Factors associated with safe food handling practices in the food service sector

Abstract

Purpose

This review assessed the methodological quality of relevant food safety studies, investigated the available evidences for factors associated with safe food handling practices (SFHPs) and suggested required improvements.

Methods

A systematic literature search was carried out in five databases. A pre-designed criteria was used for data extraction and quality assessment. Evidence synthesis was performed, based on specified criteria.

Results

Out of a total of 1768 published research articles, 28 met the inclusion criteria. Strong evidence suggests that food handlers’ knowledge and attitudes on food safety contribute to SFHPs. Moderate evidence exists about the association between a food handler’s food safety training and SFHPs. Limited evidence is available about the influence of the educational level and work experience of a food handler, to SFHPs. Insufficient evidence exists about the association between socio-demographic characteristics and SFHPs. A conceptual framework is presented to illustrate the associations.

Conclusions

Available evidence supports that in order to promote SFHPs, responsible authorities should focus on improving on the food safety knowledge and attitudes of food handlers. Future food safety studies may richly benefit from overcoming the methodological shortcomings presented in this review.

Electronic supplementary material

The online version of this article (10.1007/s40201-019-00424-4) contains supplementary material, which is available to authorized users.

Introduction

Health security means the protection of humans from urgent health threats, such as threats associated with unsafe food [1]. Unsafe food handling practices by food handlers contaminate food with disease causing organisms, and render it unsafe for human consumption [2]. Earlier studies implicate unsafe food handling practices of food handlers in most bacterial foodborne disease outbreaks [3–5]. Such outbreaks culminate in a plethora of negative consequences on human health, wellbeing and the economy [1]. For example, in the United States alone, Scharff [6] reports that the annual cost of foodborne illness is about US$ 51 billion in the basic cost-of-illness model and US$ 77.7 billion in the enhanced model. Globally, approximately 2.2 million persons die due to food and waterborne illnesses [1]. Therefore, from the primary production to consumption, safe food handling practices (SFHPs) are an essential task for the food industry [7].

A food handler is a person whose job description entails activities such as handling, storage, preparation, manufacturing, selling or serving food or cleaning equipment, surfaces and premises that come into contact with food [4, 8]. Since these activities often entail coming into direct contact with food, food handlers’ improper food handling practices are a cause of concern for persons charged with the responsibility of safeguarding public health. SFHPs can reduce the transfer of foodborne pathogens from food workers to consumers [4, 9, 10]. Consequently, an understanding of the factors associated with SFHPs by food handlers may assist food safety managers and regulatory authorities to develop and effectively implement required food safety interventions.

Studies carried out over the past two decades have attempted to determine factors that are significantly associated with SFHPs. Resultantly, rich and contradictory evidence has accumulated regarding the factors that influence SFHPs. For example, food safety training has been reported as significantly influencing [2, 5, 11, 12] and as not significantly influencing [13–15], the use of SFHPs. In order to evaluate the evidences for factors associated with SFHPs, it is necessary to review such literature using a pre-defined criteria. Therefore, the objectives of this review were to determine the methodological quality of relevant studies and systematically evaluate the available evidences for factors associated SFHPs.

Material and methods

Search strategy

A literature search strategy was developed in close consultation with two university librarians and implemented in five databases. The databases searched were the (a) Scopus, (b) Taylor and Francis, (c) PubMed, (d) Food Science and Technology Abstracts, and ISI Web of Science. In addition, references of each potential article were screened to identify more articles that met the inclusion criteria. The search terms used were various combinations of the terms: food safety, food contamination, food hygiene, food poisoning, knowledge, perceptions, attitudes, practices, training, predictors, factors and determinants, food handler, food worker, caterer, preparer, food service personnel, restaurant, foodcourt, premises, service, institution, sector, catering, establishments, commercial sector and retail food shops, (Appendix 1). The period covered was January 2010 to December 2018.

Inclusion and exclusion criteria

Studies considered in this review were: (a) peer-reviewed journal articles that assessed the relationship between at least one factor and SFHPs, (b) published in the period January 2010 to December 2018, (c) researches on food handlers primarily engaged in the preparation, processing, serving and handling of food and cleaning of food premises, cutlery and working surfaces [4, 8], (d) articles written in the English language or non-English abstracts which were translated to the English language. The following article types were excluded from this review: review articles, thesis, commentaries, editorials, letters to the editor, and correspondences. These article types were considered less relevant with regard to yielding original and sufficient data for the purposes of this review. We excluded articles that (a) assessed variables, for example knowledge and attitudes of food handlers, without directly correlating the variables to food safety practices, and (b) investigated food safety practices of personnel whose job description does not primarily entail food handling (e.g. nurses, university students).

