The intention to use E-mental health applications among people with non-Western migration backgrounds Open Access

Digitalization is a highly dominant phenomenon in society, such as in healthcare. The application of digital information and communication to improve and support health and healthcare is called eHealth (Keij et al., 2024). Whereas eHealth applies to the entire healthcare sector, technology is specifically applied in mental healthcare through electronic mental healthcare (EMH). EMH is defined as “the use of technology and the Internet to deliver mental health information and services” for health promotion, prevention, screening and treatment (Timakum et al., 2022). This includes, among others, digital self-help interventions, text messaging services and interventions that allow patients to interact with providers remotely (e.g. online therapy). With this, EMH could improve the accessibility of mental healthcare and increase the timeliness and quality of care (Mucic et al., 2016; Liem et al., 2021).

However, despite the benefits of EMH, migrants often do not have (adequate) access to digital healthcare (Kaihlanen et al., 2022) while its importance among these populations is frequently high. For example, migrants have a higher risk of developing mental disorders and a higher need for mental healthcare compared to nonmigrants (Pharos, 2025a; Fassaert et al., 2009; Gutiérrez et al., 2023; Gulgun et al., 2024). As a result, migrants were found to experience their health status as less good compared to people without a migrant background (Dekker et al., 2016). A similar result was found among Dutch migrants specifically (De Veer et al., 2015; CBS, 2024). In the meantime, migrants were found to have long waiting lists (Gutiérrez et al., 2023) and a higher drop-out rates in psychological treatment trajectories (Fassaert et al., 2009), as well as a higher threshold when asking for mental support. These barriers appear to be related to, for example, limited health and language literacy (Pharos, 2025a; Fassaert et al., 2009; Liem et al., 2021; Yılmaz et al., 2022; Gutiérrez et al., 2023; Hernandez et al., 2025) or geographical and cultural barriers (Mucic et al., 2016; Gutiérrez et al., 2023). EMH is expected to remove some of these barriers, as it is perceived to be flexible and efficient and offers the ability to provide culturally sensitive information and services (Liem et al., 2021). Herewith, the use of health technologies among vulnerable populations is endorsed by the World Health Organization (WHO, 2021). However, it is important to gain insight into the factors that influence the (intention to) use of EMH among migrants to make mental healthcare more accessible.

Among (potential) EMH users in general, important factors related to intention are performance expectancy, effort expectancy and social influence (Damerau et al., 2021; Gbollie et al., 2023; Cohen Rodrigues et al., 2024; Posselt et al., 2024). These factors stem from the unified theory of acceptance and use of technology (UTAUT) by Venkatesh et al. (2003), a widely used model in the EMH context (Damerau et al., 2021; De Veirman et al., 2022; Cohen Rodrigues et al., 2024; Posselt et al., 2024) that posits the factors that influence the acceptance and use of technology. The model posits that behavioral intention is influenced by these perceptual factors, while demographic factors act as moderators. In addition, Mucic et al. (2016) state that confidentiality, privacy and data security are essential for the acceptance of EMH: EMH can reduce confidentiality and privacy concerns as the intervention of a translator is no longer necessary, but information on data security should be very clear for users to accept EMH. In addition, Shoemaker and Hilty (2016) state that the principle of early intervention is an important factor for the intention to use EMH as EMH offers the opportunity to provide timely mental care and prevent worse mental problems relative to traditional mental healthcare. Moreover, Buelens et al. (2023) reported on the importance of the support of the therapist with EMH, increasing adherence. The use of eHealth and EMH has also been studied among refugees, asylum seekers and immigrants specifically (Pharos, 2025a; Ashfaq et al., 2020; Liem et al., 2021; Kaihlanen et al., 2022; Mabil‐Atem et al., 2024). Immigrants and refugees often have lower access to eHealth than people without a migrant background (Kaihlanen et al., 2022). Amongst others, because eHealth is often too complicated for migrants and refugees (Pharos, 2025b). As for EMH specifically, Ashfaq et al. (2020) state that a lack of digital literacy, cultural sensitivity and the provider–patient relationship pose important factors for the implementation. Liem et al. (2021) and Mabil‐Atem et al. (2024) underline the importance of digital literacy in this context. Furthermore, Mabil‐Atem et al. (2024) state that language and cultural barriers, education level, age, socioeconomic status and familiarity with technology pose important factors for the use of EMH.

