Knowledge, attitudes, and practices of anesthesiology healthcare professionals regarding crisis resource management in Shanghai, China: a cross-sectional study

Abstract

This study aimed to assess the knowledge, attitudes, and practices (KAP) of anesthesiology healthcare professionals concerning crisis resource management (CRM). A cross-sectional survey was conducted in Shanghai between November 20, 2024, and December 31, 2024. Data were collected using a self-developed questionnaire in Chinese based on published CRM/ACRM evidence summaries and educational resources. A total of 407 valid responses were obtained, yielding a valid response rate of 93.14%. Among the respondents, 281 (69.04%) were affiliated with public tertiary hospitals, 174 (42.75%) were primarily based in the operating room (OR) versus other anesthesiology work settings, and 223 (54.79%) were physicians. The mean self-perceived familiarity (knowledge), attitude, and practice scores were 12.39 ± 3.47 (possible range: 0–20), 30.85 ± 4.74 (possible range: 8–40), and 36.18 ± 6.95 (possible range: 10–50), respectively. In the prespecified path analysis (SEM) using summed KAP scores, knowledge was significantly associated with attitude and practice in the prespecified path model, and attitude was significantly associated with practice; knowledge was also indirectly associated with practice through attitude within the theoretical model (all P < 0.001). Anesthesiology healthcare professionals demonstrated moderate levels of self-perceived knowledge, generally positive attitudes, and frequent CRM-related practices, with significant associations among these components. These findings suggest that CRM-related knowledge, attitudes, and practices are statistically associated within this cross-sectional framework, although causal relationships cannot be inferred.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-57135-8.

Introduction

In modern healthcare systems, the anesthesiology department plays a crucial role in ensuring surgical safety and patient outcomes. Studies have demonstrated that human factors contribute to over 80% of critical incidents in anesthesia¹. Failures in communication are the leading cause of medical errors², and in a cohort of fatal medical accidents, 34 of 40 cases were attributed to non-technical skills rather than technical proficiency³. These findings highlight the importance of effective crisis management strategies in anesthesiology practice, where teamwork is pivotal to averting adverse patient outcomes.

Crisis Resource Management (CRM) originated in the aviation industry when investigations revealed that the majority of accidents stemmed from teamwork failures rather than knowledge or technical skill deficiencies⁴. In healthcare, anesthesiology was the first specialty to implement CRM principles through Anesthesia Crisis Resource Management (ACRM), which focuses on crucial teamwork skills such as leadership, communication, situational awareness, and resource utilization⁴. These principles have proven effective in improving team performance during perioperative emergencies. However, a recent simulation study found that nearly 45% of anesthesiologists acknowledged crises only after patients became hypotensive⁵, indicating persistent gaps in timely crisis recognition and response.

The Knowledge, Attitude, and Practices (KAP) survey methodology serves as a valuable diagnostic research tool to illuminate healthcare professionals’ comprehension, beliefs, and actions regarding specific clinical practices⁶. The KAP framework is particularly relevant to CRM implementation as it provides a structured and descriptive approach to examining how knowledge, attitudes, and practices are theoretically related within a behavioral framework⁷. In classical applications, the KAP model serves primarily as a diagnostic and exploratory tool rather than a validated causal change model. Studies have reported that greater familiarity with crisis management principles is associated with more positive attitudes and more frequent implementation in clinical settings^8,9. Furthermore, within the KAP framework, knowledge, attitudes, and practices are commonly described as interrelated components in behavioral research¹⁰. Although CRM has been widely adopted in Western healthcare systems, questionnaire-based evidence on CRM-related KAP in China remains limited. International studies have shown that structured CRM training improves team performance and communication in anesthesiology and other clinical settings^11,12. However, cultural and organizational factors may influence how CRM principles are perceived and applied in Chinese healthcare environments. In particular, hierarchical norms and communication patterns may influence how CRM-related non-technical skills, such as speaking up and closed-loop communication, are enacted in practice. These contextual characteristics underscore the importance of examining CRM-related KAP in Chinese anesthesiology settings¹³. However, empirical data specifically examining CRM-related knowledge, attitudes, and practices among Chinese anesthesiology healthcare professionals are scarce, leaving an important gap in understanding how CRM principles are perceived and operationalized in this context. Shanghai provides a high-resource urban context to assess CRM-related KAP among anesthesiology healthcare professionals; however, findings from this setting may not fully reflect other regions or hospital tiers. This study aims to conduct a comprehensive assessment of anesthesiology healthcare professionals’ KAP regarding CRM in Shanghai to inform the development of culturally appropriate educational interventions and future implementation research in anesthesiology practice.

Objectives and hypotheses

This study aimed (1) to describe anesthesiology healthcare professionals’ knowledge, attitudes, and practices (KAP) regarding crisis resource management (CRM) in Shanghai, (2) to identify factors associated with CRM-related practice, and (3) to examine a prespecified KAP-based theoretical pathway in which knowledge, attitudes, and practice are statistically associated within an exploratory analytical model.

Methods

Study design and settings

This cross-sectional study was conducted at six hospitals in Shanghai, including Shanghai Ruijin Hospital, Shanghai Ninth People’s Hospital, The First Affiliated Hospital of Naval Medical University, Shanghai General Hospital, Shanghai Sixth People’s Hospital, and Zhongshan Hospital Affiliated to Fudan University between November 20, 2024, and December 31, 2024. These participating sites are major public hospitals in Shanghai, and the majority of respondents in our sample were affiliated with public tertiary hospitals. Public tertiary hospitals typically manage higher-acuity perioperative care and more complex clinical workflows, in which effective teamwork and crisis management behaviours are particularly relevant. We selected this setting to provide an initial assessment of CRM-related KAP in a highly developed clinical context, while acknowledging that findings may differ in other regions and hospital tiers.

