A quest for quality care: Exploration of a model of leadership relationships, work engagement, and patient outcomes

Abstract Aim To explore the effects of resonant leadership, leader exchange relationships and perceived organizational support on work engagement and patient outcomes. Design A cross‐sectional survey design. Methods Data were collected in June and July 2016 from 252 nurses and clerical staff and institutional patient safety (falls rates) and patient satisfaction (Friends and Family Test) in New Zealand. Data were analysed with structural equation modelling (SEM). Results The final model was an excellent fit to the data (χ2 (22, N = 252) = 39.048, p = 0.014). Resonant leadership was significantly and positively associated with relationships at work, perception of unit care quality (β = 0.28, p < 0.001), reduced falls rates (β = −0.14, p < 0.05) and better patient satisfaction (β = −0.41, p < 0.001). A direct effect of resonant leadership was demonstrated on patient satisfaction (β = 0.20, p < 0.01). Perceived organization support (β = 0.40, p < 0.001) and leader–member exchange (β = 0.46, p < 0.001) were confirmed antecedents of work engagement. Work engagement was confirmed as an antecedent of nurse perception of unit care quality (β = 0.21, p < 0.001). Where social exchanges exist, work engagement mediates these. Three further mediated paths bypassed work engagement altogether. Conclusion Existing literature investigating the drivers and impacts of work engagement predominantly focuses on staff outcomes rather than patient outcomes. The findings identify modifiable factors to improve staff experience, patient safety, and ultimately patient satisfaction. Resonant leadership, a relational style, is a core antecedent of quality care and positively associated with staff experience and patient outcomes. Impact This investigation into a real‐world problem for nurse leaders also confirmed that an organizational focus on work engagement is not always required. Resonant leadership improves staff work experience, patient safety, and patient satisfaction. Nurse leaders should measure, foster, and develop resonant leadership in practice.


| INTRODUC TI ON
Nurse executives globally are expected to articulate the contribution of nursing to patient care within the boardroom (Mastal et al., 2007). This is becoming more important as healthcare organizations are under pressure to control costs (Francis Inquiry, 2013;Needleman, 2016). Nursing leadership is often held to account for the quality of patient care (Department of Health, 2014;Francis Inquiry, 2013;Healthcare Commission, 2006) despite an absence of research-relating nursing leadership to nurse sensitive outcome indicators. However, notwithstanding over 20 years of discourse about measuring the contribution of nursing to patient care and its importance Ausserhofer et al., 2014), there remains a lack of consensus on metrics (Dubois et al., 2013) and no single measure of ward-level quality care (Dubois et al., 2013;Hurst, 2011;Parr et al., 2018).
Nurse executives continue to be challenged with insufficient evidence to guide decisions on how to organize and lead nursing to affect gains in patient safety, clinical effectiveness and patient experience.
Evidence is emerging which supports the view that relational nursing leadership has a positive relationship with patient outcomes (Squires, 2010;Wong et al., 2013). The implication, therefore, is that nursing leadership should be a focus for organizations intent on improving patient outcomes (Wong et al., 2013). Nursing work is highly relational, where staff need to connect with patients as they provide physical and psychosocial care (Feo et al., 2017).
Critical relational components of nursing practice such as engaging with patients, being present with them, and helping them to cope (Feo et al., 2017) are highly emotional and require relational energy (Cummings, 2004). It also requires staff to be positive, fulfilled (Schaufeli et al., 2006), and willing, and able to reciprocate perceived support from employers and managers with discretionary effort (Eisenberger et al., 1997) to connect in this way. How these characteristics of nursing interact in the complex healthcare setting, however, is not well understood. Our research aim was to test a model linking resonant leadership with experiences of leader-member exchange relationships, perceived organization support, work engagement, perception of unit care quality, patient safety, and patient satisfaction.

