Understanding Technostress Worldwide: Scientific Growth, Thematic Clusters, and Implications for Social Sustainability and Organizational Resilience

Abstract

Technostress has emerged as a central psychosocial risk in digital environments, yet existing bibliometric research provides only partial accounts of its development. This study offers an updated mapping of the global intellectual structure of technostress based on 1135 articles indexed in the Web of Science Core Collection as of October 2025. Using Price’s, Lotka’s, Bradford’s, and Zipf’s Laws, alongside co-authorship and keyword co-occurrence analyses, the study examines patterns of scientific growth, author productivity, and conceptual concentration. Results reveal exponential growth, highly unequal authorship distribution, and strong terminological concentration. Four stable thematic clusters were identified: foundational theory, workplace technostress, psychological consequences, and excessive technology use. Emerging topics, including artificial intelligence, digital stress, and information overload, indicate a shift toward broader socio-technical interpretations. Keyword networks further reveal a consistent relationship between technological overload, technostress, impaired performance, and reduced well-being. Despite its growing maturity, the field remains conceptually fragmented and lacks an integrative framework encompassing cognitive, behavioral, organizational, and AI-mediated dimensions. These findings highlight the need for interdisciplinary collaboration and interventions that address both individual capacities and contextual conditions to advance digital well-being and foster socially sustainable and resilient digital environments.

1. Introduction

Since its origins, technostress has been understood as a form of psychological and organizational strain arising from difficulties in adapting to technological demands in a healthy and sustainable manner. Brod (1982) described it as a condition shaped by factors such as age, previous experience, perceived control, and organizational climate, with consequences for performance and the effective use of technology. During the 1990s, research such as Arnetz and Wiholm (1997) showed that the rapid introduction of technologies into workplaces increased cognitive demands, generated psychosomatic symptoms, and heightened mental stress, particularly when employees lacked sufficient skills or organizational recognition. From the 2000s onward, the field acquired greater theoretical consolidation through the work of Tarafdar and colleagues, who identified the classical technostress creators (overload, invasion, complexity, insecurity, and uncertainty) and demonstrated that technostress reduces productivity, increases role stress, and decreases satisfaction with information systems, although these effects can be mitigated through user participation, innovation support, and strategies aimed at reducing role conflict (Tarafdar et al. 2007, 2010). Parallel studies showed that centralization, pressure for continuous technological updating, and high-performance expectations increase technostress (Wang et al. 2008), while research in education demonstrated that teachers experience technology-related stress linked to the fit between pedagogical demands and technological environments (Al-Fudail and Mellar 2008). Collectively, this early literature established technostress as a multidimensional phenomenon encompassing cognitive, emotional, organizational, and behavioral components.

The rapid expansion of digital technologies has since transformed social, educational, and organizational life, creating new opportunities for efficiency and connectivity while also generating emerging risks to well-being. Within this broader process of digitalization, technostress, understood as the stress experienced when technological demands exceed individuals’ cognitive, emotional, or organizational resources, has become a central construct for examining how digital environments shape human behavior and social outcomes (Tarafdar et al. 2019). Contemporary digital ecosystems introduce additional pressures such as information saturation, constant availability, surveillance, erosion of privacy, and social isolation, intensifying the psychological and organizational relevance of technostress.

Over the last decade, research has advanced toward increasingly complex models that integrate technological demands, personal and organizational resources, and psychological, physiological, and behavioral responses. From the Job Demands–Resources (JD-R) perspective, recent studies show that technostress creators deplete workers’ emotional and physical resources, reducing well-being and job satisfaction, although knowledge diversity, skill flexibility, and virtual leadership may buffer these effects (Wang et al. 2023; Liu et al. 2024; Girardi et al. 2025). In enterprise systems, technostress has been conceptualized as a set of technology-driven stressors that may operate as challenges or hindrances, influencing workarounds and being moderated by organizational resources and personal resilience (Tang et al. 2024). From the Stimulus–Organism–Response (SOR) paradigm, social media design and information overload have been shown to generate anxiety, exhaustion, and information avoidance (Kumalasari and Priharsari 2023; Pang et al. 2025), while expectation disconfirmation theory explains discontinuance intentions when dissatisfaction and anxiety increase (Ma et al. 2022). Studies of older adults reveal that smartphone use can generate specific forms of strain influenced by work scheduling autonomy and contextual conditions (Van Fossen et al. 2023). In environments mediated by generative artificial intelligence, technostress reduces flow and continuance intentions while increasing switching intentions, although autotelic personality may provide psychological resilience (Chong et al. 2025). Psychometric and organizational studies have validated scales assessing technostress creators and inhibitors and demonstrated that different technological demands (emotional and functional) produce differentiated responses depending on factors such as task interdependence and ICT dependence (Kot 2022; Techmanska et al. 2024).

