Prevalence and risk factors for Salmonella infection in layer flocks in Israel between 2018 and 2023

Abstract

Non-typhoidal Salmonella infections are a leading food-borne zoonosis in which poultry and egg products play an important epidemiological role. Identification of risk factors for Salmonella infections is crucial for reducing Salmonella prevalence and protecting public health. Although Salmonella prevalence in layers is actively monitored, analysis of potential risk factors, particularly those related to proximity to other livestock farms, has not been conducted previously. This study aimed to estimate the prevalence and identify risk factors, especially proximity to dairy farms, for Salmonella and its predominant serovars infections in layer flocks in Israel between 2018 and 2023.

Overall, 24,500 sampling events from a census of 6,076 laying hen flocks, which were collected as part of the national active Salmonella surveillance program, were included in the cross-sectional analysis. We estimated overall Salmonella prevalence, the prevalence of serovars S. Enteritidis (SE) and S. Typhimurium (ST), and the association between multiple risk factors and infection. Univariable and multivariable mixed-effects regression models were applied, with nested random effects to account for shared locality, farm ownership, and repeated measures.

The overall prevalence of positive Salmonella sampling events was 17.74 % (4,347/24,500), and the cumulative flock-level prevalence was 42.56 % (2,586/6,076). SE and ST were detected in 2.98 % (181/6,076) and 2.09 % (127/6,076) of flocks, respectively. Older bird age (>85 weeks), larger flock size (>10,000 hens), caged housing system, and sampling during fall were significantly associated with increased Salmonella detection in the multivariable models (p < 0.05). SE and ST detection was associated with older bird age (>85 weeks) and sampling by the Israeli Veterinary Services vs. Egg and Poultry Board inspectors.

These findings emphasize the role of flock management and environmental context in Salmonella complex epidemiology in layer flocks. While the distance to dairy farms may not affect the risk of infection, maintaining smaller flocks (under 10,000 hens) in non-caged systems for shorter production cycles could help lower the prevalence of Salmonella in laying hens.

Introduction

Foodborne salmonellosis remains one of the leading zoonotic diseases in developed countries. Most infections are associated with the consumption of animal-derived products contaminated by non-typhoidal Salmonella (NTS) strains, with pork, poultry, and eggs being the primary sources of transmission (CDC, 2018; Chanamé Pinedo et al., 2022; EFSA and ECDC, 2024).

The genus Salmonella, a Gram-negative bacillus in the family Enterobacteriaceae, comprises two species: S. bongori and S. enterica. The latter is further classified into six subspecies: enterica, salamae, arizonae, diarizonae, houtenae, and indica. Strains are further differentiated into serotypes according to three pathogenicity factors (lipopolysaccharides, flagella, and capsular polysaccharides), with over 2,500 serotypes described to date (Grimont and Weill, 2007). Approximately 60 % of known serovars belong to S. enterica subspecies enterica, which can infect both humans and animals (Grimont and Weill, 2007). Despite this diversity, only a small fraction of NTS serovars play a central role in human infections, among them, S. Enteritidis (SE) and S. Typhimurium (ST) are the two most frequently implicated (EFSA and ECDC, 2024; Gal-Mor et al., 2014).

Between 2016 and 2023, 26,605 confirmed cases of human NTS infections were reported in Israel, corresponding to an annual incidence rate ranging between 30.7 and 53.1 cases per 100,000 population (Israeli Central Bureau of Statistics, 2024; Israeli Ministry of Health, 2025; Koch Davidovich, 2021). This rate is considerably higher than rates reported in the European Union (EU) and the United States (US). In the EU, the incidence ranged from 19.7 to 20.7 cases per 100,000 population between 2014 and 2018 (EFSA, 2019) and from 13.5 to 19.5 cases per 100,000 population between 2019 and 2023 (EFSA and ECDC, 2024). In the US, the 2023 incidence was reported as 15.79 cases per 100,000 population (Shah et al., 2024). Such differences in human salmonellosis incidence result from a combination of factors, including variations in reporting systems, data availability, consumption habits, and climatic differences. For example, poultry meat consumption in Israel is nearly 20 % higher than in the US and more than twice that of the EU (OECD data center, 2023), likely due to religious restrictions. Moreover, a higher prevalence of Salmonella in Israeli poultry compared with the EU and the US has been suggested as a contributor to the elevated human incidence (Egg and Poultry Board, 2023). In addition, Salmonella growth may be higher in the warmer Israeli climate, as the bacterium thrives in warmer conditions (Damtew et al., 2024).

Feasey et al. (2012) have previously described three levels of host adaptation in NTS serovars, which is attributed to differences in their genetic repertoire, such as the presence of virulence determinants. In general, the more a serovar is adapted to a specific host species, the more invasive it is and the more severe the disease it causes. Host-restricted serovars, such as S. Typhi in humans and S. Gallinarum in poultry, typically cause severe invasive systemic disease. Host-adapted serovars, such as S. Dublin (cattle) and S. Choleraesuis (swine), primarily affect their adapted species, usually causing milder, less invasive illness compared with host-restricted serovars. In contrast, generalist serovars, including SE and ST, typically cause self-limiting gastrointestinal disease, and many infected animals remain asymptomatic (Merkushova et al., 2023; Wallis, 2006). These two serotypes, SE and ST, are of major global importance, given their role in human infections (EFSA and ECDC, 2024; Shah et al., 2024). Although both are regarded as generalists, SE is predominantly associated with poultry (EFSA Panel on Biological Hazards, 2010) while ST is linked to a broader range of reservoirs, including rodents, reptiles, and wild birds (Dar et al., 2017; Shaji et al., 2023; Wallis, 2006). Moreover, although both can colonize the reproductive system of birds (Gantois et al., 2009; EFSA Panel on Biological Hazards, 2010), in hens, ST is rarely transmitted vertically into eggs, whereas SE transmits more frequently and efficiently in this manner (Gantois et al., 2009; EFSA Panel on Biological Hazards, 2010; Wales and Davies, 2011). This broader host spectrum increases the likelihood of cross-species transmission of ST to and from layer flocks, as was recently demonstrated with the transmission of ST strains from infected poultry flocks to an adjacent dairy farm (Perry et al., 2024). However, to our knowledge, no epidemiological studies have evaluated the possible relationship between the proximity to dairy farms and the detection of Salmonella, particularly ST, in nearby hen flocks.

