Risk perception by food handlers in the tourism sector

The objective of this study was to evaluate the perception of risk by food handlers in the food services of the tourism sector. One structured methodological instrument was used to analyze the risk perception of 108 food handlers from 19 establishments in a tourist region in the State of São Paulo, Brazil. The highest score was related to the subject “Integrated pest control” and the lowest level was related to “Eating soft-cooked eggs”. Differences (p < 0.05) were observed among the levels of risk perception, making it possible to form three sub-groups concerning the questions. The central topic of the first sub-group was safety aspects in food production (I), the second related to sanitation operations (II) and the third to integrated pest control (III). Sub-groups II and III presented the greatest level of perceived risk. Differences (p < 0.05) between the level of risk perception and the socio-demographic variables were identified. Women showed greater risk perception compared to men. The results can provide important information for public and private programs, improving development of institutional strategies directed at food safety.


INTRODUCTION
The number of meals eaten away from home is expanding in Brazil 1 , increasing from 24.0% to 31.0% in the period between 2002 and 2010 2, 3 . The tendency observed in Brazil is still below the European and North American indexes, where this sector is responsible for 50.0% to 60.0% of food consumption 1  The majority of the food services in the country are micro and small sized, including those involved with tourism 6 .
These companies have difficulties in developing food security directed programs due to deficiencies in financial, human and infrastructure resources 7,8 and such conditions give rise to problems related to inadequate sanitation practices. Studies show that food poisoning outbreaks are associated with unsatisfactory knowledge of food handlers concerning adequate hygiene practices 7,9,10,11,12 in the food services of hotels 8,13 and in the tourism sector in general 14,15 .
The World Health Organization (WHO) estimates that foodborne diseases (FBD) constitute one of the most widespread sanitation problems in the world 16 . In the USA and Europe the majority of food poisoning outbreaks occur in food services 17,18 . In Brazil the majority occurred in residences, followed by restaurants, involving predominantly mixed foods and egg-based formulations 19 .
Codex Alimentarius defines risk as the probability of danger occurring and its severity 20 . Thus, the evaluation of risk perception by food handlers in the food service is important in order to elucidate behavior that could favor food contamination.
Perception can be defined as the result of the interaction of two types of data: the physical stimuli of the external environment and the internal stimuli, which constitute predispositions based on previous experience. The perception of risk suggests that individuals are subject to cognitive biases such that the risks perceived are frequently incompatible with the information concerning the objective risk 21 .
The majority of the personality behaviors and traces of an individual are culturally evaluated, and hence some types of behavioral conduct can be considered more desirable than others 22 . Methods have been developed to quantify the degree of a socially desirable response, which allows one to identify and minimize measurement biases in self-reported surveys 23,24 , such as the Marlowe-Crowne scale 25 .
Studies related to risk perception by food handlers in food services in the tourism sector represent a gap in the literature.
Considering the above, the objective of the present study was to evaluate the risk perception by agents involved in food safety in the tourism sector.

METHOD
The research was carried out in meal producing units (

Survey instrument
A questionnaire was elaborated as methodological instrument based on the one used by Frewer et al. 27 and Cunha et al. 28 , who considered the five keys of the WHO 20 and National normative acts 29,30 (federal and state). They also considered the main non-conformities found in food services as registered by the Food and Drug Administration 31 report and by surveys carried out by other authors working in this sector 10,13,32,33,34,35,36,37,38 .
The questionnaire contemplated aspects related to safety during production and during the sanitation and storage operations of the foods. Each question (Q) started with the phrase "What is the risk of a client contracting a foodborne disease" followed by the subsequent: Q1 -if the food handler stores raw or cooked meats at room temperature; Q2 -if the handler takes advantage of leftovers from one meal in a later meal; Q3 -if one consumes non-sanitized raw vegetables; Q4 -if one consumes incompletely cooked meat; Q5 -if one consumes soft-cooked egg-white or egg-yolk; Q6 -when a sick worker handles the food; Q7 -when a worker responsible for preparing meals does not sanitize his/her hands; Q8 -when raw foods are in contact with prepared foods; Q9 -when there is no quality control of the water used in the preparations; Q10 -when there is no sanitization of the environment, equipment and utensils; Q11 -when rodents, cockroaches and flies, amongst others, are found in the installations.
The responses were given using a structured scale with values from 1 = "no risk" to 7 = "very high risk". The handlers placed an "X" on the scale to indicate the degree of risk considered for each situation.
Another questionnaire, together with a version of the Marlowe-Crowne scale 25 39 . For each question, the agents were requested to mark either "true" or "false", according to their normal behavior.
The questionnaire also included items for the socio-demographic characterization of the population studied.