These effects are not uniformly distributed across individuals or contexts. In workplaces, older or more experienced workers report higher techno-complexity, while gender influences perceptions of specific technostress dimensions (Marchiori et al. 2018). Among older adults, technostress represents a significant threat to well-being (Nimrod 2017), and age moderates the relationship between interruptions and stress (Tams et al. 2018). In remote work environments, leadership style is a critical moderator: authoritarian leadership amplifies technostress, whereas supportive leadership attenuates it (Spagnoli et al. 2020). In education, technostress arises when institutional expectations misalign with students’ competencies, increasing burnout and undermining academic performance (Wang et al. 2020). Individual characteristics such as technological self-efficacy, frequency of use, and digital literacy further shape technostress responses (Qi 2019). In everyday digital life, social media use generates tensions affecting concentration, sleep, identity, and social relationships (Salo et al. 2019), while different forms of overload influence discontinuance intentions (Chen and Wei 2019). These findings illustrate how digitalization blurs boundaries between work, leisure, and personal life, generating cross-cutting pressures that require both individual and structural responses (Bermes 2021).

Overall, technostress has evolved into a complex, multidimensional research domain encompassing psychological, behavioral, organizational, physiological, educational, and social dimensions. Digital technologies simultaneously enhance productivity, connectivity, learning, and innovation while posing risks to well-being, performance, and social cohesion. Against this background, this study uses bibliometric analysis to examine the global evolution, intellectual structure, and thematic development of technostress research, identifying emerging priorities and providing evidence to advance digital well-being and organizational resilience.

2. Materials and Methods

Bibliometric analysis operates under an epistemic logic that differs fundamentally from traditional review methodologies. Whereas narrative, systematic, mixed, rapid, exploratory, and integrative reviews rely on qualitative screening, exclusion, and synthesis of selected studies, bibliometrics emphasizes exhaustive search rigor and replaces subjective evaluation with quantitative modeling of the entire retrieved corpus (Grant and Booth 2009). Rather than reducing the dataset through interpretive filtering, bibliometric methods apply eponymous laws based on subsampling, estimation, and mathematical calculation to reveal structural patterns in scientific production, authorship, and conceptual development (Valderrama-Zurian et al. 2019). Consequently, bibliometrics is designed to map the intellectual structure of a field and to identify its social, conceptual, and historical dynamics, rather than to assess the substantive findings of individual studies.

Using this approach, a dataset was retrieved from the Web of Science Core Collection (WoSCC) on 30 October 2025. The search employed the thematic vector TS = (Technostress), which performs a simultaneous query across titles, abstracts, author keywords, and Keywords Plus^® (Clarivate 2025). This strategy ensures a high level of comprehensiveness and minimizes biases arising from partial or subjective searches. By capturing all relevant instances of the term, the resulting corpus constitutes a robust representation of global scientific output on technostress and provides a reliable foundation for analyzing its conceptual evolution and intellectual structure. WoSCC remains the preferred choice for conducting reproducible bibliometric analyses, as its standardized metadata supports classical laws and conceptual consistency. Furthermore, the creation of the ESCI in 2015 significantly expanded the number of indexed journals (Zhang et al. 2020).

Following the guidelines for advancing theory and practice through bibliometric research proposed by Mukherjee et al. (2022), the analysis combined two complementary components: performance analysis and science mapping. Performance analysis was conducted using eponymous bibliometric laws as outlined by Haddow (2018) and Valderrama-Zurian et al. (2019), specifically the formulations of Price (1976), Lotka (1926), Bradford (1934), and Zipf (1932). Science mapping focused on co-authorship networks and conceptual structures, using VOSviewer version 1.6.20 (Van Eck and Waltman 2010) and IBM SPSS Statistics 23 (IBM Corp., Armonk, NY, USA).

Price’s Law was applied to examine the exponential growth of scientific output, assessing whether annual publication trends align with an exponential curve that reflects the accumulation of a critical mass of knowledge. This law also enables the study of literature obsolescence by dividing the corpus into two semi-periods based on the median publication year, distinguishing contemporary from outdated works and identifying classical literature that remains highly cited despite its age (Price 1976; Dobrov et al. 1979). Its application allows for the evaluation of the field’s maturity by determining whether recent production consistently exceeds historical production, indicating thematic renewal and sustained expansion. This temporal distinction is fundamental for understanding how conceptual nuclei evolve and which work maintains structural influence over time.

Lotka’s Law was used to analyze author productivity patterns. This law identifies the distribution of contributions across researchers, distinguishing a small group of highly productive authors from a large base of occasional contributors. The square root of the total number of authors was used as an initial estimate of the prolific core, subsequently adjusted according to discrete publication thresholds (Lotka 1926; Nicholls 1988; Tsai 2013). This analysis characterizes the productive structure of the field and evaluates its degree of intellectual concentration, offering an indirect measure of disciplinary maturity and revealing collaborative dynamics that influence thematic consolidation.

Bradford’s Law was applied to examine the distribution of scientific publications across journals. This law identifies a core set of journals—the Bradford nucleus—that contains approximately one-third of all publications on a given topic, followed by successive zones comprising progressively larger numbers of journals that contribute similar proportions of publications (Bradford 1934; Bulick 1978; Desai et al. 2018). The Bradford nucleus is particularly relevant because it concentrates on the principal publication outlets and the most specialized authors, reviewers, and editors within a research field, thereby revealing the structural organization of its intellectual production.

