Food safety knowledge, attitudes and practices of restaurant food handlers in a lower‐middle‐income country

Abstract Global research attention appears to be focused predominantly on self‐reported than observed food safety practices. The purpose of this study was to determine the food safety knowledge, attitudes, and self‐reported and observed practices of food handlers in 22 urban restaurants in Zimbabwe. A piloted questionnaire was used to gather qualitative data regarding socio‐demographic variables, food safety knowledge (FSK), attitudes, and self‐reported food handling practices (SRFHPs). A predesigned checklist was used to observe the food handling practices. FSK scores were significantly higher in food handlers who received basic food safety training compared to those who did not (p < .05). No differences in food safety knowledge and attitudes were noted based on the socio‐demographic characteristics of the food handlers (p > .05). A significant positive correlation was observed between FSK and attitudes (r s = 0.371, p < .05), FSK and SRFHPs (r s = 0.242, p < 0.05), FSK and observed food handling practices (OFHPs) (r s = 0.254, p < .05), attitudes and SRFPs (r s = 0.229, p < .05), and attitudes and OFHPs (r s = 0.263, p < .05). About half of the food handlers washed their hands in sinks meant for washing cutlery, 57% did not use approved hand drying methods, and 19.8% did not adequately thaw frozen foods. Food was commonly defrosted either under room temperature or using hot water (>45°C). Results suggest a need for mandatory basic and advanced training to improve the food safety knowledge, attitudes, and practices.

Zimbabwe, there is currently no legal requirement for food premises to be HACCP or ISO 22000 certified. Resultantly, these preventive food safety systems are currently not widely applied in the food service sector.
Studies predominantly associate food poisoning with poor food handling practices (Clayton, 2002;McIntyre, Vallaster, Wilcott, Henderson, & Kosatsky, 2013). Food handlers seem to be a major source and means of food contamination, particularly in ready to eat food, such as that served in restaurants. Soares, Almeida, Cerqueira, Carvalho, and Nunes (2012) highlighted that the majority of food handlers had hand contamination with coagulase-positive staphylococci. In addition, an investigation by Lee, Halim, Thong, and Chai (2017) showed that 48% of food handlers whose hands were swabbed for microbiological assessment had salmonella whereas about two-thirds had ≥ 20cfu of total aerobic counts. Similarly, Illes, Toth, Dunay, Lehota, and Bittsanszky (2018) swabbed school kitchen utensils that come into contact with food and reported that most utensils were contaminated with mesophilic aerobic bacteria. They reported a strong correlation between food handlers' knowledge on food hygiene and microbiological contamination of the utensils.
Zimbabwe is a lower-middle-income country (World Bank, 2019).
Its gross domestic product (GDP) per capita is 1.411 (World Bank, 2019). Restaurant food handlers in Zimbabwe are an understudied group with regards to food safety. There are no published local studies on this category of workers. Understanding the knowledge, attitudes, and practices of food safety by restaurant food handlers enables regulatory authorities to take evidence derived measures toward the provision of safe and wholesome food to the consumer.
Such measures may include appropriate educational interventions that effectively address the food handlers' knowledge gaps, attitudes, and practices on food safety (Gillespie, Little, & Mitchell, 2000;Illes et al., 2018). The current study was, therefore, conducted to assess the knowledge, attitudes, and practices (observed and self-reported) of restaurants' food handlers on food safety and the factors that influence the use of safe food handling practices.

| MATERIAL S AND ME THODS
A cross-sectional study was carried out in Bindura town (17.30138 S, 31.31988 E) between January and May 2018. Bindura is the provincial town of Mashonaland Central Province, Zimbabwe. A total of 101 food handlers engaged in the preparation, serving, and sale of food were purposively selected from 22 commercial restaurants.
About 77% of the restaurants were small-sized and served up to 100 meals a day. The remainder were medium-sized and served over 100 meals a day. Despite preparing, serving, and handling highrisk foods (e.g., salads, eggs, meat, poultry, fish, and rice), none of the restaurants had a HACCP and/or ISO 22000 certification. The clientele comprised mainly lecturers and students from the three local universities, workers from various government departments and nongovernmental organizations, and workers from commercial and industrial sectors. Figure 1 illustrates the sample selection process. Data were collected using a questionnaire and an observation checklist guide.

