Quantitative Risk Assessment of Vibrio parahaemolyticus Toxi Infection Associated with the Consumption of Roasted Shrimp (Penaeus monodon)

In this study, a risk assessment on Vibrio parahaemolyticus infections was carried out in order to estimate the likelihood of gastroenteritis for Cameroonians after consumption of roasted shrimp (Penaeus monodon).*e Codex Alimentarius Commission framework was used in this study. Based on the distribution of total V. parahaemolyticus in shrimp and literature information indicating that nonhaemolysing carrier strains could be pathogenic to humans, the cooking, and consumption patterns, the daily exposure level generated in this study, and the dose-response model from other studies, the infectious risk was evaluated and quantified by the Monte Carlo simulation. *is simulation was realized based on 10,000 iterations using the Model Risk software, version 4.0, in combination with Microsoft Excel. To better quantify the exposure of consumers and the resulting risk of infection, several scenarios reflecting the minimal, average, and maximal exposures were undertaken. According to the results, the 90% confidence intervals for minimum and maximum exposures ranged from 15 to 24 colony-forming units per day (cells/day) and from 160 to 228 cells/day, respectively. Based on the modal scenario, 90% of the population consuming this shrimp is exposed to V. parahaemolyticus loads ranging from 74 to 110 cells/day, indicating a risk of infection ranging from 1.2 to 1.8 cases per million of consumption.*e estimated number of annual disease cases based on annual production is between 1 and 10 cases.*is reflects a relatively low risk of infection for roasted shrimp. Good hygiene practices during handling, cooking, and storage may help reduce the actual risk.


Introduction
For almost three decades, food-borne infections have been the center of concern worldwide. Europe is experiencing a decrease of infections while most underdeveloped African countries are still experiencing an increase. Raw or undercooked foods are generally the foods most implicated in food-borne illness. Among these foods, fish products (including shellfish), which are the most consumed, are the second most important source of food protein after meat [1]. Indeed, in 2016, its global production was estimated at about 171 million tonnes, valuing more than 362 billion dollars. In an estimate of 80 million tonnes of crustaceans, shrimp contributed to 65 million tonnes [2]. In 2017, in Cameroon, fishery products were estimated at about 140,100 tonnes, from which 1,150 tonnes of shrimp amounted to more than 1.3 billion CFA francs of commercial value [3].
Laws and standards are the main instruments used internationally for the adoption and prioritization of hazard criteria limits and risks to be managed. In this regard, risk assessment is the basic requirement for hazards' acceptable limits definition used in risk management [4]. Presently, in Cameroon, there is no risk management protocol regarding microbial hazards associated with shrimp, a product that is usually consumed along the sea touristic sites.
Among pathogens associated with shrimp, Vibrio parahaemolyticus seems to be the most implicated bacterium, especially when the product is raw or undercooked [5]. Tiger and pink shrimp (Penaeus monodon and Penaeus notialis, respectively), sold in national markets, are generally of nonexportable quality due to V. parahaemolyticus contamination [6,7]. Since they are caught in marine waters, where this bacterium is naturally found because of its halophytic character, contamination becomes almost unavoidable [8].
In Cameroon, there is a paucity or almost no reported cases of V. parahaemolyticus food-borne infections. However, in Africa, at the level of some countries that have access to ocean waters (Mozambique, Nigeria, and Ivory Coast), cases of V. parahaemolyticus infections related to the consumption of seafood such as shrimp have been reported [9][10][11]. According to the literature reviewed, between the years 1990 and 2010, more than 800 outbreaks caused by V. parahaemolyticus and approximately 50,000 cases of disease and some cases of death were reported in the USA, Canada, France, Malaysia, and China, all related to seafood consumption [12][13][14].
e main production and consumption cities in Cameroon are Douala, Kribi, and Limbe, with roasting being the most commonly used form of preparation of the shrimp. It is therefore important to assess if different roasting protocols impact the risk associated with V. parahaemolyticus infection. e objective of this work was to assess the risk of acquiring gastroenteritis caused by V. parahaemolyticus as a result of the consumption of roasted P. monodon shrimp in Cameroon.

