Do trained nurses feel more psychologically safe?—Results from a multi‐level modelling approach

Abstract Aim To analyse the associations between a nurse's psychological safety and her/his additional training. Design A cross‐sectional survey conducted between September 2015 and August 2016. Methods A multi‐level modelling approach was used considering unit membership. We used data from 1,239 questionnaires completed by nurses on 75 different German neonatal intensive care units, and 75 questionnaires completed by the corresponding leading nurse. Results We found the additional managerial training as a charge nurse to be a positive predictor for psychological safety (β = .346, p ≤ .05). Surprisingly, the additional clinical training in paediatric intensive care is negatively associated with psychological safety (β = −.192, p ≤ .01). Our model estimates that this negative association can be inhibited if the team's share of nurses with an additional clinical training increases (β = .313, p ≤ .05).


| INTRODUC TI ON
Patient safety is related, amongst other things, to the behaviour and attitudes of nurses (Ridelberg et al., 2014). Nurses often have to decide spontaneously whether to raise their concerns in routine care, for example with regard to a particular treatment, and thus questioning traditional practices or their superiors. If nurses are afraid that their statements may have negative consequences for themselves and therefore do not address potential shortcomings in care, this restraint jeopardizes patient safety (Rabøl et al., 2011). Nurses feel psychologically safe, if they are confident that the team will not punish them for speaking up (Edmondson, 1999); "speaking up" is defined in this context as the "communication of ideas, suggestions, concerns or opinions about work-related issues with the intent to improve organizational or unit functioning" (Morrison, 2011). Therefore, it is becoming increasingly important for hospitals to promote psychological safety amongst nurses and to use it to identify and remedy shortcomings in patient safety at an early stage. The importance of psychological safety for safe health-care organizations has recently been highlighted by the National Health Service in England (NHS England, 2019); also, global corporations like Google declare psychological safety to be their first of five key dynamics for teams to work effectively and successfully (Rozovsky, 2015).
In general, it is assumed that patient safety could be improved by a high level of education on the part of the caring nurses. It is suggested that the higher the level of education is, the higher the | 3025 SEIBERT ET al. patient safety and the quality of care is (Aiken et al., 2012;Kirwan et al., 2013). Staffing guidelines even require (a certain share of) nurses to possess additional training (for example, American

Academy of Pediatrics Committee on Fetus and Newborn and American College of Obstetricians and Gynecologists Committee on
Obstetric Practice 2017; Federal Joint Comitee 2020). In Germany, nurses complete three years of vocational training. After their vocational training and at least two years of professional experience, they might complete further additional training, for example in the field of paediatric intensive care. Within these advanced training courses, nurses deepen and broaden their theoretical knowledge and learn special forms of nursing care in a practical way. However, it is still uninvestigated whether these nurses with additional training also feel more psychologically safe and are able to apply their knowledge by speaking up even in critical situations. We will focus on the association between a nurse's additional training and her/his perceived psychological safety.

