Observing is better than asking

Ten years ago, as a bunch of University of Guelph students were barfing in their residence bathrooms with noro, Brae Surgeoner, Doug and I hatched plot to observe hand hygiene practices in situ. We wanted test whether students in the midst of an outbreak would report they were really good at washing their hands or using sanitizer. We guessed that what they said, and what we would see, would be drastically different.

It wasn’t our first foray into observational research. A couple years before we did a bunch of secret shopping at Ontario grocery stores and interacted with associates to see what they share about food safety with patrons (us). We heard a whole bunch of nonsense. Ellen Thomas advanced this style of research by training a cadre of secret shoppers throughout the U.S. to order undercooked burgers at restaurants.

As Doug wrote a while back, ‘I view the grocery store and the restaurant as my laboratory. I watch and ask questions of people, especially front-line staff. The head of food safety back at corporate HQ may know the correct food safety answer, but are they providing support to front-line staff, the people customers are most likely to interact with?’

That lab also includes the home (or simulated home) kitchen.

Asking people what they know or do is a start. But it’s never enough. People lie, forget or don’t care. Employing other methods to confirm what they say they do is necessary to confirm actions.

So we’re working with RTI International and USDA FSIS to conduct observation research on consumer food handling behavior. FSIS announced the plans in the Federal Register for comment.

To test new consumer messaging and tailor existing messaging, FSIS can help ensure that it is effectively communicating with the public
and working to improve consumer food safety practices. This behavioral
research will provide insight into the effect FSIS consumer outreach
campaigns have on consumers’ food safety behaviors. The results of this
research will be used to enhance messaging and accompanying materials
to improve their food safety behavior. Additionally, this research will
provide useful information for tracking progress toward the goals
outlined in the FSIS Fiscal Years 2017-2021 Strategic Plan.
To inform the development of food safety communication products and
to evaluate public health education and communication activities, FSIS
is requesting approval for a new information collection to conduct
observational studies using an experimental design. Previous research
suggests that self-reported data (e.g., surveys) on consumers’ food
safety practices are unreliable, thus observational studies are a
preferred approach for collecting information on consumers’ actual food
safety practices. These observational studies will help FSIS assess
adherence to the four recommended food safety behaviors of clean,
separate, cook, and chill, and to determine whether food safety
messaging focused on those behaviors affects consumer food safety
handling behaviors and whether consumers introduce cross-contamination
during food preparation. For this 3-year study, FSIS plans to conduct
an observational study each year and to focus on a different behavior,
food and food preparation task, and food safety communication product
each year. The initial study will examine participants’ use of a food
thermometer to determine if meat and poultry products are cooked to the
proper temperatures. FSIS may decide to continue to conduct these
studies annually, and if so, will request a renewal to extend the
expiration date for the information collection request.

Pork-linked Salmonella outbreak led to 192 confirmed illnesses

It’s MMWR day again. My favorite.

A few years ago we conducted a study in commercial kitchens where we acted as food safety voyeurs and watched 47 food handlers do their job for four days.

We counted and coded all the preparation actions we could see (in some kitchens we had 8 camera angles) and there was whole lot of cross-contamination.

One cross-contamination event per food handler.flyer_raw_pig

Per hour.

The MMWR note from the field detailing a 2015 Salmonella outbreak linked to a Washington State pork processor highlights the impacts of lots of cross-contamination.

A total of 192 confirmed cases were reported from five states; 184 (96%) occurred in Washington (Figure). Patients ranged in age from <1 to 90 years (median = 35 years), and 97 (51%) were female. Among 180 patients for whom information about hospitalization was available, 30 (17%) were hospitalized; no deaths were reported.

On the basis of cases investigated before August 2015, a supplemental questionnaire that went into more detail in addressing meat and livestock exposures was developed. Among 80 patients (42% of all confirmed cases) who were interviewed, 59 (74%) reported eating pork during the 7 days preceding illness. This was significantly higher than the most recently published (2007) Foodborne Diseases Active Surveillance Network (FoodNet) population survey of healthy persons, in which 43% reported eating pork in the week before they were interviewed (p <0.001) (1).

