Campylobacter spp. in chicken meat from traditional markets in Peru and its impact measured through a quantitative microbiological risk assessment

Abstract

Campylobacter is a major cause of foodborne gastroenteritis worldwide, with the mishandling of contaminated chicken meat among the main pathways for human infection. Granted the disease burden due to this pathogen, systematic assessments of its potential impact are necessary. The aims of this study were to evaluate both presence and load of Campylobacter in chicken meat sold in traditional markets, assess risk factors related with the infrastructure and hygienic conditions of market stalls, and evaluate control strategies for campylobacteriosis in Peru through a quantitative microbiological risk assessment (QMRA), a data-driven, systematic approach to quantitatively assess risks by integrating empirical contamination levels, microbial behavior, and consumer exposure. Between February and December 2022, a total of 90 chicken meat samples from traditional markets were sampled and evaluated by both culture and quantitative real-time polymerase chain reaction (qPCR). Campylobacter spp. were detected in 28% and 76% of samples with a mean quantification of 3.3 log₁₀ CFU/g and 4.9 log₁₀ GC/g through culture and qPCR, respectively. Market stalls with tap water showed higher prevalence and loads, while those without refrigeration had higher quantifications. The QMRA analysis, using the most conservative parameters and bacterial load, indicated that the entire modeled population develops campylobacteriosis at least once annually. These results highlight the public health impact of Campylobacter, potentially linked to the alarming number of Guillain–Barré syndrome cases observed in Peru. Our study suggests that consumer-level interventions, such as reducing kitchen cross-contamination and improving chicken meat storage, could substantially reduce campylobacteriosis cases in this population.

1. Introduction

Campylobacter is the leading cause of human bacterial gastroenteritis worldwide and one of the top four causes of diarrheal disease (Sheppard & Maiden, 2015; WHO, 2020). Diarrheal diseases represent the most frequent foodborne illnesses, with an annual tally of 550 million cases, affecting 220 million children under 5 years of age (WHO, 2020). In many countries, C. jejuni is responsible for most cases of campylobacteriosis, followed by C. coli (FAO/WHO, 2009; Fitzgerald, 2015). Besides causing self-limiting diarrhea, Campylobacter infections are associated with serious long-term conditions such as Guillain–Barré syndrome, Miller–Fisher syndrome, and reactive arthritis (Endtz, 2019; Kaakoush et al., 2015). High-risk groups, including young children, the immunocompromised, and the elderly, may experience severe disease, particularly in low- and middle-income countries (WHO, 2020).

Poultry is an important reservoir of Campylobacter especially for C. jejuni, with 50–80% of campylobacteriosis cases in the European Union attributed to this host (Koutsoumanis et al., 2020; Skarp et al., 2016). Main transmission pathways are cross-contamination due to mishandling of raw chicken meat contaminated with Campylobacter (neck skin being the most contaminated area) and the consumption of undercooked chicken (Ellis-Iversen et al., 2020; Mullner et al., 2009). In Peru, a major consumer of chicken meat, fresh chicken is primarily distributed through traditional markets. However, the role these markets may play in the transmission of Campylobacter remains unknown (MIDAGRI, 2022). While current surveillance focuses on prevalence data, incorporating bacterial quantification data is necessary for a more accurate assessment, even though bacterial culture is time-consuming and labor-intensive (Lv et al., 2020). In this regard, the potential use of novel methods such as the quantitative real-time polymerase chain reaction (qPCR) for Campylobacter in routine surveillance remains an area of ongoing research (Lv et al., 2020; Magana et al., 2017). Promising results suggest that qPCR could be as effective as traditional culture methods (Stingl et al., 2021).

Quantitative microbiological risk assessment (QMRA) has extensively been used to evaluate public health risk from microbiological foodborne hazards (Chardon & Evers, 2017). It has been used to estimate the impact of the broiler meat chain on human campylobacteriosis in Europe and to inform risk mitigation (Asakura et al., 2022; Koutsoumanis et al., 2020). While comprehensive QMRA of the entire food chain is costly, simplified models using retail data provide a viable alternative for resource-limited countries like Peru (Nauta et al., 2007). This study applies an improved swift QMRA methodology (Chardon & Evers, 2017) to estimate Campylobacter’s impact in Peruvian regions with retail data, evaluate control strategies, and provide insights into campylobacteriosis cases obscured by limited surveillance and underreporting.

Given the global impact of this pathogen and the high exposure levels of the Peruvian population (Munayco et al., 2020), surveillance of Campylobacter prevalence and concentration in its main food vehicle and the application of this data into risk models like QMRA is an urgent need. For these reasons, this study aimed to i) detect and quantify Campylobacter spp. in three Peruvian regions by culture and compare its performance with the faster qPCR method; ii) evaluate selected potential risk factors for Campylobacter presence in traditional market stalls; and iii) assess Campylobacter’s impact through QMRA, considering bacterial loads find by culture in selected regions of the country, cross-contamination scenarios, and the availability of refrigerator to mitigate bacterial growth and human exposure to the pathogen.

2. Materials and methods

2.1. Sample collection

Samples of fresh raw chicken (n=90) were randomly collected between February and December 2022 from traditional markets of three Peruvian regions: Huancayo, Huaral, and Tumbes/Piura. Traditional markets were selected based on convenience and feasibility of sample collection. In Huancayo, two markets were sampled; in Huaral, three; and in Tumbes/Piura, four. In each market, stalls were randomly selected, and one sample was collected per selected stall, resulting in a total of 30 samples per region. Samples were transported at 3 – 8°C in coolers containing gel packs, monitored using the temperature data−logging device TempTale^®Ultra (Sensitech Inc., Redmond, WA), and processed microbiologically within 24 h.