However, a knowledge gap remains in the empirical factors explaining the intention to use EMH among migrant groups. As understanding these factors could help increase migrants’ access to EMH, the current study aims to gain insight into the empirical factors associated with the intention to use EMH applications among Dutch first- and second-generation migrants with a non-Western background (RQ). It is expected that the intention is influenced by demographics, literacy levels and perceptual factors from UTAUT and relevant literature.

A cross-sectional survey study was performed between September 2nd and November 2nd of 2022. The survey was conducted among adults in The Netherlands with a non-Western migration background. Inclusion criteria were residency in The Netherlands, an age of 18 years and older and a non-Western migration background. Having experience with mental healthcare was not a prerequisite for participating.

The main analysis was the hierarchical multiple regression including 14 variables (see Statistical Analysis section). According to G*Power (Buchner et al., 2020), with an effect size of 0.15, an error probability of 0.05 and a power of 0.95, this results in a required sample size of 194 respondents. However, in total, 633 participants have provided informed consent and participated. From these, 236 surveys were incomplete and therefore excluded from the data set to be analyzed. In addition, another 48 participants were excluded as they did not meet the inclusion criteria: 34 participants had a Western migration background and 14 people had no migration background. This resulted in a total of 349 participants that were analyzed for this study (n = 349).

Ethical approval was obtained on July 26, 2022 by the ethical committee of the Open Universiteit (U202205743). After that, the survey was pre-tested among 8 people from a migrant background, which did not prompt any adjustments. Accordingly, the survey was imported into the online survey tool LimeSurvey (2022) and distributed by phone and digitally through the researchers’ social media and e-mail. The option was also offered to complete the survey on paper when online completion was not possible (which was the case for 17.1% of the participants). Next to personal invitations to the study, snowball sampling was used. There was no reimbursement for participation.

The survey included 20 questions assessing demographics, use of digital technologies, literacy levels, attitudes toward mental healthcare and EMH. Herewith, 16 independent factors were measured that were expected to influence the intention to use EMH. See  Appendix for the operationalization of the scales.

Age was included as a continuous variable. Gender was included as a binary variable with 1 = male and 2 = female. Next, education was measured categorically with 1 = no education, 2 = elementary school, 3 = lower vocational education, 4 = middle general secondary education, 5 = higher general secondary education, 6 = preparatory scientific education, 7 = middle vocational education, 8 = higher vocational education and 9 = scientific education. Based on the classification of the Dutch Statistics Center (CBS, 2019), this was recoded into three categories: answer options one, two, three and four were coded as 1 = lower education; answer options five, six and seven as 2 = middle education; and answer options eight and nine as 3 = higher education. This variable was dummy-coded with higher education as the reference variable. In addition, cultural background was measured categorically with 1 = Turkish, 2 = Moroccan, 3 = Antillean, 4 = Surinamese, 5 = Cape Verdean and 6= Other non-Western. These answer options were also dummy-coded with Turkish as the reference category. Finally, generation was measured categorically with 1 = participant and both parents born in The Netherlands, 2 = participant born in The Netherlands but both parents are not, 3 = participant born in The Netherlands but either parent not, and 4 = participant and both parents not born in The Netherlands. These answering options were recoded into two categories: the answer option 4 was coded as 1 = first generation (i.e. child and parents born abroad), and the answer options 2 and 3 were coded as 2 = second generation (i.e. child born in The Netherlands, one or both parents born elsewhere). Answer option 1 was a control option to exclude people without a migration background (both the child and parents were born in The Netherlands; CBS, 2025), by which the question was coded as a missing value (n = 14).

Language-, health- and digital literacy were measured and combined into separate scales. Language literacy (i.e. understanding, speaking, reading and writing in Dutch; four items, Cronbach’s α 0.967) and health literacy (i.e. understanding information from organizations on paper, online and in a conversation; three items, Cronbach’s α 0.967) were measured on a five-point Likert scale ranging from “very bad” (1) to “very good” (5). Finally, digital literacy measured the extent to which someone can use a search engine, e-mail, online documents, video calling and links and can install software or apps (six items, Cronbach’s α 0.950), which was measured on a five-point Likert scale ranging from “definitely not” (1) to “definitely” (5).