To address public health risks, Israel established an active surveillance program for Salmonella in poultry farms, based on EU sampling guidelines (EU Commission Regulation, 2011). Breeding farms have been monitored since 1992, and table-egg layer farms, since 2017. Data from the Egg and Poultry Board (EPB) indicate that between 2017 and 2023, the annual prevalence of Salmonella in layer flocks ranged from 18.3 % to 35.5 %. Specifically, SE ranged from 1 % to 2.5 % and ST ranged from 0.9 % to 1.4 % (Egg and Poultry Board, 2022, 2023). These figures are consistently higher than those reported in the EU between 2018 and 2023, where overall Salmonella prevalence in layer flocks ranged from 3.3 % to 4 %, prevalence of SE ranged from 0.88 % to 1.06 % and prevalence of ST ranged from 0.3 % to 0.43 % (EFSA, 2019; EFSA and ECDC, 2021a; b, 2022, 2023, 2024). In the US, SE monitoring is conducted as part of a surveillance program (FDA and HHS, 2009), but the accumulated data are not publicly available. According to a 2013 US Department of Agriculture study, the SE prevalence in US layer flocks was estimated at 1 % (USDA, 2013).

Given the importance of Salmonella in egg production systems, several studies have investigated risk factors for Salmonella contamination in laying flocks. Denagamage et al. (2015) reviewed key environmental and management factors associated with increased risk of Salmonella contamination in the environment of the flock, in eggs, and of SE infection in hens. The following factors were found to increase risk of Salmonella contamination: previous infections, poor hygiene, molting, caging type, in-line egg processing, and vaccination policy. Egg contamination was associated with high levels of manure contamination and diminished when hens were vaccinated with bacterin. Finally, the sampling method used affected the probability of obtaining a positive culture, with pooled dust samples more likely to yield a positive result compared with pooled feces samples. A later study by Samper-Cativela et al. (2023) outlined farm location within Spain, housing type, sampling policy, seasonality, and age of flocks as risk factors for infection. Another recent study by Hobbelen et al. (2024) described the proximity to a positive flock as a risk factor for infection. Most of these studies focused on SE, while ST-specific risk factors remain underexplored. In particular, the impact of proximity to livestock holdings, potential environmental reservoirs of Salmonella, has not yet been evaluated, although swine and cattle were cited in the past as possible sources of contamination in layer flocks (EFSA Panel on Biological Hazards, 2010).

This study evaluated risk factors for Salmonella contamination, including proximity to dairy farms, in Israeli table-egg laying hen flocks between 2018 and 2023, focusing on SE and ST serotypes.

Materials and methods

The study was designed as a cross-sectional study estimating the period prevalence of Salmonella in table-egg laying hen flocks (hereafter: layer flocks) in Israel between 2018 and 2023. Data was obtained from the national active Salmonella surveillance program.

Study population

The study population comprised a census of all laying hen farms in Israel during the study period. Individual flocks were considered the epidemiological unit, and a laying hen farm could host more than one flock over time (consecutively, not simultaneously). A total of 6,099 unique flocks were sampled between 2018 and 2023, yielding 44,524 database entries. Following the removal of duplicated and corrupted data entries (n = 5,244) and consolidating multiple entries (n = 39,280) into unique sampling events, 24,500 sampling events of 6,076 flocks in 1927 farms were included in the analysis (Fig. 1). No a priori exclusion criteria were applied, as all laying flocks were subject to mandatory sampling.

Fig. 1.

Preparation of raw data for analysis. A diagram of the dataset cleanup process performed prior to analysis.

Sampling scheme

Israel has adopted the guidelines of the EU surveillance program based on Commission Regulation (EU) 517/2011 (EU Commission Regulation, 2011) with a few modifications: (i) environmental sampling includes drag swabs, dust swabs, and dust samples instead of pooled faeces, and (ii) sampling of the food business operator is performed by inspectors of the EPB or the Israeli Veterinary Services (IVS, the competent authority), rather than farm staff. Flocks were sampled every 15 weeks, starting at 24 ± 2 weeks of age. The IVS inspectors collected four drag swabs (EP176/IR, Hylabs, Rehovot, Israel), four dust swabs (EP175/IR, Hylabs, Rehovot, Israel), and 100 g dust samples at 39 and 84 weeks. EPB inspectors sampled four drag swabs (EP176/IR, Hylabs, Rehovot, Israel) at each sampling of the remaining time points. This schedule has been maintained since the beginning of Salmonella monitoring in 2017, except during the period from April 2018 to the end of 2019, when the sampling frequency was reduced and flocks were sampled only at 24, 39, 54, and 84 weeks (approximately two tests per flock per year). In addition, out-of-schedule sampling events conducted by the IVS as part of outbreak management were also logged as routine sampling events.

