Study design

This study employed a methodological, cross-sectional design aimed at the cultural adaptation and psychometric validation of the ECCS. The cross-sectional design was chosen because it allowed assessment of the factor structure, validity, and reliability of the instrument within a single time frame, without any intervention. Participant recruitment was conducted from 20 July 2024–15 January 2025. The study population comprised all registered nurses working in seven hospitals located in the western region of Mazandaran Province and affiliated with Mazandaran University of Medical Sciences, Iran. These hospitals were selected based on their geographical accessibility and representativeness of major clinical environments, including Intensive Care Unit (ICU), Cardiac Care Unit (CCU), Post-CCU, Internal Medicine, Surgery, Emergency, and Oncology wards. This study was part of a larger project (contract code 724135862) at the Health Research Institute, Babol University of Medical Sciences.

Population and sampling

The inclusion criteria were as follows: (a) being officially employed as a registered clinical nurse in one of the hospitals affiliated with Mazandaran University of Medical Sciences located in the western part of the province; (b) having at least one year of clinical work experience; and C willingness to participate and provide informed written consent. Exclusion criteria included failure to complete the instruments fully or withdrawing from the study before submission. The sample size was determined based on the rule of thumb that there should be at least 10 participants for each scale item, i.e., an ideal ratio of respondents to items is 10:1 [25]. Considering 22 items in the ECCS instrument and allowing for possible attrition, a total of 697 nurses were included. In addition to this rule, the final sample size was set to approximately 697 to account for the elimination of incomplete responses, the 20‑participant pilot test, and the need to ensure adequate statistical power for advanced analyses including Exploratory Graph Analysis (EGA), Random Forest Model (RFM), and Measurement Invariance testing, as well as to allow for split‑sample validation (EFA n = 348 and CFA n = 349). Sampling was conducted using the convenience method.

Data collection process

The data collection process followed a structured and sequential procedure and was conducted systematically and in person. After obtaining official approval from the Deputy of Research of Mazandaran University of Medical Sciences and the nursing directors of the seven selected teaching hospitals, the first author who was solely responsible for the entire data collection process obtained a list of registered nurses employed in each hospital from the nursing administration offices. No research assistants or data collection personnel were involved, as the researcher independently managed data collection through a pre‑planned schedule across hospitals and work shifts. Accordingly, data were gathered sequentially over a six‑month period, enabling the researcher to visit each hospital in turn and cover all three work shifts (morning, evening, and night). In coordination with nursing managers and head nurses, the researcher entered each ward, introduced herself, explained the study objectives and significance, and invited nurses to voluntarily participate. Nurses who agreed to take part provided written informed consent, and the paper‑based survey instruments were distributed to them in person. To accommodate the shift‑based nature of nursing work and to prevent interference with patient care, the researcher arranged a suitable time with each participant for questionnaire completion and follow‑up. The completed instruments were collected by the researcher personally at the pre‑agreed time in each ward. On average, participants completed all sections within approximately 10–15 minutes. This method ensured that data collection was carried out in an organized, ethical, and interference‑free manner while maintaining confidentiality, voluntary participation, and the accuracy of responses.

Data were collected using the following instruments:

Demographic characteristics questionnaire

This questionnaire included variables such as age, gender, work experience, work shift, job satisfaction, income adequacy, financial status, educational level, department of service, marital status, and type of employment within the organization.

Ethical Caring Competency Scale (ECCS)

The ECCS was originally developed by Katayama et al.  (2022) to assess nurses’ ethical caring ability in clinical practice. The instrument comprises 22 items organized into four dimensions: “Expressing the sensitivity and value of good care” (4 items), “Acting while thinking about how to provide better care” (9 items), “Creating indirect effects to provide better care” (6 items), and “Acting to learn what better care is” (3 items). Each item is rated on a five‑point Likert scale ranging from 1 (never) to 5 (always). The total possible score ranges from 22 to 110, with higher scores indicating a greater level of ethical caring competency. Although the original developers did not define a diagnostic cut‑off, the current study used a distribution‑based tertile classification to facilitate interpretation: scores 22–58 reflect low, 59–86 reflect moderate, and 87–110 reflect high ethical caring competency. The internal consistency reliability of the original instrument for its four dimensions ranged from Cronbach’s α = 0.72 to 0.89 [6].

