Modeling the risk of introducing cyst-infected meat pork into the food chain in Benin

Abstract

Background

Poorly cooked pork containing Taenia solium cestode larvae contributes to human cysticercosis, a neglected zoonotic disease. Pork is consumed by a majority of the Benin population, and the cooking of popular dishes may not be sufficient to destroy cestode larvae. This study’s aim was to estimate the risk of allowing pork contaminated with Taenia solium cysticerci into the food chain.

Methods

A stochastic risk assessment was developed in R to model the risk resulting from the three scenarios leading to contaminated pork being introduced into the food chain. The model was parameterized using official animal health surveillance data, published literature and expert opinion.

Results

The overall probability of cyst-infected meat pork entering the food chain was estimated at 0.41 (95%Crl 0.34–0.48); from informal abattoirs without meat inspection: 0.16 (95%Crl 0.08–0.22) (scenario 1); from non-detection at inspection in formal abattoirs (scenario 2): 0.24 (95%Crl 0.16–0.35); and 0.007(95%Crl 0.001–0.021) from remaining contamination following partial condemnation of positive carcasses (scenario 3). The total number of cyst-infected meat pork that entered the food chain from 2020 to 2023 was estimated at 89,167 (95%Crl 60,43–115,174).

Conclusion

The risk of cyst-infected meat pork entering the food chain is high. In addition to efforts to improve sanitation and pig farming practices, mitigation measures should also reinforce slaughterhouse inspection with tests that are more sensitive.

Background

In Benin, pork is highly appreciated in most regions and widely consumed by people of all economic strata [1–3]. The number of commercial pig herds has grown since 2016, with over 60% of pigs of local breeds [4]. In a survey on food choices conducted in 2020–2021 in 10 of Benin’s 12 departments, pork ranked fourth as meat choice, and was reported as consumed by 66% of respondents [5]. Consumption of raw or undercooked meat exposes humans to the risk of zoonotic diseases, including cysticercosis, caused by ingestion of water or food contaminated by the larval form of Taenia solium [6]. Preparation or cooking methods which do not destroy the cysticerci (braised, smoked, raw or cured meats) are of particular concern [7–9]. This is the case in Benin where the pork dishes most popular with consumers are pork kebabs, fried and baked meat, grilled meat and soups (kpêtê) [2, 5], which may fail to destroy cysticerci.

Under-reporting of cysticercosis is common in developing countries [9], despite this disease being considered a major public health problem in endemic areas [9–12]. A study carried out in twenty-nine hospitals in Benin revealed that only four had reported cases of human cysticercosis over a five-year period [13]. This prevents estimating the burden of cysticercosis using surveillance data in humans.

In contrast, disease surveillance is carried out at animal level, and post-mortem inspection by veterinary officers is conducted at abattoirs and official slaughter areas. Carcass parts that tend to contain higher numbers of cysticerci are the tenderloins, tongue, heart, thigh, shoulder and masseter muscles [13–18]. To reduce the risk of transmission to humans, veterinarians condemn the whole carcass or only the infected parts, depending on the number of cysts found by inspection. Massive infestation, characterized by more than 5 cysticerci in a palm space (approx. 10 cm x 10 cm) or cysts are scattered throughout the carcass, results in whole carcass condemnation. Partial infestation, when fewer than 5 cysticerci in a palm space were found at post-mortem inspection, and only in a limited part of the carcass, leads to seizure of the infected organs or parts, while the rest of the carcass is authorized to be sold for consumption. No further analysis - autopsy or antigenic testing – is conducted as part of surveillance and control programs [18]. The overall detection level of porcine cysticercosis via post-mortem inspection remains very low at 22 p. 10,000 [18] and total condemnation in case of detection is very rare, at 0.32% [18] and 1.13% [13] of infected carcasses in two studies conducted in Benin. Post-mortem infection fails to detect a large fraction of cyst-infected carcasses [16, 18, 19], and prevalence estimates derived from surveillance data need to be adjusted accordingly. In addition to formal slaughter, informal slaughter is also commonly conducted in Benin, in which case carcasses are released directly for consumption without inspection. A prevalence survey carried out in informal areas across Benin found 7.83% of pigs positive to cysticercosis via post-mortem inspection, while 41.5% were positive with ELISA [13, 14].