Zipf’s Law was applied to Keywords Plus^® to examine the concentration of conceptual terminology within the corpus. Using the square root of the total keyword set as an estimate, the analysis identified the most prominent terms that structure the field’s conceptual landscape, reflecting the typical inverse distribution described by Zipf (1932) and later bibliometric applications (Merediz-Solà and Bariviera 2019). This approach allows for the evaluation of thematic cohesion, distinguishing between highly recurrent terms that articulate the conceptual core and a broad periphery of less frequent words, thereby revealing both terminological stability and ongoing exploration of new research areas.

The Web of Science Sustainable Development Goal (SDG) classification was used to assess the extent to which technostress research contributes to the Sustainable Development Goals, providing an indication of its broader societal and policy relevance. SDG classifications were obtained directly from the Web of Science metadata, where sustainability labels are assigned using automated classification algorithms (Clarivate 2025; Sianes et al. 2022).

Table 1 summarizes the characteristics of the analyzed corpus, which comprises 1135 articles published between 1982 and 2025, authored by 3102 researchers and indexed with 1486 Keywords Plus^®. The application of Price’s, Lotka’s, Bradford’s, and Zipf’s Laws revealed classical patterns of scientific growth, author dispersion, source concentration, and conceptual terminology, providing a robust foundation for understanding the evolution and thematic structure of global technostress research. This mapping provides structural evidence relevant for developing social sustainability indicators associated with digital well-being and organizational resilience.

Table 1. Description of the analyzed bibliographic corpus.

Variable	Value/Sample (n)	Unit of Analysis	Subsampling Criterion
Time period	1982–2025	Years	Price’s Law
Authors identified	3102	Authors	Lotka’s Law
Source identified	482	Journals	Bradford’s Law
Keywords Plus®	1486	Words	Zipf’s Law

3. Results

The scientific production on technostress began in 1982 with Brod’s seminal article “Managing Technostress: Optimizing the Use of Computer Technology” (Brod 1982). However, continuous annual output was not achieved until 2004, marking the point at which the field reached a critical mass of knowledge production. For this reason, the analysis of scientific growth focuses on the period 2004–2024, comprising 865 articles, which ensure statistical stability and avoid distortions caused by isolated early publications.

Table 2 summarizes the adjustment tests for the exponential and logistic growth models, while Figure 1 visualizes annual publication trends. Both models show a highly robust fit, with R = 0.965 and R² ≈ 0.93, indicating that more than 93% of the variance in publication output is explained by time. The standard error (~0.49) and the significance of the ANOVA (p = 0.000) further confirm the reliability of both models.

Table 2. Modeling the temporal growth of the analyzed bibliographic corpus.

Statistic	Exponential Model	Logistic (S) Model
R	0.965	0.965
R2	0.932	0.931
Adjusted R2	0.928	0.928
Std. Error of Estimate	0.489	0.492
F (ANOVA)	260.321	256.994
Sig. (ANOVA)	0.000	0.000
Predictor Coefficient	B = 0.284 (PY)	B = −1,152,128 (1/PY)
Constant	2.847 × 10−248	574.393
Dependent Variable	ln(ART) *	ln(ART) *

* ln is natural logarithm, ART: published articles, and PY: publication year.

Figure 1. Published articles in WoSCC on technostress (2004–2024).

The exponential model describes a pattern of continuous and accelerated growth, consistent with the expansion of digitalization and the increasing societal relevance of technostress as a psychosocial and organizational phenomenon. The logistic model, although statistically equivalent, implies a distant saturation point whose magnitude is so disproportionate that it does not reflect the current dynamics of the field. Thus, the exponential model better captures the historical trajectory, while the logistic model merely anticipates a long-term maturation phase.

Figure 1 illustrates the annual distribution of publications, highlighting the contemporary half period and the fitted exponential and logistic curves. Visualization shows a clear acceleration in scientific output, suggesting that technostress has become an increasingly central topic for understanding the social and psychological consequences of digitalization.

The observed points represent the total number of articles published on technostress each year, with the lighter colored point corresponding to the contemporary half-period, delimited by a segmented red line. The segmented lines in blue and green represent adjusted logistical and exponential trends. Furthermore, this visualization allows for an intuitive comparison of the stability of annual growth with the behavior projected by both models, showing how the adjusted trends capture the acceleration of the field and signal potential turning points associated with emerging digital environments such as remote work, AI-mediated systems, and pervasive information ecosystems.

3.1. Authorship Structure: Lotka’s Law

The application of Lotka’s Law to the relationship between articles (ART) and authors (AU) reveals a highly consistent power law distribution. With 3102 authors contributing to 1135 articles, the model achieves an R² of 0.954, indicating that more than 95% of the variability in author productivity is explained by the number of articles (Table 3). The F statistic (290.090, p = 0.000) confirms the robustness of the model.

Table 3. Model Summary and Parameter Estimates for Lotka Law.

Lotka Law	Model Summary					Parameter Estimates
Equation	R Square	F	df1	df2	Sig.	Constant	b1
Power	0.954	290.090	1	14	0.000	1514.478	−2.757

Dependent variable: AU, and the independent variable is ART.