Translation

The translation and cultural adaptation of the ECCS were carried out following the World Health Organization (WHO) protocol for instrument adaptation. First, the original English version of the instrument was translated into Persian independently by two bilingual translators (Forward Translation). The two forward‑translated versions were then reviewed and synthesized by a multidisciplinary panel comprising translators, nursing researchers, and medical‑ethics experts to produce a single reconciled version (Reconciliation). Next, the reconciled Persian version was translated back into English by two native English translators who were unaware of the original instrument (Back Translation). This step aimed to ensure semantic and conceptual consistency between the source and target versions. Subsequently, an expert committee consisting of researchers, linguists, and ethics specialists compared all versions and finalized the culturally and linguistically appropriate Persian version (Expert Committee Review). Finally, the pre‑final version was tested with a group of 20 nurses from the target population (Pre‑testing) to confirm clarity, comprehensibility, and cultural relevance of the translated items. After this stage, the finalized version of the ECCS was prepared for psychometric evaluation.

Face validity

In this study, the face validity of the instrument was assessed both qualitatively and quantitatively. In the qualitative stage, 12 nurses from the target population reviewed the Persian version of the ECCS and provided comments on the clarity, simplicity, and cultural relevancy of the items. According to their feedback, three items (Items 2, 4, and 20) underwent minor rewording in wording and sentence structure to enhance clarity, simplicity, and cultural appropriateness within the Iranian clinical context. After incorporating these adjustments, quantitative face validity was assessed. In this phase, the same 12 nurses rated item importance on a 5‑point Likert scale, and all item impact scores exceeded 1.5, confirming that every item was clear, comprehensible, and highly relevant to the intended construct [26].

Content validity

The content validity of the instrument was assessed both qualitatively and quantitatively. In the qualitative content validity stage, a group consisting of 12 experts in the fields of nursing, professional ethics, and psychometrics were invited to review the instrument in terms of clarity, appropriateness of the items to the intended concept, and content comprehensiveness. In this stage, items identified as inappropriate by the experts were rewritten to better align with the instrument’s objectives. To assess quantitative content validity, two indices, the Content Validity Ratio (CVR) and the Content Validity Index (CVI), were used. To measure CVR, experts were asked to rate the necessity of each item on three levels: essential, useful but not essential, and not essential. Items with a CVR value equal to or greater than 0.56 were retained as essential items in the instrument [27]. The CVR was calculated using the following formula:

Where nₑ represents the number of experts who rated the item as “essential,” and N denotes the total number of experts participating in the content validity assessment.

After determining the necessity of the items, experts rated the Clarity, Simplicity, and Relevance of each item on a 4-point Likert scale, and then the CVI value for each item was calculated. Items with a CVI value equal to or greater than 0.79 are considered to have acceptable content validity [28,29]. Furthermore, the average CVI for the entire instrument (S-CVI/Ave) was also calculated. An S-CVI/Ave value above 0.90 indicates a desirable level of content validity for the entire instrument [29].

Construct validity

Construct validity was assessed using Exploratory Factor Analysis (EFA) and Confirmatory Factor Analysis (CFA).

Exploratory factor analysis

EFA was performed on data obtained from a random subsample of 348 participants to examine the underlying factor structure of the 22 items of the ECCS. Data adequacy for factor analysis had been verified prior to extraction. The KMO value should be at least above 0.60 for the data to be considered suitable for factor analysis [30]. Also, Bartlett’s Test of Sphericity was used to examine the significance of the correlation between items. The significance of this test (p < 0.05) indicates the existence of sufficient correlation to perform factor analysis [31,32]. Principal Axis Factoring (PAF) was used for factor extraction, applying the criterion of Eigenvalues > 1. To allow for correlated factors, Promax rotation was performed. The optimal number of factors was determined based on both Scree Plot inspection and Parallel Analysis. Parallel analysis was performed with 1,000 iterations, comparing the real eigenvalues with those obtained from randomly generated datasets of identical size; factors with eigenvalues exceeding the 95th percentile of simulated data were retained. Items with factor loadings below 0.40 were excluded from the model [32].