In view of the endemicity of the disease in Benin, the pork cooking and consumption practices, the low sensitivity of post-mortem inspection, the low percentage of whole carcass condemnation in case of contamination, and the existence of informal slaughter areas, it is important to understand the risk of cyst-infected carcasses being introduced in the food chain and potentially leading to exposure of consumers of pork in Benin.

Methods

The aim of this study is to estimate the probability that cyst-infected meat pork is sent for human consumption in Benin, using a stochastic risk assessment.

Risk assessment model

Three main mechanisms (called scenarios in this study) can lead to cysticerci- infected carcasses entering the food chain (Fig. 1):

Fig. 1.

Conceptual model of the risk of introducing cysticerci-contaminated pork into the food chain in Benin

• Scenario 1, “Not inspected”: in case of informal slaughter, there is no veterinary inspection and thus all cyst-infected carcasses are sent for consumption;

• Scenario 2, “Not detected”: due to the low sensitivity of post-mortem inspection, there is a high probability of non-detection of cyst-infected carcasses by veterinary inspection in official abattoir;

• Scenario 3, “Partial seizure”: detection of infection via post-mortem inspection most often results in partial condemnation i.e., only carcass parts where cystiocerci have been found are being removed. In cases where some of the cysticerci present on the carcass are missed by visual inspection, this results in contaminated meat being authorized for consumption, despite partial condemnation following detection at inspection.

These three main scenarios were further detailed to characterize the various parameters used to estimate their probability of occurrence, as well as the overall probability that cyst-infected carcasses are sent to the food chain in Benin (Fig. 2).

Fig. 2.

Decision tree for calculating the probability of introducing cyst-infected pork into the diet, Benin 2024

Model parameters and sources of data

The model was parameterized using national surveillance data on inspection for cysticercosis detection at official abattoirs across Benin over the years 2020–2023, studies on cysticercosis in Benin, and studies on the performance of cysticercosis diagnostic methods in pigs (Table 1). In addition, expert opinion was collected for parameters that could not be informed from available data. The oral consent of the experts consulted was obtained after explanation of the context and objectives of this work.

Table 1.

Description of data sources and the model parameters for estimating the risk of introducing cyst-infected meat pork into the food chain in Benin

