THE JOINT FAO/WHO EXPERT MEETING ON MICROBIOLOGICAL HAZARDS IN SPICES AND DRIED AROMATIC HERBS

Transcription

1 SUMMARY REPORT THE JOINT FAO/WHO EXPERT MEETING ON MICROBIOLOGICAL HAZARDS IN SPICES AND DRIED AROMATIC HERBS 7-10 OCTOBER 2014 Preliminary Report 28 October 2014

2 Preliminary report of FAO/WHO expert meeting on spices and dried aromatic herbs Table of Contents 1. Introduction Foodborne pathogens of concern: Microbiological hazards associated with spices and dried aromatic herbs Public health risk from Salmonella associated with consumption of spices and dried aromatic herbs Spices and dried aromatic herbs of greatest concern Approach and methods Results Microbiological Criteria (MC) and their role in ensuring consumer health protection Performance of the existing Codex MC and associated sampling plan for Salmonella in the context of different contamination scenarios and different rates of sampling Consideration of other microbiological criteria for spices and dried aromatic herbs Impact of microbial reduction s for spices and dried aromatic herbs on public health Conclusions Appendix A...15 Appendix B

3 Preliminary report of FAO/WHO expert meeting on spices and dried aromatic herbs SUMMARY REPORT THE JOINT FAO/WHO EXPERT MEETING ON MICROBIOLOGICAL HAZARDS IN SPICES AND DRIED AROMATIC HERBS 7-10 OCTOBER Introduction This expert meeting was convened to respond to questions from the Codex Committee on Food Hygiene (CCFH) related to the microbiological safety of spices and dried aromatic herbs. The meeting considered the global evidence on the burden of illness, prevalence and concentration of selected microbial hazards with respect to various spices and dried aromatic herbs, and interventions aimed at controlling them in these commodities. This was based on a structured review and analysis of information in the public domain, information submitted in response to a call for data and information, data from some regulatory authorities, food recall and port of entry rejection data, data from industry and the knowledge and expertise of the participants. While tea (Camellia sinensis) was excluded from the scope, consistent with the Code of Hygienic Practice for Spices and Dried Aromatic Herbs (CAC/RCP ) 1, consideration was given to all spices and dried aromatic herbs used for culinary purposes, including for infusions/tisanes. The meeting focussed on bacterial hazards of concern. It was noted that viruses and parasites could also potentially be a concern but that there were little or no epidemiological data on these hazards or information on their occurrence in these commodities to warrant their specific consideration at this time. Although mycotoxins can be formed as result of fungal growth in foods, they were not considered as part of the evidence gathering or considered in the discussions as they fall outside the remit of CCFH. The following summarises the key outputs of the meeting in response to the specific issues raised by CCFH. 2. Foodborne pathogens of concern: Microbiological hazards associated with spices and dried aromatic herbs A structured and transparent systematic review of evidence in the public domain including an international outbreak database was conducted to identify and synthesize global evidence on the burden of illness, prevalence and concentration of selected bacterial hazards, and interventions to control them, in spices and dried aromatic herbs (Appendix A 2,). In addition, information received from the Call for Data issued in December was considered. The review identified 28 outbreaks of foodborne illness implicating spices and dried aromatic herbs, including herbal infusions, causing a total of 2,228 reported cases, 134 hospitalizations and two deaths, between 1973 and Most illnesses were caused by 1 The Code was revised in The detailed methodology of the review will be provided in the full report of the meeting. 3 Available at pdf 2

4 Preliminary report of FAO/WHO expert meeting on spices and dried aromatic herbs Salmonella spp. (77%), followed by Bacillus cereus (20%), and Clostridium perfringens (3%). Outbreaks were generally small: median 20 cases (range ). Bacillus subtilis and Bacillus pumilus were associated with two small outbreaks of foodborne illness in 1995 and 1997, with a total of 4 reported cases. Outbreaks associated with consumption of spices and dried aromatic herbs are probably underreported, as these foods are usually consumed as minor ingredients in meals. Although other bacterial hazards have been reported in spices and dried aromatic herbs (e.g. Cronobacter spp. and Staphylococcus aureus), no spice- associated outbreaks or cases of foodborne illness were identified due to these hazards. In addition, neither the review, nor the expert meeting identified any information on the presence of pathogenic Escherichia coli and Listeria monocytogenes in spices and dried aromatic herbs. In conclusion, Salmonella and the spore- forming organisms B. cereus and C. perfringens should be considered the foodborne pathogens of particular concern with respect to spices and dried aromatic herbs. 3. Public health risk from Salmonella associated with consumption of spices and dried aromatic herbs Estimating the public health risk associated with a specific pathogen in a particular food is challenging and can involve a detailed risk assessment requiring large amounts of data. Due to the diversity of the commodities categorized as spices or dried aromatic herbs and the limited data available, a full quantitative risk assessment was not considered to be feasible at this time. However, the meeting considered that under the typical conditions of production and processing of spices and dried aromatic herbs, there is a significant potential for contamination with a range of pathogens, including Salmonella spp. In some cases subsequent growth of these organisms could occur when spices and dried aromatic herbs are added to high moisture foods. Despite the expectation that some inactivation would occur during drying and storage, validated microbial reduction s are usually required for control of Salmonella spp. When used in conjunction with Good Hygiene Practices (GHP), Good Manufacturing Practices (GMP), and Hazard Analysis and Critical Control Points (HACCP), these s play a critical role in the microbiological safety of these products, particularly if spices and dried aromatic herbs are subsequently added to foods that would support growth of pathogens. When pathogens are present, the cooking, preparation, storage and consumption practices of consumers are important factors in relation to the potential risk to public health. 4. Spices and dried aromatic herbs of greatest concern 4.1. Approach and methods In general, the spice and dried aromatic herb industry is supported by millions of small farmers around the world. Given the production and processing practices employed, there is great potential, with few exceptions, for spices and dried aromatic herbs to be contaminated 3