At each visit, information on the infrastructure and hygienic conditions of the market stalls was collected through a survey on the Epicollect5 mobile app (Gupta et al., 2021). Field data collectors gathered the following information based on observations of the stalls: acceptability of infrastructure and hygiene, sale of chicken meat only, whether cutting or handling of samples was necessary, degree of cleanliness of the cutting table, degree of cleanliness of the cleaning cloth used on surfaces, and the availability of a refrigerator and tap water in the stall. Further description of this survey is in Supplementary Table S1.

2.2. Identification and quantification by culture technique

Campylobacter spp. detection followed the International Organization for Standardization standards (ISO) 10272–2:2017 (ISO, 2017). Briefly, 10 grams of neck-skin from each sample were diluted 1:10 with sterile 1% buffered peptone water (Merck KGaA, Darmstadt, Germany) and homogenized for two minutes. Immediately, 40 ml of the homogenized solution was stored in a falcon tube at −20°C for subsequent molecular analysis. Ten-fold serial dilutions were spread-plated onto Campylobacter agar with 10% defibrinated sheep blood and Blaser-Wang supplement (Biomark Laboratories, Pune, India) and in duplicate for quantification onto modified charcoal cefoperazone deoxycholate (mCCD) agar plate with CCDA selective supplement (Merck KGaA, Darmstadt, Germany). Plates were incubated at 41.5°C for 44h under microaerophilic conditions using CampyGen^™ 2.5L (Oxoid Ltd, Basingstoke, UK). Suspected colonies were Gram-stained and quantified according to ISO 7218:2007(E). For confirmation, colonies were streaked on Columbia agar (Becton, Dickinson and Company, Le Pont de Claix, France) with 5% defibrinated sheep blood. Standard procedures included Gram staining, motility evaluation, oxidase testing, and assessment of growth under adverse conditions. Campylobacter jejuni ATCC 33560 was used as a positive control to validate the procedure.

2.3. Molecular detection and quantification

To evaluate the potential application of qPCR in Campylobacter surveillance, we performed a parallel analysis alongside the gold standard culture method on the samples collected. From the initial dilution of each sample stored at −20 °C at the beginning of the culture technique, bacterial DNA extraction was performed according to the recommendations of the MagMAX^™ Viral/Pathogen Nucleic Acid Isolation Kit (Applied Biosystems, Thermo Fisher Scientific, TX, USA). It was carried out from a volume of 200 μL composed by 100 μL of sample and 100 μL of nuclease-free water. In addition, 200 μL of nuclease-free water was used as a negative control to validate the procedure. At the end of the extraction, 50 μL of eluate from each sample was evaluated through a qPCR performed immediately to reduce DNA loss related to storage, freezing and thawing.

The specific qPCR assay described by Josefsen et al. (2004) was used to detect and quantify Campylobacter spp. genomic copies (GC). Briefly, a reaction mix was prepared in a final volume of 30 μL using 15 μL of TaqMan^® Environmental Master Mix 2.0 (Applied Biosystems, Thermo Fisher Scientific, WA, UK), 0.44 μM of forward primer, 0.48 μM of reverse primer, 0.02 μM of probe and 10 μL of template DNA. All qPCR assays were performed in a QuantStudio^™ 3 Real-Time PCR System (Thermo Fisher Scientific, USA) with the following temperature and time conditions: incubation at 50°C for 2 minutes, initial denaturation at 95°C for 3 minutes and 45 cycles of denaturation at 95°C for 15 seconds, followed by annealing and elongation at 58°C for 1 minute. Automatic settings for threshold and baseline for the determination of positive samples were used. Standard curve was prepared from gBlocks^® synthetic DNA fragments (Integrated DNA Technologies, Inc., Coralville, IA, USA) and in a 12-point dilution series (10⁹ to 10⁻² GC). Each sample was analyzed in sextuplicate (two replicates for the undiluted sample and two replicates for each 10-fold and 100-fold dilution), in order to evaluate the effect of the inhibitory substances. In addition, a standard curve and negative controls was included in each assay. Bacterial DNA extraction, reaction mix preparation, and plate loading were performed in different environments to avoid cross contamination.

2.4. Statistical analysis

The prevalence of positive samples by region and overall across all 3 regions was calculated for both results from culture and qPCR. Differences in prevalence between regions were analyzed using chi-square test and confidence intervals for both techniques. To assess the difference in quantitative results between regions, Kruskal-Wallis and Nemenyi multiple comparison tests were performed using the PMCMRplus package (Pohlert, 2022). Concordance between culture and qPCR techniques was evaluated using McNemar’s test and Kappa statistic for qualitative results, while Spearman’s correlation was used for the quantitative results.

Furthermore, to assess the association between market stalls infrastructure and hygienic conditions and the proportions of positive samples by culture and qPCR, crude and adjusted prevalence ratios (PR) were estimated with their respective 95% confidence intervals (95% CI) through Modified Poisson regression models incorporating robust error estimates (Zou, 2004). The adjusted analysis was made based on the region variable since it was considered a potential confounder. If zero observations are found in any variable, they were randomly replaced with a minimum value (1) for the calculation of the PR. Additionally, the U Mann-Whitney test were used to evaluate the relationship between stalls characteristics and quantitative results from culture and qPCR. To assess the relationship between level of GC obtained by qPCR and the results obtained by culture (the gold standard), a Receiver Operating Characteristic (ROC) curve analysis was performed using the maximum sum of sensitivity and specificity to determine an optimal cut-off point.

