21 replies to this topic

[George @ Safefood 360°]

I’ve noticed in many of the forum threads the recurring theme of risk assessment. Questions such as how best to identify hazards, determine whether hazards are important or significant and where to find information on hazards? In particular microbiological hazards present the most difficulty since many of us who are charged with the responsibility of hazard analysis and HACCP planning are not microbiologists and when we find information on the subject we may not always be in the position to interpret it correctly.

The evolution of risk based food safety systems such as HACCP have played a major role in protecting public health and underpinning our efforts in a structured and scientific way. However the practical impact on those in our industry who are required to develop these systems is significant and often characterised by what we don’t know rather than what we do.

There is a well-documented Salmonella case were a food business producing salami products decided to develop a snack version which was smaller and had a higher mass to surface area ratio. The product dried out faster, water activity fell faster and acidity was incomplete. This gave rise to favourable conditions for Salmonella and a recall was required.

The above account highlights the difficulty with microbiological risk assessment (MRA) as part of HACCP. The simple act of producing a smaller version of the same product led to a microbiological hazard that previously did not exist. It also indicates the knowledge required to conduct proper hazard identification.
Much has been written on the subject by experts trying to understand why HACCP plans fail and lead to recalls. Research reviewed by Kane, Mayes & Mortimore identified poor hazard identification and hazard analysis as a significant reason. So how do we conduct microbiological risk assessment in food processing plants? How do we ensure that our risk assessments are robust enough to support our HACCP plans and reduce the chance of product failures?

In the next series of threads I will provide some practical support and resources for conducting MRA. It is not an easy topic to address yet it is an essential part of what we do as people responsible for food safety. Please contribute to the thread and add your own insight and advice.

There are 4 main steps in conducting a Microbiological Risk Assessment:

• Hazard Identification
• Hazard Characterisation
• Exposure Assessment
• Risk Characterisation

These steps have been defined by bodies such as the WHO/FAO and bearing in mind that no food processing plant has the massive resources of these bodies, we can use simple tools to conduct a good quality MRA depending the complexity of your products and processes. In the next thread we will look at the first step, Hazard Identification.

[George @ Safefood 360°]

STEP 1 - Hazard Identification

This is the first step in conducting microbiological risk assessment and perhaps the most important since if you don’t identify the hazards correctly no amount of subsequent HACCP planning will make your product safe.

The purpose of hazard identification is to identify the microorganism(s) of concern that may be present in the food you manufacture.
While on the face of it this may appear to be straight forward it can often prove to be difficult without expertise. Assuming you do not have a massive budget to employ the services of a microbiologist you will need to undertake this process yourself. Nor should you depend solely on the senior microbiologist in your external testing laboratory who may not be fully knowledgeable on all aspects of your product and processes.

The hazards of concern may come from a variety of sources including:

• Raw materials
• Methods of production
• Use of the food

In short you are looking to generate a list of potential microorganism that may have adverse impacts on human health should they be present in the food. And to do this you need Information. Remember this takes time but the effort will stand you in good stead and your HACCP will be all the better for it. There is a wealth of information available on food hazards and much of it is available free from quality sources on the internet. I have put together a list of some of the main sources for you in this excel workbook – just click the links and start your research.

 Risk Assessment Data Sources.xls   33.5KB   830 downloads

For example the Bag Bug Book provides excellent information of specific pathogens including details on associated food products and outbreaks.
Foodrisk.org contains a good database of food hazards organised according to commodities. The database can be easily searched.

Once you have compiled your list of potential microorganisms you now need to decide whether they are significant and require more detailed analysis (Step 2: Hazard Characterisation). To do this you will need to apply a series of logical questions which can come in the form of a decision tree not unlike that found in HACCP. However these questions are very specific for microorganisms and more robust than the generic ones found in the 4 question HACCP decision tree. I have attached a template of this decision tree which I have adapted from P. Vosey et al.

 Hazard Identification Decision Tree.doc   75.5KB   581 downloads

Work your way through the questions until you have developed your final list of pathogens. Don’t forget to keep a copy of all your work and reference sources for the auditor when he/she arrives.

