Using a Human Factors Framework to Examine Complexity as a Mediator Between IPC Work System and Healthcare Workers’ Behaviors: A Structural Equation Model

Introduction

Healthcare-associated infections (HCAIs) remain a major threat to patient safety and place substantial burdens on healthcare systems worldwide, contributing to prolonged hospital stay, increased costs, antimicrobial resistance, and avoidable morbidity and mortality.¹ Studies indicate that HCAIs can prolong hospital stays by approximately 13.89 days, raise average medical costs by $2037.46, and increase drug costs by $1612.66 per case.^2,3 Infection prevention and control (IPC) serves as a crucial cornerstone to prevent HCAIs, remaining a core priority for hospitals.^4,5

Although IPC guidelines are well established, consistent implementation in daily practice remains difficult. Healthcare workers perform IPC activities within complex work systems shaped by individual, organizational, tool-related, task-related, and environmental conditions. Barriers such as time pressure, competing clinical demands, inadequate resources, and poorly designed work environments may reduce adherence to recommended IPC practices even when knowledge and attitudes are favorable.^6,7

The Systems Engineering Initiative for Patient Safety (SEIPS) model provides a useful framework for examining how work system elements affect healthcare processes and outcomes.^8,9 In the IPC context, the SEIPS framework helps explain how person factors, organization, tools and technologies, tasks, and internal and external environments may influence healthcare workers’ behaviors.¹⁰ However, prior IPC research has focused more often on isolated determinants of compliance than on the interrelationships among work system elements. In particular, limited empirical attention has been given to the role of complexity in connecting work system conditions with IPC behaviors.^11,12

Complexity is increasingly recognized as a key characteristic of healthcare delivery.¹³ In a series of studies on complex systems commissioned by the World Health Organization (WHO), complexity describes the characteristics of working system such as being multicomponent, nonlinear and context-dependent. Dimensions of system complexity include constituent elements of the system, interactions between these elements, interactions of interventions with context and adaptation, adaptivity, emergent properties, nonlinearity and phase changes, positive and negative feedback loops and multiple outcomes and dependencies.^14–16 Evidence on complexity in healthcare remains predominantly qualitative. Quantitative studies have yielded mixed findings: some suggest that complexity enhances innovation and adaptability in clinical settings,¹⁷ whereas others identify it as a barrier contributing to medication errors, delays, falls, and infection transmission.¹³ In addition, Olsson et al¹⁸ proposed that the interactions among system components may produce unintended consequences, highlighting an alternative perspective beyond the simple facilitator–barrier dichotomy.

In IPC practice, complexity may arise from the interaction of multiple tasks, personnel, environmental demands, workflow interruptions, evolving policies, and the need to respond to rapidly changing clinical situations. Such complexity may shape how healthcare workers interpret, prioritize, and perform IPC activities.¹³ From this perspective, complexity may function as an important pathway through which work system factors affect IPC behaviors.¹⁹ Theoretical frameworks and studies suggest that IPC complexity may act as a mediating factor in the relationship between IPC system and outcome such as behaviors.¹⁹ The introduction of additional elements into working system amplifies its complexity, leading to a higher nursing workload, difficulties in maintaining care continuity, and an increased risk of adverse events such as medication errors, treatment delays, patient falls, and HCAIs.²⁰

This study used the SEIPS model to examine associations among IPC work system factors, IPC complexity, and healthcare workers’ IPC behaviors in a tertiary hospital setting in China. Specifically, we aimed to:

1. Assess the associations of IPC work system factors with IPC behaviors and IPC complexity; and
2. Examine whether IPC complexity mediates the associations between work system factors and IPC behaviors.

By clarifying these pathways, this study may provide practical evidence to support system-based strategies for improving IPC implementation.

Materials and Methods

Procedure and Sample

A questionnaire survey was conducted online in Wuhan, China in June 2021. A structured online questionnaire was distributed to clinical HCWs in 3 branches (Hankou Branch, Optical Valley Branch and Sino-French New City Branch) of Tongji hospital affiliated to Tongji Medical College, Huazhong University of Science and Technology with more than 5000 beds serving patients in central China.

Participants and Data Collection

Clinical HCWs were invited to complete the anonymous questionnaire based on their routine IPC experiences. Eligible participants were frontline HCWs who had experience using personal protective equipment and were willing to participate. HCWs who were not engaged in frontline clinical work, had no experience donning and doffing protective equipment, or declined participation were excluded.