The estimated parameters (constant = 1514.478; exponent b1 = −2.757) reflect the classic Lotkian pattern: a small group of highly productive authors and a large base of occasional contributors. The prolific core is estimated at 56 authors (SQRT (3102)), adjusted in practical terms to 80 authors with four or more publications. Among them, Monideepa Tarafdar stands out with 21 contributions, confirming her central role in defining the theoretical foundations of technostress.

Figure 2 visualizes this asymmetry, showing the concentration of output among a small group and the long tail of occasional authors (2653 authors with only one article). This structure indicates a mature yet expanding field, where a stable core of prolific authors anchors the conceptual development of technostress, while the continuous entry of new researchers contributes to its interdisciplinary diversification.

Figure 2. Authors by published articles in WoSCC on technostress (1983–2025).

The observed points represent the total number of authors by articles published on technostress, with the lighter colored point corresponding to the occasional authors. The segmented lines represent adjusted power models. This graphical representation clearly shows the marked asymmetry in scientific productivity, highlighting the concentration of contributions in a small group of prolific authors. Likewise, the adjustment using power models confirms the structural stability of this distribution over time, underscoring how technostress research combines consolidation around foundational scholars with ongoing thematic expansion driven by diverse disciplinary perspectives.

3.2. Co-Authorship Networks and Thematic Specialization

Figure 3A,B depict the co-authorship network of the 80 prolific authors and its temporal evolution (Details in Appendix A, Table A1). The field is organized around a foundational nucleus that established the core dimensions and theoretical models of technostress (Tarafdar et al. 2019, 2020; Salo et al. 2019). From this nucleus, four specialized clusters emerge:

• Foundational Cluster: Defines the conceptual dimensions of technostress and its theoretical models, serving as the intellectual anchor for subsequent thematic developments.
• Work Cluster: Focuses on organizational applications, work performance, and technostress management (Maier et al. 2019; Bondanini et al. 2020; Laumer et al. 2017).
• Consequences Cluster: Examines psychological and behavioral outcomes, including burnout and stress responses (Spagnoli et al. 2020; Pirkkalainen et al. 2019; Riedl 2022).
• Excessive Use Cluster: Investigates compulsive use, addiction, and maladaptive digital behaviors (Wang et al. 2020; Cao et al. 2018; Cao and Yu 2019).

Figure 3. (A) Co-authorship graph. (B) Co-authorship temporal graph.

Figure 3. (A) Co-authorship graph. (B) Co-authorship temporal graph.

The network shows strong interconnections between clusters, indicating that specialization does not fragment the field. Instead, shared theoretical foundations enable the circulation of constructs and models across subareas, reinforcing the coherence and interdisciplinarity of technostress research. The temporal graph (Figure 3B) highlights the emergence of new authors after 2022, reflecting the growing societal relevance of technostress within accelerated digitalization, remote work, AI-mediated environments, and pervasive information ecosystems.

3.3. Source Structure: Bradford’s Law and Source Zones

The articles were published in a total of 482 journals, indicating a low concentration of scientific output (See Table 4). However, a significant portion of the publications is concentrated in a core group of 22 specialized journals, identified through the application of Bradford’s Law, as detailed in the corresponding Bradford zones.

Table 4. Bradford’s Zones.

Zone	Number of Articles	%	Journals (Empirical Series)	%	Bradford Multipliers	Journals (Theorical Series)
Nucleus	380	33%	22	5%		22 × ($n^0$) = 22
Zone 1	429	38%	134	28%	6.09 = 134/22	22 × ($n^1$) = 94
Zone 2	326	29%	326	68%	2.43 = 326/134	22 × ($n^2$) = 400
Total	∑ = 1.135	100%	∑ = 482	100%	n = 4.26	∑ = 515
% error $(e_p)$ = $-$6.92%

The percentage error between empirical and theoretical series is expressed by Equation (1).

$$e_p=\left({\displaystyle \frac{Empirical-Theorical}{Empirical}}\right)\ast 100=\left({\displaystyle \frac{482-515}{482}}\right)\ast 100=-6.92\%$$

(1)

The percentage error between the empirical value (482) and the theoretical value (515) was −6.92%, indicating that the theoretical estimate overestimates the observed value by approximately 7%. This difference suggests a reasonably good agreement between the two values, although the model shows a slight tendency to overestimate the phenomenon under analysis. Table 5 shows that scientific research on technostress is concentrated in high-impact, interdisciplinary journals. Among the top 22 journals, the following stand out: Computers in Human Behavior (41 articles; 5081 citations), Frontiers in Psychology (37; 699), Information Technology and People (36; 1029), and the International Journal of Environmental Research and Public Health (27; 867). The predominant fields are Psychology, Information Sciences, Computer Systems, Management, and Social Sciences. Most of the journals are ranked in Q1, including publications with particularly high impact factors such as the International Journal of Information Management (31.0), Technological Forecasting and Social Change (13.5), and Technology in Society (12.9). Taken together, these results demonstrate a well-established and interdisciplinary body of work, in which technostress is addressed from psychological, technological, organizational, educational, and social perspectives, reflecting the field’s consolidation around a stable set of high-impact outlets.

