Self-Control Links Stress and Quality of Life in Healthcare Workers with Different Work Patterns

This cross-sectional study was conducted between July 2024 and February 2025 among 82 shift-working healthcare and other academic professionals, recruited from clinical and academic settings at Notre Dame of Maryland University and affiliated rotation sites. A total of 135 individuals were initially approached via email and in-person recruitment. Of these, 82 provided consent and completed the study, yielding a participation rate of 60.7% and a rejection rate of 39.3%. Non-participation was mainly due to scheduling conflicts or lack of interest. The sample consisted of student pharmacists, pharmacy technicians and other healthcare professionals, all of whom routinely engaged in irregular or night shifts. Including these diverse professional roles allowed the study to capture a broad spectrum of shift-related stressors, regulatory challenges, and quality-of-life outcomes across both trainee and practicing healthcare worker populations. The survey was open to all shift-working healthcare and academic professionals at the institution and affiliated sites. Although participants included pharmacists, student pharmacists, pharmacy technicians, and other allied healthcare staff, the exact count for each role was not recorded. Eligible participants were full-time students or healthcare workers aged 18–60 years, fluent in English, and had engaged in at least one night or rotating shift (≥8 h) within the past 30 days. This criterion reflects prior evidence that the physiological and cognitive effects of shift work can persist up to 4–6 weeks post-exposure (American Academy of Sleep Medicine, 2014; Wickwire et al., 2017). Participants were excluded if they reported a current diagnosis of schizophrenia, major neurological disorder, or use of psychoactive medications known to impair cognitive function (e.g., antipsychotics, benzodiazepines).

Participants were categorized as active shift workers (currently engaged in shift work or exposure within the last 30 days during the study period). All participants provided informed consent, and the study protocol was approved by the Institutional Review Board at Notre Dame of Maryland University (IRB # PH043024SM01YMS).

The PSS-4 is a validated short-form measure assessing global stress appraisal over the past month using four Likert-type items, with higher scores indicating greater perceived stress (Cohen et al., 1983). The 4-item PSS demonstrated acceptable internal consistency in this sample (Cronbach's α = 0.707).

The ESQ-R total score was used as a global index of self-regulation difficulty, with higher scores indicating greater impairment in executive functioning (Strait et al., 2020). All interpretations reflect this scoring direction. Although the ESQ-R includes five subscales (e.g., planning, working memory, emotional control, impulse regulation, and organization), only the total score was analyzed to capture overall self-regulatory capacity. The scale demonstrated excellent internal consistency in our sample (Cronbach's α = 0.90), supporting its reliability in this context. Nevertheless, while ESQ-R is validated in educational settings, its sensitivity in occupational stress environments requires further evaluation.

The 16-item QOLS assesses satisfaction across major life domains, including social, physical, psychological, and occupational well-being (Burckhardt & Anderson, 2003). Higher scores indicate better overall life quality. The QOLS exhibited excellent reliability in this cohort (Cronbach's α = 0.91), indicating strong internal consistency in measuring overall quality of life.

Eighty-two shift-working healthcare professionals completed the study, offering a representative snapshot of healthcare trainees managing academic and occupational demands. The majority were in the 26–45 age range and female demographics that align with broader healthcare workforce trends. Nearly three-quarters reported working in urban settings, and approximately one-third had been exposed to a stable shift schedule for over two years, suggesting potential for cumulative circadian disruption. Most participants worked shifts of 8 h or less, and a substantial portion were single, reflecting transitional life stages often associated with increased psychosocial strain.

Mean scores on validated self-report instruments reflected moderate levels of stress (PSS-4: 6.49 ± 2.72) and executive self-regulation difficulty (ESQ-R: 21.91 ± 11.63). Despite these challenges, participants reported a relatively high quality of life (QOLS: 89.18 ± 14.40), with considerable inter-individual variability. These profiles form a data-rich foundation for modeling behavioral and psychological responses to shift work, insights that can be leveraged by AI systems designed for stress prediction, resilience monitoring, and digital phenotyping (see Table 1).

