Despite education efforts, consumers often practice unsafe food handling and storage behaviors. Little is known about how these unsafe practices contribute to contamination of the home kitchen with foodborne pathogens. In addition, only a limited number of studies have examined the role of the kitchen as a reservoir for pathogens. The purpose of this study was to characterize microbial contamination and foodborne pathogens found in home kitchens and determine whether contamination was significantly associated with unsafe or unsanitary conditions observed in the kitchen. Swab samples were collected from food contact and preparation surfaces in homes (n = 100) in Philadelphia, PA. The samples were tested for coliforms, fecal coliforms, Escherichia coli, Staphylococcus aureus, Salmonella, Campylobacter, and Listeria. Fecal coliforms were found in 44% of homes (most often in samples from kitchen sinks, sponges, and dishcloths), and E. coli was found in 15% of homes (mostly in samples from kitchen sinks). Nearly half (45%) of the homes tested positive for a foodborne pathogen, and 12% had multiple pathogens present in the kitchen. S. aureus was isolated from 39% of homes, most often from countertops and refrigerator door handles. Listeria spp., including L. monocytogenes and L. innocua, were present in 15% of homes, most often in samples from refrigerator meat drawers. C. jejuni was isolated from 3% of homes. Contamination with Listeria was significantly associated with higher refrigerator temperatures. The contamination of surfaces with fecal coliforms and S. aureus was significantly associated with a lack of cleaning materials: dish soap and paper or cloth towels in the kitchen, and any type of towel in the nearest bathroom. The contamination of a sponge or dishcloth with either fecal coliforms or S. aureus was predictive of other surfaces in the kitchen having the same contamination, indicating that sponges and dishcloths are both reservoirs and vectors for bacteria in the kitchen.
In the United States, the majority of hospitalizations and deaths from foodborne illness are caused by bacterial infections (29). Five bacteria are responsible for 92% of these hospitalizations and 85% of these deaths: Escherichia coli, Staphylococcus aureus, Listeria monocytogenes, Salmonella spp., and Campylobacter spp. (29). Despite improvements to food safety–related conditions in food service and retail environments over the past 10 years (32), data from the Centers for Disease Control and Prevention Foodborne Illness Active Surveillance Network (FoodNet) indicate that the rates of foodborne illness caused by most major pathogens have not significantly decreased during this time frame (7).
The total number of infections caused by food prepared in the domestic environment is not known; however, the Centers for Disease Control and Prevention reports that private homes are the second most common location associated with outbreaks of foodborne disease (14). These data are helpful estimates but do not represent the true incidence of illness associated with private homes because these cases are more likely to be sporadic infections (21). Consumer food handling behavior in the home is considered the final defense against foodborne illness (9, 27). Pathogens may be introduced into the domestic environment via naturally contaminated raw foods, transfer from the environment (carried in by animals or insects), or transfer from another person (fecal-oral contamination and aerosolization). Under the right conditions in the home, these organisms have the potential to survive, grow, spread to other surfaces, persist for long periods of time, and contaminate food.
Although there is a large body of research documenting improper food handling and lack of knowledge among consumers, which may lead to foodborne illness, there is limited research examining whether unsafe food handling practices or lack of sanitation (elimination of pests, proper hand washing, availability of cleaning supplies, etc.) contributes directly to contamination of the domestic kitchen with foodborne pathogens (8, 18, 22).
The purpose of this study was to investigate the presence of a range of bacterial pathogens and sanitation indicators on a variety of kitchen surfaces in homes in a low-income urban environment. In addition, we sought to determine whether contamination could be predicted using or correlated with potentially unsafe or unsanitary conditions observed in the home.
MATERIALS AND METHODS
Study design and recruitment
The study was advertised and took place over a 12-month period from January to December 2013. A representative sample of 100 residents from Philadelphia, PA, was recruited for the study. All materials and methods used in this study were approved by Drexel University's Institutional Review Board.
Advertisement flyers were posted in a total of 34 locations throughout Philadelphia (in local businesses, supermarkets, libraries, and community centers). The flyers included a brief description of the study, including the eligibility criteria, time commitment, and compensation offered ($50). Potential participants who contacted the researcher were told that the purpose of the study was to collect information to help understand different types of food preparation and were discouraged from preparing or cleaning their home differently than usual. Volunteers were screened for eligibility based on the following criteria: 18 years of age or older, living in Philadelphia, preparing food at home at least three times per week, and willingness to allow researchers to visit the home and collect data. Eligible participants were enrolled in the study and a home visit was scheduled on an upcoming Monday or Tuesday. In addition to microbial swabs taken during the home visit, the home was visually audited for potential unsafe or unsanitary conditions using a previously published audit tool (2) developed for the domestic kitchen. The results of the visual audit have been reported elsewhere (3) and are discussed in this article only with respect to how the conditions observed correlated to microbial contamination.
