Cross-cultural adaptation and psychometric testing of the Evidence-Based Practice Mentoring Scale in Chinese nurses

Abstract

Objectives

Mentoring has been identified as a promising strategy for implementing and sustaining evidence-based practice (EBP) in healthcare organisation. However, no appropriate tools were specifically developed or cross-culturally adapted into Chinese context to assess nurse’s perceived EBP mentoring, impeding comprehensive evaluation of the effects of mentoring intervention studies. This study aimed to cross-cultural adapt the Evidence-Based Practice Mentoring (EBPM) scale into Mainland China and evaluate its psychometric properties, including validity and reliability.

Design

A comprehensive translation and adaptation process was adopted to achieve the Chinese version of the EBPM (C-EBPM) scale. It consists of four steps: (1) trilateral translation procedure, (2) cognitive interview, (3) psychometric testing and (4) cross-time confirmatory factor analysis (CFA).

Setting

This study was conducted in four 3A-level hospitals located in Shaanxi and Zhejiang provinces, China, during two different data collection periods.

Participants

A total of 598 registered nurses participated in this study.

Results

After two rounds of the trilateral translation procedure, a 9-item version of the C-EBPM scale was generated. Ten registered nurses participated in cognitive interview understood the meaning of all items but the response options. All items had significant critical ratio values (t=15.866~20.584, p<0.001). The item-total correlations ranged from 0.865 to 0.940 (p<0.001). The item-level Content Validity Index and item-level Translation Validity Index were 1.000. Horn’s parallel analysis suggested that one factor should be extracted, which accounted for 84.656% of the total variance. Factor loadings extracted from principal axis factoring ranged from 0.862 to 0.942. Cronbach’s α, ordinal α, McDonald’s ω and Guttman split-half coefficient all exceeded 0.900. The one-factor CFA model provided an acceptable fit: χ²/df=65.681/27<3, root mean square error of approximation=0.073 (90% CI 0.051~0.096) < 0.080, Comparative Fit Index=0.974 > 0.950, Tucker-Lewis Index=0.966 > 0.950, and standardised root mean square residual=0.026 < 0.080. Composite reliability was 0.93 (>0.70) and average variance extracted was 0.60 (>0.50).

Conclusions

The 9-item C-EBPM scale demonstrated robust reliability and validity and is suitable for assessing EBP mentoring among nurses.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• This study used a systematic cross-cultural adaptation procedure and evaluated construct validity across two independent nurse samples at different time points, enhancing methodological robustness.

• The 9-item Chinese version of the Evidence-Based Practice Mentoring Scale exhibited strong psychometric properties in Chinese nursing populations.

• As our participants were only recruited from 3A-level hospitals in two provinces in China, generalisability to all Chinese nurses may be limited.

• Concurrent validity could not be examined due to the absence of alternative scales measuring evidence-based practice mentoring in nurses.

BACKGROUND

Evidence-based practice (EBP) is a lifelong problem-solving approach to healthcare decision making that integrates the best available research evidence, clinical expertise and patient preferences to improve outcomes for individuals, groups, communities and healthcare systems.¹ It has been widely recognised as the gold standard of healthcare for its effectiveness in enhancing care quality, empowering healthcare professionals, improving patient outcomes and reducing hospital costs.^{2 3} However, despite these well-established benefits, the implementation of EBP remains suboptimal worldwide.⁴,⁶ One key strategy to advance and sustain EBP implementation is Evidence-Based Practice Mentoring (EBPM), which refers to a structured, collaborative process in which experienced mentors provide guidance, feedback and role modelling to help mentees develop EBP-related knowledge, skills and confidence, while fostering a culture of inquiry within healthcare organisations.⁷ Growing evidence indicates that EBPM enhances nurse’s EBP competencies, thereby promoting the adoption and sustainability of evidence-based care.⁸,¹⁰ Nevertheless, the effective integration of EBPM into routine nursing practice remains limited, underscoring a critical gap between evidence and implementation.

