Patterns of Avoiding Nutrition Mistakes in Metropolitan Adolescents Are Associated with Sex, Nutrition Knowledge, Physical Activity, and Family Environment

A comprehensive approach to the identification of the relationship between behaviors limiting nutrition mistakes, nutrition knowledge, and physical activity in the context of the family environment has not yet been widely explored. We aim to identify patterns of avoiding nutrition mistakes in high school students from Warsaw, Poland, and to assess their associations with nutrition knowledge (NK), physical activity (PA), body mass index (BMI), demographic, and family environment characteristics. A cross-sectional study involving 616 high school students, aged 16–19, was conducted. The data were collected by distributing questionnaires. The k-means method was used for cluster analysis, and logistic regression was used to assess the adherence to identified patterns. We identified three patterns: Prudent Ones (45%), Inconsequent (39%), and Rebels (16%). About 70% of adolescents had insufficient NK. The adherence to the Rebels pattern was lower by 85 % in girls, by 68% in students with younger siblings, and was about 4.0-times higher in children of mothers with primary education, 2.4 times higher in students with insufficient NK, and 1.9-times higher in students living in a family with more than 4 persons. The groups of adolescents with feature characteristics of the Rebels and Inconsequent are possible targets for intervention and require further in-depth research to explain their lack of attempts to avoid nutrition mistakes. The results clearly indicate the necessity of including metropolitan teenage boys in effective nutritional education for the rationalization of their dietary behavior.


Introduction
Adequate nutrition in adolescence is important for both present and future wellbeing because this period is important for preventing the consequences of nutritional deficiencies and excesses [1,2]. The diet of adolescents is not properly balanced, and the disproportions between the various nutrients are significant and have been observed in different countries [3][4][5][6]. Adolescence is a key period for the development of lifestyle and eating behaviors, which puts this group at a particularly high risk of obesity and being overweight. Unhealthy nutritional behaviors are common during adolescence and include excessive consumption of sweetened beverages, sweets, fat, and/or salty snacks, combined with reduced consumption of fruits, vegetables, milk, and dairy products [7]. These eating habits are associated with an increased risk of obesity and cardiovascular disease, which is why monitoring them is crucial because they are thought to last a lifetime [8,9]. A low-quality diet can also be caused by skipping breakfast or replacing lunch with

Study Design & Data Collection
The cross-sectional study was conducted on high school students between Autumn 2014 and Spring 2015 ( Figure 1). From 93 Warsaw public high schools, 30 schools with similar education profiles were selected. Schools with special education profiles (sport, bilingualism, arts) and organizational structures (school complex for adolescents in a wide age range) were excluded. The management of each school received an invitation to participate in the research study. The headmasters of eight high schools responded positively to the invitation (27% of the selected high schools).

Unhealthy Nutritional Behaviors and Patterns of Avoiding Nutrition Mistakes
The selected questions from The Beliefs and Eating Habits Questionnaire (KomPAN), developed and validated by the Commission of Behavioral Determinants of Nutrition from the Polish Academy of Sciences (Warsaw, Poland) [17], were used to assess the frequency of consumption of not-recommended products and unhealthy behaviors in the In each school, paper-pen questionnaires (N = 1082) were distributed in classes. The following inclusion criteria were used: the age of [16][17][18][19] years, living in Warsaw, a Polish metropolis (defined as a large urban center of high economic and social importance, hence the term "metropolitan high school students"). The following exclusion criteria were applied: diseases requiring special diets, chronic diseases, e.g., diabetes or asthma, use of medicines, and incorrectly completed questionnaires. All participants were informed about the necessary details and agreed to participate in the study. In the case of minors, written consent of the parent or guardian was obtained. A total of 981 students met the criteria (549 girls and 432 boys), but only a total of 652 students responded to the survey. Completed and returned questionnaires (N = 652) were verified by research staff, and incorrectly or incompletely filled questionnaires were excluded (N = 36). The final sample was comprised of 616 students (63% of the total sample; 34% of boys, 88% of girls).  [17], were used to assess the frequency of consumption of not-recommended products and unhealthy behaviors in the past six months (Table S1, Supplementary Materials). The frequency of consumption of the following products was examined: sweets, fast food, fried meat, sweetened beverages, energy drinks, beer, and strong alcohol. The frequency of unhealthy nutritional behaviors, as in adding salt or sugar, was tested. The following numerical codes were applied: 1-(almost) never, 2-(almost) once a week, 3-1-2 times a week, 4-3-4 times a week, 5-every day, 6-several times a day. For unhealthy nutritional behaviors such as not eating breakfasts and snacking, the following numerical codes were applied: 1-(almost) never, 2-(almost) once a week, 3-several times a week, 4-every day.

