Technostress Among Health Professionals – A Multilevel Model and Group Comparisons between Settings and Professions

ABSTRACT

Objective

Health organizations increasingly digitize. However, studies reveal contradictory findings regarding the impact of healthcare information technology on health professionals. Therefore, the aim of this study is to describe the prevalence of technostress among health professionals and elaborate on the influencing factors.

Participants

A secondary analysis was conducted utilizing cross-sectional data from the study, “Work-related stress among health professionals in Switzerland”, which included 8,112 health professionals from 163 health organizations in Switzerland.

Methods

ANOVA for group comparisons followed by post-hoc analyses, along with a Multilevel Model to identify influencing factors for technostress ranging from “0” (never/almost never) to “100” (always), were conducted.

Results

Health professionals experienced moderate technostress (mean 39.06, SD 32.54). Technostress differed between settings (p <.001) and health professions (p < .001). The model explains 18.1% of the variance with fixed effects, or 24.7% of the variance with fixed and random effects. Being a physician (β = 12.96), a nurse (β = 6.49), or the presence of an effort-reward-imbalance, increased technostress most (β = 6.11). A professional with no professional qualification (β = −7.94) showed the most reduction.

Conclusion

Health professionals experience moderate technostress. However, decision-makers should consider the cognitive and social aspects surrounding digitalization, to reach a beneficial and sustainable level of usage.

KEYWORDS:

• technostress
• healthcare professionals
• multilevel model
• healthcare information technology
• HIT

Introduction

Healthcare information technology (HIT) is increasingly being promoted to improve the working conditions of healthcare professionals and the quality of care.¹ The expected benefits are associated with an enormous acceleration of digital transformation.^2–4 HIT is “the application of information processing involving both computer hardware and software that deals with the storage, retrieval, sharing, and use of healthcare information, data, and knowledge for communication and decision making,”^,5(p38) such as decision support systems, hospital information systems, or electronic health records.⁶

Although this outlook sounds promising, the evidence regarding the expected effects on healthcare organizations, care providers, and patients remains contradictory.^3,7 On the one hand, the implementation of HIT led to a significant increase of revenue enhancement (20% – 40%) for the organization, and health professionals reported an 80% reduction of turnaround time (waiting time for process completion).⁸ Furthermore, findings of a systematic review showed an improvement in health behavior and health outcomes among patients through the use of HIT.⁹ Moreover, telehealth, one aspect of HIT, revealed several advantages for the patients, such as low costs, along with improved outcomes and increase in communication with care providers.¹⁰ On the other hand, studies demonstrated that HIT use can result in stress in up to 73% of people employed in healthcare, and up to 40% experience moderate to high stress.¹¹

This stress is also known as technostress, which was introduced by Brod^,12(p16) as “a modern disease of adaptation caused by an inability to cope with the new computer technologies in a healthy manner.” The concept is based on the transactional theory of stress and coping ¹³ and its components are: techno-invasion (employees can be reached anytime), techno-overload (technology forces users to work faster and longer), techno-complexity (users feel inadequate regarding their competences), techno-uncertainty (ongoing changes lead to uncertainty and constant learning), techno-insecurity (feeling threatened about losing one’s job), techno-unreliability (unreliability of technology used). ^14-16 These advances in its conceptual development have led to the latest definition of technostress as “a reflection of one’s discomposure, fear, tenseness and anxiety when one is learning and using computer technology.”^17(p3004) With this background, Gimpel, Lanzl ¹⁵ developed a tentative model of digital stress, which places the components of technostress described above in relation to influencing factors (e.g. support, involvement, and competence) and resulting consequences (e.g. job satisfaction, work–life imbalance).