Just one final point. The key to Hazard Identification is to review as much information and data as possible. You can do this over time and build up you own data bank of information. The more you know the better your risk assessment and resulting HACCP system will be. None of this time will be wasted.

In the next thread we will look at STEP 2 – Hazard Characterisation.

George.

[Charles.C]

Hi and Warm Welcome George,

Thks for the links and this interesting template, and particularly the series of "how to's".

Certainly not an easy challenge to analyse/segregate pathogenic micro-organisms in a template way.

Not trying to nitpick but a few comments came to mind –

(1) The (2nd) No. 5 is presumably No.6
(2) I deduce the template assumes that the process environment is not a contamination factor ?
(3) The template seems to assume that the raw material does not contain a toxinogenic pathogen at a level such that the relevant toxin may have already been produced prior to processsing. Such a situation would relate to the use of Qu6 perhaps.
(4) Qu.2 might perhaps hv acknowledged some common nasties, eg “vegetative phgs / spore formers / toxins”. (I appreciate that the intention is to offer a minimalist structure.)
(5) Qu.3 – I’m slightly confused. Surely the product is already “contaminated” by definition or perhaps the question is to a specific meaning of “contamination”, eg actual level, legislatory ?

Some of the above may hv been scheduled to be addressed in subsequent articles in which case my apologies for jumping in too quick.

“Unfectious” sounds suspiciously Irish even though Google gives a Scottish example from the 1800’s.

Best Rgds / Charles.C

[Peter Snopko]

Great discussion item. I like the simple approach and resource referencecs for a Packaging person like myself.

Is there any chance you can also make references to packaging manufacture, as this is my core focus. I have been looking for a nice Idiots guide to "Pathogen Management Risk Assessment".

Looking forward to the next installment
Cheers
Peter

[George @ Safefood 360°]

Hi and Warm Welcome George,

Thks for the links and this interesting template, and particularly the series of "how to's".

Certainly not an easy challenge to analyse/segregate pathogenic micro-organisms in a template way.

Not trying to nitpick but a few comments came to mind –

(1) The (2nd) No. 5 is presumably No.6
(2) I deduce the template assumes that the process environment is not a contamination factor ?
(3) The template seems to assume that the raw material does not contain a toxinogenic pathogen at a level such that the relevant toxin may have already been produced prior to processsing. Such a situation would relate to the use of Qu6 perhaps.
(4) Qu.2 might perhaps hv acknowledged some common nasties, eg “vegetative phgs / spore formers / toxins”. (I appreciate that the intention is to offer a minimalist structure.)
(5) Qu.3 – I’m slightly confused. Surely the product is already “contaminated” by definition or perhaps the question is to a specific meaning of “contamination”, eg actual level, legislatory ?

Some of the above may hv been scheduled to be addressed in subsequent articles in which case my apologies for jumping in too quick.

“Unfectious” sounds suspiciously Irish even though Google gives a Scottish example from the 1800’s.

Best Rgds / Charles.C

Thanks Charles,

You are of course correct. The model is not intended to be an exhaustive risk assessment but rather a simple filter of the main pathogens with a view to conducting a more detailed assessment on the main suspects. Factors such as exposure and other routes of contamination are addressed in later steps. Question 3 for example is addressing the potential of contamination after processing i.e. a lethal control step as opposed to the raw material as a source.

Feel free to adapt the model further and improve upon it. It would be great to produce a collaborative guide for members when the post is complete?

Cheers!

[George @ Safefood 360°]

Great discussion item. I like the simple approach and resource referencecs for a Packaging person like myself.

Is there any chance you can also make references to packaging manufacture, as this is my core focus. I have been looking for a nice Idiots guide to "Pathogen Management Risk Assessment".

Looking forward to the next installment
Cheers
Peter

Hi Peter

Thanks. The discussion is currently focused around the assessment of pathogenic risks in a food matrix. But I can identify with your requested. In a previous life I was responsible for quality and hazard management in a PET moulding company and always struggled to find good source data on this area. I suppose because the risks may be less or somewhat unknown this area gets less attention.

I'll do my best be make reference to packaging if possible or maybe we can come back to this as a discussion topic at a later date. I'm working on STEP 2 at the moment and hope to post it later this evening.