Salmonella detection and isolation

Regardless of the identity of the inspectors performing the sampling (EPB/IVS), all samples were analyzed by the Southern EPB Poultry Health Laboratory according to ISO 6579:2002/Amendment 1:2007, Annex D: "Detection of Salmonella spp. in animal feces and in environmental samples from the primary production stage". In brief, the pooled drag or dust swabs and 25 g dust were each placed in 225 ml buffered peptone water (BP227/225Z, Hylabs, Rehovot, Israel) and incubated for 18 ± 2 h at 37 ± 1°C. Three droplets of each pre-enrichment culture were dropped on the surface of a modified semisolid Rappaport-Vassiliadis (MSRV; BP573, Hylabs, Rehovot, Israel) agar plate containing 0.01 g/l novobiocin. The plates were incubated for 2 × 24 ± 3 h at 41.5 ± 1°C. When Salmonella growth was suspected, cultures were transferred onto a parted Petri plate of Brilliant Green (BG) and Xylose Lysine Deoxycholate agar (XLD; DD074, Hylabs, Rehovot, Israel), followed by incubation for 24 ± 3 h at 37 ± 1°C. Three presumptive Salmonella colonies on BG agar were confirmed biochemically using Hy-Enterotest (TT146, Hylabs, Rehovot, Israel). The same colonies from BG agar and three colonies from XLD agar were streaked on nutrient agar plates (PD040, Hylabs, Rehovot, Israel) and incubated for 24 ± 3 h at 37 ± 1°C before additional characterization of the strains.

Salmonella serotyping

All samples were tested to rule out autoagglutination and then serogrouped for B, C, D, E, G, or polyvalent (Becton, Dickinson and Company, Franklin Lakes, New Jersey, US; Statens Serum Institute, Copenhagen, Denmark). Further identification of ST and SE was performed for isolates from serogroups B and D, respectively, or in cases of polyvalent and autoagglutination reactions in the serogrouping. Serotyping was conducted with the Salmonella Sero-Quick ID kit (SSI Diagnostica, Hillerød, Denmark) according to the manufacturer's instructions, and verified by real-time PCR (Maurischat et al., 2015; O’Regan et al., 2008; Park et al., 2008).

Dataset availability and organization

All test results were electronically recorded in a centralized database, from which a single Excel file containing all data was downloaded on February 7th, 2024, and used for this study. For each sampling event, the following variables were recorded: test number, sampling date, sample type, flock identification number, hen breed, flock size, age in weeks, housing type (cage/non-cage), farm license number, farm location (village name), geographic area, and test results (supplementary Table S1).

Geographical information

Geographic information on the location of all commercial poultry and dairy farms was obtained from the Ministry of Agriculture GIS database (Ben Asher, 2025). QGIS software (QGIS Development Team, 2024) was used to map poultry layer farm locations and calculate Euclidean distances to nearby dairy and poultry farms. Proximity to dairy farms was defined as a dichotomous variable, with a threshold of 300 meters. The density of poultry farms within a 500 m radius was calculated as the number of commercial poultry farms (of any type) within that radius around each layer farm and then dichotomized with a threshold of 10 poultry farms.

Data analysis

The cumulative and annual prevalence of positive samples and flocks for overall Salmonella, SE, and ST were calculated. Annual flock prevalence was calculated after removal of repeated positive results in the same flock within a calendar year. In contrast, since the detection of SE and ST led to the culling of the flock, repeated positive results for these serovars did not occur.

For the purpose of the analyses, a sampling event conducted by the IVS was considered positive if any of the collected samples (i.e., dust, dust swab, or drag swab) were found positive (see results for further details).

Risk factors for infection with Salmonella, SE, and ST were assessed using the entire dataset. In addition, a subset of Salmonella-positive samples only (n = 4,347) was used to assess possible risk factors for SE and ST, in order to reduce the risk of masking possible risk factors due to the large number of negative results (n = 24,319/24,500 and 24,371/24,500 for SE and ST, respectively). However, as no major differences were found between the outcomes of the entire dataset analyses and the positive-samples-only analyses, the results of the former are presented. The outcomes of the latter are detailed in the supplemental material.

Statistical analysis

Associations between the outcome variable and single covariates were tested using glmmTMB models (Brooks et al., 2017a; b) in R software (R Core Team, 2024). In all models, the flock, farm, and village identities (in a nested formation) were used as random effects to account for repeated sampling and spatial clustering. Binomial distributions were used for all models, except for those in which SE or ST positivity were calculated in the entire dataset. In these, Poisson distribution was used due to low frequency of positive events. All factors with a p-value <0.2 in the univariable models were included in the multivariable models, which were conducted using the same models described above. Collinearities between variables to be included in the multivariable models were tested following the methodology described by Dohoo et al. (2014) for evaluating collinearity between categorical variables. In the final models, to reduce the risk of a type II error (i.e., falsely failing to reject the null hypothesis), factors were removed by the backward selection method using the likelihood statistic, and a probability of 0.05 was defined for the removal of variables.