| BACKG ROU N D
The idea of psychological safety was first introduced by Schein and Bennis in 1965, and described similarly by Kahn in 1990. Respectively, they describe it as a climate that tolerates failure without retaliation (Schein & Bennis, 1965) and the ability to "employ one's self without fear of negative consequences to self-image, status, or career" (Kahn, 1990). This means that team members are encouraged to try things out and to stand up for their opinions and ideas, and do not have to be afraid of being demoted as a result. Later, also by drawing on evidence from health-care provision, Edmondson defines psychological safety at the group level "as a shared belief that the team is safe for interpersonal risk-taking"-being safe for interpersonal risk-taking means that nurses do not fear punishment when speaking up (Edmondson, 1999). Furthermore, team members who feel psychologically safe also feel that their skills and talents are valued and mistakes are not held against them (Nembhard & Edmondson, 2006).
What all these definitions have in common is the belief that psychological safety facilitates the willing contribution of ideas and actions to a shared enterprise (Edmondson & Lei, 2014).
There are three different aspects why psychological safety is important in the context of patient care. First, employees who feel psychologically safe dare to speak up and raise their concerns, for example about an erroneously prescribed medical dosage (Bienefeld & Grote, 2014;Edmondson & Lei, 2014). By addressing their concerns, possible medical failures and treatment errors in general can be avoided and thus lead to a higher patient safety. This is particularly important in intensive care units, where the severity of the consequences of errors due to complex medical processes and vulnerable patients can be particularly high. This is the main reason why we have chosen to focus on nurses in neonatal intensive care units (NICUs). NICUs are used to treat patients who are seriously ill and sensitive to disruptions (Profit et al., 2012). Second, previous research has established that psychological safety fosters employee engagement, commitment, and job satisfaction and reduces bullying (Arnetz et al., 2019;Edmondson, 2003;Frazier et al., 2017;May et al., 2004). Nowadays, when it is important to keep good nursing staff, these aspects are crucial outcomes for hospitals. Third, psychological safety enables team members to modify and apply new technologies (Edmondson et al., 2000) and foster learning behaviour (Edmondson, 2003;Ortega et al., 2013); it is also important for information exchange amongst team members (Aranzamendez et al., 2015).
To date, various factors at an individual and an organizational level are known to affect psychological safety. Prior research has established that leader and team support are important antecedents of psychological safety. People feel their contributions are valued if they receive support from their leader and team (Schepers et al., 2008). The significance of the leader is also figured out in the meta-analytic review of Frazier et al. They outlined, for example, a positive association between psychological safety and ethical leadership, servant leadership, leader-member exchange and trust in one's leader (Frazier et al., 2017). Moreover, at the individual level the two Big Five personality constructs emotional stability (as the opposite of neuroticism) and openness to experience are theoretically linked to psychological safety. Persons with these characteristics "tend to be calm, relaxed, and secure as opposed to anxious, hostile, and vulnerable to stress" (Frazier et al., 2017). Frazier et al. also found a positive and significant relationship between a proactive personality and psychological safety (Frazier et al., 2017). Employees with a proactive personality find it easier to seek performance feedback and build social networks (Crant, 2000). Further, previous research suggests that professional experience in NICUs in general, and in the current NICU, is associated with higher psychological safety (Nembhard & Edmondson, 2006).

| THEORE TI C AL FR AME WORK AND DE VELOPMENT OF HYP OTHE S E S
Previous studies found professional status and hierarchical constraints as an antecedent of psychological safety and speaking up, respectively (for example, Morrow et al., 2016). In general, people of lower status undervalue their input and even fear negative consequences if they speak up (Bienefeld & Grote, 2014). In the healthcare context, Lyndon et al. report that nurses rate potential harm in common clinical scenarios more highly than physicians and are consequently less likely to speak up (Lyndon et al., 2012)