Statistical analysis of the collected data in this study was performed using R version 4.2.1 (R-Core-Team, 2015). All the quantifications of colony-forming units per gram (CFU/g) and GC per gram (GC/g) were expressed as the transformation of the logarithm in base 10 (log₁₀).

2.5. Quantitative microbiological risk assessment (QMRA)

The QMRA utilized the enhanced swift Quantitative Microbial Risk Assessment Model (sQMRA), specifically the sQMRA2 version developed by Chardon and Evers (2017) implemented in @Risk (Palisade Corp., NY, USA). This model tracks pathogen levels from retail through consumption, estimating the number of human illness cases. In this study, the chosen sQMRA model was used to simulate the number of cases of human campylobacteriosis in the regions studied and to evaluate possible control strategies. The input parameters needed to run the model, reflecting the situation in Peru, included Campylobacter prevalence and quantification results obtained by bacteriological culture of chicken meat purchased at traditional markets following the methodology described in section 2.2. Only culture results were utilized, as the model requires the quantification of viable bacteria to accurately represent the actual risk to the population. After identifying the regions with positive Campylobacter culture results, a literature search was conducted on population characteristics and chicken meat consumption, primarily using national surveys and censuses. Moreover, due to the sampled regions having very different climatic conditions, population size, and proportion of inhabitants owning refrigerators (INEI, 2017), the sQMRA was made region-specific. In addition, because of the knowledge gaps in the storage practices at consumer level (i.e. proportion of chicken portions stored at room temperature, in the fridge, and in the freezer by consumers) and the parameters defining the cross-contamination events (i.e. proportion of chicken portions resulting in cross-contamination in consumers’ kitchens), different scenarios were run to evaluate the relative impact of these factors.

Scenarios for “cross-contamination” were established to evaluate the risk of Campylobacter transmission during chicken preparation in consumer kitchens. These scenarios assumed that 20%, 40%, 60%, and 80% of chicken portions could potentially cause cross-contamination, where bacteria transfer from the meat to the environment, leading to ingestion through raw salads, kitchen utensils, cutting boards, water, and other surfaces. The factor “storage condition” included three levels where the relative impact of the refrigeration was simulated assuming that 10% (A), 50% (B) and 90% (C) of the people who own refrigerators freeze the chicken. In the model it was assumed that people who do not have a refrigerator store the chicken at room temperature. In addition, several scenarios were simulated in which the current prevalence of contaminated chickens in retail is reduced by 20%, 40%, 60% and 80%; these scenarios were included to evaluate the impact of hypothetical control options in the number of human cases.

The values characterizing the scenarios for cross-contamination and prevalence were selected to reflect situations of different low to high rates of contamination but without being too extreme in both directions. Values for the storage conditions scenarios were selected to reflect situations where about half (50%), almost none (10%) and almost all (90%) of the people owning a refrigerator would store the chicken in the freezer.

Each simulated scenario consisted of 10,000 iterations and the established outcome for the simulated scenarios were the cases of illness due to Campylobacter (campylobacteriosis) per 100,000 inhabitants per year. The pathogen-specific input parameters were taken from the references used by Chardon and Evers (2017) in the “Campylobacter on chicken fillet from The Netherlands” model”. They include growth and inactivation characteristics of Campylobacter based on storage conditions, probability of survival during heating, dose-response parameters and probability of illness given infection.

Ethics clearance was obtained from the Institutional Research Ethics Committee of the Universidad Peruana Cayetano Heredia (Certificate N°516-22-19).

3. Results

3.1. Detection and quantification of Campylobacter spp. in chicken meat from traditional markets

Campylobacter spp. were detected in 27.8% (25/90) of samples by culture, with an enumeration mean of 0.92 log₁₀ CFU/g (range between 0 – 4.73 log₁₀ CFU/g). On the other hand, qPCR detected a higher proportion of positive samples at 75.6% (68/90) with an enumeration mean of 3.69 log₁₀ GC/g. These mean quantifications, which include both positive and negative samples, are detailed in Tables 1 and 2. These tables also include mean quantifications of positive samples as they serve as an input parameter in the QMRA model.

Table 1:

Summary statistics of the results of chicken meat samples from Peruvian traditional markets analyzed by culture for Campylobacter spp. The results are shown by region studied and in total.

	Number of samples	Positive	Proportion (95% CI)	Enumeration mean (log10 CFU/g) (Min-Max)	Enumeration mean of positive samples (log10 CFU/g) (Min-Max)
Huancayo	30	9	30.0% a (17% – 48%)	0.99 ab (0 – 4.73)	3.29 (2.80 – 4.73)
Huaral	30	16	53.3% a (36% – 70%)	1.77 b (0 – 4.36)	3.32 (2.14 – 4.36)
Tumbes/Piura	30	0	0% b (0% – 11%)	0 a	0
Total	90	25	27.8% (20% – 38%)	0.92 (0 – 4.73)	3.31 (2.14 – 4.73)

95% CI = 95% confidence intervals

Superscripts ^a and ^b next to the proportion of positives correspond to the differences found when comparing the confidence intervals and superscripts ^a and ^b next to the enumeration mean correspond to the differences found using Kruskal Wallis post hoc test.

Table 2:

Summary statistics of the results of chicken meat samples from Peruvian traditional markets analyzed by qPCR for Campylobacter spp. The results are shown by region studied and in total.