Several questions were asked to understand and describe EMH needs. Online search behavior on mental health (four items) was assessed with four situational outlines on a five-point Likert scale ranging from “definitely not” (1) to “definitely yes” (5). Language preference was assessed with the same situational outlines (four items) and the answer options “in Dutch” (1) and “In my own -non-Dutch- language” (2). Next, the preferred mode of EMH was assessed. Here, the preference for offline, online or blended (i.e. a combination of online and offline) therapy was measured with six items on a five-point Likert scale ranging from “definitely not” (1) to “definitely yes” (5). The preference for different types of treatments (e.g. online information and tips, self-help programs, online therapy) was measured with five items on a five-point Likert scale ranging from “definitely not” (1) to “definitely yes” (5).

The factors expected to influence the intention to use EMH were assessed by the statements “I could use online psychological help only if […] ” and “What would hinder you from getting psychological help online?” measured by ten and two items, respectively, on a five-point Likert scale ranging from “totally disagree” (1) to “totally agree” (5). Here, three UTAUT factors were included: performance expectancy (one item), effort expectancy (three items; Cronbach’s α = 0.909) and social influence (two items; r² = 0.832). Performance expectancy refers to the perceived usefulness of EMH, effort expectancy refers to its perceived ease of use and social influence refers to the influence of family, peers and healthcare professionals. Based on the UTAUT model, these factors are expected to directly influence the intention to use the technology. The fourth construct, facilitating conditions, refers to the perception that an infrastructure (organizational and technical) exists to support use of the system (Venkatesh et al., 2003). This construct was not included in the current study as it did not appear to predict the intention to use EMH in the study of Damerau et al. (2021) and Cohen Rodrigues et al. (2024). In addition, perceptual factors were included based on literature: the use of the native language (one item; Hilderink et al., 2009; Schinkel et al., 2019), the possibility of starting treatment at an earlier stage through EMH (one item; Shoemaker and Hilty, 2016), the understanding of the healthcare professional (empathizing and tailoring the online conversation to the unique needs of the client; two items; r² = 0.723; Hilderink et al., 2009; Buelens et al., 2023), data security (the assurance that data and privacy are safeguarded; one item; Mucic et al., 2016) and lack of privacy (the belief that one has no privacy and can be overheard during online psychological help; 1 item; Mucic et al., 2016).

Among the participants, the use of the Internet (one item), social media (i.e. Facebook, Instagram, LinkedIn, TikTok, YouTube, Snapchat, Twitter, WhatsApp; eight items) and video calling (one item) were measured on a five-point Likert scale ranging from “never” (1) to “very often” (5).

Finally, the intention was assessed by a scale (five items; Cronbach’s α = 0.903) assessing five situations in which one would use online psychological help, measured on a five-point Likert scale ranging from “definitely not” (1) to “definitely yes” (5).

After the data was collected, the data set was imported into the statistical tool SPSS (‘IBM SPSS Statistics for Windows’, 2023). In SPSS, descriptive analyses were performed. After that, scales were constructed and tested for internal validity. Here, the variable “lack of privacy” had to be excluded from the analyses, as (based on the descriptives) the question seemed to be misunderstood by the participants, rendering the results unreliable. Next, Pearson’s correlation analyses were performed to assess the extent to which the variables were univariately related. The theoretical model was tested by a hierarchical multiple regression with “intention to use an EMH application” as the dependent variable. An Enter procedure with three blocks was performed. The first block included the demographic variables gender, age, education, cultural background and migration generation. The second block comprised language-, health- and digital literacy. Finally, the third block included the perceptual factors explaining intention according to the UTAUT (i.e. performance expectancy, effort expectancy and social influence) and additional literature (i.e. native language, understanding by the healthcare professional, starting treatment earlier and online data security). An overview can be found in Figure 1. Finally, t-tests and ANOVA were performed to explore the difference between lower, middle and higher-educated migrants and first- and second-generation migrants on the intention.

Of the participants, 36.1% was male and 63.9% was female. The mean age was 39.85 years (SD = 14.17). Regarding education level, 13.8% had a low level, 33.8% had a medium level and 52.4% had a high level of education. Most participants (57.6%) had a Turkish cultural background. In addition, participants were of Moroccan (10.9%), Surinamese (9.2%), Antillean (4.6%), Cape Verdean (2.0%) or other non-Western (15.8%) descent. Besides, 52.1% of the participants were first-generation and 47.9% were second-generation migrants. The mean language literacy was 4.53 (SD = 0.81), mean health literacy was 4.47 (SD = 0.90) and mean digital literacy was 4.69 (SD = 0.71). On a scale from 1 to 5, mean Internet use was 4.69 (SD = 0.75), mean social media use was 2.81 (SD = 0.79) and mean use of video calling was 3.62 (SD = 1.25). Furthermore, the mean intention to use EMH was 3.27 (SD = 0.96). See Table 1 for an overview of the sample description.