Data extraction

Two researchers independently screened the articles for eligibility. A standard data extraction form was used. It gathered information on the (a) author/s, (b) publication date, (c) setting, (d) research design, (e) sample size, (f) data collection instruments, (g) exposures (factors), (h) outcomes assessed, and (i) the main findings. In this review, the term outcome refers to SFHPs (self-reported or observed). Exposures refers to factors associated with SFHPs: (a) socio-demographic factors (age, gender, marital status, work experience, education and income), (b) food safety training, (c) knowledge, (d) attitudes, (e) type of handwashing facilities, (f) presence of rodents and insects in food premises, and (g) the location, adequacy and type (government or private) of the food premises. The accuracy of information in the data extraction forms was cross-checked by two authors. Discussions were held among the authors to reconcile findings and reach a consensus.

Methodological quality assessment

A framework to appraise the methodological quality of articles that met the criteria for inclusion was developed. The process of developing the concerned appraisal tool comprised two phases. In phase 1, authors reviewed, compiled and adapted items considered to minimise risk of bias in sampling, data collection and reporting, in previously used protocols [16–21]. This phase yielded a draft methodological quality appraisal tool that comprised 14 items. A good quality research provides evidence of how issues of validity and reliability have been addressed [21]. Phase 2 entailed validity and reliability assessment of the draft methodological quality appraisal tool. With regard to validity, the tool and study protocol were presented in a seminar comprising members from the environmental science department in the authors’ university. In this seminar, face validity was ensured by providing participants with opportunities to share their expert input and advice on existing criteria that have been used to evaluate the methodological quality of primary studies. Participants assessed content validity by reviewing each item to determine whether they considered it to be essential or not essential for evaluating the quality of primary studies. Results were summarised using a Content Validity Ratio score (CVRS) that denotes the extent to which participants considered each item in the draft appraisal tool to be essential. The CVRS ranges from −1.00 to +1.00 and a CVRS = 0.00 indicates that 50% of the participants considered an item to be essential [22]. For the purposes of the current study, authors considered a CRVS >0 to show that an item was essential. All items rated essential were subjected to inter-rater reliability testing, whereby two-authors independently used the tool to rate primary studies included in this review. Percent agreement between raters (equivalence) ranged from 83.3–100% (median 91.6%), which demonstrates consistency in the ratings [22]. All discrepancies in the ratings were reconciled through discussion and consensus among authors. Based on validity testing and the advice from the journal peer-reviewers for this article, we revised (deleted, added or edited) some items in the methodological quality appraisal tool. For example, the criterion concerning the study design initially assessed whether a primary study used a design capable of demonstrating causality. Based on peer review comments it was revised to assess whether a primary study used a design capable of demonstrating temporality. In this regard, the validation process helped to improve the clarity and relevance of the items contained in the methodological quality appraisal tool. The final tool comprised twelve (12) items as shown in Table 1. Each item (strength) was worth a point. A zero (0) was accorded to represent a methodological shortcoming of a study. The overall quality score of each study was calculated by expressing the study’s strengths as a percentage. This entailed dividing the number of strengths of a given study by 12 (total possible strengths in terms of our criteria) and multiplying the outcome by a 100.

Table 1.

Methodological quality assessment framework

Criterion	Description of criterion
Study design	What study design was used? Does it demonstrate temporality?
Sample size	Was an adequate sample size used (n ≥ 100)?
Sampling techniques	Were the study participants randomly sampled?
Analytical rigour	Were appropriate statistical analyses carried out and described?
Observed FHPs	Did the authors observe the participants’ food handling practices or they used participants’ self-reports? (questionnaires or interviews)
Multi-centre study	Was the study multi-centred or single-centred?
Validity	Was the study piloted? Were validated data collection instruments used?
Reliability	Did the study describe the reliability of data collection instruments?
Peer review	Is there evidence of external peer review?
Study limitations	Did the author/s acknowledge the study limitations?
Conclusions and recommendations	Evidence-supported conclusions and or/ recommendations? Were the stated conclusions and recommendations congruent with the study design and the data collection techniques used?
Research ethics	Is there evidence of ethics approval or description of any ethical issues observed?

Evidences for factors associated with SFHPs

The present review sought to answer the question: What is the status of available evidences of factors associated with SFHPs in the food service sector? In an attempt to answer this question, studies with a methodological quality score of <50% were not considered. The following criteria were designed and then applied:

1. Strong evidence: Consistent results in ≥70% of studies with a quality score ≥ 50.

2. Moderate evidence: Consistent results in 60–69% of studies with a quality score ≥ 50.

3. Limited evidence: Consistent results in 50–59% of studies with a quality score ≥ 50.

4. Insufficient evidence: association reported in <50% of studies with a quality score ≥ 50 or investigated by less than three studies.