Confirmatory factor analysis

CFA was conducted to evaluate the fit of the four‑factor model extracted from EFA using the independent subsample reserved for cross‑validation. Several fit indices were employed to assess the adequacy of model fit, including Chi‑square/df < 3, Comparative Fit Index (CFI) ≥ 0.90, Tucker–Lewis Index (TLI) ≥ 0.90, Root Mean Square Error of Approximation (RMSEA) < 0.08, and Standardized Root Mean Square Residual (SRMR) < 0.08  [33].

Normal distribution of data, outliers, and missing data

In this study, to improve the accuracy of statistical analyses, the normality of data, outliers, and missing data were examined and managed. The normal distribution of data was assessed using skewness and kurtosis indices (within the ± 1 range) and the Kolmogorov-Smirnov and Shapiro-Wilk tests [34]. To identify outliers, first, univariate outliers were identified using Z-Score ≥ ±3, and then multivariate outliers were identified and removed via Mahalanobis Distance at a significance level of p < 0.001 [32]. Furthermore, missing data were examined for randomness using Little’s test, and if no specific pattern was present, missing data were managed using appropriate methods [35].

Convergent and discriminant validity

In this study, convergent validity and discriminant validity, as subsets of construct validity, were assessed using CFA and Structural Equation Modeling (SEM). To examine convergent validity, the average variance extracted (AVE) for each construct was calculated. According to the criteria proposed by Fornell & Larcker, an AVE value higher than 0.5 indicates desirable convergent validity and signifies that more than 50 percent of the construct’s variance is explained by its items [36]. Discriminant validity was also assessed based on the Fornell & Larcker method. This method confirms discriminant validity when the square root of the AVE of each construct is greater than its correlation with other constructs. If this condition is met, it indicates that the constructs are well-differentiated from each other, and the instrument has appropriate discriminant validity [36].

Reliability

The reliability of the ECCS was evaluated through two complementary approaches. Internal consistency was examined using Cronbach’s alpha and McDonald’s omega, which assess the degree to which all items consistently measure the same construct; coefficients above 0.70 were considered acceptable [37,38]. Stability (test–retest reliability) was assessed by re‑administering the scale to a subsample of 30 nurses after a two‑week interval and calculating the Intraclass Correlation Coefficient (ICC) between the two administrations. An ICC greater than 0.70 indicated satisfactory temporal stability of the instrument [39].

Ceiling and Floor Effects

In this study, the ceiling and floor effects of the ECCS were examined to ensure that the instrument can adequately discriminate among respondents across all levels of the construct. Ceiling and floor effects occur when a considerable proportion of participants achieve the highest or lowest possible scores, which undermines the sensitivity of the instrument. To assess these effects, the score distribution was inspected, and the percentages of respondents obtaining the minimum and maximum scores were calculated. Following the quality criteria proposed by Terwee et al. (2007) a widely accepted guideline for evaluating measurement properties of health status questionnaires a ceiling or floor effect was considered present if more than 15% of respondents reached the extreme scores [40].

Standard Error of Measurement (SEM)

In this study, SEM was calculated to assess the precision of the instrument and the extent of potential variation in participants’ true scores due to random errors. SEM is a measure that indicates how much an observed score might differ from an individual’s true score (Rejas et al., 2008). For this purpose, the instrument’s reliability coefficient (Cronbach’s alpha) and the standard deviation of the total scores were calculated, and using the relevant formula, the SEM value was obtained. Subsequently, the 95% confidence interval for the true score was estimated using the SEM value for each individual [37,41].

Feasibility and acceptability

In this study, to assess the feasibility and acceptability of the instrument, several key indicators were examined [37,41]. Feasibility was evaluated through criteria such as the time taken to complete ECCS, data completeness, and clarity of items. To determine the completion time, participants were asked to record the time required to complete the instrument. Also, the percentage of instruments that were fully completed was recorded as an indicator for examining data completeness. Furthermore, in the pilot stage, participants were asked to provide their opinions on the clarity and comprehensibility of the items, and items that were deemed ambiguous or difficult by them were revised. To assess acceptability, indicators including response rate, qualitative feedback from participants, and dropout rate were examined. The percentage of individuals who responded to all items of the instrument was considered as a measure of response rate, and the dropout rate was considered as another indicator for acceptability. Additionally, qualitative feedback was collected from participants regarding their experience of completing ECCS to determine their satisfaction with the instrument.

