Previous Page Table of Contents Next Page

4. Risk Ranger

4.1 Background to developing Risk Ranger

If you are a risk manager you need to be able to compare and prioritize risks. There are a number of decision support tools that will guide you on whether a pathogen might be an important hazard in a given food or food process combination. These include various semi-quantitative scoring systems such as those by Corlett and Pierson (1992), shown in Table 15 and by Huss, Reilly and Ben Embarek (2000), which is illustrated in Table 16.

While the above approaches are able to categorize risk and direct broad mitigation strategies, neither can be used to assess an as yet undocumented risk, or to measure the effect of contributions to risk of individual factors. These schemes do not focus on the steps or variables where control could most effectively be applied.

Risk Ranger is a simple and accessible food safety risk calculation tool intended to help determine relative risks from various product/pathogen/processing combinations and is presented in Microsoft® Excel spreadsheet software. In particular, it is intended to make the techniques of food safety risk assessment more accessible to non-expert users, and to users with limited resources, both as a decision-aid and an educational tool.

Risk Ranger incorporates all factors that affect the risk from a hazard in a particular commodity including:

A number of factors affect each of the above.

Disease severity is affected by:

Hazard classification of Corlett and Pierson (1992)


Risk characteristics


Special class restricted for at-risk populations, e.g. the aged, immunocompromised, infants


Product contains sensitive ingredients


Process has no step which destroys sensitive organisms


Product is subject to recontamination between processing and packaging


Potential for abuse by distributor or consumer, which could render the product hazardous


Product is consumed without further process to kill micro-organisms

Qualitative risk assessment based on the process of Huss, Reilly and Ben Embarek (2000)

Risk criteria

Raw molluscan shellfish

Canned fish

Dried fish

Bad safety record




No critical control point for the hazard




Possibility of contamination or recontamination




Abusive handling possible




Growth of pathogens can occur




No terminal heating step




Risk category



No risk

The Risk Ranger shop front

Exposure to the food will depend on how much is consumed by the population of interest, how frequently they consume the food and the size of the population exposed.

Probability of exposure to an infectious dose will depend on:

4.2 User interface - the Risk Ranger shop front

Risk Ranger has a "shop front" with a series of list boxes into which you enter information using your computer's mouse. In total, you need to answer 11 questions, and a mathematical model then converts each answer to a numerical value or 'weighting'. The weightings are detailed in the paper by Ross and Sumner (2002). Some of the weightings are arbitrary, while others are based on known mathematical relationships, e.g. from days to weeks, or years. To help you make your responses as objective as possible, and to maintain transparency of the model, descriptions are provided and many of the weighting factors are specified. As well, in some cases, if the options provided do not accurately reflect the situation being modelled, you can enter a numerical value by using the "Other".

Behind the shop front is the model, developed in Microsoft® Excel software, using standard mathematical and logical functions. The list box macro tool is used to automate much of the conversion from qualitative inputs to quantities for calculations. For each selection you make from the range of options, the software converts that selection into a numerical value.

4.3 How to use Risk Ranger

To help you understand how Risk Ranger works, let us cover each of the 11 questions in turn and explain the scientific background behind each of them.

Question 1: Hazard severity

You are offered four choices, based on the severity of the symptoms caused by the hazard. In the Table 17 are our ideas on how seafood hazards fit into the descriptions.

If you click on the coding tab at the bottom left side of Risk Ranger you will switch to the codings for each question. You will see there is a ten-times difference in severity between each category of hazard. This is an arbitrary difference.

You may not agree with the descriptions and the way hazards are linked with them in Table 17. For example, you might say that most cases of L. monocytogenes, enterohaemorrhagic Escherichia coli (EHEC) and V. vulnificus do not require medical intervention, and it is only for susceptible groups that the description is true. Similarly, in some cases, Salmonella can be a serious infection with long-lasting consequences such as reactive arthritis. But Mead et al. (1999) state that, in the United States, there are probably 38 times more cases of salmonellosis than those that are reported, so for most people Salmonella infection obviously resolves itself without entering the medical system. By contrast, at least 50 percent of listeriosis cases are reported. And this ratio is probably higher for EHEC.

