Introduction

Quality of life is a dynamic, subjective, and multidimensional construct [1]. When quality of life is evaluated based on the health situation of the people, it is called health-related quality of life (HRQoL), that is, the qoality of life associated with the person’s physical, mental, and social well-being [2]. Bakas et al. [1] performed a systematic review to identify the most commonly used HRQoL model, finding that the model created by Wilson and Cleary [3], either in its original form or as a modification, was adequate, clear, and consistent. It is also the most commonly probed and could be applied to all individuals irrespective of age, health, or culture.

Specifically, the model includes five concepts of health in a continuum: biological and physiological factors, state of symptoms, functional state, general perception of health, and HRQoL. The model focuses on the relationships between different health domains by proposing a linear sequence of causal links that begin with the clinical level (biological, objective) at one end of the continuum and moves outward considering the interaction of individuals with the environment to perceive a quality of life level (psychological, subjective) on the other extreme. It is hypothesized that emotional and social constructs have potential causal relationships with each of the five domains; and that the characteristics of the individuals, as well as the characteristics of the environment, including the organizations that affect the individual, also influence those domains. Thus, this model integrates two paradigms, the biomedical and the social sciences and hypothesizes bidirectional causal relationships between and among the concepts [3].

In terms of application, the model provides a theoretical basis for the selection of variables according to the causality relationships between health concepts useful for both clinical and research. It is important to note that this model has shown to improve knowledge on HRQoL of a wide range of populations that face long-term health problems [4,5,6,7]. Wyrwich et al. [7] concluded that this model improved the understanding and usefulness of the health status of patients with generalized anxiety disorder. An underlying assumption of the model is that an understanding of the relationships between the concepts can guide the design of optimal clinical interventions [4].

Worldwide, there has been a great interest in the analysis of the relationship between body weight and HRQoL. Initially, studies focused on the adult population and, subsequently, on adolescents and children. Excess weights, including overweight and obesity, are public health problems which global incidence has significantly increased in recent decades. Recent estimates reveal that the worldwide prevalence of obesity doubled between 1980 and 2008. In 2014, there were over 1900 million overweight adults aged 18 years and over. Of these, over 600 million were obese [8]. In Mexico, the last National Health and Nutrition Survey found that the combined prevalence of overweight and obesity in adolescents was 39.2% in women and 33.5% in men [9].

Two systematic reviews [2, 10] and one meta-analysis [11] showed that overweight is related to a decrease in HRQoL in children and adolescents. There is consistent evidence that the dimensions of physical and social functioning are the most affected, but there is little empirical evidence regarding affection in emotional functioning and scarce evidence of academic functioning [2]. Similarly, it has been shown that HRQoL is affected at the extremes of the body composition continuum, i.e., underweight and obesity [12].

Recent outcome studies indicate that several variables play an important role in the relationship between excess weight and HRQoL. For example, women obtained lower scores in at least one of the dimensions of HRQoL compared to men, regardless of the instrument used [e.g., 13, 14]. Other variables that may regulate this relationship are diet, abnormal eating behaviors and attitudes, physical activity, educational level, and parents’ education [13], in addition to socioeconomic status [15] and the person’s perception of food. However, the evidence regarding the role of these variables is scarce and has been analyzed separately; thus, it is necessary to analyze the joint effect of these variables.

In the Latin American context, there has been a recent interest in research on HRQoL in adolescents [16], and more has been done in clinical contexts than in general samples [cf. 17,18,19,20,21]. Studies in adolescents living in low- and middle-income countries are also necessary, because most studies have been conducted in developed countries [22]. In addition, several studies have evaluated HRQoL in adolescents with excess weight and how it is affected by other specific dimensions. However, no studies were found that showed a causal model that analyzed the factors that affect HRQoL. To gain insight into variables that explain HRQoL among adolescents with different body composition, we developed and put to the test an integrative model that relies on Wilson and Clearly’s approach, which is one of the most widely cited conceptual frameworks of HRQoL. We selected this model, because it offers the most complete vision of the pathways that unite the traditional clinical variables and the most relevant concepts to understand HRQoL. In our hypothetical causal model, we focus on four of the health domains proposed in the model: biological and physiological factors, which include the Body Mass Index (BMI) and Weight Status; symptom status (SPS; drive for thinness and food concern); and functional status (FS; physical activity; estimation of food consumption; and socioeconomic status); and HRQoL. With the present study, we intend to make contributions that improve the HRQoL of people based on adequate assessments and optimal clinical interventions taking into account their physical, mental, and social functioning. Likewise, research that will conceptually explain the relationship of clinical variables with HRQoL indicators will be strengthened, and as proposed by Wilson and Cleary [3], the participation of mediating variables in such relationship will be analyzed.

