Background

Patient safety is universally recognized as a fundamental component of quality healthcare. Over the past two decades, numerous international organizations have called for the integration of patient safety competencies into the education and training of healthcare professionals [1]. Nurses, who constitute the largest proportion of the global health workforce and provide continuous bedside care, play a pivotal role in safeguarding patient outcomes [2]. Consequently, ensuring that nursing students graduate with a solid foundation in patient safety is a key priority for both educators and health systems [3].

Patient safety competence is a multidimensional construct that encompasses more than just procedural knowledge or adherence to guidelines [4]. It includes a combination of cognitive, behavioral, and affective components such as situational awareness, communication skills, teamwork, ethical judgment, and the ability to recognize and respond to risks and adverse events [5]. According to the World Health Organization’s educational frameworks [6], these competencies should be systematically taught and evaluated throughout the entire nursing curriculum, beginning in the classroom and extending into clinical practice.

However, integrating patient safety into nursing education is not without challenges. One significant difficulty lies in measuring how well students understand and internalize these competencies. Traditional assessment methods, such as knowledge-based tests or procedural checklists, often focus narrowly on factual recall and technical proficiency. While useful for evaluating cognitive and psychomotor domains, these tools lack the sensitivity to assess the complex, affective dimensions of patient safety such as professional judgment, communication in high-pressure environments, or the willingness to speak up about safety concerns. As a result, they may fail to capture students’ attitudes, values, and behaviors that are critical for safe practice [7].

In response to growing international concern about patient safety education, the Health Professional Education in Patient Safety Survey (H-PEPSS) was developed in Canada by Ginsburg et al. [8]. It was designed as a self-report instrument to assess the perceived patient safety competencies of healthcare students, particularly at the point of entry into professional practice. The development took place within the Canadian healthcare education system, which was actively promoting the integration of patient safety into curricula following national reports on medical errors and adverse events. Grounded in a sociocultural view of safety, the H-PEPSS evaluates key non-technical competencies such as teamwork, communication, and systems thinking that are often under-assessed by traditional tools focused on clinical procedures or knowledge recall [8]. The H-PEPSS is based on a theoretically grounded model and evaluates six key domains that focus on working in teams, communicating effectively, managing safety risks, understanding human and environmental factors, recognizing and responding to adverse events, and contributing to a culture of safety [8, 9]. An important strength of the H-PEPSS lies in its dual perspective: it distinguishes between what students believe they learned in the classroom and what they learned during clinical placements. This distinction is essential, as the classroom and clinical environments differ significantly in structure, content delivery, and sociocultural dynamics. In the classroom, students are introduced to patient safety concepts through formal instruction, case-based learning, and structured discussions about system-based practice and safety culture [4]. These controlled settings allow for deliberate teaching of non-technical skills such as communication, risk management, and teamwork within a clearly defined pedagogical framework. In contrast, clinical placements expose students to real-world care environments characterized by time pressure, emotional intensity, and varying degrees of supervision. Learning in clinical settings is often informal and highly dependent on local norms, organizational culture, and the behaviors of preceptors or healthcare teams [10, 11]. As a result, students may experience these two settings differently and develop varying perceptions of their own competencies. Several international studies using the H-PEPSS have reported that students tend to rate their classroom-based learning more positively than their clinical learning, particularly in areas related to safety culture, speaking up, and teamwork [8, 9, 12, 13]. This may reflect challenges such as inconsistent role modeling, hierarchical communication structures, or inadequate integration of patient safety principles in clinical routines [12]. Importantly, the H-PEPSS captures self-perceived competence, which may not directly correspond to actual performance but nonetheless provides valuable insight into students’ confidence and perceived preparedness across different learning environments. Understanding these contextual differences is vital for identifying theory–practice gaps and improving the alignment between academic instruction and clinical training in patient safety [9, 12, 13]. From a conceptual standpoint, the theoretical model underlying the H-PEPSS is consistent with the competency expectations of Slovak nursing education. Although systematic evaluation of patient safety competencies in Slovakia is still developing, the core domains assessed by the H-PEPSS including teamwork, communication, managing safety risks, understanding human and environmental factors, and safety culture are already embedded within national nursing curricula and clinical training requirements. These competencies directly reflect internationally endorsed frameworks, including those outlined in the WHO Patient Safety Curriculum Guide [6], which identifies these sociocultural domains as foundational elements of patient safety education across health professions. Furthermore, Slovak nursing programs are formally aligned with the European Union regulatory framework, including EU Directive 2013/55/EU [14], which mandates that nursing graduates demonstrate competencies in communication, interprofessional collaboration, clinical decision-making, and ensuring safe care environments. As these competencies correspond closely to the six H-PEPSS domains, the instrument’s theoretical foundation is conceptually aligned with both EU educational standards and the expectations placed on Slovak nursing graduates. Therefore, despite potential variation in local implementation, the H-PEPSS model is conceptually appropriate for application within the Slovak educational and clinical context.

