Hazard vs. Risk in Food Safety

I have got stuck more times than I can count when it comes to risk assessment and hazard analysis. For a long time I could fill in a hazard analysis form, but I couldn't honestly explain why I had scored something the way I did. I couldn't quite figure it out until my aha moment: when I learnt to distinguish a hazard from a risk.

Once those two words stopped meaning the same thing in my head, everything else in the hazard analysis fell into place: what to rate, how to rate it, and which hazards deserve real control. In this post, I'll take you through how I think about it. We'll look at the terms in general first, then through the food safety lens, and then I'll show you how I actually apply it in a warehouse.

So let's dive right in.

At a glance
Hazard = what could cause harm.
Risk = how likely that harm is, and how bad it would be.
The hazard stays the same. The risk moves with the controls you have in place.

Hazard: the thing that could hurt

I'm being tactful by starting here 😉

In general terms, a hazard is something with the potential to cause harm. In food safety, it's a biological, chemical, or physical agent in food, or a condition of the food, with the potential to cause an adverse health effect. Notice the word potential. A hazard is not an event. It's a possibility sitting quietly in your building until something lets it through.

In a warehouse, the three classic categories look like this:

•     Biological: Salmonella or Listeria brought in by pests, dirty equipment, contaminated returns or a delivery truck that isn't clean. Also bacterial growth when chilled product sits too long outside temperature.

•     Chemical: cleaning chemicals, pesticide residues, fuel or battery fumes, and allergens that cross over from one product to another (cross-contact).

•     Physical: broken glass from damaged jars, metal from racking or pallet fixings, wood splinters, plastic shards, even pest droppings.

 

The key point: having the potential to cause harm doesn't mean harm will occur. A sealed drum of sanitizer in a locked chemical store is still a chemical hazard. But it's very unlikely to hurt anyone. That takes me to risk.

Risk: how likely and how bad

In ISO terms, risk is the effect of uncertainty on objectives. That uncertainty can be positive (an opportunity) or negative (a threat). In food safety, we mostly care about the negative side, so the definition I actually work with is closer to the Codex one: risk is a function of the likelihood of an adverse health effect and the severity of that effect, resulting from a hazard in food.

Risk = Likelihood × Severity

Here is the comparison that made it stick for me. A lion is a hazard. A lion in a locked cage at the zoo and a lion in my living room are the same hazard, but the risk is completely different. The hazard stays the same. The risk changes with the circumstances and the controls in place.

In a warehouse, it's the same story. Broken glass is a hazard. Broken glass in a sealed, labelled bin in the damages cage is low risk. Broken glass on the floor next to open product on a busy picking aisle is a very different conversation.

Why this matters more in a warehouse

Most HACCP material is written for processing plants, where there's often a step (cooking, pasteurizing, metal detection) that deals with a hazard before the product leaves. In a warehouse, there usually isn't one. What leaves my building is what arrived, plus whatever happened to it while it was in my care.

That changes the questions I ask. In a warehouse, the hazards I worry about are rarely ones I create from scratch. They're ones I let in, let grow, let spread, or fail to catch. That's why I think about hazards along the product's journey through the building.

Where hazards show up along the flow

Before I score anything, I walk the product flow and list the hazards at each step. This is the table I now start from:

Table 1: Identification of potential hazards at every process flow 

Once I have a hazard list, I score each hazard on two things. Let me take each one on its own.

Likelihood

Likelihood is the chance that the hazard actually reaches the consumer in a way that causes harm. I try not to go on gut feeling. These are the questions I ask:

•     Has this happened here before? Check complaints, damage logs, temperature records, pest sightings, and audit findings.

•     How well do my current controls work, and do they work on the night shift and at weekends too?

•     How is the product held? Open, loose, or sealed? Ready-to-eat or cooked before eating?

•     How often is the product exposed? Once a week, or fifty times a day?

•     What do the people doing the job say? Forklift drivers and pickers know where the near misses really happen.

 Table 2: Likelihood of the hazard causing harm and the corresponding scores

Severity

Severity is how serious the harm would be if it did happen. I judge it by the health effect on the consumer, thinking about the most vulnerable person who might eat the product, such as children, the elderly, pregnant women, and people with allergies.

Table 3: The severity/impact of the hazard happening and the corresponding scores

What I've learnt: severity is mostly set by the hazard itself, so it rarely changes. Likelihood is the part my controls can change, and that is where most of my effort goes. It's also why I don't drop severity just because “it has never happened here.”

Risk levels

Multiply the two scores, and you get a rating. Group the ratings and you get risk levels. This is the matrix I work with:

Figure 1: Shows the risk matrix and risk ratings

Low (1–4): acceptable. I manage it through routine prerequisite programs (PRPs) such as cleaning, pest control, and good storage practices, and keep monitoring.

Moderate (5–9): manageable with defined controls and review regularly.

High (10–16): needs specific, monitored controls and management attention. This is where I decide between an operational PRP (OPRP) and a critical control point (CCP).