	Number of samples	Positive	Proportion (95% CI)	Enumeration mean (log10 GC/g) (Min-Max)	Enumeration mean of positive samples (log10 GC/g) (Min-Max)
Huancayo	30	25	83.33% a (66% – 93%)	3.21 a (0 – 4.82)	3.85 (3.02 – 4.82)
Huaral	30	30	100% a (89% – 100%)	5.80 b (4.48 – 6.84)	5.80 (4.48 – 6.84)
Tumbes/Piura	30	13	43.33% b (27% – 61%)	2.07 a (0 – 5.87)	4.78 (3.61 – 5.87)
Total	90	68	75.56% (66% – 83%)	3.69 (0 – 6.84)	4.89 (3.02 – 6.84)

95% CI = 95% confidence intervals

Superscripts ^a and ^b next to the proportion of positives correspond to the differences found when comparing the confidence intervals and superscripts ^a and ^b next to the enumeration mean correspond to the differences found using Kruskal Wallis post hoc test.

Regarding the proportions of positive samples obtained through culture and qPCR, a significant difference was found between Tumbes/Piura and Huancayo and between Tumbes/Piura and Huaral for both techniques (superscripts in the proportion column in Table 1 and 2). Additionally, when analyzing Campylobacter concentrations by culture, a significant difference in medians was found between Huaral and Tumbes/Piura, while, by qPCR, the same difference was found but there was also a difference between Huancayo and Huaral.

3.2. Comparison between culture and qPCR results for Campylobacter spp. and ROC curve analysis

The Kappa coefficient for agreement between the qualitative results was 0.22 (95% CI: 0.12–0.33), reflecting slight agreement between the procedures when interpreted qualitatively. This poor agreement is also reflected in the results of the McNemar test, which provides evidence against the null hypothesis of lack of difference in the proportion of discordant pairs (p < 0.01) (Table 3). On the other hand, when the results are treated as numerical, there is a moderate positive correlation between the outcomes of the two methods (Spearman’s Rho = 42.4%; p < 0.05).

Table 3:

Contingency table of the positive-negative results for Campylobacter spp. obtained by culture and qPCR of the chicken meat samples from Peruvian traditional markets in the evaluated regions.

	Culture		Total
qPCR	Positive	Negative
Positive	25	43	68
Negative	0	22	22
Total	25	65

Parameters of qPCR for diagnosis of positive samples (using culture as gold standard): 100% sensitivity, 34% specificity, 37% positive predictive value, and 100% negative predictive value.

Regarding the ROC curve analysis, the results show that a cut-off of 5.37 log₁₀ GC/g would achieve a sensitivity of 52% (95% CI: 0.31–0.72) and a specificity of 85% (95% CI: 0.74–0.92) for qPCR (Supplementary Figure S1). Therefore, offering, in settings with the same prevalence that was found in this study, the possibility to detect culture-negative samples with a negative predictive value of 82% if the output of the qPCR is below that cut-off. Table 3 shows the results obtained considering as a positive sample for qPCR any amplified DNA sample before the automatically configured threshold and baseline whereas Table 4 shows the results that would be obtained by qPCR if the optimal cut-off point of 5.37 log₁₀ GC/g was used.

Table 4:

Contingency table of the positive-negative results applying the cut-off point of 5.37 log₁₀ GC/g estimated through ROC curve analysis for qPCR quantifications.

	Culture		Total
qPCR	Positive	Negative
Positive	13	10	23
Negative	12	55	67
Total	25	65

Parameters of qPCR for diagnosis of positive samples (using culture as gold standard): 52% sensitivity, 85% specificity, 57% positive predictive value, and 82% negative predictive value.

3.3. Factors associated with Campylobacter presence and quantification in chicken meat from traditional markets

Traditional market stall characteristics were analyzed for their association with Campylobacter presence, determined by culture and qPCR. In the crude analysis of the culture results, significant variables (p-value < 0.05) included stalls with a refrigerator for meat storage and those with tap water. After adjusting for region, stalls with tap water presented a 3.63-fold increased prevalence of Campylobacter compared to those without tap water. Table 5 presents the results of both crude and adjusted analysis using modified Poisson regression. No significant associations were found when the same analysis was performed with the qPCR results.

Table 5:

Crude and adjusted prevalence ratios (PR) of traditional market stall characteristics for the presence of Campylobacter spp. by culture calculated using modified Poisson regression analysis.

Variables	Groups		Crude analysis		Adjusted analysis**
	Positive (n=25)	Negative (n=65)	PR (95% CI)	p-value	PR (95% CI)	p-value
Acceptable infrastructure and hygiene
Yes	28.95% (22)	71.05% (54)	Ref		Ref
No	21.43% (3)	78.57% (11)	0.74 (0.25 – 2.17)	0.58	1.23 (0.46 – 3.25)	0.68
Stall only sell chicken meat
Yes	33.33% (25)	66.67% (50)	Ref		Ref
No	0% (0)*	100% (15)	0.20 (0.03 – 1.40)	0.10	0.62 (0.94 – 4.09)	0.62
Sample ready to sell (i.e. no cutting or handling required)
Yes	35.48% (11)	64.52% (20)	Ref		Ref
No	23.73% (14)	76.27% (45)	0.67 (0.34 – 1.30)	0.24	0.82 (0.43 – 1.57)	0.55
Degree of cleanliness of the table where the meat is cut
Clean	12.50% (2)	87.50% (14)	Ref		Ref
Moderately clean	25.81% (8)	74.19% (23)	2.07 (0.48 – 8.82)	0.33	2.69 (0.72 – 10.13)	0.14
Dirty	34.88% (15)	65.12% (28)	2.79 (0.70 – 11.12)	0.15	2.41 (0.68 – 8.53)	0.17
Degree of cleanliness of the cleaning cloth used on the surfaces
Clean	12.50% (2)	87.50% (14)	Ref		Ref
Moderately clean	27.27% (12)	72.73% (32)	2.18 (0.53 – 8.91)	0.28	2.69 (0.73 – 10.00)	0.14
Dirty	36.67% (11)	63.33% (19)	2.93 (0.72 – 11.93)	0.13	3.09 (0.84 – 11.36)	0.09
Stall with a refrigerator to store meat
Yes	22.54% (16)	77.46% (55)	Ref		Ref
No	47.37% (9)	52.63% (10)	2.10 (1.10 – 4.03)	0.025	1.77 (0.96 – 3.27)	0.07
Stall with tap water
No	18.18% (8)	81.82% (36)	Ref		Ref
Yes	36.96% (17)	63.04% (29)	2.03 (0.97 – 4.26)	0.06	3.63 (1.98 – 6.63)	<0.001

^*The value 0 was randomly changed to 1 to calculate the PR presented.