Table 5. Journals in the Bradford nucleus.

Journals	Articles	Citations, WoS Core	Time (Start–End)	WoS Categories	JIF 2025 (WoS) *	Quartile of Impact (Qx)
Comput. Hum. Behav.	41	5.081	2008–2025	Psychol., Exp.; Psychol., Multidisc.	12.2	Q1
Front. Psychol.	37	699	2018–2025	Psychol., Multidisc.	3.8	Q1
Inf. Technol. People	36	1.029	2013–2025	Inf. Sci. Libr. Sci.	6.3	Q1
Int. J. Environ. Res. Public Health	27	867	2020–2023		N.A. **	N.A.
Behav. Inf. Technol.	26	337	2016–2025	Comput. Sci., Cybern.; Ergonom.	4.2	Q1
Sustainability	21	740	2017–2025	Environ. Sci.; Environ. Stud.; Green Sustain. Sci. Technol.	4.1	Q2
Int. J. Hum.-Comput. Interact.	19	434	2011–2025	Comput. Sci., Cybern.; Ergonom.	6.1	Q1
Internet Res.	16	444	2017–2025	Bus.; Comput. Sci., Inf. Syst.; Telecommun.	7.2	Q1
Educ. Inf. Technol.	15	237	2021–2025	Educ. Educ. Res.	7.2	Q1
Inf. Manage.	14	918	2016–2025	Comput. Sci., Inf. Syst.; Inf. Sci. Libr. Sci.; Manag.	7.0	Q1
PLoS ONE	12	143	2017–2025	Multidiscip. Sci.	2.8	Q2
Hum. Soc. Sci. Commun.	12	86	2021–2025	Soc. Sci., Interdiscip.	4.8	Q1
Technol. Forecast. Soc. Chang.	11	120	2015–2025	Bus.; Reg. Urban Plan.	13.5	Q1
Acta Psychol.	11	63	2023–2025	Psychol., Exp.	3.8	Q1
Behav. Sci.	11	77	2021–2025	Psychol., Multidisc.	3.2	Q1
Inf. Syst. J.	11	1623	2013–2025	Inf. Sci. Libr. Sci.	8.2	Q1
Int. J. Inf. Manage.	10	1121	2019–2025	Inf. Sci. Libr. Sci.	31.0	Q1
J. Manage. Inform. Syst.	10	2242	2007–2022	Comput. Sci., Inf. Syst.; Inf. Sci. Libr. Sci.; Manag.	7.2	Q1
J. Bus. Res.	10	670	2019–2025	Bus.	11.1	Q1
Eur. J. Inform. Syst.	10	1455	2015–2024	Comput. Sci., Inf. Syst.; Inf. Sci. Libr. Sci.; Manag.	8.1	Q1
Technol. Soc.	10	342	2017–2025	Soc. Sci., Interdiscip.; Soc. Issues	12.9	Q1
SAGE Open	10	59	2021–2025	Soc. Sci., Interdiscip.	2.4	Q1

* JIF: Journal Impact Factor, N.A.: Not Available, ** Last JIF in 2021 (4.6 and Q1).

The concentration of output in a small Bradford nucleus also indicates that technostress research is anchored in journals that shape global debates on digitalization, mental health, organizational behavior, and sustainability. This structural pattern reinforces the idea that technostress is a boundary-spanning topic whose theoretical development depends on contributions from multiple disciplines.

3.4. Conceptual Structure: Zipf’s Law and Keyword Co-Occurrence

The application of Zipf’s Law to Keywords Plus^® reveals a solid fit (R² = 0.796; F = 234.029, p = 0.000), confirming an inverse distribution typical of Zipfian patterns (Table 6). A small group of 39 high-frequency keywords (SQRT (1486)) accounts for a large portion of the articles, while 846 keywords appear only once.

Table 6. Model Summary and Parameter Estimates for Zipf Law.

Zipf Law	Model Summary					Parameter Estimates
Equation	R Square	F	df1	df2	Sig.	Constant	b1
Power	0.796	234.029	1	60	0.000	81.067	−0.764

Dependent variable: ART, and the independent variable is KWP.

The term technostress appears in 491 articles, representing the conceptual core of the field (Figure 4). This structure reflects a stable conceptual nucleus around classical categories, while the long tail of low-frequency terms signals ongoing thematic diversification and the emergence of new research directions.

Figure 4. Keywords plus^® by published articles in WoSCC on technostress (1983–2025).

Figure 5A,B show the keyword co-occurrence network, revealing three major thematic groups:

• Central Group (Red): Causes, Consequences, and Impact includes Technostress, Impact, Information, Technology Overload, and Dark Side. Represents the causal chain linking digital demands to negative outcomes and anchors the theoretical foundations of the field.
• Outcomes and Resources Group (Blue): Human Responses includes Job Performance, Mental Health, Stress, Burnout, and Resources. Highlights the interplay between demands and personal/organizational buffers, reflecting the shift toward models such as JD-R and digital well-being.
• Technological Context Group (Green): Digital Environment and Interaction includes Information Technology, Technology, Acceptance, and Adaptation. Represents the contextual conditions that trigger technostress and connects the field with research on digital transformation and technology adoption.