Participants

The study design and reporting followed the STROBE guideline. The study population comprised anesthesiology healthcare professionals. Participants were eligible for inclusion if they met the following criteria: (1) currently employed in the anesthesiology department of a hospital in Shanghai, (2) held a valid professional license relevant to their role (e.g., licensed physician or registered nurse), and (3) demonstrated the ability to accurately comprehend the questionnaire content. Exclusion criteria included (1) being on medical leave or working as a temporary employee during the study period and (2) providing incomplete questionnaire responses. Ethical approval for this study was granted by the Clinical Research Management Committee of Ruijin Hospital, Shanghai Jiao Tong University School of Medicine (Ruijin Hospital Ethics Committee: (2024) Proceedings No. (476)). Prior to participation, informed consent was obtained from all respondents.

Instrument / Measures

The questionnaire was developed by the research team in Chinese, drawing on published evidence summaries and educational resources on CRM/ACRM for anesthesiology healthcare professionals^4,12,14, and was reviewed by senior clinicians in anesthesiology to ensure conceptual relevance and clarity. It was not a direct translation of an existing questionnaire from another country; therefore, formal cross-cultural translation procedures were not applicable. The questionnaire was pre-tested in 30 participants, and the Cronbach’s α coefficients for the overall questionnaire and the knowledge, attitude, and practice dimensions were 0.949, 0.9649, 0.898, and 0.931, respectively, indicating high internal consistency. However, internal consistency alone does not establish construct validity. Formal quantitative content validity indices, such as item-level content validity index (I-CVI), scale-level content validity index (S-CVI), or content validity ratio (CVR), were not calculated during this initial development phase. Because this instrument was developed specifically for this study, broader psychometric validation, including assessment of construct validity, criterion validity, and dimensional distinctiveness, has not yet been established and warrants further evaluation in future research.

The finalized questionnaire, administered in Chinese, encompassed four key components: demographic information, knowledge, attitude, and practice dimensions. In addition, participants were asked whether they felt worried about potential anesthesia-related crises or emergency situations. This item was included to capture their perceived psychological preparedness and to explore whether such concerns were associated with attitudes and engagement in CRM-related practices. The knowledge dimension assessed respondents’ self-perceived familiarity with CRM concepts (i.e., a subjective self-assessment rather than an objectively tested measure of factual knowledge or competence) and comprised 10 items; responses of ‘very familiar’ were awarded two points, whereas ‘heard’ or ‘unclear’ received one or zero points, yielding a total possible score ranging from 0 to 20. The attitude dimension included 8 items assessed using a five-point Likert scale, where responses ranged from “strongly disagree” (1 point) to “strongly agree” (5 points), resulting in a score range of 8 to 40. Among the eight attitude items, positive items (P) were scored from “strongly disagree” (1 point) to “strongly agree” (5 points), while the negative item (N) was reverse-coded to ensure consistency in the scoring direction. Similarly, the practice dimension consisted of 10 items evaluated on a five-point Likert scale, with response options from “never” (1 point) to “always” (5 points), producing a score range of 10 to 50. No predefined categorical cutoff points were applied to the knowledge or attitude domains, and these scores were analyzed as continuous variables. Classification thresholds were specified only for the practice domain in regression analyses, as described below.

Data collection and quality control

Anesthesiology healthcare professionals who met the inclusion and exclusion criteria were invited by the study designer to complete a questionnaire online after obtaining their consent. As part of data quality control, cases with clearly unreliable or inconsistent responses were excluded. Completion time was recorded automatically by the online survey platform. The completion-time thresholds were prespecified based on pilot testing and the estimated minimum reading time required to meaningfully review and respond to all questionnaire items, rather than assumptions regarding participants’ clinical experience or expertise. Responses completed in < 60 s were considered implausible and excluded, and responses completed between 60 and 89 s were also excluded because this duration was judged insufficient to allow careful reading of all items during pilot evaluation. Inconsistent responses were defined a priori as logical contradictions between related items (e.g., conflicting answers on rotation history across items) or failing the embedded attention-check item, which instructed respondents to select a specific option. The time-based exclusion criteria were applied uniformly to all participants and were not adjusted according to years of clinical experience, in order to maintain consistent data quality standards across the sample. Specifically, we removed questionnaires with excessively short completion times (8 cases < 60 s and 16 cases < 90 s), one case with an implausible age inconsistent with anesthesiology professional status, four cases with contradictory responses regarding departmental rotation history, and nine cases that failed the embedded attention-check question. Each study subject completed the questionnaire once, with all questions being mandatory to answer. The questionnaire was required to be completed in its entirety before submission, ensuring that every question was answered 100%.

Predefined endpoints. The predefined endpoints were the knowledge, attitude, and practice scores (continuous) derived from the questionnaire. For regression analyses, the continuous practice score was used as the primary outcome to preserve statistical information. No overall composite cutoff was defined for the entire instrument. Additional analyses included categorization of practice score (≥ 35/50) for descriptive purposes and the associations among KAP assessed by correlation analysis and standardized SEM path coefficients.

