Effect of the practice environment of nurses on job outcomes and safety climate*

ABSTRACT Objective: to propose and analyze a theoretical model to measure the effect of nurses’ perceptions about the practice environment on safety climate, job satisfaction, intention to stay employed and in the profession, and burnout syndrome. Method: correlational study with probabilistic sample of 465 nurses. In the theoretical model, the dimensions of the nursing practice environment were considered as independent variables and job satisfaction, safety climate, intention to stay employed and in the profession, and burnout were considered the outcome variables. Structural Equation Modeling was the method used in the analysis. Results: small adjustments were made in the model and the dimensions of practice environment predicted job satisfaction (R2 = 43%), safety climate (R2 = 42%) and burnout (R2 = 36%), as well as the intention to stay in the job (R2 = 22%) and in the profession (R2 = 17%). Conclusion: the practice environment showed a strong impact on job satisfaction, safety climate and burnout, with a moderate impact on the intention to stay in the institution and in the profession. The findings can be used to manage care in health institutions, focusing on promoting nurse retention and improving the safety climate.


Introduction
Planning human resources in health and aspects related to patient safety assessment are priorities for health organizations. When it comes to the adequate management of human resources, there is a concern with nursing professionals with respect to the fact that, in addition to providing uninterrupted patient care and making up the majority of health professionals, they are in a position of strategic insertion at the most different levels of attention and assistance scenarios.
It is precisely because they are in this position that the assessment of the nurses' perception of the organizational characteristics of the work environment can provide information to improve the quality of care and safety climate (1) .
A broad range of studies presents the positive results of the perception of the nursing practice environment on variables called job "outcomes". Such results include increased job satisfaction, decreased intention to leave the profession and reduced levels of burnout (2)(3)(4)(5) ; among patients, there is a reduction in mortality in institutions (6)(7) .
The environments to which nurses are exposed are not always favorable to their practice. This situation highlights the importance of studies evaluating and reflecting on organizational safety culture. Researches that relate the perception of the practice environment to the evaluation of safety climate by nurses have been recommended due to the still incipient number of studies, especially national studies, on the theme (1,8) .
Considering previous research as a reference, a theoretical model based on the following research question was proposed: What are the effects of the nursing practice environment on job satisfaction, safety climate, burnout, and the intention to stay in the job and in the profession?
This study aims to propose and analyze a theoretical model that measures the effect of nurses' perceptions about the practice environment on safety climate, job satisfaction, intention to stay employed and in the profession, and burnout syndrome.

Method
This is a correlational study with quantitative approach. For the creation of a theoretical model, we considered the construct of nursing practice environment as an independent variable of second order composed of three constructs (autonomy, control over the environment and nurse-physicians relationships); the other constructs were perceived as dependent variables: safety climate, job satisfaction, intention to stay (in the current job in the next year and in the nursing profession) and burnout.
Previous studies (2)(3)(4)(8)(9) supported the raising of the hypothesis regarding the positive perception of nurses of the practice environment: 1) low levels of burnout; 2) greater intention to stay in the current job in the next year; 3) greater intention to stay in the profession; 4) greater job satisfaction and 5) more positive perception of safety climate.