Results and discussion

Study selection

A total of 1768 articles were identified, of which 1423 (80.5%) titles were retrieved from the selected databases and 345 (19.5%) from references of articles that met the inclusion criteria. The screening of the articles yielded a total of 76 (4.3%) duplicates. Titles and abstracts of the remaining 1692 articles were checked to identify relevant articles for full text download and consideration for review. In circumstances where the abstract was not provided, or where it was not clear from the abstract if a study was relevant to the objectives of the present study, the entire article study was retrieved. In total 28 (1.6%) full text articles met the inclusion criteria and were subjected to the critical appraisal tool for methodological quality assessment.

Characteristics of included studies and their methodological quality

Table 2 presents the characteristics of the 28 studies that met the inclusion criteria for review. The World Bank classifies countries into three main categories: low-, middle- and high- income [43]. This review had a fair representation of studies from each of the categories as 17.9% were from low-, 67.9% from middle- and 14.3% from high-income countries. According to Stratev et al. [44] the burden of foodborne illness is global and therefore requires global efforts with regard to issues such as collaboration, awareness, funding and commitment from governments. The study participants in this review were food handlers (n = 7342) engaged in various work settings: restaurants, meat processing plants, institutions (schools, universities, and hospitals), dairy plants, and mobile vending. Of the 28 studies, two (2) did not provide information about the gender of the participants. For the twenty-six studies (n = 7155 participants) that described the gender distribution, 54.8% were female food handlers. The top ten (10) investigated factors were food safety knowledge (64.3%), attitudes (53.6%), educational level (50%), food safety training (39.3%), gender (39.3%), work experience as a food handler (35.7%), age (32.1%), marital status (17.9%) and type of food premises (10.7%). The percentage (%) shows the proportion of studies that investigated the concerned factor (n = 28 studies).

Table 2.