. Nembhard and
Edmondson examined the effect of professional status on psychological safety. They revealed that physicians perceive higher psychological safety than nurses due to their higher professional standing (Nembhard & Edmondson, 2006). There has been less investigation into possible status differences amongst nurses, that is, within the profession, and the resulting effects on their individual perceptions of psychological safety.
The individual's status in a group depends on one's possession or embodiment of generally desired attributes or characteristics (Benoit-Smullyan, 1944). In the professional context of nursing, additional training is a desired attribute that leads to higher individual status within the group. Wolff et al. suggest that nurses' educational attainment is one of the most relevant attributes affecting their attitudes and behaviour (Wolff et al., 2010). Based on the status characteristic theory, if a nurse has completed additional training, her/his occupational prestige-and thus her/his status in the groupincreases (Bloom, 1980). Prior research has shown that employees with a higher status are more likely to participate in team decisions (Bloom, 1980), to speak up (Bienefeld & Grote, 2014) and to feel more psychologically safe (Nembhard & Edmondson, 2006). Thus, we assume that nurses with additional training also feel more psychologically safe than nurses without additional training.
In our research, we differentiate between two types of additional training that focus on different professional aspects.
The additional managerial training course as a charge nurse enables nurses to undertake management and leadership tasks.
Although these nurses are not the NICU's leading nurse despite their advanced training, they are not inferior to the leading nurse in terms of education. Thus, we expect the status difference between these nurses and the leading nurse to be lower. Furthermore, nurses also learn soft skills during their additional training, such as communication and solution-oriented responses to conflict situations.
These skills are associated with a higher psychological safety due to a higher likelihood of the nurse in question speaking up (Landgren et al., 2016). These considerations lead to our first hypothesis: Hypothesis 1: Nurses with an additional managerial training course as a charge nurse perceive higher psychological safety than nurses without an additional managerial training course as a charge nurse.
The two additional clinical training courses in paediatric intensive care and anaesthesia and intensive care enable nurses to perform core tasks in the care and support of (paediatric and) intensive care patients. Due to their additional clinical training, they possess more knowledge and are more likely to care for the patients in the most severe condition. Prior research has shown that the level of experience, the nurse's specialization and their confidence in their clinical knowledge are primary drivers for speaking-up behaviour (Aiken & Sloane, 1997;Landgren et al., 2016;Lyndon et al., 2012). We differentiate between two different types of additional clinical training: "paediatric intensive care" and "anaesthesia and intensive care." The former is more focussed on paediatric patients and therefore corresponds to the patient group in NICUs, whilst the latter is a more general type of training for nurses working in intensive care units, and does not focus on paediatric patients. Nevertheless, we expect there to be status differences between nurses with these types of additional clinical training and nurses without one of these types of additional clinical training. Thus, we hypothesize: Hypothesis 2: Nurses with additional clinical training in paediatric intensive care perceive higher psychological safety than nurses without additional clinical training in paediatric intensive care.
Hypothesis 3: Nurses with additional clinical training in anaesthesia and intensive care perceive higher psychological safety than nurses without additional clinical training in anaesthesia and intensive care.
Beside these associations with the individual level of nurses, our second purpose is to focus on nurses with additional clinical training at the organizational level. Staffing guidelines for neonatal intensive care units require additional clinical training rates, that is, a legally fixed share of nurses has to show a certain level of education and specialization. As prior research has shown, team diversity, for example due to differences in education levels, can affect team communication and performance (for example, Bowers et al., 2000;Jackson, 1996). These differences in experience-based status might limit the willingness of team members to communicate and interact with one another (Jackson et al. 1995). We aim to explore the share of NICU nurses who have additional clinical training as one indicator of team diversity. We assume that a high share of nurses with additional clinical training lowers the status differences for those who also completed additional clinical training. As more nurses possess additional clinical training, the team becomes more homogeneous, and additional clinical training becomes less likely to afford a nurse higher status within the team. Consequently, we expect the positive associations between a nurse's additional clinical training and the individually perceived psychological safety to be weakened then.
Thus, our last hypothesis is: Hypothesis 4: Nurses with additional clinical training do not perceive higher psychological safety if the NICU's share of nurses with additional clinical training is high.

| Objective
We aim to derive management implications for nursing managers and nursing teachers, which allow safer designing of the work environment for nurses. Further, we finally intend to fill the research gap and react to the call of Nembhard and Edmondson to investigate the effect of nurses' specialization on psychological safety (Nembhard & Edmondson, 2006).

| Design
A multi-level, multi-source cross-sectional survey called Safety4NICU study was conducted between September 2015 and August 2016 using data from 1,239 nurses and their corresponding leading nurses from 75 different NICUs in Germany.

| Method
Two hundred and twenty-four NICUs in Germany were identified by web-based searches and public reports. Using a simple random sampling method, all of these were contacted by post and eligible to be included in the Safety4NICU study. The participation of a NICU required the written informed consent of both the leading nurse and the head physician. Overall, 86 NICUs agreed to participate in the study. Each leading nurse received a personalized questionnaire.
Moreover, anonymous staff questionnaires and self-addressed reply envelopes were sent to the leading nurse with the request to distribute them to all nursing staff at a team meeting. All staff nurses who worked at least 50% of a full-time equivalent in the NICU were eligible to participate in the study. These nurses are most familiar with their NICU's structure, procedures and teamwork.
Both the leader and the staff questionnaires were sent back to an independent operating data trust unit located at the University of Cologne. The data collection was carried out between September 2015 and August 2016. Within this period, data from each NICU were collected once. Overall, 78 leading nurses and 1,406 nurses from 82 different NICUs sent back their questionnaires.

| Analysis
Due to the hierarchical structure of our data, we conducted three Nevertheless, only teams whose response rate met the requirement of a sampling ratio of >.32, based on Dawson (2003) and Richter et al., (2006), were included in our analyses. One questionnaire of a leading nurse was excluded due to missing data of the corresponding nursing staff, whilst 95 questionnaires of nursing staff were excluded due to missing data of the corresponding leading nurse. Further, 72 questionnaires of nursing staff and two questionnaires of the leading nurses were excluded due to missing values on relevant variables for our models. Finally, we were able to match and analyse the data from 1,239 nurses in 75 NICUs (see Figure A1 in the Appendix).