The statistical method of Structural Equation
Modeling (SEM) was used to estimate the minimum sample size calculated using the free software G*Power 3.1 (10) . The second order construct (nursing practice environment) was considered as the independent variable of the model, an average effect size of 0.15 and power of test of 0.80 were adopted, obtaining 55 cases.
It is recommended to use the triple of this value (10) ; thus, a minimum sample of 165 nurses was estimated according to this parameter.
To calculate the sample size, the population of 104,397 nurses registered at the Regional Nursing Council of São Paulo (COREN-SP), Brazil, was used. Assuming a sample error of 3% and a significance level of 5%, the sample size calculated was 1057 subjects. Nurses were selected through random simple probabilistic sampling and the draw resulted in 1516 subjects; the final sample was composed of 465 subjects.
We included nurses with active professional registration in COREN-SP and who reported developing activities of direct assistance to patients, in a position of supervision in health institutions and having a time of experience equal to or greater than six months in the institution in which they performed their professional activities. Nurses who reported to be exclusively engaged in teaching, management or supervision activities at the institution were excluded, as they did not always carry out direct care activities exclusively with patients.
Nurses who reported absence from work due to medical leave for any reason or any other type of leave in the period of data collection were also excluded. The perception of nurses on the practice environment was evaluated through the Nursing Work Index Revised (NWI-R) (11) , composed of four subscales: autonomy (5 items), control over the environment (7 items), nurse-physicians relationships (3 items) and organizational support (10 items). This last subscale, however, was not used because it contemplated the same items as the previous ones, which could compromise the measurement and the quality of the model to be tested. interpreted in the same way as Cronbach's alpha (10) .
Cross loading values were observed in order to evaluate the discriminant validity; the items should show higher factorial load values in the construct they represent.
The adherence to the criterion was assessed by the square root values of AVE that should be higher than the values of the correlations between the constructs (10) .
After checking the convergent and discriminant validity of the measurement model, the structural model analysis was started. The values and significance of path coefficients (Г) and the quality index of fitness of the tested model were analyzed, namely, the Pearson's coefficient of determination (R 2 ), the relevance or predictive validity (Q 2 ), and the size of the effect (f 2 ) for the outcome variables (10) .
The significance of path coefficients was evaluated by the Bootstrapping method, considering the value of 5000 resampling. After adjusting the model, the interpretation of path coefficients considered the theoretical reference, through which the hypotheses were tested. As for the R 2 values, it was considered that values equal or higher than 0.02 indicate a small effect, equal or higher than 0.13 indicate a medium effect; and equal or higher than 0.26, a large effect (10) .
As for relevance values or predictive validity, values above zero were considered adequate (10) . The contribution of the variables to the adjustment of the final model was evaluated by means of f 2 values for the variables, categorized as small, medium and large, respectively: 0.02, 0.15 and 0.35 (10) .

This research was approved by the Research Ethics
Committee (CAAE 2014/ 30822314.9.0000.5404). A significance level of 5% was adopted in all statistical tests.

Results
The sample consisted of 465 nurses, mostly females  (Table 1).
Regarding discriminant validity, the correlation coefficients values and square root values of the AVE for each of the constructs are presented in Table 2.
At this stage of the analysis, it was identified that the factorial load of item 7 of the subscale control over the environment ("The nursing manager is a good manager and leader") was less than 0.50 and therefore excluded to ensure discriminant validity as recommended in the model analysis.
The convergent and discriminant validity of the model was ensured after exclusion of the five items of the model, and the structural model and goodness of fit of the theoretical model were then evaluated ( Table 3).
The path coefficients values and the significance of the relationships between the constructs are presented in Figure 1 and Table 4.   job satisfaction (2)(3)(4)(5)9) . Other studies have highlighted the contribution of the evaluation of the practice environment and job satisfaction to reduce the intention of nurses to leave their work (13)(14) . Emotional exhaustion was highlighted as one of the primary dimensions of burnout (15) , since it refers to the feelings of overload and Regarding the influence of the environment on the perception of safety climate by nurses, the findings were congruent with recent studies (1,8,15) . to leave the job and job dissatisfaction (4) . Other studies indicate that nurses' positive perception of the practice environment has resulted in increased job satisfaction, lower burnout levels and lower intention to leave the current job and profession (3,5) . and burnout, showed that the environment explained 60% of the results at work (13) .
Among the potentialities and limitations of this study, the findings of the relationships between the constructs analyzed by the theoretical model represent only a small part of this complex reality to be interpreted and measured. Although using a study design of limited scope, the present research evidenced that the nurses' positive perception about the practice environment has a strong impact on the positive perception of safety climate in health institutions. These findings reinforce the hypothesis that organizational behaviors are important in promoting safety climate in health institutions (16) .
One of the limitations implied in working with a larger number of nurses in the sample, as well as the calculation of the response rate in this study, is due to the fact that the participants' electronic addresses may not be updated. It is also worth mentioning that the researchers did not have access to the data of the participants, for this information is exclusive of the

COREN-SP.
We recommend further studies to compare the perception of the nursing practice environment among nurses who provide care and take on management activities so as to enable the identification of specific groups to facilitate the focus for the creation of strategies of management of human resources in nursing.

Conclusion
All the hypotheses of the theoretical model