A coconut milk or cream spec needs different microbiological lines depending on how the product is preserved. Retort and aseptic product is judged on commercial sterility. Chilled, frozen, pasteurised and powdered product is judged on counts: total plate count (TPC), yeast, mould and Enterobacteriaceae. Putting count limits on a sterile product, or a sterility line on a pasteurised one, produces a spec the supplier cannot meet or one that protects nobody.
The reference standard does not settle the numbers. Codex CXS 240-2003 covers coconut milk and cream, sets composition and a minimum pH of 5.9, and then stops. Its hygiene clause says only that products should comply with microbiological criteria set under CAC/GL 21-1997. Every number on your spec is therefore a number you and the supplier agree, and each one needs a method and a sampling plan behind it.
What does Codex actually say about microbiology?
Very little, and that is the first thing to know before negotiating limits. CXS 240-2003 requires heat processing “so as to prevent spoilage”, before or after the container is sealed. It points to the Codex codes of hygienic practice for low-acid canned foods (CAC/RCP 23-1979) and for aseptically processed low-acid foods (CAC/RCP 40-1993). It gives no TPC, yeast, mould or pathogen limit.
The only quality sampling it defines is an AQL of 6.5 for composition and colour defects. That plan does not apply to micro results. A lot with a 6.5 AQL on fat content can still carry a failed sterility test, so the micro clause in the contract has to stand on its own.
Why does the preservation route change the spec?
Coconut milk is a low-acid food. Codex sets a minimum pH of 5.9, and under 21 CFR 113.3 a low-acid food is anything with a finished equilibrium pH above 4.6 and water activity above 0.85. That is the territory where spore-formers survive weak heat and grow in the can.
The same regulation defines commercial sterility as the condition in which heat has left the food free of microorganisms able to reproduce under normal non-refrigerated storage and distribution, plus viable organisms of public health significance. A retorted can that meets this does not need a plate count line. A count on a sterile can that shows growth is a failed can.
| Format | Preservation | Core micro lines | Lines to leave off |
|---|---|---|---|
| Retort can or pouch | Heat in sealed pack | Commercial sterility after incubation; no swollen or leaking packs | TPC, Enterobacteriaceae |
| Aseptic carton or bag-in-box | Heat then sterile fill | Commercial sterility; incubation of retained packs; filler records | TPC on finished product |
| Pasteurised, chilled | Mild heat plus cold chain | TPC, yeast and mould, Enterobacteriaceae, Salmonella, shelf-life study | Commercial sterility |
| Frozen | Freezing | TPC, yeast and mould, Enterobacteriaceae, Salmonella | Commercial sterility |
| Powder (spray-dried or reconstitutable) | Low water activity | TPC, yeast and mould, Enterobacteriaceae, Salmonella | Commercial sterility |
The table gives the structure. Numerical limits for the count lines come from the buyer’s own risk assessment, the end use and the supplier’s process validation, so they are set per contract rather than copied from a template.
Which count lines earn their place on a non-sterile spec?
TPC is a general hygiene and shelf-life indicator. It tells you how clean the raw kernel handling and extraction were and how much of the load survived the heat step. Specify the incubation temperature and the method, for example a pour-plate count at 30 °C, because the same sample reads differently at 35 °C.
Yeast and mould are the spoilage lines. Coconut milk is a nutrient-rich, near-neutral medium, so a surviving yeast or mould has plenty to grow on. The method depends on water activity: ISO 21527-1 applies to products above 0.95 and ISO 21527-2 to products at or below it, so a liquid cream and a powder use different parts of the standard. The ISO food microbiology method index lists both.
Enterobacteriaceae are the useful line for finished product that has seen a heat step. Ordinary pasteurisation kills them, so a positive result points to contamination after the heat, from a filler, a transfer line or a hand. ISO 21528-2 is the colony-count method. This is a process-hygiene signal and works better than TPC for catching a dirty filler.
Salmonella is a presence or absence test per 25 g in a stated number of samples. Specify the number of units and the method, ISO 6579-1 in most labs. A single-sample “absent” result gives much less assurance than a lot-level plan.
How bad is unpasteurised coconut milk?
A Malaysian study of 50 fresh coconut milk samples from wet-market vendors found every sample unsatisfactory on total plate count and total coliforms against the ready-to-eat guidelines used in the study (USM repository). Fresh milk is a high-risk matrix: neutral pH, high water activity, fat and protein.
That context explains why importers should not accept “freshly extracted, then frozen” product without a validated heat step and a count-based spec. Freezing preserves whatever load went in.
What belongs in the contract wording?
The strongest micro clauses name five things for every line: the organism or group, the method, the test portion, the sampling plan, and the action on failure. A line without a method is an argument waiting to happen. Our guide to writing a spec suppliers can hit covers the method-and-tolerance logic in more detail.
Three additions help on sterile formats. Require retained samples from each lot, incubated at the supplier’s site, with the records available on request. Require the thermal process authority letter and the scheduled process for the pack size shipped. Require lot coding that ties each pallet to a filler run and a retort or aseptic record.
Match the micro clause to the rest of the spec. Fat standardisation drives viscosity and heat penetration, which is one reason 17% and 22% cream are processed differently. A pack ordered ambient should ship in a dry box, as set out in reefer or dry for coconut milk, because the sterility claim assumes ambient storage.
What this means for buyers
- Procurement managers (A): Write two micro clauses, one for sterile formats and one for count-based formats. Never reuse a powder spec on a canned product.
- Importers and distributors (B): Ask for the incubation records and the process authority letter before the first container. A swollen-pack complaint after arrival is a claim you pay for first.
- Brand owners (C): If you reconstitute milk from powder or concentrate in your own plant, the Codex label rule treats it as a reconstituted product, and your micro responsibility starts at the mixing tank. Compare formats in powder versus liquid landed cost.
- Traders and brokers (D): Confirm which micro regime the contract references before agreeing a price. A cheaper lot without a sterility record is a different product.
FAQ
Does Codex set TPC or yeast and mould limits for coconut milk? No. CXS 240-2003 requires heat processing and refers to CAC/GL 21-1997 for microbiological criteria. Limits come from the contract, the buyer’s risk assessment or the destination regulator.
Should a canned coconut milk spec include a plate count? Usually the primary line is commercial sterility after incubation. A plate count is relevant for investigating a failed can, and not as a release limit on a sterile pack.
Why test Enterobacteriaceae on finished product? Heat kills them, so finding them after processing means contamination after the heat step. It flags a hygiene failure at the filler or transfer line.
What pH makes coconut milk a low-acid food? Anything with a finished equilibrium pH above 4.6 and water activity above 0.85 is low-acid under 21 CFR 113.3. Codex sets coconut milk at a minimum pH of 5.9.
Need a micro clause written into a coconut milk or cream contract? Send the desk an RFQ at /contact/.