Estimating Risk by Measuring Coliform on Common Touch Surfaces
JoAnn Xiong-Mercado, CP-FS
Food Safety Educator & Outreach
Department of Food & Consumer Safety
Marion County (Indiana) Public Health Department
International Food Protection Training Institute (IFPTI)
2017 Fellow in Applied Science, Law, and Policy: Fellowship in Food Protection
Author Note
JoAnn Xiong-Mercado, CP-FS, Food Safety Educator & Outreach, Department of Food & Consumer Safety, Marion County (Indiana) Public Health Department.
This research was conducted as part of the International Food Protection Training Institute’s Fellowship in Food Protection, Cohort VI.
Correspondence concerning this article should be addressed to JoAnn Xiong-Mercado, CP-FS, Food Safety Educator & Outreach, Department of Food & Consumer Safety, Marion County (Indiana) Public Health Department, 3840 N Sherman Drive, Indianapolis, IN 46226; Email: JXiong@MarionHealth.org
Abstract
This exploratory study surveyed five percent of Marion County, Indiana restaurants by sampling dining room common touch locations for the presence of coliform bacteria. Coliforms, although not often a direct cause of foodborne illness, were chosen as indicator organisms due to their common association with soil and feces. Testing was limited to full-service, sit-down restaurants with ten or more employees. Four environmental samples were collected at each restaurant; no food samples or food contact surfaces were sampled as a part of the study. Full, routine inspections were not performed. Restaurants that were selected were given no advance warning. All samples were submitted to the Marion County Public Health Department laboratory for analysis. The study concluded that IDEXX Colilert-18® is an effective method for establishing the presence of coliforms on non-food contact surfaces within food establishments and thus identifying the possible need for improved routine cleanings. The study also explored a number of methodological issues in testing common touch surfaces. The study recommends the increased use of collecting common touch surface samples, improvements in cleaning regimens for food establishments, reconsidering dining room design, and the development of criteria for evaluating acceptable levels of coliform contamination on common touch surfaces.
Estimating Risk by Measuring Coliform on Common Touch Surfaces
Background
Pathogens, such as viruses and bacteria, can remain and survive on surfaces for extended periods of time and those surfaces, in turn, may become temporary environmental reservoirs that facilitate the spread of illness (Scott, 2013). Studies of currency, menus, home kitchens, and common items found in hospital settings established the risk of these items serving as illness vectors (Michaels, 2002; Choi et al, 2014; Donofiro et al, 2012).
Total spending on food away-from-home increased by 66.4%, when measured as a percentage of total consumer spending, between 1970 and 2012 (USDA, 2016); 53% of foodborne illness outbreaks are attributed to sit-down restaurants (CDC, 2014).
Standard environmental assessments conducted after outbreaks require food sampling, trace backs, employee health reporting and interviews. A lack of contaminated food available for sampling and asymptomatic food workers may be barriers in locating the source of foodborne illness. However, environmental sampling in dining areas may provide clues in cases where traditional means of information gathering fall short, e.g. environmental sampling for Norovirus after outbreaks on cruise ships has proven valuable (Park, 2015).
Problem Statement
Common touch surfaces in restaurant dining areas may serve as pathogenic reservoirs but the extent of the risk is largely unknown.
Research Question
Can environmental sampling in restaurant dining areas help illuminate risks associated with common touch surfaces and offer value as an investigative tool for regulators?
Methodology
Although they rarely directly cause foodborne illnesses, coliform bacteria were selected as indicator organisms as a means of establishing contamination. According to the CDC, the presence of coliforms generally indicates contamination by soil or feces. This study conducted environmental sampling for coliforms in the dining areas of selected restaurants. A randomized list of 879 full-service, sit-down restaurants with 10 or more employees in Marion County, Indiana was generated. The first 44 restaurants on that list were selected. Sampling was conducted during unannounced visits; no inspections were performed at the time of sampling. All samples were tested at the Environmental Microbiology Lab (EML) at the Marion County Public Health Department (MCPHD).
A hierarchical list of possible common touch sampling locations was created using subjective assumptions that certain items/locations may be overlooked in the course of normal cleaning regiments as a result of their ubiquity. The hierarchy of sampling locations, in order, included: laminated menus, high chairs, soda guns/soda self-service touch screens, women’s restroom inner doorknobs, salt and pepper shakers, ketchup and mustard bottles, the undersides of chairs, and the undersides of tables. Each item on the hierarchy was given a higher priority for sampling than the ones beneath that item in order to ensure that the sampler was not biased towards items that “looked dirty” to the sampler.
