Addressing turnover intentions among medical radiation practitioners: the influence of wellbeing-focused HRM and individual resilience

This study examines healthcare management in Australia and the turnover intentions of medical radiation practitioners (MRPs). In recent years, concern surrounding the sustainability and effectiveness of Australia’s healthcare system has intensified, driven by increasing patient demand, rising rates of chronic illness, escalating healthcare costs and persistent workforce shortages (Kumah, 2025). Within the medical radiation workforce specifically, pressures are heightened by documented labour shortages and growing reports of staff burnout (Shields et al., 2021). As of March 2026, the Medical Radiation Practice Board of Australia (MRPBA, 2026) reported a total of 21,335 registered medical radiation practitioners nationwide (MRPBA, 2026) to service a population of 27.5 million (ABS, 2025). Recent data from the MRPBA have demonstrated that across nine professions including medical radiation practice, practitioners cite mental burnout, retirement, lack of recognition/feeling undervalued, lack of professional satisfaction and work no longer being fulfilling as key factors influencing their intention to resign (Tan et al., 2025).

MRPs are essential to contemporary healthcare, providing critical diagnostic and therapeutic services across medical imaging, nuclear medicine and radiation therapy (Chau et al., 2022). Their expertise underpins the detection, diagnosis and treatment of diverse conditions, making their retention vital to the functioning and quality of Australia’s healthcare system. Diagnostic radiographers produce high-quality images using ionising and non-ionising radiation across modalities such as X-ray, CT, MRI and ultrasound, supporting accurate clinical decision-making (NSW Health, 2023; ASMIRT, 2025). Nuclear medicine technologists prepare and administer radiopharmaceuticals for diagnostic and therapeutic purposes (ANZSNM, 2025). Radiation therapists plan and deliver cancer treatments, manage resources, adopt emerging technologies and contribute to multidisciplinary oncology teams to optimise patient outcomes (RANZCR, 2012). Together, these professionals ensure patients receive timely, effective imaging and treatment, with results communicated to guide subsequent medical care.

MRPs face numerous challenges in their daily work, including long working hours, escalating patient demand and persistent workforce shortages (Ku et al., 2020). Within this context, practitioners face significant surface acting demands associated with sustained patient-facing roles. For example, in a radiation therapy department, delays in appointments or unexpected machine breakdowns can leave practitioners feeling stressed and frustrated. Despite this, they must present themselves to patients as calm and composed professionals. Similarly, in a medical imaging department, patients of all ages may arrive in trauma or experiencing significant pain when they require an X-ray or CT scan. Practitioners in these situations must demonstrate empathy and remain composed, whilst supressing their own anxiety and urgency to complete the scans quickly (Ku et al., 2022; Yikilmaz et al., 2024). Emotional labour refers to the paid management of emotions in which workers are required to display organisation or role appropriate feelings while suppressing others, particularly in occupations involving frequent face-to-face interactions such as healthcare, education and service roles (Hochschild, 1983; Yeh et al., 2020). In this paper, we specifically examine surface acting, which is an emotional labour strategy in which individuals regulate their outward emotional expressions without altering their internal feelings. This is typically done by suppressing or masking negative emotions and displaying positive emotions that are not genuinely experienced (Hochschild, 1983). These challenges not only affect practitioners themselves but also have significant implications for patient wellbeing and clinical outcomes (Ku et al., 2022). Under such demanding conditions, practitioners often experience reduced resilience, which can undermine their capacity to effectively manage complex clinical situations and consistently provide high-quality patient care (Robertson et al., 2016). Resilience is defined by Masten (2021, p. 314) as “a dynamic system’s capacity to adapt successfully to disturbances that threaten system function, viability, or development.” Therefore, this is a critical attribute for wellbeing within demanding healthcare environments, sustaining quality of patient care and continuance of employment.

Using data collected from 165 MRPs in Australia, we investigate the association between wellbeing human resource management (WBHRM) practices and the turnover intentions of MRPs through the effects of surface acting, leadership support, resilience and burnout. Our study is underpinned by conservation of resources (COR) theory (Hobfoll, 1989, 2001). Wellbeing human resource management refers to a system of HRM policies and practices explicitly designed to protect, support and enhance employees' physical, psychological and social wellbeing while enabling sustainable performance for both employees and the organisation (Guest, 2017). By deepening our understanding of how WBHRM can address the surface acting demands and professional needs of MRPs, healthcare organisations will be better positioned to strengthen workforce sustainability and enhance the long-term viability of radiation services within the healthcare system (Ku et al., 2022).

Research on MRPs remains limited regarding the relationships between WBHRM, resilience, supervisor support and turnover intentions. This study makes five contributions to HRM literature. First, we extend COR theory to an under-researched cohort of allied health professionals by proposing a dual-pathway model through which WBHRM is associated with lower turnover intentions via both direct burnout reduction and indirect surface acting reduction. Second, we make an important contextual contribution by generating rare empirical evidence from one of the more data-scarce cohorts in allied health research, establishing that resource depletion dynamics identified in nursing populations extend to MRPs with meaningfully different role structures and regulatory frameworks. Third, we refine HRM theory by demonstrating that resilience functions as a first-stage moderator, clarifying where personal resources operate within causal chains rather than simply buffering final outcomes. Fourth, we illuminate resource synergies through conditional moderation, showing that personal and organisational resources realise their full potential only within supportive leadership contexts, advancing multilevel models of resource interactions. Fifth, we advance understanding of surface acting as an organisationally shaped phenomenon whose antecedents, and not just its consequences, are amenable to intervention through WBHRM, opening new directions for HRM research on emotional labour management in high-demand healthcare environments. Together these contributions inform HRM practice by highlighting how WBHRM and supportive leadership are associated with lower surface acting demands, reduced burnout and stronger retention.

