Psychometric properties of the quality nursing care scale-turkish version: a methodological study

This study aimed to evaluate the psychometrics of the Quality Nursing Care Scale in Turkish.

This was a methodological and cross-sectional study conducted at a training and research hospital. All nurses working in the hospital's inpatient wards were included in the study. The study sample consisted of 225 nurses accepted to participate in the study. The mean age of the participant nurses was 33.81 (SD = 7.70) years. Nurses’ professional and unit experiences were 8.95 (6.94) and 4.29 (4.37) years, respectively. Among the nurses, 86.7% were women, 61.8% were married, and 59.1% were undergraduates: they primarily were working at surgical units (40.0%), both day and night shifts (78.2%). Most of the nurses were working 46 or more hours weekly (73.8%) and caring for 11 or more patients in each change. Also, 66.2% of them were working evening and night shifts six or more times in a month (Table 1).

The study followed the required scale adaptation steps according to the International Test Commission and Consensus-based Standards for the Selection of Health Measurement Instruments (COSMIN) guidelines [20‐22]. Translating the items and measuring the content validity was the first step. Secondly, item-total score correlations of the items were calculated. Then construct validity was tested. Lastly, the internal consistency of the scale was analyzed.

Research data were collected online (Google Forms) between September and November 2021. The online surveys were sent online to the nurses working in the institution. They were asked to fill out the forms. Only those who filled out the informed consent form could access the survey, and only the researchers could access the results.

The literature recommends reaching out to individuals 5–10 times the number of items in reliability and validity studies [23]. Since the number of items in the scale to be validated and tested for reliability was 38, researchers aimed at reaching a sample of at least 190 people. The study sample consisted of 225 nurses working at the time of the study. The study data were collected using a questionnaire containing questions to determine personal and professional characteristics and the Turkish version of the Quality Nursing Care Scale.

It was a form consisting of 18 questions that sought data on the age, gender, marital status, and education level of the nurses participating in the research, the unit they worked in, their working style, their working hours as a nurse, their monthly duty hours, their perception of income, their satisfaction with the institution, unit, working conditions, and salary they received.

The Quality Nursing Care Scale (QNC) is a 5-point Likert-type tool developed by Liu et al. [10]. The scale consisted of six sub-dimensions and 38 items in its original form. The sub-dimensions were entitled Physical environment (six things), Staff characteristic (eight items), Precondition (seven items), Task-orientated activities (six items), Human-orientated activities (five items), and Patient outcomes (six things). The internal consistency coefficient of the scale was 0.96. The answer categories were graded between "strongly agree" (5) and "strongly disagree" (1). High scores indicate higher quality nursing care, and low scores indicate lower quality nursing care.

The researchers got permission from the original work's owner to adapt the tool into Turkish. The original scale's author also confirmed that the scale had not been previously adapted to Turkish. Approval of a university hospital's Clinical Research Ethics Committee, dated 05.05.2021 and numbered 98, was obtained. Before data collection, formal written permission was obtained from the hospital administration. Only nurses who agreed to participate in the study and filled out the online informed consent form were included.

Data were analyzed via Jamovi, an R-based open statistical software [24]. First, the Davis technique was used to analyze the content and content validity in the study. Second, Pearson correlation analysis was used for item analysis. Third, Kaiser–Meyer–Olkin and Bartlett's tests were used to evaluate sample adequacy. Fourth, exploratory factor analysis was used to determine the construct of the Turkish version. Extraction model was principal axis factoring and rotation method was direct oblimin. Confirmatory factor analysis confirmed the new structure. Finally, Cronbach's alpha internal consistency coefficient was calculated. The accepted significance level was 0.05 for a 95% confidence interval.

The Turkish version of the scale was presented to 13 nursing management and internal medicine nursing specialists. As a result of the analysis using the Davis technique, the items' content validity ratios (CVR) ranged from 0.85–1. Therefore, the content validity index of the scale was 0.96 after the content validity rate of each item was summed and divided by the total number of items obtained. The Turkish version was then translated into English by two academicians, one a medical doctor and the other a nurse with a Ph.D.

Correlation values obtained from item-total correlation analyses with 38 items regarding the Turkish version of the QNC are shown in Table 2. As a result of the investigation, the item-total score correlation coefficients of the items differed between r = 0.72 and 0.92.