5 Preliminary report of FAO/WHO expert meeting on spices and dried aromatic herbs with bacterial pathogens at levels that can present an unacceptable risk to public health. Together with the global nature of the industry and the extensive number of stakeholders for these products, it is difficult to characterize the risk associated with an individual spice or dried aromatic herb. To overcome this, the ranking approach developed by the expert meeting included the identification of a number of production and processing practices which strongly influence the prevalence and level of microbial contamination of spices and dried aromatic herbs by bacterial pathogens. Additionally, individual characteristics of pathogens were considered. Changes in pathogen levels in response to production, handling and processing, together with likelihood and severity of illness associated with specific pathogens, were explicitly considered in an estimation of relative risk. A deterministic mathematical model was constructed and implemented in Analytica 4 software to facilitate systematic consideration of the factors identified as relevant for risk ranking. The model estimates the likely contamination and change in contamination levels of Salmonella spp., B. cereus and C. perfringens on spices and dried aromatic herbs in the field, through harvest, processing and distribution, to end- use in meals for different, user- specified, scenarios. For this report, only the estimated contamination levels at the point of consumption are presented. The final report will however additionally document the combination of this information with approximate ID 5 50 values for each organism as a measure of likelihood of illness, as well as DALY values for each of the three pathogens to model disease severity. Thus, the model will ultimately produce a measure of risk for each pathogen for each scenario, which will serve as the basis for relative ranking. The factors considered for the present report are shown in Table 1 and are described in detail in Appendix B (Table A). Table 1: Key variables influencing prevalence and level of microbial contamination for a spice/dried aromatic herb and its relevant production/processing conditions. Factors Influencing Prevalence Description and Level of Microbial Contamination* 1. Pathogen Sources, growth and survival characteristics, resistance to heat 2. Agricultural contamination Level of contamination of the selected organism at pre- harvest considering agriculture inputs and environmental factors but not inputs from human handling 3. Pre- drying handling, storage Added contamination or growth during harvest and and transport post- harvest 4. Drying Added contamination during drying; potential for microbiological growth or inactivation; increase in concentration of contaminants due to dehydration 5. Post- drying processes Added contamination after drying; potential for microbiological growth after drying due to incomplete 4 Lumina Decision Systems, analytica/. 5 The dose of an infectious organism required to produce infection in 50 percent of the exposed individuals. 4

6 Preliminary report of FAO/WHO expert meeting on spices and dried aromatic herbs drying 6. Inactivation Whether pathogen reduction is applied; reduction achieved 7. Post- contamination Added contamination after ; potential for microbiological growth after 8. Food preparation and storage Growth or inactivation during preparation of food and prior to consumption during storage of prepared food prior to consumption * Full report (available 2015) will take account of additional factors (i.e., infectivity of organism and disease severity) in fully considering relevant public health risks. The meeting grouped spices and dried aromatic herbs by plant part, as this characteristic was considered to influence potential sources of contamination. A number of examples were selected to reflect the range of spices and dried aromatic herbs that are available on the market (Table 2). Table 2: Examples of different categories of spices and dried aromatic herbs based on plant parts and comments on hygiene challenges associated with their production. Plant Part Example Comments Pistil Crocus sativus (saffron) Unique hygienic production and process practices Inner bark Cinnamomum spp. Manual handling and processing (cinnamon) Fruit/berry Capsicum spp. (chilli/red pepper), Piper nigrum (black pepper) Water quality can be an issue Seed Leaf Cuminum cyminum (cumin), Myristica fragrans (nutmeg) Origanum vulgare, Origanum onites (oregano), Ocimum basilicum (basil) Water quality can be an issue (e.g., cumin); seed protected (e.g., nutmeg) Water quality can be an issue Rhizome Curcuma longa Direct contact with soil (root) (turmeric) Bulb Allium sativum (garlic) Direct contact with soil; Not suited to steam The expert meeting developed several example scenarios considering combinations of both best- case (well controlled) and worst- case (poorly controlled) production and processing conditions and practices for different spices and dried aromatic herbs. Such best- and worst- case scenarios were developed by the experts with detailed knowledge concerning current practices in the industry producing spices and dried aromatic herbs. The spices and dried aromatic herbs that were considered in the ranking presented in this report are: Cinnamomum spp. (cinnamon); Capsicum spp. (chilli/red pepper); Piper nigrum (black pepper); 5