Figure 5. (A) Co-occurrence graph of keywords plus^®. (B) Co-occurrence temporal graph of keywords plus^®.

Figure 5. (A) Co-occurrence graph of keywords plus^®. (B) Co-occurrence temporal graph of keywords plus^®.

The temporal co-occurrence graph (Figure 5B) shows that contemporary research increasingly focuses on engagement, mental health, intention, users, resources, performance, self-efficacy, and burnout, indicating a conceptual shift toward understanding technostress as a multidimensional social, psychological, and organizational phenomenon rather than solely a technological one. These thematic structures reveal patterns that are directly relevant to the social sustainability of digitalized organizations and communities, particularly regarding how digital demands, resource distribution, and psychosocial risks shape community resilience and equity in digital environments.

3.5. Concentration of Contributions by Sustainable Development Goal

Additionally, the classification of impact on the Sustainable Development Goals (SDGs) carried out by Web of Science for this research on technostress does not seem to deviate from the set of 1135 articles considered in the study. The SDGs most impacted are SDG03—Good Health and Well Being (274 articles), SDG04—Quality Education (174 articles), SDG05—Gender Equality (55 articles), SDG10—Reduced Inequality (52 articles), and SDG09—Industry Innovation and Infrastructure (36 articles). Figure 6 shows details of the contribution assigned by WoSCC of these 1135 articles to the various SDGs (note that an article could impact more than one SDG). It is important to note that SDG attribution in WoS reflects automated thematic classification rather than explicit sustainability conceptualization within each article. Figure 6 shows the number of articles associated with each SDG (radial axes), with a breakdown of the top 5.

Figure 6. SDG concentration of contributions.

This distribution reveals that technostress research aligns most strongly with global priorities related to mental health, educational equity, gender disparities, social inequality, and digital infrastructure. The prominence of SDG03 and SDG10 underscores the relevance of technostress as a psychosocial risk factor that affects well-being and contributes to unequal digital burdens across populations. Likewise, the concentration in SDG04 and SDG05 highlights how technostress intersects with educational digitalization and gendered patterns of technology use, while SDG09 reflects the field’s connection to innovation ecosystems and the challenges of digital transformation.

Taken together, these patterns indicate that technostress research contributes to the social dimensions of sustainability by illuminating how digital demands, resource distribution, and psychosocial risks shape well-being, equity, and resilience in digitalized societies.

3.6. Semantic Patterns and Emerging Research Trends

Additionally, a textual data analysis of the abstracts of all articles was conducted using VOSviewer, excluding structured labels and copyright statements. After extracting 18,632 terms, a binary counting procedure was applied at the document level, identifying 4004 terms with two or more occurrences. Three relevant patterns were distinguished: (1) the presence of occupational profiles that provide greater specificity to the broader categories of work (employee and worker) and education (teacher and student); moreover, approximately twenty studies focused on the health sector, particularly occupations such as nurse and physician (Bail et al. 2023; Ficapal-Cusí et al. 2025; Sinmaz et al. 2025); (2) the growing incorporation of artificial intelligence, addressed in more than fifty studies (Chang et al. 2024; Lițan 2025; Fan et al. 2026); and (3) more than one hundred articles that, through structural equation modeling, examine technostress in relation to other measurement scales. Within this latter group, the combined use of UTAUT and TAM is particularly prominent (Khan et al. 2024; Yaseen et al. 2025; Saeed et al. 2025), while TPACK is frequently incorporated in educational contexts (Kohnke et al. 2024; Güner et al. 2025; Wang et al. 2025).

These patterns reveal how technostress research has progressively expanded beyond general categories of “worker” or “student” toward more granular occupational profiles, particularly in health-related professions where digitalization introduces acute cognitive and emotional demands. Likewise, the growing presence of artificial intelligence reflects a thematic shift toward AI-mediated technostress, signaling the emergence of new stressors associated with algorithmic systems, automation, and generative AI. Finally, the widespread use of structural equation modeling (especially through UTAUT, TAM, and TPACK) demonstrates the consolidation of technostress within established theoretical frameworks for technology acceptance, digital competence, and behavioral intention, reinforcing its position as a multidimensional construct at the intersection of psychology, education, and information systems.

4. Discussion

The findings of this study reveal a research field that has expanded rapidly while remaining conceptually fragmented, exposing structural tensions that earlier scientometric analyses did not fully capture. Bondanini et al. (2020) provided the first large-scale scientometric assessment of technostress, but their focus on workplace risks offered only a partial view of a phenomenon that would soon diversify across educational, social, and organizational contexts. Salazar-Concha et al. (2021) broadened the scope but still characterized technostress as an emerging field due to the limited size of the corpus available at the time. In contrast, the present analysis, based on more than one thousand publications, demonstrates that these early assessments underestimated both the scale and the internal heterogeneity of technostress research. The field has entered a phase of accelerated expansion while remaining epistemically concentrated, a combination that explains the coexistence of rapid thematic proliferation and persistent conceptual fragmentation.