Material and Methods
is work was conducted in three phases: (1) a field survey on cooking, storage, and consumption patterns of shrimp, which are factors that impact V. parahaemolyticus infection levels; (2) mathematical modelling of the effect of roasting and postcooking storage parameters on the concentration variations of V. parahaemolyticus; and (3) a quantitative risk assessment of V. parahaemolyticus toxi infection during consumption.

Survey on Cooking, Storage, and Consumption Patterns of
Shrimp Related to V. parahaemolyticus Infection. A survey was conducted in the main production and consumption town of Douala on a sample size of 96 people. It was done using a questionnaire with the aim of assessing factors such as precooking handling, roasting, and postcooking patterns that are factors that can impact the risk associated with V. parahaemolyticus.

Mathematical Modelling of the Effect of Roasting and Postcooking Storage Parameters on V Parahaemolyticus Load
Variations. In the absence of standardized and validated V. parahaemolyticus enumeration methods in the literature, but inspired by the most probable number Polymerase Chain Reaction (MPN-PCR) analysis method [14], a protocol to evaluate the growth kinetics of this bacterium was set up in order to predict its initial concentration before enrichment in Alkaline Peptone Water (APW).
During this evaluation, three isolates of V. parahaemolyticus (VP 1 , VP 2 , and VP 3 ) from three shrimp samples, obtained following the protocol of [15] and identified using API 20 E, were cultured in APW for 6 hours. VP 1 and VP 2 cultures were previously incubated at 37°C overnight, while VP 3 was cultured at 37°C overnight and then stored at − 20°C for 2 h, simulating bacteria present in shrimp during storage conditions. is was done in order to simulate the transportation done by some fishermen from harvest to retail. During incubation, the microbial concentration was evaluated every thirty (30) minutes using standard counting methods on iosulphate citrate bile salt sucrose (TCBS) agar. e enumeration data collected for the three isolates and representing the growth kinetics were fitted in the Baranyi and Roberts [16] model and growth kinetic parameters (µ; Lag) were estimated. e mean of the isolates' generation times (t g ) was used in the following equation to deduce the microbial concentration before enrichment (N 0 ).
(1) e shrimp samples used for the experimental study were randomly selected from different markets, then aseptically packaged, and transported to the Food Safety Laboratory of the Biotechnology Center, Nkolbisson, University of Yaounde I. On arrival in the laboratory, shrimp samples were weighed, labelled, packaged, and stored at − 20°C. For analysis purposes, the initial concentration of V. parahaemolyticus in the samples was determined.
From the survey data, the high-risk shrimp cooking protocol was identified as being the roasting because it was the most commonly used in shrimp consumption, with the lowest mean temperature in shrimp during transformation. e roasting temperature varied between 119°C and 175°C while in the shrimp this oscillated from 80°C to 99°C. To reproduce this roasting protocol in the laboratory, a 3 (3− 1) fractional experimental plan was set up (Table 1). e factors selected for the experiment were the cooking time (min), storage time (hours), and initial concentration (cells/g) with three factors level.
From the experimental design, nine assays were performed (Table 2). e V. parahaemolyticus postcooking/ storage concentration was evaluated in each assay and mathematically modelled using a multiple and nonlinear regression analysis with the STATISTICA version 10.2 software. ere was no available simplified model for the present experimental setup in the literature; therefore, the polynomial type model was preferred and tested. A fractional factorial regression using a backward stepwise model building was used, starting from the global model. e variable selection performed with the model building takes into account the suggestions of Heinze et al. [17]. e best equation was selected based on the statistical significance of all parameters (p < 0.05) included in the model, the presence of all variables, the capacity of giving calculated data in the same order of magnitude as the observed data, the loglikelihood, and the Pearson chi-square (X 2 ) values.