Although EBP mentoring has been recognised as a promising strategy to promote evidence-based healthcare,¹¹,¹⁵ few instruments are available to evaluate the degree of EBP mentoring perceived by nurses. Over the past two decades, most instruments have focused on assessing clinical nurses’ individual EBP knowledge, skills, attitudes, beliefs or implementation behaviours (eg, the EBP knowledge scale, the EBP attitude scale and the EBP implementation scale),¹⁶,²⁰ while others examined contextual factors through instruments such as the EBP work environment scale, the EBP nursing leadership scale²¹, and the organisational culture and readiness for system-wide integration of EBP.²² However, little attention has been paid to evaluating the extent of mentoring support that nurses receive during the EBP process, such as formulating clinical questions, retrieving evidence and making practice changes. This measurement gap limits the ability to identify whether nurses have access to adequate EBP guidance, to design targeted mentoring programmes, and to evaluate their effectiveness in improving EBP adoption and sustainability. To address this gap, Melnyk and colleagues developed the EBPM scale in 2016, which demonstrated strong reliability and validity in 2344 nurses and provides a concise and effective tool to quantify EBP mentoring availability in clinical settings.²³

Similar gaps exist in Mainland China. Although several EBP-related instruments have been developed or cross-culturally adapted to assess individual EBP competencies or organisational support, no existing instrument evaluates the extent of EBP mentoring support that nurses receive throughout the EBP process, and no equivalent tool has been culturally adapted for Chinese clinical organisations.²⁴,²⁸ This also limits the ability to assess mentoring availability, identify mentoring needs, design mentoring programmes and evaluate mentoring effectiveness in Mainland China. Therefore, this study aimed to cross-culturally adapt the EBPM scale into Chinese context and to determine its psychometric properties among Chinese nurses, providing a valid tool to assess EBP mentoring for future intervention design and evaluation.

METHODS

We followed a comprehensive translation and cultural adaptation process proposed by the WHO and subsequently refined by Duan et al, to develop the Chinese version of the EBPM (C-EBPM) scale.^{29 30} It consisted of four main steps: trilateral translation procedure, cognitive interview, psychometric testing and cross-time confirmatory factor analysis (CFA) (see figure 1).

Figure 1. The translation and adaptation process. This figure illustrates the multistep procedure used to translate, culturally adapt and validate the C-EBPM scale, including forward and back translation, expert panel review, cognitive interviews, psychometric testing and confirmatory factor analysis. C-EBPM, Chinese version of the Evidence-Based Practice Mentoring.

Patient and public involvement

Patients and the public were not involved in the design, conduct, reporting or dissemination of this research. This study aimed to adapt and validate a professional instrument among registered nurses (RNs); therefore, patient or public involvement was not applicable. However, improving the assessment of EBP mentoring support may indirectly benefit patients by enhancing nurses’ capacity to deliver evidence-based care.

Study setting

This study was conducted in 3A-level hospitals located in Shaanxi and Zhejiang provinces, representing western and eastern regions of China. In Shaanxi Province, a total of 332 registered nurses (sample 1 and sample 2) were recruited from three hospitals, each employing approximately 1000 registered nurses. In Zhejiang Province, 266 registered nurses (sample 3) were recruited from one hospital with approximately 1300 registered nurses. All data collection procedures were conducted in Mandarin Chinese, the only official and working language used in public hospitals in Mainland China.

Measure — the EBPM scale

The EBPM scale was originally developed by Bernadette Melnyk and colleagues to assess the degree of EBP mentoring available to American RNs.²³ It includes 8 items rated on a 5-point Likert scale (0: None at all, 1: A little, 2: Somewhat, 3: Moderately So, 4: Very Much So). Adding the score for each item produces the total score and it ranges from 0 to 32, with higher scores indicating higher EBP mentoring support received from experts, leaders, colleagues or any other potential mentors. This one-dimensional scale was originally tested in 19 US hospitals with 2344 RNs varied in age, race and education.²³ EBPM scale has established a high internal consistency reliability of 0.99 and good construct validity. Permission to use this scale was obtained from Dr Melnyk, the author as well as copyright holder of the EBPM scale prior to the study.

Step 1: Trilateral translation procedure

The trilateral translation procedure includes forward translation (FT) and back-translation (BT). To ensure translation accuracy, an expert panel was convened, as recommended in the guidelines, including at least one methodologist experienced in instrument development and translation, one healthcare professional familiar with the construct being measured, and one linguist.^{29 30} In this study, three master students who were fluent in English (original language) and Chinese (target language) participated in the forward translation phase (see online supplemental table 1). They were responsible for translating the English EBPM scale into Chinese independently, and the first author (MW) synthesised the three translations into one forward-translated version of the C-EBPM scale, named FT1. Subsequently, another two master students and one doctoral student were asked to back-translate the FT1 into English independently. They had never seen the original EBPM scale because it was not publicly available prior to authorisation from the developer and having passed high-level English proficiency examinations. MW compared the three back-translations and synthesised them into BT1. Any discrepancies among FT1, BT1 and the original EBPM scale were discussed with the expert panel, consisting of seven experienced experts in, and iterative rounds were conducted until consensus was obtained. Consensus was defined in this study as 100% agreement among expert panel members that no further modifications were required for conceptual equivalence and translation clarity. Dr Melnyk served as an independent external consultant and was consulted only when clarification of the original instrument was required.