Results

Between 2018 and 2023, 6,099 layer flocks were raised in 1,934 layer farms in Israel, 6,076 flocks in 1,927 layer farms were included in this study. Given the small geographic area of Israel (circa 22,000 km²), farms were located in close proximity to other poultry farms as well as to dairy cattle farms, as illustrated in Fig. 2. During that period, 4,347 confirmed positive sampling events of Salmonella were identified in layer flocks in Israel, representing a prevalence of 17.74 % of the total sampling events (n = 24,500). Of the samples tested, 0.73 % (181/24,500) and 0.52 % (129/24,500) were positive for SE and ST, respectively.

Fig. 2.

Map of Israeli poultry and dairy farms. The geographical distribution of layer farms (green dots) between 2016 and 2023 in Israel. Layer flocks are marked with green dots, and poultry farm densities, including broiler, breeder, and turkey flocks, are illustrated using a red color gradient (≤10 farms within a 3.5 km radius around each poultry farm, 10–50, and>50). The cutoff of 3.5 km radius, rather than the 500 m used in the analysis, was selected for graphical presentation to allow better visualization. The location of dairy farms is indicated (blue dots).

The annual prevalence of Salmonella at the flock level was calculated after removal of repeated positive tests for the same flock within a given year, which occurred in 26 % (841/3,202) of the positive flocks. Of the 6,076 flocks sampled, 2,586 were identified as positive for Salmonella at least once, yielding a cumulative flock-level prevalence of 42.56 %. One hundred and eighty-one flocks were found positive for SE (2.98 % of the flocks), and 129 flocks were positive for ST (2.09 % of the flocks). Annual and cumulative prevalence of overall Salmonella, SE, and ST are detailed in Supplementary Table S2.

Different sample types used by the IVS inspectors vary in sensitivity, as illustrated in Fig. 3. Similar to overall Salmonella, SE exhibits varying sensitivity across different sample types, with dust being a significant contributor (data not shown; personal communication: Anat Wiseman).

Fig. 3.

Comparison of Salmonella detection by different sampling methods. A Venn diagram demonstrating the overlap in Salmonella detection by the three sample types (drag swabs, dust swabs, and dust samples) used simultaneously by the Israeli Veterinary Service inspectors during 1,965 positive sampling events in layer flocks in Israel between 2017 and 2023.

In the univariable analysis, significant associations (p < 0.2) were found between Salmonella infection and flock size (>10,000 vs ≤10,000), housing system type (cage vs. cage-free), geographic area (Center/South vs. North), breed (Hyline vs. Lohman), sampling entity (IVS vs. EPB), age in weeks (>85 vs ≤85), and distance to nearest dairy farm (<0.3 km vs. ≥0.3 km). Density of poultry farms in a 500 m radius (>10 vs. ≤10), and season (spring, summer and winter vs. fall) were significantly associated with decreased risk of Salmonella infection (Table 1). All variables, except the “Geographic area” variable, were included in the multivariable analysis. The “Geographic area” variable was excluded in the final multivariable model due to its high correlation with the variable “Density of poultry farms in 500 m radius”. The associations of all variables, except for “breed”, with Salmonella infection remained significant in the final model (p < 0.05; Table 2).

Table 1.

Univariable analyses¹ of the associations between possible risk factors and Salmonella positivity in layer flocks in Israel between 2018 and 2023.

Risk Factor		OR (CI95%)2	Cases/Total (%)		p-value
			Exposed	Not Exposed
Flock size (>10,000 vs ≤10,000)		1.80 (1.51 - 2.16)	669/2,121 (31.54 %)	3,678/22,379 (16.44 %)	<0.001
Flock type (Cage vs. Cage-free)		2.27 (1.74 - 2.97)	4,132/22,652 (18.24 %)	215/1,848 (11.63 %)	<0.001
Geographic Area (vs. North)	Center	2.39 (1.57 - 3.64)	620/2,035 (30.47 %)	2,726/17,436 (15.63 %)	<0.001
	Lowlands and mountains	1.19 (0.78 - 1.79)	728/3,801 (19.15 %)	2,726/17,436 (15.63 %)	0.420
	South	2.03 (1.04 - 3.96)	106/402 (26.37 %)	2,726/17,436 (15.63 %)	0.038
	Valleys	1.34 (0.80 - 2.26)	167/826 (20.22 %)	2,726/17,436 (15.63 %)	0.267
Breed (vs. Lohman)	DeKalb	0.96 (0.79 - 1.17)	239/1,419 (16.84 %)	3,112/17,911 (17.37 %)	0.676
	Hyline	1.17 (1.03 - 1.32)	854/4,518 (18.9 %)	3,112/17,911 (17.37 %)	0.016
	Multiple breeds	1.09 (0.83 - 1.43)	142/652 (21.78 %)	3,112/17,911 (17.37 %)	0.540
Sampling entity (IVS vs. EPB)3		1.99 (1.85 - 2.15)	1,799/7,554 (23.82 %)	2,548/16,946 (15.04 %)	<0.001
Age in weeks (>85 vs ≤85)		1.24 (1.14 - 1.34)	1,307/6,701 (19.5 %)	3,040/17,799 (17.08 %)	<0.001
Distance to nearest dairy (<0.3 km vs. ≥0.3 km)		2.14 (1.56 - 2.94)	307/1,053 (29.15 %)	4,040/23,447 (17.23 %)	<0.001
Density of poultry farms in a 500 m radius4 (>10 vs. ≤10)		0.51 (0.41 - 0.64)	2,883/18,646 (15.46 %)	1,464/5,854 (25.01 %)	<0.001
Season (vs. Fall)	Spring	0.30 (0.26 - 0.33)	722/6,240 (11.57 %)	1,431/5,416 (26.42 %)	<0.001
	Summer	0.53 (0.48 - 0.58)	1,186/6,735 (17.61 %)	1,431/5,416 (26.42 %)	<0.001
	Winter	0.48 (0.43 - 0.53)	1,008/6,109 (16.5 %)	1,431/5,416 (26.42 %)	<0.001

¹A mixed-effects regression with binomial distribution.