Prespecified hypothesis. Based on the KAP framework, we specified an exploratory theoretical model in which knowledge, attitudes, and practice were statistically associated, and attitude was examined as a potential mediating variable within this cross-sectional framework.

Sample size estimation

Although this cross-sectional study did not use a formal a priori sample size formula due to the limited prior data on CRM-related knowledge, attitudes, and practices in this context, we considered commonly referenced methodological heuristics to guide our sample size rationale. In survey research with multivariable analysis, it is often suggested that an adequate sample size can be supported by approximately 5–10 respondents per questionnaire item to provide sufficient subject–item ratios for stable estimation in descriptive and multivariable models. In the present study, with 407 valid responses and expressive variation in CRM practice scores, we interpreted our achieved sample size as reasonably aligned with these practical considerations of subject–item ratios and EPV guidance for multivariable analyses, recognizing that precise statistical power calculations for specific hypotheses would require assumptions about effect sizes that were not available prior to data collection¹⁵.

Statistical analysis

Statistical analyses were performed using STATA 17.0 (StataCorp, College Station, TX, USA). Continuous variables were expressed as mean ± standard deviation (SD). Normality of continuous variables was assessed prior to analysis. Between-group comparisons were conducted using independent-samples t-tests or analysis of variance (ANOVA) when normal distribution assumptions were satisfied; otherwise, non-parametric tests (Mann–Whitney U test or Kruskal–Wallis test) were applied as appropriate. Categorical variables were presented as frequencies and percentages [n (%)]. Spearman’s rank correlation analysis was performed to examine the associations among KAP scores, providing insights into the strength and direction of their relationships. For supplementary descriptive reporting, the practice score was categorized as high practice (≥ 35/50, i.e., 70% of the maximum) versus lower practice; however, the primary multivariable analyses were conducted using linear regression with the continuous practice score. The 70% threshold was used as a pragmatic and commonly reported benchmark in KAP studies; however, it does not represent a validated theoretical standard. Because any cutoff may reduce information and introduce misclassification, dichotomized findings were treated as supplementary and interpreted cautiously. Importantly, the practice score was not dichotomized in the SEM; we used the continuous summed practice score in SEM to retain information and to align with the analytic purpose of modeling linear pathways among KAP scores. Variables with a p-value < 0.1 in univariate analysis or deemed clinically relevant were included in the multivariate model. Structural equation modeling (SEM) was conducted as a path analysis based on summed scores to examine the interrelationships among knowledge, attitude, and practice. The model was just-identified, which inherently produces a perfect fit by definition; therefore, global fit indices are not informative for evaluating model adequacy and cannot be used to assess the validity of the specified theoretical structure. All coefficients were reported as standardized estimates. It was hypothesized that knowledge would be theoretically associated with attitude and practice, and that attitude would be associated with practice within the prespecified KAP model. Accordingly, this SEM was specified as a simple path model among the three summed KAP scores with the prespecified direct paths described above, and it is therefore just-identified. Two-sided p-values < 0.05 were considered statistically significant. Because the model is just-identified, global fit indices are not informative and were not used to evaluate model fit. Therefore, interpretation focuses on the standardized path coefficients, which should be interpreted cautiously as exploratory statistical associations within a prespecified theoretical framework. Before multivariate analysis, collinearity among independent variables was examined using the variance inflation factor (VIF), and no significant collinearity was detected (all VIF < 5). Age and work experience were analyzed separately to avoid potential overlap between the two variables. For all statistical analyses, a two-sided p-value < 0.05 was considered statistically significant.

Results

Demographic information on participants

A total of 407 valid responses were included, with an effective rate of 93.14%. Among them, 210 (51.6%) were female, and the average age was 33.54 ± 6.25 years. Most participants were married (68.3%), held a bachelor’s degree (66.6%), and worked in public tertiary hospitals (69.0%). Over half (54.8%) were physicians, and 42.8% worked in the operating room (Table 1).

Table 1.

Baseline Characteristics.

Variables	N (%)
Total score
Sex
Male	197(48.4)
Female	210(51.6)
Age (years old)[range: 22 ~ 55]	33.54 ± 6.25
Marital status
Married	278(68.3)
Unmarried	129(31.70)
Education
Associate degree or below	14(3.44)
Bachelor’s degree	271(66.58)
Master’s degree or above	122(29.98)
Type of institution
Public tertiary hospital	281(69.04)
Public secondary hospital	89(21.87)
Public community health center or private hospital	37(9.09)
Work position in the anesthesia department
Anesthesia induction room	68(16.71)
Operating room	174(42.75)
Anesthesia recovery room/ awakening room/ resuscitation room	146(35.87)
Post-anesthesia care unit (PACU)	19(4.67)
Department worked for more than 3 months (multiple choice)
Operating groom	130(31.94)
Emergency department	223(54.79)
ICU	241(59.21)
Surgery department	222(54.55)
Internal medicine department	119(29.24)
No rotations/work in other departments	16(3.93)
Years of work experience
0–3 years	7(18.92)
3–5 years	155(38.08)
5–10 years	104(25.55)
10 years or more	71(17.44)
Professional title grade
Junior	189(46.44)
Intermediate	178(43.73)
Senior	40(9.83)
Job satisfaction
Very satisfied	73(17.94)
Relatively satisfied	192(47.17)
Neutral	123(30.22)
Slightly dissatisfied	19(4.67)
Employment type
Permanent staff	196(48.16)
Contract-based employment	170(41.77)
Personnel agency	41(10.07)
Frequency of night shifts
0 times	48(11.79)
1–4 times	225(55.28)
5–7 times	108(26.54)
8 times or more	26(6.39)
Type of anesthesia-related training (multiple choice)
Case discussions	195(47.91)
Short lectures	270(66.34)
Seminars	243(59.71)
Emergency operation training	250(61.43)
Simulation training	172(42.26)
Other	17(4.18)
Experience in emergency handling
Yes	378(92.87)
No	29(7.13)
Professional role
Doctor	223(54.79)
Nurse	184(45.21)