^**Adjusted by region using modified Poisson regression analysis.

The relationship between the quantifications of Campylobacter obtained by culture and qPCR and the characteristics of traditional market stalls was analyzed using the U Mann-Whitney test. The analysis revealed that the stalls without a refrigerator for meat storage had higher concentrations of CFU/g and GC/g compared to those with a refrigerator. Additionally, stalls with tap water had higher concentration of CFU/g and GC/g compared to those without tap water. The details of the stall characteristics for which the study provided evidence of an association with Campylobacter contamination are shown in Table 6.

Table 6:

Analysis of traditional market stall characteristics and quantitative Campylobacter results by culture and qPCR using the U Mann-Whitney test, showing only variables with strong evidence of association.

	Culture		qPCR
Variables	Mean (Log10 CFU/g)	U Mann-Whitney test p-value	Mean (Log10 GC/g)	U Mann-Whitney test p-value
Stall with a refrigerator to store meat
Yes	0.77	0.05	3.47	0.032
No	1.49		4.52
Stall with tap water
Yes	1.24	0.029	4.48	< 0.001
No	0.59		2.87

3.4. Simulation of human campylobacteriosis cases and control strategies assessment using QMRA modeling

To carry out this analysis, both Campylobacter prevalence and quantifications obtained by culture in this study were used. Since the Tumbes/Piura region did not have positive results by culture, only scenarios for Huaral and Huancayo regions were simulated. After identifying the regions with positive results, the input parameters specific to the QMRA scenarios for each region were developed based on both our study’s findings and a literature review. These parameters are detailed in Tables 7, 8 and 9.

Table 7:

Input parameter values with reference/source used in the QMRA model.

Scope
Pathogen	Campylobacter
Food product	Chicken meat
Population specification	Tables 8 and 9
Population size	Tables 8 and 9
Consumption period in days	365
Question	Value	Unit	Ref
1 Portions consumed
Point estimation: portions per person per month (pppm)	30.42	/pppm	(MIDAGRI, 2022)
2 Pathogen prevalence in retail
Point estimation of prevalence	Table 9	-	-
3. Portion size
Portion size: mean	139.55	g	(MIDAGRI, 2022)
Portion size: st dev	6.71	g
4 Pathogen concentration
Concentration: mean	Table 9	log10 (CFU/g)	-
Concentration: st dev	Table 9	log10 (CFU/g)
5 Storage conditions
% of portions stored in room	Tables 8 and 9	-	-
% of portions stored in the fridge	Tables 8 and 9	-	-
% of portions stored in freezer	Tables 8 and 9	-	-
7 Cross-contamination parameters
% of portions possibly causing cross-contamination	Tables 8 and 9	-	-
8 Heating categories
% of portions done	95	%	Authors’ estimate
% of portions undercooked	5	%
% of portions raw	0	%

Table 8:

Storage conditions and cross-contamination scenarios simulated in the QMRA model.

Storage conditions scenarios		Region		Ref
		Huaral	Huancayo
A	% of portions stored in the fridge	56.93%	32.66%	(INEI, 2017)
	% of portions stored in freezer	6.33%	3.63%
B	% of portions stored in the fridge	31.63%	18.15%
	% of portions stored in freezer	31.63%	18.15%
C	% of portions stored in the fridge	6.33%	3.63%
	% of portions stored in freezer	56.93%	32.66%
A, B, C	% of portions stored at room temperature	36.75%	63.71%

A, B and C are, respectively, scenarios where 10%, 50% and 90% of the people have a refrigerator in which they keep their chicken portions in the freezer. For each scenario of storage conditions, four cross-contamination scenarios were simulated (20%, 40%, 60% and 80%).

Table 9:

Prevalence scenarios simulated in the QMRA model.

	Prevalence scenarios					Ref
Input parameters	Current prevalence	20% reduction	40% reduction	60% reduction	80% reduction
Region of Huaral
Population size			183,898			(INEI, 2017)
Pathogen prevalence in retail
Point estimation of prevalence	53.3%	42.6%	32%	21.3%	10.7%	Results of this study
Pathogen concentration
Concentration: mean		3.32 log10 (CFU/g)				Results of this study
Concentration: st dev		0.58 log10 (CFU/g)
Storage conditions
% of portions stored at room temperature		36.75% for all scenarios
% of portions stored in the fridge		56.93% for all scenarios				(MIDAGRI, 2022)
% of portions stored in the freezer		6.33% for all scenarios
Cross-contamination parameters
% of portions possibly causing cross-contamination		10% for all scenarios				Authors’ estimate
Region of Huancayo
Population size			545,615			(INEI, 2017)
Pathogen prevalence in retail
Point estimation of prevalence	30%	24%	18%	12%	6%	Results of this study
Pathogen concentration
Concentration: mean		3.29 log10 (CFU/g)				Results of this study
Concentration: st dev		0.59 log10 (CFU/g)
Storage conditions
% of portions stored at room temperature		63.71% for all scenarios
% of portions stored in the fridge		32.66% for all scenarios				(MIDAGRI, 2022)
% of portions stored in the freezer		3.63% for all scenarios
Cross-contamination parameters
% of portions possibly causing cross-contamination		10% for all scenarios				Authors’ estimate

