Background

Fresh produce plays an important role in the human diet because of its health-promoting nutritional characteristics [1]. Its antioxidants content in particular, is thought to be able to protect human cells from the attack of free radicals, which is in turn involved in the etiopathogenesis of most chronic diseases [2-4]. It is also hypothesized that vegetable antioxidants may bring further protective health effects through various mechanisms, acting as inducers of mechanisms related to cell maintenance, DNA repair and longevity [5-7]. This awareness has gradually resulted in the last years in a generalized shift of eating habits towards an increasing intake of fruits and vegetables. The data provided by the World Health Organization (WHO)/Food and Agriculture Organization (FAO) showed a 4.5% yearly increase in fruit and vegetable consumption between 1990 and 2004 [8]. Outbreaks of foodborne illnesses associated with the consumption of fresh produce are simultaneously increasing in frequency. In developed countries, changes in consumption patterns, raising numbers in elderly and immunocompromised consumers and the growing import of vegetables from countries with poor sanitary conditions have likely contributed to this epidemiological trend [9]. Moreover advanced diagnostic methods and surveillance systems have enhanced identification of fresh produce as sources of foodborne disease [10].

As a consequence, food safety of fresh produce is a matter of increasing concern; indeed, microbial contamination may occur during any of the steps in the farm-to-table continuum from environmental, animal or human sources [11-15].

Minimally processed vegetables (MPV) are processed to increase their functionality without changing their fresh properties. Preparation procedure generally includes processes such as washing, peeling, cutting, trimming and/or slicing [16]. Salmonella spp., Shigella spp., pathogenic Escherichia coli, Listeria monocytogenes and Campylobacter spp. are the most important vegetable-borne pathogens [11,15]. Data from the recent literature are inconsistent about the prevalence of these pathogens in vegetables, due to significant differences between studies in the sizes and place of sampling, local fresh produce type, seasonality and analytical methods [8]. Some authors report Salmonella spp. in less than 8% of the analyzed samples, Campylobacter spp. in 3.1% of lettuce, but E. coli O157 in up to 25% of cabbages and 19.5% of coriander and L. monocytogenes in up to 7% of cabbages, 22.7% of leafy vegetables and 20% of lettuce [8,17]. Other reports describe prevalence lower than 1-2% for E. coli O157, Campylobacter spp. and Salmonella spp. [18].

The aim of the study was to assess the hygienic quality and the prevalence of the most common pathogens in fresh produce, minimally processed vegetables (MPVs) and unprocessed vegetables (UVs), sold in retail markets in Sicily, Italy.

Results and Discussion

Microbial flora

The microbiological results of UVs are summarized in Table 1. The Aerobic Mesophilic Count (AMC) levels ranged between 2 log and 6 log cfu g−1 and the Enterobacteriaceae counts between 2 log and 6 log cfu g−1, except for the seed and bulb vegetable sample (<1 log cfu g−1). Escherichia coli was found only in leaf (lettuce) at a concentration of 2–3 log cfu g−1. Enterococci were found in leaf (lettuce), flower (cauliflowers) and fruit vegetables (marrow) at a concentration of 2–3 log cfu g−1.

Table 1 Microbiological findings obtained from 85 samples of UVs (in log cfu g −1 )
Full size table

MPVs (Table 2) yielded AMC values between 5–7 log cfu g−1, except for the fruit vegetable samples (<1 log cfu g−1) (Table 2).

Table 2 Microbiological findings obtained from 40 samples of MPVs (in log cfu g −1 )
Full size table

Enterobacteriaceae and E. coli were detected in leaf vegetables (salad and spinach) and in one chicory sample. Enterococci were detected in salad samples and in a pumpkin sample at a concentration of 2–4 log cfu g−1. Coagulase positive Staphylococci and sulphite-reducing Clostridia were not detected in any sample.

The samples of the leaf vegetables had significantly higher AMC (p < 0.001) and Enterobacteriaceae (p = 0.01) counts than fruit vegetables. The differences of the E. coli counts between the two categories of vegetables were not statistically significant (p = 0.08), probably due to the lowest proportion of contaminated samples.