| Theoretical framework
Social Exchange Theory provides a relational frame to consider patient experience and the reciprocal nature of engagement between staff and patients and families (Saks, 2006). That is, interactions among patients, family, and staff lead to obligations, which are interdependent and contingent on each other and may be of high or low quality (Cropanzano & Mitchell, 2005). As patient experience is effectively relational, there is a strong fit with considering these measures within research with Social Exchange Theory as the theoretical basis.
Within Social Exchange Theory, interactions lead to obligations which are interdependent and contingent on one another, with the potential to develop high-quality relationships (Cropanzano & Mitchell, 2005). The 'exchange' is bi-directional between two parties and includes (a) rules and norms of exchange, (b) resources exchanged, and (c) emerging relationships (Cropanzano & Mitchell, 2005, p. 875). Interdependence is characterized by 'mutual and complementary arrangements' (Cropanzano & Mitchell, 2005, p. 876). By obeying rules over time, relationships evolve into trusting, loyal, and mutual commitments. Rules of exchange may involve reciprocity or negotiation. Reciprocity is not explicitly negotiated, but understood and contingent on behaviour, may reflect cultural expectations such as expected behaviour or a norm/individual orientation. Reciprocal exchanges generate better work relationships than negotiated relationships, permitting more trust of and commitment to each other. Cropanzano and Mitchell (2005) described a model for the relationship between perceived organizational support and the Leader-Member Exchange or the quality of the relationship.
Within this, it is important to consider all the domains of leadership which include the leader, the follower, and the relationship (Graen & Uhl-Bien, 1995). Social Exchange Theory recognizes the importance of the quality of the relationship between the leader and member as the basis of the social exchange as individuals return benefits they receive and are likely to match these to the person with whom they have a social exchange relationship (Cropanzano & Mitchell, 2005). Practice environment aspects are also considered within Social Exchange Theory, in relation to Perceived Organization Support, or the degree to which the employee perceives the organization cares about their well-being and values their contribution (Eisenberger et al., 1997). An employee who perceives their employer is supportive is more likely to reciprocate.
Social exchanges are a fundamental mechanism in the interplay between leadership and engagement. The quality of the leadernurse relationship is evidenced to be predicted by resonant leadership . The individual roles that the quality of the relationship with the organization and the quality of the relationship between the leader and the nurse play as antecedents of engagement (Brunetto et al., 2014;Dasgupta, 2016;Shacklock et al., 2013) and nurse perceived quality of care (Van Bogaert et al., 2012Wong et al., 2010) have also been highlighted. Social Exchange Theory has been demonstrated as a useful perspective when investigating work relationships (Brunetto et al., 2014;Dasgupta, 2016;Saks, 2006;Shacklock et al., 2013;Squires et al., 2010;Trinchero et al., 2013). What is not evident is the importance of these constructs in relation to leadership as an antecedent and the relationships with work engagement and patient outcomes as dependent variables.
With a Social Exchange lens, we focus on the constructs of relational leadership, perceived organization support, leader-member exchange, nurse engagement and patient outcomes. The study constructs and hypothesized model (Figure 1) are reviewed in the following section.

| Resonant leadership
Relational leadership styles which focus on people and relationships to achieve the common goal are now favoured over task-oriented styles . Resonant leadership styles are described as visionary, coaching, affiliative and democratic (Cummings et al., 2005). Resonant leaders are those in tune with the people around them, they know and can communicate what to do and why to do it and have a high level of emotional intelligence (McKee & Massimilian, 2006, p. 45).
The relational leader appears to have a positive effect on relationships, safety culture and perception of exposure to adverse events such as medication errors (Wong et al., 2013). Safety climate was affected by leader-member relationships and the work environment and a small effect was seen on nurse-reported medication errors (r = −0.22; Squires et al., 2010).  demonstrated that high-resonant leadership styles were significantly associated with 26% lower odds of mortality. The nurse management at the unit level is associated with nurse perception of quality care (R 2 = 0.61, p < 0.05; Van Bogaert et al., 2009). Vogus andSutcliffe (2007) demonstrated that a combination of high 'trust in the manager' and high 'use of care pathways' is related to lower numbers of reported medication incidents. However, these patient safety outcomes were primarily nurse reported and subject to common method bias. Purdy et al. (2010) showed that fewer falls per 1,000 bed days were predicted when empowering workplaces had positive effects on nurse-assessed quality of care. This research aimed to use data that reflected the contribution of nurses to quality care (Dubois et al., 2013) and are already collected and available.
These studies led to the following hypotheses: H1 There is a negative relationship between resonant leadership and falls.
H2 There is a positive relationship between resonant leadership and perceptions of unit care quality.