The study was designed to last a full 12 months, throughout which homes were visited at a rate of approximately three per week. These measures were taken to mitigate the effects of changes in ambient temperature between the seasons on the microbiological data (24). It was estimated that a maximum of eight or nine homes could be visited each month because of the amount of time necessary to process the microbiological samples. Based on these factors, the goal for enrollment was set at 100 homes.
Sample collection and preparation
Surface samples were collected from domestic kitchens using a sterile sponge moistened with 10 mL of Dey-Engley neutralizing buffer (VWR International, West Chester, PA) and attached to a removable handle (Sani-Stick Sponge Handle Sampling System, Labplas; VWR International). These sponges were used to aseptically swab five locations: refrigerator door handles (front and back), refrigerator shelf (middle of the bottom shelf), refrigerator meat drawer (when present), and kitchen counter (portion of the counter next to the sink). At each site, an area of approximately 25 cm2 was wiped thoroughly with the sponge, which was then removed from the handle and sealed in a sterile bag. With the participant's permission, and when available, a previously used kitchen dishcloth or sponge was also collected and sealed in a sterile bag. All samples were stored on ice during transport and kept at 4°C in the laboratory. In addition to the sample collection, the refrigerator temperature was measured and recorded at each home using a pen-style infrared thermometer (VWR International). This was completed immediately after opening the refrigerator door for the first time, with the thermometer held approximately 5 cm from a food container as close to the middle of the refrigerator as possible.
All the samples were analyzed within 24 h of their initial collection. Each sample was homogenized for 2 min with either 60 mL (for the surface samples) or 200 mL (for the kitchen dishcloth and sponge samples) Dey-Engley neutralizing buffer. A paddle blender (Stomacher 400, Seward, Norfolk, UK) set at 230 rpm was used to mix all the surface samples; the kitchen dishcloth and sponge samples were systematically mixed by hand. The total volume of the homogenate (representing a 100 dilution) was then divided for microbial testing.
All samples were evaluated for the presence of four common foodborne pathogens (S. aureus, Campylobacter jejuni, L. monocytogenes, and Salmonella spp.). The samples were also tested for quality indicators (coliform bacteria, fecal coliforms, and E. coli), which were enumerated using the most-probable-number (MPN) technique (31). All tests were completed according to methods described in the U.S. Food and Drug Administration Bacteriological Analytical Manual (31). The presence of S. aureus strains was confirmed using a positive coagulase test (31). Presumptive-positive isolates of C. jejuni, L. monocytogenes, and Salmonella were confirmed using rapid bacterial identification test kits: API Campy, API Listeria, and API 20E (bioMérieux, Marcy-l'Étoile, France). Presumptive E. coli isolates were confirmed using the IMViC test series (BD, Franklin Lakes, NJ). Negative controls (sterile buffer solution) and positive controls for each organism (E. coli, ATCC 25922; S. aureus, ATCC 25923; C. jejuni subsp. jejuni, ATCC 700819; L. monocytogenes, ATCC 19115; and S. enterica subsp. enterica serovar Enteritidis, ATCC 13076) were run with each set of samples tested.
For purposes of data analysis, coliform, fecal coliform, and E. coli counts that were below the detection limit (<0.3 MPN/mL) were assigned a value halfway between the absence of cells and the detection limit, 0.15 MPN/mL. Similarly, counts that were above the detection limit (>1,100 MPN/mL) were estimated with a value 1 greater than the detection limit, 2,400 MPN/mL. The enumeration data for coliforms, fecal coliforms, and E. coli were recorded as MPN per milliliter and multiplied by either 60 or 200 mL per sample to estimate the MPN per sample. The MPN counts were finally converted to a base-10 logarithmic scale and averaged to calculate the mean log MPN per sample for each location. The MPN counts per household were also calculated by averaging the MPN counts of the individual samples collected from each home (mean log MPN per home). The presence or absence of fecal coliforms and pathogens was recorded for all samples and reported as the percentage positive per location (out of the total number of samples collected from each sampling site). The presence or absence of fecal coliforms and pathogens was also reported by household, representing the total number of homes where contamination was found as a percentage of all homes surveyed.