Characteristics of included studies

Author	Study Setting	Methodological issues	Quality score (%)	Factor (exposure)	Outcome	Results
Bou-Mitri et al. [15]	Lebanese private and government hospitals	Cross-sectional (n = 254; M = 162, F = 92) Non-random sampling Self-reported food handling practices (SRFHPs)	75	Gender Work experience Education Training Premises type (government or private) Location	SRFHPs	Reported significantly higher scores concerning SRFHPs (a) of men in comparison to women, (b) of food handlers who worked in government hospitals than those in private hospitals and (c) in relation to work experience. Grand KAP scores were significantly higher among food handlers in government hospitals than those in private ones. Training significantly associated with knowledge but not with attitudes, practices and overall KAP. Age, education and hospital location were not significantly related to knowledge, attitudes, practices or grand KAP scores.
De Souza et al. [23]	Restaurants food handlers in Brazil	Cross-sectional, Small sample (n = 70; M = 52, F = 18) Study limitations not outlined	75	Knowledge Gender Work experience Education Training	SRFHPs and observed food handling practices (OFHPs)	Reported a positive correlation between food safety knowledge and SRFHPs. No correlation between: (i) socio-demographic characteristics and SRFHPs and (ii) training and SRFHPs. Reported no correlation between food safety knowledge and OFHPs. OFHPs had no correlation with SRFHPs.
Vo et al. [24]	Canteens in Southern Vietnam	Cross-sectional (n = 909; M = 217, F = 692)	92	Education Knowledge Attitudes	SRFHPs and OFHPs	Reported no significant association between attitudes and SRFHPs Knowledge and education were associated with SRFHPs.
Zanin et al. [25]	Seafood workers in Brazil	Cross-sectional (n = 193; M = 165, F = 28) SRFHPs Study limitations not outlined Non-random sampling used	67	Age Education Work experience Knowledge Attitudes	SRFHPs	Reported significant positive correlation of SRFHPs with attitudes and age of the workers. No correlation of SRFHPs with knowledge, work experience and education.
Kunadu et al. [26]	Institutional food handlers in Ghana	Cross-sectional, (n = 278; M = 67, F = 211) SRFHPs		Gender Education Marital status Knowledge Attitudes	SRFHPs	Reported significant positive correlations between: (i) knowledge and attitudes, (ii) knowledge and SRFHPs and, (iii) attitudes and SRFHPs. Gender, education and marital status were not significantly related to SRFHPs.
Husain et al. [27]	School canteen food Handlers in Malaysia	Randomized controlled trial Small sample size (n = 79) No description of research ethics	83	Training	SRFHPS and OFHPs	Following implementation of food safety training, the SRFHPs in the intervention group were significantly higher compared to baseline. Reported a significant within-group and between-group improvement of OFHPs due to the training intervention.
Shuvo [28]	Biscuit industry, Bangladesh	Cross-sectional, (n = 110; M = 50, F = 60) SRFHPs Study limitations not outlined	75	Gender Knowledge Training	SRFHPs	Reported significant positive correlation between: (i) knowledge and SRFHPs, (ii) attitudes and SRFHPs, (iii) gender and SRFHPs, and (iv) food safety training and SRFHPs.
Jeon et al. [29]	Restaurants food handlers in South Korea	Cross-sectional (n = 140, M = 81, F = 59) SRFHPs No description of research ethics	75	Age Gender Education Income Work experience	SRFHPs	Reported significant differences in SRFHPs among (i) different age groups, (ii) educational levels and (iii) income levels. No significant differences in SRFHPs were reported in relation to the food handlers’ gender and work experience
Chen et al. [30]	Dairy plant workers in China	Cross-sectional (n = 194, M = 83, F = 110) Non-random sampling SRFHPs	75	Education Work experience Attitude Knowledge	SRFHPs	Reported a significant correlation between (a) attitudes and knowledge, (b) attitudes and SRFHPs (behaviour), (c) education with knowledge, attitudes and SRFHPs. No association between (a) attitude and SRFHPs (b) knowledge and SRFHPs and (c) work experience with knowledge, attitudes and behaviour.
Asmawi et al. [31]	Food courts in Malaysia	Cross-sectional (n = 108) SRFHPs	83	Knowledge Attitude	SRFHPs	Reported a significant correlation between (a) knowledge and attitude, (b) knowledge and SRFHPs, and (c) attitude and SRFHPs.
Derso et al. [12]	Ethiopian restaurants	Cross -sectional (n = 417, M = 93, F = 324) SRFHPs Study limitations not presented	75	Training Education Work experience Type of handwashing Age Gender Marital status	SRFHPs	Reported that factors significantly associated with SRFHPs were food safety training, education, work experience, and type of handwashing.
Woh et al. [32]	Migrant food handlers in Malaysia	Cross-sectional (n = 383, M = 364, F = 19) SRFHPs Non-random sampling	75	Gender Age Marital status Training Education Knowledge	SRFHPs	Reported a significant association between (a) training and SRFHPs, (c) knowledge and country of origin, and (d) education and knowledge Gender, age, marital status, education and knowledge were not significantly related to SRFHPs.
Al-Shabib et al. [33]	Food handlers from a university in Saudi Arabia.	Cross-sectional, Small sample (n = 87; all male) SRFHPs, No details of piloting and reliability status data collection instrument, and study limitations.	42	Knowledge Training Attitudes Work experience	Practice	Reported significant positive correlation between (a) knowledge and attitudes, (b) knowledge and training, (c) knowledge and personal hygiene (practices), (d) training and personal hygiene (e) attitudes and personal hygiene. No significant association was found between (a) work experience and knowledge, and (b) training and attitude. NB. Study did not report on the association between work experience and practices.
Ismail et al. [34]	Mobile food handlers in Malaysia	Cross-sectional (n = 320; M = 168, F = 150) SRFHPs No details of piloting and reliability status data collection instrument, study limitations and ethical issues	50	Knowledge	SRFHPs	Reported that standardized beta coefficient showed food safety knowledge as strongly influencing food safety practices
Mohd. Firdaus Siau et al. [35]	Food courts in Malaysia	Cross-sectional, (n = 274, M = 193, F = 81) SRFHPs No description of ethics and the study’s limitations	75	Age Education Training Knowledge Attitudes KAP	SRFHPs	Reported a significant correlation between (a) education and practices, (b) education and attitudes, (c) knowledge and attitudes, (e) attitudes and SRFHPs, (f) age and knowledge, and (g) training and knowledge. Study did not report on association between (a) training and SRFHPs, and (b) age and SRFHPs.
Adesokan, Raji [36]	Meat processing plants in Nigeria	Cross-sectional, (n = 190, M = 133, F = 57) SRFHPs Non-random sampling	67	Premises type (government or private) Gender Age Work experience Education Attitude	SRFHPs	Reported that the demographic variables with a significant effect on meat handling practice were age, gender, education and work experience. Reported significant relationship of (a) plant type with knowledge on meat handling and with meat handling practices (b) attitudes and practices
Tessema et al. [37]	Food and drinking premises in Ethiopia	Cross-sectional (n = 406, M = 151, F = 255) No description of the study’s limitations	83	Marital status Monthly income Knowledge Presence of insects and rodents Existence of shower and separate dressing room	OFHPs	Factors significantly associated with food safety practices were marital status, monthly income, knowledge, presence of insects and rodents and existence of shower and separate dressing room.
Sani, Siow [8]	Malaysian University cafeterias and canteens	Cross-sectional, (n = 112; M = 83, F = 29) SRFHPs Non-random sampling.	83	Attitudes Knowledge	SRFHPs	Reported significant positive correlation between (a) knowledge and attitudes, (b) knowledge and SRFHPs, and (c) attitudes and SRFHPs.
Da Cunha et al. [14]	High risk food premises in Brazil (meat products and raw salad)	Cross-sectional, (n = 183; M = 63, F = 120), No description of research ethics	83	Knowledge Training Work experience Meals served by a food handler Adequacy of buildings and facilities Nutritionist	SRFHPs and OFHPs	Variables significantly associated with use of observed desirable food handling practices were: (a) knowledge (b) work experience, (c) number of meals (d) Adequacy of buildings and facilities, (e) the nutritionist’s presence. No significant difference in food handling practice (whether observed or self-reported) or attitudes was found due to training or lack of training of food handlers.
McIntyre et al. [5]	Food handlers in British-Colombia, Canada.	Cross-sectional (n = 698, M = 264, F = 434) SRFHPs No details of piloting and reliability status data collection instrument.	67	Training Place of employment Education	SRFHPs	Trained food handlers’ self-reported handwashing practices were significantly better compared to untrained ones. Workers with tertiary education (college and university) had significantly higher food safety knowledge than those without.
Ko [38]	Restaurants food handlers in Taiwan	Cross-sectional (n = 421, M = 232, F = 189) Non-random sampling SRFHPs No description of research ethics	67	Knowledge Attitudes	SRFHPs	Reported significant positive relationship between (a) knowledge and attitudes, (b) attitudes and SRFHPs. Relationship between knowledge and SRFHPs was statistically insignificant.
Aziz, Dahan [39]	Malaysian school canteens food handlers	Cross-sectional (n = 364, M = 86, F = 278) SRFHPs No details of piloting and reliability status data collection instrument and ethical issues.	50	Attitudes	SRFHPs	Reported that food handlers’ attitude has a positive and significant impact on SRFHPs.
Tan et al. [40]	Food handlers in primary schools in Malaysia	Cross-sectional, Small sample (n = 85; M = 13, F = 72) Study limitations not outlined	75	Gender Age Work experience Education Knowledge Attitudes Glove use	SRFHPs and OFHPs	Reported that demographic variables with a significant effect on self-reported hand hygiene practice were gender and education. Reported positive correlation between (a) knowledge on food contamination and handwashing practice, (b) mean hand hygiene knowledge and hand hygiene scores No significant correlation of (a) attitudes with handwashing practice, (b) attitudes and personal hygiene, (c) work experience and hand hygiene practices, (d) age and handwashing practices.
Kibret, Abera [2]	Ethiopian food service premises	Cross-sectional (n = 455, M = 121, F = 334)	92	Gender Age Education Training License status Type of premise	Practices Sanitary conditions	Reported significant positive correlation between (a) training and practices, (b) license status and sanitary conditions, (c) type of premise (hotel, cafe, restaurant and juice house) and sanitary conditions, (d) training and food safety practices. No significant association was found between demographic variables (gender, age, education) and practices
Abdul-Mutalib [41]	Restaurants in Malaysia	Cross-sectional, Small sample (n = 64; M = 16, F = 48) Non-random sampling SRFHPs, No details on study limitations and ethical issues.	50	Gender Age Work experience Education Knowledge Attitudes	SRFHPs	Reported a significant correlation between (a) education and attitude, (b) knowledge and attitude, (c) knowledge and SRFHPs, and (d) attitude and SRFHPs. Gender, age, and experience level were not significantly related to the knowledge, attitudes or SRFHPs. Education was not significantly associated with SRFHPs.
Rahman et al. [11]	Street vendors in Malaysia	Cross-sectional (n = 361, M = 171, F = 190) Non-random sampling SRFHPs	75	Age Knowledge Attitude Training	SRFHPs	Reported that (a) age significantly influences food safety knowledge (FSK), (b) FSK and training influence attitudes, (c) FSK, training, attitudes and age influence SRFHPs. The marital status was not significantly related to food safety practices.
Ansari-Lari et al. [42]	Meat processing plants in Iran	Cross-sectional, Small sample (n = 97; M = 89, F = 8) SRFHPs Non-random sampling No details of piloting and reliability status data collection instrument and ethical issues.	42	Knowledge Attitudes Work experience Education	SRFHPs	Reported significant positive correlation between (a) knowledge and attitudes, (b) knowledge and education, (c) work experience and SRFHPs A significant negative correlation reported between (a) knowledge and SRFHPs, and (b) attitudes and SRFHPs.
Park et al. [13]	Restaurants in Korea	Pre-test post-test study Small sample (n = 90, M = 31, F = 59) Non-random sampling	75	Training Knowledge Attitudes	SRFHPs	A significant negative correlation (a) knowledge and practices, and (b) attitudes and practices. Non-significant correlation (a) training and knowledge, (b) training and food safety practice, (b) training and practices, and (c) food safety knowledge and sanitation management performances