Enumeration of V. parahaemolyticus in Raw Shrimp
Samples for Distribution and during Experimentation. In order to determine the initial V. parahaemolyticus concentration distribution in raw shrimp, 87 shrimp samples were collected from fishermen at the sea coast, while for the experimentation aimed at reproducing roasting protocols, new batch samples constituted of twenty-seven raw fresh shrimp were bought from the local market. ey were randomly divided into nine groups of three shrimp each. For the initial batch concentration, one group was cleaned and ground aseptically in an electric grinder. en an equivalent volume of APW at 1% NaCl was added to the entire ground material in an Erlenmeyer flask. e mixture was incubated for 6 h at 37°C. After the enrichment incubation, the Vibrio concentration was evaluated using the standard counting method on TCBS agar [18,19]. (1) was used to deduce the microbial concentration before enrichment (No).

Quantitative Risk Assessment of V. parahaemolyticus Infection during Consumption.
e Model Risk software package, version 4.0 [20], in combination with Microsoft Excel, was used to carry out simulations. e risk estimate was calculated by using input data in the model, data generated in this study (a mathematical model describing the relationship between postcooking microbial concentration and the cooking, storage, and consumption patterns), and assumptions based on data from previous studies [13,21]. A framework for microbiological risk assessment was applied according to the recommendation of the Codex Alimentarius Commission (CAC) [22]. All the calculations were performed by the Monte Carlo simulation method of sampling from specified input distributions and appropriately combining the sampled values to generate the corresponding output distributions.
is simulation was realized based on 10,000 iterations. e likelihood of illness following exposure to V. parahaemolyticus from consumption of roasted shrimp (P. monodon) was calculated based on the following steps: hazard identification, exposure assessment, and risk characterisation.

Hazard Identification of V. parahaemolyticus.
is was done by reviewing different papers on V. parahaemolyticus pathogenicity [1,18,19,23,24]. Based on the principle of precaution, the incapacity of the Cameroonian food safety system to selectively differentiate haemolytic and nonhemolytic carrier strains, and the fact that nonhemolytic carrier strains have been observed to cause severe illnesses, the hazard in this risk assessment was focused on the total number of V. parahaemolyticus in raw shrimp (P. monodon) and after roasting. Moreover, for the sake of comparison of the risk deviation caused by the assumption of total strains or only pathogenic strains, other calculations were performed using 8% [25] and 15% [1] pathogen species incidence. e adverse health effects include gastroenteritis. e adverse effect of daily consumption was assumed not to be cumulative.

Hazard Characterisation of V. parahaemolyticus.
e dose-response relationship in this study was developed using the infectivity parameters (α, β, and c) that were necessary for calculating the risk of infection generated by USFDA report [13,14]. For more precision, values of α, and β were those of Anse [1] produced with the same data but not rounded up, while c was the adjustment parameter applied on β in USFDA [14] in order to adapt the probability of infection obtained in FAO/WHO [21] to CDC data [26]. ese parameters were extracted from the modified Beta-Poisson relationship using the Furumoto-Mickey approximation [27] under the conditions (β >> 1 and α <β) [28]. Since there was no data on the dose-response relationship available for the  Journal of Food Quality 3 Cameroonian population, the same modified beta-Poisson parameters as those estimated during the modelling of the disease probability by the USFDA [14] were used.