| Questionnaire
A structured questionnaire was developed based on the HACCP principles, elements assessed in similar past studies (Sharif, and AI-Malki, 2010;McIntyre et al., 2013;Rebouças et al., 2016), and the requirements of Zimbabwe's food safety legislation (Ministry of Health andChild Care 1996a, 1996b). The questionnaire was designed to assess the knowledge, attitudes, and self-reported work practices of food handlers on food safety. It was designed in English, translated to, and administered in the local language (ChiShona) and then retranslated to English for data analyses and reporting. Questionnaire translation was carried out independently by two native ChiShona speakers. They exchanged the translated versions and harmonized them through discussion (discussion meeting attended by two authors) to produce one improved document.
Kappa coefficients (k) were calculated to measure the level of interrater reliability. The raters were two professional food inspectors and were given the KAP questionnaires that comprised 55 items (20 items on food safety knowledge assessment, 20 on self-reported practices and 15 items on food safety attitudes). They rated whether each item in the questionnaire needed further revision to improve on clarity. The raters either said yes (where further revision was needed) or no (where no further revision was need). Then, kappa coefficients were calculated using a procedure described in literature (Cohen, 1960;Kottner & Dassen, 2008) and values ranged from 0.76 to 0.94, which demonstrated a good measure of reliability.
Participants were advised not to write their names or any form personal identification details in order to ensure anonymity and to reduce respondent bias. Confidentiality was guaranteed by informing participants that their individual responses were not to be accessible or communicated to management. To improve on the validity of the questionnaire, it was peer-reviewed by two professional food inspectors (Abdul-Mutalib et al., 2012;Soares et al., 2012) and piloted (Pichler, Ziegler, Aldrian, & Allerberger, 2014;Soares et al., 2012;Woh, Thong, Behnke, Lewis, & Jain, 2016) with 25 respondents (24.8% of the sample size). The revised version of the questionnaire was self-administered to the study participants. Prior to administration of the questionnaire, a 10-15 min training of the respondents was carried out on how to answer the questionnaire, the importance of the reliability and completeness of collected information, the study's objectives and the respondents rights.
Questionnaires were completed in the presence of the trained research assistants in order to preclude the respondents from being assisted by workmates and looking up information (Pichler et al., 2014). In circumstances where a respondent was unable to read or write, a trained research assistant interviewed the respondent and completed the questionnaire. Each questionnaire had five sections: four (4) was accorded for a correct response or practice while a zero (0) was given for an incorrect one. A measurement scale (0 -4) was applied to interpret the mean KAP scores: 0 = poor, 1 = unsatisfactory, 2 = average, 3 = satisfactory, and 4 = excellent.

| Observation checklist guide
Visual observations of the food handling practices were carried out by three trained research assistants under the supervision of two certified food inspectors. A predesigned observation checklist guide was used. The guide contained same questions that assessed selfreported food handling practices in the questionnaire described earlier on. It provided some form of methodological triangulation by validating information pertaining to the self-reported practices.

| Statistical analyses
Analyses were performed using the Statistical Programme for Social Sciences (SPSS) version 20.0. A chi-squared test was performed to determine whether the food handlers' food safety knowledge and F I G U R E 1 Sample selection process attitudes differed with education level, age, gender, and work experience and to determine whether there were significant differences between self-reported and observed food handling practices.
The scores with respect to food handlers' food safety knowledge and attitudes were summarized using descriptive statistics (mean and standard deviation). The Spearman's correlation test was carried out to determine the association among knowledge, attitudes, self-reported practices, observed practices, and work experience. Statistically significant differences were considered at 95% level of confidence (p < .05). Table 1 shows the association of socio-demographic characteristics with food safety knowledge and attitudes. No differences in food safety knowledge and attitudes were noted based on the gender, age, educational level, and work experience of the food handlers (p > .05). This is consistent with a study by Abdul-Mutalib et al. (2012), which did not find a significant association between the respondents' knowledge level and socio-demographic characteristics. On the contrary, McIntyre et al. (2013) reported that socio-demographic factors such as the length of work experience in the food industry and food handlers' level of education were significantly associated with improved food safety knowledge. More studies are required to better understand the influence of socio-demographic factors on the food safety knowledge and attitudes of food handlers. Most of the participants (81.2%) in the current study were females aged between 18 and 37 years (27.34 ± 6.8 years) and were married. In the Zimbabwean context, the preponderance of women labor in restaurants is not surprising as this category of the population is traditionally considered primarily responsible for household food preparation activities. Consistent with this tradition, the food service sector is generally dominated by the female labor force. The large number of female handlers in our study is consistent with reports from similar past studies (da Cunha et al., 2014;Lee et al., 2017;Martins, Hogg, & Otero, 2012). Over 50% of the women had up to 3 years of work experience as a food handler in the same restaurant. The lack of varied food handling experience may increase the risk of food contamination due to nonuse of good manufacturing practices. About 32% of the food handlers had not gone beyond primary education, with over 80% (26) of them being female. Given that women perform various food handling activities at work and at home, it appears very critical to take into account their educational status and literacy level when designing and implementing food safety training programmes.