Step 2: Cognitive interview

Cognitive interview is used to explore the target populations’ understanding of the translated instruments, and the findings of it are conducive to refine the translated version.^{31 32} A sample size of no less than 10 individuals is recommended.³³ Purposive sampling was adopted to gain a sample as diverse as possible regarding age, education and work experience. Inclusion criteria were respondents who (1) had been working as a registered nurse for at least 1 year when they took part in this survey, (2) had good communication skills and (3) were able to give their informed consent. RNs who had only completed the technical nursing education were excluded because EBP education was not available in secondary schools in China. Anonymity and confidentiality were assured, and informed consent was obtained. All participants were interviewed face to face though WeChat (a Chinese social media platform) in May 2020 in Shaanxi, Shanxi and Henan Provinces, China. Each of them was asked to answer specific questions related to understanding of the instructions, items or other parts of the scale and to rate whether the items are clear. Examples of the probing questions are listed as follows: (1) Did you have any difficulty using the response options? (2) Was there any word you did not understand? (3) In your own words, what does this question ask?

Step 3: Psychometric testing

When a translated version met the priori consensus criteria in Step 1 and revisions from Step 2 were incorporated, psychometric testing was conducted. It included item analysis, exploratory factor analysis (EFA), validity analysis and reliability analysis. In this step, a convenience sample of the target population was recruited for further improving the C-EBPM scale. We used the recommended sample size of at least 10 subjects per item of the instrument.³⁴ RNs with at least 1 year of clinical experience were eligible to participate in this study. Potential eligible participants were identified by several head nurses of three 3A-level hospitals in Shaanxi Province, China. Data were collected using a professional online survey tool (www.wjx.cn) between August and September 2020. WeChat was used to distribute the electronic questionnaire, which includes a self-designed questionnaire focused on the demographic characteristics of RNs and the C-EBPM scale. Each participant received the electronic questionnaire with an opening paragraph describing the rationale for the study and the guarantee of the anonymity and confidentiality of their private information. Informed consent was obtained online.

Step 4: Cross-time CFA

In this step, CFA was conducted to re-evaluate the construct validity of the C-EBPM scale with another convenience sample of nurses collected at two different moments. A minimum of 200 samples is required for performing CFA.³⁵ Data used in this step were collected between September and October 2021 from one 3A-level hospital in Zhejiang province, China. RNs with at least 1 year of clinical working experience in the selected hospital were recruited. The professional online survey tool (www.wjx.cn) and WeChat were used to display and distribute the questionnaire. The questionnaire is comprised of a self-designed questionnaire focused on the demographic characteristics of RNs and the C-EBPM scale. All participants gave their informed consent and were guaranteed the confidentiality of private information.

Statistics analysis

SPSS (version 24.0), Mplus (version 7.0) and R (version 4.5.1) software were used for data description, analysis and graphics. Descriptive statistics were used to report the demographic characteristics of all participants, including ranges, means and SDs for continuous variables, as well as frequencies and percentages for categorical variables. All tests were two-tailed with a significance level of 0.05.

Item analysis was used for selecting discriminating items for inclusion in the final version of the scale. A critical ratio (CR) test was first performed to select the 27th and 73rd percentile of the total score as the cut-off point, representing the low and high groups, respectively.³⁶ The significance of the differences between the means of the low and high groups was determined using independent sample t-test. A CR value less than 3.0 or a p value greater than 0.05 served as the benchmarks for deleting an item.³⁷ An item-total correlation (r) test was performed to check the correlation between an item and the total score. An item with an r value less than 0.4 or a p value greater than 0.05 indicated a weak association between this item and the scale, thus leading to elimination of it.³⁷ If Cronbach’s α increased considerably on deletion of an item, the item would also be discarded. Items with ≥80% of responses concentrated in either the lowest or highest category (floor or ceiling effect) were considered for deletion from the scale.³⁷