Exposure Assessment of V. parahaemolyticus.
e exposure assessment component of a microbial risk assessment is an evaluation of the likelihood of ingesting a pathogenic microorganism via food and the likely level of exposure. In this assessment, the likelihood of exposure to V. parahaemolyticus after consumption of roasted shrimp (P. monodon) was evaluated. It was determined by integrating the initial concentration distribution function of shrimp into the predictive mathematical model for postcooking concentration and then multiplying by the shrimp weight randomly chosen from the distribution function, the average shrimp size consumed per dish, and the number of dishes per day. e general equation was as follows: where D ing/j : microbial load ingested per day (cells), f: predictive model of V. parahaemolyticus postcooking concentration per shrimp size (cells/g), ts: storage time (h), tc: cooking time (min), A: average shrimp size consumed per dish (g), B: number of dishes per day, Ci (Weibull3 (α; β; c)): distribution function of the initial shrimps concentration of V. parahaemolyticus, and Log− Logistic (α; β): shrimp weight distribution function. e estimated number of annual meals from locally harvested shrimp was obtained by dividing the annual quantities of shrimp provided by the Ministry of Fisheries produced annually by the daily meal size. We decided not to take into account the losses first because, in our opinion, estimating losses in our context introduces other bias factors since they are less directly estimated than total production. Secondly, it is better to overestimate than to underestimate.
irdly, considering total annual shrimp as consumed is the worst-case option.
Its equation was as follows: Nexp year � annual production daily meal size .
e average daily meal size was obtained by multiplying the shrimp's weight (distribution function) by the average number of shrimp consumed per dish and the average number of dishes consumed per day obtained during investigations. Its equation was as follows: Meal size/day � [shrimp weight(log − logistic (α; β))] * average shrimp dish * average number of dishes day .
e different initial concentrations of V. parahaemolyticus in each shrimp (87 shrimp in total) and their respective weights were fitted to several distribution functions using EasyFit software version 5.6. It is a software that fits more than 150 distribution functions incorporated into its system to continuous data series. e choice of the best fitting distribution function was based on three tests, namely, the Anderson-Darling test, the Kolmogorov-Smirnov test, and the chi-square test.

Risk Characterisation of V. parahaemolyticus.
In the risk characterisation, the estimated exposure is normally integrated with the dose-response model to provide a risk estimate and to determine the influence of different mitigation strategies on the risk estimate. e risk estimate was calculated by using the data generated from this study and assumptions based on data from other studies.
From the infectivity parameters (α, β, and c), and the different doses ingested daily (D ing/j ), the infection likelihood (P inf/j ) was estimated using the following equation: e number of annual disease cases was estimated by multiplying the product of the likelihood infection by the number of annual exposures. Its equation was as follows: e model simulations were implemented with Model Risk [20]. e Monte Carlo sampling method was used to perform all calculations from specified input distributions and appropriately combine the sample values to generate the corresponding output distributions.
is simulation was realized based on 10,000 iterations.

Survey on Cooking, Storage, and Consumption Pattern of Shrimp (Penaeus monodon) Related to V. parahaemolyticus Infection.
e results of the survey of 96 households in the city of Douala are shown in Table 3. ey indicate that shrimp are mostly eaten roasted; they are usually roasted using a charcoal oven for 20 min and stored at room temperature with a maximum storage time of 10 h. Most respondents visiting coastal areas usually consume two dishes per day, each containing two shrimp. e growth kinetics revealed the different growth phases: latency, exponential, stationary, and decay. Figures 1(a)-1(c) show the growth kinetics of the three isolates of V. parahaemolyticus (VP) adapted to the Baranyi and Robert [16] model. e VP 3 isolate (Figure 1(c)), which previously underwent thermal stress at − 20°C for 2 hours, presented a latency phase of 0.89 h while no latency phase was observed with VP 1 and VP 2 isolates that were not stressed (Figures 1(a) and 1(b)). e growth rates of the three isolates were 0.91 (VP 1 ), 0.82 (VP 2 ), and 0.93 (VP 3 ) log CFU/ml/h. e generation times for the three strains VP1, VP 2 , and VP 3 were 0.33 h, 0.37 h, and 0.32 h, respectively.