Healthcare management faces escalating pressures driven by rising patient demand, workforce shortages, increasing service costs and organisational inefficiencies (World Health Organization, 2022). Ineffective leadership compound delays in care and heightens patient distress, underscoring the need for strategic workforce planning and targeted retention initiatives (Ashghali Farahani et al., 2024). Yet recognition and support for the emotional labour of MRPs remain limited, along with access to mental health resources (Ku et al., 2022).

These systemic pressures permeate medical radiation practice, mirroring broader healthcare workforce challenges. In Australia, health and social assistance is among the fastest-growing labour sectors, driven by an ageing population and rising demand for complex care (National Skills Commission, 2022; Australian Government, 2023). Addressing this demand requires a flexible, resilient workforce supported by high-quality, responsive education and training systems and government investment in workforce attraction, retention, expanded training and innovative care models (Australian Government, 2023, pp. 99, 108).

The COVID-19 pandemic intensified workforce retention challenges, triggering widespread burnout, psychological distress, deteriorating work conditions and elevated turnover intentions among healthcare professionals (Ashghali Farahani et al., 2024; Ghasemi Kooktapeh et al., 2023). Systematic reviews identify excessive workloads, inadequate institutional support, pay inequity, unsafe environments and limited career-development opportunities as primary drivers of attrition across healthcare, including medical radiation practice (Ashghali Farahani et al., 2024; Australian Government, 2023). MRPs face high occupational stress, driven by workforce shortages, with only 21,335 practitioners serving Australia’s 27.5 million population (ABS, 2025; Ku, 2021).

WBHRM prioritises employees’ mental, physical and social wellbeing to enhance individual and organisational performance (Bartram et al., 2024). It fosters satisfaction through supportive work environments, work–life balance and programs that build resilience and reduce stress (Pariona-Cabrera et al., 2024; Guest, 2017). WBHRM practices, including flexible work, health programs and promotion opportunities, improve mental health, productivity and retention (Guest, 2017). Five HR practice areas influence emotional labour, burnout and patient outcomes: investment in employees, engaging work, positive social and physical environments, employee voice and organisational support (Guest, 2017). Long-term commitments, such as training, career development, job security and fair rewards, signal organisational care (Becker, 1964). Engaging work emphasises autonomy, skill variety, task significance and learning opportunities (Deci and Ryan, 2000). Managing physical and psychological risks ensures sustainable wellbeing (Demerouti et al., 2001), while employee voice enables safe participation in decisions (Edmondson, 1999). Practices promoting work–life balance help employees manage competing roles, preventing stress and supporting recovery (Hobfoll, 1989).

It is important to distinguish WBHRM from adjacent constructs, particularly high-performance work systems (HPWS). While both involve bundles of HRM practices, HPWS is primarily oriented towards maximising employee performance and organisational productivity, with wellbeing treated as a by-product of high engagement rather than an explicit objective (Bartram et al., 2012; Cooke et al., 2019). By contrast, WBHRM explicitly positions employee wellbeing as a primary organisational goal and a critical means by which employees improve their in-role job performance (Guest, 2017). This distinction is consequential: research has shown that HPWS is associated with increased burnout and job stress in healthcare contexts (Bartram et al., 2012), whereas WBHRM is theorised to reduce these outcomes by actively protecting and replenishing employee resources (Guest, 2017). The current study focuses on WBHRM rather than HPWS given the research question concerns resource protection and wellbeing sustainability, not performance maximisation.

This study applies COR theory (Hobfoll, 1989, 2001) to explore how WBHRM practices reduce turnover intentions in healthcare. COR theory suggests individuals seek to acquire, maintain and protect valued resources, such as skills, energy and supportive conditions and experience stress when resources are threatened or lost. Healthcare work, with its emotional labour, high workload and patient care demands, creates chronic resource strain (Halbesleben et al., 2014).

WBHRM practices constitute precisely such a resource caravan: a coordinated bundle of organisational investments, investment in employees, engaging work, positive and safe environments, employee voice and organisational support (Guest, 2017), that collectively replenishes depleted psychological and physical resources, reduces demand-driven resource loss and enables recovery. Their effectiveness depends on two interrelated dynamics: the resource pathways through which WBHRM operates; and the conditions enabling resource synergies. This study examines both through an integrated model of serial mediation and conditional moderated mediation.

