The U.S. Food and Drug Administration (FDA) conducted a sampling assignment in 2014 to ascertain the prevalence of Cronobacter spp. and Salmonella in the processing environment of facilities manufacturing milk powder. Cronobacter was detected in the environment of 38 (69%) of 55 facilities. The average prevalence of Cronobacter in 5,671 subsamples (i.e., swabs and sponges from different facility locations) was 4.4%. In the 38 facilities where Cronobacter was detected, the average prevalence of positive environmental subsamples was 6.25%. In 20 facilities where zone information of the sampling location was complete, Cronobacter was most frequently detected in zone 4, followed by zone 3, then zone 2, with zone 1 yielding the lowest percentage of positive samples. The prevalence of Cronobacter across the zones was statistically different (P < 0.05). There was no significant association between product type (i.e., lactose, whey products, buttermilk powder, and nonfat dried milk) and prevalence of Cronobacter in the facility. Salmonella was detected in the environment of three (5.5%) of the 55 facilities; all three facilities produced dried whey product. The overall prevalence of Salmonella in 5,714 subsamples was 0.16%. In facilities in which Salmonella was detected, the average prevalence was 2.5%. Salmonella was most frequently detected in zone 4, followed by zone 3. Salmonella was not detected in zone 1 or zone 2. The disparity between Salmonella and Cronobacter prevalence indicates that additional measures may be required to reduce or eliminate Cronobacter from the processing environment.
Cronobacter was found in manufacturing areas of 69% of 55 U.S. milk powder facilities.
Salmonella was found in 5.5% of 55 facilities.
Both organisms were most frequently isolated from zone 4 locations.
No significant association was found between product type and Cronobacter prevalence.
Additional measures may be required to reduce or eliminate Cronobacter from the processing environment.
The genus Cronobacter is currently composed of seven species: C. sakazakii, C. malonaticus, C. turicensis, C. universalis, C. muytjensii, C. dublinensis, and C. condimenti (24, 27). Cronobacter spp. are gram-negative rods and facultative anaerobes, and they have a growth range of 5.5 to 45°C. Cronobacter spp. have been isolated from a wide variety of plant- and animal-based foods, including milk powder, powdered infant formula (PIF), spices, fruit and vegetables, wheat, dehydrated noodles, tea, instant soup, candy, and various raw meats (25, 28, 32, 40, 42, 48).
All Cronobacter species, except for C. condimenti, are pathogenic to humans (19, 25). However, Eshwar et al. (12) showed that C. condimenti was as pathogenic as other species in the zebrafish model. Cronobacter gained notoriety due to infections in infants; however, Cronobacter spp. can also cause illness in older children and adults. Cronobacter infections in adults can result in septicemia, as well as respiratory, wound, and urinary tract infections (1, 22, 36). Infection of infants with Cronobacter can result in septicemia, necrotizing enterocolitis, and meningitis, and there is a high case-fatality rate (10 to 80%) (43). Cronobacter infections in infants have been associated with consumption of PIF (43); neonates (infants younger than 4 weeks of age) and infants younger than 2 months of age are at greatest risk of infection (19). Recently Bowen et al. (4) and McMullan et al. (30) reported infantile Cronobacter septicemia-meningitis infections in which the infants consumed only expressed maternal breast milk. Contaminated personal breast pumps were found to be the source of the contamination.
Salmonella enterica is a leading cause of foodborne illness in many countries, including the United States (15, 41). The symptoms of Salmonella infection usually include diarrhea, fever, and abdominal cramps. Children younger than 5 years of age have higher risk for Salmonella infection than other age groups (5). Salmonellosis has been associated with a variety of food commodities, and outbreaks of foodborne illness have been reported in foods of both plant and animal origin (35). Its ability to contaminate low-moisture foods has been recognized for some time, and outbreaks due to the consumption of milk powder, infant formula, cereals, nuts, peanut and nut butter, chocolate, and spices have been documented (37).
Salmonella can survive desiccation and survive for long periods of time (months or years) in low-moisture foods such as dried milk products, nuts, pasta, and chocolate (3, 11). There have been several multiyear salmonellosis outbreaks associated with low-moisture foods, for example, Salmonella Agona in infant formula produced in France (2005 to 2017), Salmonella Poona in infant formula produced in Spain (2010 to 2019), and Salmonella Agona in breakfast cereal in the United States (1998 to 2008) (13, 14, 26, 45). In each outbreak, Salmonella was shown to be genetically related over the time span, indicating that the same strain of Salmonella persisted within the manufacturing facility for many years.