Studies elaborating on the stress-inducing effects of HIT and the consequences for health professionals and organizations are scarce and tend to focus predominantly on electronic medical records and their effects on physicians.³ For example, among physicians, the implementation of electronic medical records with moderate functions resulted in increased stress levels, decreased levels of satisfaction ^18,19 and higher levels of frustration leading to more burnout symptoms.²⁰ Furthermore, the use of technology at work increases the level of dependence of the professional on the technology,¹⁶ thus promoting new risk factors, such as an ergonomically deficient environment or the amalgamation of privacy and work.²¹ Unfavorable working conditions related to HIT, such as work–life imbalance, high workload, job insecurity, and high physical and emotional demands, are correlated with a variety of illnesses, such as back pain, headaches, and fatigue.^15,22 There are also implications for patients which were identified, for example, in regard to the implementation of electronic medical records, which was associated with increased mortality rates among patients (odds ratio: 3.28; 95%-CI[1.94, 5.55]), due to delays in the treatment process.²³

Frey and Osborne ²⁴ predicted that the digitalization of the working environment will result in vast task shifts from humans to technology. Although the healthcare sector will likely be less affected than other working environments,²⁴ HIT can still be critically regarded as a double-edged sword by health professionals. HIT either reduces work-related stress, as promoted, or it eradicates jobs and increases stress through the improper implementation of technology.

These findings indicate that critical examination of digitization in healthcare organizations is crucial. The advantages of HIT for healthcare organizations, care providers, and patients are promising, and in times of scarce financial and human resources, they are an urgently needed solution. Moreover, contradictions regarding the current research situation, hamper implications for practice. Therefore, it is particularly important to obtain a more comprehensive description regarding the prevalence of technostress in healthcare organizations. ²⁵ Hence, this paper aims to answer the following research questions:

To what extent do health professionals experience technostress?

Does technostress differ between healthcare settings?

Does technostress differ between the health professions?

What are the influencing factors of technostress on health professions?

Materials & methods

This study is based on a secondary analysis using data from the national STRAIN study, “Work-related stress among health professionals in Switzerland.” The STRAIN study is based on a cluster randomized controlled trial (Clinical Trials registration: NCT03508596) and consists of three measurements (baseline, first, second). This study utilized the dataset of the STRAIN baseline-measurement (collected between September 2017 and March 2018), as published in Peter, Schols.²⁶

Study sample

Index-lists from all registered healthcare organizations in Switzerland were utilized for this study sample and included: acute care and rehabilitation hospitals (n = 239), psychiatric hospitals (n = 49), nursing homes (n = 1543), and home care organizations (n = 551). They were obtained through the respective association or the Swiss Federal Statistical Office, from the annual report of 2015. Small organizations (<7 employees and average number of beds <20) as well as specialized clinics (e.g. beauty clinics), were excluded. A geographically representative sample was achieved. A random sample of 100 organizations per setting was drawn from the total sample available, to ensure a sufficiently large sample size for the study.

Recruitment

An electronic invitation was sent to the Chief Executive Officers or the Human Resource Managers of the healthcare organizations and, upon request, the project was presented at a meeting. Detailed information of the recruitment process is described by Peter, Schols.²⁶ Overall, 163 organizations agreed to participate in the STRAIN study, in the following settings: acute care and rehabilitation hospitals (n = 24), psychiatric hospitals (n = 12), nursing homes (n = 86), and home care organizations (n = 41).

Data collection

Data collection in the STRAIN study was conducted as follows: An internal coordinator was appointed in each organization. This person coordinated the dissemination of information and surveys to the health professionals, which consisted of the following professional categories: physicians, medical therapeutic professions (e.g. physiotherapy, occupational therapy, nutritionists), medical technical professions (e.g. radiology, surgical technologist, laboratory assistant), nursing staff (e.g. advanced practice nurses, registered nurses, care aides, midwives), and others (e.g. administration, trainees). The internal coordinator could choose between paper and online questionnaires available in German, French, and Italian for the survey. For paper questionnaires, a pre-stamped envelope was enclosed for returning them to the project team. For online questionnaires, the link for the online survey using SurveyMonkey® and UmfrageOnline® was either sent individually by e-mail or published on the organization’s intranet by the coordinator. Two weeks afterward, a reminder was sent electronically, or a paper-version was mailed to the health professionals organized by the internal coordinator. Upon completion of data collection, the data saved on the SurveyMonkey® and UmfrageOnline ® websites were deleted.