A recognised epistemological challenge in COR research concerns the potential for circular reasoning, whereby resources are defined by their capacity to reduce stress, which then explains stress reduction (Halbesleben et al., 2014). This study addresses this challenge by specifying resources a priori from established theoretical criteria rather than deriving them inductively from outcomes. WBHRM practices constitute organisational resources by virtue of their explicit design to protect and enhance employee capacity (Guest, 2017); resilience constitutes a personal resource through its role in enabling adaptive coping under adversity (Masten, 2021); and leadership support constitutes a social resource through its function in creating conditions of psychological safety and resource availability (Edmondson and Lei, 2014). This a priori specification is consistent with Halbesleben et al.'s (2014) methodological recommendations for theoretically grounded COR applications. A further consideration concerns the distinction between perceived and implemented HRM practices. Nishii et al. (2008) demonstrate that employees' subjective attributions of the rationale behind HR practices exert independent effects on attitudes and behaviours beyond the practices as designed or implemented. Measuring perceived WBHRM therefore captures the psychological mechanism through which HRM influences employee outcomes, the experienced resource environment rather than the intended one, and is the theoretically appropriate operationalisation for a study concerned with resource protection processes at the individual level. COR theory is selected as the primary theoretical framework over alternatives such as Job Demands-Resources theory (JD-R) (Bakker and Demerouti, 2017) because the specific predictions of this model require theoretical mechanisms that are central to COR but peripheral to JD-R. COR and JD-R are theoretically complementary rather than competing (Hobfoll et al., 2018; Bakker and Demerouti, 2017), and JD-R has strong empirical support in healthcare HRM research (Pirrotta et al., 2025; Broetje et al., 2020). However, three COR principles are theoretically necessary for this model that JD-R does not provide. First, resource loss primacy explains why the chronically under-resourced MRP environment is disproportionately harmful, a prediction grounded in COR's asymmetry principle rather than JD-R's balance framework. Second, the resource caravan principle explains why WBHRM operates as an integrated bundle whose dimensions mutually sustain one another rather than as isolated demand-reducing practices. Third, and most critically for H4, COR's resource interaction principle generates the prediction that personal, social and organisational resources amplify one another multiplicatively in enabling contexts, producing synergistic effects that exceed additive contributions. JD-R's conceptual architecture, centred on demand-resource balance and the distinction between energetic and motivational processes, does not generate these multiplicative interaction predictions. Future research employing both frameworks simultaneously would provide richer insight into the complementary pathways through which organisational and personal resources shape MRP wellbeing and retention.

Burnout defined as a state of exhaustion, cynicism and reduced professional efficacy acts as a key mechanism linking workplace conditions to employee outcomes (Maslach and Leiter, 2016). Within the JD-R framework, excessive demands deplete employees’ physical and psychological resources, triggering burnout that undermines engagement, performance and retention (Bakker and Demerouti, 2017). Strong evidence across healthcare confirms burnout as a robust predictor of turnover intention. Meta-analytic findings show consistent, moderate-to-strong associations between burnout and intentions to leave among nurses and physicians, with burned-out physicians over three times more likely to consider turnover (Özkan, 2022). Recent studies further demonstrate that burnout mediates the relationship between workplace stressors and turnover intention (Li et al., 2025). Importantly, well-being-oriented HRM practice, such as flexible work, wellness initiatives, mentoring and recognition assist in restoring depleted resources, reduce burnout and strengthen resilience and retention (Galanakis and Tsitouri, 2022). Thus, burnout represents a central pathway through which workplace factors shape both individual wellbeing and organisational sustainability.

Surface acting is a prevalent and resource-depleting feature of healthcare work involving sustained face-to-face interactions (Grandey, 2003; Brotheridge and Grandey, 2002). Among emotion regulation strategies, surface acting, suppressing or faking emotions without internal change, is particularly resource-depleting (from a COR perspective) and strongly associated with higher levels of burnout, mediating the relationship between emotional demands and exhaustion (Popucza et al., 2025). Seminal work by Grandey (2000) established surface acting as a distinct form of emotional regulation in the workplace, while Brotheridge and Grandey (2002) provided foundational empirical evidence linking surface acting specifically to the depersonalisation and exhaustion dimensions of burnout among service workers. Grandey (2003) further confirmed that surface acting predicts emotional exhaustion beyond the effects of work demands alone. MRPs exemplify this dynamic, as they routinely care for patients experiencing significant distress while maintaining calm, empathic professionalism in high-stakes clinical environments. Regulatory standards require practitioners to sustain sound mental and physical health, further intensifying these demands (MRPBA, 2025, 2026). The practitioner’s emotional presentation shapes patient trust, confidence and engagement, especially during stressful treatments, positioning emotional labour as a core component of care delivery (Feng et al., 2024). However, prolonged emotional regulation is associated with psychological distress, reduced performance, increased error risk and burnout, which may ultimately contribute to workforce attrition (Kirby and Lück, 2014, Andersen et al., 2025a). These risks highlight the need for organisational strategies that support emotional regulation and practitioner resilience.

Surface acting depletes emotional resources and triggers a sequential pathway to burnout, progressing from emotional exhaustion to reduced job satisfaction, commitment and turnover intention (Grandey, 2003). This cumulative depletion reflects cascading resource loss rather than a single-step effect, consistent with the resource caravan perspective (Hobfoll, 2011). Wellbeing-oriented HRM can interrupt this process by strengthening multiple resources simultaneously through manageable workloads, social support and climate that enable authentic emotional expression thereby reducing burnout risk and sustaining employee wellbeing and performance.

Resilience refers to the capacity to convert adversity into opportunities for personal and professional growth (Luthans, 2002). Resilience involves effectively managing stress and adapting to challenges by drawing on both internal strengths and external support systems. Within healthcare, resilience is understood as a practitioners’ ability to navigate high-pressure clinical environments in ways that enable them not merely to cope, but to thrive (Robertson et al., 2016).

For MRPs, building resilience is critical to managing occupational stress and preventing burnout, especially amid workforce shortages, restricted training and professional development opportunities, and systemic barriers to wellbeing (Chau et al., 2025). Regular exposure to patient trauma and distress heightens practitioners’ risk of compassion fatigue and emotional exhaustion (Ku et al., 2022). Building resilience is therefore essential for managing demanding conditions such as workload pressure and staff shortages, while supporting adaptability, confidence and wellbeing (Hobfoll, 2011). Structured resilience training particularly for new graduates combined with ongoing professional development can equip practitioners with practical coping strategies, including stress regulation skills and access to supportive professional networks (Ago et al., 2025). Strong social support is especially critical, as evidence consistently shows that interpersonal relationships underpin emotional stability, professional growth and sustained resilience (Drageset, 2021).