Studies have also shown that Cronobacter can persist in the environment of facilities that manufacture milk powder and PIF (3, 6, 9, 17, 21, 34, 49, 50). The predominant Cronobacter species cultured from four Irish PIF production environments during an 18-month surveillance study was C. sakazakii (50), and the predominant sequence types (STs) found were C. sakazakii ST1 isolates. In another study, Chase et al. (6) found that an ST83 C. sakazakii strain had persisted in a Swiss PIF manufacturing facility for at least 4 years. Phylogenetic analysis using microarray and whole genome sequencing data showed that four of five ST83 strains from product and the environment were highly phylogenetically related, and microarray showed that between 5 and 38 genes differed from one another in these strains (6). Cronobacter can also survive in PIF for lengthy periods of time during normal product storage (10, 20). Furthermore, Yan et al. (50) posit that the adaptation of this pathogen to the PIF manufacturing environment could lead to the survival of these organisms in finished product and increase the risk of causing infections once the contaminated food is consumed.
Even though researchers have examined the presence of Cronobacter spp. and Salmonella in the processing environment in food manufacturing facilities, the studies normally targeted a single facility or a small number of facilities in a region. Large-scale or nationwide surveillance of Cronobacter in milk powder manufacturing facilities has never been reported in the United States, and there are no recently published studies of Salmonella prevalence in milk powder facilities. The U.S. Food and Drug Administration (FDA) conducts sampling assignments to update knowledge on known hazards, as well as to collect data on emerging hazards. A sampling assignment was conducted in 2014 to ascertain the prevalence of Cronobacter spp. and Salmonella in the processing environment of facilities manufacturing milk powder in the United States. In this article, we report the findings of this survey.
MATERIALS AND METHODS
Milk powder manufacturing establishments under FDA jurisdiction were identified across the contiguous United States for a sampling assignment. The inspections were conducted in fiscal year 2014 (1 October 2013 to 30 September 2014), with 58 facilities inspected and sampled during the assignment.
Sample collection and transfer
Environmental samples for Cronobacter and Salmonella were collected from each facility (the target was a minimum of 100 subsamples for each organism from each facility). FDA uses the term “environmental sample” or “sample” to denote all of the swabs and sponges collected from a food manufacturing facility during 1 day of sampling. The individual swabs and sponges from separate sampling locations in the facility are termed “subsamples” or “subs” of the sample from that facility. Subsample collection was focused primarily on zones 1 to 3 (Table 1), and environmental swabs were collected as outlined in chapter 4 of the FDA's Investigations Operation Manual (47). The swabs for Cronobacter and Salmonella were collected in tandem. Sampling was focused on potential niche areas for these organisms, such as equipment with hollow bodies due to poor hygienic design, areas that would be wet and redried, and areas that were cracked or had rough surfaces. The areas sampled included the floors, drains, and equipment frames (including near the floor); instantizing operations, including fluidizer beds and rewetting chambers, as well as vitamin addition equipment; flap or rotary valves after dryers, baghouses and sifters, dryer explosion chambers, powder hoppers in packaging areas, and powder silos interiors. In most cases, the subsamples were collected from areas of the facility where postdrying activities were conducted, but in some facilities subsamples were also collected from predrying operations.
Each FDA district had a specific servicing laboratory identified to analyze the samples. On the day of collection, the samples were placed into insulated shipping containers containing frozen gel packs and shipped via UPS next day air early morning delivery. If the samples could not be analyzed on receipt at the laboratory, they were stored at 4 ± 2°C. Sample analysis was initiated no later than 48 ± 2 h after sample collection.
Analysis of environmental samples
Cronobacter spp. analysis was conducted as outlined in chapter 29 of the Bacteriological Analytical Manual (BAM) (44). Isolate identifications were conducted using bioMerieux Rapid 32 E identification strips or the VITEK 2 (bioMerieux, Hazelwood, MO).
Frequency tables were created to examine the association between Cronobacter and zone, product, and environmental monitoring program, respectively, as well as for Salmonella and zone. A chi-square or Fisher's exact test, where appropriate, was used to test for independence. R software (https://www.R-project.org/) (38) was used for all statistical analyses.