Data was analyzed using SPSS v28.0. To test our main hypothesis, mediation analysis was performed using the PROCESS macro (Model 4) (Hayes, 2017). We examined whether executive self-regulation (ESQ-R) mediated the relationship between perceived stress (PSS) and quality of life (QOLS) and tested the reverse pathway (QOLS → PSS via ESQ-R). Indirect effects were estimated using bias-corrected bootstrapping with 5000 iterations, and significance was determined by 95% confidence intervals that did not include zero. Two mediation models were analyzed in this study dataset, i.e., a forward and a backward model is a case of multiple testing. So, a p-value adjustment (i.e., adjusted p-value = 0.05/2 (number of mediation models), and a confidence interval (CI) adjustment (i.e., changed p-value from 95% to 97.5% for bootstrap confidence intervals) were used to establish the significance level. To enhance the reproducibility and stability of our findings, we set the number of bootstrap iterations to 5126. This decision was based on preliminary analyses demonstrating that this number consistently yielded stable bias-corrected confidence intervals across repeated runs. This approach is consistent with methodological guidance in mediation analysis, which recommends using adequately large bootstrap samples to improve the precision, reliability, and replicability of indirect effect estimates (Hayes, 2017). To explore naturally occurring subgroups within the sample, unsupervised K-means clustering was conducted using standardized scores on perceived stress (PSS-4), self-regulation (ESQ-R), and quality of life (QOLS).

This analysis evaluated the mediating effect of executive self-regulation on the relationship between perceived stress and quality of life among shift-working healthcare professionals.

Perceived stress (PSS) significantly and positively predicted executive self-regulation difficulties (b = 1.60, SE = 0.36, β = 0.44, p < 0.001, 95% CI = [0.78, 2.43]), suggesting that individuals experiencing greater stress reported more pronounced impairments in self-regulatory functioning.

In turn, higher self-regulation difficulty was significantly associated with reduced quality of life (b = −0.49, SE = 0.13, β = −0.40, p < 0.001, 95% CI = [−0.80, −0.19]), indicating that diminished regulatory control may play a critical role in deteriorating well-being.

The indirect effect of perceived stress on quality of life through self-regulation was statistically significant (b = −0.79, SE = 0.33, 95% CI = [−1.63, −0.17]), supporting indirect-only mediation, wherein the effect of stress on quality of life appears to operate primarily through regulatory disruption. Effect size analysis showed that self-regulation accounted for approximately 46% of the total effect in the forward mediation (stress → self-regulation → quality of life), suggesting a substantial mediating role.

The direct effect of stress on the quality of life was not statistically significant after accounting for the mediator (b = −0.94, SE = 0.48, β = −0.21, p = 0.054, 95% CI = [−2.03, 0.16]).

These findings underscore that higher levels of perceived stress impair executive self-regulation, which in turn compromises quality of life. The results highlight self-regulation as a central psychological pathway through which stress exerts its effects, aligning with an indirect-only mediation framework. (see Figure 1 and Table 2).

A reverse mediation analysis was conducted to examine whether self-regulation mediates the relationship between quality of life and perceived stress.

Quality of life (QOLS) was a statistically significant negative predictor of self-regulation difficulty (b = −0.40, SE = 0.08, β = −0.49, p < 0.001, 95% CI = [−0.58, −0.22]), suggesting that individuals with greater life satisfaction exhibited stronger executive self-regulation.

Self-regulation significantly predicted perceived stress in this model (b = −0.09, SE = 0.03, β = −0.34, p = 0.004, 95% CI = [−0.16, −0.02]), indicating that greater regulatory difficulty was associated with higher perceived stress levels.

The indirect effect of quality of life on perceived stress through self-regulation was statistically significant (b = −0.04, SE = 0.02, 95% CI = [−0.08, −0.01]), suggesting an indirect-only mediation model. Effect size analysis showed reverse pathway proportion accounted for 44% of the total effect, but its absolute effect size was small due to a weak total effect. The direct effect of quality of life on perceived stress was marginally significant (b = −0.05, SE = 0.03, β = −0.22, p = 0.054, 95% CI = [−0.11, 0.01]) (Figure 2; Table 2).

A three-cluster solution provided a conceptually interpretable structure, identifying distinct behavioral profiles consistent with prior clustering literature (Wettstein et al., 2023; Table 3).

