Human Factor in Food Label Design to Support Consumer Healthcare and Safety: A Systematic Literature Review

: Labels play a strategic role in communication, representing the interface between consumers and the food system. Consumers’ ability to correctly understand product label information was essential for health and safety, making a sustainable choice, as also demonstrated by the Federal Regulations of USA, which deﬁne several guidelines to increase the readability of labels. Human Factor studies the interface design in order to assure users’ safety, comfort, and productiveness. So, it could be applied in label design to enhance consumers’ safeguard and healthcare. The aim of this study is to analyze the ﬁelds of application of the Human Factor in label design to evaluate the current methods of utilization in the food industry. A two-step literature review was applied, using bibliometric and content analysis. From bibliometric analysis, 6 clusters of themes and the predominance of medical terms arose. The absence of terms related to the food domain is surprising. So, a content analysis on lead papers’ sample related to the Food Industry was carried out to identify evidence about the human factor in food label design. From content analysis, 2 groups of works’ results, focused on the possibility to consider the human factor in: i) pharmaceutical label design, ii) food label design, respectively. The studies in food label design primarily focused on label readability related to content positioning and not on its understandability. The practical implications of the study include the construction of a new concept of label in terms of contents and communication channels.


Introduction
Consumers' ability to correctly understand information shown on product labels can be essential for health and safety [1] or to simply make an informed choice regarding sustainability, respective of personal principles and values [2][3][4]. Indeed, informed decision-making depends on people's ability "to accurately evaluate and understand information about risk" related to a product or a situation [5].
Although this statement can be valid for many types of products, in some fields, such as food and drugs, it assumes a stronger value, leading to the necessity of building safety into labels design [6].
In recent decades, the consumers' need to have detailed information about food ingredients has grown, because of allergies, dietary restrictions, or drug interactions [7]. As many as 98,000 people die each year due to preventable medical errors such as the selection of the wrong drug [8,9]. This evidence highlights that human error reduction in healthcare is still an urgent priority [10,11].
For this purpose, a good drug label design could be based on expertise and sound human-centred design principles [12]. Three aims can be attributed to labelling of food products [13]: (1) information Is "human factor" applied to "label" design in the current scientific scenario? What are the fields in which these concepts have been most explored? How do these concepts find application in the Food Industry? Are they used to assure consumers healthcare?
In order to give an answer to the abovementioned research questions, the literature review method was chosen. The first step considered a bibliometric analysis, which was used to identify the fields of Human Factor application, followed by a systematic content analysis to identify how this concept could be applied in the food industry.

Materials and Methods
In order to provide answers to the aforementioned research questions, a literature review method was chosen, highlighting the boundaries of knowledge [38]. A systematic literature review gains recognition as a valid and important approach to "construct evidence from evidence" [39] and, as proposed by [40], it needs an iterative process that encompasses the definition of appropriate search keywords, the literature search and its analysis, so, following a structured methodology is recommended [41]. Considering this approach and following PRISMA guidelines [42], a 3-step methodology for data collection and analysis was used to identify how Human Factor is applied to label design (Figure 1). The steps are as follows: • defining search scheme and sample definition; • initial statistical analysis on sample; • data analysis. In order to provide answers to the aforementioned research questions, a literature review method was chosen, highlighting the boundaries of knowledge [38]. A systematic literature review gains recognition as a valid and important approach to "construct evidence from evidence" [39] and, as proposed by [40], it needs an iterative process that encompasses the definition of appropriate search keywords, the literature search and its analysis, so, following a structured methodology is recommended [41]. Considering this approach and following PRISMA guidelines [42], a 3-step methodology for data collection and analysis was used to identify how Human Factor is applied to label design ( Figure 1). The steps are as follows:


defining search scheme and sample definition;  initial statistical analysis on sample;  data analysis.
Particularly, the used literature review process was iterative. Indeed, from phase 3 after bibliometric analysis, results lead us to focusing the analysis on a subsample of paper, returning to phase 1 and applying the selection criterion in Table 3.