The questionnaire

The primary outcome was technostress. Technostress was measured with a single item developed and tested by the authors, “How often do you feel stressed by the use of technologies at your workplace, e.g. electronic patient record?” It was rated using a scale with a range from “0” (never/almost never), “25” (rarely), “50” (sometimes), “75” (often), and “100” (always), adhering to the questionnaire’s scale structure. This single item was developed, since no suitable valid scale measuring technostress among health professionals was available at the time of the inquiry in the languages needed and the comprehensive STRAIN questionnaire could not be extended by multiple items for all dimensions of the concept in order not to affect the response rate negatively.

The other outcomes used in this study were all used to measure the predictor variables and stem from the STRAIN questionnaire. The STRAIN questionnaire ^26,27 used within the cross-sectional study included well known, valid, and reliable scales (Cronbach’s alpha .64 – .94), focusing on individual characteristics, work stressors, stress reactions, and long-term consequences, as defined in the underlying “Model of causes and consequences of work-related stress,” by Eurofound. ²⁸

For this study, the following scales from the STRAIN questionnaire were chosen as predictor variables to cover the dimensions of the tentative model of digital stress by Gimpel, Lanzl. ¹⁵ Their model describes the correlation between technostress, private and professional demands, stress-inducing and stress-reducing factors, as well as the resulting stress reactions: Demographic information, personal environment,²⁹ demands at work,^30,31 work organization and content,³⁰ person–work interface factors,³⁰ work environment,³⁰ and the Effort–Reward Imbalance (ERI) ³² (see Table 1).

Table 1. Scales and items used for the data analysis

	Scale	Items	Content
Primary outcome
	Technostress	1	e.g. perceived stress when interacting with technology
Predictor variables
Demographic information		7	gender, age, education, profession, work experience, childcare, caring for relatives
Personal environment
	Social Support^Citation1	3	e.g. in private life, supporting those who are close if they are experiencing personal problems
Demands at work
	Quantitative demands^Citation2	3	e.g. work at a high pace, overtime
	Emotional demands^Citation3	4	e.g. confrontation with death, aggressive patients
	Demands for hiding emotions^Citation2	2	e.g. hiding feelings
	Cognitive demands^Citation2	8	e.g. required knowledge, having to remember many things
	Physical demands^Citation4	4	e.g. lifting or moving people or heavy loads
Work organization and content
	Possibilities for development^Citation2	3	e.g. opportunity to develop skills
	Influence at work^Citation2	3	e.g. degree of influence concerning work
	Degree of freedom at work^Citation2	2	e.g. deciding when to take breaks/holidays
	Rewards^Citation2	1	e.g. work is recognized and appreciated by the superior
	Role clarity^Citation2	3	e.g. clear work tasks, objectives, area of responsibility
	Role conflicts^Citation2	3	e.g. contradictory role requirements
	Social support at work^Citation2	4	e.g. receiving support from colleagues/superior
	Feedback^Citation2	2	e.g. receiving feedback from superior
	Social relations^Citation2	1	e.g. possibility to talk to colleagues during work
	Sense of community^Citation2	2	e.g. good atmosphere, co-operation
Person-work interface factors
	Insecurity of the working environment^Citation2	2	e.g. changes in shift schedules
	Bond with the job^Citation2	1	e.g. being proud to belong to the organization
Family-work (im)balance
	Work – privacy conflict^Citation2	5	e.g. conflict between work and private life
	Lack of boundaries^Citation2	2	e.g. being available in leisure time for work issues
Work environment
	Work environment^Citation2	5	e.g. being exposed to noise, cold, chemicals
Effort–Reward Imbalance
	Effort^Citation5	3	e.g. efforts given at work
	Reward^Citation5	7	e.g. rewards received in turn
^Citation1Oslo Social Support Scale ^Citation30 ^Citation2Copenhagen Psychosocial Questionnaire ^Citation31 ^Citation3Nurses early exit study questionnaire ^Citation34 ^Citation4Sixth European Working Conditions Survey (EWCS Q30) ^Citation32 ^Citation5Effort–Reward Imbalance Questionnaire ^Citation33

Data analysis

The items included from the COPSOQ were transformed from ordinal scales with five categories, on a value range from 0 (do not agree at all) to 100 (fully agree), as proposed by the publisher.³⁰ The scale scores were included if at least half of the items had no missing values.³⁰ Nominal and ordinal variables, such as education level and profession, were dummy coded for the multilevel model (MLM). The analysis was conducted using R version 3.5.1 and included descriptive statistics, ANOVA for group comparisons, and an MLM. ³³

The ANOVA group comparisons used the Welch’s t-test because the Levene’s test showed unequal variances for comparisons of the settings (acute care and rehabilitation hospitals, psychiatric hospitals, nursing homes, home care organizations) (p = .01), as well as for the health professions (physicians, nurses, medical-technical professions, medical-therapeutic professions) (p < .001). ³⁴ Consequently, Games-Howell post-hoc analyses were computed for the group comparisons.

