We recruited a quota-based sample of 1,030 members of the UK public via an online market research panel (http://www.researchnow.com). Quotas were used to ensure representativeness in terms of age groups and social class. The quota permitted an unequal split by sex (up to 70% women) because in the United Kingdom, food preparation at home is more commonly performed by women than men. We also recruited 143 chefs through face-to-face convenience sampling at culinary shows and competitions and by online culinary forums.

All participants gave informed consent. Respondents were debriefed on the purpose of the survey after completion and given the opportunity to withdraw their data. Ethical approval was obtained from the College of Natural Science Ethics Committee at Bangor University (CNS/2014/AJ1).

To prepare cooked chicken liver dishes to serve as visual aids, we used methods similar to those used in studies of hamburgers (16) and beefsteaks (17). A chef cooked 7 batches of chicken livers for various times, recorded the maximum core temperature for each batch, and arranged each batch on a plate for photography by a professional photographer. The process was repeated (without the temperature being recorded) for 3 other meats (duck breasts, lamb racks, and beef burgers).

To determine preferences and knowledge of safe cooking practices among chefs and members of the public, we used the images of cooked chicken livers as visual aids. The images were presented in surveys (online and print), arranged in order of cooking time/rareness (Figure 1). The surveys for chefs and the public were similar, except that the chefs were asked about serving preferences and the public was asked about eating preferences.

To avoid biases (such as social desirability bias) resulting from respondents perceiving the survey to be about food safety, we described the survey as being about food preferences. Respondents were first asked preference questions about 3 of the 4 meats (in random order) to obscure the focus on chicken livers and safety. Chefs were asked to indicate which chicken liver dish was cooked “the way you would like to serve it” and “the way you think most customers would like it.” Members of the public were asked which dish they would prefer if “eating out” and “eating at home.”

Respondents were subsequently asked which chicken liver dish (if any) was the first they thought would meet FSA safe cooking guidelines. Additional questions were asked about perceived trends and influences regarding cooking meat, dining habits, and demographic information such as class and age. Chefs provided additional information about their current position, such as their training and industry experience.

To prepare a suspension of Campylobacter for experimental inoculation, we streaked Camplyobacter jejuni M1 strain (sequence type 137, clonal complex 45) on Columbia agar base containing 5% defibrinated horse blood, incubated it at 37°C under microaerobic conditions for 48–72 h, and then inoculated it into Camplyobacter enrichment broth. After subculture for another 24 h, a bacterial suspension was prepared in maximum recovery diluent to an optical density of 600 nm (?109 CFU/mL). The culture broth was diluted in Camplyobacter enrichment broth to give a suspension of ?105 CFU/mL for inoculation into fresh chicken livers.

The fresh chicken livers were purchased in packs from supermarkets and sorted into batches of 4 with similar weights. The connective tissue was cut between the 2 liver lobes, with the weight of the larger lobe recorded and assigned for inoculation with Campylobacter broth suspension; 4 livers were assigned to each cooking batch. A 1-cm2 area of each liver was scored at its thickest point by using a sterile scalpel blade and injected with 100 ?L (?104CFU) of culture broth, corresponding to the highest levels of Campylobacter reported to be found in naturally contaminated livers (18).

For each cooking time, 10 g butter was heated in a frying pan over moderate to high heat on an electric cooktop; when the butter had finished frothing, the 4 inoculated liver lobes in the batch were added. The maximum core temperature of the largest and smallest liver in each batch was recorded. To determine the survival of the inoculated M1 strain of C. jejuni within the cooked livers, we placed each liver in a sterile petri dish and a 4–5-g portion around the scored inoculated region was removed and added to a Stomacher bag (Seward BA6040, Worthing, UK); 10 mL of Exeter broth was added to each bag before Stomaching (mechanical pounding of the outer surface of the bag to remove bacteria) for 1 min. The homogenized suspension was poured into a 20-mL universal container and incubated at 41°C under microaerobic conditions (Variable Atmosphere Incubator; Don Whitely Scientific, Shipley, UK) for 24 h, after which 1 loopful of broth was plated onto Campylobacter blood-free medium (modified charcoal cefoperazone deoxycholate agar, containing cefoperazone and amphotericin) at 41°C under microaerobic conditions for 48–72 h. We picked 1 typical Campylobacter colony from at least 1 plate in each batch and confirmed it as C. jejuni by PCR; for a cooked liver to be deemed positive, 1 isolate per batch was confirmed as C. jejuni positive (19).

