Nutrition knowledge assessment studies in adults: a systematic review

The objective of this study is to perform a systematic review of published studies that assessed nutrition knowledge in adults, focusing on the methodology and content of these studies. An article search was performed on the Medline, Lilacs, and SciELO databases. The search limits were human studies; English, Portuguese, and Spanish languages; and age (over 19). Inclusion criteria were: cross-sectional studies performed on individuals over 18 years old that assessed the general nutrition knowledge of participants. The methodological quality of the articles was assessed using the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) criteria. The initial search identified 3,623 articles. After reading the titles and the abstracts and applying the exclusion criteria, 25 articles were selected. The results showed that, in most studies, nutrition knowledge was associated with socioeconomic parameters and eating behaviour. Most studies belonged to class B (92%), meeting 50-80% of the STROBE criteria. The studies have revealed a greater tendency to assess the relationship of nutrition knowledge with sociodemographic and economic parameters.

introduction Currently, the major causes of mortality worldwide are diseases that could be prevented with proper nutrition, regular physical exercise, and a healthy lifestyle that includes leisure time, stress management, and personal and environmental care 1 ; such diseases are called non-communicable diseases (NCDs). In Brazil, NCDs represent a high burden and contribute to the increased health care costs associated with illnesses such as diabetes, hypertension, obesity, and cardiovascular diseases, among other diseases; the existence of long queues for medical appointments at public health units; an increased need for specialized tests and surgeries; and other additional problems 2 .
In this sense, one aspect that can improve quality of life is increasing the ability of a population to understand health-related phenomena. The knowledge of health outcomes, in particular regarding NCDs and their risk factors, may be useful in avoiding the onset of a disease and can also influence the search for treatment when there is precise information on the established disease 2,3 .
Knowledge may be defined as factual and interpretative information that leads to understanding or that is useful in making decisions or informed actions 4 . In cognitive psychology, 2 types of knowledge can be defined: declarative and processual.
Declarative knowledge is defined as the knowledge of facts and things, the knowledge of 'what is,' for example, knowing that lemons are a good source of vitamin C or that adequate fruit and vegetable consumption may prevent hypertension. Processual knowledge is the knowledge of how actions are undertaken, for example, how to choose which of 2 snacks is healthier or how to eat a balanced diet 5,6 .
Social cognitive theory emphasizes that individual behaviour is determined by the interaction of personal and environmental factors 7 . In this sense, Motta et al. 8 explain that the personal experiences of each individual encompass cognitive assumptions that will constitute values and beliefs that, based on the reality and experiences lived by each one of us, will become meaningful and predict future events; thus, behaviour and its consequences are dependent on both conscious individual choices and events that occur in the environment in which individuals live. Therefore, social influences from the environment, a lack of motivation, low self-esteem, and societal beliefs and traditions are factors that collaborate to inhibit lifestyle changes, particularly regarding food habits 8 .
Based on this premise, educational activities have been developed to disseminate information aimed at changing individual lifestyles, considering that humans build their knowledge day by day based on what they see, hear, feel, and perceive. Humans observe, sort, and select what they consider important for their lives and begin to use what they learn in accordance with their culture. These actions are how humans change themselves and the surrounding world (or resist change) 9 . It is during the performance of their everyday activities that humans improve their actions. According to Paulo Freire 10 : "…what is essential is to do. To dive into an activity and start learning-relearning, creating-recreating…".
With respect to food, knowledge on what should be eaten and the awareness of the importance of healthy food habits are the first steps in altering eating behaviour. However, the relationship between what people know and what they do has been considered "very weak". Knowledge does not stimulate change but instead acts as an important tool when people desire to change, given that knowledge rarely anticipates a behavioural change 11 . From this perspective, nutrition knowledge (NK) may be defined as the individual cognitive process related to information on food and nutrition 12 , and it may have something to do with food selection 13 and success in NCD prevention 14 .
Triches and Giugliani 15 state that individual NK may favour healthy food consumption and thus promote changes in food habits that may reduce the risks of developing NCDs. In a systematic review on the association between NK and dietary intake in adults based on cross-sectional and quasi-experimental, randomised clinical trials, Spronk et al. 16 concluded that most studies present a weak association between NK and food consumption. However, according to the knowledge-attitude-behaviour model, knowledge might not produce positive and significant changes in eating behaviour, and it is essential to have a motivation or stimulus to make change occur 17 .
Considering all of the issue noted above, NK assessment is a relevant topic because the level of NK might be associated with eating behaviour, can promote changes in inadequate food habits, and promotes alterations in nutritional status. The objective of this study is to systematically review the literature on studies assessing the level of nutrition knowledge in adults, focusing on the methodology and content of these studies.