WADOH and PHSKC investigation into the source of pork traced the pork consumed by 35 (59%) of the 59 interviewed patients who reported eating pork back to a U.S. Department of Agriculture’s Food Safety and Inspection Service–inspected pork slaughter establishment in Graham, Washington. During the outbreak period, the establishment distributed whole hogs and pork parts, primarily from five farms in Montana and one in Washington, to Washington, Oregon, and Alaska. Among the 21 interviewed patients who did not report consuming pork before becoming ill, 13 had eaten at one of two restaurants or had shopped at one market where pork from the establishment was served. During June and July 2015, PHSKC inspections of these three facilities identified potential opportunities for cross-contamination of raw pork with other meat and produce, including inadequate employee handwashing and insufficient cleaning and sanitization of food contact surfaces and utensils used for raw meat. Food and environmental sampling by PHSKC at all three facilities yielded the outbreak strains.

There’s a lot of pork cooked and consumed in restaurants across the U.S. daily. And Salmonella in pork is a known issue, but there aren’t reported pork/Salmonella outbreaks every day. My guess is that incoming pork contamination levels were out of the ordinary as well.

Medium and message: Need to frequently change handwashing signs to be effective

I’m a food safety voyeur.

Supermarkets, farmers markets, restaurants – fancy or not – kitchens, farms, I’ve been professionally watching people for 20 years.

surprise-01Chapman likes to recount how he was invited to the GFSI Consumer Goods Forum as a last minute replacement speaker in 2013 to talk about food safety infosheets and how we evaluated them.  He said that the literature shows surprise matters when it comes to communicating risks – and a message that is up all the time, like a hand washing sign, probably doesn’t do much after the day it was posted (when it is surprising to the food handler).

The level of surprise in a message determines how successfully the information is received. In 1948, the Bell Telephone Company commissioned a study on communication as a mathematical theory to aid in the design of telephones.  In a study of brain function, Zaghloul and colleagues (2009) also showed the brain’s sensitivity to unexpected or surprising information plays a fundamental role in the learning and adoption of new behaviors.

During the Q&A session at the end of the session someone from a German retail store asked Chapman if he was suggesting that that they take down all the handwashing posters they had up, and Chapman said, yes, unless they plan on changing them every couple of days. The audience had an audible gasp.

We’ve found that posting graphical, concise food safety stories in the back kitchens of restaurants can help reduce dangerous food safety practices and create a workplace culture that values safe food.

It was the first time that a communication intervention such as food safety information sheets had been validated to work using direct video observation in eight commercial restaurant kitchens and was published in the  Journal of Food Protection.

hand_sanitizer_hospital_11We found that infosheets decreased cross-contamination events by 20 per cent, and increased handwashing attempts by 7 per cent.

Based on observations of more than 5,000 patrons at a hospital-based cafeteria, we showed that an evidence-based informational poster can increase attempts at hand hygiene.

So we gladly welcome new work on food safety messages and media in poultry processing facilities.

Signs can provide repetitive training on specific food safety practices for multicultural food processing employees. Posted signs for workers in many food processing facilities tend to be text-heavy and focus specifically on occupational hazard safety. The purpose of this study was to evaluate the effectiveness of newly-developed hand washing pictograms on employees’ hand washing behavior using video observation.