EFA was conducted using principal axis factoring (PAF) to extract common variance, with factor retention determined by Horn’s parallel analysis.³⁸ Sampling adequacy was evaluated using the Kaiser-Meyer-Olkin (KMO) measure and Bartlett’s test of sphericity. Item retention was guided by factor loadings (0.45~0.95), conceptual coherence and expert judgement, with theoretical considerations equally weighted to ensure content validity. The cumulative explained variation of factors extracted reached 60% is acceptable.³⁴

Validity analysis in our study includes content validity, translation validity and construct validity. Content Validity Index (CVI) and Translation Validity Index (TVI) rated on a 4-point scale were used to identify the content validity and translation validity for each item and the scale. Experts were invited to give their response and to provide suggestions for revising the items. The item-level CVI (I-CVI) and item-level TVI (I-TVI) were computed as the number of experts giving a rating of 3 or 4 for each item divided by the total number of experts. The average of the I-CVI or I-TVI for all items is the scale-level CVI (S-CVI) or scale-level TVI (S-TVI). It is recommended that a minimum I-CVI and I-TVI should be 0.780, and a minimum S-CVI and S-TVI should be 0.900.^{37 39}

Reliability analysis included internal consistency and test-retest reliability. Internal consistency was evaluated using Cronbach’s α, ordinal α (appropriate for Likert-type data), McDonald’s ω and Guttman split-half coefficient. Test-retest reliability (external reliability) was assessed by administering the C-EBPM scale to 30 participants twice within a 2-week interval and calculating the intraclass correlation coefficient (ICC). Values ≥0.700 were considered acceptable, ≥0.800 were considered good and ≥0.900 were considered excellent reliability, for all indices.⁴⁰,⁴²

CFA uses structural equation modelling (SEM) to test whether a new set of data fit the hypothesised measurement model proposed by EFA. Maximum likelihood (ML) estimation was used for CFA because the items of the C-EBPM scale were measured on five-point Likert scales and demonstrated acceptable item-level distributional properties (skewness: −0.578 to −0.130; kurtosis: −0.808 to 0.162), supporting their treatment as approximately continuous.⁴³ The fit indices are recommended: (1) χ²/df (χ² value/df), (2) RMSEA (root mean square error of approximation), (3) CFI (Comparative Fit Index), (4) TLI (Tucker-Lewis Index) and (5) SRMR (standardised root mean square residual). The CFA model is considered to have an acceptable fit if χ²/df<3, RMSEA <0.080, CFI >0.950, TLI >0.900 and SRMR <0.080.^{44 45} In addition to overall model fit indices, composite reliability and average variance extracted (AVE) were computed to evaluate internal consistency and convergent validity of the latent construct. Values >0.70 for composite reliability and 0.50 for AVE are considered acceptable.⁴⁶

RESULTS

Trilateral translation procedure

After two rounds of the trilateral translation procedure, a 9-item version of the C-EBPM scale was generated. The four main changes made from the expert panel to the scale were the revision of item 1, item 3 and item 7, and the addition of item 9. The original statement of item 1 was “I have access to a mentor who assists me with implementing the seven steps of evidence-based practice”. ‘A mentor’ was translated into ‘one mentor’ in FT1, which was deemed inappropriate by experts because ‘a’ in English is an indefinite article. In addition, the steps in implementing EBP vary from institution to institution, so seven may not be fully accurate. Therefore, item 1 was revised into “I have mentors who assist me with implementing the steps of evidence-based practice”. Item 3 was “Mentorship is available here to assist me in rapid critical appraisal of studies and synthesis of evidence”. The translators involved in forward translation were all confused about the meaning of ‘rapid’. Dr Melnyk replied that they developed a series of Rapid Critical Appraisal (RCA) Checklists for the sake of improving one’s understanding about how to appraise different types of evidence. Thus, item 3 was expressed as “I have mentors who assist me in critical appraisal of studies and synthesis of evidence” in the Chinese version. Likewise, item 7 was changed from “Mentorship is available here to assist me in disseminating evidence through presentations and publications” into “I have mentors who assist me in disseminating evidence by making presentations and publishing papers”. Besides, a new item (Item 9: I have mentors who assist me in retrieving evidence according to PICO questions) was proposed by experts on account of the importance of conducting a systematic search of the literature based on the PICO question, which was approved by Dr Melnyk.