Modelling the Concentration of V. parahaemolyticus Affected by Cooking and Storage
Conditions. Based on the information collected during the survey and the initial V. parahaemolyticus concentration of the sampled batch (2.02 cells/g) estimated using the protocol described previously and equation (1), an experimental design was set up to evaluate the postcooking concentration of V. parahaemolyticus.
e results obtained are shown in Table 4. For the same cooking scale, the residual microbial concentration increases proportionally to the initial concentration. After applying multiple regression analysis, a mathematical equation predicting the residual concentration after roasting as a function of tested parameters was obtained with a log-likelihood of 0.902 and a Pearson chi-square (X 2 ) of 0.431. In this equation, all the estimated parameters were statistically significant at p < 0.05.  Z � EXP − 2, 92042 + 0, 12893 * Ci − 0, 11593 * ts + 0, 01308 * ts * tc − 0, 00263 * tc * Ci , where Z is the postcooking/storage concentration of V. parahaemolyticus in cells/g, Ci is the initial concentration of V. parahaemolyticus in cells/g, ts is the storage time in an hour, and tc is the cooking time in minutes. A representation of equation (7) when the initial concentration is among the highest is given in Figure 2. e model predicts that the increase in the postcooking concentration of V. parahaemolyticus is inversely proportional to the cooking time but proportional to the storage time.

Hazard Identification of V. parahaemolyticus.
Vibrio parahaemolyticus is a halophilic bacterium that naturally occurs in the sea and brackish waters. is bacterium is often transmitted to humans through the consumption of raw, inadequate, or recontaminated cooked seafood [29]. Vibrio parahaemolyticus is rarely incriminated in a health issue in Africa and Europe, but the incidence of infections of this pathogen has increased sharply since 1996 in Asia and North America. e assumptions for this increase are global warming, eutrophication of coastal water, increased consumption of raw seafood, changes in strain virulence, and an increased proportion of subpopulations at risk [1]. However, the incidence has dropped recently in Japan. is is attributed in part to the decline in seawater contamination and especially to the hygienic measures taken downstream during the harvest [30]. Exposed for 2-3 h at room temperature, a growth of V. parahaemolyticus of 10 2 -10 3 CFU/g up to or more than 10 5 CFU/g could be observed [31]. e bacterium's viability decreases with decreasing temperatures. However, cold cannot be used for microbial deactivation in products [14,[32][33][34]. V. parahaemolyticus is destroyed by heat beyond its maximum growth temperature (43°C) [14,35,36], and Twedt [37] demonstrated that treatment at 60°C for 5 min reduced about 3-4 log of V. parahaemolyticus.
Between 1990 and 2010, in the United States of America, more than fifty outbreaks of V. parahaemolyticus were reported, causing nearly 2,500 cases of disease and some cases of death due to seafood consumption [12][13][14]. From 1995 to 2010, V. parahaemolyticus was involved in 13 confirmed outbreaks of collective food poisoning in Europe and 3 suspected outbreaks [38]. In Asia, from 1990 to 2005, about 800 outbreaks, including more than 50,000 cases of V. parahaemolyticus, were reported [13].
Pathogenicity of V. parahaemolyticus results from two hemolysin genes; tdh (thermostable direct hemolysin) and trh (thermostable related hemolysin), which lyse red blood cells by destroying their plasma membrane [39]. e incidence of Hemolysin genes may appear in environmental strains at 0.2 to 2% but can go up to 15% in some ecosystems [1]. Contrary to this, clinical samples (stools) show 95% of V. parahaemolyticus with at least one of the hemolysin genes [24]. However, in 2006, the CDC reported cases of much more serious infections associated with noncarrier strains of the two genes [23]. is can question the use of carrier genes percentage in the quantitative risk assessment of V. parahaemolyticus risk. Based on these few and contradictory data, it would be preferable to use the total microbial concentration in risk analyses.  Two type III secretion systems (T3SS1 and T3SS2) and two other haemolytic components (phospholipase A and lysophospholipase) have been identified in V. parahaemolyticus, but the exact role of the haemolytic components in their pathogenicity is not yet known [29,40].
ere are 13O antigens and 71K antigens identified in clinical strains. New pandemic clones derived from O3: K6, which appeared in Bangladesh in 1996, have been identified around the world [1]. e foods most likely to be involved in gastroenteritis are crustaceans and fish are eaten raw or undercooked. Prevalence studies, conducted in Africa [41] and Cameroon [6,7] in particular, have revealed the presence of V. parahaemolyticus in several sea products caught and sold locally and/or exported. Studies in the United States have revealed the presence of V. parahaemolyticus in oysters in wholesale and retail markets [42]. ough the levels in this study are below 100 microorganisms/g in most of the batches tested, they can exceed 10,000 microorganisms/g in certain regions of the world.

