Validity and reliability study of the Moral Distress Questionnaire in Turkish for nurses

Objective: to determine the validity and reliability of the Turkish language version of the Moral Distress Questionnaire for nurses. Method: methodological study whose sample consisted of 200 nurses working in the internal medicine and surgery clinics of a university hospital. Data was collected using the personal information form and the Moral Distress Questionnaire for nurses. Results: in the Main Components Analysis, the items were grouped under three factors. Findings regarding confirmatory factor analysis: chi-square goodness: 2.28, goodness of fit index: 0.88, comparative fit index: 0.88, non-normed fit index: 0.86, root mean square error of approximation: 0.07. The Cronbach’s alpha coefficient was found to be 0.79 as a result of the analysis conducted in order to test the internal consistency of the scale. It was seen that these three factors explained 44.92% of the total variance. Conclusion: in this present study, the Turkish version of the Moral Distress Questionnaire was found to be valid and reliable for the Turkish society. It is recommended that the Moral Distress Questionnaire for nurses should be used in future studies to be conducted with nurses in order to investigate of issues of ethical dilemma.


Introduction
The concept of moral nuisance/distress was described for the first time in 1984 as "a distress experienced when a health care professional knows the right action to be taken; but it is almost impossible to do the right action due to institutional obstacles" (1) . The nursing literature (2) described moral distress as one of the main ethical problems affecting nurses in all health care systems and defined it as a threat to the integrity of nurses and to the quality of patient care. Therefore, moral distress endangers the ability of nurses to achieve optimal patient care and to obtain high-quality outcomes for patients and, besides, nurses who have moral distress may experience burnout and eventually leave their work (3) .
In recent years, the reconstruction of health care systems around the world, the reduction in inpatient capacity, the increasing awareness about patient rights, the reforms to increase productivity in the healthcare system and the rapid improvement in therapeutical technologies and pharmacological initiatives have made the implementation of nursing care more and more complex, causing an increase the moral distress and ethical climate experienced by nurses (4) . In the process of ethical decision making and implementation by nurses, business friendship, ethicalness of the working environment, physician group, and the style of managers are influential. On one hand, the existence of moral principles that constitute the personal characteristics of nurses and, on the other hand, the existence of ethical principles and the situation of experienced distress form a basis to the situation named moral distress (5) .
The moral distress experienced by health care professionals in health care environments is related to many factors (1) . In the studies conducted, moral distress at an individual level has been associated with depression, anger, guilt, anxiety, shame, sadness, feelings of failure, despair and pain (6)(7)(8)(9) . Lack of communication and cooperation between team members, different perspectives of professionals on ethical issues, limited resources, increased workload due to personnel insufficiency, inconsistency between institutions and health policies, lack of administrative support and a negative ethical climate are among the most important reasons of moral distress at an institutional level (4)(5)(6)(7)(8)(9)(10) .
Several measurement tools for investigating the effect of ethical dilemmas on stress have been reported in the literature (7,9,(11)(12)(13) . These tools include the Moral Distress Scale which assesses moral distress among nurses (7) ; the Moral Distress Assessment Questionnaire which assesses moral distress experiences in terms of frequency, type, intensity, and duration (11) ; the Stress of Conscience Questionnaire which assesses a measurement of stress arising from a disturbed conscience (9) ; The instrument of Moral Distress which assesses the daily experience of health care personnel in various environments (12)

Method
The methodological objective of this study was to characteristics of the participants (1,6,10) . It consists of 9 three of them (item 2, 5, 12) were adopted considering qualitative findings of the Stress of Conscience Questionnaire (9) and the rest (item 1, 6, 7, 9, 14) were adopted considering the qualitative findings of The instrument of Moral Distress (12) . All the items of the questionnaire are positive and ranked in the range of 1-6 points as "Strongly Disagree", "Agree" (13) .   (15)(16) . Exploratory Factor