COR theory posits that clinicians must invest resources such as skills, time, emotional energy and social support to prevent loss and build resilience (Hobfoll, 1989). Because resource loss is more psychologically harmful than gain, clinicians in high-demand healthcare settings are especially vulnerable to stress and burnout when organisational supports constrain resource investment. Supports such as training, mentorship, protected time and strong teams enable resource accumulation, strengthening coping capacity, confidence and professional networks and are therefore central to workforce wellbeing and care quality.

Resilience functions as a personal resource that buffers early energy and emotional depletion under high job demands, reducing burnout risk by enabling more effective coping and use of available supports (Hobfoll, 1989). However, once burnout develops, cumulative resource loss limits resilience’s protective capacity, positioning it primarily as a preventive buffer rather than a moderator of later outcomes (Halbesleben et al., 2014).

COR theory further suggests that personal resources amplify organisational resources through “resource caravans,” enhancing engagement, reducing burnout and improving performance (Hobfoll, 2011; Xanthopoulou et al., 2007). In healthcare, this interaction maximises the effectiveness of organisational supports in high-demand environments.

COR theory conceptualises resources as interdependent “resource caravans” that operate synergistically rather than in isolation (Hobfoll et al., 2018). Psychological, social and material resources reinforce one another, generating gain spirals that enhance adaptation and performance, while depletion can trigger cascading loss. Evidence shows that combined resources such as resilience, leadership support and social networks produce multiplicative effects that exceed their individual contributions (Halbesleben et al., 2014). Thus, sustainable wellbeing and performance arise from the configuration and synergy of resources, not isolated resource inputs.

Leadership functions as a contextual enabler, shaping the conditions under which organisational processes operate rather than directly moderating specific relationships. Through visibility, facilitation and the creation of psychological safety, leaders signal priorities, remove barriers and enable learning, collaboration and innovation. In medical radiation practice, leadership and peer support help staff manage the emotional demands of sustained patient care, buffering emotional labour and reducing burnout risk (Anderson et al., 2025a). Supportive leadership promotes authentic emotional expression, strengthens psychological wellbeing and fosters resilience in high-pressure, collaborative environments (Konečná et al., 2026). Empirical evidence shows that leader support enhances motivation, creativity and employee voice via psychological safety (Wang et al., 2018), underscoring leadership’s indirect, enabling role and explaining why its effects are often not captured by simple moderation models.

Recent research suggests that resilience is not a universal predictor of adaptive performance but a conditionally effective resource dependent on leadership (Britt et al., 2016; Hartmann et al., 2020). Supportive leaders, clear goals and psychological safety enable resilient capacities such as flexible coping and rapid recovery to translate into performance outcomes (Edmondson and Lei, 2014). In contrast, low support or ambiguous expectations can diminish resilience’s effects (King et al., 2016). Thus, leadership critically moderates the resilience–outcome relationship.

A five-scenario framework shows how resilience and leadership jointly shape adaptive performance. Low resilience and low support produce strain and poor outcomes (Britt et al., 2016). Resilience without leadership offers limited short-term buffering but fails under structural or relational constraints (King et al., 2016), while leadership without resilience stabilises the environment but cannot fully compensate for low individual adaptive capacity (Hartmann et al., 2020). Only high resilience combined with supportive, psychologically safe leadership amplifies adaptation, sustaining performance and wellbeing (Edmondson and Lei, 2014). Therefore, leadership critically moderates the effects of resilience.

This study frames resilience and leadership support as mutually reinforcing, with multiplicative rather than additive effects. While COR theory highlights clustered resources and gain spirals (Hobfoll, 1989) and JD–R theory notes resource interactions (Bakker and Demerouti, 2017), empirical models often assume linear effects. Using a five-scenario framework, we show that resilience enhances performance and wellbeing only when activated by supportive leadership. Leadership alone provides stability but does not drive strong adaptation. These findings suggest that high performance under strain arises from synergistic resource networks, not simple resource accumulation.

The enabling role of leadership support is operationalised here through moderated mediation, treating supervisory support as a moderator of the resilience moderation effect, consistent with prior empirical research on conditional resource effects (Britt et al., 2016; Hartmann et al., 2020). This approach captures the practical mechanism through which leadership shapes resource synergy within the constraints of the research design. The full contextual process through which leadership operates, encompassing relational climate, psychological safety and structural conditions, draws on complementary theoretical frameworks including leader-member exchange and psychological safety climate (Edmondson and Lei, 2014; Wang et al., 2018), and multilevel or longitudinal designs would be better positioned to capture these dynamics in future research.

Data were collected from 165 MRPs across Australia through an online survey distributed via Qualtrics. The participating cohort comprises registered medical radiation science professionals working in medical imaging and radiation therapy departments across Australia. Participants were over the age of 18 years old and had completed their undergraduate qualification. The sample comprised 72.7% females with mean age range of 41 years. Educational qualifications ranged from Associates Degree to PhDs (M = 4.32, SD = 0.89 on a 7-point scale).

Participation was voluntary and anonymous, with informed consent obtained from all participants. The study received ethics approval from the University’s Human Research and Ethics Committee approval HREC 22395 prior to data collection commencing.