The purpose of this research was to develop and assess a causal model of HRQoL in adolescents with different body composition. The variables that were included can provide evidence that will not only allow explanatory models of the interrelation between them, but also will make it possible to generate the theoretical and empirical basis for implementing concrete actions to modify the relevant variables, enabling the HRQoL of adolescents with excess weight to be immediately improved.

Method

Participants

A cross-sectional study was conducted among Mexican adolescents in Ciudad Guzman, Jalisco, Mexico, which is considered a small city. The sample consisted of 209 students (107 women and 102 men) with a mean age of 12.91 years old (SD = 1.71). Participants attending primary schools (n = 82) and secondary schools (n = 127) and were included in the sample given its availability to participate in the study. The majority of the participants (women and men) belonged to the medium–high socioeconomic level (see Table 1). Inclusion criteria for this study were 10–15-year-old adolescents and that they were willing to comply with the study procedures and provide written informed consent/assent. Children with chronic diseases were excluded.

Table 1 Sample characteristics
Full size table

Instruments and measures

Health-related quality of life

The Screening for and Promotion of Health-Related Quality of Life in Children and Adolescents (KIDSCREEN) was used. This questionnaire has versions for different countries [23] and is a general measure of HRQoL. The Mexican version of KIDSCREEN has suitable psychometric properties [24]. For this project, the 27-item version was used, making it possible to explore five dimensions: Physical Well-being, Psychological Well-being, Autonomy and Parent Relation, Peers and Social Support, and School Environment. The response options are on a five-point Likert scale (from never to always). A higher score indicates better self-perception.

Disordered eating behaviors and attitudes

The Children Eating Attitudes Test (ChEAT-26) [25] evaluates abnormal eating behaviors and attitudes. It is composed of 26 items; each one is scored on a six-point Likert scale with response options ranging from never to always. The ChEAT was validated in Mexico by Escoto and Camacho [26] in a sample of 9–15-year-old pubescents and adolescents. The internal consistency of the instrument (α = 0.82) and its test–retest reliability (r = 0.79) were adequate, and the factor analysis yielded five factors that explained 43.74% of the variance.

Food perception in adolescents

The questionnaire Estimación y Consumo de Alimento en Niños y Adolescentes (ECAN; Food Estimation and Consumption Questionnaire in Children and Adolescents) [27] assesses how foods are perceived and how often they are consumed in a certain period. It consists of 17 items divided into two sections, estimation and consumption frequency, and uses a five-point Likert scale. For the estimation section, the options range from not healthy to very healthy, and for the consumption frequency section, the response options range from never to daily. For purposes of reproduction of the empirical model, the estimation scale, which was recoded and reverse-scored, was considered. Therefore, a higher score on the scale represents a perception estimation of less healthy food.

Body composition

The InBody 230 scale performs a bioelectrical impedance analysis in a segmental, multifrequency, and octopolar manner with an accuracy of 100 g.

Tanita HR200, a metric scale that measures the height or the length of an individual, with an accuracy of 1 mm, was used.

Body mass index (BMI) was provided by the scale. For children under 19 years of age, there are BMI tables for age and gender. Weight status was defined using the age-appropriate reference tables for men and women 2–20 years of age from the Centers for Disease Control and Prevention [28]. BMI values below the 5th percentile were considered to be underweight, from the 5th percentile to less than the 85th normal weight, from the 85th percentile to less than the 95th overweight, and from the 95th percentile obese.

Physical activity

To investigate physical activity and inactivity, three items from the 2015 Middle School Youth Risk Behavior Survey (YRBS) were used: (1) the number of days physically active over the last 7 days (i.e., whether they performed 60 min of daily activities that increased their heart rate or accelerated their breathing); (2) the number of school or community sports teams in which they played over the last 12 months; and (3) the number of days attending physical education classes at school over the last week [29]. The reliability of the YRBS questionnaire [30, 31] and its validity [32] have been analyzed, obtaining satisfactory results.

Socioeconomic status

This variable was calculated based on the parents’ education and work and was classified as: low working class, high working class, medium–low, medium–high, and high according to Rivas-Torres and Bianchi-Aguila [33].

Procedure

According to ethical principles, the research project was presented to the authorities of the educational institutions, and a formal authorization was obtained from each school. The parents signed an informed consent as a precondition for their child’s participation. In addition, we assured the voluntary nature of the children’s participation as well as the confidentiality of the data collected. The study’s purpose and all the procedures involved were explained in an understandable way to all participants. After explaining the purpose of the study, the participants completed the questionnaires during normal classes at their school in the presence of one researcher, and at the end, they were asked to move to another classroom in which trained staff individually recorded their height and body weight.