| Leader-member exchange (LMX)
LMX focuses on the two-way (dyadic) relationship between the leader and subordinate rather than the personal characteristics of the leader, the situation, or the interplay (Gerstner & Day, 1997). The concept of reciprocity is, therefore, a fundamental component. Three domains make up this theory -the leader, the follower, and the relationship, with the emphasis on all three in combination (Graen & Uhl-Bien, 1995).
Measurement of the quality of the leader-member relationship, such as the Charge Nurse Manager and registered nurses, has demonstrated that resonant leadership is associated with the quality of the relationship (correlation coefficient 0.52, pathways significant at p < 0.05; Squires et al., 2010). This led to the following hypothesis: H3 There is a positive relationship between resonant leadership and exchange relationships.

| Perceived organization support
Given the emotional nature of nursing work and the requirement to provide effort beyond the bounds of the employment contract, Perceived Organization Support becomes important. The voluntary nature of discretionary donation of resources is considered to be more highly valued than if it was not voluntary and benefits received in return are likely to be greater (Eisenberger et al., 1997).
Perceived Organization Support, therefore, reflects 'the extent to which the organization values their contribution and cares about their wellbeing and provides a basis for deciding whether increased effort for the organization will be noticed and rewarded' (Eisenberger et al., 1997, p. 818). Although no existing literature was identified demonstrating the relationship between resonant leadership and perceived organizational support, Squires et al. (2010) used the Perceived Nursing Work Environment PNWE of Critical Care Nurses (Choi et al., 2004) and revealed large effect sizes. It is, therefore, theoretically plausible to explore these relationships. This led to the following hypothesis:

| Work engagement
Work engagement is defined as 'a positive, fulfilling, work-related state of mind that is characterized by vigor, dedication and absorp-tion… a persistent and pervasive affective-cognitive state that is not focused on any particular object, event, individual, or behaviour' (Schaufeli & Bakker, 2004, p. 295). Saks (2006) demonstrated the reciprocal element of organizational support and work engagement, suggesting that there is more likelihood of trusting and high-quality relationships with their supervisor where staff are more engaged.
There is also support for work engagement being predicted by exchange relationships (t-statistic = 2.57, significant at p < 0.01; Shacklock et al., 2013). The quality of the relationship between the supervisor and the member and their perception of organizational support predict work engagement and employees more satisfied with the relationship have higher levels of work engagement (Brunetto et al., 2014;Dasgupta, 2016;Shacklock et al., 2013). These studies led to the following hypotheses: H5 There is a positive relationship between perceived organisational support and work engagement.
H6 There is a positive relationship between resonant leadership and work engagement.
H7 There is a positive relationship between exchange relationships and work engagement.

| Quality of care and patient outcomes
Quality is 'the degree to which a system of production meets (or exceeds) the needs and desires of the people it serves' (Berwick, 2013, p. 11) and comprises three domains: safety, patient experience, and effectiveness. Falls is used as a measure of patient safety in the literature (Duffield et al., 2011). Patient experience comprises several components: patient satisfaction, patient perception, patient engagement, patient participation, and patient preferences and building relationships with patients as mutual investment develops (Cropanzano & Mitchell, 2005). Therefore, we proposed the following hypotheses: H8 There is a positive relationship between level of work engagement and perceptions of unit care quality.
H9 There is a positive relationship between resonant leadership and Friends and Family Test.
H10 There is a negative relationship between level of work engagement and falls.