Measures
To assess the limitation of not-recommended product consumption (sweets, fast food, fried meat, sweetened beverages, energy drinks, alcohol) and the limitation of unhealthy behaviors (adding salt or sugar, snacking, night eating), four statements were used: (i) I do not limit myself; (ii) I am looking for a healthier alternative; (iii) I try to eat/drink this very rarely; (iv) I do not eat/drink/practice this. Due to the lack of existing measures, these items were tested in a pilot study on a group of 30 adolescents who did not take part in the main research. Thus, the statement "I do not eat/drink this" was added. The following categories and numerical codes were used for Statement (i)-not limiting (1); for Statements (ii) and (iii)-limiting (2); for Statement (iv)-not eating (3).
The data on both the frequency and limitation described above were used to identify patterns of avoiding nutrition mistakes. It was assumed that the most healthy eating pattern would be the one with less frequency, a willingness to not eat or to limit nonrecommended products, and a willingness to limit unhealthy eating behavior. The most unhealthy pattern was assumed to be eating with greater frequency and a willingness to consume nonrecommended products.

Sociodemographics, Anthropometrics Variables
Data on sex, age, weight, height, and family environment characteristics such as household size, siblings (no siblings, younger or older siblings), parental education (primary, secondary, university), accommodation (house, flat), and household structure (living with one parent, with both parents, with parents and grandparents) were collected. Body mass index (BMI) was calculated using self-reported weight (kg) and height (m). For students aged 16-18 years, the following age-sex-specific BMI cut-offs and categories were applied: BMI < 18.5 kg/m 2 is categorized as thinness; BMI = 18.5 to 24.9 kg/m 2 is categorized as normal weight; BMI ≥ 25 kg/m 2 is categorized as overweight and obesity [18]. For students aged 19, adult criteria and cut-offs were applied [19].

Nutrition Knowledge
To assess nutrition knowledge, 8 questions from the test, developed and validated by the group of Polish experts, were selected [20]. These questions verified the obligatory knowledge included in the school curriculum and were accepted by the teachers who are responsible for the high school curriculum. The subjects were required to select a proper answer to the questions on the following subjects: the number of meals in a day; the daily amount of beverages required for an adolescent; the interval between the last meal and bedtime; restrictions on the consumption of products sweetened with aspartame; health reasons to consume a low-protein diet; a starchy food product; the richest source of vegetable protein; food product rich in dietary fiber (more details in the questionnaire in Section 2; Table S1, Supplementary Materials)). The correct answer was given a mark of 1, incorrect as 0; the maximum number of points was 8. For further analysis of the results, the subjects were divided into two groups: insufficient level (0-4 points) and sufficient level (5-8 points) of nutrition knowledge.

Physical Activity in Leisure Time
The question about physical activity (PA) in leisure time is from The Beliefs and Eating Habits Questionnaire (KomPAN) [17]. The following categories were used: low PA-sitting lifestyle, TV viewing, reading books, light housework, 1-2 h a week; moderate PA-walking, cycling, gymnastics or other light physical activity performed 2-3 h a week; high PA-cycling, running, other sports activities requiring physical effort, performed over 3 h per week. The school program includes 4 h of sports activities per week; therefore, the level of physical activity in leisure time was crucial for the PA assessment.

Data Analysis
Categorical variables are presented as a sample percentage (%). Patterns of avoiding nutrition mistakes were derived using cluster analysis. The input variables were 22 data points of the frequency of consumption of not-recommended products, the frequency of unhealthy behaviors, and the limitation of consumption of not-recommended products and unhealthy behaviors. The k-means clustering algorithm, based on the Euclidean distances, was used. Three clusters were selected. The description of the cluster includes the following features: sex, age, BMI, family environment, level of physical activity in leisure time, and nutrition knowledge level. For the codes of numeric values, which refer to the frequency of consumption of not-recommended products, unhealthy nutritional behaviors, and behaviors limiting nutrition mistakes, mean and standard deviations were calculated, and an analysis of variance for comparing means of separate groups was used.
Logistic regression modeling was applied to assess the adherence to identified patterns by demographic factors, nutrition knowledge, physical activity, and family characteristics. Odds ratios (ORs) and 95% confidence intervals (95% CIs) were calculated. The ORs were adjusted for sex, BMI, characteristics of family environment, nutrition knowledge, and physical activity, excluding the modeled variable from the confounders set, respectively. The significance of ORs was assessed by Wald's statistics. For all tests, p < 0.05 was considered significant. Analyses were performed using IBM ® SPSS ® Statistics software (IBM Corp. Released 2017. IBM SPSS Statistics for Windows, ver. 25.0. Armonk, NY, USA: IBM Corp.).