Cognitive interview

We recruited 10 RNs as sample 1 (see table 1). They aged between 23 and 49, with a mean age of 33.90 years old. The mean work experience was 11.90 years, ranged from 1 to 28. The majority were female (90%) and had received a higher education (90%). Half of them (50%) earned their highest academic qualifications through continuing education. They worked in different departments, including internal medicine (10%), surgery (10%), gynaecology (20%), obstetrics (10%), et al. Few nurses (20%) possessed senior professional title, and 40% were head nurses.

Table 1. Demographic characteristics of samples 1, 2 and 3.

Characteristic	Sample 1 (n=10)	Sample 2 (n=322)	Sample 3 (n=266)
Age, years
Range	23~49	19~58	21~53
Mean	33.90	30.32	29.98
SD	8.49	6.78	7.18
Work experience, years
Range	1~28	1~37	1~30
Mean	11.90	8.23	7.44
SD	9.76	7.22	7.35
Gender
Female	9 (90%)	318 (98.76%)	261 (98.12%)
Male	1 (10%)	4 (1.24%)	5 (1.88%)
Academic qualification
Technical nursing education	–	9 (2.80%)	1 (0.38%)
Associate degree	1 (10%)	82 (25.47%)	2 (0.75%)
Bachelor’s degree	4 (40%)	220 (68.32%)	246 (92.48%)
Master’s degree	5 (40%)	11 (3.42%)	17 (6.39%)
Study mode
Full-time	5 (50%)	82 (25.47%)	212 (79.70%)
Part-time	5 (50%)	240 (74.47%)	54 (20.30%)
Department
Internal medicine	1 (10%)	55 (17.08%)	–
Surgery	1 (10%)	43 (13.35%)	21 (7.89%)
Gynaecology	2 (20%)	54 (16.77%)	100 (37.59%)
Obstetrics	1 (10%)	74 (22.98%)	104 (39.10%)
Maternity ward	3 (30%)	51 (15.84%)	38 (14.29%)
Paediatrics	1 (10%)	32 (9.94%)	–
Traditional Chinese medicine	1 (10%)	–	–
Others	–	13 (4.04%)	3 (1.13%)
Professional title
Primary	5 (50%)	237 (73.60%)	165 (62.03%)
Middle	3 (30%)	76 (23.60%)	98 (36.84%)
Senior	2 (20%)	9 (2.80%)	3 (1.13%)
Position
Head nurse	4 (40%)	29 (13.12%)	37 (13.91%)
Staff nurse	6 (60%)	192 (86.88%)	229 (86.09%)

Nurses participated in cognitive interview understood the meaning of all items but the response options. Nearly all nurses (90%) complained about the 5-point scale (none at all, a little, somewhat, moderately so, very much so) in the original scale. They said that the distance between the five options was not equal, and their understanding about ‘a little’ and ‘somewhat’ varied widely. We chose the most commonly used selection options including ‘strongly agree’, ‘agree’, ‘neutral’, ‘disagree’ and ‘strongly disagree’ with the statement to tackle this problem.

Psychometric testing

A total of 322 RNs (sample 2) responded effectively in this stage. They aged between 19 and 58, with a mean age of 30.32 years old (see table 1). The mean work experience was 8.23 years, ranged from 1 to 37 years. The majority were female (98.76%). About 71.74% had obtained at least a bachelor’s degree, while 2.8% received the technical nursing education. Most of them (74.47%) earned their highest academic qualifications through continuing education through degree programmes. Nearly three quarters (73.06%) possessed primary professional title, and 86.88% were staff nurses.

Table 2 displays the results of item analysis. All items had significant CR values (t=15.866~20.584 > 3, p<0.001), far more than the cut-off value. The item-total correlations of all items were greater than 0.4, ranging from 0.865 to 0.940 (p<0.001). Cronbach’s α coefficient of the C-EBPM scale for sample 2 was 0.9771, and the value did not increase when any of the items was deleted. Factor loadings extracted by PCA were all greater than 0.45 and less than 0.95, ranging from 0.837 to 0.937. All items had met the retention requirements and were not deleted. All five Likert scale response categories were used across items in both sample 2 and sample 3. The highest single-category endorsement observed was 57.8% (item EBPM5, response of 4) in the EFA sample (n=322) and 40.2% (item EBPM4, response of 3) in the CFA sample (n=266), both below the commonly accepted 80% threshold, indicating no severe floor or ceiling effects and supporting adequate discriminative capacity.

