Is it a CCP? The Codex decision tree worked through five South African products

Most HACCP teams can recite the decision tree, but they freeze when a real process step is on the table: is the cold room a CCP or not? The fastest way to build that judgement is to see CCP decision tree examples worked through products you know, so below we take five South African products through the tree.

What the decision tree does, and what it does not decide

Codex revised its General Principles of Food Hygiene (CXC 1-1969) in 2020 and later added an updated example decision tree to the text. Codex presents the tree as a tool, not a requirement: you do not have to use it, and another logical approach that meets the principle is acceptable. It supports the team’s judgement rather than replacing it.

The current tree is applied to each step where a significant hazard has been identified, and asks four questions in this order. We use short labels for them below:

• Q1, the PRP question: can the significant hazard be controlled to an acceptable level at this step by prerequisite programmes, such as good hygiene practices? If yes, this step is not a CCP. Codex notes these may be routine GHPs or GHPs needing greater attention, such as extra monitoring and records.
• Q2, the control question: do specific control measures for this hazard exist at this step? If no, this step is not a CCP and the team evaluates later steps.
• Q3, the later-step question: will a subsequent step prevent or eliminate the hazard, or reduce it to an acceptable level? If yes, that subsequent step should be the CCP, not this one.
• Q4, the this-step question: can this step specifically prevent or eliminate the hazard, or reduce it to an acceptable level? If yes, this step is a CCP.

If no CCP is found for a significant hazard, the tree tells you to modify the process or product, redo the hazard analysis and start the tree again. A footnote adds an important point: where the control measure at one step works in combination with a control measure at another step to control the same hazard, both steps should be considered CCPs.

Before you start: a hazard analysis worth deciding on

The tree only works on significant hazards, so the quality of the answer depends on the hazard analysis. For each step the team must have named specific hazards (not “bacteria” but Salmonella, Listeria monocytogenes, metal fragments, undeclared peanut), judged their likelihood and severity, and listed the control measures that exist. If the hazard analysis says “biological, chemical, physical” at every step, the tree will produce nonsense.

Worked examples: boerewors and biltong

Boerewors: chilled storage and metal detection

Boerewors is raw and is cooked by the consumer. Pathogens such as Salmonella and E. coli come in with the raw meat. There is no kill step in the factory, so the controls are about keeping numbers low and preventing growth.

Step: chilled storage, hazard: growth of pathogens. PRP question: many teams answer yes. A maintained cold room with temperature monitoring and alarms, stock rotation and a short shelf life is a good hygiene practice, and the consumer’s cooking is the final control. Not a CCP. Some teams with long holding times or no alarms decide the PRP is not enough and treat it as a CCP. Either answer can be defended if the reasoning is written down.

Step: metal detection, hazard: metal fragments from the mincer and filler. PRP question: maintenance and blade inspections reduce the risk but cannot remove a fragment once it is in product. No. Control question: yes, the detector. Later-step question: no, nothing after it removes metal. This-step question: yes, it is designed to find and reject metal. CCP.

Biltong: vinegar dip and drying

Biltong is eaten without cooking, so the process itself must deal with Salmonella, E. coli and Listeria monocytogenes on the raw meat. The control comes from a combination: vinegar treatment, salt and spices, and drying.

Step: vinegar dip, hazard: pathogens on the meat surface. PRP question: no, hygiene alone cannot do this. Control question: yes, the acid treatment. Later-step question: drying helps, but most processes validate drying together with the dip, not alone, so the honest answer is no. This-step question: the dip alone may not reduce the hazard to an acceptable level either.

Step: drying, same hazard. Drying lowers water activity and finishes the job the dip started, but neither step does it alone. This is the case the Codex footnote covers: because the dip and drying work in combination to control the same hazard, both steps should be considered CCPs, each with its own critical limits. Your validation must show that the combination achieves the reduction you claim.

Worked examples: a chicken mayo sandwich and fruit juice

Chicken mayonnaise sandwich

Cooking the chicken, hazard: survival of Salmonella and Campylobacter. PRP question: no. Control question: yes, cooking to a validated time and temperature. Later-step question: no, nothing after cooking will kill them. This-step question: yes. CCP.

