Organisational efficiency and co-worker incivility: A cross-national study of nurses in the USA and Italy

Abstract

Aim

To examine the relationship of co-worker incivility with organisational efficiency, workload and intention to leave in nursing samples from two different countries, the USA and Italy.

Background

Organisational efficiency has received little attention as a possible correlate of co-worker incivility in the nursing management literature. Studies on co-worker incivility have primarily been carried out in North America and no cross- national studies are available.

Method

Data were collected by a self-report questionnaire involving nurses from the USA (n = 341) and Italy (n = 313).

Findings

Organisational efficiency was negatively associated with workload, co-worker incivility and intention to leave in both samples. The path from co-worker incivility to intention to leave was also positive and significant in both samples. Workload was positively associated with co-worker incivility and intention to leave in the US sample, but not in the Italian sample.

Conclusion

The present study suggests that organisational efficiency is central to understanding both co-worker incivility and intention to leave among nurses.

Implications for Nursing Management

Nurse administrators should adopt interventions aimed at fostering organisational efficiency in an effort to reduce nurse co-worker incivility.

1 | INTRODUCTION

Co-worker incivility (CWI) is one of the subtlest forms of mistreatment, referring to any low intensity deviant behaviour(s) in violation of the norms of mutual respect, with ambiguous intent to harm the target (Andersson & Pearson, 1999). It encompasses rudeness (e.g. raising the voice), disrespect (e.g. ignoring or excluding a worker) or unfairness (e.g. doing demeaning things to a worker).

While the co-worker incivility literature has considerably grown in the past decade, knowledge on this topic needs to be further enhanced. A recent systematic review (Schilpzand, De Pater, & Erez, 2016) revealed that, whereas many studies demonstrated the role of co-worker incivility in fostering job stress (e.g. Goussinsky & Livne, 2016) and job abandon (e.g. Laschinger, Leiter, Day, & Gilin-Oore, 2009), few (three of the 36 studies in the review by Schilpzand et al.) assessed variables describing the quality of the work processes as possible correlates of co-worker incivility. In the nursing context, identifying the aspects of the work environment sustaining incivility is critical. As pointed out by Mikaelian and Stanley (2016), the forces of co-worker incivility may be compounding the staffing shortfall issue by pushing nurses out of the profession.

Another important limitation of the co-worker incivility current literature is that studies have mostly been carried out in North America (Schilpzand et al., 2016). To the best of our knowledge, no cross- national studies are available in nursing or in other work contexts.

1.1 | The present study

The present study used the conservation of resource theory (COR) (Hobfoll, 1989) as a theoretical framework. The concept of a resource is central in this theory, which states that the prime human motivation is to obtain, maintain and accumulate resources. In organisational settings, as pointed out by Hobfoll and Shirom (2001), resources are those aspects of the work environment that help in acquiring new resources, achieving work goals or in reducing workers’ physiological and psychological costs.

A central tenet of the conservation of resource theory is the concept of the ‘loss spiral’ (Hobfoll, 1989) that refers to a process of expense of resources that makes them unavailable to cope with secondary loss, thus, potentially leading to further resource loss. The notion of ‘loss spirals’ is often used to explain why individuals working in stressful or poor-resource work environments are vulnerable to secondary resource losses and, thus, are likely to take a defensive position to protect themselves from further losses.

The present study, using the loss spiral concept (Hobfoll, 1989), aimed to test in two samples of nurses from different countries, i.e. the USA and Italy, whether CWI plays a role in linking variables that describe the quality of the work process (i.e. organisational efficiency and workload) and intention to leave the job.

This study is of importance as it may contribute to building generalizable theoretical knowledge across nursing settings in various national contexts on mechanisms responsible for the spreading of co- worker incivility and its detrimental ramifications.