Table 2. Item analysis of the C-EBPM scale (n=322).

Item	Critical ration test	Tests for homogeneity
	CR	R value	Cronbach’s α	Factor loading
1	18.693**	0.928**	0.9739	0.921
2	17.172**	0.904**	0.9752	0.885
3	16.199**	0.920**	0.9743	0.929
4	20.584**	0.930**	0.9738	0.911
5	16.204**	0.933**	0.9736	0.928
6	19.814**	0.940**	0.9733	0.937
7	19.399**	0.923**	0.9741	0.915
8	15.866**	0.865**	0.9771	0.837
9	20.333**	0.935**	0.9735	0.919

Note: **represents p<0.001; Cronbach’s α of the C-EBPM scale was 0.9771. CR, critical ratio.

C-EBPM, Chinese version of the Evidence-Based Practice Mentoring.

We invited five experts, including Dr Melnyk, to evaluate the content validity and translation validity of the C-EBPM scale. By estimate, the I-CVI and I-TVI for each item were 1.000, and the S-CVI and S-TVI for this scale were 1.000 (see online supplemental table 2). EFA was performed to determine the construct validity of the C-EBPM scale. Prior to performing EFA, the KMO test (KMO=0.939 > 0.900) and Bartlett’s test (χ²=4356.465, df=36, p<0.001) were performed, indicating the suitability of EFA. Horn’s parallel analysis supported a one-factor structure. Using PAF, one factor was extracted (extraction sum of squared loadings=7.450), accounting for 82.774% of the variance (see table 2 and figure 2). As for the reliability analysis of the C-EBPM scale, Cronbach’s α coefficient was 0.977 (>0.900), ordinal α=0.988 (>0.900), McDonald’s ω=0.988 (>0.900) and Guttman split-half coefficient was 0.959 (>0.800) with this sample (n=322). Test-retest reliability with a sample size of 30 participants was 0.886 (>0.700).

Figure 2. Horn’s parallel analysis scree plot for the Chinese version of the Evidence-Based Practice Mentoring (C-EBPM) scale (n=322). This figure presents the results of Horn’s parallel analysis conducted to determine the optimal number of latent factors for the C-EBPM scale. The purple line represents the observed eigenvalues derived from the empirical data, whereas the green and yellow lines correspond to the mean and percentile eigenvalues generated from randomly simulated datasets. Factors were considered meaningful when the observed eigenvalues exceeded the corresponding random eigenvalues.

Cross-time CFA

We subsequently conducted CFA with sample 3 (n=266) to identify whether the one-factor structure derived using EFA was empirically supported. They aged between 21 and 53, with a mean age of 29.98 years old (see table 1). The mean work experience was 7.44 years, ranged from 1 to 30 years. The majority of nurses were female (98.12%). Nearly all RNs (98.87%) had obtained at least a bachelor’s degree, and 79.70% earned their highest academic qualifications through full-time degree programmes. More than half (62.03%) possessed primary professional title, and 86.09% were staff nurses.

The one-factor CFA model provided an acceptable fit for the sample 3 data: χ²/df=65.681/27<3, RMSEA=0.073 (90% CI 0.051~0.096) < 0.08, CFI=0.974 > 0.950, TLI=0.966 > 0.950, and SRMR=0.026 < 0.080. As presented in figure 3, all items had salient standardised loadings on this latent factor, ranging from 0.683 to 0.829 (p<0.001), demonstrating the stability of the one-factor structure of the C-EBPM scale. Cronbach’s α, ordinal α, McDonald’s ω and Guttman split-half coefficient with this sample were 0.927, 0.945, 0.945 and 0.905, respectively. Besides, composite reliability was 0.93 (>0.70) and AVE was 0.60 (>0.50), both exceeding the recommended cut-off values, further indicating strong internal consistency and convergent validity of the one-factor C-EBPM scale (see online supplemental table 3 and 4 for the items of C-EBPM).

Figure 3. Confirmatory factor analysis of the Chinese version of the Evidence-Based Practice Mentoring (C-EBPM) scale (n=266). This figure displays the standardised confirmatory factor analysis (CFA) model of the C-EBPM scale. The latent construct (C-EBPM) is represented by the oval, and the observed items (items 1–9) are shown as rectangles. Single-headed arrows indicate the factor loadings of each item on the latent construct, with standardised estimates displayed along the paths. Circles labelled e1–e9 represent the measurement error terms associated with each observed item. EBP, evidence-based practice. PICO: Population, Intervention, Comparison, and Outcome.