Measurement Invariance across Gender

In this research, to assess the consistency of the instrument across gender groups, multi-group analysis based on SEM was used. This approach allows for the examination of an identical factor structure between different groups. In this regard, three levels of invariance, including configural invariance, metric invariance, and scalar invariance, were tested according to the guidelines proposed by Cheung & Rensvold [42]. SEM is a fundamental principle in psychometrics that determines whether a measurement instrument functions in the same way across different groups (such as gender) [43].

Exploratory Graph Analysis (EGA)

To identify the factor structure of the ECCS, we employed EGA, a novel network‑psychometric approach that represents variables as nodes and their conditional dependencies as edges in a graphical model. After computing the item‑correlation matrix, we used the Graphical LASSO to estimate the partial correlation network, from which the number and composition of factors were derived. This data‑driven, network‑based technique served as a cross‑validation of the factor structure obtained from EFA. The convergence between EFA and EGA results provides empirical evidence for the structural stability of the scale, indicating that the identified dimensions are not sample‑specific but reflect robust latent constructs. For readers less familiar with advanced psychometrics, EGA offers a transparent, visual representation of how items cluster together, thereby enhancing confidence in the scale’s factorial integrity and supporting its practical applicability in diverse nursing contexts [44].

Random Forest Model (RFM)

RFM was employed as a complementary machine‑learning approach to evaluate the relative importance of each component and item of the ECCS in predicting the total ethical‑caring‑competency score. The model builds an ensemble of decision trees through random sampling of both observations and predictors, which enables it to capture complex, non‑linear relationships among items while effectively preventing overfitting [45]. Beyond its predictive accuracy, the RFM output provides a practical ranking of influential items, thereby identifying which dimensions of ethical caring competency contribute most substantially to the overall score. This information directly supports targeted interventions and assessment prioritization in clinical and educational settings, making the scale not only a measurement tool but also a decision‑aid for nursing managers and educators.

Statistical analysis

Prior to statistical analyses, data were screened for completeness and accuracy. Incomplete or inconsistent responses (e.g., surveys with more than 20% missing items) were excluded from the final analysis. EFA was performed using SPSS v26 to examine the underlying factor structure of the ECCS. CFA and measurement invariance testing were conducted using the maximum likelihood estimation method in AMOS v24. EGA was implemented using the EGAnet package in R, and RFM analyses were performed using the randomForest package in R to assess item and component importance. Descriptive statistics, skewness and kurtosis evaluation, and ceiling/floor effects were calculated using SPSS. Reliability indices, including Cronbach’s alpha, were estimated via SPSS, and McDonald’s omega was computed using the psych package in R. Convergent and discriminant validity were assessed based on the Fornell and Larcker criterion, with AVE and CR calculated in AMOS v24.

Ethics approval and consent to participate

The study was conducted in accordance with the Declaration of Helsinki and received ethical approval from the Ethics Committee of Babol University of Medical Sciences (Code: IR.MUBABOL.HRI.REC.1403.047). All methods were carried out in accordance with the relevant guidelines and regulations. The objectives of the ongoing study were explained to all participants and informed written consent was obtained from them. In details, obtaining permission for audio recording, explaining the objectives and methodology to participants, informing participants of their right to withdraw at any stage of the research, adhering to the principle of non-maleficence, maintaining confidentiality and anonymity of participant information, and offering to share the research results with participants if they wished was part of the ethical considerations in the study.

Descriptive statistics

In this study, nurses with a mean age of 35.17 ± 8.41 years, ranging from 22 to 57 years, were examined. Also, the mean work experience of nurses was 10.55 ± 7.24 years, with a range of 1–30 years. The results show that most nurses, in terms of gender, 562 (80.60 percent) were female; in terms of marital status, 439 (63 percent) were married; in terms of education level, 623 (89.4 percent) had a bachelor’s degree; in terms of place of service, 195 (28 percent) were in the emergency department; in terms of work shift, 103 (14.8 percent) had a fixed morning shift; in terms of job satisfaction, 374 (53.7 percent) were average; in terms of income adequacy, 470 (67.4 percent) reported less than sufficient; in terms of financial status, 458 (65.7 percent) were average; and in terms of employment type, 369 (52.9 percent) were permanent.