Group and Cluster Characteristics
The majority of the subjects were girls (76%), students aged 16-17 years. The body mass index in the vast majority of participants indicated a normal weight. Almost 20% of students were thin, and only 8% had excessive body weight. Almost 70% of students had insufficient nutrition knowledge. Most of the participants in the study group lived in households with no more than four people (77%), in flats (61%), with both parents (69%). (Table 1).
In the study group, three clusters were identified, whose sociodemographic characteristics, PA level in leisure time, and nutrition knowledge are shown in Table 1.

Patterns of Avoiding Nutrition Mistakes and Their Determinants
Patterns of avoiding nutrition mistakes, derived using cluster analysis, were the following: Cluster 1. (Prudent Ones: I know and I limit)-the most healthy eating pattern. This group includes subjects who, significantly, seldom consumed not-recommended products and snacks and most often limited or did not eat not-recommended products ( Table 2).
Cluster 2. (Inconsequent: I know something and I limit sometimes); this group includes students who most often added sugar and salt but limited most not-recommended products and unhealthy behaviors ( Table 2). Cluster 3. (Rebels: I do not know and I do not limit)-the most unhealthy eating pattern. This group includes students with the highest frequency of consumption, who do not limit not-recommended products and snacking and rarely consume breakfast (Table 2). Girls (aOR = 6.34; 95% CI (3.89-10.34); p < 0.001) and students with moderate PA in leisure time (aOR = 2.12; 95% CI (1.14-3.94); p = 0.018) had over six-and two-times higher adherence to the Prudent Ones pattern ( Table 3). The adherence was lower by 62% in students with younger and older siblings (aOR = 0.38; 95% CI (0.23-0.62); p < 0.001) and by 57% in students with insufficient nutrition knowledge (aOR = 0.43; 95% CI (0.29-0.62); p < 0.001).
In the crude model, the adherence to the Prudent Ones pattern was 1.5 times higher in students living in flats (OR = 1.48; 95% CI (1.08-2.09); p = 0.017), while it was 49% lower in students with older siblings (OR = 0.51; 95% CI (0.32-0.81); p = 0.004); however, those associations became nonsignificant after the adjustment (Table 3). Table 3. Associations between patterns of avoiding nutrition mistakes with demographic factors, family characteristics, nutrition knowledge, and physical activity (odds ratios: crude and adjusted, 95% confidence intervals).