Ethical considerations

This study was approved by Southern University Center Ethics Committee. All the study was conducted according to the Mexican Ethical Code of Psychologists [34], the ethical principles of the American Psychological Association [35], and the Declaration of Helsinki of the International Ethical Guidelines for Biomedical Research Involving Human Subjects of the World Medical Association, and it was considered a low-risk study according to the General Health Law of Mexico.

Data analysis

Descriptive statistical analysis was performed for the variables recorded. For continuous variables, the mean and SD are presented, whereas, for categorical variables, the percentages are presented. Subsequently, a structural equation modeling (SEM) analysis was performed to evaluate the HRQoL model with structural equation software (EQS version 6.1). SEM was performed using the maximum-likelihood estimation method based on the unprocessed data matrix and from the standardized Mardia’s coefficient. The indicators of goodness of fit included: the Chi-Square statistic (χ2), χ2 by degrees of freedom ratio (χ2/df), the non-normed fit index (NNFI), the comparative fit index (CFI), the incremental fit index (IFI), and the goodness-of-fit index (GFI). The cut-off value considered for the NNFI, CFI, IFI, and GFI was ≥ 0.90. In addition, the standardized root-mean-square residual (SRMR) and the root-mean-square error of approximation (RMSEA), with values < 0.08, were considered [36,37,38,39,40].

Results

The descriptive analysis of the characteristics of the participants (see Table 2) showed that the combined percentage of excess weight (overweight and obesity) was higher in women (40.6%) than in men (28.4%), whereas low weight was more prevalent in men (41.2%) than in women (22.4%). The difference was significant (Chi-square = 8.68, p < 0.05). The BMI mean was 20.95 (DS = 4.23) for total sample, 21.50 (DS = 4.35) for women, and 20.37 (DS = 4.03) for men.

Table 2 Descriptive statistics for body composition and questionnaires
Full size table

Prior to performing the SEM analysis, a multivariate normality test was performed. The standardized Mardia’s coefficient was − 0.42, which suggests that the model data did not deviate from multivariate normality. The SEM was then tested to analyze the contribution of the variables of biological status [BS; body mass index (BMI) and weight status (WS)], symptomatic psychological status (SPS; drive for thinness and food concern), and functional status (FS; physical activity, estimation of food consumption, and socioeconomic status) on HRQoL in the sample of adolescents.

As shown in Table 3, the initial model showed a significant Chi-square likelihood ratio adjustment (χ2 = 73.24, df = 40, p = 0.001). Then, we proceeded with the adjustment of the model considering the suggestions of observed variables and construct disturbances based on the Lagrange multiplier tests and the highest standardized residuals (> 0.10). In this manner, the fit of the model was obtained with a significant likelihood ratio Chi-squared (χ2 = 51.88, df = 38, p = 0.07). The final model was adjusted with three movements (Fig. 1), a correlation between errors of the variables from socioeconomic status with social support and peers (− 0.24); a correlation between FS disturbances and HRQoL was also established (− 0.57), and the effect of BS on FS (0.34) was also determined.

Table 3 Goodness-of-fit indexes of the model to predict HRQoL
Full size table
Fig. 1
figure 1

Paths related to the HRQoL model. Ovals represent latent constructs, and rectangles represent observed variables. Unidirectional arrows connote “causal” relationships, and bidirectional arrows indicate correlations. BS biological status, SPS symptomatic psychological status, FS functional status, HRQoL health-related quality of life, BMI body mass index, WS weight status, DT drive for thinness, FC food concerns, PA physical activity, FCE food consumption estimation, SS socioeconomic status, PWB psychological well-being, ARP autonomy and relationship with parents, SSP social support and peers, AE academic environment

Full size image

In the final adjusted model, the explained variance of HRQoL was 13%, for FS was 19%, and for ES was 19%. Thus, two pathways can explain HRQoL. As initially proposed, BS showed a positive effect on SPS (0.44) but a negative trend in FS (− 0.45), and FS had a positive effect on HRQoL (0.21). In addition, the model revealed a second explanatory pathway with a positive effect of BS on FS (0.34) and of FS on HRQoL (0.21).

Discussion

The purpose of the present research was to develop and assess a causal model of HRQoL in adolescents with different body composition. The model evaluated was based on the quality of life model proposed by Wilson and Clearly [3], which has been analyzed extensively in this field of study. The model began with the variables of body composition and ended with HRQoL. Intermediate variables along this continuum included psychological aspects and functional status. The fit indices indicated that the hypothesized model represents an adequate solution to the data; that is, it provides a good representation of the relationship between the variables analyzed in adolescents with different body composition.