The KMO value was high at 0.975, and Bartlett’s test was significant at the < 0.001 level. Then, confirmatory factor analysis was performed first. However, the × 2/df value was 9.82 for the six factored structures. Then an explanatory factor analysis was made to understand the new structure in the Turkish language. The results showed that the factor loads of the items ranged from 0.42 to 0.93. Three subscales explaining 80.806% of the total variance were revealed. First, items 4, 5, 6, 7, 8, 9, and 10 were cross loaded to factors 1 and 2. Next, items 24, 26, and 27 crosses loaded in Factor I and 3.

However, the differences in the factor load values of the same items in different factors were over 0.300. Therefore, those items remained in the factors that got higher factor loads.

Then confirmatory factors analysis was repeated for the new structure, and the × 2/df value was calculated as 3.85 (Table 3). The Comparative Fit Index, Standardized Root Mean Square Residual and the Root Mean Square Error of Approximation indices were 0.90, 0.041, and 0.011, respectively (Fig. 1).

Cronbach’s alpha coefficient of the Turkish version of the Scale was 0.99. The same coefficient for the first subdimension was 0.95, and there were 0.99 for the second and third subdimensions (Table 4).

Since the translated items might not mean the same in the adapted language [23, 28], experts were asked to evaluate the restated items in terms of meaning. Minor revisions were made considering the experts' recommendations to clarify their meanings in Turkish. For example, wording arrangements were made for Item 5 (I provide a quiet ward environment for patients staying in the hospital), Item 14 (I work well with my team [other nurses and healthcare providers]), Item 16 (I master the clinical, technical operations to meet the needs of nursing care), Item 20 (I can manage drugs well), and Item 21 (I intend to help patients whenever the help is needed). Then, two academicians who knew both languages backtranslated the items into English.

Davis technique, a frequently used method developed by a nurse researcher, was used for the content validity analysis [29]. The original and Turkish items were presented to the experts and asked to compare based on their meanings and grammatical structure. It was observed that the experts mainly assessed the Turkish items in the scale as “quite appropriate.” The lowest CVR value was 0.80, an acceptable value in the literature [29, 30].

This study evaluated the compatibility of 38 items using item-total score correlations. It was seen that the correlation, or coherence, of each item of the scale with the whole scale.

Although it is recommended that the scale's construct validity be evaluated using confirmatory factor analysis for adaptation studies [31], this study used explanatory factor analysis because the original model did not fit in Turkish based on the confirmatory factor analysis results. Since the literature indicated that the scale's construct would not have the same structure in the local context [23, 32], the authors decided to perform a new explanatory factor analysis to explore the system in Turkish. Unlike in the original study, the items were distributed into three subscales instead of six (Table 5). Only the Physical environment and Patient outcomes subscales remained the same. Staff characteristic, Precondition, Task-orientated activities, and Human-orientated activities subscales merged into one subdomain in the Turkish context. When the statements were evaluated carefully, the meanings of the statements were precise in Turkish. However, it was assessed that nurses perceived the statements under one subdomain because they perceived all items under Staff characteristic, Precondition, Task-orientated activities, and Human-orientated activities subscales related. The authors evaluated that those items differed from others because they were mainly associated with primary nursing roles and tasks. For example, the items in the physical environment subdomain might also be related to the other staff and the hospital environment. Patient outcomes were related to the customers' perspective. However, the items that emerged with one subdomain were directly associated with nurses and nursing. Therefore, the authors named that domain "Nursing."

The Cronbach's alpha internal consistency test, the most commonly used test to assess the internal consistency of Likert scales, evaluated the reliability of the subscales and scale in Turkish [33]. Although the acceptable Cronbach's alpha coefficient level has varied in previous reports, the commonly recommended threshold value is 0.70 [34‐36]. The Cronbach's alpha coefficients of the scale and subscales in Turkish form were relatively high and indicated high reliability. The coefficients were also higher than the original work's values.

Although we planned to reach out more participants and to perform confirmatory and exploratory factor analysis on different sample groups in study setting, we could not reach out enough number of nurses because of the huge working conditions during COVID-19 Pandemic.