Parameter	Description	Source	Distribution	Mean (95%Crl)
P1	Probability of informal slaughter, P(Informal_slaughter)	Expert opinion on the proportion of pigs slaughtered in informal areas, based on the knowledge and experience of inspectors involved in control at strategic level (Expert 1: Min1 = 0.15, Max1 = 0.45, Mode1 = 0.3) and operational level (Expert 2: Min2 = 0.4, Max2 = 0.5; Mode2 = 0.5)	Combination of Pert (0.15, 0.3, 0.45) and Pert (0.4, 0.5, 0.5) with equal weights	0.39 (0.21–0.50)
P2	Prevalence of cyst-infected pigs in informal slaughter areas via Ag-ELISA P(Infection|informal)	Tassou et al., 2022 [14]: Out of 460 (n_testELISA) animals tested, 191 (s_pos.ELISA) were positive results	Beta (s_pos.ELISA + 1, n_testELISA - s_pos.ELISA + 1) or Beta (191 + 1, 460 − 191+1)	0.41 (0.37–0.46)
P3	Probability of official slaughter, P(Official_slaughter)	Deducted of P(informal slaughter)	1- P(informal slaughter)	0.61 (0.50–0.79)
P4_not adjusted	Probability of cyst-infected pigs in official slaughter areas after veterinary inspection P(Infection|official slaughter) not adjusted	Ministry of Agriculture and Livestock animal’s health surveillance data from 2020 to 2023 [20]: Out of 128,364 (N_inspected) carcasses inspected in official slaughter areas, 520 (S_detected) were cyst-infected carcasses detected via post-mortem inspection	Beta (S_detected + 1, N_inspected - S_detected + 1), or Beta (520 + 1, 128,364–520 + 1)	0.0040 (0.0037–0.0044)
Adjustment factor	Adjustment factor for P(Infection|official slaughter)	Ndri et al., 2022 [21] Out of 93 (n_no_detect) animals no detected cyst-infected at post-mortem inspection, 41(s_pos_test) were positive results for ELISA	Beta (s_pos_test + 1, n_no_detect - s_pos_test + 1), or Beta (41 + 1, 93 − 41 + 1)	0.44 (0.34–0.54)
S_detected_adjusted	Cyst-infected carcasses estimated via ELISA	Out of 127,844 (N_inspected -S_detected) carcasses not detected by inspection in official slaughter areas, 127,844 x Adjustment factor were cyst-infected	S_detected + (N_inspected -S_detected) x Adjustment factor or (520+ (128,364–520) x Beta (93 + 1,41–93 + 1))
P4	Adjusted prevalence of cyst-infected pigs in official slaughter areas P(Infection| official slaughter)	Prevalence of cyst-infected pigs slaughtered in official abattoirs after adjusting for the sensitivity of post-mortem inspection	Beta (S_detected_adjust + 1, N_inspected - S_detected_adjust + 1)	0.44 (0.35–0.54)
P5	Probability of no detection cyst-infection at post-mortem inspection, P(Non detection|infection)	Gounassou, 2014 [18]: Out of 17 animals positive with ELISA test (n.pos.test), only one was detected at post-mortem inspection (s.detect = 1)	Beta(s.detect + 1, n.pos.test –s.detect + 1) or Beta(1 + 1, 17 − 1 + 1)	0.89 (0.73–0.99)
P6	Probability of detection cyst-infection at post-mortem inspection, P(Detection|Infection)	Deducted of P(Non detection|infection)	1- P(Non detection|infection)	0.11 (0.01–0.27)
P7	Probability that carcasses underwent partial seizure (vs. full carcass condemnation) after detection via post-mortem inspection at official abattoirs P(Partial seizure | detection)	Tassou et al., 2022 [14] with: Out of 62 (n_postmort.pos) cyst-infected pigs found at inspection, 9 (s_pos.tot.seiz) were partially condemned	Beta(s_pos.tot.seiz + 1, n_postmort.pos - s_pos.tot.seiz + 1) or Beta (9 + 1, 62 − 9 + 1)	0.84 (0.74–0.92)
P8	Probability of carcasses remained contaminated with cysts after partial condemnation at post-mortem inspection, P(Contaminated meat | partial seizure)	Expert opinion on the proportion of carcasses that remain contaminated with cysts after partial condemnation, based on the knowledge and experience of an animal health executive	Pert( min, mode, max) Min = 0.05, Max = 0.5, Mode = 0.3	0.29 (0.13–0.44)

Adjustment of official inspection data

The performance of post-mortem inspection has been compared against ELISA a test frequently used for the diagnosis of cysticercosis [17–19]. Such comparison studies comparing diagnostic methods for detection of cyst-infected carcasses in Africa [18, 21, 22] were reviewed, and the study by N’dri et al. [21] was selected to adjust the proportion of cyst-infected carcasses at official abattoirs (Table 1). The adjustment was conducted as described below:

• Step 1: First, we determined the Adjustment factor using data from N’dri et al [21]. We estimated the proportion of false negative results by inspection compared to the ELISA test (41 out of 93), assuming that the ELISA test can be considered a perfect test with 100% sensitivity and specificity. The Adjustment factor was modelled using a Beta distribution Beta (s+1, n-s+1) so Beta (41+1, 93-41+1) (Table 1).

• Step 2: This Adjustment factor was then applied to Benin's official slaughter data to adjust for false negatives amongst the pigs deemed negative at veterinary inspection between 2020 and 2023. Out of 128,364 carcasses inspected during that period, 127,845 were negative at post-mortem inspection, and 521 were identified as cyst-contaminated. Thus, S_detected_adjusted = 521+ (127,845 x Adjustment factor) (Table 1).

• Step 3: The adjusted prevalence of cyst-infected pigs slaughtered under official slaughter was then estimated using S_detected_adjusted as the number of cyst-contaminated carcasses, thus taking into account the sensitivity of post-mortem inspection. The adjusted prevalence, P4, was modeled using a Beta distribution (Table 1).