Participants were asked about the preferred mode of EMH when experiencing psychological complaints or problems; 54.7% would search online for information, tips and experiences, 49.8% would follow an online self-help program, 65.3% would want blended therapy and 33.2% would desire fully online therapy. When searching online for information, participants indicated that they would search for information on physical complaints (79.9%; e.g. headaches, stomachaches, back pain and skin problems), treatment possibilities (77.0%), lifestyle (76.0%; e.g. losing weight, quitting smoking, exercising, healthy eating and drinking and addictions) and psychological complaints (61.6%; e.g. feeling depressed or anxious, worrying, sleep problems and stress). Herewith, 76.5% of the participants indicated searching for information in Dutch, 17.5% searched for information in their native language (other than Dutch) and for 6.0% of participants, the language depends on the information being searched for. See Table 1 for an overview.

Pearson correlation was used to gain insight into the correlation between the variables in our research model (i.e. independent variables and the dependent variable intention to use EMH). There was a small correlation between the dependent variable intention to use EMH and the variables gender, migration generation, language literacy, health literacy, digital literacy, native language, performance expectancy and social influence. There was a medium correlation with the independent variables treatment at an earlier stage, online data security and understanding from healthcare professionals. Overall, there were no strong significant correlations. Table 2 shows all correlations.

A three-step hierarchical multiple regression was run to gain insight into the explanatory factors of the intention to use EMH. The first block included five demographic variables (age, gender, education level, cultural background and migration generation), the second block included three literacy variables and the third block included seven perceptual variables expected to facilitate or impede EMH use. All the assumptions for the regression were met. The third (final) model was statistically significant, R² = 0.316, F(20, 327) = 7.57, p < 0.001, adjusted R² = 0.275. Herewith, the variables Cape Verdean cultural background (β = −0.115, p = 0.015), treatment at an earlier stage (β = 0.190, p = 0.004), performance expectancy (β = 0.245, p < 0.001) and perceived understanding by the healthcare professional (β = 0.216, p = 0.002) appeared to be explanatory of the intention to use EMH. Higher performance expectancy, a stronger belief that treatment can be started at an earlier stage with EMH than offline treatment, and a higher perceived understanding of a healthcare professional increased the intention to use EMH. Finally, a negative relationship was found between the Cape Verdean background and the intention to use EMH, indicating that Cape Verdean migrants show a lower intention than participants with the (reference) Turkish cultural background. See Table 3 for the regression model.

A one-way ANOVA was performed to explore the difference between lower, middle and higher-educated participants in behavioral intention. Lower educated respondents had a mean intention of 2.99 (SD = 1.07), middle educated respondents 3.37 (SD = 0.85) and higher educated respondents 3.28 (SD = 0.98). This difference was not significant, F(2, 346) = 2.83, p = 0.061. Next, an independent t-test was performed to explore the difference between first- and second-generation migrants in intention. This was a significant difference, t(347) = −2.26, p = 0.024, with a higher mean intention of 3.39 (SD = 0.94) for second-generation migrants than for first-generation migrants with a mean intention of 3.16 (SD = 0.97). Based on the results of the correlation analyses, three additional exploratory independent-sample t-tests were performed to explore the difference in language, health and digital literacy between first- and second-generation migrants (see Table 4). All literacy levels appeared significantly higher among second-generation participants than first-generation participants.