Data are presented as mean ± SD or n (%). P-values were calculated using independent t-tests or one-way ANOVA for continuous variables and χ² tests for categorical variables, as appropriate.

The mean knowledge, attitude, and practice scores were 12.39 ± 3.47, 30.85 ± 4.74, and 36.18 ± 6.95, respectively. Significant differences in KAP scores were observed mainly across sex, job satisfaction, work experience, and night shift frequency (all P < 0.01). Demographic and professional variables presented in Table 1 were retained because they were considered potentially relevant to CRM-related KAP and were examined in subsequent univariate and multivariable analyses. These factors, which showed significant between-group differences, were included in subsequent multivariate and structural equation modeling analyses (Table 2). The statistical tests used for these comparisons are specified in the table footnotes.

Table 2.

Endpoints characteristics.

	Knowledge score		Attitude score		Practice score
	Mean ± SD	P	Mean ± SD	P	Mean ± SD	P
Total score	12.39 ± 3.47		30.85 ± 4.74		36.18 ± 6.95
Sex		0.004		0.007		< 0.001
Male	12.97 ± 2.82		31.34 ± 4.45		37.67 ± 5.73
Female	11.83 ± 3.90		30.37 ± 4.96		34.77 ± 7.67
Age (years old)[range: 22 ~ 55]
Marital status		0.005		0.002		0.001
Married	12.70 ± 3.31		31.20 ± 4.88		37.06 ± 6.37
Unmarried	11.70 ± 3.70		30.06 ± 4.34		34.28 ± 7.73
Education		0.597		0.935		0.210
Associate degree or below	13 ± 3.25		31.35 ± 3.54		33.14 ± 9.13
Bachelor’s degree	12.26 ± 3.37		31.00 ± 4.34		36.54 ± 6.76
Master’s degree or above	12.59 ± 3.69		30.42 ± 5.63		35.71 ± 7.01
Type of institution		0.007		0.001		0.429
Public tertiary hospital	12.62 ± 3.67		31.36 ± 4.57		36.10 ± 7.55
Public secondary hospital	11.92 ± 3.17		30.14 ± 4.90		36.44 ± 5.48
Public community health center or private hospital	11.70 ± 2.05		28.56 ± 4.79		36.10 ± 5.17
Work position in the anesthesia department		0.002		0.018		0.005
Anesthesia induction room	12.69 ± 3.59		31.89 ± 4.61		36.75 ± 6.76
Operating room	12.18 ± 3.37		30.41 ± 5.24		35.55 ± 6.98
Anesthesia recovery room/ awakening room/ resuscitation room	12.19 ± 3.59		30.65 ± 4.24		36.07 ± 7.12
Post-anesthesia care unit (PACU)	14.68 ± 1.79		32.47 ± 3.09		40.63 ± 3.81
Department worked for more than 3 months (multiple choice)
Operating groom
Emergency department
ICU
Surgery department
Internal medicine department
No rotations/work in other departments
Years of work experience		0.070		0.011		0.050
0–3 years	12.66 ± 3.46		31.63 ± 3.87		37.35 ± 7.26
3–5 years	12.01 ± 3.07		30.32 ± 4.46		36.67 ± 6.11
5–10 years	12.36 ± 3.60		30.23 ± 5.64		35.39 ± 6.99
10 years or more	12.95 ± 3.99		32.01 ± 4.49		34.97 ± 7.98
Professional title grade		0.252		0.875		0.672
Junior	12.29 ± 3.78		30.92 ± 4.83		35.47 ± 7.73
Intermediate	12.30 ± 3.13		30.80 ± 4.59		36.82 ± 6.10
Senior	13.22 ± 3.26		30.65 ± 5.03		36.62 ± 6.37
Job satisfaction		< 0.001		< 0.001		< 0.001
Very satisfied	14.43 ± 2.69		33.39 ± 3.20		39 ± 4.64
Relatively satisfied	12.24 ± 3.43		30.76 ± 4.91		36.05 ± 6.82
Neutral	12.04 ± 3.07		30.10 ± 4.70		35.82 ± 7.30
Slightly dissatisfied	8.210 ± 4.06		26.63 ± 3.35		28.89 ± 7.59
Employment type		0.094		0.007		0.031
Permanent staff	12.49 ± 3.11		30.19 ± 5.26		35.98 ± 6.13
Contract-based employment	12.07 ± 3.75		31.14 ± 4.20		36.11 ± 7.52
Personnel agency	13.21 ± 3.76		32.68 ± 3.54		37.36 ± 8.07
Frequency of night shifts		< 0.001		0.009		< 0.001
0 times	11.33 ± 4.59		30.81 ± 4.01		30.41 ± 8.57
1–4 times	12.05 ± 3.37		30.48 ± 5.08		36.38 ± 6.43
5–7 times	13 ± 2.70		30.96 ± 4.57		37.40 ± 5.82
8 times or more	14.69 ± 3.49		33.53 ± 2.31		39.92 ± 6.60
Type of anesthesia-related training (multiple choice)
Case discussions
Short lectures
Seminars
Emergency operation training
Simulation training
Other
Experience in emergency handling		0.010		0.008		0.013
Yes	12.49 ± 3.46		31.01 ± 4.71		36.44 ± 6.78
No	11.06 ± 3.28		28.58 ± 4.61		32.68 ± 8.15
Professional role		0.230		0.174		0.010
Doctor	12.65 ± 3.20		31.08 ± 4.91		36.99 ± 6.49
Nurse	12.07 ± 3.75		30.54 ± 4.52		35.19 ± 7.35