| Association of socio-demographic characteristics with food safety knowledge and attitudes
The food safety knowledge of food handlers who had received basic food training significantly differed from those who did not (p < .05; Table 1 food safety training is not yet mandatory in Zimbabwe (Ministry of Health and Child Care, 1996a,b), which may account for the low number of trained food handlers. We recommend widening the scope of Zimbabwe's food safety laws to include a requirement for managers in the food service sector to provide periodic food safety training to food handlers. Such training should be provided every 6-12 months (da Cunha, Rosso, Pereira, & Stedefeldt, 2019), and its effectiveness should be evaluated (Soares et al., 2012;ISO, 2018).

| Knowledge of participants on food safety
The descriptive statistics concerning the study participants' food safety knowledge is shown in Table 2. The highest knowledge score was 3.96 ± 0.40 and pertained to the risk to food contamination by food handlers suffering from diseases such as diarrhea, sore throat, syphilis, and flu. This shows that participants understood the risk to food contamination that an unhealthy food handler posed. Similar studies have demonstrated that food handlers have good knowledge with regards to this issue (da Cunha et al., 2014;McIntyre et al., 2013;Pichler et al., 2014). However, some studies reported that food handlers lacked adequate knowledge about the risk of contamination of food by diarrheal food handlers (Clayton, 2002;Osaili, Obeidat, Jamous, & Bawadi, 2011).
At least 85% of the participants had satisfactory to excellent (3-4) mean knowledge scores concerning the statements: inadequate thawing of food can contribute to bacterial food poisoning (BFP), contact between raw and cooked foods contributes to food contamination, not wearing rings, watches, necklaces minimizes food contamination, cleaning, and sanitizing utensils reduces the risk of food contamination and keeping nails short and unpainted reduces the risk of food contamination (Table 2). McIntyre et al. (2013) reported that most food handlers had excellent knowledge about how to thaw frozen foods such as red meats. Smigic et al. (2016) and Bou-Mitri et al. (2018) demonstrated that over 90% of the food handlers knew that separate storage of cooked and raw foods is necessary to prevent bacterial food contamination. In Brazil, da Cunha et al., (2014) reported that about 92% of the food handlers knew that use of earrings, rings, and watches could contribute to food contamination. In Portugal, Martins et al. (2012) reported significantly higher food handlers' knowledge on surface and utensils hygiene than the overall food hygiene knowledge.
The least mean food safety knowledge score of 2.30 ± 1.99 was recorded in response to the use of dish towels to wipe hands as source of food contamination (Table 2). This means that the food handlers lacked knowledge regarding cross contamination of dish towels by pathogens from fingers, nails, and palms from the drying with dish towels. About 32% of the food handlers were not aware that handwashing in sinks for washing cutlery increases the risk of food contamination. This highlights the need to strengthen food handlers' food safety knowledge through the provision of basic food safety training, as the majority of them never attended such training (Table 1). However, such training should be carefully designed and implemented in order to yield significant improvements in food handlers' attitudes and/or practices. Zanin et al. (2017) conducted review of food handlers' knowledge, attitudes, and practices and reported that in about 50% of the reviewed studies (n = 36) knowledge and/or attitudes were commonly not translated into practices.
Consequently, to enhance transformation of food safety knowledge and/or attitudes into practices, factors such as the training strategy and characteristics of training venue should be taken into account when designing and implementation food safety training programmes (Zanin et al., 2017).
At least 30% of the food handlers in the current study lacked awareness that the use gloves to handle raw foods reduces the risk of food contamination, as shown in Table 2. Bou-Mitri et al. (2018) also reported substantial deficits in food safety knowledge of hospital food handlers, with regards to this particular issue. Approximately 40% of the food handlers did not know that refreezing defrosted food contributes to bacterial food poisoning (Table 2). This is consistent with previous food safety studies (Abdul-Mutalib et al., 2012;Sani & Siow, 2014).
About 73% of the food handlers indicated that the safe operating temperature for a refrigerator is 1-5°C, as shown in Table 2.
(2014) recommend that the food safety training should be conducted more regularly, such as every six to twelve months, in order to refresh food handlers on learnt content. In recognition of the value of food safety training the European Union requires that member states offer the training at least annually to food handlers (Regulation EC, 2004). However, in resource-constrained settings, a more regular frequency, such as every quarter, may be required due to higher turn overs of workers. Restaurant food handlers may also benefit from advanced food safety training such as HACCP and ISO 22000:2018. Overall, the grand mean score concerning knowledge on food safety was 3.03 ± 1.71 and the average percentage of correct responses was 75.8% (Table 2). This is comparable with the work of da Cunha et al. (2014). Both studies cement the conclusion that food handlers' food safety knowledge is insufficient and require major improvements as highlighted earlier on in this article.