The MLM approach considers the natural structure in the data with health professionals (the lowest, level 1 units) nested in organizations (level 2 units). Hence, it is expected to result in a more accurate model compared to simple linear regression, as it ignores the hierarchy.³⁵ The dependent variable for the MLM was “technostress.” The predictor variables for the working conditions were as follows: the ERI, commitment, demands (quantitative, cognitive, emotional, physical, hiding emotions), role clarity, role conflict, insecurity of working conditions, work-privacy conflict, lack of boundaries, working environment, setting, and employment level. The predictor variables for the stress-reducing factors at the workplace were as follows: possibilities for development, influence at work, freedom at work, appreciation, feedback, support at work, social support, and sense of community. Regarding the private conditions, the predictor variables were childcare and caring for relatives. The predictor variables for the individual characteristics were gender, age, education, profession, and work experience. The second level variable was clinic (see Figure 1). In order to minimize internal dropouts, the missing data for the numerical predictor variables in the MLM were filled based on multiple imputation with expecting data to be missing completely at random, using the mice package.³⁶

Figure 1. Scales used for the MLM based on the model by Gimpel, Lanzl.¹

In a first step, a stepwise model selection with the MASS package was conducted with a higher Akaike Information Criterion, as the inclusion criteria.³⁷ The selected variables were then fitted using lme4 package.³⁸ For the MLM, beta coefficients with according p-values (2 tailed) and 95% confidence intervals (CI), as well as the marginal R-Squared (associated with fixed effects) and the conditional R-Squared (associated with fixed and random effects), were computed.³⁹ The assumption of heteroskedasticity was not met for the model. Hence, standard errors, p-values, and CI were bootstrapped (r = 999, bias corrected and accelerated, 95% CI).

Ethical considerations

The local Swiss ethical board confirmed, on 24^th October 2016, that the study did not warrant a full ethical application and did not fall under the Swiss Federal Act on research involving human beings (Req-2016-00616). The participants are professionals and can take their responsibility for their participation. They were informed in writing at the beginning of the questionnaire about the contents and the voluntary nature of their participation.

Results

Overall, data of 8,112 health professionals were included in this analysis. Among the participants, 7% were physicians (n = 463), 4% medical-technical professionals (n = 241), 9% medical-therapeutic professionals (n = 628), 75% nurses and midwives (n = 4925), as well as 5% others (n = 346). Among the participating health professionals, 42% worked in acute care and rehabilitation hospitals, 26% in psychiatric hospitals, 21% in nursing homes, and 11% in home care organizations. The mean age of the participants was 42 years (SD 12) and the majority were female (82%).

Technostress

In total, health professionals reported on the range from 0 (never/almost never) to 100 (always), a mean score for technostress at work of 39.23 (SD = 32.54). Table 2 summarizes the mean and standard deviation of technostress according to setting and profession.

Table 2. Technostress experienced among healthcare professionals

Setting	Profession	N	Technostress* Mean (SD)
Acute care and rehabilitation hospitals	Total	3398	46.38 (31.95)
	Nurse	1905	49.67 (31.56)
	Physician	229	53.85 (30.74)
	Medical-therapeutic profession	237	35.85 (30.82)
	Medical-technical profession	241	39.96 (29.49)
	Other	103	30.78 (28.98)
Psychiatric hospitals	Total	2075	37.90 (31.86)
	Nurse	952	43.01 (32.74)
	Physician	204	40.97 (30.35)
	Medical-therapeutic profession	319	32.55 (28.97)
	Other	158	25.15 (28.76)
Nursing homes	Total	1693	31.65 (32.49)
	Nurse	1317	32.80 (32.80)
	Physician	30	22.41 (30.14)
	Medical-therapeutic profession	72	20.96 (28.20)
	Other	13	25.51 (28.18)
Home care organizations	Total	946	31.90 (30.92)
	Nurse	751	32.40 (30.59)
	Other	72	34.34 (31.32)
Total		8112	39.23 (32.54)
*Range: 0–100