Data are presented as mean ± SD or n (%). P-values were calculated using independent t-tests or one-way ANOVA for continuous variables and χ² tests for categorical variables, as appropriate.

Self-perceived familiarity, attitude, and practice

Item-level responses for the knowledge, attitude, and practice dimensions are provided in Supplementary Tables S1, S2, S3. Overall, respondents showed high endorsement of CRM-related attitudes and generally frequent CRM-related practices, while a minority reported time/workload constraints and less frequent participation in CRM training or team discussions.

Correlations between KAP

In the correlation analysis, significant positive correlations were found between knowledge and attitude (r = 0.6035, P < 0.001), knowledge and practice (r = 0.5626, P < 0.001), and attitude and practice (r = 0.5872, P < 0.001), respectively (Table 3).

Table 3.

Correlation analysis.

	Knowledge	Attitude	Practice
Knowledge	1
Attitude	0.6035 (P<0.001)	1
Practice	0.5626 (P<0.001)	0.5872 (P<0.001)	1

P-values were obtained using Spearman’s rank correlation analysis.

Univariate and multivariate linear regression analysis of continuous practice score

Multivariable linear regression analyses were conducted using the continuous practice score as the dependent variable. In the adjusted model, self-perceived knowledge score (β = 0.712, 95% CI: 0.541–0.883, P < 0.001), attitude score (β = 0.423, 95% CI: 0.297–0.55, P < 0.001), and age (β = 0.255, 95% CI: 0.141–0.37, P < 0.001) were positively associated with practice score. Compared with participants with 0–3 years of work experience, those with 5–10 years (β = −3.448, 95% CI: −5.096 to − 1.8, P < 0.001) and ≥ 10 years (β = −5.748, 95% CI: −7.897 to − 3.598, P < 0.001) demonstrated lower practice scores. Higher night shift frequency was also positively associated with practice score (e.g., ≥ 8 times/month: β = 4.307, 95% CI: 1.85–6.765, P = 0.001) (Table 4). The results of the multivariate analysis excluding “Age” are provided in the Table S4.

Table 4.

Univariate and multivariate linear regression analysis of continuous practice score.

	Univariate analysis		Multivariate analysis
	β (95%CI)	P	β (95%CI)	P
Knowledge score	1.217 (1.062,1.373)	< 0.001	0.712 (0.541,0.883)	< 0.001
Attitude score	0.81 (0.681,0.938)	< 0.001	0.423 (0.297,0.55)	< 0.001
Sex
Male
Female	-2.905 (-4.233,-1.577)	< 0.001	-1.061 (-2.037,-0.085)	0.033
Age (years old)[range: 22 ~ 55]	0.192 (0.085,0.299)	< 0.001	0.255 (0.141,0.37)	< 0.001
Marital status
Married
Unmarried	-2.775 (-4.207,-1.343)	< 0.001	-0.974 (-2.189,0.24)	0.116
Education
Associate degree or below
Bachelor’s degree	3.4 (-0.334,7.134)	0.074
Master’s degree or above	2.576 (-1.27,6.422)	0.189
Type of institution
Public tertiary hospital
Public secondary hospital	0.346 (-1.32,2.011)	0.683
Public community health center or private hospital	0.005 (-2.39,2.399)	0.997
Work position in the anesthesia department
Anesthesia induction room
Operating room	-1.195 (-3.133,0.743)	0.226	-0.577 (-1.995,0.841)	0.425
Anesthesia recovery room / awakening room / resuscitation room	-0.675 (-2.663,1.314)	0.505	0.029 (-1.404,1.462)	0.968
Post-anesthesia care unit (PACU)	3.882 (0.367,7.396)	0.03	2.25 (-0.234,4.734)	0.076
Years of work experience
0–3 years
3–5 years	-0.68 (-2.576,1.216)	0.481	-0.824 (-2.228,0.579)	0.249
5–10 years	-1.953 (-4.001,0.096)	0.062	-3.448 (-5.096,-1.8)	< 0.001
10 years or more	-2.379 (-4.616,-0.141)	0.037	-5.748 (-7.897,-3.598)	< 0.001
Professional title level
Junior
Intermediate	1.352 (-0.073,2.778)	0.063
Senior	1.152 (-1.222,3.525)	0.341
Job satisfaction
Very satisfied
Relatively satisfied	-2.948 (-4.755,-1.141)	0.001	-0.199 (-1.541,1.142)	0.77
Neutral	-3.171 (-5.111,-1.23)	0.001	0.08 (-1.388,1.549)	0.915
Slightly dissatisfied	-10.105 (-13.487,-6.723)	< 0.001	-2.214 (-4.892,0.464)	0.105
Employment type
Permanent staff
Contract-based employment	0.122 (-1.312,1.556)	0.867
Personnel agency	1.376 (-0.972,3.724)	0.25
Frequency of night shifts
0 times
1–4 times	5.972 (3.914,8.03)	< 0.001	4.043 (2.426,5.66)	< 0.001
5–7 times	6.991 (4.746,9.235)	< 0.001	4.098 (2.302,5.893)	< 0.001
8 times or more	9.506 (6.356,12.657)	< 0.001	4.307 (1.85,6.765)	0.001
Experience in emergency handling
Yes
No	-3.759 (-6.368,-1.149)	0.005	-1.373 (-3.244,0.497)	0.15
Professional role
Doctor
Nurse	-1.8 (-3.152,-0.448)	0.009	-0.239 (-1.241,0.764)	0.64