Data from 3 of 58 inspected facilities were excluded from analysis because microbiological analysis was not conducted on the samples from one facility and two of the facilities were not producing milk powder at the time of sampling. Therefore, data were available from 55 facilities.
In these 55 facilities, 5,671 environmental subsamples were collected and analyzed for Cronobacter, an average of 103 subsamples per facility. The minimum subsample number collected from one facility was 55, the maximum was 176. Cronobacter was detected in the environment of 38 (69%) facilities. In the 38 facilities where Cronobacter was detected, 4,005 subsamples were collected, of which 250 were positive for Cronobacter, yielding a prevalence of positive environmental subsamples of 6.25%. The overall prevalence of Cronobacter in the 5,671 subsamples from 55 facilities was 4.4%. The number of subsamples that tested positive for Cronobacter within a single facility ranged from 0 to 41 (0 to 41%).
For Salmonella, 5,714 subsamples were collected and analyzed, an average of 104 per facility. The minimum subsample number collected from one facility was 55, the maximum was 176. Salmonella was detected in the environment of 3 (5.5%) of the facilities (359 subsamples were collected, of which 7 were positive). In the three facilities in which Salmonella was detected, the average prevalence was 2.5%. The overall prevalence for Salmonella in the 5,714 subsamples collected from 55 facilities was 0.16%.
For statistical analysis regarding the presence of Cronobacter in milk powder facilities, a subset of data from 20 facilities (1,989 subsamples) was used. This subset of facilities was selected because it had the most complete information on zones sampled and product types during sampling and because Cronobacter was detected in at least one environmental subsample in all 20 facilities. In these 20 facilities, most of the subsamples that were collected and tested for Cronobacter were from zones 2 and 3 (Table 2). Cronobacter was most frequently detected in zone 4 (14.3% of subsamples positive), followed by zone 3 (8.7%), zone 2 (4.5%), and zone 1 (1.1%). The prevalence of Cronobacter across the zones was not the same (P < 0.05), although the sampling was biased toward the collection of samples from zones 2 and 3.
The prevalence of Cronobacter as a function of the product manufactured at the time of sampling was also analyzed (Table 3). The products were lactose, whey products, buttermilk powder, and nonfat dried milk (NFDM). In some facilities, samples were collected across different product categories (i.e., the facility was manufacturing more than one product during the inspection); therefore, the “number of facilities” in Table 3 adds to greater than 20. There was no significant association between product type and prevalence of Cronobacter spp. in the facility (P = 0.167).
Information about the environmental monitoring program in a facility was collected during some of the inspections (Table 4, these are the same 20 facilities used for the statistical analysis of Cronobacter, plus one additional facility that tested positive for Salmonella). The information was collected during an interview with facility personnel; the details were not verified by reviewing the facilities' programs. Of the 10 facilities with data available, all except one had a program that included testing for Salmonella. Two of the facilities reported that their plan included testing for Cronobacter, and one other facility reported testing for Enterobacteriaceae (Table 4). Statistical analysis was not included due to the small size of the data set.
Salmonella was detected in the environment of three facilities (Table 5). All three facilities where Salmonella was detected were producing a dried whey product at the time of sampling. In these three facilities, 89.7% of the subsamples that were collected and tested for Salmonella were from zones 2 and 3. Salmonella was most frequently detected in zone 4 (21.7% of subsamples positive), followed by zone 3 (2.6%). Salmonella was not detected on surfaces in zone 1 or zone 2 in any of the facilities. As with Cronobacter, a majority of the Salmonella subsamples were collected from zones 2 and 3.
The prevalence of Salmonella and Cronobacter in the environment of three milk powder production facilities that yielded Salmonella-positive samples is compared in Table 6. Due to the low number (three) of facilities with Salmonella detected in the environment, statistical comparisons were not made between Cronobacter and Salmonella data in the facilities. No clear pattern between the presence of Salmonella and Cronobacter was noted. For example, Cronobacter was not detected in the environment of facility A, which had greatest number of Salmonella positives (Table 6).
This sampling assignment was conducted in 2014, prior to the FDA's publication in 2015 of the regulation “Current Good Manufacturing Practice, Hazard Analysis, and Risk-Based Preventive Controls for Human Foods” (21 CFR Part 117). This regulation applies to milk powder manufacturers and defines the term “environmental pathogen.” It requires that manufacturers producing ready-to-eat food consider the risk of contamination by environmental pathogens during their hazard analysis and, if these hazards require a preventive control, that they implement appropriate controls and verification of these controls. In 2014, an environmental monitoring program to verify control of environmental pathogens was not a regulatory requirement.