All constructs were measured using established scales with 5-point Likert response formats (1 = Strongly Disagree to 5 = Strongly Agree), except where noted. Specific details for each measure follow:

WBHRM was measured using the 35-item scale developed by Pariona-Cabrera et al. (2024), which operationalises Guest's (2017) five-dimensional WBHRM framework across investment in employees, engaging work, positive and safe environment, employee voice and organisational support. This scale was assembled from ten established sub-scales drawn from prior HRM literature and validated across Australian and Chinese healthcare samples. In the current study, CFA confirmed the factor structure in the MRP population, yielding excellent fit: χ²/df = 1.338, CFI = 0.950, TLI = 0.945, RMSEA = 0.045, with strong composite reliability for each dimension (Voice CR = 0.925, Investment in Employees CR = 0.909, Engaging Work CR = 0.910, Organisational Support CR = 0.889, Positive Environment CR = 0.874). All items were rated on a five-point Likert scale (1 = Strongly Disagree to 5 = Strongly Agree).

Surface acting was measured using items adapted from Morris and Feldman's (1996) emotional labour scale, focusing specifically on the surface acting dimension. This scale measures the extent to which employees fake or suppress their genuine emotions to display organisationally required emotions.

Leadership support was operationalised at the direct supervisor level, capturing two theoretically related dimensions of line manager behaviour. The first dimension, leader emotional support, reflects supervisor responsiveness to staff distress and active assistance in managing workplace stress, behavioural expressions of care directed towards individual practitioners (Andersen et al., 2025b). The second dimension, perceived supervisory support, reflects practitioners' global perceptions that their immediate supervisor values their contributions and cares about their wellbeing (Lee and Shin, 2023). Together these dimensions capture both the behavioural and relational climate aspects of direct supervisory support. This operationalisation focuses on immediate supervisors as the primary conduit through which organisational resources reach practitioners, consistent with research demonstrating that direct supervisory relationships are the most proximal and influential source of leadership support for frontline healthcare workers (Edmondson and Lei, 2014). Items were rated on a five-point Likert scale.

The Oldenburg Inventory (OLBI) is a self-report measure of burnout developed by Evangelia Demerouti and colleagues in 1998 (Demerouti and Bakker (2008). This tool measures the two dimensions of burnout namely exhaustion and disengagement. This tool is widely used in healthcare, education and business.

In this study, the exhaustion and disengagement subscales were combined into a single composite burnout score, consistent with prior research treating the OLBI as a unidimensional measure of overall burnout (Demerouti and Bakker, 2008). This approach is justified given that the two dimensions are theoretically interdependent and empirically correlated, and the study's focus is on burnout as a resource depletion state rather than its dimensional structure.

In this study, the resilience component of the Psychological Capital Questionnaire Scale by Luthans et al. (2007) was utilised. This scale draws from positive organisational behaviour and have been adapted from established measures by Wagnild and Young (1993). The questions on resilience were rated on a Likert-based scale representing six items.

Intention to leave was measured using a four-item measure with sample questions “I often think of leaving the organisation” as per Ang et al. (2013).

To minimise common method variance, several procedural controls were implemented during data collection following Podsakoff et al.'s (2003) recommendations. Participants were assured of full anonymity and informed that there were no right or wrong answers, reducing evaluation apprehension. Predictor and criterion variables were separated within the survey instrument and presented under distinct headings to reduce item proximity effects. Demographic items were included at the end of the survey to assist in reducing drop off rates, minimise bias and reduce respondent fatigue (Dobosh, 2017). At the analysis stage Harman's single-factor test was conducted as a statistical diagnostic, with the first factor accounting for 35.2% of total variance, well below the 50% threshold, suggesting common method variance is unlikely to account for observed patterns (Podsakoff et al., 2003).

We first conducted confirmatory factor analysis (CFA) using AMOS 26.0 to assess the measurement model's fit, reliability and validity. Hypothesis testing employed Hayes' (2022) PROCESS macro-Version 4.2 for SPSS with 5,000 bootstrap samples. We used Model 4 for simple mediation (H1), Model 6 for serial mediation (H2), Model 7 for first-stage moderated mediation (H3) and Model 11 for three-way moderated mediation (H4). All variables were mean-centred before creating interaction terms (Aiken and West, 1991).

Table 1 presents descriptive statistics, zero-order correlations and reliability coefficients. All measures demonstrated acceptable to excellent internal consistency, with Cronbach's alpha coefficients ranging from 0.74 (wellbeing HRM practices) to 0.97 (leadership support) (Nunnally and Bernstein, 1994). Correlations were consistent with theoretical predictions. Wellbeing HRM practices showed strong negative correlations with burnout (r = −0.83, p < 0.01) and turnover intentions (r = −0.72, p < 0.01), and a moderate negative correlation with surface acting (r = −0.45, p < 0.01). Burnout demonstrated a strong positive correlation with turnover intentions (r = 0.79, p < 0.01), consistent with meta-analytic findings (Swider and Zimmerman, 2010). Surface acting was positively correlated with both burnout (r = 0.63, p < 0.01) and turnover intentions (r = 0.36, p < 0.01), supporting the proposed serial mediation pathway. Employee resilience showed modest negative correlations with burnout (r = −0.21, p < 0.01) and turnover intentions (r = −0.18, p < 0.05). Notably, resilience and leadership support were essentially uncorrelated (r = −0.01, ns), indicating these personal and contextual resources are independent constructs, strengthening the rationale for examining their interactive effects.

The hypothesised six-factor measurement model (Table 2) demonstrated excellent fit: χ² (233) = 323.56, p < 0.001; χ²/df = 1.39; CFI = 0.97; TLI = 0.96; IFI = 0.97; RMSEA = 0.049 [90% CI: 0.035, 0.061], PCLOSE = 0.557; SRMR = 0.070. All factor loadings were significant (p < 0.001) and ranged from 0.583 to 0.966 (M = 0.76), supporting the proposed factor structure. Composite reliability values ranged from 0.737 to 0.965, exceeding the 0.70 threshold (Fornell and Larcker, 1981). While AVE values for WBHRM (0.486) and burnout (0.492) fell slightly below the conventional 0.50 threshold, they exceeded the 0.40 criterion acceptable when CR > 0.70 (Fornell and Larcker, 1981), and all individual factor loadings exceeded 0.60, indicating adequate convergent validity.