Approximately 2000 HCWs were invited and 906 questionnaires were returned, yielding a response rate of 45.30%. After exclusion of questionnaires with illogical responses or excessively short completion times, 833 questionnaires were included in the final analysis, which provided an adequate sample size for the structural equation modeling (SEM).

Measures

The instrument was developed through a three-phase procedure: item generation, content validation and psychometric evaluation. The detailed development process has been described in the dissertation.²¹ For scales originally developed in English, a back-translation procedure was used for Chinese adaptation. Detailed measurement items were listed in Table A.1.

IPC Work System Factors

IPC work system factors included 6 domains: person, organization, tools and technologies, tasks, internal environment, and external environment. All work system items were rated on a 5-point Likert scale, with higher scores indicating more favorable conditions.

IPC person was assessed using items developed from IPC guidelines, prior literature, and field interview findings, capturing individual-level IPC-related knowledge, attitudes, perceptions, and motivation.²²

IPC organization was measured using adapted items from the Safety Attitudes Questionnaire.²³

IPC tools and technologies were assessed using adapted items from the Comfort Questionnaire for Hand Tools, modified for the IPC context.²⁴

IPC tasks were measured using adapted items from the psychological job demands dimension of the Job Content Questionnaire.²⁵

IPC internal environment and IPC external environment were developed based on the literature and field interviews relevant to the IPC setting.^26–28

IPC Behaviors

IPC behaviors were assessed primarily using self-reported hand hygiene behaviors. Based on the specification of hand hygiene for healthcare workers in China, which included the compliance of World Health Organization’s “5 Moments for Hand Hygiene” and other criterions.²⁹ Items were rated on a 10-point Likert scale, with higher scores indicating better self-reported compliance.

IPC Complexity

To assess IPC complexity, we drew on prior conceptualizations of healthcare system complexity as clarified in the introduction section.¹⁴ The scale was developed to reflect multiple dimensions relevant to IPC practice, including the number and interdependence of system components, variability in work situations, adaptation to changing conditions, and the need to manage unexpected events. Items were developed and refined to fit the IPC context. Higher scores indicated that respondents perceived IPC work as involving more coordination, adaptation, and interacting demands.

Statistical Analysis

Descriptive statistics were used to summarize participant characteristics and study variables. Common method bias was assessed using Harman’s single-factor test. A largest-factor variance below 40% indicates that common method bias is unlikely to be a serious concern. Exploratory factor analysis was initially performed to examine the factor structure of the questionnaire items. Confirmatory factor analysis with maximum likelihood estimation was then used to evaluate the measurement model.

Internal consistency was assessed using Cronbach’s α and composite reliability. Values ≥0.70 were considered acceptable Convergent validity was assessed using average variance extracted (AVE), with AVE ≥0.50 indicating adequate convergent validity. Discriminant validity was also evaluated by comparing the square roots of the AVE values to values in their respective rows and columns. Correlation coefficients were also examined as a preliminary check for potential multicollinearity, with values greater than 0.80 considered indicative.³⁰

SEM was used to examine direct and indirect associations among IPC work system factors, IPC complexity, and IPC behaviors. Model fit was examined using the following indices: relative chi-square (χ2/df), comparative fit index (CFI), Tucker-Lewis index (TLI), incremental fit index (IFI), root mean square error of approximation (RMSEA), and standardized root mean square residual (SRMR). Acceptable thresholds were: CFI/TLI/IFI ≥0.90, RMSEA < 0.08, and SRMR < 0.08.

Mediation effects were tested using bias-corrected bootstrap analyses with 5,000 resamples. Indirect effects were considered statistically significant if the 95% confidence interval did not include zero.

As a sensitivity analysis, the behavioral outcome was replaced with items related to the use of protective equipment while other measurement items were unchanged. All analyses were performed using IBM SPSS Statistics version 24.0 and Amos version 24.0.

Results

Participant Characteristics

Of the 906 returned questionnaires, 833 were included in the final analysis. Most participants were female (90.04%) and nurses (88.12%). The majority were younger than 35 years (68.79%) and held a bachelor’s degree (83.19%). Participants were relatively evenly distributed across internal medicine (39.86%), surgery (38.3%), and other departments (21.85%) (Table 1).