We completed the statistical analyses using the Statistical Package for the Social Sciences, release 22.0.0 (IBM Corporation, Armonk, NY). We calculated the descriptive statistics and frequencies for all the qualitative data. Independent t tests were run to identify differences between in-group means for numerical variables (refrigerator temperature and enumeration counts). Dichotomous variables (participant demographics and presence or absence data) were compared using cross-tabulations and Pearson's chi-square test to evaluate the differences between groups. We calculated the relative risk to evaluate trends in microbial contamination associated with the observed conditions.
A total of 100 participants completed this study. The participants were mostly female (86%) and represented a wide variety of ages, education levels, economic statuses, and racial and ethnic backgrounds (Table 1). The subjects varied in age from 18 to 84 years old, with the highest proportion (29%) falling in the range 45 to 54 years. The most commonly reported annual income was less than $15,000 (27%), and African American was the most common race and ethnicity (47%). Chi-square goodness-of-fit tests show no significant differences between the sample and the population (of Philadelphia, PA) patterns for race and ethnicity (χ2 goodness-of-fit = 6.203; 4 df; P = 0.18). However, the sample pattern for household income level was significantly different from the population pattern, with a greater representation of low-income subjects (χ2 goodness-of-fit = 15.341; 4 df; P = 0.004). The education level was almost evenly split between participants who had completed college (41%) and participants who had completed high school or a GED (43%). When asked what language was used at home, 8% of subjects reported speaking a language other than English at home at least half of the time. The majority (60%) of participants had experience working in the food service or food industry, either from previous or current employment.
Among the 100 homes assessed in this study, the refrigerator temperatures ranged from below freezing, −3.3°C (26.0°F), to very warm, 12.0°C (53.6°F), with a mean temperature of 4.3°C (39.7°F). Almost half (48%) of homes had a refrigerator that was too warm (>4.4°C [40.0°F]), and 11% had a refrigerator temperature at or above 7.2°C (45.0°F). These data were previously reported elsewhere (3), but we include them here because there are associations between the refrigerator temperature and microbial contamination.
Contamination present in domestic kitchens
Fecal coliforms were present in 44% of the domestic kitchens, and E. coli present in 15% (n = 100). Nearly half (45%) of homes had at least one target foodborne pathogen (S. aureus, Listeria, or C. jejuni) present, and 12% had multiple pathogens present in the kitchen. S. aureus was found most often, isolated from 39% of the domestic kitchens. Listeria spp. were present in 15% of the kitchens, and C. jejuni was present in 3% of the kitchens. Salmonella spp. were not found in any homes. Three homes had L. monocytogenes present in the kitchen, and in one case, L. monocytogenes was isolated from multiple sites in a single kitchen.
Locations of contamination in domestic kitchens
Samples were collected from up to six locations in each home, yielding a total of 557 samples. We used coliform levels to evaluate the overall sanitation of the kitchens and to compare cleanliness of the different areas (Table 2). Dishcloths and sponges had the highest levels of coliform bacteria, with an average of 4.62 log MPN per sample; 64% of the samples had coliform counts above 5.0 log MPN per sample. Refrigerator door handles had the lowest levels of coliforms among all the sites sampled (mean of 1.36 log MPN per sample), and the highest proportion (66%) of samples with MPN counts below 1.0 log MPN per sample.
Fecal coliforms were isolated from 15% (n = 83) of all the samples collected, and were found most frequently in samples from kitchen sinks (Table 3). Among these positive samples, the highest levels of fecal coliforms were found on dishcloths and sponges, and the lowest levels on refrigerator door handles. A total of 3% (n = 18) of all the samples tested positive for E. coli. The presence of E. coli was highest in the samples from kitchen sinks, followed by dishcloth and sponge samples and samples from refrigerator shelves (Table 3).
Foodborne pathogens were isolated from 17% (n = 94) of all samples collected during this study. S. aureus was isolated most frequently and was found in 14% (n = 75) of the samples, most often from kitchen counters and refrigerator door handles (Table 3). Listeria spp. were isolated from 3% (n = 16) of the samples, most commonly from refrigerator meat drawers. These 16 samples represent several different species of the genus Listeria, including L. innocua, L. welshimeri, L. grayi, L. seeligeri, and L. monocytogenes. L. monocytogenes was isolated from one sample each from a refrigerator door handle, refrigerator drawer, kitchen sink, and dishcloth or sponge. C. jejuni was detected from only three samples (0.5% of the total), one sample each from a kitchen sink, sponge or dishcloth, and counter.