Setting comparison

The extent of experienced technostress differed significantly between settings, Welch’s F(3, 3148.8) = 99.39, p < .001. The Games-Howell post-hoc analysis revealed a significant difference (p < .001) between technostress experienced among health professionals, as follows: in acute care hospitals and psychiatric hospitals (8.48, 95%-CI[6.00, 10.90]), in nursing homes (14.73, 95%-CI[12.20, 17.30]), and in home care organizations (14.47, 95%-CI[11.50, 17.40]). This reveals, for example, that on average, health professionals working in acute care and rehabilitation hospitals have higher technostress (14.47 points), in comparison to health professionals working in home care organizations. Moreover, the psychiatric hospitals showed a significant difference in technostress among health professionals in comparison to nursing homes (p < .001) (6.25, 95%-CI[3.40, 9.10]), as well as when compared with home care organizations (p < .001) (5.99, 95%-CI[2.80, 9.20]). The difference between the nursing homes and home care organizations was not significant (p < 1) (.25, 95%-CI[−3.50, 3.00]).

Comparison of the health professions

The Welch’s Test revealed a significant difference of technostress between the health professions, Welch’s F (4, 933.04) = 47.30, p < .001. The Games-Howell post-hoc analysis also showed a significant difference of technostress between health professions. Physicians had significant higher technostress than medical-therapeutic professions (p < .001) (14.7, 95%-CI[9.54, 19.80]), medical-technical professions (p = .003) (7.00, 95%-CI[0.37, 13.50]) and nurses (p < .001) (5.80, 95%-CI[1.72, 10.00]). Furthermore, nurses had a significantly higher technostress than medical-therapeutic professions (p < .001) (8.80, 95%-CI[5.30, 12.34]).

Influencing factors on technostress

In regard to the regression analysis, cases with missing data in the included and not imputed factor variables (e.g. education, profession) were excluded. Hence, the dataset comprised 7,230 cases (89.13%). The estimated MLM explains 18.1% of the variance with fixed effects (marginal R-Squared) or 24.7% of the variance with fixed and random effects (conditional R-Squared). Working as a physician (β = 12.96, p < .001) or a nurse (β = 6.49, p < .001), or having a higher ERI was associated with increased technostress (β = 6.11, p < .001). However, working in a profession with no professional qualification, such as trainees, civilian service, and volunteers (β = −7.94, p < .001), was most significantly associated with a decrease in technostress (see Table 3). Furthermore, higher social support was associated with decreased technostress (β = −0.64, p < .01). Regarding binary variables, for example, with physicians, the data is interpreted as follows: if the individual is a physician, the technostress increases by 12.96 points. The interpretation for numerical variables, for example, with social support, is different: if social support increases by one point, technostress decreases by 0.64 points.

Table 3. Model for technostress in healthcare

Coefficient	β	Std. Error	T value	p-value (*with bootstrap)	CI
Intercept	−16.06	4.07	−3.94	<.001	−24.04 – −8.07
Physician	12.96	1.88	6.90	<.001	9.28–16.65
Nurse	6.49	1.18	5.52	<.001	4.18–8.79
Effort–Reward Imbalance Ratio	6.11	1.58	3.86	<.001	3.01–9.21
Medical-therapeutic profession	5.47	1.65	3.32	<.001	2.25–8.70
Work experience	0.31	0.04	8.32	<.001	0.24–0.39
Working environment	0.24	0.02	9.56	<.001	0.19–0.28
Emotional demands	0.12	0.03	4.04	<.001	0.06–0.18
Physical demands	0.12	0.02	5.15	<.001	0.08–0.17
Role conflict	0.12	0.02	5.01	<.001	0.07–0.16
Work-privacy conflict	0.11	0.02	4.69	<.001	0.07–0.16
Quantitative demands	0.09	0.03	2.92	.01*	0.03–0.15
Cognitive demands	0.08	0.04	2.24	.03*	0.01–0.15
Insecurity of working conditions	0.08	0.02	4.22	<.001	0.04–0.11
Feedback	0.05	0.02	2.19	.02*	0.00–0.09
Possibilities for development	0.04	0.03	1.26	.19*	−0.02–0.10
Appreciation	0.03	0.02	1.56	.13*	−0.01–0.06
Employment level	−0.03	0.02	−1.58	.11*	−0.07–0.01
Lack of boundaries	−0.06	0.02	−3.06	.01*	−0.10 – −0.02
Social support	−0.64	0.23	−2.82	.01*	−1.08 – −0.19
Education secondary II	−3.10	1.04	−2.99	<.001*	−5.13 – −1.07
No professional qualification	−7.94	2.50	−3.17	<.001*	−12.85 – −3.03
Random Effects
Marginal R2:	0.181
Conditional R2:	0.247
Variance intercept	71.27
Variance Residual	812.34
ICC	0.08