Associating hazards with Risk Ranger descriptions at Question 1


Consequences of the hazard



Death in most cases

Tetrodotoxin, Botulinum toxin


Most cases require medical treatment

Listeria monocytogenes, Vibrio vulnificus, Vibrio cholerae, EHEC


Sometimes medical treatment is needed

Vibrio parahaemolyticus, Hepatitis A, Norwalk-like viruses, histamine, ciguatera, algal biotoxins, Salmonella


Medical treatment rarely required

Staphylococcus aureus, Clostridium perfringens

This is one limitation of Risk Ranger and, if you believe a description is wrong for the specific country or system you are working on, then by all means move the hazard to the category of your choice.

Question 2: Susceptibility of the population in which you are interested

Risk Ranger allows you to select one of four populations that vary in their level of susceptibility. Groups that are slightly (five times) more susceptible than the general population to food-borne hazards are small children (1-5 years old) and people over 65 years old. In the "very susceptible" category are newborn babies, children under one year and people with conditions such as diabetes, cancer and liver damage, which predispose them to infectious diseases. They are rated 30 times more susceptible than the general population. People with AIDS or who are recovering from transplant surgery have very impaired immune systems, which place them in the "extremely susceptible" category, 200 times more likely to succumb to hazards than the general population. The various weightings, 5x, 30x and 200x, are loosely based on the relative susceptibility of each population subgroup to Listeria monocytogenes. Consequently, they may give unexpected results if applied to hazards that all people are more or less equally susceptible to, for example, S. aureus enterotoxin. If you want more details of the reasons for these weightings, see Ross and Sumner (2002).

When you select this subpopulation, Risk Ranger automatically makes changes in two other questions:

Absolute risk is based on the population size, the proportion of the population consuming the food and how frequently people eat the food, and this information is selected in Questions 3-5.

Question 3: Frequency of consumption

Obviously, the more often we face a hazard, the more likely we are to be affected by it, and this question reflects the popularity of a seafood product. The selections you can make are set in absolute terms, based on annual consumption, and this is obvious from the coding used.

Question 4: Proportion of population consuming the product

The proportion consuming the product is set for all (100 percent), most (75 percent), some (25 percent) and a few (5 percent) of the population.

It is best to link your selections for Questions 3 and 4, such as "Everyone eats the product daily", which might apply to consumption of reef fish by the population living on a Pacific atoll. By contrast you may select "Some people (25 percent) eat the product weekly", which might apply to oyster consumption in a European country.

You can answer Questions 3 and 4 using either of two methods:

Question 5: Size of consuming population

Risk Ranger has several country populations already programmed into Question 5 and, if you want to select another country just select "Other" in the list box, and type the population of that country in the "Other" box. Alternately, if you want to make the list box specific, click the tab for CODINGS and you will see instructions on how to put in your own populations.

If you select a subpopulation from the general population in Question 2, Risk Ranger automatically estimates the number in that category. Because Risk Ranger was developed in Australia, the proportions refer to that country, and they also fit many other countries with similar lifestyles, particularly in North America and Europe. However, you may need to recalibrate the coding for this question for countries in which certain diseases are rampant, e.g. for countries with a high prevalence of AIDS.

Question 6: Probability that a serving of raw product is contaminated

To answer this question you obviously need data. For example, if you were considering viruses in oysters, it is important to know how many servings have sufficient viruses to infect you. Similarly, you may want to know how prevalent is a bacterial pathogen, such as Salmonella, in raw shrimp. If you have data from a properly designed survey you can insert the exact level by selecting "Other" in the list box, then typing the percentage in the box below. Alternatively you may not have an accurate idea on proportion contaminated and, in this case, you can select the most appropriate category in the list box.

