INTRODUCTION

Healthy lifestyles (EVS) are associated with better health that reduces mortality, increases life expectancy⁽¹⁾, improves psychological well-being and reduces symptoms of depression, anxiety⁽²⁾ and the risk of developing chronic diseases⁽³⁾. Therefore, EVS are considered one of the main strategies for health promotion and disease prevention.

Although the promotion of EVS is important for all age groups, the availability of approach and measurement strategies vary among them. Young people, particularly Mexican university students, are a particularly important group to target because they are in the process of developing health-related habits and behaviours; their lifestyles can negatively affect their health⁽⁴⁾ due to inadequate diet, low levels of physical activity, poor quality of sleep, high levels of stress and consumption of tobacco or alcohol⁽⁵⁾. University students who live with excess weight present low levels of appreciation for life and poor eating, studying and resting habits, especially when compared to students of normal weight⁽⁶⁾.

Prior to the COVID-19 pandemic, differences in health habits between men and women were already evident: women had less substance use, while men had better nutrition, study and rest ⁽⁶⁾. During the COVID-19 pandemic, differences continued: women increased the number of meals per day, and men increased their consumption of sweet foods⁽⁵⁾. The pandemic also affected young people: the time of exposure to electronic devices exceeded 10 continuous hours, physical activity decreased⁽⁵⁾ and sedentary behaviours increased⁽⁷⁾.

These changes are a serious concern considering the state of health of the Mexican population before the pandemic . In 2018, the combined prevalence of overweight and obesity in the population over 20 years of age was 75.2% (39.1% overweight and 36.1% obesity), which was higher in women (76.8%) than men⁽⁸⁾. In 2021, during the pandemic, the prevalence was 72.4% (75.0% in women and 69.6% in men), and although there was a decrease of 3% in the general population and almost 2% among women⁽⁹⁾, the percentages are still high. Among college students, higher combined prevalences of overweight and obesity have been reported in men (42.9% to 45.75%) than in women (36.27% to 37.2%), respectively⁽⁶⁾,⁽¹⁰⁾, while other studies have reported overweight prevalences of 41.1% in women⁽¹¹⁾. Thus, excess weight continues to be a public health problem, especially among university students, and this is largely determined by inadequate health behaviours.

EVS are related not only to physical health but also to psychological factors that influence mental health, such as psychological well-being⁽¹¹⁾,⁽¹²⁾ and the self-regulation of eating habits⁽¹³⁾, which are central components of the change process towards healthier lifestyles.

However, the study of EVS has prioritized food, exercise and substance use, leaving aside the exploration of other dimensions such as health responsibility, social support, stress management and appreciation for life^{(5) (6) (7)}. This may be especially related to the limited availability of instruments that allow their comprehensive evaluation.

In Mexico, there are five instruments: the FANTASTIC lifestyle checklist⁽¹⁴⁾, the Instrument to Measure the Lifestyle of Diabetics⁽¹⁵⁾, the Instrument to Evaluate the Lifestyle of Mexican Adolescents⁽¹⁶⁾, the Health-Promoting Lifestyle Profile-II⁽¹⁷⁾ and the Healthy Lifestyle Scale for University Students (EEVSUV)⁽¹⁸⁾,⁽¹⁹⁾. However, only the last two have been validated in university students, and none have compared men and women because there are cultural and behavioural components that could foresee differences in the way the construct is measured.

The Health-Promoting Lifestyle Profile-II has six dimensions and 46 items that explain 49.9% of the variance⁽¹⁷⁾. However, the validation procedures have some limitations: they do not consider that the scale was created for adults 18 to 92 years of age, the process of translation into Spanish or the cultural adaptation for the Mexican population, including students in upper and upper secondary education; these students are socially, educationally and psychologically different, and in terms of the results, these procedures present inadequate adjustment indices of inadequate magnitude (e.g., RMSEA = 0.08), keep items with low factorial load (0.273), estimate reliability using only Cronbach's alpha coefficient without considering confidence intervals and not performing an analysis of measurement invariance by sex⁽¹⁷⁾.

For its part, the EEVSEU was created for Chinese university students from Pender's theoretical model of health promotion and contains 38 items distributed across eight dimensions⁽¹⁹⁾. This scale was translated into Spanish and adapted for female Mexican university students⁽¹⁸⁾. Subsequently, the original model was evaluated under a confirmatory factor analysis (CFA), and when it did not receive support, an exploratory factor analysis (EFA) was performed. Then, after successive EFAs, a new model was proposed with 14 items distributed across four dimensions that explain 64.4% of the variance⁽¹⁸⁾; thus, this version of the EEVSEU has better psychometric support, seems to be a more promising scale for use in the Mexican population, and has already been used in previous research in the population of interest, presenting adequate levels of measurement⁽⁶⁾,⁽¹¹⁾.

Therefore, the main objective of this research was to determine the psychometric properties of the EEVSEU in male and female Mexican university students.

MATERIALS AND METHODS

This is a cross-sectional, instrumental study to review the psychometric properties of the EEVSEV.