George

[George @ Safefood 360°]

STEP 2 – HAZARD CHARACTERISATION

Once you have identified the specific pathogens of concern the next step is hazard characterisation. This should be conducted for each pathogen identified.

In its simplest terms, hazard characterisation is an assessment of the pathogen and the nature of the problems it can cause. We are trying to answer a number of questions in order to develop the character of the hazard. These questions include;

• What is the disease caused by the pathogen?
• What are the symptoms and how long before their onset?
• What are the range and likelihood of adverse outcomes e.g. death?
• What is the minimum dose required to produce symptoms?
• Who are the main 'at risk' groups in the population?

To answer these questions you will need good sources of information. Remember, this data may not always be available so there will likely be a degree of uncertainty with your answers. This is unavoidable, however what is important is that you identify this uncertainty.

The New Zealand Food Safety Authority has produced some excellent datasheets on the main food pathogens where you will find much of the information required for hazard characterisation. The image below is part of the datasheet for E.Coli O157. By way of an example, I have highlighted some sections relevant to hazard characterisation.

 E Coli Datasheet.JPG   247.6KB   106 downloads


You can view the full list of datasheets here: New Zealand Food Safety Authority - Microbiological Data Sheets


An important concept of hazard characterisation is ‘Dose-response’. This is the minimum level of the pathogen required to be ingested to cause an adverse response e.g. 10 cells. Again, this information can be found in the above mentioned data sheets.

I have prepared a MRA model using Excel. It is based on a variety of sources and tools that I have referenced. It can be used to capture and score the required questions. The model contains all the steps required for good quality risk assessment with the final total score providing a measure of the risk. The model is easy enough to work with but any questions let me know. I will address the remaining steps of MRA in the coming days.

 MRA Profiling.xlsx   70.68KB   414 downloads
 MRA Profiling (Excel 97-2003).xls   109KB   285 downloads


George

[George @ Safefood 360°]

STEP 3 - EXPOSURE ASSESSMENT

In the previous two steps we a) identified the likely hazards of concern and b) characterised these hazards in terms of the dose and response. When conducting a microbiological risk assessment the next step is exposure assessment.

The aim of exposure assessment is to determine the level of the microorganism (or toxin) likely to present in the food at the time of consumption. Here we must take into account a number of potenital paths or routes of contamination and the impact of various processing steps on microbiologcal levels. The following are important:

• The microbiology of the raw material e.g. raw meat will have certain pathogens associated with it.
• Initial contamination levels of raw materials.
• The effects of production, processing, handling etc on the levels of pathogens in the final product.
• Sanitation standards in your processing plant.
• Potential for re-contamination after a specific control point e.g. cooking.
• Characteristics of the food being produced.
• Product usage and instructions.

Data required to conduct Exposure Assessment may be found from the following:

• Pathogen data sheets (see previous posts for references)
• In house micro testing reports and history.
• Outbreak data.
• Complaints data.
• Guides, standards and codes of practices.
• Micro modelling, challenge testing etc.

The Hazard Identification Decision Tree provided in the previous post gives you a structured approach to the above and ensures you cover all the exposure paths. You use a scoring system to reflect your assessment and the sum of your score provides you with an assessment of risk.

While a score is applied to your answer this does not mean your MRA is quantitative. It is in reality a qualitative risk assessment since only large agencies and research centres have the resources to conduct full quantitative MRA's. Also be aware that uncertainty in risk assessments is as important as what you know. Don't be scared to document any uncertainty and be open about it i.e. 'there is not enough available data to conduct a clear assessment...' Understanding the uncertainty allows you to be more cautious on certain aspects of the process.

STEP 4 - RISK CHARACTERISATION

This is the final step and is basically the total score produced at the bottom of the Excel model. It is the measure or character of the risk as assessed by you.

Microbiological Risk Assessment like HACCP and food safety management in general requires good quality data. I am currently working on a more comprehensive database of hazard reference sources from the web which I will post before the end of the week.

Please come back to the forum with any questions or clarifications you might have.

George.

[ads78]

Fantastic stuff, and very useful.

Thanks.

[Peter Snopko]

Fantastic stuff, and very useful.