²Odds ratio with confidence intervals of 95 %.

³Israeli Veterinary Services vs. Egg and Poultry Board inspectors.

⁴Poultry farm density was calculated by creating a 500 m buffer around the flock and counting the number of poultry farms (layer, broiler, turkeys and reproduction of both layer and broilers) located in this area. The analysis was performed on a categorized variable derived from this count with a threshold of 10 poultry flocks within a 500 m radius buffer.

Table 2.

Multivariable analysis⁵ of the associations between possible risk factors and Salmonella positivity in layer flocks in Israel between 2018 and 2023.

Risk Factor		OR (CI95%)6	p-value
Flock size (>10,000 vs ≤10,000)		1.73 (1.42 - 2.10)	<0.001
Flock type (Cage vs. Cage-free)		3.16 (2.42 - 4.13)	<0.001
Season (vs. Fall)	Spring	0.29 (0.26 - 0.32)	<0.001
	Summer	0.49 (0.45 - 0.55)	<0.001
	Winter	0.48 (0.44 - 0.54)	<0.001
Sampling entity (IVS vs. EPB)7		2.04 (1.89 - 2.21)	<0.001
Age in weeks (>85 vs ≤85)		1.24 (1.14 - 1.35)	<0.001
Distance to nearest dairy (<0.3 km vs. ≥0.3 km)		1.86 (1.36 - 2.54)	<0.001
Density of poultry farms in a 500 m radius8 (>10 vs. ≤10)		0.54 (0.43 - 0.69)	<0.001

⁵A mixed-effects regression with binomial distribution.

⁶Odds ratio with confidence intervals of 95 %.

⁷Israeli Veterinary Services vs. Egg and Poultry Board inspectors.

⁸Poultry farm density was calculated by creating a 500 m buffer around the flock and counting the number of poultry farms (layer, broiler, turkeys and reproduction of both layer and broilers) located in this area. The analysis was performed on a categorized variable derived from this count with a threshold of 10 poultry flocks within a 500 m radius buffer.

Sampling entity (IVS vs EPB inspectors) and bird age (>85 vs ≤85 weeks) were significantly associated with SE infection in both the univariable (p < 0.05; Table 3) and multivariable models (p < 0.05; Table 4). Flock size (>10,000 vs ≤10,000), geographic area (Lowlands and mountains vs. North) and season (Spring vs. Fall) were included in the initial multivariable regression model (p < 0.2) but were excluded from the final multivariable regression model as they were non-significant.

Table 3.

Univariable analyses⁹ of the associations between possible risk factors and Salmonella Enteritidis positivity in layer flocks in Israel between 2018 and 2023.

Risk Factor		PR (CI95%)10	Cases/Total (%)		p-value
			Exposed	Not Exposed
Flock size (>10,000 vs ≤10,000)		1.43 (0.83 - 2.47)	23/2,121 (1.08 %)	158/22,379 (0.71 %)	0.198
Flock type (Cage vs. Cage-free)		0.93 (0.25 - 3.47)	166/22,652 (0.73 %)	15/1,848 (0.81 %)	0.919
Geographic Area (vs. North)	Center	0.26 (0.02 - 3.23)	3/2,035 (0.15 %)	116/17,436 (0.67 %)	0.292
	Lowlands and mountains	1.87 (0.82 - 4.29)	51/3,801 (1.34 %)	116/17,436 (0.67 %)	0.139
	South	1.10 (0.07 - 17.26)	3/402 (0.75 %)	116/17,436 (0.67 %)	0.946
	Valleys	1.40 (0.24 - 8.09)	8/826 (0.97 %)	116/17,436 (0.67 %)	0.706
Breed (vs. Lohman)	DeKalb	0.89 (0.18 - 4.42)	9/1,419 (0.63 %)	127/17,911 (0.71 %)	0.887
	Hyline	1.18 (0.48 - 2.86)	37/4,518 (0.82 %)	127/17,911 (0.71 %)	0.719
	Multiple breeds	1.60 (0.26 - 9.88)	8/652 (1.23 %)	127/17,911 (0.71 %)	0.614
Sampling entity (IVS vs. EPB)11		2.42 (1.72 - 3.41)	90/7,554 (1.19 %)	91/16,946 (0.54 %)	<0.001
Age in weeks (>85 vs ≤85)		11.59 (6.53 - 20.57)	76/6,701 (1.13 %)	105/17,799 (0.59 %)	<0.001
Distance to nearest dairy (<0.3 km vs. ≥0.3 km)		1.09 (0.45 - 2.64)	8/1,053 (0.76 %)	173/23,447 (0.74 %)	0.848
Density of poultry farms in a 500 m radius12 (>10 vs. ≤10)		0.75 (0.43 - 1.30)	134/18,646 (0.72 %)	47/5,854 (0.8 %)	0.305
Season (vs. Fall)	Spring	1.55 (0.95 - 2.52)	45/5,416 (0.83 %)	46/6,240 (0.74 %)	0.081
	Summer	0.94 (0.57 - 1.53)	42/6,735 (0.62 %)	46/6,240 (0.74 %)	0.800
	Winter	1.22 (0.75 - 2.01)	48/6,109 (0.79 %)	46/6,240 (0.74 %)	0.425

⁹A mixed-effects regression with Poisson distribution.