Discussion
Among the three patterns of avoiding nutrition mistakes among high school students from a Polish metropolis, two opposite patterns were identified: the Prudent Ones pattern, with limited unhealthy behaviors (45%), and the Rebels pattern, with no limits to unhealthy behaviors (16%). As we hypothesized, the avoidance of nutrition mistakes was associated with sex, nutrition knowledge, physical activity, and family environment. Higher adherence to the Prudent Ones pattern was found in girls and students with moderate physical activity, while adherence was lower in students with siblings and students with insufficient nutrition knowledge. The adherence to the most unhealthy pattern (Rebels) was found in students with families larger than 4 persons, children of mothers with primary education, and students with insufficient nutrition knowledge. Having siblings and insufficient nutrition knowledge increased the chance of adherence to the Inconsequent pattern. An unexpected finding of the study was the lack of association of the identified patterns with BMI.
Our results indicated that sex was a significant determinant of the adherence to the identified patterns, and metropolitan girls were more likely to avoid nutrition mistakes. This confirms a strong association of the female gender with the most healthy pattern and the male gender with the most unhealthy pattern. This is consistent with previous results indicating that sex significantly determines the choice of food and eating behavior. Young women more often avoid high-fat foods, eat fruits and vegetables, and report the greater importance of healthy eating, so dietary choices depend on the perception of femininity and masculinity [21]. Women often report significantly more prohealth behaviors than men and show healthier lifestyle patterns [22]. The results of other studies have shown that boys are the main consumers of junk food, sweets, and sweetened drinks and girls less so [22][23][24][25].
Living in a household with more than 4 people was associated with higher adherence to the Rebels patern. This relationship is difficult to explain directly. There are models of home eating environments based on sociocognitive and socioeconomic theories that suggest that the influence of the family, including parenting practices and other aspects of home, will shape the adoption of healthy eating behavior [26,27]. The available data explain the relationship between the size and structure of the family and qualitative and quantitative nutrition [28] but do not explain the relationship with the motivation of adolescents to specific eating behaviors; this is a field for further research. Therefore, the explanation and description of the relationships between complex aspects of adolescent environments (family, home, peers) and dietary behaviors remain a challenge.
We found an association between patterns of avoiding nutrition mistakes and having siblings. Having siblings was strongly associated with higher adherence (by 97-98%) to the Inconsequent pattern, characterized by less frequent attempts to limit the consumption of sweets, sugar, and salt. The number and profile of siblings may be related to compliance with specific nutritional patterns or health-related markers, both positive and negative [29,30]. A Spanish study showed that having more than one sibling is associated with a higher nutritional risk [28]. In the research of Cook et al., cited by [29], persons having siblings showed lower consumption of all nutrients, with the exception of carbohydrates and added sugar, which was confirmed by our results. One possible explanation is that more siblings means that the family has less to spend on each family member [31]. The only child, especially if it is a girl, may have better economic and social conditions to follow a healthy lifestyle. However, it may also be related to the age and sex of the siblings. In our study, students with siblings had greater adherence to the Inconsequent pattern, and those with younger siblings had lower adherence to the Rebels pattern. The complexity of the adolescent-siblings-parents relationship and dietary behavior should be taken into account in further research for greater insight.
Our study also confirmed that the adherence to patterns other than the Prudent Ones pattern was associated with the lower education of mothers. It is consistent with van Ansem et al.'s results [32], which indicated that children of mothers with higher education were characterized by prohealth eating behaviors more often than children of mothers with lower education. Unhealthy foods were more available at home for adolescents of mothers with a low educational level. Bere et al. [33] found that adolescents' fruit and vegetable consumption was linked to their mothers' level of education, while Watts et al. [26] showed that high-fat food availability was lower in families where the mothers had higher education. As they are more responsible for food supply and choice, mothers are shown as a "factor" affecting and shaping children's eating behaviors [34]. On the one hand, this result points out the importance of a mother's education, but on the other, high school students who have achieved a certain degree of education and have greater experience will acquire the skills of formal, operational, and critical thinking towards norms and restrictions to which they often disagree [35] and oppose. Without a doubt, this association indirectly points out the role of the mother's education and knowledge, but it is probably more important in the earlier stages of child development; this issue requires a longitudinal study.
Despite the presence of elements of nutrition sciences during school education, the majority of students were characterized by insufficient nutrition knowledge. This indicates the lack of effectiveness of the health program in improving the lifestyle of adolescents. Our study shows that high school students with insufficient nutrition knowledge have higher adherence to unhealthy patterns (Rebels or Inconsequent) and lower adherence to the most healthy pattern (Prudent Ones). Despite the fact that some authors have shown that nutrition knowledge has a weak correlation with actual food choices [36], successful health promotion should be a better fit for the needs, attitudes, and characteristics of the target group and should focus on increasing nutrition literacy (skills and behavior) during adolescence [37,38]. Our results highlight the importance of nutritional knowledge for prohealth behaviors of high school students. Additionally, higher adherence to the Prudent Ones pattern is associated with moderate physical activity in leisure time; the coexistence of physical activity and a healthy diet in adolescents [6,39,40] and in girls [41] has been confirmed in the literature.
In our study, patterns of avoiding nutrition mistakes were not associated with BMI. The explanation for the lack of such a relationship can be found in the specificity of the studied group, which mostly declared high physical activity. Additionally, the percentage of high school students who were thin was two-times more than those with excess body weight (17% vs. 8%). Furthermore, in our study, 20% of the girls and 10% of the boys were classified as thin. Thinness is more common among adolescent girls and young women in European countries such as the Czech Republic, Sweden, Greece, and France than among boys. Additionally, the prevalence of thinness among Polish children is higher than in other European countries or Australia and the USA. This tendency may be related to sociocultural conditions, the fashion for a slim figure created in the media, and greater weight control in adolescent girls. The authors also point to the role of psychological factors such as self-esteem, body image, and emotional state [42].
Our results show that Polish high school students from the metropolitan area made typical nutrition mistakes such as unrestricted eating of sweets, fast food, and fried meat, drinking sweetened beverages and energy and alcohol drinks, and also adding salt or sugar, snacking, and eating at night. Such behaviors are characteristic for this age group and are associated significantly with the adolescents' peer behaviors [9,[43][44][45][46][47]. This is a natural phenomenon, as adolescence is a period of independent decision-making and increased autonomy in personal choices, which include practicing unhealthy behaviors. The results of the conducted research showed that some of the adolescents were aware of their nutrition mistakes and tried to limit them, but most of them did not avoid nutrition mistakes. Bryan et al. [48] pointed out that persuading adolescents to change their eating behavior often does not bring the expected result because they face a behavioral obstacle in the form of a sense of their own autonomy. Adolescents make the right food choices when they have a sense of independence from adult control and associate these behaviors with improved social status [48]. Both modifiable determinants revealed in our study, physical activity and nutrition knowledge, should be used as the basis for the development of educational and public health promotion programs for adolescents. Future programs need to be better suited to the sociodemographic characteristics of the groups and should include changing adolescent lifestyles.