1.2 | The relationship of organisational efficiency with co-worker incivility

According to Arnetz, Lucas, and Arnetz (2011) and Arnetz, Zhdanova et al. (2011), organisational efficiency may be defined as the employees’ perception of how well work processes function at their work-place. It includes an evaluation of decision-making processes, the degree to which goals are shared, how work is planned and resources are utilized. Those studies indicated that organisational efficiency helps to understand job well-being both in the general working population and in the nursing context. Despite that, it remains an understudied dimension and no prior studies have considered the relationship between efficiency and CWI.

Conservation of resource theory suggests that a reduced level of organisational efficiency, being a condition of poor resources, may work as an initiator of a loss process, thus favouring the spread of co-worker incivility. Although there is no empirical evidence in support of this statement, there are logical arguments, consistent with the COR theory, that suggest its plausibility. Co-worker incivility may arise because of an inefficient work environment, which may encourage different visions among nurses regarding how to carry out the job. In fact, in an organisation in which resources are not optimally utilized or goals are not shared, workers may be more prone to secondary loss, such as being a target of co-worker incivility, as they are more likely to be overloaded, to make mistakes, or to experience disagreement with other workers (Bolino & Turnley, 2005). Based on this, we propose that:

H1: Efficiency is negatively associated with co-worker incivility.

1.3 | The relationship of workload with co-worker incivility

The principle of the loss spiral also suggests that workload may mediate the relationship between efficiency and co-worker incivility. A workplace in which organisational processes do not work well can be characterized by an increased workload and, as a result, CWI may be more frequent as well.

Workload is a job demand and refers to the amount of work assigned to or expected from a worker in a specified time period. In line with the loss spiral concept, work overload can be seen as a compensation strategy that the employee may be required to put into action in order to compensate for the demands related to inefficiency. Although job demands are not necessarily negative, they are generally associated with the loss process because workers, in order to accomplish demands, are required to use personal resources (i.e. energy, time, etc.) that may not be recovered (Hobfoll & Shirom, 2001). Moreover, a high workload may make workers more vulnerable to further losses, such as being a target of co-worker incivility. This may occur because employees, when faced with excessive demands, may behave in ways that encourage others to treat them rudely (Taylor & Kluemper, 2012).

From an empirical point of view, there is evidence supporting significant associations both between workload and CWI (e.g. Magnavita, 2014; Taylor & Kluemper, 2012) and efficiency and workload (Arnetz, Lucas et al., 2011). However, the mediating role of workload (Ha), despite its plausibility and its consistency with the principle of loss spiral, has not been previously tested. The present study aims to expand knowledge in this direction. To confirm Ha, in addition to H1, this study will also examine the following hypotheses (Figure 1).

H2: Efficiency is negatively associated with workload.

H3: Workload is positively associated with co-worker incivility.

FIGURE 1.

Graphical representation of the study hypotheses

1.4 | The relationship of co-worker incivility and intention to leave the job

A work environment characterized by poor efficiency, high workload and co-worker incivility may lead workers to develop negative attitudes and behaviours driven by the goal of minimizing losses and to protect themselves from further losses (e.g. energy, career opportunity). In that sense, a possible outcome is intention to leave. In COR terms, it represents a withdrawal strategy that individuals may plan to put into action in order to interrupt the loss spiral (Sliter, Sliter, & Jex, 2012).

The predictive roles of CWI and workload on intention to leave were widely supported in studies carried out in many working settings, including nursing (e.g., Laschinger et al., 2009; Viotti & Converso, 2016). Accordingly, we hypothesized:

H4: Co-worker incivility is positively associated with intention to leave.

H5: Workload is positively associated with intention to leave.

Although there are no studies specifically focused on the association between efficiency and intention to leave, many studies demonstrated that the quality of the organisational processes affects worker attitudes toward the job, including intention to leave (Hayes et al., 2006). Hence, we hypothesized:

H6: Efficiency is negatively associated with intention to leave.

Despite their consistency with the principle of the loss spiral and the presence of empirical evidence that suggests their plausibility, no studies examined whether CWI mediated the effect of efficiency (Hb) and workload (Hc), respectively, on intention to leave. The present study is intended to fill this gap. Accepting Hb implies confirming H1, H4 and H6. To accept Hc, H2, H4 and H5 need to be confirmed (Figure 1).