7 Preliminary report of FAO/WHO expert meeting on spices and dried aromatic herbs Cuminum cyminum (cumin); Ocimum basilicum (basil); and Curcuma longa (turmeric). For the full report that is under development, a broader range of spices and dried aromatic herbs will be ranked Results Table 3 shows the results of the relative ranking using the model described in 4a. The outputs of the ranking were expressed as relative microbial change through a poorly or well- controlled process with and without a microbial inactivation, i.e. the higher the relative rank, the higher the estimated level of contamination. Since the relative ranking is primarily associated with the estimated level of contamination by pathogens, the model does not necessarily illustrate the actual level of risk due to impact on public health, severity of the illnesses, etc. However, it is often the case that the higher the estimated level of contamination of a certain pathogen, the higher the potential risk of illness by that pathogen. While the relative ranking of scenarios for each individual pathogen is probably realistic, comparison between pathogens are not yet possible due to differences in diseases severity and infectivity of the three pathogens considered. The eventual risk ranking model, which will be documented in the full report of the meeting, will, however, take into consideration the infectivity and illness severity for the three pathogens of concern. For example, the ID 50 values for Salmonella spp., B. cereus and C. perfringens will highlight the difference in infectivity between these three pathogens. The incorporation of the information on infectivity and severity will allow a comparative ranking of the pathogens. Table 3: Results of the scenario evaluation for three microbial hazards using the ranking model.* Relative C. perfringens B. cereus Salmonella spp. ranking 1 Chilli (Capsicum), poorly controlled, no Chilli (Capsicum), poorly controlled, no Chilli (Capsicum), poorly controlled, no 2 Black pepper, poorly controlled, no Turmeric, poorly controlled, no Turmeric, poorly controlled, no 3 Turmeric, poorly controlled, no Turmeric, poorly controlled, delicate product Black pepper, poorly controlled, no 4 Turmeric, poorly controlled, delicate product Black pepper, poorly controlled, no Chilli (Capsicum), well controlled, no 5 Chilli (Capsicum), well controlled, no Chilli (Capsicum), well controlled, no Cinnamon, poorly controlled, no 6 Cinnamon, poorly controlled, no Cinnamon, poorly controlled, no Cumin, poorly controlled, no 7 Cinnamon, poorly controlled, delicate product Cinnamon, poorly controlled, delicate product Turmeric, poorly controlled, delicate product 8 Cumin, poorly controlled, no Cumin, poorly controlled, no Turmeric, moderately controlled, no 9 Black pepper, poorly controlled, strong Chilli, poorly controlled, strong Chilli, poorly controlled, strong 10 Basil, moderately controlled, no Black pepper, poorly controlled, strong Cinnamon, poorly controlled, delicate product 11 Turmeric, moderately controlled, Turmeric, moderately controlled, Cinnamon, well controlled, no 6

8 Preliminary report of FAO/WHO expert meeting on spices and dried aromatic herbs no no 12 Chilli (Capsicum), poorly controlled, strong Turmeric, moderately controlled, delicate product Cumin, well controlled, no 13 Basil, moderately controlled, delicate product Basil, moderately controlled, no Black pepper, well controlled, no 14 Turmeric, moderately controlled, delicate product Basil, moderately controlled, delicate product Black pepper, poorly controlled, strong 15 Cumin, well controlled, no Cumin, well controlled, no Basil, moderately controlled, no 16 Black pepper, well controlled, no Black pepper, well controlled, no Turmeric, moderately controlled, delicate product 17 Chilli (Capsicum), well controlled, strong Chilli (Capsicum), well controlled, strong Basil, moderately controlled, delicate product 18 Cinnamon, well controlled, no Cinnamon, well controlled, no Cumin, poorly controlled, strong 19 Cumin, poorly controlled, strong Cumin, poorly controlled, strong Chilli (Capsicum), well controlled, strong 20 Cinnamon, well controlled, delicate product Cinnamon, well controlled, delicate product Cinnamon, well controlled, delicate product 21 Black pepper, well controlled, strong Black pepper, well controlled, strong Black pepper, well controlled, strong 22 Cumin, well controlled, strong Cumin, well controlled, strong Cumin, well controlled, strong *Note: 1. Poorly controlled: handling variable is poorly controlled. Other variables on drying, growth and recontamination during process depend on commodities but the categories selected based on Table A (Appendix B) are always worse than moderately controlled or well controlled within a certain commodity (e.g. gross recontamination is selected in post drying process before : contamination). 2. Well controlled: handling variable is well controlled. Other variables on drying, growth and recontamination during process depend on commodities but the categories selected based on Table A are always better than moderately controlled or poorly controlled within a certain commodity (e.g. no recontamination is selected in post drying process before : contamination). 3. Moderately controlled: This is applicable to turmeric and basil and handling variable is moderately controlled. Other variables on drying, growth and recontamination during process depend on commodities but the categories selected based on Table A are always better than poorly controlled or worse than well controlled. 4. Delicate product : equal to validated process, delicate product in the variable on microbial reduction (Table A). This is applicable to turmeric, cinnamon and basil. 5. Strong : equal to validated process, sporicidal in the variable on microbial reduction (Table A). The results above illustrate, that irrespective of the specific spice or dried aromatic herb, poor controls and practices can result in a commodity that may pose a high estimated level of contamination. Generally, the scenarios indicate that the estimated level of contamination may be reduced when microbial reduction s are applied. However, the success of a microbial inactivation in eliminating/reducing the contamination depends on efficacy and production and processing practices. Pre- pathogen levels can exceed the capabilities of process s, particularly when significant pathogen growth has occurred. Post- re- contamination or growth can potentially erase reductions achieved by a process. Ranking was conducted based on the scoring process of the current model (the estimated level of contamination). The actual scores and their implications will be further described in the full report. In addition, the expert meeting identified Saffron (Table 2) as one of the rare examples of a spice that is expected to have a very low level of contamination because of its particular handling and processing practices. This will be further examined by FAO/WHO and the result will be included in the full report. 7