Four samples were collected from each of the 44 restaurants for a total of 176 samples. At least 4 of the 8 sampling locations included in the hierarchy were available at every sampling location. All of the samples for the study were taken with 3M Hydrated Sponges and transferred to 3M Petrifilm Coliform Count (CC) Plates.
The 3M Hydrated Sponge is pre-hydrated with a Letheen broth solution to retain viable coliforms. Efforts were made to take samples in a consistent manner with roughly equivalent amounts of surface area sampled in each instance. Due to inconsistent sizes and shapes, consistent sampling area sizes were not always possible to achieve for items such as door knobs. The ideal surface area to sample per sample was set at 187 in²--or the equivalent of two standard sheets of paper. Additionally, some items, such as soda guns, were consistently too small to meet the 187 in² standard and thus some caution should be made when comparing results across sample types.
Additionally, a one mL serial dilution at 10-7 of E. coli was used to establish a control. The same quantity was routinely plated and counted on a Petrifilm plate to ensure that no irregularities in sampling or incubation caused inaccurate results. Taking the same quantity, plating and recounting the bacteria on the plate after incubation each time other samples were done yielded colony counts between 35 and 106 for the control in each instance. A secondary control demonstrated that sponge-based capture resulted in a 1000-fold reduction in colony forming units (CFU) counted. For every bacterium on the plate, 999 remained in the sponge or on the surface that was sampled.
Samples were plated on the same day or refrigerated at 1°C - 5°C (34°-41°F) until plated. After 24 hours, the 3M Petrifilm CC Plate® was counted for the total number of visible colony units with gas production per the Interpretation Guide provided by 3M. Colony units without gas production were also noted; however, these colony units are not indicative of coliforms and were not included in coliform colony count.
Additionally, each sample was further tested against an enzyme substrate coliform test known as Colilert-18® as a secondary means of coliform detection. The 3M Hydrated Sponge® was submerged in single-strength Trypticase Soy Broth (SS TSB), incubated and visually inspected for any growth, cloudiness or particulate matter. Cloudiness indicated bacterial growth and prompted a Colilert test. In 100% of the samples, the SS TSB contained cloudiness and particulates and thus 100% of samples were subjected to the Colilert test.
Colilert is a coliform sampling test whose efficacy for testing water of all types is long-established and well-documented. Colilert’s use with regards to food sampling is rare but not unique to this study; a 2014 field study tested its efficacy for food sampling with good results (Rodrigues, 2014).
When Colilert is added to a liquid sample, an enzymatic reaction occurs causing the liquid to turn yellow in the presence of coliforms and fluoresce in the presence of E. coli specifically. The enzymatic reaction remains consistent if the quantities of enzyme and liquid being tested are maintained at an appropriate ratio. Accordingly, the test can be scaled down and a single test kit can be used to conduct dozens of tests in individual test tubes—making Colilert very cost-effective within this context. A provided “Quantitray” can also allow Colilert to quantify the number of coliforms per sample; however, the quantitray will not work if the test is performed without the standard amount of liquid per sample.
In instances where a positive Colilert result was noted when <1 CFU/mL had been counted on the plates, RapID was used as the final layer of testing to potentially validate the Colilert positive reading. Due to the costs associated with the RapID test, 9 out of the 67 potential false-positives were randomly selected for further testing. Another three samples with more than one visible colony and one with no visible colonies and a negative Colilert test were also run through RapID.
Due to a limited number of RapID test kits available, all RapID samples were given an oxidase test to ensure that the bacteria they contained were oxidase-negative. One additional sample “failed” the oxidase test by indicating the presence of oxidase-positive bacteria and was not tested. The oxidase-positive sample had been a non-typical (not clear or yellow) negative Colilert sample. Additionally, two other samples fluoresced indicating the presence of a β-glucoronidiase positive bacteria and were plated on an Eosin Methylene Blue (EMB) plate in order to isolate suspected E. coli instead.