Defining the Search Scheme and Initial Search Results
Keeping in mind the objective of this paper, keywords that fully cover the theme in exam were selected: "Human Factor" and "Label." Only a combination was used: "Human Factor" AND "Label." The scope is to identify all works coming from literature that treat the two items at the same time, showing or not, any links between them.
The terms were searched into in the following indexed electronic scientific databases:  Web of Science, published by Thomson Reuters, is considered the most important source of scientific data [43]; and  Scopus, managed by Elsevier publishing, more comprehensive than the first one that encloses only ISI-indexed journal [44] and recommended as a good source [45].
The paper research in the database took place in October 2018. Particularly, the search in the Web of Science was conducted in "Title" and "Topic," choosing "all database," meanwhile the research in  Particularly, the used literature review process was iterative. Indeed, from phase 3 after bibliometric analysis, results lead us to focusing the analysis on a subsample of paper, returning to phase 1 and applying the selection criterion in Table 1.

First
The paper must contain the word "food." Because the objective of the study is focused on the Food Industry.

Second
The paper must treat the issue related to label as a communication tool in any industry.
Because good approaches are already applied in other industries, could be transferred to the Food Industry for food label design.

Third
Availability of document in English language. Because we had to be able to evaluate its contents.

Defining the Search Scheme and Initial Search Results
Keeping in mind the objective of this paper, keywords that fully cover the theme in exam were selected: "Human Factor" and "Label." Only a combination was used: "Human Factor" AND "Label." The scope is to identify all works coming from literature that treat the two items at the same time, showing or not, any links between them.
The terms were searched into in the following indexed electronic scientific databases: • Web of Science, published by Thomson Reuters, is considered the most important source of scientific data [43]; and • Scopus, managed by Elsevier publishing, more comprehensive than the first one that encloses only ISI-indexed journal [44] and recommended as a good source [45].
The paper research in the database took place in October 2018. Particularly, the search in the Web of Science was conducted in "Title" and "Topic," choosing "all database," meanwhile the research in Scopus was conducted in "Article title, Abstract, Keywords" following the standard setting of database. In order to identify all works about Human Factor and Label, any temporal restriction was set.
The search results were exported in CSV and txt formats, storing all needed paper information: title, authors' names, authors' affiliations, abstract, keywords, and references. Scopus was conducted in "Article title, Abstract, Keywords" following the standard setting of database.

Initial Data Statistics
In order to identify all works about Human Factor and Label, any temporal restriction was set. An initial sample of 219 papers was identified, excluding duplicates (Web of Science data-base: 50; Scopus database: 189; Duplicates: 20).
The search results were exported in CSV and txt formats, storing all needed paper information: title, authors' names, authors' affiliations, abstract, keywords, and references.  The first result that arose from analysis of the publishing trend is the consistent increase of publications on human factor and label from 2000 till now. This result confirms the major attention given by the scientific panorama to the association between human factor and label issues, probably as a result of the increase, in the same historical period, of general awareness about health, safety, and sustainability issues [46][47][48]. The first statistics on the initial sample of works show that 136 journals have contributed over time to the publication of 219 papers. It was found that 10 journals have published 83 of these identified articles, representing approximately 38% of all papers published. Figure 3 shows the journals in which these papers appeared.

Initial Data Statistics
Moreover, starting from the information available on indexed electronic scientific databases, the sample was analysed in order to identify the research areas of the works.
Thirty-nine categories of research area arose. Through an aggregation process that regrouped the research areas in the same branch of knowledge, 10 macro-categories were identified (for example: haematology, immunology, paediatrics, oncology, neuroscience, neurology were grouped in the Medicine macro-category). Each paper can treat themes related to more categories as shown in Table 1. The first result that arose from analysis of the publishing trend is the consistent increase of publications on human factor and label from 2000 till now. This result confirms the major attention given by the scientific panorama to the association between human factor and label issues, probably as a result of the increase, in the same historical period, of general awareness about health, safety, and sustainability issues [46][47][48]. The first statistics on the initial sample of works show that 136 journals have contributed over time to the publication of 219 papers. It was found that 10 journals have published 83 of these identified articles, representing approximately 38% of all papers published. Figure 3 shows the journals in which these papers appeared.