We modeled the probability of survival for the 60 livers for which temperature and Campylobacterpresence/absence after cooking were recorded. We used logistic regression to model the relationship between the core temperature of the livers and the survival of Campylobacter. The probability of Campylobacter survival as a function of core temperature was modeled via estimation of a logit model, which captured the nonlinear temperature-survival relationship (Figure 2). Parameter estimates were obtained by using logistic regression (Stata logit command; StataCorp LP, College Station, TX, USA) on the binary variable indicating Camplyobacter survival (1 = survival, 0 = nonsurvival) in a sample of 60 cooked chicken livers. Temperature was the maximum core temperature recorded for the batch from which the chicken liver was taken. This model was used to assign predicted survival rates for each photographed chicken liver dish.

We used the Kolmogorov Smirnov 2-sample test to compare differences in the distribution of knowledge and preferences between groups (chefs and the public). We investigated within-person differences by using the Wilcoxon signed-rank test for paired data. Ordered logit models (20) were estimated to determine the effects of observable characteristics on respondents’ preferences for chicken liver rareness and their choices of FSA-compliant livers.

We discuss the results of the Campylobacter survival experiment first because an understanding of those results is useful for interpreting the preferences and knowledge analyses. The relationship between core temperature and Campylobacter survival rate was inverse (Table; Figure 2). Of the 32 batches of 4 inoculated livers, the shortest cooking time was 1 minute, leading to a mean core temperature of 36°C and a 100% Campylobacter survival rate. At the maximum mean core temperature (72°C), Campylobacter survival rate was 8.3%.

The logistic model predicted a survival rate of 98% in liver with core temperature that reached 52°C (liver 1) and equivalent survival rates of 95% and 48% at core temperatures of 56°C and 66°C (livers 2 and 3). Liver 4 reached a maximum temperature of 70°C, but the temperature was not held for the recommended 2 minutes; predicted Campylobacter survival rate was 22%. Livers 6 and 7 met the FSA guidelines, and their predicted Campylobactersurvival rate was <0.001%.

Of the 1,030 members of the public surveyed, 43.0% ate chicken livers and hence were asked to select the chicken liver dishes they preferred and which they thought met FSA guidelines. Half (49.3%) of all male respondents and 38.4% of all female respondents ate chicken livers. Rates of chicken liver consumption varied by age group: 18–34 years, 34.7%; 35–44 years, 44.7%; 45–54 years, 49.0%, 55–64 years: 51.5%; and >65: 42.9%. Chicken livers were eaten by half (51.0%) of respondents belonging to UK socioeconomic grouping ABC1 (upper, middle, and lower middle class) and 32.3% of those belonging to C2DE (working class and those at the lowest level of subsistence).

Members of the public poorly identified whether a chicken liver met FSA guidelines for safe cooking (Figure 3). Thirty percent identified livers 1–3 as being safe to eat; the predicted rates of Campylobacter survival in these livers were 48%–98%. Another 22% thought that liver 4 (Campylobacter survival rate 22%) was safe to eat.

No significant difference was found between the public’s choices of FSA-compliant livers and their preferences when dining out (p = 0.776, Wilcoxon signed-rank test; n = 386) (Figure 4); respondents were consistent between what they wanted to eat and what they thought was safe. Respondents showed a significant preference for pinker livers when eating out rather than at home (p = 0.007, Wilcoxon signed-rank test; n = 446). Paradoxically, respondents reported being more concerned about food safety when eating out than at home (p<0.001, Wilcoxon signed-rank test; n = 999).

Ordered logit results (not reported) identified no systematic differences in rareness preferences by respondent sex, age, or class. Livers that were more pink were preferred by respondents who described themselves as adventurous (p<0.030, n = 444) and who were less concerned about restaurant food safety (p<0.001, n = 444).

Among the 143 chefs, of those who indicated their sex, 134 (88%) were male. Among the 141 who indicated their type of work, 31.9% worked in fine dining, 17% in contract catering, 11.3% in casual restaurants, 5.7% in pubs, and 19.1% in multiple kitchen types. The most commonly held position among 131 chefs who responded was head chef (54.0%), followed by chef trainer (11.5%), chef de partie (10.7%), commis chef (6.9%), and sous chef (6.1%).

Chefs were much better than members of the public at

download now