Psychological safety
Our dependent variable psychological safety was measured with four items on a 7-point Likert scale (1 denoting complete disagreement; 7 denoting complete agreement), whereas higher scores indicate high psychological safety. The scale was adapted from Edmondson's psychological safety scale (Edmondson, 1999) and has already been applied in a NICU setting by Nembhard and Edmondson (2006). The scale was translated into German by in-depth discussion with a bilingual German/English) psychologist with expertise in item translation. For example, nurses were asked whether the people in their NICU valued the unique skills and talents of others. A onefactor confirmatory factor analysis (CFA) model showed good fit to the data (χ 2 = 8.75, df = 2, p < .01, comparative fit index (CFI) = .995, Tucker-Lewis index (TLI) = .986, root mean square error of approximation (RMSEA) = .052). The internal consistency of the scale was satisfactory (α = .76).

Additional training
Nurses were asked to state whether they had completed additional managerial training as a charge nurse and/or additional clinical training in paediatric intensive care and/or anaesthesia and intensive care.

High share of nurses with additional clinical training
We calculated the share of nurses by summing up all nurses in each NICU with additional clinical training, that is, paediatric intensive care and/or anaesthesia and intensive care, and dividing this sum by the number of respondents in each NICU. Using a median split to identify a high share of nurses in the NICU with additional clinical training, we generated a new dummy variable. NICUs whose share of nurses with additional clinical training was 46.67% or higher had a high share of nurses with additional clinical training.

Job tenure
Nurses were asked to state their job tenure in their current NICU in years and months.

Leader tenure
The NICU's leading nurse was asked to state her/his job tenure in her/his current position in years and months.

Collective team tenure
Nurses were asked to state their job tenure in their current NICU in years and months. Based on these data, we calculated the units' mean and used this as a unit-level measure for team tenure.

Perceptions of management
The perceptions of management within the team were measured by the Safety Attitude Questionnaire (SAQ) (Sexton et al., 2006) for each nurse. The SAQ dimension perceptions of management consist of four items on a 7-point Likert scale (1 denoting complete disagreement; 7 denoting complete agreement), whereas higher scores indicate a high perception of management. We used a German version validated by Zimmermann et al., (2013) and slightly rephrased the items for our NICU setting (for example, "the leading nurse of this NICU supports my daily efforts").

Team size
The NICU's leading nurse was asked to state the number of nurses working in her/his NICU.

| Respondent characteristics
Our final data set consists of responses from 1,239 nurses in 75 NICUs (see Table 1) and covers about one third of all German NICUs.
The average response rate per team was 52.1% (SD: 0.182). 98.14% nurses were female. The average job tenure in the current NICU was 11.78 years. 5.08% of nurses had completed additional training as a charge nurse; 35.59% had additional training in paediatric intensive care; and 21.15% had additional training in anaesthesia and intensive care. The average rating on individually perceived psychological safety was 4.94 on a 7-point Likert Scale.

| Analyses
We tested for multicollinearity by examining the Pearson correlation coefficients (r) between all relevant variables (see Table A1 in the Appendix). As there is no correlation coefficient between the independent variables higher than .4, we do not expect multicollinearity.
The null model shows an intraclass correlation coefficient of .1941, which means that 19.41% of all variance about the individually perceived psychological safety can be explained by NICU membership. Table 2  .210-.420, p ≤ .001). Model 2 reveals that, at the individual level, the additional managerial training as a charge nurse is significantly positively related to a nurse's psychological safety (β = .346, 95% CI: .070-.622, p ≤ .05). Therefore, Hypothesis 1, which posited a positive relationship between additional managerial training as charge nurse and psychological safety, is supported. However, Hypothesis 2, which posited a positive relationship between additional clinical training in paediatric intensive care and psychological safety, could not be supported. We even found that additional clinical training in paediatric intensive care is negatively associated with psychological safety (β = −.192, 95% CI: −.325 to −.060, p ≤ .01). Hypothesis 3, which posited a positive association between additional clinical training in anaesthesia and intensive care and a nurse's psychological safety, could not be supported either. We did not find any significant association between these two variables. We also tested the association between psychological safety and two completed additional clinical training courses, that is, paediatric intensive care and anaesthesia and intensive care. Nonetheless, there was no significant association.