Face and content validity

The results of the qualitative face validity assessment showed that most items were deemed appropriate in terms of clarity, simplicity, and relevance to the intended concept. However, some items underwent minor revisions due to cultural incongruity or conceptual ambiguity. For instance, minor linguistic simplifications were made to enhance clarity and cultural appropriateness for example, refining the phrase “having a bird’s‑eye view of caring based on laws, rules, and social trends” to a more accessible wording for nurses while preserving its original meaning. In the quantitative face validity, all items had a value higher than 1.5. This indicated that all items were sufficiently important from the respondents’ perspective, and therefore, no items were deleted at this stage. In the qualitative content validity assessment, experts evaluated all items in terms of conceptual appropriateness, clarity, and comprehensiveness. Some items were revised after receiving corrective feedback, but no items were completely deleted. The CVR value for all items was equal to or higher than 0.56, indicating the necessity of all items according to the experts. The CVI value for all items was also equal to or higher than 0.79, indicating confirmation of the instrument’s content validity. The overall S-CVI/Ave mean was calculated to be higher than 0.90, which confirmed the desirable level of the instrument’s content validity.

Exploratory factor analysis

The results show that the sample size is adequate for performing factor analysis, with a KMO equivalent to 0.948, indicating that the data are sufficient for analysis. Also, Bartlett’s test was significant (). Therefore, it indicates that there is sufficient correlation between the items to proceed with factor analysis. The EFA results indicate the presence of four factors in the ECCS, including: Acting while thinking about how to provide better care, creating indirect effects to provide better care, Expressing the sensitivity and value of good care, and acting to learn what better care is. These four latent factors explain 19.56 percent, 13.37 percent, 10.41 percent, and 9.46 percent of the variance, respectively, and in total, they explained 52.80% of the total variance of ECCS. All items had an acceptable factor loading higher than 0.4 (Table 1). The final Persian version of the ECCS retained all the 22 original items and confirmed the same four‑factor structure identified in the original scale. No items were deleted after factor extraction, and the Likert‑type scoring (ranging from 1 = Never to 5 = Always) as well as the total score range (22–110) remained unchanged.

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Extracted Exploratory Factors after Promax Rotation.

https://doi.org/10.1371/journal.pone.0346672.t001

To confirm the four factors in ECCS, a scree plot was also used. As observed in Fig 1, the presence of four factors is also confirmed based on the scree plot.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Scree plot based on eigenvalues for factor extraction.

https://doi.org/10.1371/journal.pone.0346672.g001

Confirmatory factor analysis

To evaluate the model fit, all indices: PCFI = 0.737, PNFI = 0.713, CMIN/DF = 2.76, RMSEA = 0.071, IFI = 0.934, CFI = 0.933, and GFI = 0.918 confirmed the appropriate fit of the model. Therefore, the four-factor model of ethical care competency was confirmed (Table 2).

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. Fit Indices of the ECCS Confirmatory Factor Analysis Model.

https://doi.org/10.1371/journal.pone.0346672.t002

Fig 2 shows the standardized factor loadings between the items and the factors of the ethical care competency construct in the first-order factor analysis. All factor loadings in the model are above 0.4.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. First-Order Confirmatory Factor Analysis of ECCS.

https://doi.org/10.1371/journal.pone.0346672.g002

Fig 3 shows the standardized factor loadings of each factor and the construct in the second-order confirmatory factor analysis. In the second-order confirmatory factor analysis, the values of the fit indices indicate an acceptable fit of the proposed model with the data.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. Second-Order Confirmatory Factor Analysis of ECCS.

https://doi.org/10.1371/journal.pone.0346672.g003

Convergent and discriminant validity

The results obtained from calculating AVE showed that all constructs have values higher than 0.5, and therefore, the convergent validity of the instrument was confirmed. Also, comparing the square root of AVE for each construct with its correlation with other constructs showed that the condition proposed by Fornell and Larcker holds for all constructs. These findings indicate that the instrument has desirable discriminant validity, and the measured constructs are well-differentiated from each other (Table 3).