H11
There is a positive relationship between level of work engagement and Friends and Family Test.
Nurse-reported perceptions of unit care quality (Lake, 2002) is often used to understand quality of care. This may be due to the significant challenges of evaluating nursing care due to the laborious nature of identifying and measuring nurse-sensitive measures which persist decades after Donabedian highlighted them (Parr et al., 2018). A significant correlation was found between nurse perception of unit care quality and nurse-reported falls and patient satisfaction (Purdy et al., 2010). Although no existing literature was identified to demonstrate relationships between falls and the Friends and Family Test and perception of unit care quality and Friends and Family Test, the obligations and mutual investment generated within these social exchanges led to the following hypotheses: H12 There is a negative relationship between perceptions of unit care quality and falls.

H13 There is a negative relationship between falls and Friends and
Family Test.
H14 There is a positive relationship between perceptions of unit care quality and Friends and Family Test.

| Hypothesized model
Resonant leadership is evidenced as an antecedent to the quality of the leader-nurse relationship . The work environment has been investigated in the context of patient outcomes but not in research involving leadership styles. What is also not evident is the importance of these constructs in relation to leadership as an antecedent and the relationships with work engagement and patient outcomes as dependent variables. The

| Aim
The aim of this study was to explore the effects of resonant leadership, leader/member exchange relationships and perceived organizational support on work engagement and unit-level patient outcomes.

| Design
Data from a cross-sectional self-report survey of nurses and clerical staff called the Leadership and Engagement of Nurses (LEON) survey and institutionally collected patient safety (falls rates) and patient satisfaction (Friends and Family Test) data were analysed using structural equation modelling (SEM). SEM models the relationships among multiple independent and dependent constructs and simultaneously allows researchers to answer a set of interrelated research questions in a single, systematic, and comprehensive analysis contrary to first-generation statistical tools such as regression (Anderson & Gerbing, 1988). This approach uses a measurement model specified a priori to assess and confirm convergent and discriminant validity and a structural model to undertake a confirmatory assessment of nomological validity (Anderson & Gerbing, 1988).

| Participants
The participants, 252 registered nurses, enrolled nurses, and healthcare assistants, as well as administrative and clerical staff, worked in 1 of 20 units across adult inpatient medical surgical wards at two hospital sites in urban New Zealand. These staff were all managed by their unit manager and considered to contribute to the unit's quality outcomes. The inclusion of clerical staff is consistent with the approach taken by White, Wells and Butterworth (2014) who considered that all team members contribute to the quality of care on the ward.
Considering the complexity or size of the model, a sample size of 10-20 cases per included measured variable is appropriate (Bentler & Chou, 1987;Lomax & Schumacker, 2004). As this research had eight variables, a sample of 200 was acceptable (Squires, 2010).

| Survey
Data were collected over 2 months, June -July 2016. An information sheet explaining the research, voluntarily participation, and contact details of the researchers in case of questions was provided to all eligible staff. Participants were asked to complete the online survey, with an option to complete a paper survey and return in the internal post. An independent person using the work email system and the LEON email address contacted participants. A poster was displayed, and reminders were sent to units to remind staff that the research was still seeking participants and to highlight the remaining time for completion at handover and ward meetings. This approach, recommended by Dillman (2000) and Babbie (2013), was repeated during the 2 months of collection.

| Institutional data
The falls and Friends and Family Test data were routinely collected by the institution in the process of service delivery and service improvement and were also collected for the period of June-July 2016.

| Measurements
The study was comprised of eight variables; four independent variables, three dependent variables, and one marker variable. Table 1 describes the variables, constructs, and psychometric properties of the LEON survey scales (Table 1).