Using the input parameters described above, we employed the sQMRA model to simulate annual campylobacteriosis cases per 100,000 inhabitants in each region. The simulations considered different scenarios, including chicken portion storage conditions, the percentage of portions that could cause cross-contamination at the consumer level, and the reduction of the prevalence identified in this study of Campylobacter-contaminated chicken sold in traditional markets. This approach allowed us to quantify the impact of these parameters on the annual incidence rate of campylobacteriosis in each region. As a result, in the most conservative scenario, the annual number of campylobacteriosis cases was around 104,405 per 100,000 habitants.

Additionally, through simulated scenarios, we evaluated how control strategies focusing on improving storage conditions, reducing cross-contamination, and lowering the prevalence of positive samples could potentially reduce case numbers. Regarding the simulated scenarios for storage conditions, the median number of campylobacteriosis cases decreases as more people store chicken meat in the freezer. When the percentage of refrigerator owners who store chicken at freezing temperatures increases from 10% to 90%, campylobacteriosis cases decrease by 40% in Huaral and by 22% in Huancayo, approximately. Additionally, when the percentage of chicken servings likely to cause cross-contamination at the consumer level is reduced, the decrease in median campylobacteriosis cases is similar across all three storage condition scenarios in both regions. Cases approximately decrease by 25% when cross-contamination is reduced from 80% to 60%, by 50% when reduced from 80% to 40%, and by 75% when reduced from 80% to 20% (Figure 1).

Fig. 1.

Simulated cross-contamination scenarios with different storage conditions. The X-axis represents cross-contamination scenarios at the consumer level, with each color representing different percentages of people storing chicken meat at freezing temperatures. Points indicate the median incidence rate of campylobacteriosis, and error bars represent the 90% credible intervals for each scenario.

Finally, in the simulated scenarios where the current prevalence of contaminated chicken meat at traditional markets is decreased, the reduction in median campylobacteriosis cases closely matched the percentage decrease in the prevalence of contaminated chicken meat for both regions. Specifically, when the current prevalence is reduced by 20%, 40%, 60%, and 80%, the median cases decrease by approximately 20%, 41%, 61%, and 81%, respectively (Figure 2).

Fig. 2.

Simulated scenarios for the percentage reduction of the current prevalence of contaminated chicken meat for Huaral and Huancayo. The X-axis represents current prevalence levels in markets and their reduction by 20%, 40%, 60%, and 80%. Points indicate the median incidence rate of campylobacteriosis, and error bars show the 90% credible intervals for each scenario.

4. Discussion

In this study, Campylobacter spp. were detected and quantified in chicken meat from traditional markets in Peru and QMRA was used to assess the risk of consuming contaminated chicken in the evaluated regions. Campylobacter was studied since it is a major cause of gastroenteritis globally, and because despite being associated with an outbreak of Guillain-Barre syndrome in Peru in 2019, it remains poorly understood in the country (CDC MINSA, 2023; Munayco et al., 2020). Furthermore, reliable quantitative data on viable pathogenic bacteria are necessary for assessing food-related health risks and understanding bacterial interactions to inform decision-making (Newell & Fearnley, 2003). Although Peruvian regulations require an absence of Campylobacter in chicken sold for consumption (MINSA-Peru, 2008), no efforts have been made yet for the situational diagnosis or quantification of this microorganism. In contrast, the European Union sets a limit of up to 1000 CFU/g in broiler carcasses, which has been shown to reduce human campylobacteriosis incidence by two-thirds (Swart, A., Mangen, M-J., Havelaar, 2013). Given that eradicating this pathogen at the farm level is not yet possible in European countries (Georgiev et al., 2017), a similar situation could be expected in Peru.

4.1. Campylobacter spp. in chicken meat from traditional markets

This study found a Campylobacter concentration of 3.31 log₁₀ CFU/g in 27.8% of positive samples through the culture technique. These results align with concentrations reported globally, including 3 log₁₀ CFU/g in Eastern Europe (Mäesaar et al., 2014), 2.7 log₁₀ in United Arab Emirates from 28.6% of contaminated samples (Habib et al., 2022) and 3.1 log₁₀ CFU/g in Brazil (Pozza et al., 2020). On the other hand, we found higher concentrations than in Australia which reported 1.8 log₁₀ CFU/g (Habib et al., 2018) and Lithuania which reported concentrations between 1.99 – 2.11 log₁₀ CFU/g (Bunevi et al., 2010). However, there are limited studies worldwide on quantifying Campylobacter in chicken meat using culture techniques, and to our knowledge, this is the first study of its kind in Peru.

Campylobacter concentration in chicken meat depends on initial flock contamination, handling during transport, slaughter process, time between slaughter and sale, and storage methods until consumption (Newell et al., 2017). Although our analysis focused on traditional markets, it is important to consider the time the chicken is being transported since crates are an important factor in their exposure to pathogens and subsequent colonization. For example, chicken consumed in Huancayo are transported alive for approximately 10–12 hours to be immediately slaughtered, this could increase stress in chickens which might reduce their immunocompetence making them more susceptible to Campylobacter. Research indicates that transport stress increases Campylobacter excretion rates, with colonization occurring within 4 hours in contaminated crates (Rasschaert et al., 2020). This was not found in poultry transported for only 2 hours (Slader et al., 2002). In contrast to Huancayo, chicken from Huaral experience shorter transport times, suggesting other factors may influence Campylobacter concentrations found. Therefore, transport crates may play an important role in cross-contamination of chicken carcasses in the markets from Huancayo, but less so in the markets from Huaral.