Download:

• PNG
  larger image
• TIFF
  original image

Table 3. Convergent Validity, Discriminant Validity, Internal Consistency, and Stability of ECCS Factors.

https://doi.org/10.1371/journal.pone.0346672.t003

Reliability

The findings related to Reliability showed that ECCS has very good internal consistency and temporal stability. Cronbach’s alpha (α) and McDonald’s omega (Ω) coefficients for all factors were reported to be higher than 0.85, indicating high consistency among the items of each factor. Also, CR was greater than 0.84 for all factors, indicating the desirable structural stability of the instrument. Furthermore, ICC values are in the range of 0.83 to 0.89, which signifies high stability in repeated assessments (Table 3).

Ceiling and floor effects assessment

In this study, the ceiling and floor effects for the four factors are less than 15%; therefore, no ceiling and floor effects are observed, and the total score is normally distributed (Table 4).

Download:

• PNG
  larger image
• TIFF
  original image

Table 4. Ceiling and Floor Effects.

https://doi.org/10.1371/journal.pone.0346672.t004

Standard Error of Measurement (SEM)

The examination of SEM in the ethical competency construct shows that the instrument used has appropriate precision for evaluating different components. SEM values in all subscales are relatively low, ranging from 0.70 to 2.14, which indicates acceptable measurement error. Also, the Minimum Detectable Change (MDC) values in all components are greater than SEM, and the Minimum Important Change (MIC) values are also lower than MDC in all cases, which indicates the instrument’s ability to identify meaningful and clinical changes. The presence of positive agreement in all components also strengthens the construct validity of the results (Table 5).

Download:

• PNG
  larger image
• TIFF
  original image

Table 5. Standard Error of the Ethical Competency Construct.

https://doi.org/10.1371/journal.pone.0346672.t005

Feasibility and acceptability

The findings showed that ECCS performs adequately in terms of Feasibility and Acceptability. The average time required to complete the scale was 10 minutes. 96.4% of the scales were fully completed, indicating data completeness. In the pilot phase, only 3.2% of participants reported some items as ambiguous, and these items were revised after review. In terms of acceptability, 94.8% of participants responded to all items of the instrument, and the dropout rate throughout the instrument completion process was 1.7%. Also, 89% of respondents reported their overall experience of completing the instrument as “satisfactory” or “completely satisfactory.” These results indicate that the ECCS instrument not only has conceptual accuracy but is also simple to implement, understandable, and acceptable to the target population.

Measurement invariance across gender

The results of the multi-group confirmatory factor analysis (MGCFA) showed that the 4-factor model of ECCS has a good fit in both male and female groups. In the first stage, by imposing the constraint of equality of factor loadings between the two groups, no significant difference in the value was observed (), which indicates the establishment of factor loading invariance. Subsequently, by constraining variances and factor covariances, the results still indicated a good fit of the model and no significant difference (), which indicates structural invariance. In the third step, by also imposing constraints on the measurement residuals, the model still had a good fit (). Also, the index in all stages of analysis (for factor loadings = 0.001, for covariances = 0.002, and for residuals = 0.005) was less than the threshold of 0.01, which means maintaining model invariance at a desirable level. However, although invariance in the factor structure and relationships between factors was established in the male and female groups, at the level of measurement residuals, signs of difference were observed, which may be due to cultural or experiential differences between the two groups (Table 6).

Download:

• PNG
  larger image
• TIFF
  original image

Table 6. Results of Multi-Group Confirmatory Factor Analysis Designs in Different Subgroups.

https://doi.org/10.1371/journal.pone.0346672.t006

Exploratory Graph Analysis (EGA)

EGA identified and estimated the four-factor structure of ECCS. The results showed that all items were correctly clustered with their respective factors, and this structure aligns with the findings from EFA. Furthermore, bootstrap analysis with 1000 iterations showed that the median number of factors was four, with a standard error of 0.060. The 92% confidence interval for the number of factors was between 4.19 and 6.47. The results indicated that in 99.4% of the bootstrap samples (994 out of 1000 samples), a four-factor structure was identified. These findings indicate the stability and predominance of the four-factor structure in the data and support the factorial validity of the scale at the network level (Fig 4).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 4. Results of Network Analysis Factors for ECCS.