Resonant leadership
Resonant leadership was measured using the 10-item Resonant Leadership Scale which is a subscale of the Alberta Context Tool (Cummings, 2004;Cummings et al., 2008;Estabrooks et al., 2009).
Participants were asked to rate the extent to which their immediate supervisor displays leadership behaviours using a 5-point Likerttype scale from 'strongly disagree' (1) to 'strongly agree' (5). A sample statement is 'the leader in my clinical program or unit acts on values even if it is at a personal cost'.

Perceived organization support
Perceived Organization Support was measured using the 8-item Perceived Organization Support scale (Eisenberger et al., 1997).
Participants were asked to indicate the extent of their agreement TA B L E 2 Descriptive statistics, average variance estimates, composite reliability coefficients, and inter-correlations for the study variables with each item on a 7-point Likert-type scale from 'strongly agree' (1) -'strongly disagree' (7). A sample question is 'My organisation cares about my opinions'.

Leader-member exchange
The validated 7-item Leader-Member Exchange (LMX-7) scale developed by Graen and Uhl-Bien (1995) was used to measure the satisfaction of employees with their relationship with their leader.
Participants respond on a 5-point scale ranging from 'to a very little extent' (1) to 'to a very great extent' (5). A sample statement is 'How effective would you characterize your working relationship with your supervisor?'

Work engagement
Work engagement was measured using the shortened form of the Utrecht Work Engagement Scale. Participants were asked to answer statements about how they feel at work on a scale of 'never' (0) to 'always/every day' (6). A sample statement is 'at my work I feel bursting with energy'.

| Dependent variables
The perception of unit care quality The perception of unit care quality was measured using a 4-item short scale originally used by Aiken et al. (2002). A sample question is 'In general, how would you describe the quality of nursing care delivered to patients on your unit?' (excellent, good, fair, or poor).

Patient safety
Falls is the proxy measure for patient safety and is measured by the number of falls recorded by the institution reported as the number per 1,000 bed days (Purdy et al., 2010).

Patient satisfaction
The Friends and Family Test is the proxy measure for patient satisfaction. The Friends and Family Test asks the question 'How likely are you to recommend our ward to friends and family if they needed similar care or treatment?' (Department of Health, 2013).
It is reported as a percentage of promoters (score 5) over detractors (score 1 & 2) across a 5-point scale. Single-item global measures can allow respondents to consider all aspects of a phenomenon (Patrician, 2004).

Marker variable
Common method bias is a concern when combining multiple self-report variables into independent and dependent variables (Podsakoff et al., 2003). To avoid potentially misleading findings, a 'marker variable' is suggested by Podsakoff et al. (2003) to be used as a statistical remedy for common method bias. The marker variable must be theoretically unrelated to one or all of the constructs in the research (Podsakoff et al., 2003). We selected the willingness to try new food products DSI scale (Barcellos et al., 2009) as a 'marker variable' (social desirability scale). An example of an item in this scale was 'I buy new, different or innovative foods before anyone else I know'.

| Ethical considerations
Approval for the study was obtained from the Auckland University of Technology Ethics Committee (19 April 2016) and locality approval was granted from the organization involved in the study (January 2016).

| Data analysis
Data analysis was conducted using IBM SPSS Statistics 25.0® software and IBM AMOS 25.0® software for structural equation modelling. Confirmatory factor analysis using the two-step approach suggested by Anderson and Gerbing (1988) was employed to test the significance of the scales as the instruments were being used in New Zealand for the first time (Hinkin et al., 1997). One factor congeneric models were reviewed for goodness of fit using the chisquared statistic of goodness-of-fit cut-off criteria recommended by Hu and Bentler (1999). The structural equation model was tested with the data. Path coefficients are interpreted as suggested by Cohen: absolute values from 0.10 to 0.30 are considered small, 0.30-0.50 medium, and 0.50 and above large (Cohen, 1992