Discussion

This study revealed that health professionals in Switzerland experience moderate technostress in their daily work, which is comparable to the findings of Gimpel, Lanzl ¹⁵ from Germany. However, the technostress experienced among the health professions differs between settings and professions. Health professionals working in the acute care or psychiatric hospitals reported especially higher technostress than professionals in the other healthcare settings. This might be related to the fact that in Switzerland, the settings with higher technostress are also more advanced in terms of digitization. Therefore, they might be more exposed to the adjunct influencing factors of implemented HIT.⁴⁰

In this study, physicians showed significantly higher technostress in comparison to other included health professions, followed by the nurses. Additionally, the MLM revealed that with an increase in individuals’ educational levels, the experienced technostress significantly increased. To our knowledge, comparable studies from the healthcare sector are missing. Other studies focusing on different sectors revealed contradictory findings regarding the correlation between education level and technostress.⁴¹ Therefore, we assume that the influence of education level on technostress is sector specific.⁴² For example, in industry, personnel with low education levels interact a great deal with technology, whereas personnel with a lower education level (e.g. care aides) in the healthcare system have less interaction with technology, which may explain their lower levels of technostress. The higher technostress reported among physicians could be explained to some extent, by the unwanted time spent with electronic medical records.⁴³ Previous studies showed that physicians spent more time with documentation than other health professionals.⁴⁴⁴⁹ This is related to the fact that they have an increasing number of mandatory forms to complete due to reimbursement regulations. This is also because they have an increasing number of patients to care for, with increasing levels of complexity in care.⁴⁵

According to the findings in the MLM, the ERI variable has been identified as a relevant predictor, regarding its impact on technostress, which is supported by Stadin, Nordin.⁴⁶ Considering the tentative model of technostress,¹⁵ the dimension techno-overload and techno-unreliability could contribute to an explanation. Techno-overload might cause health professionals to conduct more and more tasks with HIT, without a noticeable increase in rewards (e.g. increasing reporting to health insurance companies). Moreover, the techno-unreliability of HIT (e.g. system crashes, connection errors) can also increase the effort required to achieve a task. ¹⁴

However, Patel, Ryoo ^47(p3) highlighted the “dual role of [technology] as a job demand and a job resource.” They argue that when elaborating on ERI’s association with technology, the ERI variable fails to differentiate between technostress-inducing and technostress-reducing resources of technology. Thus, they propose the use of the job demands-resources model in place of the ERI variable.

The MLM revealed that a higher level of social support (resource) results in decreased technostress. This corresponds with the proven fact, that social support has a stress-reducing effect.⁴⁸ Hence, having a supportive community helps with managing HIT, broadening the theory of IT support as being a technostress-reducing factor. Health professionals might seek support from non-IT colleagues to manage HIT, thereby, respectively, also enhancing their digital competence.^17,47,49

This relationship between technostress and digital competence is supported by Gimpel, Lanzl ¹⁵, explicating that a mismatch of the digitization level with the individual’s digital competence, led to an increase of technostress. International recommendations such as the Technology Informatics Guiding Education Reform (TIGER) Initiative or the DACH-recommendations for German-speaking countries propose a framework for the required digital competences. However, these recommendations require more elaboration and evaluation, along with further research.^50,51

Terminio and Gilabert ⁵² stated that most professionals are not aware of the consequences of the ongoing disruptive processes regarding digitization. This lack of awareness might be noteworthy, along with the fact that Switzerland’s healthcare system is much less digitized than several other countries.⁵³ This could underline the experienced low technostress among health professionals working in nursing homes and home care organizations in this study, which are sparsely digitized.