Cooling the cooked chicken, hazard: growth of spore formers such as Clostridium perfringens. Cooking does not destroy spores, and slow cooling lets them germinate and grow. PRP question: in a small kitchen with a blast chiller and a cooling procedure, some teams decide PRPs cover it. In a factory cooling large batches, most teams answer no. Control question: yes, controlled cooling. Later-step question: no, the sandwich is not heated again. This-step question: yes. CCP for most manufacturers.

Assembly, hazard: contamination with Listeria monocytogenes and Staphylococcus aureus from hands and surfaces. PRP question: yes. Hygienic zoning, handwashing, cleaning and environmental monitoring are prerequisite programmes. Not a CCP. This is where many sites wrongly add a CCP called “hygiene”.

Fruit juice

Pasteurisation, hazard: survival of pathogens such as E. coli and Salmonella. PRP question no, control question yes, later-step question no, this-step question yes. CCP.

Fruit receiving and sorting, hazard: patulin from mouldy apples. Patulin is not destroyed by pasteurisation. PRP question: supplier approval, a raw material specification, removing rotten fruit and testing are mostly prerequisite programmes. Where the team judges these sufficient, it is not a CCP. Where sorting is the only barrier, a team may decide sorting is a specific control measure and a CCP.

Washing the fruit, hazard: pathogens on the skin. PRP question: no. Control question: yes, washing does reduce surface contamination. Later-step question: yes, pasteurisation will reduce pathogens to an acceptable level. So pasteurisation is the CCP for this hazard, and washing is not.

Worked example: roasting and metal detection in peanut butter

Peanut butter is a low moisture food, and Salmonella can survive in it for a long time even though it cannot grow.

Roasting, hazard: Salmonella on raw peanuts. PRP question: no. Control question: yes, roasting at a validated time and temperature. Later-step question: no, nothing after roasting kills Salmonella. This-step question: yes. CCP. The validation must be for your roaster and your load, not a figure from the internet.

Post-roast handling, hazard: recontamination. PRP question: yes. Separating raw and roasted areas, keeping the area dry and dry cleaning are prerequisite programmes, verified with environmental monitoring. Not a CCP, but a vital PRP.

Metal detection after milling and filling follows the same logic as boerewors. CCP.

Too many CCPs: why it happens and how to fix it

We regularly see HACCP plans with twelve or fifteen CCPs. A bakery in Gqeberha had CCPs for receiving, every storage area, handwashing, cleaning and pest control. The team could not monitor them all properly, so none was really controlled. Common causes:

• The hazard analysis did not separate significant hazards from minor ones.
• The PRP question was skipped, so every hygiene control became a CCP.
• The team confused “important” with “critical”.
• Someone assumed auditors or customers like more CCPs.

The fix is to redo the hazard analysis with honest likelihood and severity ratings, ask the PRP question seriously and strengthen the PRPs. A focused plan with three or four well monitored CCPs is stronger than a long list nobody checks.

Recording the decision so an auditor can follow it

An auditor does not need your answer to match theirs. They need to see how you reached it. For each significant hazard and step, record:

• The hazard, its likelihood and severity, and why it is significant.
• The answer to each question in the tree, with a short reason.
• The outcome: CCP, PRP or (under ISO 22000 and FSSC 22000) OPRP.
• The team members present and the date.

A table in your HACCP plan does this well. The HACCP (SANS 10330) and PRPs document templates toolkit includes formats you can adapt. If you want the team to apply the tree with confidence, our HACCP training for supervisors and HACCP teams builds the skills to run a hazard analysis and decide on CCPs. For hands-on help, see our HACCP and GMP implementation consulting.

What to do this week

• Ask the PRP question again for each existing CCP and decide honestly whether a PRP would do the job.
• Check that every CCP has a recorded reason, not just a tick.
• Confirm that combined controls (like biltong dipping and drying) are validated together.
• Ask for an independent internal audit to test your CCP decisions.