Discriminant validity was assessed using heterotrait-monotrait (HTMT) ratios (Table 3). All 15 construct pairs met the 0.90 threshold, with 93% meeting the more conservative 0.85 criterion. The single exception was burnout-turnover intentions (HTMT = 0.872), which is theoretically expected given burnout's role as a proximal predictor of turnover (meta-analytic ρ = 0.54; Swider and Zimmerman, 2010). Harman's single-factor test indicated the first factor accounted for 35.2% of total variance, well below the 50% threshold (Podsakoff et al., 2003), suggesting common method variance is not a major concern.

Simple mediation analysis (PROCESS Model 4) revealed that wellbeing HRM practices were significantly associated with lower burnout (b = −1.16, SE = 0.06, t = −19.51, p < 0.001, 95% CI [−1.28, −1.04]), explaining 70.8% of variance in burnout (R² = 0.708). Burnout, in turn, significantly predicted turnover intentions (b = 1.38, SE = 0.19, t = 7.09, p < 0.001, 95% CI [0.99, 1.77]). The bootstrap analysis revealed a significant indirect effect (ab = −1.60, SE = 0.22, 95% CI [−2.04, −1.18]), with a completely standardised effect of −0.53, indicating a large effect. The direct effect of wellbeing HRM on turnover remained significant (c' = −0.61, SE = 0.27, p = 0.026), indicating partial mediation. The indirect effect substantially exceeded the direct effect, consistent with burnout as the primary mechanism through which WBHRM is associated with retention. H1 was fully supported.

Serial mediation analysis (PROCESS Model 6) supported the hypothesised sequential pathway. Wellbeing HRM practices significantly reduced surface acting (b = −0.78, SE = 0.11, p < 0.001; R² = 0.245), supporting H2a. Both wellbeing HRM (b = −0.97, SE = 0.06, p < 0.001) and surface acting (b = 0.25, SE = 0.04, p < 0.001) were significantly associated with burnout (R² = 0.774), supporting H2b. Burnout strongly predicted turnover intentions (b = 1.71, SE = 0.22, p < 0.001). Interestingly, surface acting showed a negative direct effect on turnover when controlling for burnout (b = −0.36, SE = 0.11, p = 0.001), representing a suppression effect. This unexpected finding suggests that the association between surface acting and turnover is more complex than the theoretical model anticipated and warrants investigation in future research.

Three specific indirect pathways emerged: (1) WBHRM → surface acting → turnover (b = 0.28, 95% CI [0.11, 0.46]), a positive but suppressed effect accounting for 13% of total mediation; (2) WBHRM → burnout → turnover (b = −1.66, 95% CI [−2.09, −1.25]), the dominant mechanism accounting for 75% of mediation; and (3) WBHRM → surface acting → burnout → turnover (b = −0.33, 95% CI [-0.50, −0.19]), the serial pathway of theoretical interest accounting for 15% of mediation. The completely standardised serial indirect effect was −0.11, indicating a medium effect. The total indirect effect was −1.71 (95% CI [−2.16, −1.26]), representing 77% of wellbeing HRM's total effect on turnover. The final model explained 65.7% of variance in turnover intentions. H2a, H2b and H2c were all supported.

Moderated mediation analysis (PROCESS Model 7) examined whether resilience moderates the WBHRM → burnout relationship. The WBHRM × resilience interaction approached conventional significance (b = −0.24, SE = 0.13, t = −1.89, p = 0.060, 95% CI [-0.49, 0.01]), accounting for an additional 0.6% of variance in burnout. While the p-value exceeded 0.05, the effect was in the predicted direction, and the confidence interval barely included zero.

Conditional effects analysis revealed that wellbeing HRM's effect on burnout strengthened with increasing resilience: low resilience (b = −1.04, 95% CI [−1.20, −0.89]), mean resilience (b = −1.15, 95% CI [−1.26, −1.03]) and high resilience (b = −1.25, 95% CI [−1.41, −1.09]). The effect was 20% stronger for high-resilience compared to low-resilience employees. Correspondingly, conditional indirect effects on turnover through burnout increased across resilience levels: low resilience (b = −1.44, 95% CI [−1.89, −1.02]), mean resilience (b = −1.58, 95% CI [−2.04, −1.13]) and high resilience (b = −1.72, 95% CI [−2.25, −1.23]), representing a 19.4% difference between high and low resilience. The index of moderated mediation was significant (−0.33, 95% CI [−0.74, −0.01]), providing formal evidence for moderated mediation. H3a received marginal support, while H3b was fully supported.

Three-way moderated mediation analysis (PROCESS Model 11) tested whether leadership support conditions the resilience moderation effect. The full model explained 73.4% of variance in burnout (R² = 0.734). The omnibus three-way interaction (WBHRM × resilience × leadership) was not statistically significant (b = −0.07, SE = 0.11, p = 0.522, 95% CI [-0.29, 0.15]) and accounted for minimal additional variance (ΔR² = 0.0007).

However, conditional analyses revealed theoretically meaningful patterns consistent with regional significance (Spiller et al., 2018). Table 4 presents conditional indirect effects across nine combinations of resilience and leadership levels. All nine effects were significant, but the pattern revealed systematic strengthening: effects increased as both resilience and leadership increased. The strongest effect occurred at high resilience and high leadership (−1.79, 95% CI [−2.61, −1.18]), representing a 40% improvement over the baseline condition of low resilience and low leadership (−1.28, 95% CI [−1.76, −0.81]). This 40% enhancement (difference = 0.51) translates to a completely standardised difference of approximately 0.61 standard deviations in turnover intentions. Table 5 provides a summary of the hypotheses tests undertaken in this analysis.

