, delivered within the time and temperature your lab accepts. Temperature recorded on receipt. |
| Always sampling the first pack of the shift | Represents the cleanest moment of the day, not the lot. | Vary the time of sampling. |
| Poor labelling and no chain of custody | A result that cannot be tied to a batch is of no use in a dispute. | Batch code, date, time, sampling point and sampler on every sample. |
The same discipline applies to environmental swabs. See our surface swabbing guide.
What do method validation and ISO/IEC 17025 accreditation tell you?
Method validation shows that a test method does what it claims, for example that a rapid method performs as well as the reference method. Accreditation to ISO/IEC 17025 shows that an independent body has assessed the laboratory’s competence to run specific methods. In South Africa that body is SANAS. Neither says anything about your sampling.
Three checks take ten minutes.
- Accreditation is granted per method, so check that the lab’s SANAS schedule of accreditation lists the test you are paying for. A lab can be accredited for aerobic plate count and not for Listeria detection.
- Where the lab uses a rapid or alternative method, ask what it was validated against. The ISO 16140 series is the usual framework.
- Ask for the measurement uncertainty on the tests you use most. An accredited lab has to estimate it.
How do you write a specification that accounts for uncertainty?
A specification that accounts for uncertainty states the method, the units, the sampling plan, the point of sampling and the maximum sample temperature on receipt. Agree in writing how results are compared with the limit and how disputes are settled, and set your process target far enough below the limit that normal variation does not produce failures.
- Method and units. “Aerobic plate count, ISO 4833-1 or equivalent validated method, cfu/g.”
- A three class plan where it fits. With n, c, m and M, the band between m and M is the tolerance for normal variation. A single hard limit judged on one sample invites the dispute described above.
- A decision rule. The common convention in food microbiology is to compare the reported result directly with a limit that was set with variability in mind. Whatever you choose, both parties sign it before the first delivery.
- A referee. An agreed third laboratory, accredited for the method, and a properly stored retained sample.
- A process target. If the limit is 5 log, a supplier whose trend sits at 4 log or lower will seldom trouble you. One who runs at 4.8 log will fail regularly on uncertainty alone.
Take that last point to the chicken supplier. Which groups deserve a limit at all is covered in indicator organisms versus index organisms.
What does the Advanced Food Microbiology Course cover, and what does it cost?
The Advanced Food Microbiology Course from ASC Food Safety Training is an 8 hour online course that costs R1 750. Its five modules are micro-organisms in a food handling environment, growth and reproduction, indicator organisms, sampling, enumeration and testing, and food quality and safety in microbiology.
Module 4 holds this post’s material: aseptic technique, interpretation of laboratory results, and method validation and uncertainty of measurement.
Course facts
| Course | Advanced Food Microbiology Course. Provider: ASC Food Safety Training. Instructor: Mthokozisi Nkosi |
|---|---|
| Level | Advanced, the third step in the ASC microbiology ladder |
| Time | 8 hours |
| Price | R1 750 |
| Format | Online, self paced, lifetime access |
| Assessment | An assessment for each module, then a final assessment |
| Certificate | Certificate of completion |
| What you will be able to do | Sample aseptically, interpret laboratory results with their uncertainty in mind, and apply testing and laboratory techniques to safety, quality and shelf life decisions |
| Who it is for | QCs, QA and technical managers, food technologists and supplier quality staff who write specifications, take samples or argue about results |
Enrol in the Advanced Food Microbiology Course, R1 750
Questions people ask before they enrol
How long does the course take, and what does it cost?
The course is 8 hours and costs R1 750, paid by card or EFT through PayFast. It is online and self paced. You keep access for life, which suits a topic like uncertainty that you may only need twice a year, usually mid dispute.
Do I need to complete the earlier microbiology courses first?
This is the advanced step. It follows Introduction to Food Microbiology (R1 195) and Intermediate Food Microbiology: Principles and Application (R1 450). If you already read certificates of analysis at work, you can start here. If not, begin lower down.
Will the course make me a laboratory analyst or a registered microbiologist?
No. It is a training course with a certificate of completion, and it does not register you with any professional body. It aims to make you a competent user of laboratory results: someone who can sample properly, question a certificate and write a sensible specification.
How is the course assessed, and do I get a certificate?
Each of the five modules has an assessment, and a final assessment closes the course. You receive a certificate of completion, which you can file as training evidence in your food safety team records.
Can I enrol if I am outside South Africa?
Yes. Pricing is in rand, with an indicative USD, EUR and GBP equivalent shown on the course page. Payment settles in rand and your bank handles the conversion. The course is delivered fully online.
The next time two certificates disagree, you should be the person in the meeting who can explain why, and what to do about it. One wrongly rejected load, or one supplier dispute that drags on for a week, costs more than R1 750. Enrol in the ASC Advanced Food Microbiology Course and get there in 8 hours. To build up to it, the food microbiology training hub sets out all three steps.
ISO and ISO/IEC standards are published by ISO and the IEC. SANAS is the South African National Accreditation System. This course is an independent training product and is not approved or endorsed by ISO, SANAS or any certification body.