Five employee hand washing behaviors (soap use, wash completeness, wash time, complete rinsing, and towel use) were evaluated with (a) no intervention, company signs posted and considered the baseline; and compared to (b) hand washing behavior the next day (short term) and two weeks (long term) after experimental hand washing signs were displayed at a raw poultry slaughter facility (Facility A) and a poultry further processing facility (Facility B).

sponge.bob.handwashingBoth facilities showed a significant increase (p < 0.05) in soap use after the new sign was introduced at both short and long term time periods. There was a significant increase (p < 0.05) in washing, time of washing, and rinsing observed by Facility B employees, when baseline data was compared to the short term. This indicates that a new sign could increase hand washing compliance at least in the short term. Sign color also had a significant effect (p < 0.05) on employee behavior for washing and time of washing. Behavior for four of the five variables (soap, wash, time of wash, and towel use) was significantly different (p < 0.05) between baseline and either experimental observation period.

While signs can be a useful tool to offer as recurring food safety training for food processing employees, employees tend to revert back to old habits after several weeks.

Evaluation of how different signs affect poultry processing employees’ hand washing practices

Food Control, Volume 68, October 2016, Pages 1–6

Matthew Schroeder, Lily Yang, Joseph Eifert, Renee Boyer, Melissa Chase, Sergio Nieto-Montenegro



Watch the chickens; predict the campy

When I had campylobacteriosis I didn’t want to move much for fear of unleashing what was in my bowels. According to Yahoo News, chickens infected with campy also move less.

Using cameras to track how the birds move around can predict which flocks are at risk of being infected, according to research by Oxford University.images-1

Lead author of the study Dr Frances Colles, from Oxford University’s Department of Zoology, said: “Humans consume nearly 60 billion chickens a year, more than any other animal.

“At the same time, there is a worldwide epidemic of human gastroenteric disease caused by campylobacter.

“It is estimated that up to four fifths of this disease originates from contaminated chicken meat.”

The findings, published in the journal Proceedings of the Royal Society B, showed campylobacter-positive birds had less movement and different behaviour to those without the bacteria.

Professor Marian Dawkins, Professor of Animal Behaviour at Oxford University, said: “The findings are compatible with the growing evidence that campylobacter may be detrimental to chickens’ health, rather than simply being harmless gut bacteria.

“Use of this optical flow information has the potential to make a major impact on the management of commercial chicken flocks, for the benefit of producers, consumers and the birds themselves.”

Researchers collected data for 31 commercial broiler flocks and tested for the presence of campylobacter at different ages.

Surveys suck: Consumers don’t accurately report what they do in kitchen

Research utilizing both survey and observational techniques has found that consumers do not accurately report their own food handling behaviors. The goal of this study was to objectively observe conditions related to food safety risks and sanitation in domestic kitchens in an urban environment.

survey-saysSubjects (n = 100) were recruited from Philadelphia, PA. Homes were visited over a one-year period by two trained researchers using a previously developed audit tool to document conditions related to sanitation, refrigeration, and food storage.

Potential food safety risks identified included evidence of pest infestation (65%), perishable food stored at room temperature (16%), storage of raw meat above ready-to-eat foods (97% of homes where raw meat was present), and a lack of hot running water in the kitchen (3%). Compliance with correct refrigeration practices was also low, with 43% of refrigerator temperatures ≥ 41°F, and only 4% of refrigerators containing a thermometer. Consumers of minority race/ethnicity were more likely to have evidence of pest infestation in the home, lack a dishwasher and lack a cutting board in the kitchen, while Caucasian consumers were more likely to have an animal present in the kitchen during the audit visit.

 Visual audit of food safety hazards present in homes in an urban environment

Food Protection Trends, vol. 35, no. 4, pp. 290-301, July 2015

Patricia A. Borrusso, Shauna Henley, Jennifer J. Quinlan



Toilet psychology: why do men wash their hands less than women?

By hiding in toilet cubicles for a new study, psychologists observed how long people spend using the loo, and how long they wash their hands for afterwards. That men usually wash their hands less conscientiously than women is a well-established finding. Thomas Berry and his colleagues wanted to find out more about the reasons for this gender difference.

toilet.hideFor one day, between 10am and 4pm, a male researcher secreted himself inside one of three cubicles in a gents toilet facility at a U.S. University. For optimal observational purposes he chose the cubicle adjacent to a row of three urinals. Nearby, in a similarly designed female toilet facility, a single female researcher positioned herself in one of the three cubicles available. Don’t worry, both researchers were provided with a “customised wooden bench” for comfort.