The colonies isolated from Violet Red Bile Glucose Agar proved to belong to the following bacterial species: Klebsiella oxytoca, Pantoea spp., Serratia odorifera, Serratia liquefaciens, Serratia ficaria, Raoultella terrigena, Rahnella aquatilis and Erwinia spp.

Our findings showed that leaf vegetables had the highest bacterial counts among vegetable products, with particular reference to the mesophilic bacteria, Enterobacteriaceae, E. coli and Enterococci. Despite the high counts, no products reported visible signs of organoleptic alterations. Accordingly, Ragaert et al. [19] reported that evident organoleptic alterations occur in vegetables only when bacterial count is higher than 7 or 8 log cfu g−1.

High bacterial counts could likely be associated with the morphology of leaves which have a broad and rather rough surface. Indeed, both the large surface, easily coming in contact with the ground and the irrigation water, and its roughness facilitate the accumulation of dirt and adhesion of bacteria, as reported by some authors [10,20,21]. Our results agree with those reported by literature that identify salads as the most contaminated vegetable product. Seow et al. [10] in a study conducted on fresh vegetables and fruit samples, both UVs and MPVs, showed that samples of lettuce, along with sprouts, have the highest levels of mesophilic bacteria and coliforms. They reported mesophilic bacteria counts ranging from 3.4 to 7.3 log cfu g−1 and coliforms between 1.6 and 5.9 log cfu g−1, with 50% of samples containing more than 5 log cfu g−1 of coliforms. Allende et al. [22], while assessing the microbiological quality of red lettuce marketed in Spain, found 3.67 log cfu g−1 of coliforms. In Canada, Bohaychuk et al. [23] assessed the prevalence of E. coli in fresh vegetables produced in Alberta and found this organism in 8.2% of lettuce, spinach and carrots samples with counts ranging from 0.48 to 3.04 log MPN g−1. About UVs, it can be observed that Enterobacteria were present in almost all samples; counts higher than 2 log cfu g−1 were detected in 97.6% of the examined samples. However, this bacterial counts does not appear to be of special concern, as the identified species are commonly found in soil, water and in environments characterized by an excessive organic load. Thus, such microbial flora can be primarily attributed to an environmental source [10,24]. Regarding E. coli contamination, there was a low percentage of contaminated UVs samples (3.5%) with bacterial counts ranging from 2 to 3 log cfu g−1. Moreover, in our study, 5.0% of MPVs were contaminated by E. coli, but only one sample did not comply with the limits set by law [25], as the bacterial count was 4 log cfu g−1. Of interest, E. coli is included in the Regulation (EC) No 2073/2005 as the indicator of quality of hygienic processing for MPVs (pre-cut fruit and vegetables) [25]. Our data agree with those reported by literature, indicating low percentages of vegetable products contaminated with E. coli and with low counts. Abadias et al. [21] found E. coli in 7.1% of vegetable products of UVs and in 11.4% of products of MPVs, with only 0.8% of products exceeding 100 MPN g−1. Santos et al. [24] found 2.6% of contaminated samples (lettuce and spinach) with counts ranging between 1 and 2 log cfu g−1.

Bacterial pathogens

Our results showed a total of three samples out of 125, one UV and two MPVs, yielding foodborne pathogens, in particular Salmonella spp., L. monocytogenes and Y. enterocolitica.

All samples tested negative for Campylobacter spp., Shigella spp. and E. coli O157.

Salmonella veneziana was detected in one sample of green salad (UVs). It yielded also an AMC of 6.98 log cfu g−1, an Enterobacteriaceae count of 5.73 log cfu g−1, an E. coli count of 3.95 log cfu g−1 and an Enterococci count of 2.07 log cfu g−1 (Table 3).

Table 3 Characteristics of vegetables which tested positive for foodborne pathogens
Full size table

Listeria monocytogenes serotype 4b was detected in association with L. innocua in a pumpkin sample. This was a minimally processed vegetable sample and the quantitative analysis yielded a concentration of Listeria less than 1 log cfu g−1. This sample had also an AMC of 6.47 log cfu g−1 (Table 3).

Yersinia enterocolitica O:3 was detected in a MPV spinach sample which yielded also Aeromonas hydrofila. Moreover, this sample showed an AMC of 6.95 log cfu g−1 and an Enterobacteriaceae count of 4.90 log cfu g−1 (Table 3).