¹⁰Prevalence Ratio (PR) with confidence intervals of 95 %.

¹¹Israeli Veterinary Services vs. Egg and Poultry Board inspectors.

¹²Poultry farm density was calculated by creating a 500 m buffer around the flock and counting the number of poultry farms (layer, broiler, turkeys and reproduction of both layer and broilers) located in this area. The analysis was performed on a categorized variable derived from this count with a threshold of 10 poultry flocks within a 500 m radius buffer.

Table 4.

Multivariable analysis¹³ of the associations between possible risk factors and Salmonella Enteritidis positivity in layer flocks in Israel between 2018 and 2023.

Risk Factor	PR (CI95%)14	p-value
Sampling entity (IVS vs. EPB)15	2.37 (1.63 - 3.45)	<0.001
Age in weeks (>85 vs ≤85)	11.33 (6.33 - 20.27)	<0.001

¹³A mixed-effects regression with Poisson distribution.

¹⁴Prevalence Ratio with confidence intervals of 95 %.

¹⁵Israeli Veterinary Services vs. Egg and Poultry Board inspectors.

Sampling entity (IVS vs EPB inspectors) and bird age (>85 vs ≤85 weeks) were significantly associated with ST infection, with this association consistent across all models (Table 5, Table 6).

Table 5.

Univariable analyses¹⁶ of the associations between possible risk factors and Salmonella Typhimurium positivity in layer flocks in Israel between 2018 and 2023.

Risk Factor		PR (CI95%)17	Cases/Total (%)		p-value
			Exposed	Not Exposed
Flock size (>10,000 vs ≤10,000)		0.65 (0.10 - 4.36)	8/2,121 (0.38%)	121/22,379 (0.54%)	0.661
Flock type (Cage vs. Cage-free)		1.07 (0.19 - 6.07)	120/22,652 (0.53%)	9/1,848 (0.49%)	0.940
Geographic Area (vs. North)	Center	1.50 (0.34 - 6.67)	14/2,035 (0.69%)	76/17,436 (0.44%)	0.591
	Lowlands and mountains	1.65 (0.54 - 5.06)	30/3,801 (0.79%)	76/17,436 (0.44%)	0.380
	South	2.19 (0.15 - 31.26)	4/402 (1%)	76/17,436 (0.44%)	0.564
	Valleys	1.33 (0.13 - 13.89)	5/826 (0.61%)	76/17,436 (0.44%)	0.812
Breed (vs. Lohman)	DeKalb	1.26 (0.21 - 7.43)	9/1,419 (0.63%)	88/17,911 (0.49%)	0.801
	Hyline	1.14 (0.37 - 3.55)	25/4,518 (0.55%)	88/17,911 (0.49%)	0.822
	Multiple breeds	1.87 (0.22 - 16.15)	7/652 (1.07%)	88/17,911 (0.49%)	0.569
Sampling entity (IVS vs. EPB)18		2.43 (1.60 - 3.67)	59/7,554 (0.78%)	70/16,946 (0.41%)	<0.001
Age in weeks (>85 vs ≤85)		10.54 (5.15 - 21.56)	44/6,701 (0.66%)	85/17,799 (0.48%)	<0.001
Distance to nearest dairy (<0.3 km vs. ≥0.3 km)		1.60 (0.26 - 9.77)	9/1,053 (0.85%)	120/23,447 (0.51%)	0.613
Density of poultry farms in a 500 m radius19 (>10 vs. ≤10)		0.67 (0.26 - 1.76)	86/18,646 (0.46%)	43/5,854 (0.73%)	0.418
Season (vs. Fall)	Spring	1.07 (0.62 - 1.87)	31/5,416 (0.57%)	39/6,735 (0.58%)	0.800
	Summer	0.71 (0.39 - 1.27)	28/6,240 (0.45%)	39/6,735 (0.58%)	0.246
	Winter	1.00 (0.57 - 1.76)	31/6,109 (0.51%)	39/6,735 (0.58%)	0.997

¹⁶A mixed-effects regression with Poisson distribution.

¹⁷Prevalence Ratio (PR) with confidence intervals of 95%.

¹⁸Israeli Veterinary Services vs. Egg and Poultry Board inspectors.

¹⁹Poultry farm density was calculated by creating a 500 m buffer around the flock and counting the number of poultry farms (layer, broiler, turkeys and reproduction of both layer and broilers) located in this area. The analysis was performed on a categorized variable derived from this count with a threshold of 10 poultry flocks within a 500 m radius buffer.

Table 6.

Multivariable analysis²⁰ of the associations between possible risk factors and Salmonella Typhimurium positivity in layer flocks in Israel between 2018 and 2023.

Variable / Risk Factor	PR (CI95%)21	p-value
Sampling entity (IVS vs. EPB)22	2.07 (1.33 - 3.22)	0.001
Age in weeks (>85 vs ≤85)	9.15 (4.49 - 18.68)	<0.001

²⁰A mixed-effects regression with Poisson distribution.

²¹Prevalence Ratio with confidence intervals of 95%.

²²Israeli Veterinary Services vs. Egg and Poultry Board inspectors.