Table 1 Demographic Information of the Participants [N=833]

Reliability, Validity and Correlations

Harman’s single-factor test indicated that no single factor accounted for more than 40% of the total variance, suggesting that common method bias was unlikely to be a serious concern. All constructs showed acceptable to excellent internal consistency, with Cronbach’s α coefficient ranging from 0.804 to 0.962. In the CFA, factor loadings exceeded 0.70. Composite reliability values were all above 0.80, and the minimum AVE was 0.512, supporting acceptable convergent validity. The square roots of the AVE values exceeded all inter-construct correlations, supporting discriminant validity. The final items were attached in the Table A.1.

The measurement model demonstrated acceptable fit: χ²/DF=5.296, IFI =0.910, TLI =0.902, CFI =0.910, and RMSEA =0.072. Mean scores indicated generally favorable perceptions of IPC person factors, organizational conditions, and internal and external environments. IPC complexity was also rated relatively high. Most IPC work system factors and IPC complexity were significantly correlated with IPC behaviors (Table 2). All coefficients were less than 0.8, suggesting that multicollinearity was unlikely.

Table 2 Level and Correlation Analysis of IPC Work System Factors, Complexity and Behaviors

Structural Equation Modeling

In the SEM, IPC person (β=0.110, P=0.003), internal environment (β=0.417, P<0.001), and external environment (β=0.350, P<0.001) were positively associated with IPC complexity.

With respect to IPC behaviors, IPC tasks (β=0.172, P=0.003), internal environment (β=0.221, P<0.001), and IPC complexity (β=0.106, P=0.048) were positively associated with IPC behaviors. (Table 3 and Figure 1).

Table 3 Estimation of the Association of IPC Work System Factors with IPC Complexity and IPC Behaviors Using SEM

Figure 1 SEM with the pathway. Notes: ***p < 0.001, **p < 0.01, *p < 0.05. Structural paths are indicated by thick single-headed arrows, whereas covariance relationships are denoted by thin double-headed arrows.

Mediation Analysis

Bootstrap analyses showed significant indirect associations of IPC person factors, internal environment, and external environment with IPC behaviors through IPC complexity. The 95% confidence intervals for these effects were as follows: IPC person: [0.001, 0.033], IPC internal environment: [0.009, 0.088], IPC external environment: [0.007, 0.075]. The confidence intervals did not include zero, indicating a significant mediation effect. (Table 4).

Table 4 Estimation of Mediation Effect of IPC Complexity Between IPC Work System Factors and IPC Behaviors Using Bootstrap Method

Sensitivity Analysis

In the sensitivity analysis, replacing hand hygiene behaviors with protective equipment yielded similar directions and significance patterns for the main parameter estimates. These findings support the robustness of the model. The analysis result was attached in the Tables A.2 and A.3.

Discussion

Main Findings

This study examined how IPC work system factors and IPC complexity were associated with healthcare workers’ IPC behaviors based on SEIPS model. Three main findings emerged. First, IPC tasks and the internal environment were directly associated with IPC behaviors. Second, IPC person, internal environment, and external environment were positively associated with IPC complexity. Third, IPC complexity showed a significant indirect linking role between work system factors and IPC behaviors. Overall, these findings support the value of a systems perspective for understanding IPC implementation.

Task Characteristics and IPC Behaviors

Task-related factors were directly associated with IPC behaviors in the present study. This finding is consistent with prior IPC research showing that workload, competing demands, time pressure, and interruptions may interfere with adherence to recommended practices.^31,32 When IPC activities are difficult to integrate into routine clinical workflows, healthcare workers may be less able to carry them out consistently. Recent ergonomics-relevant research has similarly shown that task-related factors (workflow disruption, task flow and sequence) are closely related to safety and infection prevention performance in complex care processes.^33,34

Internal Environment as a Key System Factor

The internal environment was directly associated with IPC behaviors and was also indirectly associated through IPC complexity. This finding highlights the importance of the physical and operational care environment in supporting IPC practice. In hospital settings, the location and accessibility of sinks, alcohol-based hand rub, protective equipment, and disposal facilities can strongly shape whether IPC behaviors are carried out efficiently and consistently.³⁵ Ward layout, crowding, visibility of supplies may also facilitate or hinder IPC implementation.³⁶

Interpreting the Role of IPC Complexity

A notable finding was the positive association between IPC complexity and IPC behaviors. This result should be interpreted cautiously. In many contexts, complexity is assumed to impede implementation because it increases workload and uncertainty.¹³ However, in this study, perceived IPC complexity may not have reflected burden alone. Instead, it may also have captured the extent to which IPC work involved active coordination, attention to protocols, responsiveness to changing conditions, and engagement with multiple interacting system components.