Critically, the pattern was non-additive. Moving from low to high resilience alone (with low leadership) produced a 19% improvement (−1.28 to −1.53), and moving from low to high leadership alone (with low resilience) produced only an 8% improvement (−1.28 to −1.38). However, moving from low to high on both simultaneously produced a 40% improvement – substantially exceeding the sum of individual improvements (19% + 8% = 27%), supporting resource synergy.

Indices of conditional moderated mediation at each leadership level provided direct evidence for conditional moderation: low leadership (−0.30, 95% CI [−0.89, 0.24], ns), mean leadership (−0.39, 95% CI [−1.02, 0.03], approaching significance) and high leadership (−0.49, 95% CI [−1.37, −0.03], significant). This pattern demonstrates that resilience significantly moderates the indirect effect only when leadership support is high, but not when leadership is low. H4a was not supported based on the non-significant omnibus interaction; however, H4b was fully supported: resilience moderation emerges only in contexts of high leadership support.

This study examined the mechanisms and boundary conditions through which WBHRM practices is associated with employee turnover intentions in healthcare settings. Drawing on COR theory, we tested a serial mediation model and a moderated mediation model. Four key findings emerged. First, burnout serves as a critical mediator linking WBHRM practices to turnover intentions. The effect of WBHRM practices operates predominantly through their capacity to prevent resource depletion, rather than through alternative pathways. This finding situates burnout not merely as a correlate but as the primary conduit through which WBHRM influences retention. Prior work in this area has been extended by demonstrating that the effectiveness of such practices is rooted in resource protection processes consistent with COR theory (Hobfoll, 2001).

Second, serial mediation through surface acting and burnout provides evidence for a sequential resource depletion process. WBHRM practices reduce surface acting demands, which in turn reduces burnout, ultimately decreasing turnover intentions. Although secondary to the direct burnout pathway, it represents a distinct mechanism through which organisational practices influence retention. This finding suggests that WBHRM practices operate not only by directly replenishing resources but also by reducing the resource drains associated with emotional labour. The identification of this two-stage depletion process contributes to emotional labour theory by clarifying the pathway from display rules to withdrawal behaviours (Grandey, 2003; Pinkawa et al., 2024).

Third, employee resilience moderates the effectiveness of WBHRM practices in reducing burnout (first-stage moderated mediation). Individuals higher in resilience derive greater burnout-reducing benefits from WBHRM practices, resulting in stronger indirect effects on turnover intentions. This pattern supports COR theory's resource investment principle: individuals with greater personal resources are better positioned to leverage organisational resources (Hobfoll, 2011; Brennan et al., 2024). Importantly, resilience did not buffer the burnout to turnover relationship (second stage), suggesting that once burnout develops, its negative consequences are relatively uniform regardless of resilience levels.

Finally, conditional analyses provided exploratory evidence of resource synergy involving resilience and leadership support. While the omnibus three-way interaction was not statistically significant (b = −0.07, p = 0.522), consistent with H4a not being supported, conditional analyses following Spiller et al.'s (2018) regional significance framework revealed a theoretically meaningful pattern: resilience moderated the indirect effect of WBHRM on turnover only when leadership support was high. We interpret this pattern cautiously, as regional significance testing cannot substitute for a significant omnibus interaction, and the non-significance of H4a is likely partly attributable to limited statistical power with n = 165 for a three-way interaction (Aguinis, 2004). Nonetheless, the conditional pattern is directionally consistent with COR theory's resource interaction principle and warrants replication with larger samples. The practical implication, that WBHRM, resilience-building and leadership development should be treated as coordinated rather than independent investments, remains theoretically grounded even where formal confirmation of the interaction awaits future research.

This study makes five key theoretical contributions. We extend COR theory in healthcare management by demonstrating that WBHRM supports employees through two complementary pathways: direct resource replenishment via burnout reduction; and indirect resource conservation via surface acting reduction upstream of burnout. This dual-pathway model clarifies the sequential nature of resource protection and suggests that comprehensive WBHRM practices address multiple sources of resource depletion simultaneously, consistent with Hobfoll's (2011) resource caravan principle.

Second, this study makes an important contextual contribution by generating rare empirical evidence from MRPs, one of the more data-scarce cohorts in allied health research. With a total nationally registered workforce of only 21,335 practitioners dispersed across Australia, MRPs are structurally difficult to access for research, due to systemic constraints in the workforce and work environment, yet face some of the most acute surface acting demands in the healthcare system. Practitioners routinely manage patients experiencing cancer diagnoses, trauma and end-of-life treatment while maintaining clinical precision and professional composure under conditions of chronic workforce shortage. Despite these demands, MRPs remain almost entirely absent from the HRM literature, which has concentrated on nursing and medical populations. This study establishes a foundation for MRP-specific HRM research and provides evidence that the resource depletion dynamics identified in nursing populations extend to allied health professionals with meaningfully different role structures and regulatory frameworks. Third, we extend moderated mediation research in HRM by demonstrating first-stage rather than second-stage moderation effects of resilience. Rather than buffering the stress–outcome link (Cohen and Wills, 1985), personal resources enhance the acquisition and utilisation of organisational resources, highlighting the importance of where moderators operate within causal chains, representing a novel application of COR’s resource investment principle.