The milk powder manufacturing process uses pasteurized milk, which is then concentrated and dried through evaporation and spray drying, respectively. Cronobacter and Salmonella are inactivated during pasteurization of milk; therefore, presence of these organisms in the final milk powder product indicates contamination from the postpasteurization environment or from the addition of contaminated ingredients after pasteurization. Depending on the application, milk powder can be considered a ready-to-eat product, which can be prepared and consumed without a kill-step (e.g., in the home of the consumer). It is also used as an ingredient in the manufacturing of numerous food and beverage products, including products for infants (such as PIF), the elderly, and the immunocompromised. Depending on the application, the milk powder may or may not undergo an additional lethality treatment. Therefore, if milk powder is to be used as an ingredient in ready-to-eat products for infants, the immunocompromised, or the elderly, it is important to source this material from manufacturers that apply strict hygienic control measures and production strategies to prevent recontamination by environmental pathogens such as Salmonella and Cronobacter.
In the current study, environmental samples were collected from 55 milk powder production facilities and tested for Cronobacter and Salmonella. In other published literature investigating Cronobacter in milk powder processing facilities, samples were taken from a smaller number of facilities, typically in the range of three to five (e.g., references 9, 16, 21 ). In the present study, Cronobacter was isolated from the environment of 38 (69%) of 55 milk powder production facilities. In previous studies, Cronobacter was detected in the manufacturing environment of all milk powder facilities that were tested, e.g., five of five facilities in Australia (9); three of three facilities in China (16), four of four facilities in Europe (29).
Although Cronobacter was detected in at least one environmental subsample in 69% of the facilities, the prevalence within the facilities was generally low, with an average of 6.25% of the subsamples positive in the 38 facilities where Cronobacter was detected. When considering the overall prevalence in the samples collected from the 55 facilities, 4.4% of the subsamples tested positive for Cronobacter. These numbers are lower than what has been reported in some other studies. For example, Craven et al. (9) reported that 32% of their 298 samples were positive; Kandhai et al. (29) collected 68 samples and reported a range of 9 to 35%. (Note: In the literature on this topic, the term “sample” is used to denote a location within a manufacturing facility where a single sponge or swab is taken and tested; it is equivalent to the FDA term “subsample.”) The prevalence observed in the current study is more in line with the findings of Fang et al. (16), who reported that 10.6% of the samples collected from goat milk powder facilities were positive for Cronobacter. Other studies have focused on sampling in one facility, in some instances with multiple visits over an extended period (e.g., months). In these studies, the prevalence of Cronobacter on environmental surfaces has varied; for example, Reich et al. (39) reported 0%, in contrast with Hein et al. (21), who reported 39.4%. These studies have been conducted with different sampling and testing methodologies and have been conducted in different countries, which may contribute to the variation in results. Another possible reason for a higher prevalence in studies conducted more than 10 years ago is that industry awareness regarding Cronobacter has increased, which may have led to improved control measures in recent years.
In contrast to Cronobacter, Salmonella was found in only 3 of the 55 milk powder production facilities in the United States. These results are consistent with earlier studies targeting Salmonella in milk powder plants. In 1985, a study was undertaken by the National Research Council of the U.S. Department of Agriculture Salmonella surveillance program for nonfat dried milks. The study found that numerous milk powder plants were still not designed to ensure the containment of microbiological contamination, in particular Salmonella (31). The level of Salmonella contamination in skim milk powder in the United States was found to drop from 0.44% in 1980 to 0.06% in 1988 (31). Environmental samples collected as part of the Salmonella surveillance program from dry skim milk facilities from 1966 to 1985 ranged from a high of 8.2% in 1967 to a low of 2.3% in 1985 (31). The reduction during this period was attributed to the gradual introduction of preventive measures in the form of codes of practice by organizations such as Codex Alimentarius (7). These control measures focused on (i) prevention of ingress by Salmonella into the plant; (ii) avoidance of growth and spread in case of entry; (iii) application of hygienic controls, equipment and facility hygienic design, and hygienic zoning principles; and (iv) establishment of a raw material control program and use of Salmonella-free ingredients (18). Today, the incidence of Salmonella in dry milk powders can be considered to be rare. The few events that are reported are typically attributed to errors in the application of the preventive hygiene measures or a breakdown of hygienic controls.