9 Preliminary report of FAO/WHO expert meeting on spices and dried aromatic herbs Table 3 shows that the spice categories of root (rhizomes) and fruit/berry tend to have higher ranks (higher estimated level of contamination than the spice categories inner bark and seeds. Leaves had two scenarios with only moderately controlled based on the actual handling practices and tended to rank lower. In terms of combinations of control and, the poorly controlled + no scenario ranked highest, followed by poorly controlled + strong, well controlled + no and well controlled + strong. The trends with the categories based on plant parts reflecting similarities in production and processing are summarized below: Upper rank (higher risk of contamination): Fruit/berry: poorly controlled + no, well controlled + no Root: poorly controlled + no, poorly controlled + delicate product Middle- upper rank Inner bark: poorly controlled + no, poorly controlled + delicate product Fruit/berry: poorly controlled + strong Root: moderately controlled + no Seed: poorly controlled + no Middle- lower rank Inner bark: well controlled + no Fruit/berry: well controlled + no Root: moderately controlled + delicate product Leaf: moderately controlled + no, moderately controlled + delicate product Seed: poorly controlled + strong, well controlled + no Lower rank (lower risk of contamination) Inner bark: well controlled + delicate product Fruit/berry: well controlled + strong Seed: well controlled + strong 5. Microbiological Criteria (MC) and their role in ensuring consumer health protection The safety of foods is principally achieved through the implementation of control measures throughout the production and processing chain. Such controls are based upon the consistent application of GHPs, GMPs and HACCP, including the application of validated microbial inactivation s where appropriate. This preventive approach offers much more control than reliance on microbiological end-product testing, which is limited by the number of samples tested, the available methodology and cost and can only provide a specified degree of confidence in the result. In a modern food safety management system, microbiological testing is optimally used for assessing adherence to GMP/GHP and the suitability of a food or ingredient for its intended purpose, while end-product testing is only a final verification that the measures put in place are effective in producing safe products. 8

10 Preliminary report of FAO/WHO expert meeting on spices and dried aromatic herbs The value of testing for Salmonella spp. or any other pathogens in terms of ensuring consumer health protection is difficult to quantify. As noted above the most relevant application of testing is in the context of a preventative approach where it is one of many factors contributing to consumer health protection. However, in some cases when information on the history of a product or commodity is not available and there is the potential to have high levels of contamination, testing and rejecting positive lots of product may have a direct positive impact on consumer health. However the efficacy of testing is also closely linked to the stringency of the microbiological criteria and associated sampling plan. As average contamination levels decrease, the value of the sampling and testing as a means of removing contaminated product from the market diminishes Performance of the existing Codex MC and associated sampling plan for Salmonella in the context of different contamination scenarios and different rates of sampling. The current Codex sampling plan for spices and dried aromatic herbs (n=10 (x25g) c=0) is intended to be applied to ready- to- eat spices and dried aromatic herbs at points in the food chain after any microbial reduction has been applied and prior to use by consumers. The performance of this plan was assessed using Salmonella contamination data available for Capsicum spp. from port of entry testing in the United States of America (Van Doren et al., 2013) 6, and the FAO/WHO microbiological sampling tool ( The Operating- Characteristic (OC) curve from this sampling plan and two variations are presented in Figure 1. This illustrates the probability of acceptance of the tested lot according to the microbial load. Figure 1: OC- curve for Salmonella sampling plans with c=0 and (a) n=1, (b) n=30 and (c) n=60; sample size=25g (within lot standard deviation = 0.5). The current sampling plan (10 samples of 25g) will detect, with 95% probability for lots tested, those with approximately an (arithmetic) mean concentration of 1 Salmonella in 62g 6 Van Doren et al Prevalence, level and distribution of Salmonella in shipments of imported capsicum and sesame seed spice offered for entry to the United States: Observations and modeling results. Food Microbiology 36:

11 Preliminary report of FAO/WHO expert meeting on spices and dried aromatic herbs of product 7. The performance of the plan, in comparison to testing 30 and 60 samples and with application to 1% and 100% of lots is presented in Table 4. With the same contamination scenario, increasing the number of samples to 30 will enable the sampling plan to detect approximately 1 Salmonella in 224g of product, while increasing to 60 samples will enable the sampling plan to detect approximately 1 Salmonella in 468g of product. The OC curves for a sampling plan with 30 and 60 samples respectively are also presented in Figure 1. The figure illustrates that increasing the number of samples taken (while sample weight remains constant) increases the probability of rejecting a lot at lower levels of contamination, i.e. the OC- curve moves to the left. Table 4: Overview of the performance of a range of sampling plans for Salmonella in spices and dried aromatic herbs when 100% or 1% of lots are tested, using a contamination profile based on port of entry testing data for Capsicum 67. Number of samples Sample mass Detectable microbial load* Probability of accepting 1 cell per 1000g % lots tested % of load remaining ** % load removed ** Log removal from the overall supply % of lots rejected ** 10 25g 1 cell/62g 79% 100% g 1 cell/224g 48% 100% g 1 cell/468g 23% 100% g 1 cell/62g 79% 1% g 1 cell/224g 48% 1% g 1 cell/468g 23% 1% *Intra- lot distribution is lognormal distribution in Van Doren et al., 2013; within lot standard deviation set to 0.5 log cfu/g; Performance of sampling plan is slightly improved when the within lot standard deviation is reduced i.e. the lot is more homogeneous, performance is concentration rejected with 95% probability. ** Assumes a between lot distribution, lognormally distributed, a geometric mean of log units and a standard deviation of 1 log As illustrated in Table 4, applying such a sampling plan (n=10) to 100% of lots, with a contamination profile based on port of entry testing data for capsicum 67, would remove 86% of the contamination from the supply through the approach of removing the most contaminated lots. This would have the effect of reducing the amount of Salmonella in the supply by just under 1 log (i.e. by a factor of 10). When 30 or 60 samples are taken, 93% and 96% respectively of the contamination is removed from the supply. This reduction is entirely driven by assumptions regarding the level and variability of contamination among lots 67. However, rejection of this proportion of lots is not economically feasible from a producer/processer perspective. While this illustrates what a sampling plan can achieve, in reality it will drive the producer/processer towards achieving a lower level of contamination in order to ensure that a large percentage of lots are not rejected in this way. As noted above the level of contamination assumed here was based on port of entry data for Capsicum 67, which may or may not have been heat- treated. The mean level of contamination was assumed to have a geometric mean (or median) concentration of 3 cells per 10,000g, with a standard deviation (on the log 10 scale) of 1, and with an arithmetic average concentration of 1 cfu per 250g. In the case of one spice- attributed outbreak 7 The within lot standard deviation is assumed to be 0.5 log cfu/g. 10

12 Preliminary report of FAO/WHO expert meeting on spices and dried aromatic herbs (paprika, Capsicum sp.), in which the level of contamination in the paprika was estimated to be 2.5 MPN/g, the current sampling plan (n=10) would have rejected all lots with this level of contamination. Another scenario was considered where the contamination of the product was slightly greater than in the case of Capsicum (median concentration of 1 cell per 1000g, average of 1 cell per 80g) (Table 5). Results for this scenario emphasize that sampling is more likely to detect, and as a consequence remove, contaminated product when the mean concentration of contamination of the lots increases. Table 5: Overview of the performance of a range of sampling plans for Salmonella in dried spices and aromatic herbs when 100% of lots are tested based on an average contamination of 1cfu/1000g. Number of Sample Detectable % lots % of load % load Log % of lots samples mass microbial tested remaining** removed** removal rejected** load* from the overall supply 10 25g 1 cell/62g 100% g 1 cell/224g 100% g 1 cell/468g 100% * The within lot standard deviation is 0.5 log cfu/g; Performance of the sampling plan is slightly changed when the within lot standard deviation is reduced i.e. detection is more likely when the lot is more homogeneous ** Assumes a between lot distribution, lognormally distributed, a geometric mean of - 3 log units and a standard deviation of 1 log Several of the above scenarios assume 100% of lots of a commodity are tested. However, in reality, at least in the context of port of entry testing, sampling may only apply to, for example, approximately 1% of lots. In this case, using the same contamination scenario 56 (Table 4) only 1% of the microbial load would be rejected as a result of testing. This clearly illustrates that the role of testing in the regulatory context where only a small fraction of product is tested does not directly protect consumer health but rather serves as a means of verification of control measures applied to the lot tested and provides an incentive to producers/processors to ensure their product meets the required specifications. Some industry organizations currently test 1 sample of 75g for Salmonella. While on an individual lot basis this is not as effective at finding Salmonella as the current Codex sampling plan, the fact that it is applied to 100% of lots means that it would remove (by lot rejection) 76% of the microbial contamination from the overall supply for those organizations, again considering a contamination profile based on port of entry testing data for capsicum 56. The expert meeting also examined the probability of detecting 1 Salmonella in 1000g of product (Table 5). With the current Codex sampling plan (n=10), there is a 79% chance of accepting a lot with an average level of contamination of 1 Salmonella cell per kilogram of product. In the case of 30 and 60 samples this decreases to 48% and 23% respectively. If we consider that typical serving sizes of spice or dried aromatic herb range from 0.2 g to 2.5g per person per meal, and that a typical average serving size is 0.5 g per person, per meal, a spice lot with one Salmonella per kilo of spices would mean that one in every 2,000 servings will be contaminated. If we further assume, for simplicity of calculation, a 1 in 100 chance of illness as a result of exposure to 1 Salmonella, this equates to one illness for every 40,000 servings of spices consumed. Considering the use of the spice, i.e. whether it is consumed 11