The data for the presence or absence of fecal coliforms and target organisms, average coliform level (MPN per home), and refrigerator temperature were grouped by season and analyzed to determine the effect of warm ambient temperatures on microbiological samples (using cross-tabulations and the chi-square statistic for presence or absence and independent t tests for the numerical data). Samples were collected consistently over 12 months, resulting in 17 homes being visited during the winter (28 January to 18 March 2013), 27 homes during the spring (25 March to 17 June 2013), 25 homes during the summer (24 June to 16 September 2013), and 31 homes during the fall (23 September to 16 December 2013). The presence of target organisms and fecal coliforms did not vary by season, nor did the average refrigerator temperature recorded in the homes. One significant relationship between season and level of contamination was observed for the average level of coliforms in the homes (3.25 log MPN per home in the summer compared with 2.68 log MPN per home in the fall; t = −2.223; P = 0.03).
Factors associated with microbial contamination in homes
At the time of the microbial sampling, each home was also visually audited using a previously published audit tool (2) developed for the domestic kitchen. The results of this visual audit have been previously published (3). The results we describe here indicate where microbial contamination was found to be significantly (P < 0.05) associated with an unsafe or unsanitary condition observed in the home.
Relative risk is a proportional measure that describes the probability of an event occurring in an exposed group, comparing the results with the probability of the same event in a nonexposed group. This method can be used to estimate the effect of specific conditions on microbial data (17). In this case, we consider the microbial contamination to be the event and the observation of an unsafe or unsanitary condition in the home to be the exposure. A relative risk of 1.0 indicates there is no difference between the two groups, whereas a score of more than 1.0 suggests that the event (contamination) was more likely to occur when the unsafe or unsanitary condition was observed.
Independent t tests indicated that refrigerators that tested positive for Listeria spp. had significantly higher mean operating temperatures (6.7°C [44.0°F]) than uncontaminated refrigerators (4.2°C [39.5°F]) (P = 0.004). Similarly, refrigerators that tested positive for fecal coliforms had significantly higher mean operating temperatures (6.9°C [44.5°F]) than uncontaminated refrigerators (4.2°C [39.5°F]) (P = 0.022).
Coliform counts were higher in homes that lacked sanitizer or disinfectant products and dish soap in the kitchen, as well as in homes without hand-drying towels in the bathroom (Table 4). Only homes with bathrooms on the same floor as the kitchen were observed for soap and towels in the bathroom (n = 48). Similarly, a lack of cleaning materials (sanitizer products, dish soap, and hand-drying towels) in the kitchen and bathroom, as well as a lack of observed surface cleanliness, was associated with the presence of fecal coliforms, E. coli, and S. aureus in various locations in the home (Table 5). Interestingly, the presence of contamination by fecal coliforms and S. aureus that was found in certain areas (refrigerator shelves, kitchen counters, sinks, and refrigerator door handles) of the consumer homes was significantly associated with the presence of the same type of contamination isolated from a sponge or dishcloth in the same home (Table 6).
This research identified the unsafe cold storage of food by consumers due to refrigerator temperatures above the recommended 4.4°C (40°F), and these results are consistent with previously reported research in this area (13, 19). However, the research reported here also shows that refrigerators with higher temperatures were themselves significantly more likely to be contaminated with Listeria spp. and fecal coliforms.
Coliform bacteria were detected in all locations sampled in the kitchens, with a wide range of contamination levels. Kitchen sinks and kitchen dishcloths and sponges were the most contaminated areas (46 and 64% of samples >5 log MPN), while refrigerator door handles were the least contaminated areas (66% of samples <1 log MPN). Recent studies (20) have reported similarly high amounts of coliform bacteria present in dishcloths and sponges. These results indicate that kitchen sponges and dishcloths may serve as reservoirs and potential vectors for contamination in consumer kitchens and that there may be a need to focus food safety education efforts on how to properly care for and handle these common household items.