Therefore, the questions that should be asked and answered are, is it possible to achieve the same level of control for Cronobacter as that achieved for Salmonella? Are there differences between the control measures for Cronobacter and Salmonella? Cordier (8) suggests that the preventive control measures described above for Salmonella form the basis of the management and control for Cronobacter and are also a prerequisite to control this environmental pathogen in dry powder processing operations. Cordier (8) goes on to recommend that control measures more stringent than the existing strategy will be necessary to control Cronobacter. The application of existing hygienic practices targeted toward control of Salmonella will only minimize, and not completely suppress, the presence of Cronobacter. The most successful additional strategy to impact the control of this pathogen is the very strict management of moisture in the processing environment, with the target being complete elimination of any water. This is a difficult task; the cleaning and sanitation practices currently observed for some dairy powder operations serve to introduce moisture into the manufacturing environment and equipment interiors, for a variety of reasons: e.g., formulation changeover, allergen control, equipment design. This results in a risk for both Salmonella and Cronobacter proliferation within manufacturing environments and equipment. Consequently, manufacturers may wish to consider utilizing additional hygienic measures and implementing additional environmental monitoring for drying systems, especially with equipment postdrying chambers. In respect to Cronobacter and Enterobacteriaceae in general, even the slightest traces of water can lead to rapid increase in the population and a higher probability of process contamination (8). Therefore, it is imperative that strict management of water be constantly applied for the processing environment, near the process line and wherever the product may be exposed to the environment (such as filling), and for packaging areas, in areas where condensation might occur with cooling within equipment after shutdown (e.g., baghouses), and in equipment where moisture is intentionally introduced (e.g., instantizers). These areas of the process are more commonly managed at a higher hygienic control than warehouses or nonproduction areas of a dry powder plant. Where Cronobacter has been identified as a hazard of concern, facilities should consider the hygienic design of their processing equipment and modify sanitation procedures specifically to address this pathogen. For example, they could employ dedicated lines where strict hygienic control measures are applied to prevent line recontamination from the environment.
In the current study, both Cronobacter and Salmonella were detected more frequently in zones 4 and 3. Zone 4 areas are those outside the food production areas, including employee locker rooms, dry goods storage warehouses, finished product warehouses, cafeterias, hallways, and loading dock areas. Although the percentage of zone 4 subsamples collected in the present study was low compared with zone 2 and 3 subsamples, the results of this study are consistent with other studies. Craven el al. (9) also reported that the occurrence of Cronobacter was higher in “nonprocessing areas” (49% positive rate). Zone 4 areas are not likely to be cleaned with the same regularity or thoroughness as food processing areas; zone 4 areas may also accumulate dust and can be subject to higher moisture conditions due to roof leaks, faulty sprinklers, leaking water or steam valves, or a drain backup. Dust has been reported to be a source of Cronobacter (32). In the current study, the data collection format was not conducive to making conclusions about the specific sites where Cronobacter was isolated. Other studies that were designed to explore this question have reported areas that frequently tested positive for Cronobacter: spray drying rooms (9, 16), entrances to spray drying rooms (9), and packaging rooms (16), as well as air filters (33), floors, and vacuums (23).
This study presented data on the prevalence of Cronobacter and Salmonella in milk powder production facilities in the United States. Understanding the prevalence of Cronobacter and Salmonella can help manufacturers design programs to control these pathogens in these facilities. It would appear that many U.S. milk powder production facilities are doing a good job of controlling Salmonella in the environment. However, Cronobacter was found in many of the facilities. The disparity between Salmonella and Cronobacter prevalence indicates that controls for Salmonella may not eliminate Cronobacter from the environment. Where Cronobacter is a hazard of concern, facilities should strictly manage moisture in the dry processing operations close to product flow and should design sanitation procedures to specifically address this pathogen. Manufacturers should also identify and address possible routes of contamination for Cronobacter into the finished product intended for infants and sensitive consumers, including the potential of contamination from zone 4 areas into processing areas.
The authors thank FDA colleagues for their assistance in designing and executing this project. In particular, we acknowledge Edette Newby, Tom Hammack, and Stuart Chirtel. We also thank Jenny Scott for reviewing the document.