These behavioral phenotypes may serve as a foundation for developing future screening tools and personalized wellness interventions in pharmacy and healthcare education settings.

The forward mediation pathway observed in this study aligns with established stress and resilience frameworks, which emphasize that cognitive and emotional regulation capacities shape how individuals respond to stress (Gross, 1998; Lazarus & Folkman, 1984; McEwen, 1998). Recent evidence also shows that adaptive regulation strategies can mitigate burnout among healthcare professionals (Pálfi et al., 2024), underscoring the importance of targeting these skills in interventions. Our findings support this view, demonstrating that the indirect effect of stress on quality of life operates primarily through regulatory mechanisms, while the reverse pathway is weaker, highlighting an asymmetrical but bidirectional relationship.

Our findings add empirical support to this view in the context of shift-working healthcare professionals, who face compounding risks such as exposure to bright light, sleep deprivation, circadian disruption, and excessive caffeine consumption (Kalmbach et al., 2018; Salahuddin et al., 2025a; Soleimani et al., 2024; Wickwire et al., 2017; Zeek et al., 2015). These stressors can impair prefrontal cortex functions responsible for executive control, while simultaneously heightening amygdala reactivity and dysregulating hypothalamic–pituitary–adrenal (HPA) axis function, mechanisms that degrade working memory, emotion regulation, and behavioral inhibition (Kinlein et al., 2015; McEwen & Morrison, 2013). Our data support the hypothesis that these impairments are not merely side effects of stress but active mediators of its impact on quality of life, underscoring the neurocognitive cost of chronic stress exposure in this population.

Building on our earlier findings related to sex differences in stress biomarkers among student pharmacists (Salahuddin et al., 2025a, 2025b), this study further addresses a critical gap in the literature concerning shift-working healthcare professionals. While prior research has shown that pharmacy students experience high stress levels, poor sleep quality, and frequent stimulant use (Zeek et al., 2015; Cates et al., 2015; Hagemann et al., 2020), most studies in this space have focused on descriptive statistics or direct correlations, rather than underlying explanatory mechanisms. Furthermore, burnout interventions in healthcare training have tended to emphasize mindfulness and stress awareness, while often overlooking executive functioning and behavioral regulation skills that may buffer against real-time decision fatigue, emotional overload, and performance lapses (Rompilla et al., 2023). By centering self-regulation as a mediating mechanism and assessing it via the ESQ-R, this study adds theoretical depth and practical utility to the conversation on pharmacy and related health professions education and clinical workforce resilience (Langran et al., 2022; Murry & Witry, 2020). It also lays conceptual groundwork for integrating digital monitoring tools that can track self-regulatory breakdowns and predict quality-of-life outcomes. (Oudin et al., 2023).

The findings point to clear educational and institutional implications. As burnout and stress-related impairment rise among healthcare trainees and practitioners (Batanda, 2024), efforts to promote resilience must move beyond self-care slogans and toward evidence-based interventions that strengthen regulatory capacity (Luthar & Cicchetti, 2000). Training modules in cognitive reappraisal, flexible goal-setting, and impulse control, key subdomains of self-regulation, can be integrated into pharmacy and related health profession curricula, residency onboarding, and continuing education (Cutuli, 2014; Deci & Ryan, 2000). These skills are particularly important during transitional periods when regulatory demands are high and cognitive load increases (Luciana, 2016). Furthermore, brief executive functioning screeners may help identify at-risk individuals during times of transition (e.g., final year placements and shift rotations; Johnson et al., 2007). Tailoring interventions to support those with high cognitive load and poor sleep, especially shift-working healthcare professionals, may offer a cost-effective way to prevent burnout and promote quality of care (Hagemann et al., 2020; Razai et al., 2023).