Discussion

We studied the prevalence and risk factors associated with Salmonella contamination in table-egg layer flocks in Israel between 2018 and 2023, with emphasis on two prominent serovars, SE and ST. We found positive sampling event prevalences of 17.74 %, 0.73 % and 0.52 % for Salmonella, SE, and ST, respectively. The annual flock-level prevalence values ranged between 18.33 % and 35.52 %, 0.97 % and 2.40 %, and 0.88 % and 1.31 % for Salmonella, SE, and ST, respectively. Consequently, between 2018 and 2023, approximately two million birds were culled (1,360,924 due to SE detection and 634,111 due to ST detection (Egg and Poultry Board, 2023)).

Layer flocks included in this study were hatching from fertilized eggs laid by their progenitor layer breeder flocks as early as 2016. Between 2016 and 2023, 147 layer breeder flocks (each flock represents a growing cycle in a single poultry house) were active in 26 farms (with an average life span of 68 weeks), of which 27 % tested positive for overall Salmonella. During this period, only one case of SE was detected and culled, and no instances of ST were found (personal communication: Anat Wiseman). While vertical transmission is considered a significant source of SE and ST in layer flocks (Gantois et al., 2009; Shaji et al., 2023), the near absence of these serovars in layer breeding flocks in Israel suggests that their descendants, the layer flocks, acquire SE and ST through horizontal transmission from the environment.

We found that larger flock sizes (>10,000 birds), caged housing systems (compared to cage-free), older bird age (>85 weeks), proximity to dairy farms (<0.3 km), and sampling by IVS (compared to EPB) were significantly associated with an increased risk of Salmonella detection. Seasonality showed a strong protective effect, with reduced odds observed in spring, summer, and winter compared to fall, and so did high density of poultry farms within a 500 m radius.

Various studies have reported large flock size as a risk factor for Salmonella infection, with flocks larger than 20 to 30 thousand hens at higher risk of infection (Huneau-Salaün et al., 2009; Namata et al., 2008; Snow et al., 2010; USDA, 2013). We found similar results even when comparing flock size at a much lower threshold (10,000 hens/flock).

Although cage-free systems may introduce exposure to additional risk factors such as increased access of vermin and wildlife, several studies have shown that these housing systems pose a lower risk for Salmonella compared with conventional caging (Coleman et al., 2005; Samper-Cativiela et al., 2023; Van Hoorebeke et al., 2011), consistent with our results.

Previous studies have linked bird age with increased risk of infection in recently hatched chicks that are at risk for clinical salmonellosis (particularly if infected close to hatching) (Wallis, 2006), or in older birds (>60–80 weeks of age), particularly after molting (Coleman et al., 2005; Samper-Cativiela et al., 2023; USDA, 2000; Wales et al., 2007). In our study, an increased risk of Salmonella, ST and SE serovar infections was found among older birds (>85 weeks). However, as deliberate molting of laying hens has become less frequent in recent years (personal communication: Anat Wiseman), and was legally prohibited in Israel as of 2024 (Ministry of Agriculture, 2022 (effective of July 2024)), it is unlikely that molting accounts for the increased risk for infection; instead, other age-related physiological processes, such as increased susceptibility to infection due to a decrease in immune response as the hen ages (Schmucker et al., 2021), changes in the gastrointestinal tract's microbiota (Khan et al., 2020) as well as its mucosal barrier function (Hou et al., 2025) are more plausible underlying causes. In addition, assuming the infection source is horizontal, the probability of infection increases over time as environmental exposure increases.

The increased risk of infection during the fall, as observed in our study, aligns with seasonality reported by the European surveillance program (EFSA Panel on Biological Hazards, 2010) and with a specific research describing risks for Salmonella in Spanish flocks (Samper-Cativiela et al., 2023). In contrast, an increased risk was reported during the summer months in another country-specific study from the UK (Wales et al., 2007). Given the substantial variation in climatic factors across regions and seasons, it is difficult to identify the precise drivers of differences in the seasonal risk of Salmonella detection. However, the increased risk in the fall season observed in Israel and Spain – both with relatively similar climates – versus the summer peak in the cooler UK climate, likely reflects seasonal differences in temperature and humidity that favor Salmonella survival and growth.

Proximity to dairy farms (<300 m) was found to be a risk factor for general Salmonella. Still, it was not significantly associated with increased risk of ST or SE, contrary to our hypothesis that ST, given its wide host adaptability (Dar et al., 2017; Shaji et al., 2023; Wallis, 2006), would be more strongly associated with proximity to dairy farms. Yet, the lack of statistical significance may reflect the limited number of positive detections for these two serovars. Also, we did not evaluate proximity to ST positive dairy farms (as this information was not available), and this analysis could have contributed further insights to this association. The significantly elevated risk for Salmonella near dairy farms could suggest potential environmental or interspecies transmission pathways, as was presented before with dairy farms (Perry et al., 2024). Such transmission between adjacent farms may be a result of shared vectors or farm infrastructure (EFSA Panel on Biological Hazards, 2010; Snow et al., 2010). Moreover, the increased risk for Salmonella infection observed in our study could be attributed to exposure to generalist serovars, other than ST, which were not evaluated in this study, originating from these facilities; further investigation of these serovars could clarify the role of dairy proximity as a risk factor for Salmonella infection in layer flocks.