Fourth, we illuminate resource synergies through conditional moderation. Resilience is associated with stronger WBHRM effectiveness only under high leadership support, departing from additive moderation models and supporting COR's resource caravan principle. Personal and organisational resources realise their full potential only within supportive organisational contexts, advancing multilevel models of resource interactions. This finding is interpreted as exploratory given the non-significant omnibus interaction, and replication with larger samples is needed to establish the conditional pattern with greater confidence.

Fifth, this study advances understanding of surface acting as a specific and consequential emotional labour strategy in clinical settings. Rather than treating emotional labour as a broad construct, this study isolates surface acting as the focal mechanism and traces its sequential role in the resource-depletion process from WBHRM practices through burnout to turnover intentions. Drawing on Grandey (2000) and Brotheridge and Grandey (2002), who established surface acting as distinct from deep acting in its resource costs, this study demonstrates that the organisational antecedents of surface acting, and not just its consequences, are amenable to intervention through WBHRM. This reframes surface acting not only as an individual coping strategy but as an organisationally shaped phenomenon, opening new directions for HRM research on emotional labour management in high-demand healthcare environments.

Our findings provide actionable guidance for healthcare organisations aiming to reduce turnover. Burnout accounted for 75% of mediated turnover effects, highlighting the need for interventions such as adequate staffing, manageable workloads, flexible scheduling, mental health resources and early burnout screening. Preventing burnout is more effective than trying to retain already burnt-out employees.

Second, the significant serial mediation pathway highlights a 15% variance explained by surface acting, indicating its overlooked impact. Organisations should audit emotional labour requirements and display rules, train supervisors to allow authentic emotional expression, provide emotional regulation resources (e.g. debriefings, peer support) and reduce unnecessary “service with a smile” mandates.

Third, the moderation findings reveal that high-resilience employees gain 19% more from WBHRM practices than low-resilience ones. Organisations should combine broad WBHRM implementation with resilience-building interventions (e.g. strengths-based coaching, cognitive reframing) to maximise benefits for all staff.

Fourth, the resource synergy finding that resilience moderation emerges only under high leadership support, has implications for the implementation of WBHRM by line managers. Resilience is most effective under strong leadership support, with combined effects improving outcomes by 40%. WBHRM, leadership development and resilience training should be coordinated, not delivered in isolation, to create resource-rich contexts and maximise return on investment.

Fifth, the finding that resilience does not buffer the burnout-to-turnover relationship (second-stage moderation) carries an important warning: resilience does not buffer turnover once burnout occurs. Organisations must focus on proactive WBHRM practices rather than relying on employees to “push through” burnout.

This study has several limitations. Its cross-sectional design limits causal inference; longitudinal research is needed to establish temporal relationships. Reliance on self-reported measures raises common method variance concerns, though diagnostics were acceptable. Incorporating objective data such as actual turnover, supervisor ratings or organisational records would strengthen findings.

The sample size of 165, while sufficient for detecting the large effects observed in the primary mediation pathways, represents a recognised constraint for higher-order interaction testing. Detecting small interaction effects in moderated mediation models requires substantially larger samples, and the marginal moderation result for H3 and non-significant omnibus interaction for H4 are consistent with limited statistical power for small effects, a well-documented challenge in moderation research independent of sample size (Aguinis, 2004). These findings should therefore be interpreted with appropriate caution, and replication with larger samples is a clear priority.

Achieving larger samples in medical radiation practice research presents genuine structural challenges that are important to contextualise. The total nationally registered MRP workforce comprises only 21,335 practitioners (MRPBA, 2026), making this one of the smallest regulated health professions in Australia. Unlike nursing or medical research, where large institutional databases and established research networks facilitate high-volume data collection, MRP research must reach a geographically dispersed, time-pressured specialist workforce without the infrastructure of major hospital research programmes. The achieved sample represents a meaningful proportion of an inherently small and hard-to-reach population, and the robustness of the primary mediation findings suggests the core theoretical relationships are reliable within this cohort. Future research partnerships with the regulatory authority the MRPBA, professional associations such as ASMIRT and ANZSNM, or multinational collaborations across Asia–Pacific MRP workforces, would provide the sample sizes needed to adequately power higher-order interaction testing.

We focused on surface acting, a resource-depleting emotional labour strategy, rather than the full spectrum of emotion regulation. Healthcare workers may also engage in deep acting or authentic emotional expression, which could respond differently to WBHRM and affect burnout outcomes. Future research should explore how organisational support interacts with diverse emotional labour strategies.

We measured turnover intentions rather than actual turnover or broader organisational outcomes. Linking WBHRM to staff retention, performance and patient care would enhance practical relevance.

Despite these limitations, the study offers actionable insights. WBHRM is associated with lower turnover intentions primarily through burnout reduction, but effectiveness depends on employee resilience and leadership support. Wellbeing initiatives are most effective when delivered as integrated systems with trained leaders, highlighting that investments in employee wellbeing yield strongest returns when combined with leadership development and resilience-building programs.

This study demonstrates that WBHRM practices reduce employee turnover intentions among MRPs primarily through burnout prevention, with surface acting reduction as a secondary mechanism. Employee resilience enhances the effectiveness of these practices by enabling better resource acquisition and utilisation, but this resilience advantage manifests only in contexts of strong leadership (line managers) support, revealing a resource synergy effect. These findings support the COR theory's application to HRM-retention linkages and provide actionable guidance for healthcare organisations struggling with retention challenges. By investing in comprehensive WBHRM practices while simultaneously building employee resilience and developing supportive leadership, organisations can achieve substantial reductions in turnover intentions through efficacious psychological mechanisms.

Institutional Human Research and Ethics Committee approval HREC 22395.