Data Analysis
Data analysis is conducted in two consecutive parts including a "bibliometric analysis" and a "systematic content analysis." Bibliometric analysis is becoming a fundamental methodology for analysing research [49]. This kind of analysis has been widely used in the past (e.g., [50,51]) and has been used in many disciplines to better understand the nature of scientific activity [51][52][53] and it has also been applied to Human Factor studies [54].
So, in this work, bibliometric analysis was carried out in order to recognize the most recurrent themes in the Human Factor and Label fields, their relations, and the timing by which these themes are treated. For this scope, a network analysis and graphical investigation on text data was realized in order to create a term co-occurrence map using VOSviewer software [55][56][57]. Unlike most computer programs that are used for bibliometric analysis and mapping, VOSviewer is more focused on the graphical representation of bibliometric maps, functionally useful for displaying large maps in an easy-to-interpret way [58].
Bibliometric analysis was chosen because, when performed at the macrolevel, it is able to supply a general assessment of a research field as a whole, thus obtaining an answer to the following research question: What are the fields in which these concepts have been most explored?
The results obtained from bibliometric analysis were surprising, since the food field does not appear among the other ones. So, a content analysis was carried out iterating the literature review process on a lead subsample of papers to investigate how Human Factor is applied in food label design and if it was used to assure consumers' healthcare in food eating. The content analysis is a qualitative method for analysing the meaning of a study, allowing researchers to describe the topics and themes that are most meaningful to the research objectives of the study [59]. Moreover, starting from the information available on indexed electronic scientific databases, the sample was analysed in order to identify the research areas of the works.
Thirty-nine categories of research area arose. Through an aggregation process that regrouped the research areas in the same branch of knowledge, 10 macro-categories were identified (for example: haematology, immunology, paediatrics, oncology, neuroscience, neurology were grouped in the Medicine macro-category). Each paper can treat themes related to more categories as shown in Table 2. Medicine and Engineering were the fields in which the relationship between "label" and "human factor" was more evaluated. Starting from the objective of the study focused on the food industry, it is interesting to underline that many categories related to the food system emerged, except for the category "Agricultural and Biological Science," encompassed into the "Environmental Science" macro-category.

Data Analysis
Data analysis is conducted in two consecutive parts including a "bibliometric analysis" and a "systematic content analysis." Bibliometric analysis is becoming a fundamental methodology for analysing research [49]. This kind of analysis has been widely used in the past (e.g., [50,51]) and has been used in many disciplines to better understand the nature of scientific activity [51][52][53] and it has also been applied to Human Factor studies [54].
So, in this work, bibliometric analysis was carried out in order to recognize the most recurrent themes in the Human Factor and Label fields, their relations, and the timing by which these themes are treated. For this scope, a network analysis and graphical investigation on text data was realized in order to create a term co-occurrence map using VOSviewer software [55][56][57]. Unlike most computer programs that are used for bibliometric analysis and mapping, VOSviewer is more focused on the graphical representation of bibliometric maps, functionally useful for displaying large maps in an easy-to-interpret way [58].
Bibliometric analysis was chosen because, when performed at the macrolevel, it is able to supply a general assessment of a research field as a whole, thus obtaining an answer to the following research question: What are the fields in which these concepts have been most explored?
The results obtained from bibliometric analysis were surprising, since the food field does not appear among the other ones. So, a content analysis was carried out iterating the literature review process on a lead subsample of papers to investigate how Human Factor is applied in food label design and if it was used to assure consumers' healthcare in food eating. The content analysis is a qualitative method for analysing the meaning of a study, allowing researchers to describe the topics and themes that are most meaningful to the research objectives of the study [59].
Following the approach proposed by ref. [60], a subsample composed by lead papers was defined and carefully analysed from a contents point of view. The criterion used to define the lead papers subsample are shown in Table 1. The subsample of analysis is composed by 7 works, as shown in Table 3.

Results
In the following subsections, results coming from the two different analyses conducted, Bibliometric analysis and Systematic content analysis, are shown.

From Bibliometric Analysis
The sample of 219 papers was analysed in order to identify the research areas. Bibliometric analysis was carried out in order to count the recurring terms in "Title" and "Abstract" and create a term co-occurrence map using Network analysis and graphical investigation. The minimal recurring term frequency chosen is 2. So, according to the full counting method, 622 terms recurred, but only 117 recurred the last 2 times. The term list was cleaned from numbers, common terms ("what," "fact," etc.), articles, verbs, conjunctions and other terms with low meaning, leaving active 113 terms. The network was composed of 6 clusters of themes, each one indicated by a different colour. Figure 4 encompasses the 6 clusters, the evolution over time of the general terms network and the focus of Human Factor node using the related image coming from VOSviewer. leaving active 113 terms. The network was composed of 6 clusters of themes, each one indicated by a different colour. Figure 4 encompasses the 6 clusters, the evolution over time of the general terms network and the focus of Human Factor node using the related image coming from VOSviewer. Cluster 1 (red colour) was composed of 34 items referred to as the patient sphere, such as "patient," "prevention," "treatment," "trial," "safety," "human CL RHFVIII," etc.
Cluster 6 (light blue) was composed of 9 items referred to as the medical studies sphere, such as "study," "use," "risk," "mortality rate," "cardiac surgery," etc. Figure 4 shows how these themes have evolved over time, where the blue nodes have been treated for a long time (since 2000) and the green and yellow ones are more recent (respectively, since 2010 and 2015).
It is interesting that all clusters denote a medical print but are never related to the Food Industry. Moreover, even if the term "label" was used as a selection keyword, it does not compare in any cluster. However, "human factor," present in Cluster 3, was related to the following concepts: "performance," "company," "success," "district," "process," "factor," "study," "mortality rate," "efficacy" (Figure 3).
Analysing this concept from a time point of view, it arose that "human factor" compares in the scientific scenario from 2012, underlying the novelty of the issue.