Since its development, the H-PEPSS has undergone psychometric testing in multiple countries, including Australia, Belgium, Canada, France, Italy, and Switzerland [8, 11, 15,16,17]. Previous cross-cultural validation studies have consistently supported the H-PEPSS’s theoretical six-factor structure, strong internal consistency across domains, and acceptable model fit, reinforcing its utility as a tool for assessing self-perceived patient safety competencies in diverse educational systems. At the same time, differences in factor loadings, model performance, and inter-factor correlations have been reported, likely reflecting variations in curriculum design, healthcare culture, and clinical training practices [8, 15, 16]. Given these contextual influences, the present study did not assume an exact replication of psychometric properties. Rather, it aimed to evaluate whether the instrument demonstrates acceptable validity and reliability within the Slovak nursing education system, where no prior psychometric evaluation of the H-PEPSS or any similar instrument has been conducted. This represents a significant gap in the regional literature and underscores the need for localized validation to ensure meaningful and accurate assessment of patient safety competencies among nursing students.

Slovak nursing education is formally aligned with international standards, including those of the European Union and the Bologna process. While patient safety topics are included in both theoretical and practical instruction, there is still limited evidence on how well nursing students perceive their preparedness in this area. As a result, it remains difficult to evaluate the effectiveness of current educational strategies in developing patient safety competencies [18]. Without a validated, culturally appropriate tool to measure patient safety competence, it is difficult for educators and policymakers to identify gaps in learning or to evaluate the effectiveness of curricular interventions [19]. Given the linguistic, cultural, and educational specificities of the Slovak context, it is essential to conduct rigorous psychometric testing to ensure that the H-PEPSS functions as a reliable and valid tool for measuring nursing students’ self-perceived patient safety competencies in Slovakia.

The validation of measurement tools is a critical prerequisite for their use in both research and practice. This process ensures that the instrument accurately captures the underlying constructs it is intended to measure, and that the results can be meaningfully interpreted across different groups and settings. Confirmatory factor analysis (CFA) is commonly used to test whether a proposed factor structure fits the data, allowing researchers to assess the adequacy of the theoretical model and evaluate alternative model specifications [20]. In cross-contextual applications, such as comparing classroom versus clinical learning environments, it is essential to test measurement invariance using Multi-Group CFA (MG-CFA). This ensures that the instrument operates equivalently across subgroups, and that observed differences in scores reflect actual differences in perceptions or learning – not measurement bias [21]. The H-PEPSS has proven useful in other countries, and this study aims to extend its utility by adapting and validating it for the Slovak nursing education context. Importantly, it goes beyond previous research by conducting a full test of measurement invariance across classroom and clinical learning environments using multi-group confirmatory factor analysis (MG-CFA), which is a methodological approach that remains rare in the literature and was previously applied only to a limited extent by Ginsburg et al. [8]. In doing so, the study will not only support improvements in local educational quality but also contribute to the global effort to enhance patient safety through evidence-informed teaching and assessment.

Methods

This cross-sectional, validation study aimed to examine factor validity, reliability, and cross-setting measurement equivalence of the Slovak version of the H-PEPSS (H-PEPSSSVK) and to examine nursing students’ perceived patient safety competencies across classroom and clinical learning environments. The study was carried out according to the STROBE checklist [22] and was approved by the Ethics Committee of the Constantine the Philosopher University in Nitra, Slovakia (UKF/917/191013:002).

Participants

All nine public universities in Slovakia offering bachelor’s or master’s degree programs in nursing were invited to take part in the study. Following approval, a convenience sampling approach was used to recruit participants among undergraduate and postgraduate nursing students. The eligibility criteria included the completion of at least one semester of clinical practice and the provision of informed consent. A total of 1734 questionnaires were distributed, for a response rate of 58.7%. Prior to analysis, missing data were handled in accordance with established procedures.