| Validity reliability and rigour
The seven steps outlined in Hinkin et al. (1997) were followed to ensure the measures used in the LEON survey were valid and reliable.
All variables of interest, measures, number of items retained in the final model, means, standard deviations, alphas, and score ranges are described in Table 2. The measurement model was tested for discriminant validity, demonstrated (AVE > 0.5) convergent validity and fit to the data (Hu & Bentler, 1999). Tests for common method bias suggested by Podsakoff et al. (2003) were undertaken. The psychometric properties of the variables of interest are presented in Table 2.

| Hypothesis testing
The initial path model demonstrated a very good fit (χ 2 (19,

| Path and mediation analysis
Path and mediation analysis identified four indirect mediated paths (Table 4b). The

| D ISCUSS I ON
This research explored the effects of resonant leadership, leader exchange relationships, and perceived organizational support on work engagement and patient outcomes. Our findings suggest that resonant leadership is a core antecedent of quality care. Resonant leadership also has a direct relationship with the socio-emotional mutual investment social exchange resource between staff and patients.
It also indicates that when resonant leadership is high, staff report higher quality care being delivered, associated with lower falls rates, and higher Friends and Family Test. Only two studies had previously investigated the relationship of resonant leadership to patient outcomes: 30-day mortality  and reported medication errors .
These findings confirmed the role of work engagement as an emerging social exchange in reciprocity to perceived organization support and the quality of leader relationships. This extends the findings from other research where Perceived Organization Support and Leader-Member Exchange were antecedents of work engagement in relation to staff outcomes such as job satisfaction (Shacklock et al., 2013), team commitment (Dasgupta, 2016), and affective commitment (Brunetto et al., 2014;Dasgupta, 2016). Falls are a concrete and tangible example of social exchange resources (Cropanzano & Mitchell, 2005). This results from a greater mutual A strength of the current study was the use of institutional data to evaluate the quality of care being provided as the predominant approach in the literature was to investigate nurse-sensitive indicators using nurse reported exposure to adverse events (Kutney-Lee et al., 2009;Purdy et al., 2010;Squires et al., 2010;Wong et al., 2015).  (Niederhauser & Wolf, 2018).
A focus on resonant leadership is supported by the associations with lower falls rates and higher patient satisfaction (Friends and Family Test) suggesting leadership was not solely restricted to how people feel about their work and practice environment, but is translated to higher quality, particularly, patient satisfaction.

| Limitations
The research was a cross-sectional study with data collected at one period in time. It may therefore, be susceptible to prevalence-incidence bias (Levin, 2006 Future research is indicated to explore these relationships further.

| CON CLUS ION
This research aimed to explore the effects of resonant leadership, leader/member exchange relationships, and perceived organizational support on work engagement and patient outcomes, as nurses are held accountable (Francis Inquiry, 2013). The findings suggest that resonant leadership is a core antecedent of quality care and reinforce the unequivocal expectation of nurse leaders to assure quality care (Pegram et al., 2014). The influence of highor low-quality social exchanges on patient outcomes in highly relational contexts such as acute inpatient settings is a significant finding.
Our findings identify modifiable factors to improve staff experience of work, the safety of patient care, and ultimately pa-

ACK N OWLED G EM ENTS
The authors would like to thank the Director of Nursing, Data Manager, and Workforce Development Manager at Waitemata District Health Board.

CO N FLI C T S O F I NTE R E S T
No conflict of interest has been declared by the author(s).

AUTH O R S' CO NTR I B UTI O N S
JP undertook this project in partial fulfilment of a Doctorate of Health Sciences. JKM and ST were supervisors of the DHSc. ST provided statistical advice and guidance. All authors read and commented on major drafts and signed off the final manuscript.

PE E R R E V I E W
The peer review history for this article is available at https://publo ns.com/publo n/10.1111/jan.14583.