1.5 | Testing for cross-national invariance

Co-worker incivility prevalence rates considerably vary across cultures (Schilpzand et al., 2016). Empirical evidence indicated that US nurses are significantly more exposed than Italian nurses to co- worker incivility (USA about 80%, e.g. Lewis and Malecha (2011); vs. Italy: about 40%, e.g. Magnavita & Heponiemi, 2012). An explanation for these differences can be given with the help of the Hofstede theory (2001) which proposed that one of the dimensions by which a culture can be classified is the extent to which its members try to control their impulses and desires. Co-worker incivility tends to be higher in cultures so called ‘indulgent’ (e.g. Anglo-Saxon) countries in which control over impulses are weaker rather than in ‘restrained’ cultures (e.g. Mediterranean countries) in which control is stronger (Hofstede, 1984). These cultural differences suggest that, in order to advance in co-worker incivility literature, it is decisive to establish whether it is appropriate or not to assume that factors relating to co-worker incivility can be generalized across cultures (Schilpzand et al., 2016). Accordingly, we chose to test our hypotheses across nursing samples from two different countries, i.e. the USA and Italy, which are typical examples of an indulgent and a restrained culture, respectively (Hofstede, 1984). The resulting knowledge may help nurse administrators to identify measures to prevent and manage co-worker incivility, while considering possible culture-based characteristics.

Given the absence of previous cross-cultural studies on co- worker incivility, statements on differences or similarities between the two national samples in the tested relationships cannot be made. Therefore, we included an exploratory hypothesis that all the relationships hypothesized (Figure 1) would be cross-nationally invariant (H7).

2 | METHOD

2.1 | Data collection and participants

In the USA, data were collected via a survey in a large, Midwestern hospital system. In 2013, 1389 questionnaires were mailed home to employees, along with a postage-paid return envelope. A total of 341 (24.5%) nursing employees responded to the survey.

In Italy, data were gathered via a survey in two medium-sized hospitals in the North-Western area of the country in 2016. Questionnaires were distributed during working hours. Employees were instructed to enclose the completed questionnaire in an envelope and to leave it in a box placed by the researchers in each ward. A total of 557 questionnaires were distributed and 313 questionnaires were returned to the research team (response rate: 56.1%).

Both national subsamples were made up of nurses and nurses aides engaged in direct patient care work activities.

2.2 | Ethical considerations

In the survey in the US hospital system, ethical approval was granted by the Human Investigation Committee of the University and the Research Review Council of the hospital system.

In Italy, the research protocol, designed in agreement with the Helsinki Declaration (and subsequent revisions) and the Italian regulations on data protection and privacy (Law number 196/2003), was approved by the Hospital Administrations and Nurse Coordinators. The questionnaire was accompanied by a letter that openly described the research purpose, the voluntary nature of participation, and the anonymity of the data collection and analysis. In accordance with the country law, no further ethical approval was required since no vulnerable individuals, patients, or minors took part in the study.

2.3 | Measures

Organisational efficiency and workload were measured using two subscales from the Quality-Work-Competence questionnaire (QWC) (Arnetz, Lucas et al., 2011). Organisational efficiency consisted of four items, using a response scale ranging from 1, strongly agree to 4, strongly disagree (e.g. “[At my workplace] Resources are optimally utilized”; α_US = .75; α_ITA = .76).

Workload consisted of three items, employing a response scale ranging from 1, often to 4, never (e.g. “Do you have time to plan your work tasks ahead of time?”; α_US= .75; α_ITA= .76). Previous studies (Arnetz, Lucas et al., 2011; Arnetz, Zhdanova et al., 2011) demonstrated satisfactory reliability in nursing settings for both subscales (α > .80).

Co-worker incivility (CWI) was measured using a scale adapted by Sliter et al. (2012). It contains four items aimed at capturing the frequency of being target of CWI during the past 12 months (e.g. “How often do co-workers ignore you or exclude you at work?” response scale: 1, never, 5, always; α_US= .86; α_ITA = .77). The reliability and validity of this scale were previously demonstrated by international studies involving various types of frontline workers (e.g. Rhee, Hur, & Kim, 2017).