Cases with more than three missing responses (i.e., more than 3 out of 23 core questionnaire items) were excluded from the analyses. Among the remaining cases, the overall proportion of missing data at the item level was low, and no participant had more than 1.3% missing responses. For these cases, missing values were imputed using the median of each item. This approach was selected because the indicators were ordinal and all item medians were identical (Mdn = 4), making median imputation a conservative strategy that preserves the ordinal scale structure and avoids introducing implausible category values. Given the uniform central tendency across items, this procedure minimized distributional distortion while maintaining consistency across indicators. Little’s MCAR test indicated that the missing data were unlikely to be completely at random (p < .001). However, CFA was estimated using the WLSMV estimator, which is robust to non-normality and does not require the MCAR assumption, being appropriate under MAR conditions. Moreover, when the proportion of missing data is small, simple imputation methods have been shown to yield results comparable to more complex approaches in structural models, while tending to attenuate rather than inflate covariances [23].

The final analytic sample consisted of 1,017 participants. Participant recruitment was facilitated by designated contacts at each institution, who also provided accurate enrollment numbers for the winter and summer semesters of 2024; this determined how many questionnaires were to be distributed per university. Students who were absent at the time of data collection, such as due to illness, were not followed up, which may have influenced the overall response rate; however, the potential for bias introduced by their non-participation is likely minimal [24].

Data collection

Data collection took place between February and December 2024 using the Slovak version of the Health Professional Education in Patient Safety Survey (H-PEPSSSVK, see Supplementary File 1 and 2), whose face and content validity were reported elsewhere [18]. A previous pilot study explored initial trends in a smaller sample; those respondents were not included in the present validation study. Additionally, the original version of the survey is available in the validation study and may be used with permission from the Canadian authors [8].

In this study, the survey was administered in a paper-and-pencil format. To ensure accuracy of data entry, two authors (DK, TS) participated in the data-checking procedure. After initial entry by the primary researcher (DK), a random sample of 10% of questionnaires was independently re-entered by another author (TS) and cross-checked item-by-item (including all H-PEPSS items and sociodemographic variables). The H-PEPSS instrument comprises 37 items and is structured into three main sections. The first section assesses knowledge and experience related to clinical aspects of patient safety, including hand hygiene, infection control, and safe medication administration. The second section evaluates six dimensions of patient safety competencies, each measured in both classroom and clinical settings: Working in teams with other healthcare professionals (6 items), Communicating effectively (3 items), Managing safety risks (3 items), Understanding human and environmental factors (3 items), Recognizing and responding to reduce harm (4 items), and Safety culture (4 items). The third section explores students’ comfort with speaking up about patient safety concerns (3 items). Responses were rated on a 5-point Likert scale ranging from 1 (completely disagree) to 5 (completely agree).

In addition, sociodemographic data, including age, previous vocational education, study program, form of study, year of study, supervision during the most recent clinical placement, current clinical placement, previous work experience in healthcare, outcome expectations, and satisfaction with clinical placement (measured on a Likert scale from 0 – not satisfied to 10 – very satisfied), were collected.

Data analysis

Descriptive and inferential analyses were performed using JASP version 0.19.0.0 [25]. Descriptive statistics (means, frequencies, SDs) were used to describe perceptions of patient safety competencies in both the classroom and the clinical setting and the characteristics of the sample. Wilcoxon Signed‑Rank Test was used to test differences in two settings.

To rigorously evaluate the structure of the Health Professional Education in Patient Safety Survey (H-PEPSSSVK), a series of confirmatory factor analyses (CFAs) were conducted. All analyses were conducted in R [26] using the lavaan package [27]. The H-PEPSS focuses on the sociocultural dimensions of patient safety, including safety culture (F1), teamwork (F2), communication (F3), managing risk (F4), understanding human and environmental factors (F5), and responding to adverse events (F6).

Model fit was assessed using the Comparative Fit Index (CFI), the Root Mean Square Error of Approximation (RMSEA), and Standardized Root Mean Square Residual (SRMR). According to established thresholds, CFI values > 0.95 and RMSEA values < 0.06 were interpreted as indicators of good model fit, which is consistent with prior research in medical and nursing education [28]. Because the indicators were ordinal, the CFA models were estimated using the WLSMV estimator (diagonally weighted least squares with robust standard errors and mean–variance adjusted chi‑square statistics) [29]. This estimator is recommended for ordinal data and is implemented in lavaan using a DWLS weight matrix with WLSMV corrections. Due to ongoing debate over the reliability of chi-square statistics in large samples, χ² values are reported and discussed in the online technical appendix only.