Campylobacter was not detected in Tumbes/Piura through culture technique. However, the genetic material found through qPCR demonstrates the possibility of the presence of dead, non-viable bacteria, or concentrations below the detection limit of the culture method. This region has an average annual temperature of 24.2 °C (INEI, 2022b), so farms usually have more ventilation control, and most traditional market stalls have refrigerators, which may reduce bacterial concentrations in the analyzed samples. On the other hand, the viability of Campylobacter could have been affected at some point between the slaughterhouse and the retail sale. Fresh chicken sold may have been frozen without consumer notification, and this could have led to Campylobacter entering a viable but non-culturable state, which has been described under stress conditions such as low temperatures and aerobic environments, making it undetectable by culture techniques (Lv et al., 2020). It would be important to continue monitoring the presence and load of viable Campylobacter in chicken meat, as Tumbes/Piura is a region where an outbreak of Guillain-Barre Syndrome was reported shortly after our evaluation (CDC MINSA, 2023).

qPCR detected more positive samples than culture, but the results should be interpreted with caution since they may not reflect actual population risk. This is because the qPCR assay used in this study detects both viable and non-viable cells, potentially inflating prevalence estimates (Josefsen et al., 2004; Magana et al., 2017). Although promising qPCR assays capable of distinguishing alive from dead cells are under development, they have not yet replaced the gold standard of bacteriological culture (Dubovitskaya et al., 2023; Thornval & Hoorfar, 2021). Thus, qPCR was used in this study to complement culture and to evaluate its potential as a tool for routine Campylobacter surveillance in chicken meat, as discussed in Section 4.3.

4.2. Role of traditional market stall characteristics in Campylobacter spp. prevalence and quantification

Regarding the traditional market stall characteristics, access to tap water was associated with higher Campylobacter prevalence and higher quantifications by both culture and qPCR analysis. Studies have demonstrated that this pathogen can rapidly spread on food and nearby surfaces (Habib et al., 2020; Sarjit & Dykes, 2017), suggesting that tap water in the stall might represent a risk factor by facilitating the spread of bacteria when washing utensils used on contaminated carcasses, such as chopping boards, knives, and cleaning cloths. In fact, campaigns have highlighted the importance of not washing chicken at home to reduce cross-contamination risks (Henley et al., 2016). Moreover, sellers could be washing the chicken meat to remove organic residues, making it more appealing to consumers. On the other hand, not having a refrigerator for chicken storage was related to higher Campylobacter quantifications by culture and qPCR, despite not being associated with a higher prevalence. This could be attributed to refrigeration reducing bacterial concentrations by up to 0.81 log₁₀ CFU per gram (Bhaduri & Cottrell, 2004), while freezing meat for a few days decreases Campylobacter concentration by up to 1.4 log₁₀ CFU/g and if it is done for 28–31 days decreases it by up to 2.33 – 2.9 log₁₀ CFU/g (Georgsson et al., 2006; Maziero & de Oliveira, 2010). In this manner, traditional market stall characteristics might be playing an important role in Campylobacter transmission to consumers through contaminated chicken meat.

4.3. Potential use of qPCR in Campylobacter surveillance

Surveillance of Campylobacter spp. in many countries is hindered by the complex methodology required for isolation and quantification via culture, which demands immediate processing after sampling and can take up to seven days (Magana et al., 2017). Conversely, conventional molecular techniques enable detection within a day but only identify genetic material, leaving uncertainty regarding the pathogen’s viability in the meat and its potential risk for consumers. Our evaluation, using ROC curve analysis, suggests a cut-off point of 5.37 log₁₀ GC/g to determine if a sample would be negative by culture with a specificity of 85% (95% CI: 0.74–0.92). This could allow a rapid assessment of chicken meat batches using molecular techniques, ensuring they meet safety standards before reaching retail markets, while culture analysis is conducted concurrently. However, the proposed cut-off point requires validation before being implemented as a complementary diagnostic tool in Campylobacter surveillance.

4.4. Simulated cases of human campylobacteriosis through sQMRA based on prevalence and quantification data from traditional markets

In our QMRA simulations we obtained risks that were 600 times higher those reported in the baseline “Campylobacter on chicken fillet from The Netherlands” model (Chardon & Evers, 2017). In Peru, Consumption, storage conditions, and cross-contamination levels for chicken differ from those in the Netherlands. A hypothetical scenario using only the prevalence and concentration of Campylobacter from this study but maintaining Dutch parameters shows a risk 35 times higher, highlighting the impact of bacterial load. However, the remaining impact of this disease may also stem from chicken being Peru’s primary protein source, with per capita consumption 15 times higher than in the Netherlands (MIDAGRI, 2022). In Peru as in other low- and middle-income countries, washing raw chicken to remove visible residues is common, as many believe it makes it safer (Joao et al., 2021). However, this practice can spread bacteria to other kitchen surfaces (Sarjit & Dykes, 2017), especially if the same cutting board is used for raw chicken and vegetables, increasing transmission risk. This reflects the paramount importance of maintaining good kitchen hygiene practices to avoid cross-contamination during handling of Campylobacter-contaminated chickens to reduce the risk of campylobacteriosis (Eriksson et al., 2023).