Intention to leave was measured by a single item by Cammann, Fichman, Jenkins, and Klesh (1983) which states: “I often think about quitting”. Responses were given on a five-point Likert type (1, strongly disagree, 5, strongly agree).

Control variables. As suggested by previous literature (Cortina, Magley, Williams, & Langhout, 2001; Magnavita & Heponiemi, 2011; Viotti, Gilardi, Guglielmetti, & Converso, 2015), gender, age and occupation (nurses vs. nurse aides) were included as control variables as they may work as potential confounders in the relationships under study.

2.4 | Analysis strategy

Analyses were performed using SPSS 22 (IBM, 2013) and AMOS (Arbuckle, 2006).

Preliminary analyses included t tests and Pearson’s correlations to examine univariate relationships among study variables.

Multi-Sample Structural Equation Modelling (MS-SEM) was employed to analyse simultaneously data from two national samples to assess invariance of paths across groups. Testing for invariance entails the estimation of a model in which certain parameters are constrained to be equivalent across samples and, then, the comparison of this model with a less restrictive model (non-invariant) in which these parameters are allowed to vary across samples (Byrne, 2013).

Based on Byrne (2013) suggestions, we developed our analyses in a series of steps. In a first step, we estimated the model across two samples simultaneously without imposing any equality constraints. In a second step, we imposed equality constraints factor loadings (λ)¹ across the two subsamples (λ_US = λ_ITA). Through this step we checked whether for multi-item scales, factor-item saturations were invariant across groups. Next, in order to assess the relationships hypothesized, the regression paths (β) were equality constrained across groups (β_US = β_ITA).

The goodness of fit of models were evaluated using the ratio of chi-square to the degrees of freedom (χ²/df), the comparative fit index (CFI), the Tucker-Lewis index (TLI), the standardized root mean square residual (SRMR), and the root mean square error of approximation (RMSEA). According to Schreiber, Nora, Stage, Barlow, and King (2006), model fit is satisfactory if a value of χ²/df is ≤3, CFI and TLI are ≥.95, SRMR and RMSEA are ≤.08.

The CFI-difference test (Meade, Johnson, & Braddy, 2008) was used to determine the presence of invariance. A difference higher than .002 in the CFI value between the less constrained and the more constrained model indicates lack of invariance.

In order to ascertain mediation, we assessed the indirect effects through the bootstrapping procedure (Cheung & Lau, 2008). Bootstrapping is a test that involves ‘resampling’ the data many times with replacements to generate an empirical estimation of the entire sampling distribution of a statistic. According to Cheung and Lau (2008), in order to minimize the risk to incur Type I error, the use of SEM with the bootstrapping procedure requires a sample size ≥200.

3 | RESULTS

3.1 | Descriptive analyses

Table 1 reports sociodemographic data in the two national samples. According to χ² tests, age class distribution significantly differed across subsamples (younger classes were more numerous in the US sample), whereas the Italian sample included more nurses than the US sample (Italy, 92.2%; USA, 76.8%).

TABLE 1.

Background variables

	US sample	Italian sample	χ2-test
Gender
Male	37 (10.9%)	44 (14.1%)	1.52 (n.s.)
Female	304 (89.1%)	269 (85.9%)
Age
<30	102 (29.9%)	30 (9.6%)	52.59***
30–39	64 (18.8%)	45 (14.4%)
40–49	86 (25.2%)	117 (37.4%)
50–59	60 (17.6%)	91 (29.1%)
≥60	29 (8.5%)	30 (9.6%)
Occupation
Registered nurses	262 (76.8%)	307 (92.2%)	12.65***
Nursing aides	79 (23.2%)	26 (7.8%)

Significant at .001***; n.s., not significant; χ², chi-squared.