P-values were derived from univariate and multivariate logistic regression analyses.

Structural equation model

The effect estimates between KAP were detailed in Table S5. All reported β coefficients are standardized estimates based on the summed KAP scores. Here, the ‘knowledge’ (Ksum) score reflects self-perceived familiarity with CRM rather than objectively assessed factual knowledge or competence. SEM results showed statistically significant associations between knowledge and attitude (β = 0.69, P < 0.001) and between knowledge and practice (β = 0.91, P < 0.001), and attitude was significantly associated with practice (β = 0.43, P < 0.001). Knowledge was also indirectly associated with practice through attitude within the specified model (β = 0.29, P < 0.001). (Table 5 and Fig. 1).

Table 5.

SEM results.

Model paths		Total effect (Direct + Indirect)		Direct Effect		Indirect effect (via Asum)
		β (95% CI)	P	β (95% CI)	P	β (95% CI)	P
Asum <-
	Ksum	0.69(0.58,0.79)	< 0.001	0.69(0.58,0.79)	< 0.001	–	–
Psum <-
	Asum	0.43(0.29,0.56)	< 0.001	0.43(0.29,0.56)	< 0.001	–	–
	Ksum	1.21(1.06,1.37)	< 0.001	0.91(0.74,1.09)	< 0.001	0.29(0.19,0.40)	< 0.001

P-values were calculated based on standardized estimates in structural equation modeling. Total effect equals the sum of direct and indirect effects; minor differences may occur due to rounding. ‘—’ indicates that no indirect effect was specified for that path.

Fig. 1.

Path analysis model (SEM) based on summed KAP scores with standardized coefficients (β). SEM: Structural Equation Model.

Discussion

In this sample, participants reported generally positive attitudes and frequent CRM-related practices, with knowledge (self-perceived familiarity) at a moderate level. Self-perceived familiarity was associated with practice both directly and through attitude within the theoretical model, suggesting that these components are interrelated. Given the significant associations between knowledge, attitude, and practice, the observed relationships provide an empirical basis for further research exploring how CRM-related familiarity and attitudes relate to clinical behaviours in different contexts.

Our findings describe the current profile of CRM-related KAP among anesthesiology healthcare professionals in Shanghai. Prior studies report that anesthesiology professionals often understand CRM concepts, but day-to-day use can be shaped by institutional support, training opportunities, and workload¹⁶. Similar perioperative KAP studies in other clinical settings have also revealed persistent gaps between available evidence and routine practice, despite generally positive attitudes. For example, a national survey of Swedish perioperative healthcare professionals using the KAP model found that many respondents had limited knowledge of opioid-free anesthesia evidence and that practice patterns often remained opioid-centric, with barriers such as conceptual ambiguity and lack of clear guidelines reported by participants¹⁷. These findings illustrate that translating clinical evidence into consistent practice remains a common challenge across different healthcare systems and domains. The observed relationships among knowledge, attitudes, and practices are consistent with the KAP framework, which describes these components as interrelated within a behavioral model^18,19. This pattern has been extensively documented in healthcare research, particularly in studies examining the effectiveness of training interventions aimed at improving emergency response preparedness and team-based communication^20,21. In the Chinese healthcare context, hierarchical team structures and education systems that emphasize technical proficiency over communication and leadership may influence the application of CRM principles. Such cultural and organizational features can discourage open dialogue and reduce the translation of knowledge into practice. In highly hierarchical settings, junior staff may be less likely to challenge decisions or raise concerns, even when clinical deterioration is recognized. This may weaken core CRM behaviours such as closed-loop communication and speaking up during time-critical events. Hierarchical norms may also frame CRM as secondary to technical expertise, reducing its perceived priority for training and debriefing, particularly under high workload. In our item-level findings, time constraints and workload pressure were cited by a subset of respondents, and some reported less frequent participation in CRM training or discussions with colleagues (Tables S2, S3), which may reflect these contextual barriers. Creating psychologically safe teams and supporting communication openness at the departmental and institutional levels may be relevant considerations for strengthening CRM-related practice in Chinese anesthesiology settings, informed by the observed associations and contextual barriers¹³. Therefore, promoting interprofessional communication and integrating CRM-related training into medical education may be considered as potential strategies to address these contextual barriers, although the present cross-sectional findings cannot establish causal effects¹³.

