University students may experience academic stress, defined as “that which is suffered by students in secondary or higher education and whose exclusive source is stressors related to the activities to be carried out in the school environment” (Barraza and Silerio, 2007). The main academic stressors are exams, presenting assignments, tutorials, work overload, lack of time, competition among classmates, group work, lack of incentives, conflicts with classmates, high demand academic tasks, little autonomy in academic work, and work pressure (Barraza and Silerio, 2007; Gutiérrez Rodas et al., 2010; González Cabanach et al., 2010a). The greatest academic stress is caused by exams, distribution of time, meeting stipulated deadlines, and academic overload (Feldman et al., 2008; González Moreno, 2017; García-García, 2019). Epidemiologically, previous studies estimate that between 47% and 55% of university students suffer from moderate levels of academic stress (Rivas et al., 2014; Estrada et al., 2021).

The consequences of academic stress on health are similar to those of psychosocial stress. The student body can suffer consequences in their physical and psychological health, since the students’ lifestyle is modified by the academic demands and the stress that, in most cases, determines the acquisition of health risk behaviors (i.e., excessive consumption of caffeine, tobacco, stimulants or tranquilizers, hypercaloric, hypersodic or hyperlipidic intake). These lifestyle changes make them susceptible to headaches, sleep disturbances, irritability, lack of concentration, anxiety, depression, and burnout (Extremera Pacheco et al., 2007; Balanza et al., 2009).

Likewise, the university population may be a group vulnerable to the stress generated by the use of information and communication technologies (ICTs) in both academic and personal spheres. In the current university context, the use of ICTs has been consolidated and increased, given that the vast majority of students use laptops, tablets or smartphones in the classroom (González Elices, 2021). The term technostress was used in 1984 by the psychologist Craig Brod to denominate the physical and emotional alterations that some people suffered when faced with handling computers (Jiménez, 2010). It is defined as a negative psychological state that is related to the use of ICTs, exposure to them, or an anticipatory fear or threat of their use in the future (Salanova, 2003). It derives from a mismatch between demands and available resources to use ICTs, which produces a high level of unpleasant psychophysiological activation, tension, and discomfort when using technologies or thinking about their use (techno-anxiety), eventually reaching the point of rejecting ICTs (techno-phobia) (Salanova et al., 2006).

Technostress affects both those with little technological knowledge who may reject the use of ICTs and perceive them as something negative, as well as those accustomed to its use, who experience frustration due to continuous training, recycling and the rapid acquisition of new knowledge (Minaya Lozano, 2008). In addition, it entails consequences for physical and mental health, such as eyestrain, headaches, backaches, digestive problems, irritability, frustration, demotivation, anxiety, memory and concentration problems, addiction, burnout, and lower satisfaction (Tams et al., 2014; Samaha and Hawi, 2016; Mahapatra and Pati, 2018; Rodríguez-Vásquez et al., 2021; Sánchez-Macías et al., 2021; Wang et al., 2021).

It must be taken into account that ICTs alone are “neutral” since they do not themselves generate negative or positive consequences. Just as a person can experience techno-stress when using ICTs, one could also experience flow and engagement; that is, a psychological state in which the user focuses with dedication on that activity with which they experience pleasant sensations and forgets their other thoughts (Sharafi et al., 2006; Barrera-Algarín et al., 2022).

Following the model of interaction between demands and resources proposed by Karasek and Theorell (1990), high levels of technostress are connected to high ICT-related demands and an individual’s lack of technological resources (Salanova, 2003). Along the same lines, the model of interaction between demands, control, and social support would be applicable (Johnson and Hall, 1988). According to this model, social support acts as a modulator of stress if the individual has the quantity and quality of social support they need. If, on the other hand, their social support is scarce or they find themselves in a situation of discrimination or intimidation, social support can become a factor that generates stress.

Social support, defined as the degree to which people’s needs for belonging, affiliation, affection, identity, security and approval are satisfied through interaction with others, can directly or indirectly affect stress and health (Clemente, 2003). Social support can promote adaptive health behaviors, provide well-being, and inhibit the negative effects of stress, and its stress-buffering effect is widely accepted by the scientific community (Barra Almagiá, 2004; González Cabanach et al., 2010b).

Thus, 3 models are proposed to explain the association between social support, stress, and health. First, the “direct effect” model, in which social support directly and positively influences health and well-being to the extent that it contributes to satisfying human needs like safety, social contact, belonging, esteem, affection, etc., regardless of the individual’s stress levels (Cohen and Wills, 1985; Pérez Bilbao and Martín Daza, 1997). In the “indirect effect” model, social support can directly reduce work stress levels and improve coping and, therefore, indirectly improve health (Forbes and Roger, 1999; Schwarzer et al., 2004). Finally, in the “buffer effect” model, social support does not have a direct effect on stress or health, but rather allows the stressful situation and coping strategies to be redefined, protecting individuals from the consequent negative effects (Strom and Egede, 2012; Fachado et al., 2013).