Table 2 reports the findings of the univariate analyses. Pearson’s correlations carried out in each subsample between major study variables were all significant and in the expected direction. The t tests conducted on each major study variable across subsamples revealed that, whereas workload was significantly higher in the Italian subsample (m_US = 2.22; m_ITA = 2.58), co-worker incivility (m_US = 1.91; m_ITA = 1.66) and the intention to leave (t = 12.25, m_US = 2.92; m_ITA = 1.91) were significantly higher in the US subsample. No significant difference was observed in organisational efficiency.

TABLE 2.

Univariate analyses between study variables within and between subsamples

	US sample				Italian sample				US sample	Italian sample	t test
	1	2	3	4	1	2	3	4	M (SD)	M (SD)
1. Organisational efficiency	1				1				2.77 (.60)	2.71 (.71)	−1.17 (n.s.)
2. Workload	.44***	1			.34***	1			2.22 (.73)	2.58 (.59)	7.01***
3. Co-worker incivility	.28***	.30***	1		.28***	.21***	1		1.91 (.73)	1.66 (.68)	4.55***
4. Intention to leave	.39***	.41***	.33***	1	.23***	.16**	.29***	1	2.92 (1.28)	1.91 (.73)	12.25***

Significant at .001***, .01**, .05*; n.s., not significant; M, mean; SD, standard deviation.

3.2 | Hypothesis testing

The structural model (i.e. MS-SEM) used to test our hypotheses was developed by modelling efficiency, workload and CWI (i.e. multi-item measures) as latent variables, and intention to leave (i.e. single-item measure) as an observed variable. As no serious violations from the normality distribution were found (all the skewness and kurtosis values of the study variables were within ± 2), maximum likelihood, was employed as an estimation method.

In a first step, the hypothesized model was tested across the sub-samples, without imposing any equality constraints. The model fit was satisfactory (Table 3, model a). Hence, a more constrained model was tested by equally constraining all factor coefficients (λ_US = λ_ITA) across the two subsamples. Model fit was satisfactory (Table 3, model b) and all factor loadings were statistically significant in both national samples; however, a CFI difference of .16 (>.002) between this and the unconstrained model (a) was observed, indicating that factor loadings were not invariant across the US and the Italian subsamples.

TABLE 3.

Multi-sample structural equation models (MS-SEM): Goodness-of-fit indexes

	χ2(df)	χ2/df	CFI	TLI	SRMR	RMSEA
(a) Unconstrained model: all parameters are freely estimated	166.45 (98)	1.70	.975	.97	.04	.03 [.02–.04]
(b) Constrained factor loadings (λ)	216.99 (106)	2.05	.959	.95	.05	.04 [.03–.05]
(c) Constrained regressed paths (β)	314.32 (104)	3.02	.923	.90	.10	.05 [.05–.06]
(d) Partially constrained regressed paths (β)	174.33 (101)	1.72	.973	.96	.05	.03 [.02–.04]
(e) Partially constrained regressed paths (β) + control variables	301.35 (155)	1.94	.955	.93	.05	.04 [.03–.04]

χ², chi-squared; df, degree of freedom; CFI, comparative fit index; TLI, Tucker-Lewis index; SRMR, standardized root mean square residual; RMSEA, root mean square error of approximation; λ, lambda, standardized factor loading; β, standardized regressed path.