Thanks.

Hello George
Did you ver get the database hazard reference sources from the web to complete this thread?

I really appreciate the effort you have put into this subject.
Cheers
Peter

[Charles.C]

Dear George,

I have prepared a MRA model using Excel.

Sadly, not recognised by Excel 2003. Any other options ?

Rgds / Charles.C

[George @ Safefood 360°]

Hi Peter

I haven't forgotten about it. We will be developing a rich Hazard Plus add on module for Safefood 360 where our users will be able to benefit from an updated hazard web reference bank. It will be used to support the HACCP planning module and assist in robust hazard analysis.

I plan to post an Excel copy of the list for non-Safefood 360 members on the forum but it will a little more time...

George.

Hello George
Did you ver get the database hazard reference sources from the web to complete this thread?

I really appreciate the effort you have put into this subject.
Cheers
Peter

[Philip @ Safefood 360°]

Dear George,


Sadly, not recognised by Excel 2003. Any other options ?

Rgds / Charles.C

Hi Charles,

Please find Excel 2003 attachment below.

 MRA Profiling (Excel 97-2003).xls   109KB   148 downloads

Note that you can also install the Microsoft Office Compatibility Pack which will allow you to open and edit the newer file formats for Excel, Word & PowerPoint. This is probably a more long-term solution for you as no doubt you will be exposed to the newer file formats again in the near future.

HTH

Philip.

[Charles.C]

Dear Philip / George,

Thks for conversion.

Looks intriguing but seems to require a key for some non-specific questions illustrating how to select the "yes-no" or levels of quantitative data to input for scoring , eg 11.1? (added - I guess that's the commercial part )

I am maybe over-cautious but I hv found that MS free facilities hv an unfortunate habit of installing other MS-friendly apps in parallel. Like Sir Adobe. I usually look for a 3rd party viewing app but was too lazy this time (never run into the xlsx before .)

Rgds / Charles.C

[ximon198657]

Hello George
Did you ver get the database hazard reference sources from the web to complete this thread?

I really appreciate the effort you have put into this subject.
Cheers
Peter

Hello george
In short thanks for great effort,it's deeply meaningful work ,especially provide the MRA template that simplify our risk analysis steps.thank u!
Cheers
ximon

[Wellsy1]

STEP 3 - EXPOSURE ASSESSMENT

In the previous two steps we a) identified the likely hazards of concern and b) characterised these hazards in terms of the dose and response. When conducting a microbiological risk assessment the next step is exposure assessment.

The aim of exposure assessment is to determine the level of the microorganism (or toxin) likely to present in the food at the time of consumption. Here we must take into account a number of potenital paths or routes of contamination and the impact of various processing steps on microbiologcal levels. The following are important:

• The microbiology of the raw material e.g. raw meat will have certain pathogens associated with it.
• Initial contamination levels of raw materials.
• The effects of production, processing, handling etc on the levels of pathogens in the final product.
• Sanitation standards in your processing plant.
• Potential for re-contamination after a specific control point e.g. cooking.
• Characteristics of the food being produced.
• Product usage and instructions.

Data required to conduct Exposure Assessment may be found from the following:

• Pathogen data sheets (see previous posts for references)
• In house micro testing reports and history.
• Outbreak data.
• Complaints data.
• Guides, standards and codes of practices.
• Micro modelling, challenge testing etc.

The Hazard Identification Decision Tree provided in the previous post gives you a structured approach to the above and ensures you cover all the exposure paths. You use a scoring system to reflect your assessment and the sum of your score provides you with an assessment of risk.

While a score is applied to your answer this does not mean your MRA is quantitative. It is in reality a qualitative risk assessment since only large agencies and research centres have the resources to conduct full quantitative MRA's. Also be aware that uncertainty in risk assessments is as important as what you know. Don't be scared to document any uncertainty and be open about it i.e. 'there is not enough available data to conduct a clear assessment...' Understanding the uncertainty allows you to be more cautious on certain aspects of the process.

STEP 4 - RISK CHARACTERISATION

This is the final step and is basically the total score produced at the bottom of the Excel model. It is the measure or character of the risk as assessed by you.