This analysis indicates that, using conservative parameters and the bacterial load identified by culture in our study, every consumer in the evaluated regions would develop campylobacteriosis at least once annually. This alarming finding highlights the potential severity of the problem. On the other hand, limitations in Peru’s syndromic surveillance of acute gastroenteritis and foodborne illnesses lead to underreporting of campylobacteriosis cases, making it difficult to fully rule out annual incidence rates in the simulated scenarios. Consequently, the QMRA section in this study focuses on discussing mainly the control strategies evaluated in the analysis rather than the simulated incidence rates. Nevertheless, it is important to emphasize the need to characterize the agent in both acute gastroenteritis outbreaks and routine surveillance.

4.5. Control strategies for human campylobacteriosis evaluated through QMRA simulations

Among the campylobacteriosis control strategies evaluated through QMRA simulations, reducing cross-contamination at the consumer level and increasing the fraction of the population with refrigerators that store chicken at freezing temperatures are realistic options. These strategies could be implemented through educational campaigns without requiring investment in equipment. This study assumes that the percentage of chicken portions causing cross-contamination in Peru ranges from 40 – 80%. Despite this high figure, we demonstrated that reducing cross-contamination from 80% to 20% could decrease the median annual incidence rate of campylobacteriosis by up to 75%. This is consistent with previous studies indicating that effective control of cross-contamination markedly reduces the risk of Campylobacter transmission (Cardoso et al., 2021; Lindqvist & Lindblad, 2008). On the other hand, while having a refrigerator improves quality of life (Karlsson & Subramanian, 2023), only 55.1% of Peruvians currently own one, and expanding this coverage would require substantial economic investment (INEI, 2022a). Nevertheless, increasing the proportion of people who store chicken in the freezer from 10% to 90% while maintaining the current number of refrigerator owners could decrease campylobacteriosis cases by nearly 40%, likely due to the significant reduction in pathogen concentration after freezing (Georgsson et al., 2006). This finding aligns with the studies that demonstrate that cold and freezing temperatures effectively reduce Campylobacter loads, thus lowering the risk of infection (Nastasijevic et al., 2020; Thames & Sukumaran, 2020). Therefore, it is necessary that corresponding authorities implement education programs to promote proper raw chicken meat handling practices, raise awareness about the role of cross-contamination in the transmission of campylobacteriosis, and encourage the storage of chicken in the freezer to reduce the incidence rate.

The percentage reduction in simulated cases relative to the decrease in the prevalence of contaminated chicken on traditional markets highlights the importance of efforts to reduce the level of contamination in primary production in order to provide safer meat to consumers. Nevertheless, even with an increase of biosecurity at the farm level to extremely high standards, a high proportion of poultry batches would still be contaminated (Georgiev et al., 2017). This issue lacks a single solution; instead, there are multiple vulnerable points in the production chain where the different stakeholders, including the authorities, can intervene. Therefore, improvements must be developed at the level of the different critical points throughout the production chain to mitigate the public health impact of this pathogen.

Our study had limitations, including the potential selection bias related to the traditional markets sampled, which could overestimate the prevalence of Campylobacter. To minimize it, we selected markets with the highest attendance in each region. Furthermore, this study did not differentiate between Campylobacter species in the chicken samples, and the sQMRA model used was parameterized with C. jejuni dose-response data. As species distribution in the evaluated regions is unknown, extrapolating disease impact to the Peruvian context without specific epidemiological data introduces a potential bias that could overestimate or underestimate the results of the presented QMRA.

It is also important to acknowledge that the QMRA outputs rely on a number of simplifications and assumptions and require further validation in real-world contexts. Despite this, these results offer a valuable foundation for understanding campylobacteriosis control dynamics. The reliance on input parameters that may not fully reflect the diverse consumer practices regarding cross-contamination and storage could affect the applicability of the results. Variations in consumer behavior across different regions may lead to over- or underestimations of the effectiveness of the proposed interventions. Consequently, current information on cross-contamination levels and chicken meat storage practices at the consumer level in Peru is needed for more accurate estimations of campylobacteriosis cases.

Finally, the QMRA model used in this study does not consider immunity due to previous Campylobacter infections potentially leading to an overestimation of simulated cases. While recurrent infections can occur in adults, exposure to this pathogen at an early age in endemic settings may induce protective immunity (Kaakoush et al., 2015). Therefore, future studies should focus on the efficiency and duration of immunity produced by Campylobacter to enhance available QMRA models.

5. Conclusions

In conclusion, Campylobacter is present in chicken meat sold in traditional markets in the three regions evaluated, at least by molecular analysis. Culture results indicated a high mean concentration of CFU/g, which does not comply with any of the sanitary regulations for the sale of chicken meat. Additionally, our findings indicate that improper use of tap water and inadequate refrigeration practices in market stalls are likely contributors to the observed Campylobacter contamination. This represents an important risk to public health, which was corroborated with the development of QMRA models that demonstrated that cross-contamination and storage conditions at the consumer level play an important role in the transmission dynamics of human campylobacteriosis. In light of the high levels of contamination reported in this study and the control strategies assessed in the QMRA, it is necessary to enhance food hygiene practices along the poultry production chain and among consumers to mitigate the risk of acquiring campylobacteriosis. While the outputs of our QMRA model need further validation, they represent a valuable starting point for understanding these dynamics. For these reasons, the findings of this study could contribute to the establishment of prevention and control measures by the corresponding authorities.

Abbreviations:

QMRA

Quantitative Microbiological Risk Assessment

sQMRA

swift Quantitative Microbial Risk Assessment Model

qPCR

quantitative real-time polymerase chain reaction

CFU

colony-forming units

GC

genomic copies

DNA

deoxyribonucleic acid

mCCD

modified charcoal cefoperazone deoxycholate

ISO

International Organization for Standardization standards