https://doi.org/10.1371/journal.pone.0346672.g004

Random forest model

The results show that the component “acting while thinking about how to provide better care” with 37.56% was the most important component, and the items “learn about good care from reflection and insights based on the experience case” and “exploring diverse values and awareness without sticking to own values” with 27.21% and 26.92% respectively, were identified as the most important items (Fig 5).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 5. Variable importance plot derived from the Random Forest model for the ECCS.

https://doi.org/10.1371/journal.pone.0346672.g005

Strength and limitations

This study is considered one of the most comprehensive research projects conducted on the cultural adaptation and evaluation of the psychometric properties of ECCS in the Iranian nursing community. To ensure the accuracy, validity, and cultural compatibility of the instrument, standard cross-cultural adaptation frameworks, including forward and backward translation, review by a panel of experts, and pilot testing in the target population, were utilized. This research examined all psychometric dimensions, including face, content, and construct validity (through EFA, CFA, and EGA), as well as convergent and divergent validity. Furthermore, the Feasibility and Acceptability of the instrument were assessed, and its reliability was evaluated from two aspects: Cronbach’s alpha and McDonald’s omega for internal consistency, and through test-retest for assessing the temporal stability of the instrument.

Among the prominent innovations of this study was the use of EGA as a novel method for examining the factorial structure, which confirmed the high stability of the ECCS four-factor structure. Also, the examination of measurement invariance based on gender showed that the factorial structure of the instrument is stable between male and female nurses and allows for inter-group score comparisons. Moreover, in this research, RFM was used for the first time to analyze the relative importance of the factors and items of the ECCS instrument; an approach not employed in the Japanese version of the scale. This analysis enabled the precise identification of the most effective components influencing ethical care competency and introduced a novel approach to the psychometric content analysis of ethical instruments. The use of a large and diverse sample (n = 697) from various hospital departments has also increased the generalizability of the findings to the Iranian nursing community.

Despite its numerous strengths, this research is also accompanied by limitations. The cross-sectional design, while appropriate for psychometric validation, precludes causal inference and does not allow examination of the temporal stability of ethical caring competency. The use of a convenience sampling method may reduce the generalizability of the results to the entire population of nurses. Furthermore, the research’s focus on nurses working in hospitals in a specific geographical region (west of Mazandaran province) could subject the findings to the influence of regional cultural and organizational factors. However, it should be noted that Iran’s healthcare delivery system is highly centralized; all public hospitals nationwide operate under standardized policies and professional protocols mandated by the Ministry of Health and Medical Education. Therefore, organizational culture and professional practices are largely similar across provinces, minimizing potential regional bias. In addition, because the vast majority of Iranian nurses share a common religious background, where respect for patients and protection of patients’ rights are emphasized as fundamental moral and religious duties, the ethical orientation underlying nursing practice is relatively homogeneous across the country. This cultural and religious consistency may further reduce regional variation in ethical caring behaviors. It is recommended that in future studies, this instrument be evaluated in more geographically and professionally diverse populations. Future longitudinal or repeated-measure designs would be valuable to examine changes in ethical caring competency over time and to establish causal relationships.

Clinical implications

The results of this study confirmed that ECCS, due to its ability to identify weaknesses in nurses’ ethical competency, can be used in in-service training, monitoring nurses’ performance, and as a basis for developing professional improvement programs. Hospital managers and nursing managers can use ECCS to assess the level of ethical competency of nurses and utilize its results in designing managerial strategies to promote ethical standards in healthcare settings. Furthermore, this instrument is applicable in developing specialized educational programs in the field of professional ethics. Conducting skill-enhancement courses and training workshops based on ECCS data plays an important role in improving ethical decision-making processes and strengthening nurses’ professional skills. On the other hand, the self-assessment capability provided by ECCS allows nurses to independently gauge their ethical competencies, identify their strengths and weaknesses, and, based on that, develop individual improvement strategies. This process helps increase ethical self-awareness and enhance the quality of decision-making in complex clinical situations. Finally, the widespread use of ECCS can provide valuable information to health policymakers and contribute to the development of guidelines based on professional ethics. At a macro level, this will lead to the improvement of care standards, enhancement of the quality of nursing services, and strengthening of ethical principles in the healthcare system.