Extreme (20–25): unacceptable. I don't proceed until the risk has been brought down.

Note: This matrix is only an example. Your team sets the scales and thresholds, then documents them

Why risk levels matter

Risk levels tell me where to spend limited time and resources. If I treat every hazard as a CCP, my plan collapses under paperwork, and nobody can keep up with the monitoring. If I treat everything as low, I miss what actually matters. The rating is what separates a significant hazard, one that needs control in the HACCP plan, from one that general good practice already handles.

Rate it twice

The biggest improvement I made to my own hazard analysis was to score each hazard twice: once as if no controls existed, and once with the controls I actually have. The first score shows how serious the hazard is by nature. The second shows how well my controls are doing.

 Figure 2: Shows the significance of rating the hazard twice, first without controls and second with controls applied

Notice what the “after” score is telling me. It only holds as long as those controls are working. That's why the controls I rely on to push a risk down are the ones I need to monitor, record and verify. If the dock time limit isn't enforced, the likelihood goes straight back up and the “Moderate” was never real.

Hands-on tip: the first time I did this properly, I realised several of my “Low” scores were only Low because I assumed controls worked that I had never actually checked. Scoring twice, and asking for evidence behind the second score, made that obvious.

A simple practical routine I follow

1.     Map the flow. Draw or walk the process from the gate to the gate. Receiving, storage, picking, dispatch, returns.

2.     List hazards at each step. Biological, chemical, physical, and allergens. Include food fraud and tampering, which I treat separately through VACCP and TACCP.

3.     Score before controls. Likelihood × Severity, with the reason written next to each score.

4.     List the controls in place. Be specific. For example, “Pest control” is not a control. “Monthly contractor inspections with bait station map and trend report” is.

5.     Score after controls. Be honest about how well each control really works.

6.     Decide what's significant and how to manage it. Low and moderate hazards go under PRPs. Higher ones get a specific control: an OPRP or a CCP.

7.     Document and review. Revisit it whenever something changes.

Here is the kind of register I keep.

Table 4: Shows a simple hazard register

This leads us to the decision-making section of this whole conversation

PRP, OPRP or CCP

The rating isn't the end of the story. It tells me where each hazard should be managed. This is the rule of thumb I use. I'll go through it in detail in the next post.

•     PRP: the general conditions and practices that keep the environment hygienic, such as cleaning, pest control, maintenance and training. Most warehouse hazards live here.

•     OPRP: a specific control for a significant hazard, where I can monitor it, set action criteria, and take action if it goes wrong, but there isn't a critical limit.

•     CCP: a step that is essential to prevent or reduce a significant hazard to an acceptable level, with a measurable critical limit, and no later step to fix it. Warehouses often have few CCPs. Where product relies on temperature control, such as chilled or frozen storage, temperature is commonly treated as one.

Mistakes I made

•     Confusing a hazard with its cause. “Pests” is not the hazard. The pathogens and contamination they carry are. Naming the real hazard changes the controls I choose.

•     Copying someone else's plan. A manufacturer's hazard analysis is built for a manufacturer. Mine has to reflect my building, my products and my people.

•     Scoring on gut feeling. I now write down why I picked each score. If I can't explain it in a sentence, I don't understand it yet. And auditors will ask.

•     Scoring everything high “to be safe.” It sounds careful, but it buries the real priorities and creates monitoring nobody can sustain.

•     Ignoring existing controls, or trusting them blindly. I rate risk with the controls I have, and I check they actually work.

•     Doing it alone. When I score a hazard with the QA, operations and warehouse team in the room, the scores are more honest, and the plan is more likely to be followed.

•     Treating the assessment as one-off. I review it whenever I change a supplier, product, layout, equipment or system, such as moving from a manual process to a WMS, and after any incident, complaint or recall.

What auditors actually look for

In my experience, an auditor reading a hazard analysis is rarely checking whether I have picked the “right” numbers. They want to see that the method is defined, applied the same way every time, and that the reasoning is written down. They also want to see that the significant hazards link to controls, and that I have reviewed the analysis when something changed. A simple method, applied consistently and backed by records, beats a complicated one nobody follows.

Takeaway: A hazard is what could cause harm. A risk is how likely it is to cause harm and how bad that harm would be. Once I stopped using the two words interchangeably, hazard analysis stopped feeling like guesswork, and started feeling like a way of deciding where my attention should go.

If you're getting started, my advice is simple: walk your flow, list the hazards honestly, score them twice, write down your reasons, and involve the people who do the work every day.

In the next post, I'll take a significant hazard and walk through deciding whether it needs a PRP, an OPRP or a CCP.

Note: From experience, you are better off using Excel or another electronic version of hazard analysis than physical paper. Why: it makes updating and scoring effortless. Think about it, you don't have to print the hazard analysis table every time you make an adjustment or update. But remember: it must be easy to retrieve, and you should keep a copy. Always save a copy separately.

Have you struggled with the hazard-versus-risk distinction? Tell me in the comments.