Definition of the sample
The definition of the survey universe was based on the number of accommodation companies in the region studied, extracted from documents made available by the Municipal Secretariats of Tourism of the municipalities involved. The companies were classified according to the typologies of the hotels and inns.
The size was defined by the number of employees, according to the Serviço de Apoio às Micro e Pequenas Empresas 41 classification for services companies, as follows: up to 9 employees is a micro-company; from 10 to 49 is a small company; from 50 to 99 is a medium company; and with more than 100 employees is a large company.
Municipality A was composed of 11 hotels and 6 inns; municipality B of 6 hotels and 14 inns; all the companies were invited to take part in the survey. Of the hotels and inns in Municipality A, 72.3% and 16.6%, respectively, opted to take part; in Municipality B, 85.7% and 35.7%, respectively, for the same typologies, opted to take part. This gave a grand total of 108 food handlers in 19 meal producing units. The questions were responded by the handler in the presence of a researcher, so that the latter could clear up any doubts.

Statistical analysis of the data
The perception data did not follow a normal distribution (Kolmogorov-Smirnov test). The Friedman test, followed by the Nemenyi paired multiple comparison test were applied to examine the differences between the risk perception questions.
The Mann-Whitney U test (for 2 groups of independent variables) and the Kruskal-Wallis test (for more than 3 groups of independent variables) followed by Dunn's multiple paired comparison test were used to check for differences between the risk level perceived in relation to the socio-demographic variables In order to identify the tendencies of the individuals for social desirability as a function of the socio-demographic variables, the following parametric tests were carried out (based on the fact that the desirability data followed a normal distribution according to the Kolmogorov-Smirnov test): the student t test for independent samples for the variables of gender, education in the food area and training in hygiene; and ANOVA, followed by the Tukey multiple comparisons test for the other socio-demographic variables. The non-parametric Spearman correlation coefficient test was also used. A 5% significance level was used in all the tests.

RESULTS ANS DISCUSSION
The questionnaires used in the present study were considered satisfactory according to that explained by Hair et al. 42 in relation to the reliability of the groups of questions, using Cronbach's Alpha (α) coefficient for the items of risk perception evaluated (Cronbach's α = 0.774) and social desirability (Cronbach's α = 0.644), with indices above 0.6 for these variables.

Population characterization and social desirability
With respect to the socio-demographic characterization, in micro-companies. A total of 56.5% were from Municipality A and 43.5% from Municipality B (Table 1).
Studies concerning food services in Brazil have shown a low educational level amongst food handlers 9 , as in other developing 32,44 and developed 45 countries.
The tendencies for social desirability, the means obtained were within normality 25 . The values found for the socio-demographic variables of the handlers were not statistically different (p < 0.05).

Risk perception
Social desirability did not influence the responses of the interviewees with respect to risk perception, since no correlation (p < 0.05) was found between the two variables.
A mean score of 6 to 7, considered as high or very high risk according to the scale used, was attributed by most of the respondents (Figure). a high risk in spreading foodborne diseases to the clients, and more than half considered the same for the questions concerning "making use of leftovers" (Q2), "cooking temperature" (Q4) and "cross contamination" (Q8). For the subjects "storage at room temperature" (Q1) and "eating non-sanitized vegetables", the percentages were relatively higher (about 65.0%). showed that the highest percentages of perceived risk were related to "consumption of soft-cooked eggs" (64.8%), "making use of leftovers" (68.2%), "cooking temperature" (81.8%) and "cross contamination" (75.0%), values superior to those found in the present study. In this same study by Cunha et al. 28 , the lowest percentages were related to "eating non-sanitized raw vegetables" (34.1%) and "serving food at room temperature" In the present study the questions presenting the greatest percentages of risk perceived as between high and very high were the subjects "food handling by sick worker" (Q6), "hand sanitization"  (Q7), "water potability" (Q9), "sanitization of the environment, equipment and utensils" (Q10) and "integrated pest control" (Q11).
Studies evaluating food handlers with respect to their knowledge and practice of adequate hygiene in food services showed that, in general, the percentages in relation to aspects of hygiene were greater than those related to food preparation 9,10,35,45,52 .
However, the authors highlighted the fact that in relation to hand sanitization, the handlers knew that the lack of this procedure could lead to risks, but part of the handlers had no knowledge of the need for a disinfection step 53,54 or of the correct technique and frequency to carry it out 9 , even though this was a subject always highlighted in training for this public concerning food safety 50  In the present study, the fact that the individuals considered questions related to meal preparations to have a lower risk level, could presuppose a conviction that FBD rarely happen in their work environment. This situation is indicated in the literature as an "optimistic bias" and a cognitive error 27,58 . Weinstein 58 , and Lion et al. 59 believe that the discrepancy between risk perception and the behavior of an individual or of the collectivity could also be related to a sentiment of control over the reality perceived, due to a limited knowledge concerning the risks, as proven by Meysenburg et al. 57 .
From the statistical analysis, it is possible to notice a grouping of subjects of the questions and a random setting of 3 subgroups.
The   Significant differences were observed between Q11 and the other questions. The subject "hand sanitization" (Q7) of subgroup II presented differences (p < 0.05) with respect to the subjects "making use of leftovers" (Q2), "cooking of meat and eggs" (Q4 and Q5) and "cross contamination" (Q8) of subgroup I. The subjects concerning "food handling by sick worker" (Q6) and "sanitization of the environment, equipment and utensils" (Q10) of subgroup II presented differences when compared to Q2 and Q5 of subgroup I.