The correlation and SEM analyses provide further insight into the dynamics between knowledge, attitudes, and practices. The positive associations suggest that greater perceived CRM familiarity aligns with more favorable attitudes and more consistent CRM-related practice. These findings align with broader trends in medical education, where improvements in training and exposure to structured crisis management frameworks have been associated with higher levels of adherence to evidence-based protocols^12,22. The mediation effect observed in the SEM model suggests that attitudes play a crucial role in linking knowledge with practice, reinforcing the notion that awareness alone may not be sufficient, and that attitudinal orientation may play an important role in reported practice engagement²³. It should be noted that the knowledge dimension in this study represents self-perceived familiarity with CRM concepts (based on respondents’ self-rated familiarity), rather than objectively assessed factual knowledge or professional competence. As such, this construct may partially overlap conceptually with perceived confidence or attitudinal orientation, which could contribute to stronger observed associations with self-reported practice. Accordingly, this ‘knowledge’ construct should be interpreted as perceived familiarity, and some conceptual proximity to self-reported practice cannot be completely excluded, which may partly account for the magnitude of the standardized association. Therefore, the large standardized path from knowledge to practice should be interpreted as reflecting an association between perceived familiarity and reported practice, rather than a deterministic effect of objective knowledge on behavior. Notably, the standardized knowledge-to-practice path coefficient in our model was relatively large. Because this SEM was a just-identified path analysis based on summed self-reported scores from a self-developed instrument, the magnitude should be interpreted cautiously, and overlap between domains cannot be excluded. In addition, because the model was just-identified, conventional global fit indices could not provide meaningful evidence regarding model adequacy or structural validity. Replication in independent samples and further psychometric evaluation (including assessment of construct distinctiveness) are needed for external validation of this relationship. Therefore, the SEM findings should be viewed as exploratory and hypothesis-generating rather than confirmatory evidence of a robust structural mechanism. This mediating effect has been reported in other areas of clinical training, particularly in studies examining adherence to safety protocols and implementation of structured team communication strategies^24,25.

Multivariate analysis further identified several factors influencing practice engagement. Both self-perceived familiarity and attitude were statistically associated with practice, suggesting that these factors are related to reported CRM-related engagement. Age and work experience also emerged as significant factors. However, the multivariate analysis showed that longer work experience—particularly > 5 or ≥ 10 years—was associated with lower self-reported CRM-related practice scores. This inverse association suggests that accumulated clinical experience may not be linearly reflected in self-reported engagement with structured crisis management strategies. Possible contributing factors may include reliance on routine practices, limited exposure to updated training, or reduced perceived need for formal protocols. Alternatively, this pattern may reflect reporting differences across career stages (e.g., more stringent self-evaluation among senior staff), cohort effects related to differences in formal CRM exposure during training, or role-based variation in how CRM-related activities are documented and perceived. Measurement factors may also contribute, as self-reported practice items may capture “visible” behaviors that differ by seniority rather than overall crisis management competence. These findings suggest that periodic refresher opportunities and tailored professional development across career stages may merit further evaluation, while recognizing that the observed association does not imply behavioral deficits among senior staff.

The impact of work-related factors on practice engagement was also significant. Frequent night shifts were associated with higher CRM practice engagement, which may reflect greater exposure to emergency situations in clinical settings. This finding aligns with research indicating that high-acuity environments often encourage greater reliance on structured decision-making frameworks to manage unpredictable scenarios (reference). However, high workload and fatigue may also affect CRM engagement, as excessive shift hours have been linked to reduced adherence to safety protocols in other clinical settings²⁶.

Responses to the knowledge and practice dimension items revealed specific gaps in understanding and inconsistencies in application. A significant proportion of respondents demonstrated uncertainty or incorrect responses regarding key CRM concepts, such as proactive risk identification and preemptive intervention strategies. These findings reflect broader challenges in medical training, where theoretical instruction alone is often insufficient to ensure long-term retention and application of crisis management skills²⁷. Studies in similar healthcare contexts have shown that simulation-based training and interactive case-based learning are more effective in reinforcing knowledge and improving real-time application of crisis protocols²⁸. In terms of attitudes, while most respondents expressed support for CRM training, barriers such as time constraints and workload pressures were frequently cited, mirroring findings in other studies on healthcare training implementation challenges^29,30. The practice dimension revealed a strong overall engagement, but some professionals reported difficulties in consistently applying CRM strategies due to institutional constraints, underscoring the importance of an organizational commitment to fostering a culture of safety and teamwork.

These findings suggest that structured CRM-related learning opportunities may be relevant in supporting consistent practice engagement. Previous medical education research has shown that periodic reinforcement and simulation-based approaches are associated with improved crisis management performance and team communication^31–33. The observed variation across experience levels may reflect differences in training exposure, role expectations, or professional development pathways. These interpretations warrant further investigation in diverse institutional contexts.

Organizational context may also influence CRM-related practice engagement. Prior hospital management research indicates that institutional support mechanisms, such as structured communication practices and post-event debriefing, are associated with improved safety culture and team performance^34,35. At the same time, workload demands and fatigue may contribute to variability in practice engagement, as suggested in other high-stress medical settings^36,37. These contextual factors should be considered when interpreting the present findings.

This study also has several strengths. First, it provides one of the few empirical assessments of CRM-related knowledge, attitudes, and practices among anesthesiology healthcare professionals in a Chinese metropolitan context, thereby addressing an underexplored area in the literature. Second, the study employed a prespecified analytic framework grounded in the KAP model, integrating correlation analysis, multivariable linear regression, and structural equation modeling to examine interrelationships among domains in a coherent manner. Third, data quality control procedures, including completion-time thresholds and embedded attention checks, were transparently defined a priori, enhancing the credibility of the findings. Together, these methodological features contribute to the robustness of the study and offer a structured baseline for future research in diverse institutional settings.