Hazard Characterisation of V. parahaemolyticus.
e hazard characterisation of V. parahaemolyticus is presented here by dose-response curves, already used for the USFDA report [14] and several other studies that deal with risk assessment regarding this pathogen around the world [1,12,13] e initial concentration values of V. parahaemolyticus in shrimp fitted well into the Weibull 3p distribution whose function is given by the following equation: where x is the initial concentration of V. parahaemolyticus in shrimp; α is the slope or the shape parameter; β is the scale parameter; and µ is the the location parameter. e values of the parameters α, β, and μ of this Weibull 3p distribution function are 0.76, 0.53, and 0.016, respectively. e weights of the shrimp were well fitted by the loglogistic distribution, whose function is given by the following equation: where x is the weight of the shrimp; α represents the slope or the shape parameter; and β is the scale parameter. e values of the parameters α and β of this log-logistic distribution function are 5.40 and 67.67, respectively. e summary statistics for the fitted distributions are presented in Table 5.

Assessment of the Meal Size Consumed per Day and the Average Number of Meals Consumed per
Year. Based on the results of the survey, it was possible to simulate distributions of the average daily consumed meal sizes ( Figure 3) and the average number of meals per year ( Figure 4). It was observed that for the average size of the daily meal, 90% of the values were between 279.58 g and 629.97 g (Figure 3), with an average of 407.4 g. Regarding the average number of meals per year, 90% of the values were between 3.65 × 10 7 and 8 × 10 7 meals (Figure 4), with an average of 5.9 × 10 7 meals.
3.3.3.3. Evaluation of the Daily Exposure to V. parahaemolyticus. Several possible scenarios were performed to illustrate minimal, average, and maximum predicted risk conditions. is exposure is, in general, proportional to the size of the meal, and the storage time but inversely proportional to the cooking time (Table 6). It was observed that 90% of the predicted conditions for the minimum exposure were between 14.73 cells/d and 23.38 cells/d, and those associated with predicted conditions for maximum exposure were between 159.86 cells/d and 228.80 cells/d, while for the average scenarios, the values were between 47 and 181.05 cells/d. Figure 5 represents the simulation of the exposure trend of the consumers with modal behaviour. ese predicted simulations were based on the modal values of the survey. It was observed that 90% of the predicted exposure values were between 73.10 and 109.38 cells/d with a mean of 86.03 cells/d.

Risk Characterisation of V. parahaemolyticus.
e probabilities of infection were calculated based on the exposure degree of the different scenarios (Table 6). It was observed that 90% of the simulated values for the minimum and maximum risk were between 2.41 × 10 − 7 and 3.80 × 10 − 7   Figure 6 represents the distribution simulation of the risk of infection to which consumers with modal behaviours are exposed. It was observed that 90% of the predicted risk values were between 11.96 × 10 − 7 and 17.79 × 10 − 7 with an average of 1.41 × 10 − 6 . In general, an average of 14 people in a ten million is likely to contract gastroenteritis following the ingestion of their meal.
Annual distributions of disease cases were simulated at minimum, maximum, and modal trend behaviour scenarios. Ninety percent of the simulated disease cases per year for minimum, maximum, and general risk scenarios are between 0.36 and 0.90 cases (minimum risk scenario), between 2.33 and 23.6 cases for maximum risk scenario, and between 1.74 and 4.28 cases for modal risk scenario (Figure 7). From the average value of the distribution, it can be noted that about 3 consumers are likely to suffer from gastroenteritis in a year.