The methodological quality of the studies was assessed using pre-designed criteria (Table 1). Based on the criteria, none of the reviewed studies scored more than 92%. The methodological quality scores for studies ranged from 42 to 92%, as shown in Table 2. A summary of the overall methodological issues of the studies is presented in Table 3. In terms of the review criteria, few of the studies (7.1%) used a study design capable of demonstrating temporality. Most were cross-sectional (92.9%) studies with data collected generally within an average period of about three (3) months. A well-known limitation of cross-sectional designs is their inability to demonstrate temporality. This suggests that cohort studies are required to address this gap. An intervention study of school canteen food handlers in Malaysia [27] used randomisation but the sample sizes were small (n < 100) and ethical considerations were not stated. Few studies (32.1%) observed the FHPs of studied food handlers. Studies commonly relied upon self-reported FHPs by food handlers. Fundamentally, relying on self-reported FHPs presents some challenges about the reliability of the collected data. For example, self-reported practices are not necessarily the actual practices of food handlers [15]. This means that in such studies, associations between the concerned factor and SFHPs were not immune from information bias. In light of this finding, we encourage further food safety studies to verify self-reported food handling practices through observations. An earlier review of training studies [45] on food hygiene and safety also reported that studies rarely incorporated researcher observation of food handling practices. Other limitations of studies reviewed in the present investigation pertained to non-random sampling of the study participants and usage of small samples (n < 100 participants). Some of the studies did not describe the study limitations (42.9%) and any research ethics observed (32.1%). On the positive, the majority of studies (82.1%) had measures to for data quality control, such as, the use of validated and reliable data collection instruments. However, studies rarely described other measures such as training rendered to data collectors, monitoring of data collection. Studies (100%) used rigorous statistical methods, were multi-centred, and conclusions drawn were supported by results.