Differences in the levels of risk perceived by the respondents
Corroborating with the information found in the present study, Osaili 36 evaluated food handlers with respect to knowledge concerning hygiene, and found that about 90.0% of the sample presented better results in relation to the subject "hand sanitization" when compared to "cross contamination" (72.9%), "making use of leftovers" and "cooking temperatures for foods" (52.6%). They also observed greater knowledge concerning health problems that affected food safety (60.4%) when compared to those related to the use of leftovers and the cooking temperatures of foods. It is important to take in consideration that the mentioned less known procedures have a fundamental importance to the food safety, once they are usually related to food poisoning outbreaks.
Meysenburg et al. 57 showed that hand washing was reported as the most common practice to prevent foodborne diseases (96.0%) when compared to other aspects of food hygiene. McIntyre et al. 45 showed greater knowledge by trained and non-trained food handlers of both sexes in the evaluation of questions involving sanitization of the hands and utensils when compared to those related to use of leftovers, adequate cooling and heating of foods, control of cooking time and temperature and distribution.
The subgroups were also analyzed in an independent way, observing differences (p < 0.05) between Q5 and Q1 (p = 0.039) of subgroup I and between Q3 and Q7 (p = 0.002) of subgroup II. Thus it can be seen that the perception of risk concerning "eating soft cooked eggs" was less than that for "storing foods at room temperature". Similarly, the perception of risk for "eating non-sanitized salads" was less than that for "lack of hand sanitization".
When the socio-demographic variables of the respondents were correlated with the level of risk perception, it was observed that individuals educated in the food area presented higher levels for the questions concerning "storing meat at room temperature" (Q1) and "eating soft cooked eggs" (Q5). Women presented greater risk perception than men for the questions "making use of leftovers" (Q2) and "eating raw salads without sanitization" (Q3). Women and individuals educated in the food area perceived the risks related to water potability (Q9) more than men, corroborating the findings of Lazou 10 and Gomes-Neves 33 .
Contracted employees educated in the food area presented a Table 3. Comparison between the groups of socio-demographic variables of the agents involved, as a function of the levels of risk perception for questions 1, 2, 3 and 5.  (Tables 3 and 4) with respect to a lack of environmental, equipment and utensil hygiene (Q10). Cho 52 reported that in general temporary employees had no training in hygiene and were not trained before starting to work there, which could explain the lower levels of risk perceived by these workers. No differences were observed (p < 0.05) between the socio-demographic variables and the perception of risks related to "eating incompletely cooked meat" (Q4), "handling by sick worker" (Q6) and "integrated pest control" (Q11) (data not shown).

Variables
For the risk "food contamination due to a lack of hand sani-  at room temperature" than those with lower levels, a fact also observed in school meals services 28 .
Women showed higher levels of risk perception than men. Lordelo et al. 62 explains the differences related to gender with respect to risks based on the human evolution theory, which predicts differences according to the domain evaluated (environmental challenge, mating and resources, intergroup competition, fertility and reproduction). Similarly studies related to health found differences between the genders, and verified that women showed lower frequencies than men in risky behaviors 63 . Other study indicated greater involvement of men than of women in risky behaviors and situations, especially when considering age, history of the individual's life and cultural particularities 64 . Lordelo et al. 62 affirmed that risk behavior is oriented by certain regularities and should be defined in relation to different ambiences, with consequences on the behavioral results.

CONCLUSIONS
Based on the facts exposed by this survey, it can be inferred that the lowest levels of risk perception by food handlers concerned "eating soft-cooked eggs" (Q5), "making use of leftovers" (Q2), "cooking temperature" (Q4) and "cross contamination" (Q8), which represents important barriers to food safety and reinforces the lack of knowledge and consciousness about the subject. The formation of subgroups amongst the risk perception questions could also be observed, with the conclusion that the respondents perceived less risks for the questions related to food preparation than for those related to operational hygiene and integrated pest control. In addition, women showed a greater level of risk perception than men.
The results indicate a worrying scenario related to food safety in the tourism sector. Moreover, they point to the need for a greater sanitary fiscalization and for the development of public policies that consider educational actions to adjust this sector, as realized in commercial and institutional restaurants around the country.
Despite the sample have been a non-probabilistic one and the survey have been conducted just in a touristic region of the São Paulo state -which limits the results generalization -, the compilation of several aspects can be considered in future studies to identify differences on risks of food safety in hotels between regions.
The authors hope that the methodological instruments for the analysis of risk perception by handlers in food services in the tourism sector presented and discussed in the present research, contribute to minimizing the lack of information concerning this subject. In addition, they hope the results contribute to the development of public programs aimed to help the sector have a better understanding of the behavior of handlers in the operations that represent health risks to the consumers.