13 Preliminary report of FAO/WHO expert meeting on spices and dried aromatic herbs directly or added to a food that is subsequently cooked, this figure will in some cases be dramatically lower. Attributing illnesses to spice consumption is complicated because these commodities are only minor ingredients in foods, single lots can be dispersed to a large number of end users, and the shelf life of spices is such that ingredients from a single batch can be added to foods anytime over a year or more Consideration of other microbiological criteria for spices and dried aromatic herbs Apart from Salmonella, the expert meeting considered the value of establishing additional microbiological criteria for spices and dried aromatic herbs. Consideration was given to B. cereus, generic E. coli, aerobic plate counts and Enterobacteriaceae. Further information about these was considered to be potentially useful to risk managers in informing the safety of these products, as they could be used to reflect the manner in which these commodities were produced or whether or not they had been subjected to a validated microbial inactivation. Based on the systematic review, B. cereus was the second most important bacterial pathogen associated with spices and dried aromatic herbs. While high levels of B. cereus can be found in spices and dried aromatic herbs, the distribution of contamination can be highly variable and in some cases the organism may only be present in very low numbers. This variability suggested that B. cereus may not be the optimal indicator of overall product safety or, when detected in low numbers, as an indicator that a spice has been subjected to a validated microbial inactivation. Nevertheless, considering the potential of this organism to grow when present in certain high moisture foods, it was considered important that the organism be present in low numbers in spices and dried aromatic herbs. Data available from one spice processer in Europe indicated, based on almost 7000 samples of a range of spices 8 in the past year, that 93% of samples had B. cereus at levels </=100 cfu/g and 99.8 of samples had levels </= 500 cfu/g (Personal communication British Pepper and Spice Co Ltd). Generic E. coli, aerobic plate counts or Enterobacteriaceae were considered as possible indicators of process hygiene, particularly in the case of those commodities that cannot be subject to certain microbial inactivation s (e.g., garlic, onions). One of the challenges of establishing microbial limits for these indicators is a good understanding of the levels present in spices and dried aromatic herbs before and after inactivation s. Very little of such data are available in the public domain; however, the industry indicated that they do collect such data, for both heat- treated spices and spices which cannot be heat- treated. Data available from two spice processors in Europe indicated similar results, based on sampling a range of spices in the past year. One processer reported that 92% of samples 7 had </=10 cfu/g coliforms (n=5747) and over 99% of samples had </=10 cfu/g E. coli (n=6556) (Personal communication British Pepper and Spice Co Ltd), with another processer 8 Heat- treated cumin, heat- treated black pepper, heat- treated oregano, heat- treated basil, heat- treated turmeric, heat- treated ground nutmeg, whole nutmeg (not heat- treated), garlic powder (not heat- treated), onion powder (not heat- treated), cinnamon bark (not heat- treated) 12

14 Preliminary report of FAO/WHO expert meeting on spices and dried aromatic herbs reporting 91% of samples 9 had </=10 cfu/g coliforms (n=103), 100% of samples 8 had </=10 cfu/g E. coli (n=104). With regard to aerobic plate counts, data from industry indicated 35% of 6,485 samples had </= 1,000 cfu/g, 36% of samples had 1,001-10,000 cfu/g and 27% of samples had 10, ,000 cfu/g. Data on total viable count at 30 C indicated that 48% of samples (n=98) had </= 100 cfu/g, 20% of samples had 101-1,000 cfu/g, 9% of samples had 1,001-10,000 cfu/g, 21% of samples had 10, ,000 cfu/g and 1% had 100,001-1,000,000. (Personal communication Drogheria & Alimentari Spa). The expert meeting considered that it could be useful to provide guidance on potential limits for one or more of these organisms as an indicator of either implementation of good practices or whether an adequate inactivation had been applied where appropriate. However, further consideration of the available data, as well as data from the broader spice and dried aromatic herb industry, was considered necessary before any specific guidance on potential limits could be provided. 6. Impact of microbial reduction s for spices and dried aromatic herbs on public health Contamination of untreated spices and dried aromatic herbs with Salmonella is not uncommon. The prevalence and levels found will depend strongly on the methods of production, harvest, and processing of these commodities. Much of the international spice and dried aromatic herb industry relies on GHPs, GMPs, and HACCP, including microbial inactivation s, for product safety. Given the diversity of products and processes involved, it is not possible to generalize the impact of different levels of microbial reduction s on public health. 7. Conclusions Under typical conditions of production and processing of spices and dried aromatic herbs, the meeting considered that there is a significant potential for contamination with a range of pathogens, including Salmonella spp. With certain uses of spices there is potential for subsequent growth of these organisms to occur. Given the global nature of the spice and dried aromatic herb industry and the diverse ways in which spices and dried aromatic herbs are produced, processed and used, it is difficult to characterize the risk associated with an individual spice or dried aromatic herb. Instead, the approach developed by the expert meeting included the identification of a number of production and processing practices which strongly influence the estimated level of contamination in spices and dried aromatic herbs. This estimated contamination information was used as the basis for the ranking presented in this summary report. The final report will further consider the probability and severity of infection in the ranking production and processing scenarios for a range of spices and dried aromatic herbs. 9 Treated oregano, treated black pepper, delicate treated black pepper, treated chilli, garlic (only heat treated during the production process) 13