Table 3.

Methodological quality assessment of the 28 included studies

Criterion	Yes	No	Comments
	n (%)	n (%)
Study design capable of demonstrating temporality	2 (7.1)	26 (92.9)	92.9% were cross-sectional in design
Adequate sample size (n ≥ 100)	21 (75)	7 (25)	Small sample sizes used by 25% of the studies
Study participants randomly sampled	16 (57.1)	12 (42.9)	Over 40% of the studies were not immune to sampling bias
Analytical rigor	28 (100)	0 (0)	Statistical analysis appropriate
Multi-center study	28 (100)	0 (0)	At least three food premises studied
Used piloted and valid data collection instruments	23 (82.1)	5 (17.9)	Validated data collection instruments used
Described or tested reliability of data collection instruments	23 (82.1)	5 (17.9)	Most studies used reliable data collection instruments
Peer-review	28 (100)	0 (0)	All articles were published in peer-reviewed journals
Observed the food handling practices	9 (32.1)	19 (67.8)	Self-reports of food handlers commonly used
Acknowledged the study limitations or biases	16 (57.1)	12 (42.9)	Study limitations not provided in over 40% of the studies
Provides evidence-supported conclusions and/or recommendations	28 (100)	0 (0)	Conclusions supported by results
Provided evidence of ethics approval	19 (67.9)	9 (32.1)	Ethics approval evidence missing in over 32% of the studies

Factors associated with SFHPs

Food safety knowledge

The factors associated with SFHPs are presented in Table 4. There is strong evidence on the influence of food safety knowledge on a food handler’s SFHPs. Specifically, 76.5% of the studies with a quality score ≥ 50%, indicated that the food safety knowledge of food handlers is significantly associated with SFHPs. This finding reinforces the need to improve on the food handlers’ food safety knowledge through training. According to the Knowledge, Attitudes and Practices (KAP) model practices are dependent on knowledge and attitudes [46]. Essentially, the KAP model reinforces that the provision of information (knowledge) contributes to changes in practices. On the other hand, some studies reviewed in this study reported that the food handlers’ food safety knowledge was not significantly associated with SFHPs [13, 25, 30, 38]. This shows that although the evidence of association is strong, there is no complete consensus on the role of food safety knowledge in effecting desirable food safety practices. Therefore, further research in this regard is necessary.

Table 4.