Fecal coliforms, E. coli, and S. aureus were isolated frequently from consumer homes during this study (44% were positive for fecal coliforms, 15% for E. coli, and 39% for S. aureus). Although these frequencies are lower than reported by some—Carrasco et al. (5) and Josephson et al. (18) isolated fecal coliforms from 48 to 67% of the dishcloths and sponges and from 63% of the kitchen sinks they tested —others found similar frequencies: E. coli in 3 to 11% of dishcloth and sponge samples (23, 25) and S. aureus in 41 to 43% of homes tested (4, 19). S. aureus was found inside 14% of the domestic refrigerators sampled in this study and was isolated from multiple surfaces in 2% of these refrigerators. The presence of S. aureus in domestic refrigerators represents a specific food safety concern due to the organism's ability to grow and produce toxins when exposed to mild temperature abuse (15, 16). Conflicting reports have been published regarding this contamination. Three studies (6, 16, 25) isolated S. aureus infrequently (1 to 6%) from refrigerators, while other findings (12, 19) suggest a much higher prevalence (33 to 41%) of contamination on these surfaces.
Fecal coliforms, E. coli, and S. aureus are all associated with personal hygiene, and in this study, they were all frequently isolated from hand contact sites (dishcloths and sponges, countertops, and refrigerator door handles). A lack of cleaning products in the home, including paper and cloth towels, was strongly associated with several types of contamination, including coliforms, fecal coliforms, E. coli, and S. aureus. These findings suggest that a significant transfer of these organisms occurs in domestic kitchens, assisted in some cases by consumers' failure to wash their hands correctly. In other cases, this transfer may occur via the cleaning utensils, such as sponges and dishcloths, because we found a significant association between contamination on kitchen sponges and dishcloths and contamination in other parts of the kitchen of the same home. This pattern suggests that, rather than cleaning surfaces as they are intended to do, these items may actually do the opposite if they are not cleaned properly and regularly.
Due to the severity and high mortality rate (19.5%) associated with listeriosis (7), the prevalence of L. monocytogenes in domestic kitchens may be a cause for concern. Five species of Listeria were isolated from 15% of the homes during this study; of these, 3% of homes had L. monocytogenes present in the kitchen. Listeria was isolated from 9% of the refrigerators sampled, and one refrigerator (1%) harbored L. monocytogenes. Several other studies (1, 10, 11, 16, 19, 28, 30) confirm low levels of surface contamination by Listeria spp. (1 to 9%) and L. monocytogenes (1 to 6%) in domestic refrigerators. In 2013, Macias-Rodriguez et al. (20) reported that 60% of the refrigerator shelves they sampled in consumer homes in Mexico tested positive for L. monocytogenes. This high frequency of contamination is unprecedented for Listeria and may be related to the greater consumption of unpasteurized milk and cheese that is common among Hispanic and Latino populations. Our results indicate Listeria contamination was more likely to occur in refrigerators operating above the recommended temperature (4.4°C [40°F]). Proper refrigeration practices are clearly lacking among consumers and should be emphasized to reduce risk of listeriosis.
Campylobacter is notoriously fragile in the environment and difficult to isolate and culture successfully. Despite this fragility, we were able to isolate viable C. jejuni from three kitchens. Only one other study reported the presence of Campylobacter on kitchen surfaces during “everyday activity” (i.e., not during or directly following meal preparation): Josephson et al. (18) isolated C. jejuni from 2% of kitchen sinks. Although these numbers are low, the fact that viable C. jejuni can be isolated from kitchens in which food was not actively being prepared indicates that kitchens can serve as a reservoir for C. jejuni that may later cross-contaminate other foods.
In our study, no samples tested positive for Salmonella. Although not typically considered to be as fragile as Campylobacter, Salmonella has also not been frequently identified in consumer homes. Most studies reported Salmonella as being present on less than 5% of samples from dishcloths and refrigerators (6, 12, 16, 18, 19, 22, 28), and two reported a much greater frequency (between 10 and 33%) of positive samples (20, 26). Considering that only one-third of these studies was conducted in the United States, the differences in methodology and sample populations may account for the discrepancies between these studies and our results.
It is evident that a wide range of foodborne pathogens and contamination may be present in the domestic kitchen at any given time and that they can persist long after food preparation takes place. These findings highlight the importance of all safe food handling practices in the home, especially practices that can reduce microbial transfer between surfaces: proper cleaning, sanitation, and personal hygiene. Research on consumer education should investigate the lack of cleaning supplies as a possible barrier to performing recommended food safety behaviors, especially among low-income populations. Further research is needed to elucidate the prevalence of Campylobacter, Salmonella, and other major pathogens in the domestic environment to improve ongoing efforts to reduce the burden of foodborne illness among consumers.
This research was supported by the U.S. Department of Agriculture, Cooperative State Research, Education, and Extension Service grant no. 2009-51110-05853. Special thanks to Dr. Shauna Henley for help with this research.