To further investigate the heterogeneity of experiences within the sample, unsupervised K-means clustering revealed three distinct stress–regulation–well-being profiles. These included a "resilient cluster" group with high stress, moderate regulation difficulty, and preserved quality of life; a "low-strain cluster" group with low stress, low regulation difficulty, and high quality of life; and a "high-strain cluster" group marked by high stress, high regulation difficulty, and diminished QOL. Prior studies have shown that data-driven profiling using clustering methods can identify meaningful psychological subtypes across stress, coping, and mental health dimensions (Drysdale et al., 2017; Hudon et al., 2023; Wang et al., 2021). These findings provide empirical support for the development of AI-enabled screening tools that could classify frontline professionals into high- and low-risk categories, supporting early identification of burnout trajectories (Ghosh et al., 2022; Kamath et al., 2022; Lazarou & Exarchos, 2024). Clustering also aligns with precision mental health frameworks, suggesting that resilience interventions may benefit from stratification based on behavioral phenotypes rather than assuming a one-size-fits-all model (Insel, 2014; Kapur et al., 2012; Mohr et al., 2017).

The present findings highlight the utility of these clusters in guiding tailored intervention strategies. For instance, individuals in the low-stress, high-regulation, high-QOL cluster may require minimal intervention, focusing instead on maintenance and periodic well-being monitoring. In contrast, those in the burnout-risk cluster, with elevated stress and compromised self-regulation, may benefit from intensive support such as cognitive-behavioral therapy, stress inoculation training, or digital mind–body interventions. Intermediate profiles, such as a resilient cluster, could benefit from resilience training to maintain function under continued stress exposure. Therefore, the cluster-based approach not only enhances conceptual understanding of stress vulnerability but also offers a practical framework for optimizing resource allocation and intervention delivery, saving time, reducing cost, and improving outcomes through targeted mental health support.

The reverse mediation model, in which quality of life was hypothesized to improve stress levels via stronger self-regulation, was partially supported. While higher life satisfaction was modestly associated with better regulatory capacity, this improvement did not translate to reduced perceived stress, supporting an indirect-only pathway with a marginal direct effect. This aligns with resource-based theories such as the Strength Model of Self-Control (Baumeister et al., 2007) and Ego Depletion Theory (Baumeister et al., 1998), which suggest that regulation is more often depleted by stress than bolstered by well-being. Neurobiological evidence also shows that chronic stress impairs the very neural circuits involved in control, meaning that top-down recovery often requires specific intervention, not just environmental or hedonic improvements (McEwen & Morrison, 2013; Mika et al., 2012; Wolff et al., 2021). In healthcare, this reinforces that even satisfied professionals may not be shielded from stress-related cognitive breakdown unless they are actively engaging in regulation-enhancing strategies.

Nonetheless, several limitations must be noted. First, the cross-sectional design restricts causal interpretation, although the model aligns with prior longitudinal frameworks. Notably, while both mediation models yielded statistically significant indirect effects, the small sample size may limit the stability of these findings and inflate the risk of Type I and II errors. These results do not appear to be explained by multicollinearity, suppression effects, or model misspecification. Future work should use larger sample sizes and temporal or experimental designs to validate these pathways. Additionally, digital health approaches such as wearable sensors or mobile-based behavioral analytics may offer scalable methods to monitor regulatory functions in real time. Second, reliance on self-report instruments may introduce bias; incorporating physiological (e.g., cortisol), behavioral (e.g., cognitive tasks), or passive sensing data (e.g., actigraphy) could improve measurement accuracy and reduce self-report bias. Third, the sample was limited to shift-working healthcare professionals, potentially reducing generalizability to daytime workers or those in less cognitively demanding occupations. The moderate response rate (60.7%) and non-random recruitment process may have introduced selection bias, limiting generalizability. Additionally, the absence of detailed counts for specific occupational roles limits the ability to explore role-specific differences in stress, regulation, and quality-of-life outcomes. Furthermore, the small sample size, predominantly urban and female composition, and reliance on voluntary participation constrain the representativeness of findings. Future research should include more diverse and larger samples to confirm these results. Moreover, while descriptive data on shift length were collected, subgroup analyses comparing different shift durations (e.g., <8 h vs. ≥12 h) were not conducted. Shift exposure intensity may influence stress and regulation outcomes, and future studies should investigate whether variations in shift duration alter these relationships. Finally, unmeasured contextual factors such as job control, social support, and organizational culture may moderate the observed effects and should be examined in future studies.