15 Preliminary report of FAO/WHO expert meeting on spices and dried aromatic herbs The expert meeting grouped spices and dried aromatic herbs into a series of categories based on plant parts reflecting similarities in production and processing. The ranking approach considered the microbial pathogens Salmonella spp., B. cereus and C. perfringens, the full range of production and processing practices that might affect the level of contamination, and a set of representative spices and dried aromatic herbs that illustrate the diversity of these commodities. The scenarios examined highlighted the importance of production and processing methods on potential contamination and growth of pathogens in spices and dried aromatic herbs. These scenarios also illustrate the importance of mitigation and control measures throughout the supply chain, and demonstrate that, while microbial inactivation s are critical, these s must be combined with other mitigations and controls as part of GHPs, GMPs and HACCP to ensure the safety of spices and dried aromatic herbs. In considering microbiological criteria for spices, the meeting noted that the microbiological safety of food is principally achieved through the implementation of control measures throughout the production and processing chain. The performance of the existing Codex sampling plan for Salmonella was assessed against several contamination scenarios. While the sampling plan was seen to have some value when all lots are tested, this value was dramatically reduced when only a small portion of lots are tested. The meeting positively considered the establishment of an indicator microbiological criterion which would reflect production and processing conditions to which a particular commodity had been exposed but at the time of the meeting there were insufficient data to consider appropriate limits. Despite the expectation that some inactivation is expected to occur during drying and storage, validated microbial reduction s are usually required for control of Salmonella. When used in conjunction with GHPs, GMPs, and HACCP, these s play a critical role in the safety of these products, particularly if spices and dried aromatic herbs are subsequently added to foods that would support the growth of pathogens. 14

16 Appendix A: Preliminary report of FAO/WHO expert meeting on spices and dried aromatic herbs Summary Card: Spices, Dried Herbs and Tea (Burden of Illness, Prevalence and Interventions) Low-moisture food category description Spices are dried parts of fruits, seeds, bark, roots, leaves, or flowers of plants and herbs (EFSA, 2013; US FDA, 2013). They are often ground, crushed, or otherwise processed and used for seasoning, flavouring, and/or preserving foods (EFSA, 2013; US FDA, 2013). For the purposes of this summary, and due to their similar nature, spices (including dried herbs) have been combined with tea an aromatic beverage prepared by mixing hot water with dried leaves of the tea plant and/or other dried herbs such as chamomile. To facilitate summary and interpretation of this large area of research, spices have been grouped into hierarchical categories based primarily on the part of the plant from which they originated (Sagoo et al., 2009; US FDA, 2013; Van Doren et al., 2013a). Categories were also created for mixed/unspecified spices and dried herbs, and for tea (Appendix G: Spice Classification Table). Evidence summary In total, 129 articles 1 and outbreak reports 2 were identified that investigated the burden of illness related to spices, the prevalence or contamination of selected microbial hazards in spices, and/or interventions to reduce contamination of microbial hazards in spices. The distribution of identified research stratified by microbial hazard investigated and research focus is shown in Appendix F: Summary Card Evidence Charts. Salmonella spp. was the most frequently investigated microbial hazard in spices for burden of illness (n=13 articles and outbreak reports), prevalence (n=42 articles), and intervention (n=12 articles) information. Burden of illness Burden of illness evidence related to spices includes 28 reported outbreaks and non- outbreak burden of illness information in 1 cohort study and 2 case- control studies. Outbreaks affected 2228 individuals, including 134 hospitalizations and 2 deaths between 1973 and Outbreaks were generally small: median 20 (range ); however, they can be very large. Spice outbreaks, shown in the summary table below, were reported from Demark (9), the United States (4), Finland (3), the United Kingdom (2), Germany, Norway, Canada, France, Hungary and Belgium. Several outbreaks occurred where the spice was added to the food product after the final pathogen reduction step. Spice outbreaks are likely significantly under- reported as they are usually consumed in mixed ingredient foods and in small amounts. Salmonella spp. accounted for 77% of illnesses associated with spices > B. cereus 19.7% > C. perferingens 2.8% > > C. botulinum 0.04%. A case- control study examining source association with Salmonella Enteritidis cases (n=719) in Germany found the consumption of dried herbs was associated with infection; OR 1.4 (95% CI: ) (Ziehm et al., 2013). Ten of the 28 outbreaks ( ) implicated black or white pepper as the contaminated ingredient. Other spices were implicated in 1 or 2 outbreaks each. 1 Articles refer to peer- reviewed journal publications as well as government and research agency reports. 2 For burden of illness information, multiple articles often reported complementary and/or overlapping information on the same outbreak. In addition, outbreak data were supplemented from other literature sources, including line lists from various countries, news reports, or annual summaries of country outbreaks. Thus, to avoid counting the same outbreak more than once, the term outbreak report is used instead of article to count the total number of unique outbreaks. 15