In particular, the current findings in this sample of 10–15-year-old adolescents are remarkable and it is important to analyze several results obtained in this empirically assessed model. This study suggests that BS, SPS, and FS are important components to consider when targeting improvements in the HRQoL of adolescents, being BS (BMI and weight status) the most important component. Specifically, two pathways of HRQoL involvement were found. In the first pathway, BS influences SPS particularly in cognitions such as drive for thinness and food concerns, and these have a negative effect on FS (physical activity, food estimation, and socioeconomic status), which has a direct effect on the perception of HRQoL. This finding confirms the linear form in which the variables are related to explain HRQoL in the original model of Wilson and Cleary [3]. The second pathway indicated that BS has an indirect effect on HRQoL through FS; that is, BS has two indirect influences on HRQoL. Similarly, the results indicated that BS alone does not affect HRQoL; other variables must be present to have an effect. There are studies that support this finding. For example, Gowey et al. [41] found that abnormal eating behaviors and attitudes along with body weight are the best predictors of HRQoL in 8–12-year-old children.

Regarding the considerable evidence that BMI is associated with HRQoL and its multiple domains in children and adolescents [2, 10, 11], findings from this study revealed that BMI and weight status are not the sole determinants of HRQoL. There are psychological and sociodemographic variables that contribute to explain HRQoL. It means that the relationship between BS and HRQoL is complex and it is necessary to consider other variables to offer a complete explanation for this relation to develop effective intervention strategies. Excess weight people are at risk of developing health-compromising behaviors that may compound associated medical and social problems [42]. The condition in their experience of life is likely to continue due to the poor effectivity of obesity treatments. Our findings evidence the necessity of conducting studies to search effective ways to address problematic BMI in the early life through precise interventions. Such efforts possibly reduce the risk for physical health consequences in the long run, but have more immediate effects on the HRQoL of adolescents.

In the first pathway, SPS was important and included the drive for thinness and food concerns, variables that have an important role in the presence of abnormal eating behaviors and for the development of eating disorders [43]. It has been demonstrated that these variables disturb food and eating behaviors and increase the motivation to reduce body weight. These findings suggest considering, in addition to the body composition, the presence of other variables that contribute to the diminishing of HRQoL [e.g., 44]. These variables, drive for thinness and food concerns, should be taken into account before implementing prevention or treatment programs.

An original contribution of the present study is that, to the best of our knowledge, this is the first proposal to identify, by means of an empirical model, variables related to HRQoL in adolescents with different body composition. The factors influencing HRQoL in obese adolescents have been analyzed using multiple linear regression models [cf. 13]. This analysis can be employed to test conceptual models, with the limitation that the variables considered cannot be simultaneously independent and dependent. SEM is an extension of multiple regression analysis that allows these relationships between the variables be more clearly distinguished.

Another relevant finding was the confirmation of high percentages of adolescents with excess body weight, being higher among women (40.6%) than men (22.4%). In two senses, these results coincide with those found in the most recent National Health and Nutrition Survey [9]. On one hand, approximately 40% of adolescent women present these important public health problems. On the other hand, regarding the relationship between genders, it is also confirmed that these health problems are more prevalent in women than in men. The high percentages of overweight adolescents have urged the implementation of health policies in the short term, because this health condition will generate important physical, psychological, social, and economic consequences [45, 46]; and these consequences increase considerably if the excess weight begins to develop in the early ages. These findings also show the need for more research to understand the problem, as in the case of the present research.

The limitations identified for the present study are discussed below. First, it is a cross-sectional research; thus, it is not possible to analyze the temporality of the relationships identified. It is suggested that future research with long-term designs should be conducted to achieve a broad understanding of the relationships found in the model. Second, caution has been adduced to perform body composition analysis in children and adolescents through bioelectrical impedance analysis (BIA), due to accelerated changes in weight, size, and fat indicators; however, validity studies have been carried out suggesting the pertinence and convenience of the BAI to assess the nutritional status of minors, putting it before possible limitations, such as ethnical differences [47]. Third, regarding the age of the participants and minimizing fatigue when completing the questionnaires, it was decided to assess only some variables, resulting in the exclusion of variables that may also be relevant.

Finally, we conclude that the results support the empirically evaluated model. The variables of body composition, drive for thinness, and food concerns have an indirect effect on HRQoL through physical activity, food estimation, and socioeconomic status.