In the current study, the intention to use EMH applications among non-western migrants was studied in terms of its explanatory demographic-, literacy- and perceptual variables. Of the demographic variables, only the Cape Verdean background was negatively related to the intention to use EMH, indicating a significantly lower intention than participants of Turkish descent. However, only 2% of the participants (n = 7) were of Cape Verdean descent, which may have affected the validity of this result. Gender was found to be explanatory of intention in the regression until the perceptual factors were added to the model, which appeared to have a greater explanatory value. No other demographic variables (i.e. age, education level and migration generation) were found to be explanatory for the intention to use EMH. However, it was notable that second-generation migrants had a significantly higher intention than first-generation migrants. In The Netherlands, second-generation migrants are on average over 17 years younger (CBS, 2024) and have higher digital literacy (Buisman et al., 2024) and language literacy (Levels et al., 2017; Buisman et al., 2024) than first-generation migrants, in line with the results of the current study. Thus, based on the literature and significant negative correlations, it is suggested that there is a relationship between migration generation and age and between migration generation and literacy levels, which suggests some form of relationship between demographic factors and intention to use EMH.

Regardless of the differences between the two generations, literacy levels appeared to be relatively high in general. This may be explained by the relatively young age of the participants (M = 39.85 years, SD = 14.17) and the majority (52.4%) being highly educated. However, no literacy levels (i.e. digital-, language- and health literacy) were found to be explanatory for the intention to use EMH. Perceptual variables appeared to be more explanatory as starting at an earlier stage, performance expectancy and understanding by the healthcare professional appeared to be explanatory for the intention to use EMH, in line with the literature (Shoemaker and Hilty, 2016; Damerau et al., 2021; Buelens et al., 2023; Cohen Rodrigues et al., 2024; Posselt et al., 2024). In other words: among the participants, the intention to use EMH was higher when people thought that online treatment is more helpful and earlier accessible than offline treatment. In addition, participants were more likely to use EMH when they would feel empathy and cultural sensitivity during online treatment. This highlights the importance of effectiveness the one hand and the importance of inclusive approaches in digital mental healthcare on the other.

Interestingly, in contrast to the UTAUT framework and studies of Damerau et al. (2021) and Posselt et al. (2024), but in line with Cohen Rodrigues et al. (2024), effort expectancy and social influence did not significantly contribute to the intention to use EMH in the current study. This might suggest that, for non-Western migrants, performance expectancy (and therewith the expected health outcomes) may outweigh concerns about ease of use and the influence of relatives. However, the effect may also be explained by the relatively high literacy levels within the sample, reducing the relevance of effort expectancy and social influence. The scale of effort expectancy consisted of the need for technical support and a manual, whilst social influence consisted of encouragement and support from relatives. These forms of support may be less relevant for individuals who possess the knowledge and skills and do not rely on support. Moreover, support or information sharing from friends and family may be less desirable due to the sensitive nature of the problems (Becker, 2016; Schomakers et al., 2019) or expected stigma and perceptions of others (Borghouts et al., 2021). Participants did, however, value empathy and cultural sensitivity from the healthcare provider, which aligns with findings by Posselt et al. (2024) and highlights the importance of “understanding by the healthcare professional” as an explanatory factor in our model. This could also be interpreted as past of UTAUT’s construct of facilitating conditions as such understanding might represent a form of support that enables EMH use.

Within EMH, participants indicated to predominantly search online information on physical complaints, (self-help) treatment possibilities and lifestyle, while fewer participants were interested in information on psychological complaints. Because all non-Western migrants could participate in the study, regardless of their mental status (experience with mental health problems or mental healthcare were no prerequisite for participating), the majority of the participants may have been more focused on prevention rather than treatment. Because mental status or issues were not questioned, it was not possible to study this. However, when having (or imagining having) psychological complaints or problems, participants also indicated a preference for online information on the problem, a self-help program and offline or blended therapy rather than fully online therapy. In line with previous research (Mabil‐Atem et al., 2024; Posselt et al., 2024), this suggests that participants do not consider EMH as a replacement for face-to-face therapy, but as a way to prevent mental illness, obtain information prior to treatment and to be able to start treatment earlier through self-help programs. However, follow-up research is recommended to obtain insight into the EMH applications non-western migrants prefer.