Methods
This study is a systematic literature review. The question this review aimed to answer is the following: 'What methods are being used in cross-sectional observational studies to assess the level of general NK in adults?'

Search strategy
Relevant articles were identified in April 2013 by performing a search on three electronic databases: Medical Literature Library of Medicine (Medline) via PubMed; Scientific Electronic Library (SciELO); and Latin-American and Caribbean Literature (Literatura Latino-Americana e do Caribe -Lilacs) via Virtual Health Library (Biblioteca Virtual em Saúde -BVS). Using PubMed as reference, the article search was conducted using the limits 'humans'; 'age' -applying the three available filters 'Adult: 19+ years, Young Adult: 19-24 years, and Adult:19-44 years'; and 'language' -English, Portuguese, and Spanish. When necessary, these limits were adapted to resources available in other databases.
The terms used for the article search were identified in Health Sciences Descriptors (Descritores em Ciências da Saúde -DECS). The following keywords were used: nutritional status, body mass index, nutrition assessment, food habits, food consumption, knowledge and health knowledge, attitudes, and practice. The search strategy was organized according to the characteristics of each database. The logical operators OR and AND were used to combine the terms used for the article search.

eligibility criteria
Eligible studies met the following criteria: cross-sectional observational study; participants aged > 18 years, regardless of gender, country, ethnicity, and socioeconomic status; and description of the methodology used to assess the level of general NK.
The following were excluded: review articles; case-control, cohort, ecological, and interventional epidemiological studies; studies with children and adolescents; duplicate studies; exclusively qualitative studies; studies on the development, validation, and reproducibility of NK questionnaires; studies that assessed specific aspects of nutrition and food knowledge such as only fruits, vegetables, fat, fibres; and articles that did not address NK.

Article selection
Two authors independently reviewed the title and abstract of the articles identified by the initial search strategy, using the predefined inclusion and exclusion criteria. Disagreements that occurred during the selection step were settled by consensus; if consensus was not reached, more experienced researchers were consulted.
Each study was described based on the following information: year of publication; journal and year in which the study was performed; study location; population; sample size; objective; anthropometric, dietary, socio demographic, and economic variables; the tools used to assess NK, including how the tools were used; classification and assessment criteria; and associations between NK and the analysed variables.

Methodological quality assessment
Eligible articles were reviewed by two authors according to the STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) 18 criteria. As in Mendes et al. 19 , each of the 22 criteria was given a score of 0 (meets the criterion) to 1 (does not meet the criterion). After assessing all of the criteria, each article received a score from 0 to 22 from each reviewer. The final score was obtained by calculating the mean score of both reviewers. The manuscripts were ordinally ranked according to their final score. The global score was converted into a percentage to assess article quality and to classify the articles into three categories, according to Mataratzis et al. 20 : A -study meets more than 80% of the STROBE criteria; B -study meets 50-80% of the STROBE criteria; and C -study meets less than 50% of the STROBE criteria. In case of disagreement, two other researchers were consulted.

results
A total of 3,623 articles were identified. Following the analysis of titles and abstracts and the application of exclusion criteria, 25 studies were selected for inclusion in this systematic review, as shown in Figure 1. Regarding the assessment of methodological quality using the STROBE 18 criteria, none of the studies analysed was classified as quality C; 8% (n = 2) and 92% (n = 23) were classified as quality A and B, respectively (Table 1).
The articles reviewed in this study are presented in Charts 1 and 2 by study year/location, sample description, analysed variables, objectives, summary of the tools used for NK assessment and classification, and associations with NK.

Study year/location
Articles in this area began to be published in 2000 and continued in low numbers until 2003 (a mean number of 1 article per year), returning in 2005 with the studies by Lin and Lee 28 and Schaller and James 32 . From this year on, the mean number of published articles increased to three articles/year. The studies were performed from 1999 to 2015. However, 36% of the authors (n = 9) did not report the year of the study. Interestingly, the study by Mimiran et al. 43 was performed in 1999, but there is a difference of 11 years between the year in which the study was performed and the year of publication.
The journals with the highest number of published articles were Asia Pacific Journal of Clinical Nutrition and Journal of The American Dietetic Association, with three articles each (12%).
NK assessment studies were performed on five continents. Brazil is the country with the highest number of studies (n = 4; 16%), followed by the US and Australia, each with 12% (n = 3) of the articles. Included studies: n = 25 Exclusions: -Excluded for being a duplicate: n = 24 -Excluded for being a systematic review: n = 1 -Excluded due to type of study and because it did not address nutritional knowledge: n = 59 Full-text analysis: n = 53 Exclusions: -Excluded due to not addressing general nutritional knowledge, age group, and type of study: n = 31 Selected studies: n = 22 Selected and included based on the reference list: n = 3