In general, previous studies confirm the benefits of social support on health in the university environment. The results indicate how social support is related to a better quality of life, a lower level of academic stress, and better mental health (de Almeida et al., 2018; Roming and Howard, 2019), and present variations according to both gender and source of social support. On the one hand, women tend to prefer emotional support (shows of empathy, love, and trust) over informative support (receiving useful information to deal with the problem) (Forbes and Roger, 1999). On the other, social support from peers and friends is related to better management of mental health and academic stress (Feldman et al., 2008; Muirhead and Locker, 2008; Fernández-González et al., 2015). However, more recent studies find only moderate associations between social support and degree of mental health (Pasinringi et al., 2022), and even negative associations with psychological well-being in women (Luna et al., 2020).

Since the beginning of the COVID-19 pandemic, there has been an increase in the use of ICTs. During that period the classes were held online, through google meet, email and discussion forums. Likewise, the interaction between teachers and students, and among the students themselves, was carried out through video conferences, social networks, and email. This situation could have modified ways of relating and communicating generally, key elements for the generation of social support and its impact on health and stress. That is why it is necessary to look in depth at the process by which social support exerts its effect, in the interest of the human, labor, and social development of university youth.

For this reason, this research aims to study the role of social support as a mediator in the relationship between technostress or academic stress and health in Spanish university students following the COVID-19 pandemic.

A descriptive, quantitative, cross-sectional study was conducted on university students using a self-reported survey.

The study has been carried out with a sample of the university student population over 18 years of age.

According to the Government of Spain and the Ministry of Universities (2021), the total number of students enrolled in the Spanish University System (SUE) in the 2020–2021 academic year was 1,679,518, including Bachelor, Master, and Doctorate students. For a margin of error of 5% and a 95% probability of success, with a confidence level of 95% and an accuracy of 3%, a sample of at least 213 individuals was needed. Once the term for administering the survey was over (April 30, 2022), a final sample of 389 participants was obtained.

The inclusion criteria were the following: individuals over the age of 18, of both sexes, who understood written and spoken Spanish, who have provided their informed consent and who were enrolled in the 2021/2022 academic year in a Bachelor’s, Master’s, or Doctorate program at a Spanish University.

The following instruments were used to collect the information necessary for this study:

We collected sociodemographic variables (sex, age, place of residence, marital status, cohabitation status, type of studies, university enrolled in, and work) through an ad hoc questionnaire.

Technostress questionnaire (technoanxiety and technofatigue) (created by the Work and OrganizationalNeTwork, WoNT) (Salanova et al., 2006). It consists of 26 items, which evaluate technostress as psychosocial harm with three dimensions: (1) Affective (anxiety vs. fatigue), (2) Attitudinal (skeptical attitude towards technology), and (3) Cognitive (beliefs of ineffectiveness in the use of technology). The items are responded to through a Likert-type frequency scale ranging from “0” (not at all/never) to “6” (always/every day). Obtaining high scores in these three dimensions is an indicator of technostress in its two manifestations: technoanxiety and technofatigue. To diagnose technoanxiety, high scores in anxiety, skepticism and ineffectiveness should be obtained and for technofatigue, high scores in fatigue, skepticism and ineffectiveness. High scores in any of the individual dimension does not necessarily indicate technostress, but it could develop or appear in the future if appropriate measures are not taken. The questionnaire has an adequate internal consistency, exceeding in all cases the minimum Cronbach’s Alpha score of 0.70, which ensures the validity and reliability of the measures (Salanova et al., 2006).

Goldberg General Health Questionnaire (GHQ-28). It is subdivided into 4 subscales, with 7 questions each, referring to somatic symptoms, distress/anxiety, social dysfunction, and depression (Lobo et al., 1986). According to Godoy-Izquierdo et al. (2002), the questionnaire has been taken as a positive indicator of current level of health or well-being: the higher the scores, the better the general physical and psychological health. Thus, obtaining a high score in the subscale of “physical state” or somatic symptoms indicates a good level of physical health, in the subscales of “anxiety” and “depression” indicates the absence of anxious and depressive symptoms, and in the subscale of social dysfunction or “daily well-being” indicates that one has the personal capacity to develop a healthy and functional daily life. The responses are presented in Likert format with 4 possibilities (from 0 to 3). The individual scores for each item, which, when added together, form the scores for each subscale, as well as the total (general health) obtained from the sum of the latter, were entered into the statistical analysis. The questionnaire presents good psychometric qualities with a Cronbach’s alpha score of 0.97 for the full scale, 0.93 for somatic symptoms, 0.92 for distress/anxiety, 0.91 for social dysfunction and 0.97 for depression (Godoy-Izquierdo et al., 2002).