The findings of the current study offer several theoretical insights and practical implications in the context of migrants’ intention to use EMH, which may contribute to improving mental healthcare access among non-Western migrants. Firstly, the correlations and regression suggest that the UTAUT model provides a good starting point for explaining the intention to use EMH. In line with the UTAUT framework, perceptual variables explained intention to use EMH more strongly than demographic factors. Moreover, our findings confirm a significant correlation between social influence and the intention to use EMH, as well as a significant influence of performance expectancy on the intention to use the technology. At the same time, the current study extends the framework by identifying additional factors derived from the literature of particular relevance to migrants, namely the perceived understanding of the healthcare professional and the possibility to start treatment at an earlier stage. These findings suggest that while the UTAUT framework provides a solid foundation, it may require further contextualization to capture the nuances of EMH adoption among migrant populations. As there is a 27.5% explained variance in the final model, it is also suggested that there will be additional factors explanatory of intention that were not included in the theoretical model. Therefore, the theoretical model could be expanded for follow-up studies. Nonetheless, the results of the current study show that initiatives aimed at increasing the perceived usefulness of EMH may increase the intention to use EMH. For example, by offering the possibility to start treatment at an earlier stage -which importance was also emphasized by the study results- through providing information (on physical complaints, treatment possibilities and lifestyle) and self-help activities before therapy. Moreover, in the current study, the influence of literacy (particularly among first-generation migrants) and understanding of the healthcare professional were highlighted. This shows that it is important to enhance literacy levels, strengthen confidence in the effectiveness and accessibility of EMH and increase the involvement, empathy and cultural sensitivity of mental healthcare providers. Nonetheless, future research is recommended on the perceived understanding of the healthcare provider (both linguistic and cultural), and the conditions to increase the sense of involvement, empathy and cultural sensitivity in non-Western migrants using EMH.

With 349 participants, the study was sufficiently powered which can be considered a strength. In addition, with 52.1% of participants being first-generation migrants and 47.9% second-generation migrants, the distribution seems representative of the Dutch migrant population (CBS, 2022). The number of highly educated people seems to be somewhat higher than the national trend of non-western migrants (CBS, 2024), which may limit representativeness in terms of education levels. However, with second-generation migrants having higher intention and literacy levels, the results suggest that the challenges related to EMH adoption might not be uniform across the total migrant population but may be more pronounced in specific subgroups, such as male first-generation migrants with limited literacy. Rather than framing migrants as a single, homogeneous group facing healthcare challenges, it is advised for future research to focus on subpopulations experiencing higher barriersp, as these groups may require targeted interventions. Specifically, it is recommended to conduct follow-up research among more diverse migrant populations (e.g. lower educational and literacy levels, older age and diverse cultural backgrounds) to investigate whether the explanatory factors identified in the current study extend to the entire non-Western migrant population with varied demographic characteristics.

Next, as this study entails a cross-sectional design, it is not possible to make statements about causal relationships. This requires a cautious interpretation of the results. It would be interesting to conduct a longitudinal follow-up study to gain insight into the cause-and-effect relationships. Herewith, it would also be interesting to include the actual use of EMH to be able to make inferences about the explanatory factors of EMH use and the causal relationship between intention and use, as posited by the UTAUT model. Moreover, the factor habit from the extended UTAUT2 model (Venkatesh et al., 2012) is recommended to include as this was not taken into account in the current study. According to Lipschitz et al. (2023), habit strength has associations with both technology adoption and treatment adherence. Hence, it might be interesting to gain insight into participants’ habit strength concerning EMH use and their habit of, for example, looking up symptoms or information on the Internet or keeping track of information in a diary, app or through a smartwatch or other smart device. The UTAUT2 factors hedonic motivation and price value are expected to be of less importance because EMH predominantly concerns a necessity and insured care, by which financial incentives and intended pleasure may play a minor role.

The findings of this study provide insight into the demographic, literacy-based and perceptual factors influencing the intention to use EMH among non-western migrants. No demographic variables were found to explain the intention to use EMH. However, second-generation migrants showed a significantly higher intention than first-generation migrants, which is likely related to their higher digital, health and language literacy. Moreover, performance expectancy, starting treatment earlier and perceived understanding of the healthcare professional appeared to be facilitators. Here, EMH seems mostly valuable for providing information (on physical complaints, treatment possibilities and lifestyle) and self-help activities to precede (offline or blended) therapy, allowing one to start treatment earlier. However, future research should explore EMH use and the EMH applications desired among the non-western migrant population to guide better development and tailoring of EMH interventions.

Tessi Hengst, Senay Gönül-Taymaz, Viviane Thewissen and Catherine Bolman developed the research proposal. Senay Gönül-Taymaz collected the data. Tessi Hengst performed the analysis and wrote the drafts of the manuscript. Lilian Lechner and Catherine Bolman revised the manuscript critically. All authors approved the final version of the manuscript to be published.