RapID testing requires a pure isolate in order to identify a species. To obtain an isolate from a broth sample, MacConkey Agar (MAC) plates were used to separate bacteria by type. The agar in a MAC plate contains food mediums to support various bacteria and a dye which is pH sensitive. Coloration changes in the agar caused by bacterial activity allow the plate to sort bacteria into three groups. If the bacteria present metabolize lactose by fermenting it into acid, the pH of the plate will lower causing the dye to turn purple. If peptone is metabolized instead, ammonia is produced causing the dye color to fade or turn white. Otherwise, the sample retains the default yellow color. Most coliforms ferment lactose and so purple colonies were selected for RapID testing.
In one instance, no color changes were observed on a MAC plate containing a sample that had generated as Colilert positive reading. This sample was tested via RapID anyways and was identified as Enterobacter (a coliform). This strain of Enterobacter possibly had a mutation that allowed the strain to survive by some means other than fermenting lactose.
Of the ten Colilert positives to undergo secondary testing, 5 were positively identified as coliforms (including E. coli, Klebsiella, and Citrobacter). The five that were not positively confirmed as coliforms could indicate a false-positive by the Colilert test, however a false positive is unlikely given that the MAC plates confirm that they were indeed lactose-fermenting bacteria and the RapID database only contains the coliform species considered to have “clinical significance”. Thus, in effect, RapID validates Colilert in at least 5 of 10 instances, and its inability to validate the positive Colilert test in the other 5 cases does not mean RapID disputes those results. Given that the MAC plates consistently supported the Colilert positive results and the RapID “unknown” result do not necessarily dispute those results, those results are assumed to be mostly valid. Therefore, in instances where Coliert was positive but no colony units were visible on the Petrifilm, it is most likely that coliforms were indeed present at rates <1 CFU/mL.
Results
The 3M Petrifilm CC Plate Interpretation Guideline was used to count gas producing colonies likely to be coliforms. Instances where no colonies are observed but secondary testing still indicates the presence of coliforms are also noted. Table 1 shows the results.
Table 1
Distribution of Colony Forming Units Across Samples
Note: The two counts with 101+ colonies were a soda gun and a ketchup and mustard bottle.
Table 2 shows how many samples of each type were taken (based on the pre-established priority list) and how many tested positive for Colilert.
Table 2
Colilert Results by Sampling Location
Thirteen samples were tested and were identified with RapID testing including 4 Klebsiella, 2 Enterobacter, 1 Citrobacter, and another two fluoresced indicating E. coli. The two E. coli results were isolated on an Eosin methylene blue (EMB) plate. EMB plates were used after a positive Colilert test with fluorescence, a characteristic for E. coli that can detect β-glucoronidiase positive bacteria. The two E. coli positive samples came from the underside of a chair and a laminated menu.
One Petrifilm with a colony without a gas bubble was tested and positively identified as Acinetobacter. There were 36 other 3M CC Petrifilm plates with the same characteristics (colonies with no gas production). Acinetobacter is not a coliform and is the probable bacteria in those 36 instances as well. However, anaerogenic (non-gas producing) coliforms, or other anaerogenic bacteria, cannot be ruled out without further testing.
Conclusions
This study verifies the efficacy of Colilert as an effective testing method for establishing the presence of coliforms on non-food contact surfaces within food establishments. Plate counts enumerate the extent of coliform, but current guidelines for evaluating the safety profile of those counts do not exist for common touch environmental samples. Safe levels of coliform contamination on common-touch surfaces, if any, are unknown. The “Guidelines for the microbiological quality of some ready-to-eat foods sampled at the point of sale” suggest that anything under 100 CFU’s per gram of food is acceptable, but translating these guidelines to apply to one milliliter of liquid from a sample taken by swabbing a 187² inch surface area is difficult. Furthermore, the lack of food substrate on common-touch surfaces will naturally suppress bacterial populations while no such limitations exist for food samples. Accordingly, the thresholds established for bacteriological contamination on food may be too high to be comparable even if the challenges in establishing a comparable sampling technique could be conquered.
The large gap between contamination found between high chairs and women’s restroom inside-door knobs implies a link between the Colilert positive detections and cleaning regimens. Many restauranteurs believe consumers view restroom cleanliness as a strong indicator of overall sanitation and so discovering that an item sampled inside a restroom was cleaner than everything sampled within the dining room is not exceptionally surprising.
The results of the study suggest that there are surfaces, particularly high chairs, to which restaurants need to pay more attention in their normal routine cleanings. While the risks associated with various levels of coliform contamination for common touch objects within restaurants may be hard to quantify, the risks for some pathogens are well-known. For instance, Norovirus, also associated with fecal contamination, is notorious for being infectious even at comparatively low levels of surface contamination and the role of surface contamination in facilitating outbreaks has been documented (Wu, et al 2005)—therein lies the value of an indicator organism.