Sample description
In most studies, the population was composed of adults of both genders, aged from 18 to 75 years old; the exceptions are the studies by Zawila et al. 35 ,Freitas et al. 26 ,Pon et al. 25 ,Holdsworth et al. 33 ,Nuss et al. 23 ,De Vriendt et al. 40 ,Castro et al. 30 , , which were performed with women only. These studies correspond to 36% (n = 9) of the reviewed articles.
The smallest sample size was of 21, and the largest sample was composed of 2,027 participants. It is noteworthy that these samples were obtained almost entirely by convenience.

Analysed variables and objectives
Most studies did not analyse dietary, anthropometric, socioeconomic, and demographic variables simultaneously; only 32% of the studies did so 25,26,29,[39][40][41][42][43] . However, sociodemographic and economic data were assessed by all studies. Of these, 25% (n = 6) assessed only these data, with the most frequent variables analysed being age, gender, educational level, occupation, income, marital status, number of children, and ethnicity. Regarding nutritional assessment, the most studied variables were as follows: weight, height, body mass index (BMI), waist circumference (WC) and hip circumference (HC), and food consumption. BMI was assessed in all articles that analysed anthropometric data (n = 14). However, WC and HC were analysed simultaneously only by Pon et al. 25 and Mirmiran et al. 43 .
In general, the objective of the studies was to describe and/or assess NK and to correlate it with nutritional status, eating behaviour, or economic and sociodemographic variables. Although, in 40% of the studies 25,26,[31][32][33][35][36][37][38]43 , the authors did not state in the article that one of the objectives was to test the existence of associations, most of them presented data on this matter in the results section.

Nutrition knowledge assessment tools
A total of 44% (n = 11) of the studies used questionnaires specifically designed for their own investigations 6,23,25,28,31,33,39,[41][42][43] . In general, these assessment questionnaires addressed nutrients (carbohydrates, lipids, proteins, vitamins, minerals, and fibres); the health benefits of food; the diet-disease relationship (obesity, cardiovascular disease, and non-communicable diseases in general); fruits and vegetables; salt; and the food pyramid. The Nutritional Knowledge Questionnaire (NKQ) 45 was used by 32% of the studies (n = 8); The NKQ was developed and validated for use in the English-speaking population, aged 18 or over, and assesses knowledge on dietary recommendations, nutrient sources, daily food choices, and the diet-disease relationship. The Harnack et al. 46 NK scale, translated into and validated for Portuguese by Scagliusi et al. 13 , was used in 16% (n = 4) of the articles 21,26,27,30 ; this scale assesses knowledge on the diet-disease relationship, the amount of fibres and lipids in food, and the recommendations of fruit and vegetable consumption. Schaller and James 32 studied nurses and used the questionnaire designed by Henderson Sabry et al. 47 specifically for these professionals. Gender, educational level, life conditions, and eating arrangements. Diet using FFQ.

Objective
To assess the possible relationship between NK and the intake of fruits, vegetables, and fat To analyse NK and its demographic variations To assess NK and attitudes related to food intake and nutrition in marathon runners To understand NK, attitudes related to nutrition, and attitudes and behaviours related to food restriction in elderly Taiwanese To determine NK among nurses To assess health and NS in urban middleaged women and to study the relationship between NK and health To assess knowledge on dietary factors and behaviours related to non-communicable disease.
To assess NK and food intake frequency among vegetarian and non-vegetarian female university students To investigate food habits of university students and their relationship with knowledge of dietary guidance.

To investigate the correlation between NK level and BMI in individuals residing in urban areas
To assess NK and its effect on weight retention in women 1-year postpartum To investigate general NK and its demographic variations To assess NK and to correlate NK with BMI and educational level To examine the relationship between NK and food intake in university students Age, gender and residential area (geographical area).
Educational level, marital status, occupation, income, place of birth, language spoken at home, children and mother's date of birth. Prepregnancy weight and height (self-reported), and BMI. Diet using FFQ. Age, gender, educational level and children/no children at home.