Risk estimation

The stochastic risk assessment model was developed in R version 4.4.2 [23]. Output distributions, means and 95% credibility intervals (95%) were estimated using a Monte Carlo simulation with 100,000 iterations.

Scenario analysis

Two experts provided an estimate of the probability of informal slaughter, P1, based on their knowledge and experience. To assess the impact of these opinions on the probability of each scenario as well as the overall probability, we ran the model with each expert opinion separately, as well as with their combination (Table 1), assuming that the two experts’ opinions carry the same weight, and compared the credibility intervals of the model outputs.

Sensitivity analysis

A sensitivity analysis was performed to determine the influence of the input parameters on the overall probability using conditional means. Results were visualized with a Tornado graph.

Results

Estimated probability that cyst-infected pork enters the food chain in Benin

Based on available data and under current conditions, any pig carcass entering the food chain in Benin has a probability of 0.41 (95% Crl 0.34–0.48) of being infected with cysts (Table 2). For the period 2020–2023, this corresponds to 89,167 95% Crl (60,943 − 115,174) cyst-infected carcasses sent to the food supply chain in Benin out of the 216,601(95% Crl 161,716 − 256,213) pigs slaughtered both in official and informal slaughter places.

Table 2.

Probabilities of sending cyst-infected carcasses into the food chain, Benin 2020–2023

Scenarios	Probability of occurrence	Number of cyst-infected carcasses entering the food chain in 2020–2023
Not inspected	0.16 [0.08–0.22]	36,698 (95% Crl 13,683 − 56,094)
Not detected	0.24 [0.16–0.35]	51,001 (95% Crl 37,241 − 64,944)
Partial seizure	0.007 [0.001–0.021]	1,480 (95% Crl 162- 4,383)
Overall	0.4 [0.34–048]	89,191 (95% Crl 60,752 − 115,096)

The largest number come from official slaughterhouses, where non-detection at post-mortem inspection resulted in 51,013(95%Crl 37,313 − 64,763) cyst-infected carcasses authorized for consumption, while the informal sector contributed 36,673(95%Crl 13,744 − 56,062) cyst-infected carcasses, and the remaining came from partially condemned carcasses.

The three scenarios considered contribute differently to the overall probability that cyst-infected pork enters the food chain in Benin (Table 2). Scenario 2, non-detection of cyst-infected pigs during post-mortem inspection in official slaughter places, has the highest probability of occurrence at 0.24 (95% Crl 0.16–0.35), followed by Scenario 1 (Non-inspected), with a probability of cyst-infected meat being released for consumption via informal slaughter of 0.16 (95% Crl 0.08–0.22). Scenario 3, where a partially condemned carcass remains infected with cysts, has a low probability of occurrence at 0.007 (95% Crl 0.001–0.021).

Scenario analysis

Whether the model was run with the combined distribution of the two experts, or with Pert distributions from each expert opinion, the 95%Crl of the resulting overall probability distributions largely overlap (Fig. 3). We therefore chose to parameterize the model with the combined distribution of the two expert opinions.

Fig. 3.

Density curves of overall probabilities of introduction to consumption of cyst-infected meat pork, Benin 2024

The conditional means sensitivity analysis indicates that the prevalence of cysticercosis in pigs slaughtered at official slaughter places, P4 adjusted for the sensitivity of post-mortem inspection (Mean 0.44, 95%Crl 0.35–0.54), and the probability of non-detection of cyst-infected pork via post-mortem inspection, P5 (Mean: 0.89, 95%Crl 0.73–0.99) ; are the parameters that most influence the overall probability of introduction of cyst-infected pork carcasses into the food supply chain in Benin, followed by the prevalence in informal slaughter places P2 (Mean:0.41, 95% Crl 0.37–0.46) and the probability of detection via post-morten inspection P6 (Mean: 0.11, 95% Crl 0.01–0.27) (Table 1). The results also indicate that the overall risk is most sensitive to the uncertainty of Scenario 2, i.e. the probability that cyst-infected pork enters the food chain in Benin due to non-detection of cyst-infected pigs during post-mortem inspection in official slaughter places (Mean: 0.24, 95% Crl 0.16–0.35)] (Fig. 4).