For multiple-group CFA (MG-CFA), model fit was evaluated by examining the χ² difference test and changes in CFI. Invariance was supported when the ΔCFI was less than 0.01 and the χ² difference test was non-significant [30]. Discriminant validity was examined using the heterotrait–monotrait ratio of correlations (HTMT), which provides a stringent assessment of construct distinctiveness by comparing heterotrait–heteromethod correlations to monotrait–heteromethod correlations. Consistent with recommendations by Henseler et al. [31], HTMT values below 0.85 were considered indicative of strong discriminant validity. Finally, the internal consistency of the six patient safety competency dimensions (based on the CFA-2 model) was assessed using McDonald’s omega coefficient.

Results

A total of 1,017 nursing students participated in the study. Most respondents were female (95.2%), with only 4.8% identifying as male. The mean age of the participants was 22.88 years (SD = 5.31), ranging from 18 to 59 years. See Table 1 for additional details on the sociodemographic characteristics of the sample.

Table 1 Sample characteristics (N = 1017)
Full size table

Evaluation of competing factor models using CFA

Confirmatory factor analyses (CFA) were conducted to evaluate the structure of the H-PEPSSSVK across classroom and clinical settings (Table 2). Three models were tested in each group. The models represent the current international perspective of patient safety competence measurement by H-PEPSS, where internationally most preferred model represents 6 correlated factors (as indicated in the original study). Model 1 (6 Fr) specified six correlated factors: Working in teams, Communicating effectively, Managing safety risks, Understanding human and environmental factors, Recognizing and responding to reduce harm, and Safety culture). Model 2 (6 Fr_modified) used minor modification of Model 1, including a correlated error path items 10 and 11. Model 3 (2nd order factor) represented a second-order factor model, in which the first-order dimensions were hypothesized to load onto a single overarching latent construct of patient safety competence. All models yielded statistically significant results according to the chi-square (χ²) test (p < .001). As noted in the literature [32, 33], this outcome is not unexpected when evaluating models with many items and degrees of freedom. To address this limitation, alternative fit indices such as the Comparative Fit Index (CFI), Tucker-Lewis Index (TLI), Root Mean Square Error of Approximation (RMSEA), and Standardized Root Mean Square Residual (SRMR) were used to evaluate model adequacy.

Table 2 Results of confirmatory factor analyses
Full size table

Model 1 demonstrated acceptable fit, however not according to all indices (e.g. IFI, CFI, RMSEA at the threshold value). The inspection of modification index values (MI = 170.0 in the classroom settings and 89.4 in the clinical setting) showed strong suggestion to adjustment. The rationale for adding this parameter stems from item 10 and 11 shared content – both address specific aspects of conflict management and adverse events within the “Working in Teams with Other Health Professionals” factor. All other items in this factor address working in teams content in a more general way.

Modified Model 2 show substantial improvement in model fit (all indices reached acceptable levels: CFI ≥ 0.95; χ²/df approached 3). Moreover, in both settings, the modified six-factor correlated model (Model 2) demonstrated the best fit, with RMSEA = 0.058, and SRMR = 0.030, indicating good model performance (Fig. 1).

Fig. 1
Fig. 1
Full size image

CFA-2 model of H-PEPSSSVK

The second-order factor model (Model 3) showed borderline fit in both settings, supporting the hierarchical structure of patient safety competencies.

Factor loadings and inter-factor correlations

The factor loadings for Model 2 across both the classroom and clinical settings strongly support the hypothesized factor structure of the H-PEPSSSVK (Table 3). All items loaded significantly on their respective latent factors, with p-values < 0.001, indicating robust relationships between the observed variables and their underlying constructs. All factor loadings fall within the moderate to strong range (0.68–0.85), suggesting that the items are reliable indicators of their respective dimensions. The consistency of loadings across both settings supports the measurement of the instrument, reinforcing its applicability in diverse educational contexts. The pattern of loadings aligns well with theoretical expectations and supports the construct validity of the H-PEPSSSVK across both the classroom and the clinical environment.