Available evidences for factors associated with SFHPs

Factor	Total studies included	Studies with score ≥ 50 %	Comments	Status of evidence
Gender	11	11	Association reported in 36.6% of studies with quality score ≥ 50%.	Insufficient
Marital status	5	5	Association only reported in one out of the five studies (20%)	Insufficient
Income	2	2	Association reported in both studies.	Insufficient
Age	9	9	Association reported in 44.4% of studies.	Insufficient
Work experience	12	10	Association reported in 40% of the studies and with quality score ≥ 50.	Insufficient
Educational level	15	14	Association reported in 50% of the studies with quality score ≥ 50.	Limited
Food safety Training	12	11	Association reported in 63.6% of the studies with quality score ≥ 50.	Moderate
Food safety knowledge	19	17	Association reported in 76.5% of the studies with a quality score ≥ 50. The score of studies with quality ≥ 50% ranged from 50 to 92%.	Strong
Food safety attitudes	16	14	Association reported in 71.4% of the studies with a quality score ≥ 50. The score of studies with quality ≥ 50% ranged from 50 to 92%.	Strong
Type of food premises	3	3	All the three studies (100%) found a positive association between type of premises (government or private) and safe food handling practices. Their score ranged from 67 to 92%.	Limited
Type of handwashing	1	1	Reported association between type of handwashing facility and SFHPs.	Insufficient
Presence of insects and rodents	1	1	Association reported in the study.	Insufficient
Location	1	1	No association between the location of a food premise and safe food handling practices	Insufficient
Adequate buildings & facilities	2	2	Association reported in both studies.	Insufficient

Food safety attitudes

Several studies (n = 16) investigated food safety attitudes as a determinant of safe food handling practices, of which fourteen were of sufficient methodological quality. The evidence of association was classified as strong, as a significant positive relationship was reported in 71.4% of the studies that had a quality score ≥ 50. This finding emphasises that efforts targeted at improving food handling practices must target the food safety attitudes of food handlers. Unless food handlers fully appreciate the food safety risks associated with their FHPs, chances are high that they won’t change their unsafe food handling [10]. A statistically significant relationship between food safety attitudes and safe food handling practices, was however, not established in some studies [24, 30, 35, 40]. This demonstrates that the evidence of association between food safety attitudes and safe food handling practices is not only strong but debatable. More research is warranted in this dimension. Most of the studies reviewed in the present article used a pre-designed questionnaire to measure the self-reported food safety attitudes and practices of food handlers [8, 11, 31, 35, 36, 39, 41]. Such studies [8, 11, 31, 35, 36, 39, 41] commonly reported that the majority of food handlers expressed positive food safety attitudes that were significantly associated with self-reports of their food safety practices. The use of questionnaires without observation presents some limitations with respect to the assessment of food safety practices. First, self-reported practices are not necessarily the actual practices of food handlers [15]. Second, where completed in the absence of the researcher, respondents have room to look up information or consult colleagues, which may yield misleading and less reliable information about their attitudes and practices. Last, some concerns relate to issues of reliability and validity of the questionnaires (Table 3). Taken together, the limitations of self-reported data seem to underscore the need for observation of the food safety practices. Few studies observed the food safety practices of food handlers (Table 3). In one study [14], trained food handlers had higher knowledge scores but did not significantly differ from the untrained ones regarding the food safety attitudes, self-reported and observed practices. In another study [23], the average scores for self-reported and observed food handling practices had minor differences (80.71 versus 75.40%) Based on these findings, questionnaires seem to yield data not very different from that gathered with the use of an observation checklist. Vo et al. [24] observed inadequacies of canteens with respect to hygiene materials, equipment, facilities and buildings. Since most of the inadequacies are primary responsibility of food safety managers and not food handlers, it is difficult to compare such findings to food handlers’ knowledge and self-reported practices. Finally, although the observation method is essential in addressing the limitations of questionnaires, it has own shortcomings. For example, there is no guarantee that when aware that they are under researcher observation, food handlers won’t change their usual food handling practices to what they perceive as desirable. Therefore, methodological triangulation with regard to the measurement of food safety practices may be useful.

Educational level

Studies investigated the association between a food handler’s educational level and SFHPs. A significant relationship was reported in half of the studies with a quality score ≥ 50%, which yields limited evidence of association in terms of our review criteria. Food handlers with a higher level of education were commonly reported to practice safe food handling practices than those with a lower level [5, 12, 24, 29, 30, 35, 36]. In such studies [5, 12, 24, 29, 30, 35, 36], the educational level of a food handlers is portrayed as a determinant of SFHPs. For instance, in one study [29], the kitchen workers with technical college or higher degrees were reported to have significantly higher: (a) sanitation knowledge with regard to inventory and storage, and preparation and cooking areas, and (b) sanitation practices with respect to equipment and facilities. In three studies [12, 35, 36] the safe food handling practices were reported to be higher among food handlers who had formal education compared to those with no formal education. On the contrary, other studies reported that a food handler’s level of education does not significantly affect his or her use of SFHPs [2, 15, 23, 25, 26, 32, 41]. This finding calls for a rethinking of the practical value and contribution of a food handler’s educational qualifications to food safety enhancement.