Fig. 4.

Tornado plot of parameters and probabilities of the introduction of cyst-infected pork, Benin 2024

Discussion

The aim of this study was to estimate the risk of introducing cyst-infected pork into the food supply chain in Benin. This is a key step towards assessing quantitatively the risk of exposure to T. solium via consumption of pork, and exploring the cost-effectiveness of potential mitigation measures pre- and post-harvest.

The assessment, parameterized using data available for Benin, shows that the risk that pork intended for consumption is infected with cysticerci is 0.41 (95%Crl 0.34–0.48). Our study suggests that the probability of cyst-infected meat entering the food chain associated with informal settings 0.16 (95%Crl 0.08–0.22) is not significantly higher than that associated with official slaughterhouses 0.25(95% Crl 0.16–0.37). In the informal sector, all cyst-infected carcasses are authorized for consumption, due to the absence of inspection. In contrast, in official abattoirs, thanks to the work of inspectors, the risk of sending infected carcasses for consumption is lower. In case of visible signs of infection, this risk is almost zero. However, lesions are not always easy to detect, and thus inspection does not fully rule out the risk of cysticerci being present in the carcass. Although pre- and post-mortem inspections in official abattoirs reduce the risk that a cyst-infected carcass is authorized for consumption, it does not prevent it. This, combined with the larger number of pigs slaughtered in official slaughterhouses, resulted in non-significant differences in the estimated number of cyst-infected carcasses authorized for consumption in official slaughterhouses (51,013; 95% CI: 37,313 − 64,763) and informal slaughterhouses (36,673; 95% CI: 13,744 − 56,062) for the period 2020–2023. Our work highlights that whether or not pork meat is inspected, the risk of introducing cysticerci into the food chain is significant, and thus control strategies need to consider both official and informal sectors.

The estimated prevalence of cysticercosis in pigs sent to official slaughter places in Benin, 0.41(95%Crl 0.37–0.46), is similar to that reported in a risk assessment of exposure to Taenia solium via consumption of pork in Kenya [24] (Mean: 0.38, 95%Crl 0.24–0.51).The probability of detection of infected carcasses by post-mortem inspection was very low in both countries: 0.00(95%Crl 0.00-0.07) in Kenya according to the same study versus 0.11(95% Crl 0.01–0.27) in our study. This combination of high prevalence of cysticercosis in pigs and low probability of detection via post-mortem inspection greatly influence the overall risk of introducing cyst-infected carcasses into the food chain. In addition, expert opinion collected for this analysis suggests that in Benin, informal pork slaughter is frequent 0.39 (95%Crl 0.21–0.50), and significantly higher than in Kenya 0.04 (95%Crl 0.00-0.19) [24]. Given that there is no inspection in such settings, this is also an important contributor to the overall risk that we estimated for Benin. However, in Benin, the fact that there is no significant difference between the prevalence of infected pigs in official and informal slaughter places does not give rise to concerns that farmers are preferentially sending animals they believe to be infected to informal slaughterhouses to avoid the risk of condemnation.

Post-mortem inspection is not a sensitive method for detecting cysticerci in pig carcasses, as observed by several other authors in Benin [18], Madagascar [21] and Rwanda [17]. In this study, we used data collected in Madagascar on slaughterhouse surveillance of porcine cysticercosis to estimate the sensitivity of post-mortem inspection [16], and adjust our prevalence estimate accordingly. When making this adjustment, we assumed that the ELISA test is perfect (i.e., 100% sensitivity and specificity), although studies have reported values between 68% and 100% for sensitivity, and 67% and 94.55% for specificity [15–18, 21, 22]. However, the fact that our results are comparable to those from Kenya in terms of prevalence seems to indicate that this possible over- or underestimation may not have a large effect on results.