Objective
To assess NK, the role of sociodemographic and lifestyle variables in NK, and the association between NK and eating behaviour To assess NK in judo athletes and to identify the factors that can influence NK To describe attitudes, behaviours, and NK; to identify obstacles to healthy lifestyle changes and the need for educational intervention To assess NK in physically active women and to correlate it with educational level and anthropometric characteristics To assess knowledge and beliefs related to nutrition and food habits To assess the prevalence of misconceptions on how to have a healthy diet and correlate these misconceptions with eating behaviour To identify NK, its association with overweight or obesity, and sociodemographic conditions and characteristics To assess NK and frequency of food consumption To investigate NK, attitudes, and behaviour in adults (19-64 years) To assess NK and its association with socioeconomic status and diet quality To assess knowledge on characteristics of foods that affect health/well-being and its relationship with food labelling * Medline, Lilacs and SciELO; ** Sociodemographic and economic, anthropometric, and dietary, respectively.   Kolodinsky et al. 29 used the questionnaire based on the MyPyramid Food Guidance System 48 and on the Diet and Heath Knowledge Survey from the US Department of Agriculture 49 ; it addresses knowledge on energy, total fat, saturated and trans, added sugars, fibres, and the importance of keeping a healthy weight and of eating different fruits and vegetables.
How the answers to the questionnaires were assessed is not stated in 44% (n = 11) of the articles. In 28% (n = 7), only the maximum score that could be reached if all questions were answered correctly was given 22,32,36,38,40,42,44 . The remaining 28% classified NK as low, poor, or weak (0-9 points); moderate or medium (7 to 10 points); and high (above 10). These scores varied according to the methodology used by the authors, with different points in the classification scores.

NK Associations
Associations between NK and socioeconomic and demographic variables (age, gender, educational level, and family income) were identified in 64% of the studies. However, Holdsworth et al. 33 did not find a relationship between NK and education. By contrast, BMI, which was assessed in 56% of the studies (n = 14), was associated with NK in 42.8% of the studies, and such association could be positive 27,41 , negative 40 , and not present 22,33 ;in Nuss et al. 23 , women with less than 5% of weight retention (compared to pre-pregnancy weight, assessed by BMI) during the first year postpartum had better NK.
Associations were found between NK and several positive aspects of diet (n = 10): healthy eating 29,34 ; increased fruit and vegetable consumption 6,36,40 ; and a low intake of simple sugars, fat, and salt, among others 28,36 .
In light of the results presented in Charts 1 and 2, some conclusions, according to the population studied, were: for women, the most important NK determinants were educational level, age, and occupation 40 ; nutritional education is important for the elderly, and the development of nutritional education programmes must consider the low educational levels of elderly people 28 ; NK is an important target for health education and has the potential to contribute to a better dietary quality 34 ; the knowledge of food, nutritional properties, and recommendations on the size and frequency of consumption must be the primary goals of nutritional education programmes throughout the life of individuals 36 .