From Systematic Content Aanalysis
In order to identify how the relationship between "label" and "human factor" was treated in the Food Industry, the papers in Table 3 were analysed using systematic content analysis.
Ref. [8] propose a study focused on medications label design considering concepts related to Human Factor. The word "Food" arose only in relation to the regulation "US Food and Drug Administration (FDA)" that establishes the guidelines for labelling and packaging in the related industries. The authors proposing usability and validation tests, respectively on a sample composed of 8 and 25 people. Thanks to the information from the experiment, a new label with a high level of readability was defined. Also, ref. [61] consider the Human Factor in labelling of injectable medications in order to understand if bad readiness is the cause of the high number of in-hospital medication errors. In a survey conducted on doctors and medications, it arose that many medication labels do not comply with the Canadian Food and Drug Regulations for labelling. Also, ref. [62] focus on the same issue about medical errors related to naming, labelling, and packaging of pharmaceuticals, discussing how the medication-use system and drugs packaging do not consider the human factor.
Ref. [63,64] are similar studies where the authors analyse the velocity of the consumer to identify the ingredient list depending on its position on the label. Several hypotheses of influencing factors from literature are considered and several layouts were created for the same food product. Particularly in the layouts analysed were 4 and 30, for the first and second one, respectively. It arose that there is a correlation between the speed in ingredients list recognition and: i) its positioning in the label (top, bottom, right, etc.); ii) its adjacency with other elements (nutritional facts panel and other). Moreover, in ref. [63], authors study if positioning in label layout, items with semantic relatedness increase the readiness of label. Twelve label items were identified and, using a survey method, its semantic relatedness was identified. So, 3 layouts with similarity, dissimilarity, and random positioning were designed and administered to a sample of people. The results showed that search task performance improved according to semantic relatedness.
Ref. [65] investigate the relationship between nutrition label format and decision-making of consumers. The study was based on the theoretical principle of the Proximity Compatibility Principle (PCP). Three experiments were realized to present nutrient data with different levels of proximity in the design of label format. It arose that the decision performance in the context of nutrition labels was better supported when the visual and cognitive processing requirements of decision-making were matched.

Discussion and Conclusions
The study starts from the assumption that labels represent the interface between the food system and consumers, and it could be designed according to Human Factor principles to assure safety and health. The results coming from the literature review were presented above and demonstrate how Human Factor was applied to label design. The following subsections discuss the results of the work and propose the conclusions of the study, including implications and potential applications, limitations, and future trends, to the best of our knowledge.