| Food safety attitudes
The food safety attitudes of restaurant food handlers are presented in  (Jenner et al., 2006). We propose that restaurant managers should commit themselves to building a positive food safety culture (PFSC) among food handlers. A PFSC is defined as an organization's culture that is strongly supportive of food safety and is perceived to be important to the accomplishment of the organization's vision (Griffith, 2013). Improving the PFSC within an organization yields hygiene compliance not only of existing workers but also potentially for new ones (Griffith, 2013).
Further studies are required to investigate the key factors that perpetuate the negative attitude toward issues such as reporting of illness and taking of sick leave, and to determine required remedies.
About 23% of the participants considered microbiological swabbing (surface sampling) of their palms and nails as not useful with regards to assessing the effectiveness of handwashing.
Previous researches support the conclusion that food handlers'
palms and nails pose a major risk to food contamination. For example, in a study of food handlers working in school kitchens in Brazil, Soares et al. (2012) reported hand contamination with coagulase-positive staphylococci. In addition, an investigation by Lee et al. (2017) showed that 48% of food handlers whose hands were swabbed for microbiological assessment had salmonella whereas about two-thirds had ≥ 20cfu of total aerobic counts. Therefore, it may be necessary for restaurant managers to motivate workers toward a culture of adopting positive food safety attitudes.
Worker motivation has been associated with desirable outcomes such as having positive food safety attitudes and use of recommended food handling practices (Seaman, 2010). From a public health perspective, regular health inspections of food workers and closure of food premises that do not comply the relevant health and safety standards has been recommended to minimize spread of food-borne diseases (Woh et al., 2016). Table 4 shows the association between self-reported and observed food hygiene practices. Data from field observations indicate that contrary to self-reported information, a substantial proportion of food handlers did not use a detergent or disinfectant to wash their hands before food handling or post-handling potentially contaminated materials (p < .05; Table 4). Nonsanitization of hands can contribute to food contamination with pathogenic bacteria (Abdul-Mutalib et al., 2012). Handwashing with an antibacterial soap can remove over 95% of coliform counts (Toshima et al., 2001). Therefore, improved hand hygiene should be promoted to reduce the risk of transmission of bacterial food-borne diseases.

| Association between self-reported and observed food hygiene practices
On the other hand, our findings are consistent with previous food safety studies that showed that food handlers' self-reports with Note: Scores (0-4): 0 = least score (poor); 1 = unsatisfactory, 2 = average, 3 = satisfactory, 4 = highest score (excellent).

TA B L E 3 Food handlers' food safety attitudes
respect to the use of desirable hand hygiene practices were largely not supported by findings from observation checklists (Dharod et al., 2007;Reboucas et al., 2016). Self-reporting has been de- A significant proportion of the food handlers did not dry their hands using approved methods such as disposable towels, air drier, and frisk drying (p < .05;  Table 4). Using the same cutlery such as cutting boards and knives to prepare raw and cooked foods can contribute to cross contamination of food (Abdul-Mutalib et al., 2012). In addition, there were significant mismatches between self-reported and observed food handling practices such as smoking, eating, sneezing, and nose-poking when preparing food (p < .05;

| Strengths and limitations of the study
This study used a piloted and validated questionnaire whose reliability was satisfactory (k = 0.76-0.94). In addition, each questionnaire was completed in the presence of trained research assistants.
Methodological triangulation of self-reported data concerning the food handling practices was carried out using task performance observations. Furthermore, the study was carried out among restaurant food handlers in the lower-middle-income country, which helps to fulfill the lack of research in such contexts. On the other hand, the current study is subject to some methodological limitations. First, we did not examine the knowledge, attitudes, and practices of restaurant managers regarding food safety. Since food safety managers may have significantly higher performance level than food handlers (Illes et al., 2018), our findings may not be generalizable to them.
Further research may need to assess the food safety KAP of restaurant managers. Second, cross-sectional designs lack the capacity to definitely demonstrate cause-effect relationships (Ncube et al., 2017

ACK N OWLED G EM ENTS
The authors are grateful to all food handlers who participated in the present study.

CO N FLI C T S O F I NTE R E S T
All authors declare no conflicts of interest in this article.

E TH I C A L S TATEM ENT
The study conforms to the Declaration of Helsinki, US and does not involve any human or animal testing. The study protocols and procedures were ethically reviewed and approved by Bindura municipality and the authors' university.