Recommendations
Based on the conclusions of this study, the following recommendations are offered for the consideration of interested parties:
1. Regulators should consider using routine environmental sampling to uncover risk factors for outbreaks or as an investigative tool when other methods fail or when otherwise indicated (such as cleaning verification after an outbreak has occurred).
2. Regulatory agencies should adopt a standardized method for common surface sampling so as to better compare results and risk factor implications
3. Restaurant operators should examine their cleaning practices regarding areas which are frequently touched but may not be frequently cleaned such as high chairs, soda guns, laminated menus and condiment bottles.
4. Restaurant operators should consider food safety as a part of dining room design and material selection as they do kitchens. For instance, many kitchen items are made of metals such as steel or copper. Such metals are naturally biocidal due to a process known as oligodynamic effect in which positive charged metal ions interact with negative ions in living cells leading to cell death.
5. Increased guidance regarding common surface cleaning for retail food establishments should be offered by regulatory authorities.
6. Establishments should focus on cleaning high chairs with relation to high risk population.
Acknowledgements
I would like to thank everyone who had a part in making my research project successful. Especially, I would like to thank John Jai, the microbiologist who mentored me in the lab and Janelle Kaufman, the MCPHD Department of Food and Consumer Safety administrator for supporting me throughout the whole research process. I would also like to give a special thanks to Dr. Paul Dezendorf, research subject matter expert, Charlene Bruce, mentor, and all other IFPTI mentors and staff for the opportunity to participate in Cohort VI. Lastly, I’d like to say thanks to each of the IFPTI Fellows in Cohort VI for a one of a kind experience.
References
3M. (2014). 3M Petrifilm Coliform Count Plate Interpretation Guide.
Barker, J., Vipond, I. B., & Bloomfield, S. F. (2004). Effects of cleaning and disinfection
in reducing the spread of Norovirus contamination via environmental surfaces. Journal of Hospital Infection, 58(1), 42-49.
Centers for Disease Control and Prevention. (2014). Surveillance for foodborne disease outbreaks United States, 2014: Annual report. Retrieved from https://www.cdc.gov/foodsafety/pdfs/foodborne-outbreaks-annual-report-2014-508.pdf
Choi, J., Almanza, B., & Neal, J. (2014). A strategic cleaning assessment program:
Menu cleanliness at restaurants. Journal of Environmental Health, 76(10), 18-24.
Donofrio, R.S., O’Malley, K., Bestervelt, L.L., & Saha, N. (2012). Are we aware of
microbial hotspots in our household? Journal of Environmental Health, 75(2),
12-19.
Gilbert, RJ., et al. (2000). Guidelines for the microbiological quality of some ready-to-eat foods sampled at the point of sale. Communicable Disease and Public Health, 3(3), 163-167.
Lapage, S.P., Estratiou, A., & Hill, L.R. (1973). The ortho-nitrophenol (ONPG) test
and acid from lactose in Gram-negative genera. Journal of Clinical Pathology, 26, 821-825.
Michaels, B. (2002). Handling money and serving ready-to-eat food. Food Service
Technology, 2, 1-3.
Park, G. W., Lee, D., Treffiletti, A., Hrsak, M., Shugart, J., & Vinjé, J. (2015). Evaluation
of a new environmental sampling protocol for detection of human norovirus on inanimate surfaces. Applied and environmental microbiology, 81(17), 5987-5992.
Rodigues, M.J. (2014). Colilert® applied to food analysis [Abstract]. Frontiers in Marine
Science. 1.
Scott, E. (2013). Community-based infections and the potential role of common touch
surfaces as vectors for the transmission of infectious agents in home and community settings. American journal of infection control, 41(11), 1087-1092.
United States Department of Agriculture Economic Research Service. (2016). Food-
Away-from-Home. Retrieved from https://www.ers.usda.gov/topics/food-choices-health/food-consumption-demand/food-away-from-home.aspx
Wu, H. M., Fornek, M., Schwab, K. J., Chapin, A. R., Gibson, K., Schwab, E., &
Henning, K. (2005). A norovirus outbreak at a long-term-care facility: the role of environmental surface contamination. Infection Control & Hospital Epidemiology, 26(10), 802-810.