Authorship patterns reinforce this tension. Although the distribution follows Lotka’s Law (Lotka 1926), with Tarafdar et al. (2019) and a small group of prolific authors shaping the theoretical core, the surge of new contributors after 2020 has intensified thematic diversification without producing conceptual convergence. This mirrors patterns observed in adjacent digital-behavior fields, such as affective computing (Ho et al. 2021), where rapid growth generated innovation but also thematic drift. The four clusters identified in this study reflect divergent assumptions about whether technostress is primarily a cognitive overload, a psychosocial hazard, or a behavioral dysregulation (Wang et al. 2020; Cao and Sun 2018; Spagnoli et al. 2020; Salo et al. 2019; Cao and Yu 2019; Pirkkalainen et al. 2019; Maier et al. 2019; Laumer et al. 2017; Mache and Harth 2020; Riedl 2022). Neither Bondanini et al. (2020) nor Salazar-Concha et al. (2021) resolved these divergences, and contemporary studies continue to treat these dimensions in isolation (Arya et al. 2025; Akar et al. 2024; Li et al. 2024; Louzán and Torrano 2024). This pattern suggests an epistemic dependency on a small set of conceptual anchors that stabilizes communication but constrains theoretical pluralism, producing parallel research trajectories that rarely converge.

Recent bibliometric studies indicate a clear conceptual reorientation of technostress research. The field is consolidating around information systems, stress theory, technology acceptance, adaptation, and organizational psychology while incorporating constructs such as techno-eustress, techno-distress, and digital stress (Temur et al. 2026; Almakrob and Alduais 2026). Artificial intelligence is accelerating this diversification through emerging stressors, including AI anxiety, job insecurity, and role ambiguity (Hossain et al. 2026; Rani et al. 2026). At the organizational level, AI-supported HRM introduces concerns regarding algorithmic transparency, fairness, data governance, and personalization-related technostress (Ubeda-Garcia et al. 2025). Meanwhile, information overload research increasingly emphasizes algorithmic filtering, recommender systems, and digitally mediated work environments (Schmutzer and Vrabcová 2026). Technostress emerges as a multidimensional socio-technical phenomenon shaped by AI, digital transformation, workplace changes, well-being, adaptation, and resilience.

Educational research further intensifies this fragmentation. Li et al. (2024) and Louzán and Torrano (2024) show that technostress surged during the pandemic, yet their analyses remain focused on teachers and students, generating population-specific insights without fully integrating educational, psychological, organizational, and socio-technical dimensions into a broader explanatory framework for understanding digital stress and adaptation processes. Özmen and Yalçin (2026) demonstrate that the COVID-19 period produced a field-level reorientation toward mental health, psychological well-being, social isolation, technostress, and resilience. This shift is reflected in the keyword network, where terms such as engagement, intention, and self-efficacy coexist with burnout and overload without a clear explanatory hierarchy. These population-specific foci provide valuable empirical detail but hinder the emergence of cross-population constructs and cumulative theory building.

Organizational studies also reveal unresolved contradictions. Arya et al. (2025) conceptualizes technostress as a homogeneous construct, whereas Akar et al. (2024) situates it within the Job Demands–Resources model, emphasizing implications for occupational health standards such as ISO 45003:2021(en) International Organization for Standardization (2021). Marino and Capone (2021) further complicate this picture by showing that smart working can either enhance or undermine well-being depending on the alignment between technological demands, autonomy, and organizational support. Rani et al. (2026) add that AI maturity and employee resilience are critical moderators of whether AI functions as a resource or a demand. Yet none of these studies fully explain the dual nature of technostress, which can generate both strain and innovation (Tarafdar et al. 2019). Addressing this gap requires models that explicitly specify mediators (e.g., perceived control, cognitive load) and moderators (e.g., leadership style, institutional support) to predict boundary conditions for harm versus adaptation.

Coping research adds another layer of inconsistency. Reactive strategies such as venting and distancing coexist with proactive strategies such as reinterpretation and control (Pirkkalainen et al. 2019), but existing studies do not clarify when each strategy is adaptive or how coping interacts with contextual moderators such as leadership style or institutional support (Spagnoli et al. 2020; Özgür 2020). This omission is reflected in the bibliometric structure, where resources appear as a central node but remain theoretically underdeveloped. Clarifying the adaptive value of coping strategies requires longitudinal and experimental designs that test interactions between individual strategies and organizational resources, moving beyond cross-sectional associations.

Population-specific vulnerabilities also remain insufficiently integrated. Studies consistently identify heightened susceptibility among students, teachers, older adults, and individuals with anxiety or compulsive tendencies (Wang et al. 2020; Nimrod 2017; Horwood and Anglim 2019). The textual analysis conducted in this study further reveals occupational profiles, particularly nurses and physicians, that face acute cognitive and emotional demands associated with digitalization. From a sustainability perspective, these unresolved tensions weaken the resilience of organizations and communities, as technostress undermines well-being, equity, and the capacity to adapt to digital transformation. This is consistent with the SDG distribution observed in this study, where SDG03 (health), SDG04 (education), SDG05 (gender), SDG10 (inequality), and SDG09 (digital infrastructure) emerge as central domains affected by technostress.