Food safety training

Studies on the association between food safety training and SFHPs yielded contradictory results. Seven (63.6%) of the eleven studies reported a statistically significant positive relationship [2, 5, 11, 12, 27, 28, 32]. Resultantly, the evidence of association was classified as moderate. Improving workers’ knowledge through training has been reported to be extremely useful in strengthening positive attitudes and SFHPs [38]. The success of a food safety training programme predominantly depends on the motivation, support and facilities provided by supervisors and management [8]. In addition, food safety training must be hands-on, goal oriented and continuous [8] and hinged on a correct analysis and evaluation of the training needs [4]. Further, care must be taken to ensure that the training initiatives must be carefully planned, utilise a varied mix of training methods and strategies to motivate workers [4]. On the other hand, Park et al. [13], Da Cunha et al. [14], Bou-Mitri et al. [15] and De Souza et al. [23] found that food safety training did not influence SFHPs by food handlers. Therefore, further research needs to be undertaken about training issues such as the course content, site and its duration [45].

Work experience

Ten (10) studies of sufficient methodological quality investigated the association between a food handler’s work experience as a food handler and SFHPs. A statistically significant association was reported in less than 50% of the studies with a score ≥ 50%, which gives insufficient evidence about the association between length of employment as a food handler and safe food handling practices. Their score ranged from 67 to 75%. However, other studies reported that work experience was not significantly associated with SFHPs [23, 25, 29, 30, 40]. It is noteworthy that there was no consistency on the length of work experience investigated, which may account for the differences in reported results. There is need to standardise on the length of work experience that can be used as a criterion to classify a food handler as either experienced or inexperienced.

Socio-demographic factors

The promotion of food safety in the food service sector must be hinged on a clear understanding of the factors that influence a food handler’s decision to use SFHPs. The present review found out that there is insufficient evidence to suggest that socio-demographic factors of food handlers (age, gender, marital status, and income) significantly influence SFHPs. The same applies to other factors such as (a) the location and type of food premises, (b) type of handwashing facility in the food premises and (c) presence of insects and rodents. This is because the association was reported in <50% of the studies with a methodological quality score ≥ 50%. Therefore, less focus should be placed on socio-demographic factors, as the evidence from this review does not support that they significantly influence SFHPs. However, more research is required to determine interactions of socio-demographic factors and other factors with SFHPs, knowledge and attitudes.

Conceptual framework

A conceptual framework of the available evidences for factors associated with SFHPs, based on the findings of the present review, is presented in Fig. 1. According to this framework, factors contribute to SFHPs at varying degrees of influence, but strong evidence of association was with food handlers’ food safety knowledge and positive attitudes.

Fig. 1.

Conceptual framework illustrating the available evidences for factors associated with SFHPs. Arrows in bold: strong association; moderate bold: moderate; non-bold: limited evidence; absence of arrows linking socio-demographic and other factors to safe food handling practices: insufficient evidence

Methodological considerations

The use of a wider search strategy that comprised five databases and a reference search could have improved our chances of identifying most articles that met the inclusion criteria. In addition, data extraction, quality assessment and evidence synthesis for the articles that met the inclusion criteria were performed using pre-designed criteria. This enabled the authors to uniformly rate each study, thus enhance the probability of yielding reliable results. Moreover, this review summarises studies from a wide range of countries and a total population 7342 food handlers. In contrast, the present study is not immune to some methodological limitations. First, the exclusion of studies published prior to 2010 could have impinged on the capacity of this study to identify some relevant conclusions regarding some factors associated with SFHPs. For example, contrary to most studies reviewed in the present study, Pilling and colleagues [47] reported a lack of translation of knowledge from training courses, into positive changes in attitudes and behaviour. Second, other relevant articles published in other databases could have been left out. Third, the consideration of 100 study participants as a cut-point for enough sample size maybe in question as sample size is not always a direct indicator of quality and reviewed studies did not necessarily investigate same outcomes or use same methods of measurement. Therefore, the present study could have unintentionally misclassified some evidence from available studies. Fourth, the quality of studies in this setting may not necessarily have attributes that are designable from other settings, which may limit the capacity to generalise this study’s findings to such contexts. In addition, the methodological quality appraisal tool used in the current study may richly benefit from further validation. Last, the exclusion of articles that did not gather primary data (e.g. review articles, editorials, commentaries) may have limited our opportunities to learn from such studies.

Conclusions

Available evidence support that in order to promote safe food handling practices, responsible authorities need to focus mainly on improving the food safety knowledge and attitudes of food handlers. While the importance of a food handler’s educational level and work experience cannot be underestimated, the evidence of association of the factors with SFHPs is limited and not strong. Future food safety studies may richly benefit from overcoming the methodological shortcomings presented in this review. Further investigation of factors that currently show insufficient evidence of association with SFHPs, such as socio-demographic factors, are required.

Funding information

This study was self-funded.

Compliance with ethical standards

Conflict of interest

The authors declare that they have no conflict of interest.

Ethical considerations

This study received ethics approval from the institutional review board of the author’s university (Ref. 010/2018).