Table 3 Factor loadings of model 2
Full size table

The combined correlation matrix presents inter-factor relationships within the H-PEPSSSVK across classroom and clinical settings as evaluated in Model 2. Overall, the correlations are consistently high, indicating strong conceptual coherence among the six factors (Table 4).

Table 4 Factor correlation matrix
Full size table

In both settings, the strongest correlations were observed between: Understanding Human Factors and Adverse Events (r = .837 classroom; r = .857 clinical) and Managing Safety Risks and Adverse Events (r = .825 classroom; r = .817 clinical). These associations suggest that competencies related to recognizing and responding to adverse events are closely linked with understanding human and environmental contributors to safety risks. The Safety Culture factor shows moderately strong correlations with all other dimensions (ranging from 0.688 to 0.828 in the classroom; 0.686 to 0.857 in the clinic), reinforcing its foundational role in shaping safety-related attitudes and behaviors.

Notably, correlations in the clinical setting tend to be slightly greater, particularly between Working in Teams and Communicating Effectively (r = .851), which may reflect the increased interdependence and communication demands in real-world clinical environments. These patterns support the construct validity of the instrument and suggest that the factor structure is stable and meaningful across both educational contexts. Moreover, HTMT values ranged from 0.68 to 0.84, all below the recommended 0.85 threshold, indicating adequate discriminant validity among the factors (Table 5).

Table 5 Heterotrait–Monotrait ratio of correlations
Full size table

Measurement invariance across classroom and clinical settings

To directly evaluate the measurement invariance of the factor structure across classroom and clinical settings, a multi-group confirmatory factor analysis (MG-CFA) was conducted. The results demonstrated good model fit at all levels of invariance. The configural model showed acceptable fit (χ² = 1331.5, df = 428, RMSEA = 0.058, CFI = 0.951), indicating that the basic factor structure was consistent across settings. Metric invariance was supported by no change in CFI (ΔCFI = 0.000) and RMSEA (ΔRMSEA = -0.001), suggesting equal factor loadings. Scalar invariance was also acceptable (ΔCFI = 0.003), allowing for meaningful comparison of latent means. Finally, strict invariance was confirmed with changes in fit indices (ΔCFI = 0.000, ΔRMSEA = 0.000), indicating equal residual variances. These findings support the stability of the measurement model across both educational contexts, validating within subject comparisons.

Internal consistency of the H-PEPSSSVK

The internal consistency of the H-PEPSSSVK instrument was assessed using McDonald’s omega (ω) across both classroom and clinical settings (Table 6). All six factors demonstrated acceptable to high reliability, with ω values consistently exceeding the recommended threshold of 0.70. The highest reliability was observed for the Working in Teams with Other Healthcare Professionals factor (ω = 0.82), indicating strong coherence among items measuring teamwork competencies. Similarly, Communicating Effectively and Recognizing and Responding to Reduce Harm showed robust reliability (ω = 0.83), reflecting consistent measurement of communication and event response skills. The Safety Culture factor yielded slightly lower but still acceptable reliability (ω = 0.79), which may be attributed to its broader conceptual scope. Notably, total scale reliability was excellent in both settings (ω = 0.95), supporting the overall consistency and structural integrity of the instrument.

Table 6 Internal consistency of the H-PEPSSSVK
Full size table

Comparison of perceptions of patient safety competencies among nursing students

Because several assumptions for using parametric tests (including Tests of Normality, p < 0,001) were not met, a series of Wilcoxon signed‑rank tests was conducted to examine differences in perceived patient safety competencies in the classroom and clinical settings across the six dimensions of the H-PEPSSSVK. Statistically significant differences were found for three of the six factors. In the classroom setting significantly higher scores in Safety Culture (Z = − 2.54, p = .011, d = 0.18), Working in Teams (Z = − 2.97, p = .003, d = 0.21), Communicating Effectively (Z = − 3.07, p = .002, d = 0.22), were found. No significant differences were observed for Recognizing to Reduce Harm, Managing Safety Risks and Understanding Human Factors (all p > .05), indicating comparable perceptions across settings. Overall, although several comparisons reached statistical significance, the magnitude of change across factors was minimal.