Next, the invariance of the regression paths across groups was tested (Table 3, model c, β_US = β_ITA). Since factorial invariance was not achieved (λ_US ≠ λ_ITA), at this step, factor loadings were allowed to vary across subsamples. The model fit was not completely acceptable. The ratio χ²/df was slightly above 3 and SRMR greater than .07. Moreover, the CFI difference between the present model and the model with the unconstrained regressed paths (a) was greater than .002. These findings suggested that the relationships among the study variables were not invariant across samples. Hence, we proceed by inspecting the CFI differences for each path separately in order to detect the source of invariance. The paths from efficiency to workload (CFI-difference = .035), from workload to CWI (CFI- difference = .004) and from workload to intention to leave (CFI- difference = .005) were found non-invariant. On the other hand, the following paths demonstrated invariance: efficiency to CWI (CFI- difference = .001), CWI to intent to leave (CFI-difference = .000) and efficiency to intention to leave (CFI-difference = .001). Accordingly, we tested a model in which while the former three paths were unconstrained and the latter three paths were equally constrained across samples (Table 3, model d). As this model reached a satisfactory fit, to further strengthen findings, it was adjusted by the effect of control variables. No equality constraints were imposed for control variables. The model fit was good as well (Table 3, model e) and values of the regressed paths were analogous to those obtained in the non- controlled model (in terms of directions and significance). Whereas in the US subsample all the regressed paths were found to be significant and in the expected direction, in the Italian subsample the paths from workload to CWI and from workload to intention to leave did not show significant associations (Table 4). These results suggested accepting H1, H2, H4, H6 in both subsamples. However, invariance across the subsamples (H7) was confirmed regarding H1, H4, and H6 only. H3 and H5 were accepted in the US samples but not in the Italian samples (Table 4).

TABLE 4.

Regressed coefficient paths in the final model (e)

Path	US sample		Italian sample		Hypotheses
	β	p	β	p
Organisational efficiency-> co-worker incivility	.28	.001	.24	.001	H1
Organisational efficiency-> workload	.53	.001	.38	.001	H2
Workload-> co-worker incivility	.24	.002	.13	.061	H3
Co-worker incivility -> intention to leave	.13	.001	.27	.001	H4
Workload->intention to leave	.36	.001	.02	.774	H5
Organisational efficiency->intention to leave	.12	.001	.23	.001	H6

Italic indicates significant paths; β, standardized path coefficient; p, probability value.

Model was controlled for age, gender and type of occupation. Significant path(s) in the US subsample: Age->workload (β = .22, p = .001); type of occupation (1, nurse aides)->organisational efficiency (β = .14, p = .021). Significant path(s) in the Italian subsample: Age->organisational efficiency (β = .14, p = .021).

Bootstrap tests (Table 5) confirmed the mediating role of workload in the relationship between efficiency and CWI (Ha), CWI in the relationship between efficiency and intention to leave (Hb) and between workload and intention to leave (Hc), respectively, in the US subsample. In the Italian subsample, the only significant indirect effect was observed between efficiency and intention to leave, confirming the mediating role of CWI between these two variables (Hb).

TABLE 5.

Indirect effect estimated by means of bootstrap test

	US sample				Italian sample
	Standardized indirect effect	Confidence interval		p	Standardized indirect effect	Confidence interval		p
		Upper bound	Lower bound			Upper bound	Lower bound
Organisational efficiency -> incivility (Ha)	.13	.06	.21	.01	.05	.01	.12	.15
Efficiency ->intention to leave (Hb)	.24	.17	.31	.01	.09	.04	.14	.01
Workload -> intention to leave (Hc)	.03	.06	.01	.01	.04	.01	.08	.15

Italic indicates significant indirect effects; p, probability value.

4 | DISCUSSION

One of the main contributions of the present study was to analyse the mechanisms linking the characteristics of work processes, co- worker incivility, and the intention to leave in a cross-national perspective. Organisational efficiency has received little attention in previous nursing management literature. The present study revealed its centrality in understanding both co-worker incivility and intention to leave phenomena. In particular, the mediating role of co- worker incivility in the relationship between organisational efficiency and intention to leave was supported in both the US and the Italian samples.

On the other hand, different mechanisms were highlighted regarding workload across the two national samples. In the USA sample, workload was found to contribute to shape the hypothesized loss spiral by mediating the relationship between efficiency and CWI and by affecting intention to leave though CWI. On the contrary, in the Italian sample, workload was not associated with either CWI or intention to leave. An interpretation for these results can be given considering the macro socioeconomic context in each nation. Regarding the relationship between workload and intention to leave, it is important to consider that in Italy, in the past 15 years, a generalized hiring freeze due to the welfare crisis has led to a resulting increase in workload throughout the entire Italian health system. This generalized alarming situation is confirmed by a study involving ten European countries, where Italian nurses emerged to be one of the groups most exposed to high job demands (Hasselhorn, Tackenberg, & Muller, 2003). In this context, nurses might have become accustomed to a heavy workload, therefore not consider it as a motivating factor for quitting. On the other hand, in the USA, where the unemployment rate is very low (USA: about 5% vs. Italy: about 20%), nurses may feel they have more opportunities to improve their working conditions regarding workload, by changing their job/organisation (Eurostat, 2016; Mazurenko, Gupte, & Shan, 2015).