Literature data are conflicting about the prevalence of pathogens in vegetables. Studies on UV fresh vegetables in USA, United Kingdom, and Malaysia showed a highly variable prevalence of Salmonella contamination, with values between 0% and 35%, respectively [18,26,27]. During 2005–2006, in Spain, Abadias et al. [21] detected Salmonella spp. and L. monocytogenes, respectively, in 1.3% and 0.7% of their samples. Furthermore, Sant’Ana et al. [28] studied the prevalence of Salmonella spp. in MPVs and found that 0.8% of the samples was contaminated by serotypes Enteritidis and Typhimurium. In Norway, Johannessen et al. [29] isolated L. monocytogenes and Y. enterocolitica, respectively, in 0.3% and 3% of the samples. Santos et al. [24] tested MPV samples for pathogens and found A. hydrophila in 11 samples (7.3%), L. monocytogenes in one sample of spinach (0.6%) and L. innocua in a further sample of spinach and salad (1.3%). Pathogen contamination in MPVs suggests failures in risk assessment and management systems in processing facilities, where contamination sources may be very heterogeneous: soil, irrigation waters and processing environment. Moreover washing and disinfection may not provide a complete removal of pathogens from MPVs. It is, indeed, of the uttermost importance to accurately follow Good Manufacturing Practices (GMP) through the entire production cycle (from primary production to distribution), in order to consistently meet food safety objectives.

Conclusion

Community regulation about the hygienic quality of vegetables is quite recent. Regulation (EC) 2073/2005 and subsequent modifications, has established microbiological limits for some types of vegetable products. In particular for ready-to-eat pre-cut fruits and vegetables, E. coli and L. monocytogenes/Salmonella spp. were defined as microbiological criteria of process hygiene and food safety, respectively. No further specific regulation is in place in EU for other types of vegetable products, such as UVs, or other pathogen microorganisms (e.g. Yersinia spp., Campylobacter spp.). Even though there are worldwide reports of outbreaks associated with the consumption of vegetable products, data concerning the microbial contamination level of these foodstuffs are still few and discrepant; in particular this lack of knowledge affects most MPV and UV vegetables.

This work is to some extent a preliminary investigation which will allow for targeting specific groups of products with a higher risk profile in the next future.

Methods

Vegetable samples

One hundred twenty-five vegetable samples, (40 MPVs and 85 UVs), were collected from four supermarkets and two greengrocer’s shops (Table 4) in Palermo district.

Table 4 Number and type of vegetable samples under analysis
Full size table

These variety of places allowed products commonly available to consumers to be sampled, thus making results more representative.

Sampling was carried out by randomly taking 250–300 g of sample from each box, according to the Regulation (EC) No 333/2007 [30]. The samples of MPVs were bought within 1–2 days from their packaging. All samples were subjected to microbiological analysis after an inspective check to evaluate the following features: for UVs, freshness and absence of spoilage signs; for MPVs, labeling information and presence of extraneous materials into the wrapping.

The samples were transferred to the laboratory in cooler boxes at temperature between 1°C and 8°C and the microbiological assays were performed within 24 hrs since the sample collection.

Enumeration of bacteria

Each sample was analyzed for Aerobic Mesophilic Count (AMC), Enterobacteriaceae, E. coli β-glucuronidase positive, coagulase positive Staphylococci, sulphite-reducing anaerobic organisms, Enterococci and L. monocytogenes. Microbial analyses were carried out using the standard methodologies described in Table 5.

Table 5 Standard methods used for microbial analyses
Full size table

Thirty g of each sample were weighed into sterile stomacher bags and homogenized with 270 ml Saline Peptone solution (NaCl 8.5 g l−1, Peptone l.0 g l−1) in a stomacher (Type 400; Seward London, UK). Decimal dilutions were prepared with the same diluent and 1 ml of each was used as inoculum. Results were reported in terms of colony forming units (cfu g−1).

A laboratory internal method (Rapid Enterococcus Agar, 44°C, 48 hrs, followed by catalase and esculin hydrolysis test on the suspected colonies) (Biorad) was used for Enterococci.

Representative colonies of all discernible morphologies were picked up from the Violet Red Bile Glucose Agar plates, subcultured and biochemically identified by API 20E (bioMerieux, Marcy-l’Etoile, France).