Discussion

The present study employed a rigorous cross-sectional design to conduct the first Slovak psychometric evaluation of the Health Professional Education in Patient Safety Survey (H-PEPSSSVK), with the aim of validating its factor structure, reliability, and cross-setting measurement equivalence among nursing students. Our approach was guided by current best practices in psychometric validation, including confirmatory factor analysis (CFA), multi-group confirmatory factor analysis (MG-CFA) for measurement invariance testing, and multiple indices of internal consistency (McDonald’s ω). The robustness of these methodological strategies underpins the validity and reliability of the findings and positions the H-PEPSSSVK as a culturally adapted instrument for assessing patient safety competence in Slovak nursing education.

The H-PEPSS instrument is grounded in a socio-cultural view of patient safety, aligning with global educational frameworks [6] that call for the integration of patient safety into both academic curricula and clinical experiences. Its focus on safety culture, teamwork, communication, risk management, human factors, and error recognition precisely reflects those socio-cultural competencies that health systems increasingly regard as essential for safe practice [8, 19]. Because these competencies are shaped by context, not just cognitive knowledge, the instrument’s dual-context structure (classroom vs. clinical) is theoretically sound; it captures how students perceive their safety competencies in environments that differ in supervision, pressure, and opportunity for application [8, 34, 35]. This theoretical duality, however, also introduces a methodological demand: the instrument must not only exhibit a valid internal structure, but also function equivalently across contexts [36]. In other words, measurement invariance across the classroom and clinical versions is essential so that we can validly compare levels of competence across learning domains [8, 34]. Without invariance, observed differences could be artefacts of context-dependent measurement rather than true differences in latent competence. Considering these requirements, the present study adopted a rigorous and context-sensitive approach for adaptation and validation.

The heart of the methodological contribution lies in the application of CFA to assess the factor structure of the H-PEPSSSVK across classroom and clinical settings. Model 2 (a modified six-correlated factor), which incorporated a theoretically justified correlated error between items related to conflict and debriefing within the teamwork domain, demonstrated the best fit in both contexts. These findings are consistent with prior international validations, e.g. in Canada [8], China [34], the Netherlands [15], Italy [17], and France [16] and underscore the robustness of the six-factor structure. In their original development, Ginsburg et al. [8] reported inter-factor covariances, and subsequent validations in China [34], Persia [36], and Belgium [15] confirmed that the six domains are empirically distinguishable yet statistically interrelated. Numerous studies in nursing education and patient safety research suggest that the core domains of safety competence such as teamwork, communication, risk management, and understanding of human factors are inherently intertwined rather than orthogonal. Mortensen et al. [19] highlighted that existing instruments often include overlapping constructs, reflecting substantive co-variation among domains. Similarly, Bressan et al. [17] advocate for a holistic interpretation of safety competence, emphasizing the interaction between relational, cognitive, and cultural elements. This interconnectedness is further evidenced in empirical studies, e.g., Alidousti-Shahraki et al. [37] reported correlated performance across safety domains in multidisciplinary student samples, and Zhang et al. [38] used a unified, multi-domain safety competence score to predict adverse events. Additionally, the second-order model (Model 3) also showed good fit, suggesting that the six domains reflect a higher-order construct of patient safety competence. While this hierarchical structure may be useful in some applications (e.g., aggregate reporting), the six-factor correlated model remains the optimal structure for detailed evaluation and intervention planning in practice. Moreover, the HTMT results confirmed discriminant validity, further supporting the first‑order factor solution (Model 2).

