Common mental disorders through the eyes of German employees: attributed relevance of work-related causes and prevention measures assessed by a standardised survey

Purpose Common mental disorders (CMDs) are becoming increasingly relevant in the working world. Numbers of risk factors have been confirmed by mostly correlative cross-sectional studies. Comprehensive and effective prevention is urgently needed. There is little knowledge about employees’ own perceptions on causes of CMDs and prevention measures. Therefore, a survey was conducted in 2016. Methods A standardised instrument was developed for an online survey in a commercial access panel, targeted employees in different job types. We assessed two outcomes: perceived relevance of (1) work-related demands to the development of CMDs, and (2) prevention approaches in the workplace and on individual and societal levels. Possible predictive aspects were analysed exploratively by multivariate linear regression analysis. Results The response rate was 75% (n = 610). Job types were categorised as “blue”, “grey” and “white-collar” workers (n = 193, 169, 248). The majority of respondents rated both outcomes consistently as “quite” or “very relevant”; societal prevention strategies were more relevant for non-white-collar workers. Perceived relevance of individual predisposition to develop a CMD was the strongest predictor for both outcomes, indicating the perception that people with higher personal vulnerability might suffer a higher strain from work-related risk factors than others. Conclusion We assume that participants in our survey judged the relevance of work-related causes of CMDs independently of their own workload. The perceived relevance of prevention measures in different areas is consistent with official guidelines. A possible selection bias due to characteristics of access panel collectives and own direct or indirect experiences with CMD should be critically questioned. Electronic supplementary material The online version of this article (10.1007/s00420-019-01414-7) contains supplementary material, which is available to authorized users.

1. Perceived "Relevance of work-related demands to the development of CMDs" and 2. Perceived "Relevance of prevention approaches at workplace, individual and societal level".
The corrected item-total correlation (CITC) as an indicator of the selectivity of the items and the overall Cronbach's α as an indicator of internal consistency were calculated for all items under outcomes 1 and 2. A Cronbach's α between 0.70 and 0.90 was considered as ideal; a CITC of 0.30 and α of 0.60 were considered the minimum for sufficient selectivity and internal consistency, respectively [Bühner 2010].
To gain information about the structural validity, all items were subjected to an exploratory factor analysis (principal component analysis (PCA), eigenvalue criterion >1, Varimax rotation). Kaiser-Meyer-Olkin criteria (KMO) test values for sampling adequacy of at least 0.5 and a significant Bartlett's sphericity test for homogeneity of variances were considered as appropriate conditions for this approach. Items were regarded as good for factor loading on their own factor if > 0.5 and on other factors if < 0.30 [de Vet et al. 2011]. Firstly, the development from self-constructed items to scores assessing the importance of relevant prevention strategies (outcome 2) was satisfied by finding good psychometric properties and structural validity.
The same was found for the dimensions of outcome 1 (perceived relevance of work-related causes to the development of CMDs) by factor analysis, but with some limitations. We accepted these in order to maintain comparability with the PHOEBE I-samples covering healthcare provider and human resource managers in further publications. In the former study, no structural validity analysis was performed.

Outcome 1: Perceived relevance of different work-related demands
Sufficient factor loading on factor 1 of at least 0.50 in a 2-factor solution (explained total variance: 52.4%) was found for all items grouped under "Interpersonal relations/ leadership", although 2 out of 8 loadings just failed to match the criterion of > 0.50 on their own factor and on other factors with < 0.30 (no. 9 and 10; see Online Resource 3 (see Table O1)).
Most items grouped under the levels "Work content" and "Organisation of work processes" met the 0.50 criterion, but with some limitations: in total, 3 out of 6 items had higher loading than the required maximum of 0.30 on factors other than their own, and "Quantitative job demands" showed too little loading on both factors. The item "Work-privacy conflict" (level "Organisation of work processes") showed higher loading on factor 1, just reaching the 0.50 limit.
KMO and Bartlett's test values in this and in the other models had good quality. All CITC and Cronbach α values met the criteria for sufficient selectivity and internal consistency.  1 quantitatively high demands, e.g. many tasks, take care of everything at once 2 difficult work content/technology in relation to qualification 3 dissatisfied customers 4 concerning "Influence on the job": influence on work pace, work content, work breaks; concerning "Development potential on the job": further education 5 operational processes 6 shift work, length of daily working hours 8 transparency of decisions and responsibilities 9 flat / steep hierarchy structures 10 working atmosphere in the team/in the company, mobbing 14 related to work distribution, decision-making 15 lighting conditions/noise level

Outcome 2: Perceived relevance of workplace prevention activities
For the 17 items under "Workplace prevention activities" (see Table O2), a predominantly sufficient structure could be found, resulting in the three dimensions "Work organisation", "Coaching and training" and "Behavioural prevention". All factor loadings met the 0.50 loading criterion. However, some factor loadings failed to match a loading on other factors with < 0.30. All CITC values were sufficient and an ideal internal consistency was found (Cronbach α at least 0.70, with the exception of dimension B3 in Table 3, covering only two items).
Measuring "Individual prevention activities" (see Table O3), similar results were found, resulting in the dimensions "Support by specialists", "Mental e-health support " and "Support in private life". The only item lacking factor-loading quality ("Expanding one's knowledge about CMDs by reading"; no. 6 in Table 3) was nonetheless included in the score construction and attributed content-wise to the dimension "Support by mental e-health applications".
For the five items under "Societal prevention activities", a 1-factor solution with sufficient psychometric properties was found (see Table O4).

Legend
(1) Principal component analysis (PCA), varimax rotation; eigenvalue criterion > 1 in parts A and C. In part B, a better solution was found with 3 factors in the model. Loadings > 0.3 not depicted; factors named e.g. DA1 for "Dimension 1 of prevention field A" (2) Value ranges: 4-point Likert scale from 1 "very unimportant" to 4 "very important"