Question 7: Effect of processing

To answer this question you need to know about the process and how it affects the hazard. Here are some examples:

Question 8: Potential for recontamination after processing

Recontamination is particularly important for those products that have received heat treatment during the process. Such products have low bacterial levels and any contaminant will be able to grow with little competition. Examples of where recontamination is important include:

To answer Question 8 you really need data generated from surveys, and this can be typed in the "Other" box. If you do not have data on recontamination you can make an assumption based on observation or on comparison with similar processes that have been surveyed in countries with conditions similar to your own. For example, if you observe operators peeling shrimp with their bare hands you can predict that up to 50 percent of the product will be contaminated, because 30-50 percent of food handlers carry S. aureus on their hands and nose.

Question 9: How effective is post-processing control?

In this question you need to know how the product is handled during storage, distribution and retailing. Also you need to know something about the hazard and how it responds to those controls. Here are some examples:

Question 10: What level of increase is needed to cause illness?

To answer this question you need to know something about the amount of the hazard that would be required to cause illness. Table 18 presents some data on the number of organisms it takes to make a healthy person ill. The numbers are given for a 100 g serving, so the count/g of food is 100x lower. It is with great trepidation that these numbers are presented here because not every microbiologist will agree with them. These numbers should be regarded as guidelines, and do not forget that vulnerable consumers require much lower doses to make them ill. And if you believe a number is wrong and have good evidence, please use your own data at Question 10.

Levels of pathogenic bacteria that are likely to cause illness in healthy people


Infective dose in a 100 g serving


10 000 000


10 000 000 000

Viruses (Hepatitis A, Norwalk)


Enterohaemorrhagic E. coli, Shigella

1 000

Staphylococcus aureus

100 000 000

Knowing the number of micro-organisms surviving after processing, divide this number into the number required to cause illness (from Table 18 or more updated data) and you have the answer to Question 10.

For example, if you are considering Hepatitis A contamination in oysters, then select "None", because the infective dose is already contained in the serving. The same answer is selected for ciguatera, since the toxin is already present at processing.

If you are considering L. monocytogenes in smoked seafood, you probably will not know the contamination level after processing and recontamination. The literature tells us that contamination level will probably not exceed 10 g, so if we consume 100 g there are 1 000 cells in our serving just after processing. If we assume that we need around 10 000 000 000 to make us ill, the increase to infective dose is 10 000 000-fold and we can enter that in the "Other" box at Question 10.

Question 11: Effect of meal preparation

This question considers the form of cooking and preparation for cooking. Here are some examples of how you can answer Question 11:

4.4 Risk estimates

Risk Ranger combines the factors in Questions 1-11, including some logical tests to generate three estimates of risk:

Full details of the logic and equations leading to the risk estimates are shown in the paper by Ross and Sumner (2002), which is included in the Resources Bank.

Risk ranking

The risk ranking value is scaled logarithmically between 0 and 100. The former (Risk Ranger = 0) is equated to a probability of food-borne illness of less than, or equal to, one case per 10 billion people (greater than current global population) per 100 years. At the upper limit (risk ranking = 100), every member of the population eats a meal that contains a lethal dose of the hazard every day. A risk ranking change of 6 corresponds to a tenfold difference in the absolute risk. Thus an increase in risk ranking from 36 to 48 means that the risk increases 100-times.

Predicted annual illness

Risk Ranger estimates the total number of illnesses in the population you select at Question 5. Obviously, the higher the risk ranking, the greater the proportion of the population that will become ill. The absolute numbers of illnesses, however, will depend on the population size.

Probability of illness per day in target population

Risk Ranger targets the proportion of the population that you select at Question 2. Risk ranking remains the same, irrespective of whether you are considering the general population, or a highly susceptible subpopulation. But the probability of illness increases in the target population. This output tells you where illness is focused.

4.5 Evaluating risk ranger

To evaluate the performance of the tool, the authors modelled several scenarios and compared the results with actual data or with other risk assessments. In the first evaluation, conditions leading to an outbreak of Hepatitis A from consumption of oysters in Australia in 1997 were simulated using the model and compared with the epidemiological data reported by Conaty et al. (2000). In the second evaluation, the data and assumptions of the quantitative risk assessment of Cassin et al. (1998) for the risk of illness from enterohaemorrhagic E. coli due to consumption of hamburgers in North American culture were used to derive answers to the questions of the risk assessment spreadsheet. The results of both assessments were compared and are detailed in Ross and Sumner (2002). In general, Risk Ranger predicted illness of the same order of magnitude as in the actual events.