Microbiological Risk Assessment like HACCP and food safety management in general requires good quality data. I am currently working on a more comprehensive database of hazard reference sources from the web which I will post before the end of the week.

Please come back to the forum with any questions or clarifications you might have.

George.

Hi George

Thank you for this work its great however i am a little stuck on the Step 2. When you say put a score in what should it be? The prompts are between 1 - 5 so should each question be scored between 1 - 5? What if you dont know the answer to a question should you leave the score out or go for a mid range score ie 3? your help would be much appricated kind reagrds Karen

[George @ Safefood 360°]

Hi George

Thank you for this work its great however i am a little stuck on the Step 2. When you say put a score in what should it be? The prompts are between 1 - 5 so should each question be scored between 1 - 5? What if you dont know the answer to a question should you leave the score out or go for a mid range score ie 3? your help would be much appricated kind reagrds Karen

Hi Karen
Thanks for the question. You are correct. Many of the questions require a numerical answer, say 1 to 5. The value or expression for each number can be found by hovering your cursor over the cell under the 'ANSWER OPTIONS' column. A box will appear with the list of numbers and their description. These are the cells with the red triangle in the top right hand corner of the cell. Once you know the value you want, simply enter it into the SCORE cell adjacent.

For questions that are not applicable etc you simply enter a value of 0.

If you need more clarification just let me know.

George.

[cazyncymru]

Hi George

I've read the risk assessment with great interest, as i've just had to do microbiological risk assessment for my MSc. There were some very good pointers which i have utelised when designing mine.

Thanks

Caz xx

[George @ Safefood 360°]

Hi Caz


You're more than welcome. Hope the MSc goes well for you.


George.

[Charles.C]

Dear George,

This is a personal view and I apologise if somewhat tangential to the main line of the thread. I don’t wish to be too negative over MRA but it appears to me that for many people who use HACCP, the quantitative aspects of this subject are about as accessible as a black hole.

I totally agree with yr opening post on this topic that it is difficult (and unfortunately potentially dangerous) to present workable (simple) generic menus for application to specific microbiological situations. But unfortunately that is often exactly what official microbiological specifications, standards, etc are obliged to do.

There are some painful critical surveys of the validity of existing micro.standards which existed at the inception of HACCP’s application. I suspect that many of those standards may well still exist today. Unvalidatable on reasoned expert assessment but unchangeable for fear of the (political?) unknown / non-comprehendable, alternatives.

It seems to me that concepts like FSOs and the like were, a decade ago, seen as as a magic wand to bridge the gap between the QMRAs and, for example, the simplistic but intelligible Codex Tree (and its minute footnotes). But in terms of general food application, the beautiful and valuable microbiological insights involved have IMO never taken off, despite the (rare?) inspired attempts to present the information in common parlance in publications such as the Micro-organisms in Food / ILSI texts. And except where QMRA conclusions led to actual numbers which, even if their basis might be somewhat mystical, could be officially seized upon to escape from situations where the tendency to make everything “Zero tolerant” had caused biological niceties to collide with economic realities. (L.monocytogenes is perhaps the classic example.)

A (very) few locations did embody explicit FSO-type concepts in their official HACCP thinking, eg New Zealand, but I deduce from later publications that this philosophy was subsequently reversed for application of HACCP in the food industry at general manufacturing level.

Similarly, in ISO 22000, the ideas are encouraged, perhaps even promoted, to be applied for Hazard Analysis. I have yet to see a single example of detailed published usage other than the Heggum/milk series which appear to hv not been elsewhere developed in freely accessible media. In fact the tendency has been to go back to “Lowest” Common Denominators such as the Codex Tree, probably because they are relatively easy to (mostly) understand, give rapid, agreed-on answers for many standard situations and, perhaps most of all, are “official”. All very acceptable reasons, particularly from an auditorial perspective.

I applaud development of techniques such as the implementation procedure described in this thread but I fear that widespread usage will be conditional on easily understandable/demonstrable benefits as compared to existing prescriptions. And a change in official thinking.

Rgds / Charles.C

[George @ Safefood 360°]

Hi Charles.C

Many thanks for your post. I think you have opened up some very important issues regarding MRA. Issues which I believe are overdue for debate within the food safety community. You raise a number of points, so forgive me if I don’t address them in logical sequence.