We observed decreased risk of detecting Salmonella in flocks located in areas with high density of poultry farms within a 500 m radius, contrary to our hypothesis that higher density would increase contamination risk. The literature on the role of density of poultry farms in Salmonella risk is limited and inconsistent: Snow et al. (2010) demonstrated that the presence of poultry farms less than one km away from the tested farm was significantly associated with Salmonella infection, and recently, Hobbelen et al. (2024) showed that proximity to SE and ST-positive neighboring poultry farm increases the risk of infection. However, research by Fagbamila et al. (2018) found such proximity to be protective from Salmonella infection. The inconsistency reflected by the literature is further highlighted by our findings. Moreover, Fagbamila et al. (2018) suggested increased biosecurity in areas with a high density of poultry farms as a possible explanation for their findings, as it would reduce transmission via people and fomites; however, this is unlikely to account for the situation in all Israeli egg-producing poultry farms, given that strict biosecurity measures are often enforced mainly in large farms. In addition, unmeasured confounding variables may influence the outcomes of the analyses conducted here and elsewhere, leading to these inconsistent results.

The sampling entity (IVS compared to EPB inspectors) was also significantly associated with higher odds of detecting Salmonella, including ST and SE. Differences in sampling methodology may partly explain these findings, as IVS inspectors collected three sample types simultaneously, which vary in sensitivity and result in partially non-overlapping detection results. In our analysis, we consolidated all collected samples into a single sampling event, effectively converting the scheme to parallel sampling and thereby increasing sensitivity and the number of detections by IVS inspectors. However, this is only a partial explanation since the positivity rates of IVS and EPB inspectors differed significantly even when comparing only drag swabs (23.9 % positive in IVS drag swabs vs. 17 % in EPB drag swabs) (Egg and Poultry Board, 2023). Another factor potentially contributing to the observed difference could be biased out-of-schedule sampling conducted by the IVS inspectors during outbreaks in certain localities as part of outbreak containment measures. Such samplings were recorded in the database identically to routine surveillance samples, preventing differentiation between the two in our analysis. Moreover, the described variations in sample type and inclusion of sampling during outbreaks as part of the routine surveillance results may also partially account for the differences in prevalence observed between Israel and the EU, where sampling is conducted by farm personnel using faeces samples, which have lower sensitivity for Salmonella detection (EFSA, 2007). Other studies also show differences in sensitivity rates between sample types, with swabs and dust having better sensitivity compared to faeces samples (Gole et al., 2017).

This study was based on data obtained from the national surveillance program, which collects extensive information from all layer flocks in the country. This dataset provides a powerful, population-wide source of information, free from the selection bias associated with targeted sampling (Thygesen and Ersbøll, 2014). Since this dataset is based on routine surveillance, it does not include observational details, such as the presence of rodents, cleaning schedule, truck movements and molting time which have been identified as risk factors in smaller sample-sized studies (EFSA Panel on Biological Hazards, 2010; Denagamage et al., 2015). Future studies incorporating these variables could help refine our understanding of risk factors associated with Salmonella in layer flocks.

In contrast to overall Salmonella, only three variables remained significant in the multivariable model for SE: sampling entity, bird age, and geographic area. Older bird age (>85 weeks) was associated with over twice the risk of SE detection, and IVS sampling also showed higher odds of detecting SE. Previous research on SE epidemiology in Israel has identified the presence of wild birds, a distance of less than 100 m to the nearest neighboring flock, short clearance time between consecutive flocks, and mixed-breed flocks as risk factors for SE detection (Berman, 2005). However, in our study, the density of poultry farms within a 500 m radius was not associated with increased risk of SE. Similarly, other studies have found that proximity to neighboring farms is not a significant risk factor for SE, though it has been linked to general Salmonella risk (Snow et al., 2010).

In the analysis of ST, only sampling entity (IVS in comparison to EPB inspector) and bird age (>85 weeks) were associated with increased risk of ST detection in the multivariable analysis. Other known contributors to ST infection in poultry (layers and breeders) have been summarized in a review by Dar et al. (Dar et al., 2017), such as contaminated feed and water, presence of rodents and wild birds, poor biosecurity, vertical transmission, contaminated equipment and litter, humans acting as physical vectors, stress factors in hens, and persistence in the environment, often due to inadequate cleaning between production cycles. These variables could not be included in our analysis; therefore, their potential association with ST positivity could not be estimated. Yet, in the Israeli context, vertical transmission can be ruled out, as breeder flocks in Israel are free of ST (Egg and Poultry Board, 2023).

Conclusion

This analysis highlights the multifactorial nature of Salmonella contamination in layer flocks, with flock management, environmental exposure, sampling methodology, and bird age playing pivotal roles. Outcomes reflect the current prevalence and risk factors for Salmonella infection in the intensively managed egg-laying sector in Israel and can be extrapolated to other global regions with similar intensive egg-laying production systems. Notably, older age (>85 weeks) was found as a consistent risk factor across all models and serovars, suggesting that it should be considered when determining the optimal duration of a flock's productive cycle, in addition to other considerations such as decline in egg quality as layers get older (Wiseman et al., 2024). Based on the results of this study, we recommend maintaining smaller flocks for shorter productive cycles and using non-caged housing systems, as these measures may reduce Salmonella contamination.

Ethical statement

Data analyzed in this study is of isolates collected as part of a mandatory routine surveillance program that was conducted according to the ethical standards and in compliance with Israel law.

CRediT authorship contribution statement

Bat Shachaf: Writing – review & editing, Writing – original draft, Visualization, Software, Formal analysis, Data curation. Anat Wiseman: Writing – review & editing, Writing – original draft, Validation, Supervision, Resources, Project administration, Methodology, Data curation, Conceptualization. Ehud Elnekave: Writing – review & editing, Writing – original draft, Validation, Supervision, Project administration, Methodology, Conceptualization.

Disclosures

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.