Strength and Limitation of the Study
Our study had a number of strengths and limitations that should be mentioned. This study is novel due to its comprehensive approach, taking into account lifestyle behaviors, the level of nutrition knowledge, as well as family environment characteristics. A major strength is the relatively large sample of metropolitan adolescents. The study area is a metropolis, the capital city of Poland, so the results of this study may be a good basis for further research in high school students from big cities. This is due to the fact that after the completion of primary school, almost half of the graduates continue their education in high schools, of which over 28% are located in the 18 largest Polish cities [49]. Ideally, the sample would have included a larger number of boys. There was a lower response rate for boys vs. girls, which was taken into account by poststratification weights, but, in the future, alternative recruitment strategies to encourage boys and to reach groups that vary in sex are needed. Most nutrition studies suffer from selection bias, in which there are more females and individuals with more health/nutrition knowledge participating [50]. Additionally, a relatively lower percentage of students with excessive body weight and a higher percentage with high physical activity levels seem to confirm the existing tendency; it should be noted that data about body mass and height as well as physical activity were self-reported. However, the data on health-related behaviors obtained by self-reporting has been found to offer satisfying reliability [51].
This study was also limited by the cross-sectional design, which precludes the investigation of casual relationships. However, it provides data to more complex study designs or to create interventions targeted at improving health-related behaviors.

Conclusions
Our research revealed that avoidance and no-avoidance nutrition mistakes were associated with sociodemographics, PA in leisure time, and nutrition knowledge but not with BMI. Among the three identified nutrition-mistakes avoidance patterns, two opposite patterns were identified: the Prudent Ones pattern, with limited unhealthy behaviors, and the Rebels pattern, with no limitation to unhealthy behaviors.
Female gender, no siblings, nutrition knowledge, and greater PA in leisure time increased the adherence to the Prudent Ones pattern, which can be described as "know and avoid nutrition mistakes". Having siblings, mother's secondary education, and insufficient nutrition knowledge increased the adherence to the Inconsequent pattern, who "know something about healthy eating and sometimes avoid unhealthy behaviors". Male gender, living in large families, insufficient nutrition knowledge, and mother's primary education increased the adherence to the Rebels pattern, who "do not know and do not avoid unhealthy behaviors".
From a public health perspective, implementing interventions that increase nutrition knowledge and physical activity in high school students are important for improving prohealth dietary habits. The groups of high school students with the feature characteristics of the Rebels and Inconsequent are possible targets for intervention. Further in-depth research is needed to explain their lack of attempts to avoid unhealthy behaviors.
Author Contributions: This manuscript represents original work, which has not been published previously and is not being considered by another Journal. The authors' responsibilities were as follows: conceptualization-M.D.; methodology-M.D. and M.G.; software-K.G.; data analysis M.G. and K.G.; investigation, data curation-S.K.; writing-M.D., M.G., and S.K. All authors have read and agreed to the published version of the manuscript.