RESULTS

When analysing the goodness-of-fit indices of the first two measurement models in men and women (separately), it was found that M1 presented an adequate fit in women (CFI = 0.923; RMSEA = 0.050, CI 90% [0.044-0.055]; WRMR = 1.158) but not in men (CFI = 0.890; RMSEA = 0.057, CI 90% [0.051-0.063]; WRMR = 1.176). With respect to M2, a good fit was obtained in the group of women (CFI = 0.961; RMSEA = 0.080, CI 90% [0.066-0.093]; WRMR = 0.983), showing acceptable factor loadings (Table 1), but in the group of men, no results were obtained because the model presented estimation problems associated with items 13 (Drinking alcohol in excess) and 14 (Smoking or consuming tobacco or marijuana), so it did not receive support in that group.

Therefore, in M1, the items related to health risk behaviours (CRPS; items 13, 14, 15 and 16) showed difficulties. First, items 15 (Listen with headphones continuously for more than 30 minutes) and 16 (Use the computer continuously for more than one hour) presented low factorial loads (λ <0.50). In turn, items 13 (Drinking excess alcohol) and 14 (Smoking or consuming tobacco or marijuana) presented negative factor loads in men and positive factor loads in women (Table 1). Thus, the CRPS dimension was excluded from the analysis.

In this sense, a new model was conceived that considers all the dimensions of the original model except dimension CRPS (M3), which showed an acceptable fit in both men (CFI = 0.893; RMSEA = 0.063, CI 90% [0.056-0.069]; WRMR = 1.175) and women (CFI = 0.928; RMSEA = 0.053, CI 90% [0.048-0.059]; WRMR = 1.132), although Item 4 (Perform vigorous exercise 30 minutes after eating) and Item 26 (Talk to others about my problems) also presented inappropriate factorials (λ ≤ 0.50) (Table 1).

Table 1: Factor loadings of each item by measurement model in men and women. 

λ: factorial load; H: men; M: women; M1: measurement Model 1; M2: measurement Model 2; M3: measurement Model 3; NE: not estimated.

After this, favourable evidence of measurement invariance was obtained in relation to M3 (Table 2).

Table 2: Measurement invariance with respect to gender. 

RMSEA: mean square error of approximation; CFI: comparative adjustment index; Δ: variation.

Then, considering that M3 is invariant between men and women, the group was analysed jointly, obtaining adequate fit indices (CFI = 0.912; RMSEA = 0.059, CI 90% 0.055, 0.062; WRMR = 1.385) and acceptable factor loadings in most cases, as well as adequate reliability coefficients.

Regarding the reliability of the scores, in most cases, an acceptable magnitude was found, except in health responsibility. With respect to the reliability of the construct, in all cases the magnitudes were adequate (Table 3).

Table 3: Factor loadings and reliability of M3 in the total sample. 

Finally, associations are observed between the dimensions of the EEVSEU and psychological well-being and self-regulation of eating habits, although to a lesser degree with the self-assessment dimension (Table 4).

Table 4: Correlations between healthy lifestyles, psychological well-being and self-regulation of eating habits. 

^Note.In italics: nonsignificant correlations (<.20)

DISCUSSION

The EEVSEU shows adequate psychometric properties for its use in the Mexican university population, which will allow us to strengthen the research in this group and, above all, the implementation and improvement of strategies for the promotion of EVS to improve health.

This version is shorter than the original model since when eliminating the dimension of health risk behaviours due to inadequate adjustment, four items are lost. This loss of items was also observed in the Iranian version due to low factor loadings of items 15 and 16, in addition to low reliability values of the scores⁽¹⁹⁾,⁽²⁵⁾. In the Mexican version, these items were also eliminated⁽¹⁸⁾. These two items are related to negative health effects⁽²⁶⁾,⁽²⁷⁾; however, the processes derived from the COVID-19 pandemic may be related to its loss, since both are directed to behaviours closely linked to the modality of online classes and distance learning (Item 15: Listen with headphones continuously for more than 30 minutes; Item 16: Use the computer continuously for more than one hour). Therefore, it is probable that the psychometric behaviour of these items originates because what could be considered an unhealthy behaviour before the pandemic is currently valued as a tool to access education.

Items 13 and 14 (drinking alcohol in excess and smoking or consuming tobacco or marijuana, respectively) also belong to the eliminated dimension; although they are behaviours directly related to EVS, there are other questionnaires available for their measurement^{(28) (29) (30)} that can be used in conjunction with this questionnaire to study and explain in greater depth the VE of Mexican university students.

The analysis of measurement invariance with respect to sex has not been reported in previous research^{(17) (18) (19) (25)}, which represents one of the main strengths of this research. This measurement model offers the possibility of use as a standardized tool to measure the EVs of both female and male Mexican university students and make fair comparisons that reflect the differences or similarities between the sexes.

Finally, the relationship of the EVS to psychological well-being and the self-regulation of eating habits provides greater theoretical support to the measurement model proposed in this research; it seems that high scores of psychological well-being and self-regulation help the development, consolidation and maintenance of EVS or vice versa ⁽⁶⁾,⁽¹¹⁾.

However, and despite these strengths, there are limitations to this study. One of the main limitations is that only young university students participated. In Mexico, many younger people are engaged in the formal and informal labour market, which is why subsequent studies should consider their inclusion.

Finally, the scale used in this research should be applied to other variables and adapted for other population groups, especially young people who are not students or the general population.

CONCLUSIONS

This study's findings provide a comprehensive proposal for measuring EVS in a standardized way to objectively measure variables of interest. These results have potential implications for nursing research, health education interventions and disease prevention, as well as follow-up measures to gauge the recovery process of the health service user.