University academic stress questionnaire (CEAU) (García-Ros et al., 2012). The potential stressors relating to the university environment were studied using a Likert-type scale with 5 possibilities (0 being no stress at all and 5 being very stressful). This questionnaire consists of 21 items grouped into 4 stress-generating factors during the university period: academic obligations, academic record and future prospects, interpersonal difficulties, and expression and communication of ideas. The items in the first factor category have to do with academic obligations that evaluate the level of academic stress in relation to the completion of compulsory tasks and assignments, academic overload, study-related activities, and performance on evaluation tests. The second factor evaluates the stress caused by anticipating future academic situations or problems such as concluding studies within the stipulated time frame, obtaining good grades, maintaining or obtaining a scholarship, or choosing subjects during the degree course. Thirdly, the factor category on interpersonal difficulties evaluates the stress generated by conflicts with professors and classmates, as well as competitiveness with the latter. The last stress-generating factor evaluates the stress produced by presentations of work, participation in class activities such as debates, and dealing with professors during tutoring hours. Thus, a high score indicates the presence of academic stress. The questionnaire presents an adequate internal consistency for the four dimensions, with a Cronbach’s Alpha value of 0.70 (García-Ros et al., 2012).

Questionnaire to measure the frequency and scope of TIC use at the educational level (CUTIC-28). This questionnaire evaluates the usefulness of ICTs and the emotion generated by their use or non-use among young university students (Jiménez Rodríguez et al., 2017). The questionnaire collects data on digital behaviors and opinions on the usefulness of ICTs in the educational setting in two media: computer (computer or laptop) or tablet, and cell phone (cell phone). It consists of 28 items distributed into two groups of 14 items and three dimensions. The items in both groups are identical, but one of the groups corresponds to computer or tablet use and the other to cell phone use. The first-dimension groups the items that measure the frequency of ICT use for games, messaging, and social networks; the second measures the usefulness of ICTs in the educational setting in relation to group work, research, classroom work, and information search; the third dimension measures the behavior/emotion generated by ICT: irritability, relaxation, and addiction. Responses are recorded in time frequency intervals (hours per day) or with a 5-point Likert response (from never to always). Following the recommendation of Jiménez Rodríguez et al. (2017), values above 2 in frequency and behavior exceed the mean value; in effect, the average is above 3.5. The questionnaire presents a good internal consistency with an alpha coefficient of 0.86 (Jiménez Rodríguez et al., 2017).

Questionnaire to measure perceived social support (Spanish version of the Social Support Questionnaire-Short Form) (Martínez-López et al., 2014). This questionnaire is composed of 6 items, representing moments of stress or need in different situations. For each item, it assesses the number of people that each individual perceives as willing to help and support him or her in a given situation, and the degree of satisfaction with this support. The items relating to the degree of satisfaction are answered on a 6-point Likert scale (very dissatisfied to very satisfied) and the number of people on a 9-point scale, with 1 being one person and 9 being nine or more people. This questionnaire measures two different aspects of perceived social support: availability and the index of satisfaction with perceived availability, calculated by averaging the scores obtained, with a maximum of 36 for the satisfaction score and 54 for availability. The questionnaire presents good psychometric properties with Cronbach’s Alpha figures of 0.89 and 0.94 (Martínez-López et al., 2014).

Data collection was carried out employing an online survey, conducted through the Google Forms platform and disseminated through social networks (LinkedIn¹ and WhatsApp²) and institutional distribution lists (the University of Zaragoza and student organizations). Finally, only those surveys that were fully answered were taken into account.

To request participation in the study, the link to the anonymous survey was sent, complying with the regulations in force regarding data protection (Organic Law 3/2018, of December 5, on Personal Data Protection and guarantee of digital rights). Google’s privacy conditions were observed, providing the link to the information and requesting prior approval as a requirement for participation in the study. Participants had to give their informed consent, being free to opt out of the survey at any time. The dissemination period took place during the months of March and April 2022.

This project has been approved by the Research Ethics Committee of the Autonomous Community of Aragon (CEICA) (no. PI22-114) and by the Data Protection Office of the University of Zaragoza (no. RAT 2022–49).