4.6 How Risk Ranger can be used

Risk ranger was originally developed as a means of quickly assessing the performance of various conceptual models for food safety risk assessment. However, it is also a useful tool for risk assessment and risk communication. It can be used by risk managers and others without extensive experience in risk modelling:

Tools such as this can help managers to think about how risks arise and change and to help to decide where interventions might be applied with success.

4.7 Let us work through some examples

Working through some examples together will help you see how Risk Ranger is used.

Let us use consumption of chilled, hot-smoked salmon contaminated with L. monocytogenes in the United States. For more than ten years the United States risk management strategy has been to operate zero tolerance for this organism in this product (in a 25/g sample), reflecting the seriousness that the risk manager (Food and Drug Administration) ascribes to this hazard: product pairing.

We can summarize the inputs in Table 19 and you should open Risk Ranger and work through the inputs for the general population. You will see that the Risk Ranking is 41 with predicted illnesses of 400 per annum from the United States population of 270 million. The probability of illness per day per consumer in the general population is 8-8.

Important questions and assumptions are:

Question 7, where we assume that hot smoking eliminates all L. monocytogenes.

Question 8, where we assume first that 1 percent of servings are recontaminated and second that the level of recontamination is 0.1 cell/g (10 cells/serving).

Question 9, where we assume the shelf-life at around 5 °C is long (several weeks in the distribution and retailing chain) and the population increases to 100/g (10 000/serving).

Question 10, where we assume that the infective dose for the general population is 10 000 000/g (10 000 000 000/serving). So for Question 10, you need to select "Other" and insert 1 000 000 for the increase needed for an infective dose.

Now let us consider the effect of contaminated smoked salmon on a more susceptible subpopulation, the very young and very old. When you select this subpopulation, Risk Ranger automatically makes changes in two other questions:

The consumption frequency remains the same, as do the contamination levels in raw product and the effect of processing, recontamination and post-process control. Risk Ranking remains the same as for the general population; the number of illnesses falls slightly to around 350 per annum and the probability of illness per day in this vulnerable group is 2-6. The latter two outputs tell you that the vulnerable subpopulation bears almost all the risk of illness with 75 percent of all listerioses and much higher probability of illness on any one day.

Reality check

Let us do a reality check on listeriosis in the United States to see if the annual illnesses we are predicting from Risk Ranger are approximately correct. Statistics indicate around 1 000 cases notified each year (4 cases/million population), which, when we take into account under-reporting, extends to around 2 000 cases. Risk Ranger predicts around 400 annual cases due to the consumption of smoked salmon, a prediction which is in an acceptable range.


The main reason for doing this exercise is to help you understand how Risk Ranger automatically selects subpopulations for you, plus how important it is to understand how to calculate the increase to infectious dose (Question 10). The more you understand a food process and the behaviour of the target pathogen, the better outputs you will get from Risk Ranger.

Risk Ranger inputs for L. monocytogenes in chilled, hot-smoked salmon in the United States

Risk criteria

General population

Very young and old

Dose and severity

Hazard severity

Moderate - usually requires medical attention

Moderate - usually requires medical attention


General - all population

Very young and very old

Probability of exposure

Frequency of consumption

Few times a year

Few times a year

Proportion consuming

Few (5%)

Few (5%)

Size of population

270 million

ca 50 million

Probability of contamination

Probability of raw product contaminated



Effect of processing

Hot smoking eliminates all contaminants

Hot smoking eliminates all contaminants

Possibility of recontamination

Minor (1%)

Minor (1%)

Post-process control

Poor (>3 log increase)

Poor (>3 log increase)

Increase to infective dose

1 000 000x

1 000 000x

Further cooking before eating

Not effective in reducing hazard

Not effective in reducing hazard

Predicted annual illness in the population considered



Probability of illness per day per consumer of interest

8.22 x 10-8

2.47 x 10-6

Risk ranking (0-100)



Previous Page Top of Page Next Page