Taken together, the reviewed studies and earlier bibliometric analyses (Bondanini et al. 2020; Salazar-Concha et al. 2021; Ho et al. 2021; Li et al. 2024; Louzán and Torrano 2024) contribute valuable empirical insights but also reveal a field marked by conceptual fragmentation, inconsistent theoretical grounding, and limited integration across populations and contexts. The persistent tension between well-being and productivity remains unresolved because technostress is simultaneously a psychological strain, a behavioral pattern, and a structural feature of digital environments. By mapping the structural relationships among themes, authors, and concepts, this study clarifies these tensions and underscores the need for a conceptual framework that integrates technological demands, psychological resources, organizational conditions, and population-specific vulnerabilities. Such integration is essential for supporting healthier, more equitable, and more resilient digital environments in the context of accelerated digitalization.

5. Conclusions

This bibliometric analysis provides a comprehensive mapping of the intellectual structure of technostress and demonstrates its consolidation as a significant field of inquiry within the broader study of digitalization and social change. The results reveal a consistent causal chain across thematic clusters: technological overload increases cognitive and emotional demands, which in turn trigger technostress and negatively affect performance and mental health. This pattern confirms that technostress constitutes a central psychosocial risk in contemporary digital environments, where the boundaries between work, education, and personal life are increasingly blurred. The integration of recent research trends, particularly those related to artificial intelligence, digital stress, and information overload, further indicates that technostress is evolving into a multidimensional socio-technical phenomenon shaped by algorithmic systems, hybrid work arrangements, and accelerated digital transformation.

A key academic implication is the need to strengthen interdisciplinary collaboration. Although the field has grown rapidly and diversified, conceptual fragmentation persists, limiting the development of integrative models. Future theoretical frameworks must explicitly articulate the contributions of Information Systems (focused on design, usability, and technological architecture) with those of Psychology, centered on cognition, emotion, and well-being. Incorporating insights from organizational studies, education, and emerging AI-related research is also essential for capturing the full spectrum of digital demands and resources. Only through this convergence will it be possible to move beyond risk characterization toward unified theories that explain how digital technologies generate both strain and opportunities for human development. This requires expanding current models to include mediators such as cognitive load and perceived control, and moderators such as leadership style, institutional support, and AI maturity. Such integration will enable the field to advance toward a more comprehensive understanding of digital well-being as a core research and policy objective.

From a practical standpoint, the findings offer clear guidance for educational institutions, labor organizations, and policymakers. Educational institutions can reduce technostress by implementing progressive digital literacy programs, strengthening teacher support, and adopting technological platforms aligned with the actual capabilities of students and educators. Labor organizations can conduct periodic assessments of digital psychosocial risks, redesign workflows to prevent information overload, and establish clear rules for digital disconnection. In environments increasingly mediated by artificial intelligence, organizations should also evaluate AI-related stressors, promote algorithmic transparency, and ensure that automation enhances rather than undermines employee autonomy. At the policy level, it is urgent to incorporate technostress into occupational health regulations and to promote digital well-being standards that guide the design of more humane and sustainable technologies. These actions allow scientific evidence to be translated into concrete practices that protect mental health and foster healthier digital ecosystems.

Future research should advance interventions that act simultaneously on individuals and their environments. At the individual level, strengthening technological self-efficacy, digital literacy, and adaptive skills through continuous training and emotional regulation strategies is essential. At the organizational level, the challenge is to create genuine techno-stress inhibitors, including proactive technical support, robust ICT infrastructure, balanced usage policies, and facilitative leadership capable of reducing tension and promoting healthy digital practices. Emerging evidence also highlights the importance of employee resilience, AI maturity, and socio-technical design principles as key determinants of whether digital demands become harmful overloads or manageable challenges. Ultimately, the focus must shift from diagnosing the problem to implementing structural solutions that integrate individual training, organizational support, and responsible technological design.

This study also has limitations that must be acknowledged. Bibliometric analyses depend on the indexing criteria of the selected databases, which may exclude relevant works. The methods used describe structural patterns of co-occurrence and collaboration but do not assess the methodological quality or empirical robustness of individual studies. Moreover, the rapid evolution of digital technologies means that the patterns identified may change quickly, requiring periodic updates. The emergence of new digital stressors, such as AI anxiety, techno-distress, and algorithmic overload, suggests that future bibliometric analyses should incorporate dynamic monitoring approaches capable of capturing fast-moving conceptual shifts. These limitations do not undermine the validity of the findings but highlight the need to complement future bibliometric work with qualitative approaches, systematic reviews, and empirical studies that deepen the understanding of the psychological, organizational, and sociotechnical mechanisms underlying technostress. By integrating these insights, this study contributes to the broader agenda of strengthening the social dimensions of sustainability and supporting the development of resilient organizations and communities in an increasingly digitalized world.