First, a descriptive analysis (frequencies for categorical variables; means and standard deviation for continuous variables) was carried out to examine the composition of the sample, and the normality or non-normality of the data was tested using the Kolmogorov–Smirnov test with the Lillierfors modification since the sample was greater than 30 cases. Second, in order to see if there were differences between men and women in the study variables, mean differences were studied using Student’s t-test for quantitative variables and chi2 (χ²) for qualitative variables. Thirdly, in order to analyze the association between the principal variables of the study (criterion, outcome, and mediator variables), correlations between them were also calculated. Finally, to determine the mediator effect between health and stress variables (technostress and academic stress), various regression analyses were carried out using the procedure designed by Baron and Kenny (1986). This procedure requires that the predictor (stress variables), criterion (health), and mediator variables (perceived social support) be positively correlated with each other. Due to the positive associations observed between health and the covariables (age, living together, studies, job, university), we decided to control for the effects of these variables by entering them in the first step of the regression. To determine whether perceived social support mediates the relationship between self-rated health and stress variables, four mediation analyses were carried out using the Process macro for SPSS (Hayes, 2018), one for each stress variable and others for women and men with the goal of seeing how the variables of interest behave in men and women separately in order to be able to later analyze gender inequalities. The bootstrapping technique with 10,000 subsamples was used to estimate the confidence interval (95%).

Tests were considered significant when p < 0.05. The analyses were carried out using IBM SPSS Statistics 25 (IBM Corp, 2017) and Stata 16 (StataCorp, 2019), both licenses from the University of Zaragoza.

The percentage of variance explained by social support in the relationship between technostress and health oscillated between 10% to 20% in both sexes when the stressor analyzed was technostress. As shown in Figures 1, 2 for men and women respectively, the analyses revealed a direct significant effect between technostress and health (β_men: −0.341, CI95%: −0.553; −0.074; β_women: −0.483, CI95%: −0.701; −0.266). Similarly, the direct effects of the mediator variable (social support) on health were significant in both sexes (β_men: 3.415, CI95%: 1.787; 5.043 β_women:1.915, CI95%: 0.571; 3.260).

In the analysis of the global model, the indirect effect of technostress on health was not significant in either sex (β_men: −0.026, CI95%: −0.113; 0.086; β_women: −0.020, CI95%: −0.084; 0.016). The covariables analyzed were also not found to influence the relationship between technostress and health in either sex.

When we took academic stress into account, the percentage of variance explained by social support in the relationship between academic stress and health oscillated between 10% to 30% in both sexes. As shown in Figures 3, 4 for men and women respectively, the analyses revealed a direct significant effect between academic stress and health (β_men: −0.469, CI95%: −0.701; −0.264; β_women: −0.443, CI95%: −0.551; −0.102). Similarly, the direct effects of the mediator variable (social support) on health were significant in both sexes and (β_men: 2.545, CI95%: 0.967; 4.123 β_women:1.407, CI95%:0.133;2.682).

In the analysis of the global model, the indirect effect of academic stress on health was significant in men (β_men: −0.303, CI95%: −0.759; −0.016); but not in women (β_women: −0.089, CI95%: −0.272; 0.019). The covariables analyzed were also not found to influence the relationship between academic stress and health in either sex.

The present study is not without limitations. As we used non-probabilistic snowball sampling, it would be interesting to use a random sampling method in order to generalize the outcomes. Since the questionnaire was disseminated online, the representativeness of the sample could not be controlled and people without internet access are automatically excluded (Arroyo Menéndez and Finkel, 2019). Due to the cross-sectional nature of the study is difficult to establish causal relationships and the associations identified might be difficult to interpret (Wang and Cheng, 2020).

It would be advisable to carry out a study with a sample that includes other universities with a greater offer of online degrees, since the University of Zaragoza, excepting the period of confinement due to the pandemic, is a face-to-face university. In addition, to extend the sample to various countries since most of the respondents belong to one of the four university campuses of the University of Zaragoza, distributed throughout the Autonomous Community of Aragón. Likewise, the starting age of the studies completed should be analyzed, since the average age is relatively high (25 years), as should the course that the participants are taking. It also must be taken into account that, in the present study, no differentiation has been made between in-person social support and social support through ICTs.

The present study is one of the first to analyze how social support mediates the relationship between technostress and academic stress and self-perceived health, which is why the data must be interpreted with caution and more multi-center studies with larger samples are necessary to be able to corroborate these results. In addition, there are other variables such as personality traits and resilience, which could also influence the present results (Conti et al., 2018; Zhang et al., 2018; Hu et al., 2020).