Strengths and limitations

This study compares, for the first time, technostress between the settings as well as between the health professions. Thus, it contributes to a more comprehensive understanding of the extent to which technostress is experienced in the healthcare sector. Moreover, the analysis gained a large study sample for each health professional’s discipline and language region of Switzerland, as well as for the chosen analysis. Through conducting the hierarchical model analysis, the authors verified the added value of the analysis, by considering the natural structure of the data.

Technostress was, however, measured with a single item. This aggregated information offers only an insight into the complexity of technostress, which consists of multiple stress-inducing and -reducing dimensions.^15,47 The use of this single item limits the interpretation of the findings, since no reference values exist and no measurement reliability has been estimated. However, the statistical differences could indicate a sufficient discriminant validity.⁵⁴⁵⁵ To test this hypothesis, further research is needed for psychometric testing of the single item using a reference questionnaire.

Moreover, not all factors which were required (as promoted in the tentative theoretical framework by Gimpel, Lanzl ¹⁵) could be included into the MLM, since the used questionnaire was comprised of partly differing dimensions. This might have led to a lower explanation of variance for the MLM. Furthermore, the participation within the primary study was voluntary, which may have caused a selection bias. This could indicate that organizations, respectively, health professionals, which experience a higher technostress, did not participate in the study. In addition, the study sample comprised of healthcare organizations from Switzerland is less digitized than other industrial countries.⁵³ Moreover, no causal conclusion can be drawn, as this study utilized cross-sectional data. These implications need to be considered when interpreting the results.

Conclusions

The data provide a first insight into the prevalence of technostress among different health professions. To our knowledge, there are no other studies available on technostress comparing various settings and health professions. This study promotes awareness of this topic among health professionals and managers of healthcare organizations. HIT must be evaluated for reliability over a sufficient period of time before implementation, along with the involvement of the target group testing them to prevent techno-unreliability. Tasks could, furthermore, be assigned to new professions, or interfaces could be simplified for greater user-friendliness to manage techno-overload. Moreover, IT-specialists are gaining knowledge concerning the avoidable accompanying effects that HIT can have on health professionals. The findings suggest that IT-specialists and managers should consider the cognitive and social aspects of affected health professionals, to achieve sustainable and beneficial usage of HIT. Specifically, this means, considering the needs of the health professionals affected, involving them in the development and evaluation of HIT, offering continuous support, and formulating a long-term digitization strategy for the organization.

The healthcare sector is increasingly being digitized. Accompanying this process, an increase of technostress among health professionals is expected. Therefore, even though this study revealed moderate technostress among health professionals, longitudinal approaches as well as intervention studies to elaborate the change of technostress over time with regard to evidence-based measures (e.g. enhance digital competence), are needed.

The findings of this study need to be validated with further research, focusing in the first instance on physicians and nurses, as those professions showed the highest technostress among the professions included. Moreover, measures in intervention studies should address social support within teams, since it is expected to have a mitigating impact on technostress. Specifically, the relationship between ERI and technostress should be elaborated more comprehensively for a better understanding of its origins.

At this stage, technostress is an emerging topic in research. Its theoretical framework is still in development and will continue to evolve, due to the rapid pace of changes caused by digitization. Further research is needed to identify stress-inducing and -reducing factors of HIT among health professionals, and to develop a theoretical framework based on these findings. This is relevant, as digitization is on the agenda of healthcare organizations worldwide. Hence, the findings of this study should be compared to other studies internationally, thus broadening the discussion and facilitating international exchange. This is important since the transferability of technostress-reducing measures between the countries is expected to occur.

Declaration of interest

The authors report no conflict of interest. The authors alone are responsible for the content and writing of the paper.

Acknowledgments

We wish to thank the organizations as well as the health professionals for their participation. Additionally, we thank Dr. Reto Bürgin, who shared his expertise in statistical analysis with us.