In regard to the model presented in the thread (and I suspect you already appreciate this) it is important to say that it does not attempt to provide a quantitative risk assessment of microbiological hazards. A proper quantitative risk assessment requires significant resources and tools which are beyond the wherewithal of most food businesses. We will leave this for international agencies,governments and research centres. The model presented is qualitative in nature. Although it uses numbers to represent various answers the output remains qualitative and is ultimately expressed in the form of ‘High’, ‘Medium’ and ‘Low’.

There is always the danger that the development and application of structured models to specific situations may result in unsatisfactory decisions and actions – even when those models are intended to make the subject accessible to a wider community of practitioners. Your opinion on this matter is one I would share.

I would also agree with your historical perspective on QMRA. Starting in the mid-90’s the FAO/WHO published a series of documents dealing with food related risk analysis. This was driven by the General Agreement onTariffs and Trade (GATT) and the establishment of the World Trade Organisation’s (WTO) Agreement on the Application of Sanitary and Phytosanitary Measures (SPS Agreement). This required all countries who signed up to made risk analysis the basis of their local food safety laws. The more cynical members of the audience might suggest that this was driven solely from fear of countries using public health as the basis for an unofficial (and illegal) protectionist economic policy. In any event the result was to place science and risk assessment at the heart of food standards. Recognising that ‘zero tolerance’ in all cases would be as unworkable as it sounds, science would be used to establish an Appropriate Level of Protection (ALOP) for the consumer and everybody would be happy. Countries would be able to trade their goods across boarders and the consumer would be protected. This approach was adopted from that already used at the time for chemical hazards in foods and to take account of the biological nature of these hazards. Following a risk assessment, Food Safety Objectives(FSO) could be set and developed into specific process, product and storage criteria to inform and drive local HACCP systems.

It has always been my opinion that the principles set out in the various FAO/WHO documents on risk analysis and risk assessment are sound, valuable and logical. If followed they can ensure risk assessments whether conducted at a government or local level are consistent in their methodology even when the outputs are of questionable value for HACCP. However as you point out all has not turned out as planned and the example of L.monocytogenes supports this. Microbiological criteria, no matter how well intended, are useless if they cannot in reality be achieved. Situations where containers of food are detained at entry ports because they contain levels of L.monocytogenes illegal in that county but perfectly legal in the country from which they came demonstrates the difficulty.

This, of course, reflects what in my mind is the real issue.That which is ‘safe’ food and that which is ‘legal’. A recent food incident in Europe highlighted how the risk assessment wheel had eventually turned full circle. This incident concerned the removal from the market of a significant quantity of food products containing dioxin levels above that legally allowed. This legal level was determined from the application of risk assessment. However, the international authority concerned conducted a risk assessment and determined there was no measurable risk from consuming the contaminated product – yet the product was recalled anyway. It was safe but not legal. This is where I believe risk assessment has damaged its repetition. The dominance of allergen recalls in contemporary food regulation could arguably be a symptom of the same issue.

Having said all that, in my opinion QMRA’s provides excellent information and data (Hazard characterisations / Exposure assessments) required for informing HACCP studies and should be valued on that basis alone. HACCP is a risk management tool and ultimately depends on quality data where available. The model presented can provide practitioners with a structure within which to work and a tool to drive more informed decisions. It can also help to ensure that all relevant factors are at the very least considered. All prescribed models are in some way deficient by their nature and I believe this holds true for simple models like the codex tree which in my experience breaks down for certain hazards in specific situations. Our objective has to be to improve on these models. Some years ago I did some work on developing a simple model for integrating risk assessment and CCP determination. I will dig this work up and share my thoughts on it with you.

Cheers,

George

[israa99]

Dear all, greeting

 

I'm a PhD student doing microbial risk assessment for raw produce,

I read all Dr. George posts, but still I can't apply QMRA implemented with @risk to measure the risk illness estimate, 

I didn't collect all the data required by the model to measure the risk, and I used the default data, Is that OK??

any recommended source to practice the model and evaluate my results?

 

Thank you 

Regards

ISR