17 Appendix A: Preliminary report of FAO/WHO expert meeting on spices and dried aromatic herbs All outbreaks associated with tea were in infants less than 18 months old in Germany, Serbia and Portugal and are detailed in the summary table below. One case- control study implicated tea in association with B. cereus infection in child cancer patients (El Saleeby et al., 2004). In contrast, a cohort study of Mexican infants from 0-1 year old (n=98) found that herbal tea was protective against diarrhea; hazard ratio 0.11 (95% CI: to 0.62) (Long et al., 1994). Summary table of globally reported outbreaks on spices Spice category/ Country (year) b specific spice (source) Bark/flowers Cinnamon (EU, no date) Root Turmeric (EFSA, 2013) Fruit/seed Cumin (EFSA, 2013) Capsicum spp. Dried chilies (EU, no date) Red Pepper (EU, no date) Paprika (Anon., no date) (Lehmacher, 1995) Piper nigrum Black pepper (EU, no date; EU, 2012a) (EFSA, 2013; Van Doren, 2013b) (Gieraltowski, 2013; Gustavsen, 1984; Little, 2003; Van Doren, 2013b) Microbial hazard(s) Outbreaks/ cases/ hospitalized/ deaths a Comments: susceptible populations/ attack rate/ concentration of microbial hazard in the product B. cereus 1/30 c /0/0 Denmark (2011) Concentration: 5000 organisms/g. B. cereus 2/23 c /0/0 Finland (2011) B. cereus C. perfringens Salmonella Caracas 1/3 c /0/0 Finland (2011) Concentration: B. cereus CFU/g, C. perfringens 180 CFU/g and S. Caracas presence/25 g. C. perfringens 1/3 c /0/0 Denmark (2011) C. perfringens 1/37 c /0/0 Denmark (2011) B. cereus 1/48 c /0/0 Denmark (2009) Salmonella Saintpaul, Rubislaw, Javiana (94 serovars isolated) 1/1000 c /0/0 Germany (1993) Implicated paprika on potato chips. Attack rate= 1/1000. Mostly affected children <14 years old. Concentrations: chips MPN/g; paprika 2.5 MPN/g; spice mixture MPN/g. C. perfringens 2/19 c /0/0 Denmark (2011) Concentration 330 mill. / g of pepper. B. cereus 2/164 c /0/0 Denmark (2010 E Salmonella Weltevreden, Oranienburg, Enteritidis PT4, Montevideo, Seftenberg & Rissen & 2011) 6/521 c /94/2 Canada (1973), Norway (1981), United Kingdom (1996), United States (2009, 2009, 2008) Black pepper originated from India, Brazil [0.1 to >2.4 MPN/g], Vietnam & China. White pepper from Vietnam. Red pepper from India implicated in 2 outbreaks with black pepper. Mixed spices Garlic salt & black pepper mix (Raevuori, 1976) BBQ spices (EU, no date) Seasoning mix (Sotir, 2009) B. cereus 1/18 c /0/0 Finland (1975) Attack rate 50%, Concentration: garlic salt 100 organisms/g, white pepper 4500 organisms/g. C. perfringens 1/4 c /0/0 Denmark (2011) Salmonella Wandsworth & Typhimurium 1/87 c /8/0 United States (2007) Seasoning applied to commercial puffed vegetable coated ready- to- eat snack after final pathogen reduction step. 16

18 Appendix A: Preliminary report of FAO/WHO expert meeting on spices and dried aromatic herbs Spice blend (Van Doren, 2013b) B. cereus 1/146 c /0/0 France (2007) Outbreak in school children. (EU 2012b) Salmonella Enteritidis 1/41/6/0 Hungary (2012) EU category of herbs and spices. Curry powder Salmonella 1/20 c /1/0 United Kingdom (Van Doren, 2013b) Braenderup (2002) (EU, 2010) B. cereus 1/7 c /0/0 Belgium (2009) a Superscipt C indicates confirmed cases, p indicates presumptive cases. b Superscript E indicates the link between human cases and implicated product was epidemiological only, otherwise the link was laboratory confirmed. Spice originated from India. Summary of globally reported outbreaks related to tea Tea category/ specific tea Tea Chamomile tea (Saraiva, 2012) Anise seed in tea (Koch, 2005) Fennel seed in tea (Ilic, 2010) Microbial hazard(s) Outbreaks/ cases a / hospitalized/ deaths a Superscript C indicates confirmed cases, p indicates presumptive cases. b Superscript E indicates the link between human cases and implicated product was epidemiological only, otherwise the link was laboratory confirmed. Country (year) b Comments: susceptible populations/ attack rate/ concentration of microbial hazard in the product C. botulinum 1/1 c /0/0 Portugal (2009) Case of infant botulism, both honey and chamomile tested positive. Salmonella 1/42 c /21/0 Germany (2002) Cases, infants <13 months. Anise seed (Pimpinella anisum) from Turkey. Concentration: MPN/g. Salmonella 1/14 c /4/0 Serbia (2007) Cases, infants <12 months. Fennel seed (Foeniculum vulgare) Prevalence A total of 77 studies containing 1,275 unique trials were identified that investigated the prevalence and/or concentration of one or more selected microbial hazards in spices. The median publication year was 2009 (range ). Most studies (>69%) were conducted in Europe (n=32) and Asia/the Middle East (n=21). Most studies (84%) sampled products during a specific or defined period of time, while 2 conducted sampling over multiple time points, and 10 reported on the results of systematic surveillance programmes. Studies primarily sampled products at retail (e.g. markets, grocery stores) and/or from manufacturing plants (75%). Only 8 studies specified the country(s) of product origin, while 12 studies sampled products produced in the country where the study was conducted. Salmonella spp. was the most commonly investigated microbial hazard across most spice categories. Both Salmonella and S. aureus were infrequently isolated from most trials; in many cases only one or a few trials found positive results for these pathogens. However, the prevalence estimates and ranges shown in the summary table indicate the potential for high contamination if appropriate good production and manufacturing practices are not followed (ASTA, 2011; US FDA, 2013). A summary of United States FDA spice recalls ( ) recorded 17 recalls all due to Salmonella contamination in spices and dried herbs (Vij et al., 2006). Generic E. coli was also infrequently found in prevalence trials except in the mixed/unspecified spice category, where it was found in 75% of trials with a median prevalence of 11% and range of 0-33%. 17