The application of multi-group confirmatory factor analysis (MG-CFA) to assess measurement invariance across classroom and clinical contexts represents a critical methodological strength of this study. In educational and psychological measurement, testing for invariance is essential to determine whether an instrument operates equivalently across different groups or conditions ensuring that observed score differences reflect true differences in the underlying construct, not distortions caused by measurement bias [30, 39]. Without such testing, comparisons across contexts could be misleading, particularly in cross-cultural or dual-setting assessments where learning environments differ substantially. Our analyses confirmed configural, metric, scalar, and strict invariance, with ΔCFI values consistently below the recommended threshold of 0.01, and minimal changes in RMSEA and SRMR across all levels – criteria widely accepted as evidence of measurement stability [40, 41]. This finding is both methodologically robust and educationally meaningful. Although the study employed convenience sampling, the large sample size (n > 1000) and the inclusion of participants from most regions of the country enhance the diversity of the data. These features strengthen the representativeness of the sample and support the generalizability of the findings, while still warranting caution in interpreting them beyond the studied population. First, the Slovak version of the H-PEPSS (H-PEPSSSVK) retains the same factorial structure and scale properties when it is administered in both classroom and clinical learning environments. In practical terms, it validates the fairness, reliability, and interpretability of the instrument when used to evaluate perceived patient safety competencies across diverse educational settings. Second, this level of invariance has rarely been reported in previous H-PEPSS validation studies, which often confirm internal consistency and construct validity but do not extend to the full MG-CFA [8, 15]. By confirming strict invariance, this study contributes to a small but growing body of literature emphasizing the importance of testing latent measurement equivalence in educational research [42]. In the Slovak context, theoretical instruction and clinical placements represent distinct learning environments in nursing education, each with unique expectations, interactions, and assessment contexts. Confirming full measurement invariance across these two core educational conditions strengthens confidence that the Slovak version of the H-PEPSS operates equivalently in both settings. Although institutional contexts may vary in terms of infrastructure, supervision, or safety culture, our study focused specifically on instructional context (classroom vs. clinical) rather than broader organizational-level differences. These were not systematically measured, and the interpretation of results is thus limited to the educational conditions assessed. Additionally, these findings suggest that the H-PEPSSSVK can be used to compare students’ perceptions of patient safety competence across different learning settings, supporting both curriculum evaluation and evidence-informed improvements in nursing education. In short, the MG-CFA findings not only strengthen the psychometric credibility of the instrument but also enhance its utility as a diagnostic and developmental tool in national nursing education reform.

Although previous Slovak research by Kohanová and Sollárová [18] examined patient safety competencies and identified age as a potential influencing factor, it did not test whether the H-PEPSS functioned equivalently across groups. Without measurement invariance, comparisons across variables like age may reflect bias rather than true differences. The present study addresses this gap by confirming strict invariance across classroom and clinical contexts, ensuring valid comparisons. This strengthens the instrument’s structural integrity and provides a foundation for future research to explore sociodemographic predictors more reliably. Furthermore, strong factor loadings across all six domains (ranging from 0.686 to 0.852) provided compelling evidence for the construct validity of the H-PEPSSSVK. All loadings were statistically significant at p < .001, and their magnitudes met established psychometric thresholds for well-specified latent constructs [43, 44]. Loadings above 0.50 are generally considered acceptable, whereas those above 0.70 indicate strong item-construct relationships [45]. Importantly, the factor loadings were remarkably consistent across both classroom and clinical settings, highlighting the structural stability of the instrument across learning environments. This consistency affirms the theoretical expectation that patient safety competencies, particularly those related to teamwork and communication, should manifest robustly in both academic and applied contexts. For example, strong loadings for teamwork and communication in the clinical version reflect the heightened salience of these domains in real-world care delivery, where interprofessional interaction and collaborative decision-making are essential for safe practice [11, 13]. These results contribute to growing evidence that patient safety competencies, although abstractly defined, are empirically measurable and contextually grounded, supporting both theoretical coherence and practical relevance.

The observed pattern of inter-factor correlations (r = .686–0.857) further reinforces the construct validity of the instrument. These correlations are not only statistically strong but also theoretically coherent with the sociocultural foundations of the H-PEPSS. The HTMT results supports the discriminant validity of the H-PEPSS. In particular, the highest correlations – between Understanding Human Factors and Recognize, Respond to and Disclose Adverse Events and Close Calls, and between Managing Safety Risks and Recognize, Respond to and Disclose Adverse Events and Close Calls – reflect the natural conceptual overlap between situational awareness, human limitations, and the ability to recognize and respond to harm [8, 12]. These synergies mirror findings from international studies that also report strong associations between these domains [11, 15]. The slightly stronger correlations observed in the clinical setting, especially between teamwork and communication, likely reflect the contextual demands of applied healthcare, where high-stakes, time-sensitive interactions make interdependence between competencies more pronounced. This result is theoretically supported by situated learning theory [46], which posits that knowledge and skill development are shaped by authentic social participation in real-world environments. In this context, the interrelatedness of safety competencies in clinical settings is not a psychometric artifact, but rather a reflection of how safety practices are taught, learned, and enacted in complex healthcare systems.