| D ISCUSS I ON
The objectives of our study were to examine the associations between a nurse's psychological safety and her/his additional training, and investigating the moderating effect of a NICU's share of nurses with additional clinical training. Our data show an individually perceived psychological safety level of 4.94. This average value is slightly lower than the reported value of 5.31 perceived by nurses, physicians, respiratory therapists and other health-care professionals in NICUs (Nembhard & Edmondson, 2006), but is higher than the reported value of 3.50 perceived by nurses in public hospitals (Ortega et al., 2013).
The most striking result to emerge from the data is that the associations between psychological safety and the three various additional types of training are different.
First, the additional managerial training as a charge nurse is significantly positively associated with a nurse's individually perceived psychological safety. Although these nurses with the additional managerial training are not their NICU's leading nurse, they feel more psychologically safe than nurses without this managerial training. Therefore, the reasons for this positive association might be based primarily on their acquired skills and their associated standing in the team. There are two likely causes for this positive association.
On the one hand, leadership skills and other managerial issues that nurses have learnt during their additional managerial training could empower nurses to speak up (Landgren et al., 2016). For example, part of this additional managerial training focuses on acting as a leader, communication skills and the right way to deal with conflicts.
These nurses may find it easier to address conflict situations and unclear procedures, which in turn increases their psychological safety (Erkutlu & Chafra, 2015). On the other hand, managerial training might reduce the status differences between these nurses and the other superiors on the NICU, especially the leading nurse, as there is little to no difference in their respective educational levels. TA B L E 2 Results of model 1 and model 2 Second, the association between the psychological safety and the additional clinical training in paediatric intensive care is significantly negative. This result highlights that the additional training according to the working environment does not immediately lead to increased psychological safety. On the contrary, the finding suggests that these nurses feel less psychologically safe. This result might be explained by possible difficulties for these nurses to implement their knowledge and expertise. As a result, these nurses feel that their skills and expertise are not valued, and consequently feel less psychologically safe (Nembhard & Edmondson, 2006). For example, possible treatment options that are proposed by these nurses might be viewed sceptically by nurses without a clinical training. This explanation is consistent with the findings of Kvarnström (2008), which indicated that a lack of consensus is a typical difficulty in teamwork when new and unknown skills are added to the team (Kvarnström, 2008). The moderating positive effect of a high share of nurses with additional clinical training underlines this assumption. If the share of nurses with additional clinical training is high, nurses with the additional clinical training in paediatric intensive care feel more psychologically safe. In NICUs where a high share of nurses possess additional clinical training, it might be easier to implement expertise and experience peer support due to the relatively even distribution of up-to-date knowledge (Kvarnström, 2008). There are also two further explanations for this result. Firstly, prior research has shown that individuals of similar background and abilities are attracted to one another (Bowers et al., 2000). It is easier and more desirable for them to interact (Williams Phillips & O'Reilly, 1998). It can be expected that nurses with the same educational background have high-level relationships based primarily on their shared knowledge, which increase their psychological safety (Carmeli & Gittell, 2009). This is also confirmed by research on person-environment fit, which shows that there is a positive association between the supplementary fit, that is, the similarities between a person and her/his team members, and their psychological safety (Cooman et al., 2016;Seong et al., 2015;Stonefish, 2019). Secondly, if we link this result to social network theory, a further explanation might be that a person feels more psychologically safe the more so-called advice ties she or he receives (Schulte et al., 2012). This means that nurses with the specialized additional training in paediatric intensive care might receive less advice and support from other nurses in their NICU, as there are not many other nurses whose level of expertise is equal to or higher than their own. The probability that these nurses receive advice and support may increase if the share of nurses with additional clinical training is high, which in turn increases their psychological safety once more (Schepers et al., 2008;Schulte et al., 2012). Thus, these nurses might be less likely to feel that their skills and expertise are not valued, which is one typical characteristic of psychological safety (Edmondson, 1999). Compared to nurses with additional clinical training in paediatric intensive care, nurses with additional clinical training in anaesthesia and intensive care may be more likely to experience peer support and advice from nurses who are more specialized. These support and advice increase their sense of psychological safety (Schulte et al., 2012).
Additionally, our control variables emphasize the importance of the NICU's leading nurse to promote psychological safety. As con-

| CON CLUS ION
The current study provides new insights into the professional education and specialization in nursing about psychological safety.

| RE S E ARCH E THI C S COMMIT TEE APPROVAL
This article draws on data from the cross-sectional survey-the Leading nurse n = 75