Discussion
This investigation is the first systematic review on cross-sectional studies that assessed the level of general NK in adults; however, similarly to Sarno et al. 50 , it was not possible to perform a meta-analysis due to the heterogeneity of the populations, methods, and tools used in these studies.
The identification of 25 articles that met the inclusion criteria demonstrates that NK assessment has been gaining attention among researchers worldwide. Although no year of publication filter has been used to include all of the studies on the topic of interest, publications in this field began recently (year 2000), and the number of articles/year has been increasing since 2005, reaching a mean number of three articles/ year, which also demonstrates the current importance of this topic. According to Verbeek et al. 51 , the number of published articles can be used to quantify scientific progress and evolution.
In the assessment of the STROBE criteria, all of the selected articles presented percentages higher than 50%, which corresponded to a classification of A or B. In the studies by Mirmiram et al. 43 and Kresić et al. 44 , which scored 80% and belonged to the STROBE A category, NK was associated with age (p < 0.01), educational level (p < 0.01), and gender (p < 0.05) and with the consumption of food servings in accordance with that recommended by the food pyramid (p < 0.01), respectively.
The fact that studies were performed in countries from five continents; that sample size varied widely among the studies that almost no articles assesses dietary, anthropometric, sociodemographic, and economic variables simultaneously; and that the methods used were different limits our discussion of the results by making comparisons between studies more difficult. Silveira and Santos 52 and Barros et al. 53 have also noted the difficulty of not being able to establish comparisons between results that use different methods. In addition, each article cites its own limitations, which include the following: the use of NK questionnaire not validated for that specific population 35,41 ; self-reported data 29,42 ; the non-validation of NK questionnaire scoring methods 42 ; the sampling technique used 22,33 ; the fact that some participants participated in health and nutrition educational programmes 27; ; and the sample size -the use of a small sample 27,29 ; regarding this last limitation, Schaller and James 32 state that, although the sample size was sufficient to estimate NK prevalence with proper statistic power, it was inadequate to perform regression analysis. Thus, the limitations of this review, together with those presented by each article, have hampered the compilation and comparison of results; thus, it is important to perform studies on this subject that use tools and methods that are validated and/or adequate to each study.
The analysis of the sociodemographic and economic variables performed by all of the authors indicates that these phenomena are an important and relevant aspect for NK and the general well-being of participants. One study performed in three low-income neighbourhoods of Fortaleza (state of Ceará, Brazil), conducted by the World Bank, assessed the social risk of poverty and found that the risk factors were low educational and social capital levels, family breakdown, and early pregnancy, among others 54 . Poverty exists when part of a population cannot make a sufficient income to have sustainable access to the basic resources (water, health, education, food, housing, income, and citizenship) that ensure a good quality of life. Low socioeconomic status interferes with healthcare and food habits because, by affecting the family structure, life and food quality are also affected 54-56 .
Regarding the tools used for NK data collection, a wide range of tools was used; however, most authors (44%) used a questionnaire designed specifically for their study 6,23,25,28,31,33,35,39,41,43 . Researchers design their own questionnaires to include items relevant to their specific study by considering the characteristics of the population being studied 12,57 . However, if there is already a data collection tool that is appropriate for a particular population, then there is no need to develop another tool (to be applied to another population), provided that the psychometric properties have been reassessed and are considered reliable 57 . Nevertheless, even when different NK assessment questionnaires are used, the content of these questionnaires does not diverge much, and all of them provide a general view of the participants' NK.
Of all the studies that included the cut-off used for NK classification (Chart 2), only those using the NK scale of Harnack et al. 46 , translated into and validated for Portuguese by Scagliusi et al. 13 , have used the same cut-off for NK assessment 21,26,27,30 , which indicates that the dynamics of NK are not well understood.
Regarding the associations between the analysed variables, NK was shown to be asso-ciated with age, income, and educational level 23,24,28,31,32,37,38,40,42,43 . Nicastro et al. 58 ,Obayashi et al. 59 , and Sapp and Jensen 60 also showed a positive association between educational level and the NK questionnaire score, which clearly shows that education is a basic tool for obtaining knowledge on nutrition. According to Zawila et al. 35 , the greater NK observed in females supports the hypothesis that women are more concerned with aesthetic aspects, which leads them to search more for food-related information.
The studies that investigated a possible association with anthropometric data (Chart 2) show different results regarding BMI, and do not suggest a protector effect against increased body mass. Even when the evidence shows that NK may influence food choices, promote the adoption of healthy food habits, and, consequently, prevent an increase in body mass 13,30 , this association is not universal. It is not clear whether NK leads to healthy food habits 13 . According to Dattilo et al. 27 , the available literature on NK and the nutritional status of a population is very limited, and few studies have used dully validated methods.
Regarding dietary aspects, greater NK was generally associated with healthy food choices and habits. However, even if NK might be a factor that improves eating behaviour, it is not the only determinant factor. For example, according to Montero Bravo et al. 61 , a study performed on 105 college students studying health-related subjects (including nutrition) revealed that, even when students thought they had good NK, the NK was not compatible with their food habits. This study demonstrates the complex nature of this research area and explains why conclusions from similar studies are in apparent disagreement; finally, it shows that further studies are necessary to assess the relationship between NK and eating behaviour, particularly due to the limitations observed in some of the articles reviewed in this study.

Conclusion
Studies assessing the level of nutrition knowledge in adults remain rare and use different assessment tools and scores, in addition to different variables, objectives, and sample sizes, hampering the analysis of its determinant factors. There is a greater tendency to assess the relationship between NK and sociodemographic and economic variables than to assess the relationship between NK and anthropometric and dietary data, including food habits and choices. Most studies here reviewed present a methodological quality classified as B according to the STROBE criteria, and only few are classified as A, which is compatible with the observed profile.

Collaborations
LB Barbosa assisted with the study design, database search, and manuscript writing; LOS Correia assisted with the study design and manuscript writing and revision; RC Ferreira participated in the database search and manuscript revision; and SML Vasconcelos assisted with the study design and manuscript writing and final revision.