They were also equipped with stopwatches. The researchers used an “unobtrusive sight procedure” – that is, they spied on other visitors to the lavatories using the gaps beneath and by the side of the cubicle doors (for some reason, US toilet cubicles always have a gap of about a centimetre either side of the door). The researchers also used an “acoustic procedure”. That is, they listened to the visitors’ actions. The study authors explained:

“… research assistants recorded the facility [urinal or cubicle], and then started a stopwatch when the patron’s feet stood relatively still. For the men, the research assistants also recorded the orientation of the feet to gauge the patron’s use of the commode (i.e. as a commode or a urinal). When research assistants heard the flushing of the patron’s commode or urinal the stopwatch was turned off … and the duration of the restroom event was recorded.”

Similar procedures were followed for recording each visitor’s “hand washing event” if there was one. A clever twist was that for part of the study, the researchers put “out-of-order” signs over the men’s urinals. This was to see how much they’d hand wash if they were forced to urinate in a cubicle, rather than at a urinal.

The psychologists managed to observe the toilet behaviour of 34 women using cubicles; 32 men who used a cubicle to defecate; 40 men who had no choice but to use the cubicles for urinating (because of the out-of-order signs); and 64 men who used a urinal. The bare statistics show that the hand-washing rates for these four groups were 91 per cent, 87.5 per cent, 75 per cent and 59.4 per cent, respectively.

The difference in hand washing rates between women using a cubicle and men using a cubicle (for defecating) was not statistically significant. In contrast, both women using a cubicle, and men using a cubicle (for defecating), showed significantly higher hand-washing rates than men who used a urinal.

The data are somewhat compromised because, as the researchers delicately put it – the women’s “facility use is a constant (i.e., commode) and their behaviour (urination, defecation, or menstrual care) is confounded within the one environment.” However, taken together, the results suggest that the reason men wash their hands less than women overall, is not because of gender norms (i.e. men are less bothered about being clean), but because of the differences in the toilet environment and toilet behaviour for men and women. In fact, after using a toilet cubicle to defecate, men tended to wash their hands for longer than women (but remember we don’t know what the women had been doing).

Surveys still suck, here’s an alternative: video observation and data coding methods to assess food handling practices at food service

Ben Chapman, who was a Phd student with me at Guelph and is now plying his trade at North Carolina State University, Tanya MacLaurin, who used to be at Kansas State and is now at Guelph, and me, who used to be at Guelph and now is at Kansas State, got together to create a how-to paper for video observation to measure food safety behaviors. Abstract below.

Eating at foodservice has been identified as a risk factor for foodborne illness. The World Health Organization (WHO) has identified four food handler-related factors that contribute to foodborne illness: improper cooking procedures; temperature ben.video.observation.13abuse during storage; lack of hygiene and sanitation by food handlers; cross-contamination between raw and fresh ready to eat foods.

Evaluation of food handler behaviors, important for risk assessment calculations and for the effectiveness of training strategies, has historically been limited to self-reported data, inspection and participatory observation. This article describes the framework of a video observation methodology, novel to food service situations used capture and code food handler practices for analysis.

Through the piloting of this technique in a working foodservice establishment, a number of lessons were learned, including best equipment to use, equipment location and configuration, as well as pitfalls in coding practices. Finding and working with partner organizations and navigating institutional ethics review is also discussed.

Chapman, B., MacLaurin, T. and Powell, D.  2013. Video observation and data coding methods to assess food handling practices at food service. Food Protection Trends. 33 (3). 146–156.