Additional measures, such as testing, could be considered to improve detection of cyst-infected meat. In Madagascar, N’DRI et al. recommend using Western Blot rather than ELISA testing at official abattoirs, because it is a highly specific test, and would avoid false-positive carcasses being seized and disposed of, which the authors argue could result in an increase in informal slaughtering to avoid inspection [21]. This is an important consideration in the context of African countries, where livestock farmers and butchers are not compensated for condemned carcasses, and where a test that is not specific enough would cause them too many economic losses. On the other hand, the Elisa test has better sensitivity than the Western-Blot, and so in the future it will be important to use quantitative risk assessment tools such as the one developed here to assess the cost-effectiveness of these mitigation options to inform decision-making. Given the fact that the main contributor to the overall the risk of introducing cyst-infected pork into the food chain in Benin is the probability of non-detection via post-mortem inspection, such evaluation of interventions at official slaughter places is key. Indeed, authorities and decision-makers can more easily influence the formal sector through the implementation of laws, standards, regulations and prevention strategies.

The risk estimations presented in this work are based on the limited data available for Benin. For example, official inspection data from slaughterhouses was incomplete for three out of twelve departments, which may have slightly affected the prevalence estimate. However, this was the best data available, and prevalence estimates were found comparable with other risk modeling studies. The prevalence estimate for informal slaughter areas was based on a single published study [14] conducted in Benin. It was assumed to be representative of the pigs slaughtered in similar slaughter areas across the country, but may have resulted in a biased estimates if the sampling approach was not appropriate. Finally, due to the absence of data on the proportion of animals slaughtered in informal settings, in this study the proportion of animals slaughtered in official vs. informal slaughter places was based on expert opinion. However, such estimates may be biased by the experts’ experience [25]. As more data becomes available, the estimates from this study should be verified and updated as needed, to address potential issues of representativeness and biases discussed here. The impact of data scarcity was captured in this study via confidence and credibility intervals.

The risk of exposure to T. solium for consumers of pork could not be assessed in this study due to the lack of data on level of infection in carcasses, pork consumption, storage and cooking in Benin. Studies to collect relevant data should be conducted in Benin, so that the risk of infection with the zoonotic parasite T. solium can be fully assessed in Benin, and mitigation strategies evaluated.

Conclusion

This study has highlighted a high risk of introducing cyst-infected meat pork into the food chain in Benin. This could lead to infection with T. solium if viable cysticerci are not destroyed during storage or cooking, as illustrated by the high level of reported convulsive seizures (neurocysticercosis, often confused with simple epileptic seizure) [6, 13]. Data on pork consumption and cooking practices in Benin are needed, so that the present study can be complemented with an exposure assessment to viable cysticerci to estimate the risk of human infection with T. solium resulting from consumption of pork.

Due to the nature of T. solium infection cycle, and the partial effectiveness of individual control interventions, a One Health integrated approach is needed to control cysticercosis. Control strategies should combine, a range of interventions such as: (i) awareness campaigns targeting the various stakeholders of the pork supply and consumption chain to educate on the disease, the risk of selling uninspected pork and consuming undercooked pork, and prevention and control practices; (ii) improvement of the sensitivity of abattoir inspections through ongoing training of staff at official slaughterhouses and actors involved in informal slaughter during major ceremonies; (iii) prevention measures to decrease the likelihood of contact of pigs with human feces, to reduce prevalence: latrines in pig farming areas, confinement of pigs, etc. In addition, interventions in high-risk areas could also include (i) treatment of human taeniasis, (ii) anthelmintic campaigns in pigs and (iii) complementing visual inspections in abattoirs with more accurate detection methods. Risk assessment models can be used to evaluate the cost-effectiveness of combinations of pre- and post-harvest interventions and inform the development of integrated strategies to control cysticercosis.

Authors’ contributions

RA and BAH designed the study, analyzed data and wrote the article. BMG, MG, AMA assisted in the design of the study, data analysis and the amendment of the article. SC oversaw the development of the model, contributed to literature review and selection of relevant data, verified the model and results, and reviewed the manuscript. All authors have read and approved the final version of the manuscript.

Funding

Work funded by One Risk, a technical assistance program made possible through a contribution from USDA-APHIS (Agreement #: 23-5000-2155-CA), in collaboration with WOAH.

Data availability

The data and code are available from the corresponding author on reasonable request, with the exception of the surveillance data in official slaughter place.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.