Analysis/Main Components Analysis, and Confirmatory
Factor Analysis were used for factor construct validity. calculated probability of error (p-value) of the Bartlett sphericity test is below 0.05 (15,(17)(18) .
Items with a factor load value of 0.30 or higher in the Confirmatory Factor Analysis (CFA); and 0.32 or higher in the exploratory factor analysis were taken to factor constructs.  (17,19) . Reliability is the power of a measurement tool to present sensitive, coherent, and stable measurement results.
200 randomly selected participants were asked to find a nickname for themselves and to indicate it on the questionnaire during their first participation. The same scale was applied to the test group after 2 weeks and they were asked to re-write the same nickname.
Afterwards, the questionnaires with the same name were matched and re-test results were obtained. for factor analysis was used. As a result of this analysis, the eigenvalue of three factors was found to be above 1 and it was seen that these three factors explained 44.92% of the total variance. According to the confirmatory factor analysis, the factor loads for the model are shown in Figure 1.

51.5% (n=103) of the nurses participating in
The significance p value gives information about the difference (value) between the observed covariance matrix and the expected covariance matrix. The p value is expected to be significant in CFA (17,19) . The values on  Table 2.   *χ^2⁄sd = Chi-square goodness; † RMSEA = Root mean square error of approximation; ‡ CFI = Comparative fit index; § NNFI = Non-normed fit index; || SRMR = Standardized root mean square residuals; ¶ GFI = Goodness of fit index; **AGFI = Adjusted goodness of fit index The fit index, which should be examined first in CFA, is Chi-square (X 2 ) fit statistic, and it says that if the ratio to the degree of freedom is less than 3, it shows perfect fit; and if it is below 5, it shows good fit (20) . This ratio was found to be 2.28. RMSEA is the square root of the mean of error squares and it says that, in order for the model to be significant, a value below 0.05 means perfect fit and a value below 0.10 means good fit (20)(21)(22) . The RMSEA value was found to be 0.08 and it shows good fit ( Table 2). CFI is a fit index that compares the covariance matrix predicted by the model with the covariance matrix of the null hypothesis model (17)(18)(19)(20) . The CFI takes values ranging between 0 and 1. It can be concluded that a model with a CFI value between 0.95 and 1 has a good fit and a model with a CFI value between 0.90 and 0.95 has an acceptable fit (17,(20)(21)(22) . Some researchers have taken the value of 0.80 as a more flexible limit (23) . Although the CFI (0.88) and NNFI (0.86) values calculated for the best model that can be established in this study are below the generally accepted value, it can be said that the model is acceptable due to its complexity ( Table 2).
The GFI shows the amount of general covariance between the variables calculated and observed by the assumed model. The GFI value ranges from 0 to 1. It is considered as a good model if the GFI value exceeds 0.90. This means that enough covariance has been calculated among the observed variables (20) . The GFI value was found to be 0.88, and it indicates good fit.
In addition, AGFI means adjusted fit index and it was found to be 0.85, indicating good fit (  (16,24) . The very high level of correlation coefficient between the first and second implementation demonstrates the reliability of the participants' responses.
The exploratory factor analysis method, which was one of the factor analysis methods suggested in the literature (25)(26) , was used to evaluate the construct validity for the scale development study. In the literature (25)(26) it is stated that, as a result of the sampling adequacy test result, the KMO value should be, at least, 0.50; a value between 0.50-0.60 is considered bad; a value between 0.60-0.70 is considered weak; a value between 0.70-0.80 is considered fair; a value between 0.80-0.90 is considered good and a value higher than 0.90 is considered perfect, when the factor analysis method is used (15,18,22) . The KMO value was found to be 0.77 for the MDQ. It was concluded that the sampling size was moderately enough for factor analysis. The decision on whether the structure would be divided into factors or not after the adequateness of the sampling size is analyzed with the Bartlett sphericity test. It is stated that the factors can only be revealed when the significance value obtained from the analysis is less than 0.05 (16,22,26)  It is stated in the literature (19,22,(25)(26) that the Cronbach's alpha coefficient changes between 0 and 1; a coefficient value between .60 and .80 indicates that the scale is very reliable and a coefficient value of .80 and above indicates the scale is perfectly reliable.
In this study, the Cronbach's Alpha coefficient of the This situation was evaluated positively in terms of the reliability of the items and no addition or subtraction process was carried out on the items of scale (18)