Campylobacter in the kitchen: observational trial of safe food handling behavior during food preparation

Austrian researchers report on an observational trial of safe food handling behavior during food preparation using the example of Campylobacter spp.

chicken.thermJournal of Food Protection®, Number 3, March 2013, pp. 376-551 , pp. 482-489(8)

Hoelzl, C.; Mayerhofer, U.; Steininger, M.; Brüller, W.; Hofstädter, D.; Aldrian, U.


Campylobacter infections are one of the most prominent worldwide food-related diseases. The primary cause of these infections is reported to be improper food handling, in particular cross-contamination during domestic preparation of raw chicken products. In the present study, food handling behaviors in Austria were surveyed and monitored, with special emphasis on Campylobacter cross-contamination. Forty participants (25 mothers or fathers with at least one child ≤10 years of age and 15 elderly persons ≥60 years of age) were observed during the preparation of a chicken salad (chicken slices plus lettuce, tomato, and cucumber) using a direct structured observational scoring system. The raw chicken carcasses and the vegetable part of the salad were analyzed for Campylobacter. A questionnaire concerning knowledge, attitudes, and interests related to food safety issues was filled out by the participants. Only 57% of formerly identified important hygiene measures were used by the participants. Deficits were found in effective hand washing after contact with raw chicken meat, but proper changing and cleaning of the cutting board was noted. Campylobacter was present in 80% of raw chicken carcasses, albeit the contamination rate was generally lower than the limit of quantification (10 CFU/g). In the vegetable part of the prepared product, no Campylobacter was found. This finding could be due to the rather low Campylobacter icarly.chicken.cell.handscontamination rate in the raw materials and the participants’ use of some important food handling behaviors to prevent cross-contamination. However, if the initial contamination had been higher, the monitored deficits in safe food handling could lead to quantifiable risks, as indicated in other published studies. The results of the observational trial and the questionnaire indicated knowledge gaps in the food safety sector, suggesting that further education of the population is needed to prevent the onset of foodborne diseases.

More on pink and thermometers

Surveys still suck.

Using I-own-a-thermometer as an indicator of thermometer use is as useful as I-own-a-sink therefore I wash my hands. Or, I own a toilet, so I always hit the bowl. Or … use your imagination.

Researchers at the U.S. Food and Drug Administration report in the Journal of Food Protection that the use of a food thermometer is the best way to ensure that meat, poultry, and other foods reach an internal temperature sufficient to destroy foodborne pathogens.

The 1998, 2001, 2006, and 2010 Food Safety Surveys were used to analyze changes in food thermometer ownership and usage for roasts, chicken parts, and hamburgers in the United States.

But surveys still suck.

The paper notes that when E. coli O157:H7 was first associated with ground beef in the 1980s, the U.S. Department of Agriculture (USDA) recommended that consumers cook hamburgers until the meat was ‘‘brown or pinkish brown in the center. However, as a result of research that showed that one out of four hamburgers may be brown in the center before reaching a safe internal temperature, the USDA changed its advice to consumers— instead of using color as an indicator of doneness in hamburgers, consumers should use a food thermometer to ensure that a safe temperature has been reached. In May 2000, the USDA launched the Thermy educational campaign to encourage consumers to use a food thermometer when cooking small cuts of meat, such as hamburgers and chicken parts. The USDA also provided guidance to consumers about the safe temperature for various cuts of meat and poultry.

Ho Phang and Christine Bruhn reported earlier in JFP that in video observation of 199 California consumers making hamburgers and salad in their own kitchens, handwashing was poor, only 4% used a thermometer to check if the burger was safely cooked, and there were an average of 43 cross-contamination events per household. They concluded Thermy had not been successful.

We did our own survey with 40 people brought in to cook a chicken meal in a Kansas State kitchen and videotaped their behaviors. Many participants reported owning a food thermometer (73%) and nearly half (42.5%) of participants reported knowing the suggested end temperature for cooking poultry to ensure doneness. When asked the final recommended internal temperature for chicken, the mean response was 214°F with a range of responses from 140°F to 450°F. (The correct answer is 165F)

Of those participants observed measuring the internal temperature of the product, only three used the thermometer correctly. During observation, two individuals who used the thermometers failed to remove protective casings prior to taking internal temperature readings, and therefore used the instruments incorrectly.