Under this interpretation, higher perceived IPC complexity may indicate a more active and structured IPC system rather than a purely more difficult one.¹⁴ For example, healthcare workers who work in settings with stronger IPC oversight, more frequent feedback, and greater interaction around IPC tasks may perceive IPC work as more complex, while also demonstrating better IPC behaviors.

Indirect Pathways Through Complexity

IPC person factors, internal environment, and external environment were indirectly associated with IPC behaviors through IPC complexity. This finding suggests that some work system characteristics may influence behaviors not only directly but also IPC complexity. Individual engagement with IPC, supportive local environments, and broader policy and regulatory contexts may increase IPC complexity (coordination demands, interactions, and adaptation within the IPC system), which may influence behaviors.¹⁹ The results are similar with recent evidence showing that infection prevention practice is shaped by the broader psychosocial and safety context in which staff work,³⁷ while infection prevention and environmental management should be understood as integrated work systems rather than separate domains.³⁸ These results extend the application of the SEIPS framework in the IPC field by suggesting that complexity may be a useful intermediate construct for understanding how system conditions are translated into frontline practice.

Significance-Relevant Findings

The findings suggest that IPC behaviors are associated with a layered work system in which some elements act through direct, indirect or insignificant pathways. IPC task and IPC internal environment appear to be more immediately tied to behavioral execution, whereas IPC person and IPC external environment may influence behavior through their contribution to IPC complexity. The absence of significant effects for IPC organization and IPC tools and technology should not be interpreted as evidence that both factors are unimportant for IPC. Rather, this may suggest that the influence of both factors may operate through work-system components. Recent studies on safety climate and infection prevention suggest that organizational influences may be transmitted through unit-level climate rather than appearing as isolated direct effects.³⁹ Additionally, research suggested that technology can support behavior change, but its effect depends heavily on implementation strategy, staff engagement, and fit with daily work processes.⁴⁰ Another possible explanation relates to the study context. Because the sample was drawn from three branches within the same tertiary hospital system, organizational arrangements, training systems and managerial expectations may have been relatively homogeneous across participants.

Implications

This study has several practical implications for infection prevention programs. First, interventions should address task design. IPC procedures should be feasible within routine clinical workflows, especially in busy settings. Reducing unnecessary complexity in execution, clarifying responsibilities, and ensuring adequate staffing may help improve adherence. Second, hospitals should strengthen the internal work environment by improving the accessibility, visibility, and usability of IPC resources such as hand hygiene facilities and personal protective equipment. Unit layout and workflow should be reviewed from an IPC perspective. Third, infection prevention efforts should move beyond individual education alone and adopt a systems-based approach. Infection preventionists and hospital leaders may benefit from complexity management. Regular audit and feedback, multidisciplinary review of IPC problems, context-specific protocol adjustment, and timely communication of IPC targets may help translate system support into sustained behaviors. Additionally, further investigation of these potential suppression patterns would be worthwhile in future studies.

Limitations

This study has several limitations. First, the cross-sectional design precludes causal inference. Second, IPC behaviors were self-reported, which may have introduced social desirability bias and overestimation of compliance. Third, the study was conducted in branch hospitals within a single tertiary hospital system, which may limit generalizability to other settings, especially primary care institutions or smaller hospitals. Fourth, the measurement of IPC complexity only captured the aspects included in WHO‑commissioned studies. Future studies should incorporate direct behavioral observation, longitudinal designs, multicenter samples and system-related measurements to further test the role of complexity in IPC implementation.

Conclusions

This study explored the influence of IPC system on IPC behaviors, while considering the mediating role of IPC complexity, expanding the explanatory scope of SEIPS model. The finding indicated that IPC tasks are directly associated with IPC behaviors; IPC internal environment is directly associated with IPC behaviors through IPC complexity, and IPC external environment and IPC person are indirectly associated with IPC behaviors through IPC complexity. IPC organization and IPC tools and technology did not show significant result, suggesting that their influence may be conditional, upstream, or embedded within other system components. These findings support a more targeted systems approach to IPC improvement, with particular attention to task feasibility, environmental support, and the management of complexity in frontline practice.