Isolation of foodborne pathogens

Twenty five g of each sample were weighed into sterile stomacher bags and homogenized with 225 ml of enrichment broth for each pathogen.

Shigella spp. detection was performed by the ISO 21567:2004 method consisting of a first enrichment step in Shigella Broth with novobiocin at 41.5°C for 16–20 hrs, followed by subculture in MacConkey Agar, Xylose Lysine Deoxycholate (XLD) Agar and Hektoen Enteric Agar at 37°C for 20–24 hrs.

Yersinia enterocolitica detection was performed by the ISO 10273:2003 method with a selective enrichment in Pepton Sorbitol Bile Salts (PSB, Biolife) at 25°C for 5 days and an additional selective enrichment in Irgasan Ticarcillin (ITC, Biolife) Broth at 25°C for 48 hrs. The PBS broth was inoculated in Cefsulodin Irgasan Novobiocin Agar, while ITC Broth was streaked onto Salmonella-Shigella Agar with Sodium Deoxycholate and Calcium Chloride (Biolife, Sigma). All the plates were incubated at 30°C for 24 hrs.

Campylobacter spp. detection was performed by the ISO 10272–1:2006 method with a selective enrichment in Bolton Broth at 37°C for 4–6 hrs and then at 41.5°C for 48 hrs microaerobically. The Bolton Broth was inoculated in modified Campylobacter Charcoal Differential Agar (Biolife) and Skirrow Agar at 41.5°C for 48 hrs microaerobically.

Detection of Salmonella spp., L. monocytogenes and E. coli O157 was carried out by an enzyme linked fluorescent assay (ELFA) in an automatic system VIDAS (bioMerieux, Marcy-l’Etoile, France). In particular, the following methods were used:

  • for Salmonella spp., the AFNOR BIO 12/23-05/07 method including a pre-enrichment step in Buffered Peptone Water at 37°C for 16–20 hrs and a subsequent step performed by VIDAS Immuno-Concentration Salmonella II (ICS2). The samples which tested positive were then confirmed by subculturing them in XLD Agar at 37°C for 24 ± 3 hrs;

  • for L. monocytogenes, the AFNOR BIO 12/11-03/04 method was performed with Half Fraser broth at 30°C for 24–26 hrs and then Fraser Broth (FB) at 37°C for 24–26 hrs. One portion of the FB culture was then used for the L. monocytogenes VIDAS test (LMO2). The samples which tested positive were then confirmed by subculturing them in Listeria Aloa Agar (ALOA) (Biolife) and Listeria Oxford Agar at 37°C for 24 ± 3 hrs;

  • for E. coli O157, the AFNOR BIO 12/8-07/00 method was used with a first step in Tryptone Soya Broth with novobiocin, incubated at 41.5°C for 6–7 hrs, and a second step in MacConkey Broth with cefixime-potassium tellurite, incubated at 37°C for 18 hrs. After heating at 95-100°C, an aliquot was used for VIDAS (ECO).

All culture media were from OXOID except otherwise stated.

Biochemical and serological identification

The suspected Salmonella spp. colonies were subcultured for purity and identified by the Biolog automatic system (Biolog Inc., Hayward, CA). All suspected Listeria spp. colonies were submitted to catalase and β-hemolysis test and definitively identified by the Biolog automatic system.

Listeria monocytogenes isolates were subjected to serotyping by specific antisera (Denka Seiken, Tokio, Japan). The suspected Yersinia spp. colonies were submitted to oxidase, urease, indole test and definitively identified by API 20E (bioMerieux, Marcy-l’Etoile, France).

Serotyping of Salmonella, L. monocytogenes and Y. enterocolitica isolates was carried out at the Regional Reference Centre for Enteric pathogens, University of Palermo, Italy, by using commercial antisera (Staten Serum Institut, Denmark, and Denka-Seiken, Japan).

Statistical analysis

Quantitative values were categorized in two classes (equal/higher vs lower than the median value for AMC and Enterobacteriaceae and detected vs undetected for E. coli). Association between these classes and the most represented type of vegetables was assessed by contingency tables. Statistical significance was calculated by the chi-square test or the Fisher’s exact test, when appropriate. p value less than 0.05 was considered statistically significant.