Discussion from Results
The first issue worthy of debate emerging from the bibliometric analysis is the predominance of terms coming from the medical world. This is also confirmed by the initial statistical analysis where the Medicine and Engineering fields were the ones that treated the relationship between "label" and "human factor" more. However, it is necessary to underlie the ambiguity of the meaning that words like "label" and "human factor" have in several contexts. In fact, even if in this work the "label" is understood as the physical interface between product and user/consumer, the term has been given a different meaning, particularly in the medical and pharmaceutical spheres (i.e., "open-label," "off-label"). In the same way, the terms "factor" and "human," not connected to each other, take on a more different meaning, most of which stem from medical and biomedical spheres (i.e., "human CL RHFVIII"). This could be the reason why there are many works from the medical field. Moreover, the term "human factor" appears in Cluster 3, with other terms referred to the organizational system. It comes related to concepts like "performance," "company," "success," "district," "process," "factor," "study," "mortality rate," "efficacy." Starting from this point of view, a second opportunity of debate emerges. Each term becomes more meaningful if located, another time, in the medical sphere. In fact, regarding the health organization, the terms deriving from the analysis could refer to an organizational system in which a multiplicity of different components (personnel, technology, environment) interacts in complex ways, always leaving the people (patients) at the centre of the system. The term "mortality rate," tied with "human factor," leads to a clinical risk management issue.
From systematic content analysis on the 7 papers that satisfy the selection criterion, two groups of works arose. The first one focused on the possibility to consider the human factor in pharmaceutical label design, in order to decrease the medical errors and preserve the patient's health. Even if these studies do not focus on food products, they could be considered because a high similarity between food and drugs products exists. Almost two common points could be identified: i) both kinds of products can be swallowed; ii) both products generate positive or negative effects on the human body, thus affecting consumers' healthcare. Therefore, it is not surprising that the U.S. Food and Drug Administration (FDA) treats the regulation about food and drug combinedly.
The second group of papers focused on the possibility to consider the human factor in food label design. All works study several layouts where the same contents are positioned in different ways in order to highlight several scopes, such as: increasing the recognition speed of content in a label, improving the label readiness, fostering the product decision-making. It arose that many studies show how much better the information on the label items represents the content; for example, the use of a picture, graph, or text to represent the same concept. Moreover, the label items were identified starting from the current food labels (e.g., list of ingredients, batch number, choking hazard, company logo, expiration date, nutrition fact, product logo, and so on) and no study identified the label items starting from the real needs of consumers about food product.
Effectively, even if an item is positioned for a faster recognition or a label is designed with a high level of readiness, it does not mean that the information is really interesting for the final consumers or really understandable.
In summary, what can be drawn from this research is:

•
The "human factor" is applied to "label" design in the current scientific scenario.

•
The main fields in which the relationship between "human factor" and "label" was more treated are Medicine and Engineering.

•
The concept of "human factor" finds application in food label design as a strategy to increase the recognition speed of contents in a label, improve the label readiness, foster the product decision-making. These strategies could have an indirect effect on consumers' healthcare but this cannot be considered as assured.

Implications and Potential Application
From the study, it arose that the Human Factor is widely applied in the Medicine domain to improve the design of pharmaceutical labels in order to decrease medical errors, thus preserving the patient's health. Starting from the similarities between Drugs and Food products, discussed above, the connection between the two domains exists and is also underlined by FDA regulations. So, the results from drugs labelling design could be transferred to obtain the same objective in food labelling design: assuring the consumers health and safety.
The application of the Human Factor in label design, in particular in the food industry, shows some practical implications. Modern consumers are increasingly interested in health attributes and sustainability (environmental, economic, and ethical). It follows that a type of label, consisting of this information, needs to be presented to consumers, and can be applied not only in the medical and pharmaceutical fields but also in the agri-food sector. In this field, it will be possible to review the concept of labels in terms of contents and communication channels, considering also smart labelling. For reasons of space, current packaging and labels do not allow the consumers to be provided with Supplementary Information (carbon footprint, water footprint, company certifications, health claim, fair-trade, ecolabel, transformation product mode) any more than the mandatory one (Statement of Identity, Net Contents, Nutrition Facts Panel, Ingredient Statement, Statement/Warning on Allergen, Country of Origin, Name and Place of Business).
This limit would be surpassed by smart labels that could contain all the history related to the product. Furthermore, the use of a label that considers the principles of the Human Factor shows implications of public health and food safety. The possibility of communicating in detail the inputs and the technology used in the production process, the method of conservation, ingredients and allergens, the correct storage and use of the product, offers to the user a useful tool for reducing the health risk. For example, phenomena of chemical and microbiological alterations, with consequent damages to human health, can be avoided with correct information on cooking and storage methods.

Limitations and Future Trend
Despite the fact that the 3-step methodology was well established and two databases were considered in order to amplify the research spectrum, as widely justified in the Method section (Section 2), discrepancies or shortcomings should be noted. These derive from the different research standard settings in the databases, which for Web of Science were "Title" and "Topic," while for Scopus was "Article title, abstract, keywords." Even if "Topic" and "keywords" could be assimilable, the field "Abstract" in the first database does not compare, excluding, probably, several published works. However, this discrepancy does not exist in data analysis, where the term co-occurrence map was created, starting from the analysis of same fields for all works in the sample (Title and Abstract).
The Medicine field turned out to be the first field where human factor was applied in drug label design. So, it could be interesting to identify the best practices in this design and to study how these could be transferred in food label design.
As discussed above, all works in the food industry analyse several layouts in order to investigate how to better position the contents in a label. None of the work focussed on how to present the