The significant association between workload and co-worker incivility in the USA but not in the Italian context is consistent with the Hofstede theory (1984): in indulgent cultures (e.g. USA), co- workers may be less able to control themselves from rudely treating a co-worker who has trouble getting the job done due to excessive demands (Taylor & Kluemper, 2012). Another feasible explanation for this difference across national settings, consistent with the conservation of resource principle (Hobfoll, 1989), comes from Bolino and Turnley (2005). Specifically, they noted that aggression typically goes beyond an employee’s formally prescribed job duties, and engaging in such acts is likely to require additional employee resources (e.g. time, energy). As such, when job demands are chronically high, such as in the Italian nursing context, employees may need to conserve energy to get the job done. On the other hand, in a context (such as the US nursing setting) in which workload is high but not excessive, workers may use CWI as a “strategy” (even though dysfunctional) to push other co-workers in getting the job done.

However, the present study suggests that the role of workload in loss spiral needs to be further explored. In particular, since previous studies suggested that job demands—even if they deplete energy—are not necessarily negative, future research should try to clarify under which conditions they may represent a risk by fostering CWI and intention to leave.

5 | LIMITATIONS

The most relevant study limitation is the cross-sectional design. Future research should employ a longitudinal design to explore the cross- lagged associations between the constructs examined. Longitudinal studies may also be useful for understanding whether and how the relationships between these constructs changes over time.

Another limitation is that all the measures employed were self- reported. Data coming from a single source may introduce the issue of common method variance. Future studies may benefit from employing research designs that include a combination of objective (e.g. administrative data on nurse turnover) and subjective measures or using data from multiple sources (i.e. co-workers and supervisors).

Finally, response rates differed significantly between the national samples (USA = 24.5%; Italy = 56.1%). Cultural factors may have determined differences in nurse attitudes to the questionnaire surveys. Future research should plan strategies (e.g. communication) aimed at enhancing participation across national settings.

6 | IMPLICATIONS FOR

NURSING MANAGEMENT

The present study has helped to shed light on the key role of co- worker incivility in linking variables describing the quality of the work process and intention to leave the job. The complex interrelations that emerged among these phenomena should be carefully considered when interventions to address CWI in the nursing context are planned. In line with previous studies, our findings suggest that interventions aimed at improving the quality of leadership may help to prevent CWI and intention to leave (Mikaelian & Stanley, 2016) by contributing to enhanced efficiency. The development of leadership skills among frontline nurse managers, through specific training, may contribute to the creation of conditions conducive to a more efficient work environment. Such training would also enable managers to enhance the sharing of work goals, to foster the quality of communication between nurses and to optimize the utilization of job resources in the unit.

Moreover, our finding suggest that reflective practice group (Knight, 2015) may represent a useful tool to sustain both the quality of the work process and relationships among co-workers. Reflective practice group is a technique that provides an opportunity for a work team to collectively examine the daily practice and learn through and from experience, with the end of identifying shared visions regarding ways to improve the working processes. It is a technique widely used in the nursing context, but so far, not specifically considered to address work efficiency and CWI.

Finally, other types of interventions aimed at improving the organisational process may help to reduce CWI as well. For example, regularly scheduling meetings among unit members specifically aimed at organising work activities and sharing objectives may be beneficial (Mikaelian & Stanley, 2016). In addition, employing participative systems to make decisions both at unit and at an organisational level may reinforce the quality of the process and thus the quality of the relationships among co-workers (Laschinger, Leiter, Day, Gilin-Oore, & Mackinnon, 2012).