Surveys do not measure behaviors: they give an indication of what people think their behavior is, or what the survey person wants to hear, but that isn’t going to get people to use a thermometer (tip-sensitive, digital).

We don’t need no education: burger preparation, what consumers say and do in the home

I cringe when someone says, ‘food safety is simple.’

A review of existing studies by the U.K. Food Standards Agency found that, although people “are often aware of good food hygiene practices, many people are failing to chill foods properly, aren’t following advice on food labels and aren’t sticking to simple hygiene practices that would help them avoid spreading harmful bacteria around their kitchens.”

Yes, individuals are impervious to risk; been known for decades.

And there’s that word, ‘simple’ again.

I especially cringe when someone says, ‘cooking a hamburger is easy with these simple food safety steps.’

Ho Phang and Christine Bruhn report in the current Journal of Food Protection that in video observation of 199 California consumers making hamburgers and salad in their own kitchens, handwashing was poor, only 4% used a thermometer to check if the burger was safely cooked, and there were an average of 43 cross-contamination events per household.

There’s some good data in the paper about what consumers do in their own kitchens, and the results are an additional nail in the self-reported-food-safety-survey coffin: people know what they are supposed to do but don’t do it.

But what the paper doesn’t address is how to influence food safety behaviors. Instead, the University of California at Davis authors fall back on the people-need-to-be-educated model, without out providing data on how that education – I prefer compelling information – should be provided.

The authors state:

• educational materials need to emphasize the important role of the consumer in
preventing foodborne illness and that foodborne illnesses can result from foods prepared in the home.;

• the gap between the awareness of the importance of hand washing and the actual practice of adequate hand washing should be addressed by food safety educators.

• food safety educators should address the lack of reliability of visual cues during cooking (stick it in — dp);

• food safety educators should emphasize faucet cleaning with soap and water as a way of preventing cross-contamination; and,

• ignorance about food irradiation point to a further need for education.

The authors do correctly note that program to promote the use of thermometers when cooking burgers, initiated by the introduction of Thermy in 2000, has not been successful. So why do more education?

And the E. coli O157:H7 outbreak in Jack-in-the-Box hamburgers happened in Jan. 1993, not 1994 as stated in the paper; someone should have caught that.

Burger preparation: what consumers say and do in the home
Journal of Food Protection®, Volume 74, Number 10, October 2011 , pp. 1708-1716(9)
Phang, Ho S.; Bruhn, Christine M.
Ground beef has been linked to outbreaks of pathogenic bacteria like Escherichia coli O157:H7 and Salmonella. Consumers may be exposed to foodborne illness through unsafe preparation of ground beef. Video footage of 199 volunteers in Northern California preparing hamburgers and salad was analyzed for compliance with U.S. Department of Agriculture recommendations and for violations of the U.S. Food and Drug Administration’s Food Code 2009. A questionnaire about consumer attitudes and knowledge about food safety was administered after each filming session. The majority of volunteers, 78%, cooked their ground beef patties to the Food Code 2009 recommended internal temperature of 155°F (ca. 68°C) or above, and 70% cooked to the U.S. Department of Agriculture consumer end-point guideline of 160°F (ca. 71°C), with 22% declaring the burger done when the temperature was below 155°F. Volunteers checked burger doneness with a meat thermometer in 4% of households. Only 13% knew the recommended internal temperature for ground beef. The average hand washing time observed was 8 s; only 7% of the hand washing events met the recommended guideline of 20 s. Potential cross-contamination was common, with an average of 43 events noted per household. Hands were the most commonly observed vehicle of potential cross-contamination. Analysis of food handling behaviors indicates that consumers with and without food safety training exposed themselves to potential foodborne illness even while under video observation. Behaviors that should be targeted by food safety educators are identified.