This study has several limitations. First, although the study included multiple hospitals in Shanghai and a large proportion of public tertiary institutions, the sample was restricted to a single metropolitan area in China. Therefore, the findings may not be generalizable to anesthesiology healthcare professionals in other regions, rural settings, or different hospital tiers, and external validation in broader and more diverse settings is required. Because most participants were recruited from urban public tertiary hospitals, potential selection bias cannot be excluded, and CRM-related KAP patterns may differ in lower-resource or non-urban environments. Second, the cross-sectional design precludes causal inference; therefore, the directionality among knowledge, attitudes, and practices cannot be confirmed, and the observed associations may be influenced by unmeasured factors. Third, all measures were self-reported, which may be subject to recall error and social desirability bias, potentially resulting in overestimation of CRM-related attitudes and practices. In addition, because knowledge, attitudes, and practice were assessed using self-reported responses collected at a single time point within the same questionnaire, common method variance cannot be excluded. Such shared measurement characteristics may have inflated the observed correlations and standardized path coefficients in both the correlation analysis and SEM. Additionally, although internal consistency was satisfactory, broader questionnaire validity has not been fully established. Formal quantitative content validity indices (e.g., item-level content validity index [I-CVI], scale-level content validity index [S-CVI], or content validity ratio [CVR]) were not calculated during the initial instrument development phase, and construct validity was not formally evaluated, which may limit the comprehensiveness of psychometric validation. Furthermore, while high Cronbach’s α values suggest internal consistency, they do not confirm construct validity or dimensional distinctiveness among knowledge, attitude, and practice domains. Given the conceptual proximity inherent in KAP instruments, further psychometric evaluation—such as exploratory or confirmatory factor analysis and tests of construct validity—is warranted in future research. Finally, CRM-related practice was assessed using self-reported questionnaire items rather than objective behavioral or observational measures (e.g., simulation-based assessment or clinical observation), and future studies incorporating such objective measures alongside longitudinal designs are warranted to better evaluate CRM implementation. In addition, dichotomization of the practice score using a 70% threshold may reduce information and introduce misclassification; therefore, greater emphasis is placed on the continuous linear regression results, and any binary categorization should be interpreted cautiously³⁸. In addition, the knowledge score was based on self-perceived familiarity rather than objectively assessed factual knowledge or competence, which distinguishes it from standard KAP surveys and may have inflated associations with other self-reported constructs. For future development, these findings can serve as a baseline description of CRM-related KAP in a metropolitan, predominantly tertiary-hospital context. They may inform the design of context-appropriate CRM education (e.g., focusing on communication openness, teamwork behaviours, and feasible training formats under workload constraints) and provide a reference for benchmarking across hospital tiers. Further multicentre studies in other regions and in non-tertiary hospitals are needed to externally validate the findings and to guide broader implementation strategies. In addition, caution is warranted when generalizing these findings beyond Shanghai, as regional differences in healthcare resources, institutional structures, and training environments may influence CRM-related KAP patterns. Future studies may also consider predefined subgroup analyses and mixed-methods designs, including qualitative approaches such as focus group discussions, to explore the contextual and sociocultural dimensions underlying statistically significant associations.

Anesthesiology healthcare professionals demonstrated moderate self-perceived knowledge, positive attitudes, and relatively frequent CRM-related practices, with significant associations among these components. These findings provide an exploratory basis for CRM-related educational planning, although causal relationships cannot be inferred due to the cross-sectional design.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

CRM

Crisis resource management

KAP

Knowledge, attitudes, and practices

ACRM

Anesthesia crisis resource management

SEM

Structural equation modeling

OR

Operating room

SD

Standard deviation

ANOVA

Analysis of variance

RMSEA

Root mean square error of approximation

IFI

Incremental fit index

TLI

Tucker-Lewis index

CFI

Comparative fit index

CI

Confidence interval

P

Probability value

BA

Bachelor of Arts

PhD

Doctor of Philosophy

Author contributions

Zhang Junyu, carried out the studies, participated in collecting data, and drafted the manuscript.Luo Yan and Zhu Weiyi, grasp the research direction (etc.). Xu Yiqiong, Huang Yanhua and Wang Hongwei performed the statistical analysis and participated in its design. Du Yuanyuan and Zhang Yuhao participated in acquisition, analysis, or interpretation of data and draft the manuscript. All authors read and approved the final manuscript. All authors read and approved the final manuscript.

Funding

2022 High-level Local University Construction Project - Nursing Discipline - High-level Specialized Nursing Base Construction Project. The 2024 Shanghai Jiao Tong University School of Medicine Nursing Research Project (Project No. : Jyhz2403).

Data availability

All data generated or analysed during this study are included in this published article.

Declarations

Competing interests

The authors declare no competing interests.

Ethics approval and consent to participate

All procedures were performed in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki and its later amendments. Ethical approval for this study was granted by the Clinical Research Management Committee of Ruijin Hospital, Shanghai Jiao Tong University School of Medicine [(Ruijin Hospital Ethics Committee: (2024) Proceedings No. (476)); President of the ethics committee: Yufang Bi; Date: 2024/11/14]. Prior to participation, informed consent was obtained from all respondents. The study was carried out in accordance with the applicable guidelines and regulations.