The reliability estimates for the H-PEPSSSVK, which were calculated using McDonald’s ω coefficient, exceeded the widely accepted threshold of 0.70 for all six competency domains, with total scale reliability surpassing 0.94 in both the classroom settings and the clinical settings. These results confirm the internal consistency of the instrument and affirm its psychometric suitability for use in high-stakes educational research, curricular evaluation, and competency-based assessment. The consistency of reliability results is noteworthy, as unbalanced item loading can distort construct representations and weaken interpretability in applied psychometric tools [47, 48]. Furthermore, such high reliability across both settings enhances confidence in the stability of measurement across varied learning environments, thereby supporting the use of the H-PEPSSSVK for longitudinal tracking, program benchmarking, and outcome-based education.

One of the study’s most significant contributions lies in its ability to differentiate students’ perceived patient safety competencies across classroom and clinical learning environments. The differences in three of the six domains, with consistently higher scores reported in the classroom context suggest a persistent disjunction between theoretical instruction and its practical application, which aligns with extensive international evidence on the theory–practice gap in nursing education [7, 10, 49]. Such discrepancies may reflect structural challenges within clinical placements, including limited supervision, variable quality of preceptorship, and inconsistent reinforcement of safety principles in real-world settings [50]. These findings underscore the urgent need for integrated educational strategies that facilitate the transfer of classroom-based knowledge into clinical competence such as simulation-based learning, structured mentorship, and guided reflective practice [51, 52]. However, the magnitude of the differences was rather small according to the Cohen d statistic.

Interestingly, the domains of Managing Safety Risks and Understanding Human and Environmental Factors did not exhibit statistically significant differences across settings, suggesting that these competencies may be either more robust to contextual variation or more effectively reinforced across both instructional formats. It is possible that these domains are more cognitively anchored and grounded in conceptual frameworks that students can generalize across situations, or that they benefit from repeated exposure in both classroom discussions and clinical observations. Alternatively, these results may point to curricular strengths in how these domains are taught, or to students’ increasing recognition of human limitations and systemic risk factors regardless of setting. Future qualitative research could provide deeper insight into how students internalize these specific competencies and how educational design influences their perceived preparedness. Understanding these mechanisms is essential for tailoring interventions that close the theory–practice gap and promote meaningful, transferable learning in patient safety education.

Study limitations

This study has several limitations. First, its cross-sectional design does not capture changes in patient safety competence over time and therefore limits causal interpretation. Second, although the sample included students from all Slovak public nursing faculties, participation was voluntary and based on convenience sampling. While this approach enabled broad institutional coverage and a large sample size, it may limit the representativeness of the sample and the generalizability of the findings within the Slovak context. Constructing a proportionally representative sample or excluding participants to achieve balance could have introduced additional bias and reduced statistical power; therefore, all eligible responses were retained, and the sample structure is reported transparently. Third, the exclusive reliance on self-reported data introduces potential measurement bias, as perceived patient safety competence may not fully correspond to actual clinical performance. Although the H-PEPSS is a well-established instrument for assessing perceived competence, future research would benefit from incorporating complementary methods such as simulation-based assessments, objective structured clinical examinations, or supervisor evaluations to strengthen validity. Another limitation concerns our handling of missing data. While we excluded cases with extensive missingness, we used median imputation for the small number of remaining missing values. Although this approach is pragmatic and commonly used when missingness is minimal, it may reduce variance and underestimate standard errors. Given the uniform median value across all H-PEPSS items and the ordinal scale structure, the risk of distortion was likely low. Nonetheless, in future studies with higher missingness or longitudinal designs, the use of multiple imputation for ordinal data would be a more statistically robust alternative. Finally, the study was conducted solely in Slovakia, and differences in educational systems, clinical training structures, and patient safety cultures should be considered when applying these findings to other cultural or educational contexts.

Conclusion

This study provides important evidence for the factorial validity, internal consistency, and cross-setting measurement invariance of the H-PEPSSSVK instrument among Slovak nursing students. Using advanced psychometric methods, we confirmed that the instrument performs reliably across theoretical and clinical educational environments. These findings support its use in assessing perceived patient safety competencies and guiding curriculum development within Slovak nursing education. While the validation focused on selected aspects of construct validity and did not include all types of validity evidence, the results offer a solid foundation for further research. The study also highlights the potential for culturally adapted instruments to retain structural integrity and fairness when carefully localized. Future research should explore longitudinal changes in competence, incorporate objective performance measures, and include broader stakeholder perspectives. In conclusion, the H-PEPSSSVK represents a psychometrically sound and context-sensitive tool that can meaningfully contribute to the development of patient safety education and support safer nursing practice in the Slovak context.