Discussion
According to the literature we could access, no data related to V. parahaemolyticus toxi infection has yet been reported in Cameroon, unlike in other African countries    [9][10][11] and the world [1,13,14,38]. is does not exclude the fact that the problem may exist. e risk quantification in this work is based on the results of the survey and the predictive mathematical model proposed with several scenarios representing minimum, average, maximum, and modal-consumer behaviours. e values obtained by simulation on the basis of 10,000 iteration risk show that the exposure is proportional to the initial load, the size of the meal, and the storage time but inversely proportional to the cooking time. In the same light, the risk of infection and the number of gastroenteritis cases follow these same trends. In fact, the higher the exposure, the higher the risk of infection, and the greater the number of disease cases [1,12].  Risk of illness (x1E-5)  In this study, it was generally noted that the Cameroonian population consuming the product has an average exposure of 86 cells/d with an average risk of infection of 1.41 × 10 − 6 and a number of annual gastroenteritis cases estimated at 3 cases/year based on annual shrimp production. is shows that the risk is low on the Risk Ranking scale proposed by Summer and Gallagher [43], and thus, there is a low incidence of the disease. In Australia in 2009, V. parahaemolyticus was downgraded from the second category (intermediate) of the Risk Ranking for the hazards associated with seafood [43] to the first category (low) [44]. Abdullah et al. [12] also observed a low risk associated with cooked black shrimp in Malaysia. Indeed, to induce a disease with a 100% probability, an exposure value of 10 6 CFU/meal is required [14], which is not achieved even in the maximum exposure scenarios of this study. As indicated in the hazard identification part of this study, the choice of using total V. parahaemolyticus was deliberate. For comparison's sake, simulation based on the same worst scenario ( Table 6, condition 15), taking into consideration total V. parahaemolyticus, 8%, and 15% pathogen species incidences, has led to the following infection risk: 3.18 × 10 − 6 , 2.16 × 10 − 7 , and 2.58 × 10 − 7 , respectively. ese simulation results indicate that extrapolating haemolytic carrier percentage species obtained from other regions in this context would result in a ten-fold reduction of the risk. e influence of variables on the risk of infection per day and the observed shifts due to modulation of the studied factors indicate that these are possible ways of implementing mitigation policies. e risk model obtained can be implemented in a simple excel sheet in order to help decisions-makers simulate different patterns of cooking and handling of shrimp, and this will orientate risk communication.
is study was performed while accepting some limitations. Firstly, the use of the Baranyi and Robert model to adapt our growth data with respect to the 3 isolates of V. parahaemolyticus and the use of the estimated parameters to predict their initial concentration in shrimp before enrichment. In a dilemma, the choice that could lead to little overestimation of the risk was preferred. Secondly, the doseresponse relationship of V. parahaemolyticus was based only on previous studies on a small number of volunteers (20 people) and modified to fit US CDC statistics. e bacteria load was administered to the volunteers with a pH-neutralizing buffer rather than with a food matrix [21]. irdly, the exposure assessment model was predicted from total V. parahaemolyticus (with and without virulent genes) in roasted shrimp.

Conclusion
For the Cameroonian population consuming roasted Penaeus monodon, the mean estimated exposure trend for most households is 86 cells/day. 90% of the predicted values are between 73 and 110 cells/day. 90% of the risk of infection to which most households are exposed falls between 11.96 × 10 − 7 and 17.79 × 10 − 7 with an average of 1.41 × 10 − 6 . e estimated number of annual disease cases based on annual production is between 1 and 10 cases.
is hazard characterisation indicates that roasted shrimp have a very low incidence of V. parahaemolyticus health effects on populations. is hazard could be further reduced by controlling roasting temperature as well as the storage conditions before consumption.

Data Availability
e data used to support the findings of this study are included within the article. Untransformed data are available on demand to the authors.

Conflicts of Interest
e authors declare no conflicts of interest.