DISCUSSION

In this study, we systematically translated, culturally adapted and psychometrically validated the EBPM scale for use among Chinese nurses. Following a rigorous four-step procedure and testing the adapted instrument in two independent nurse samples, the 9-item C-EBPM scale showed excellent internal consistency, stable unidimensional structure and good convergent validity. Our findings indicate that the C-EBPM scale is a concise and psychometrically sound instrument suitable for quantifying perceived EBP mentoring support in clinical nursing settings. To the best of our knowledge, this is the first instrument specifically designed to measure EBP mentoring in the Chinese context, filling a critical measurement gap in EBP research and practice.

A recent systematic review involving 40 EBP-related instruments across healthcare disciplines confirmed that no available measure captures EBP mentoring, despite nurses representing the profession with the greatest number of EBP measurement tools.⁴⁷ This absence is notable because mentoring has been increasingly identified as a core mechanism enabling EBP uptake.⁹ The Advancing Research and Clinical practice through close Collaboration (ARCC model) positions EBP mentors as key drivers for advancing system-wide EBP by providing hands-on guidance throughout evidence translation,^{48 49} and evidence from a quasi-experimental study showed that structured EBP mentoring significantly improved nurses’ competencies and EBP culture.¹⁵ Without a valid tool to assess mentoring, organisations may invest in EBP training programmes without determining whether nurses actually receive EBP mentoring during implementation. The C-EBPM directly fills this gap, allowing organisations to quantify mentoring availability, detect mentoring deficits and evaluate the impact of mentoring interventions.

The C-EBPM scale has several important implications for research and practice. First, its brevity (9 items, <2 min to complete) enhances feasibility in busy clinical settings, increasing response rates and enabling repeated use in quality improvement cycles. Second, because the items correspond to key steps of the EBP process, the scale can evaluate not only the presence of mentoring, but also whether mentoring effectively supports EBP at each stage. Third, the addition of one culturally relevant item (“mentors assist me in retrieving evidence according to PICO questions”) indicates that the C-EBPM is capable of detecting contextual mentoring needs. Chinese nurses consistently report database navigation and literature searching as major obstacles in conducting EBP, which differs from findings in Western settings where nurses receive more formal training in evidence retrieval.⁵⁰

In terms of methodological contribution, this study adopted an EFA-first and CFA-second strategy using two independent samples. Although the original English version had an established unidimensional structure, EFA was necessary because cross-cultural adaptation may alter item functioning, and one additional culturally relevant item was added during the adaptation. Conducting EFA before CFA ensured that the expanded C-EBPM scale retained factorial coherence, and the subsequent CFA provided strong confirmation of factor stability. This sequential approach aligns with recommended practices in cross-cultural psychometric research.³⁵ Besides, the combination of cognitive interviewing and cross-time CFA is still underused in cross-cultural adaptation studies,⁵¹ yet these procedures revealed conceptual issues not detectable through translation alone and confirmed factor stability across independent samples. Nonetheless, several limitations must be acknowledged. Convenience sampling from 3A-level hospitals limits generalisability to primary care or community settings. Concurrent validity could not be assessed because no comparable EBP mentoring tool exists in China. This study was based on classical test theory, and applying item response theory in future studies could further examine item discrimination and measurement precision across different mentoring levels.⁵² Future study should examine the C-EBPM scale across diverse disciplines (eg, physicians, pharmacists and rehabilitation therapists), particularly as interprofessional EBP implementation becomes more prevalent. Future studies may extend the present findings by applying robust or categorical estimators to examine item thresholds and potential response asymmetries in the C-EBPM scale.

CONCLUSIONS

The C-EBPM fills a notable measurement gap by providing a brief, reliable and culturally adapted instrument to assess EBP mentoring perceived by nurses. Its availability facilitates more nuanced research into mentoring as a modifiable organisational strategy and supports practice efforts to identify and strengthen mentoring resources that underpin sustained EBP. We recommend that future studies validate the C-EBPM in more diverse healthcare settings (including primary care and non-nursing professions), examine its sensitivity to change in mentoring interventions and pair it with outcome measures (eg, EBP implementation scales, patient-level indicators) to test whether increases in perceived mentoring translate into measurable clinical improvements.

Data availability statement

Data are available upon reasonable request.