Insulin resistance is a pathological condition characterized by inadequate tissue response to normal insulin action, causing impaired glucose uptake and metabolism [1]. Insulin resistance affects approximately 26.5% of the global adult population [2], representing the central pathophysiological mechanism underlying Type 2 diabetes, and substantially increases the risk of cardiovascular diseases, obesity, hepatic steatosis, and metabolic syndrome [3]. This asymptomatic metabolic dysfunction precedes overt Type 2 diabetes by 10~15 years [1], during which progressive beta cell compensation masks deteriorating glucose homeostasis [4]. In Korea, the adult prevalence of insulin resistance is approximately 23.3% [2], while metabolic syndrome affected one-third of the adult population as of 2020, reflecting a 22.5% increase over the past two decades [5]. Insulin resistance remains clinically silent until complications manifest [6]; thus, identifying at-risk populations and modifiable predictors is critical for preemptive intervention.

Among the occupational factors that may influence insulin resistance risk, shift work has emerged as a significant exposure, affecting approximately 25% of the global workforce and many Korean workers across the manufacturing, healthcare, transportation, and service sectors [7,8]. Shift work disrupts the alignment between endogenous circadian rhythms and external environmental cues, causing chronic circadian discordance [9]. This desynchronization triggers multiple pathophysiological pathways that affect diverse health outcomes. At the metabolic level, disrupted suprachiasmatic nucleus signaling impairs vasopressin-mediated glucose uptake, while suppressed nocturnal melatonin secretion reduces pancreatic insulin synthesis [10]. Beyond glucose metabolism, circadian disruption dysregulates lipid metabolism, promotes systemic inflammation through elevated pro-inflammatory cytokines (TNF-α, IL-6), and impairs vascular endothelial function [11,12]. These mechanisms collectively increase susceptibility to insulin resistance, cardiovascular diseases, inflammatory disorders, and metabolic syndrome [13,14]. A recent meta-analysis of over 235,800 participants demonstrated that shift workers exhibited a 30% increased risk of Type 2 diabetes compared with the risk among day workers [13]. Additionally, an umbrella review has confirmed elevated risks of obesity (25% increase), hypertension (30% increase), and mental health disorders (32%~42% increase), establishing shift workers as a particularly vulnerable occupational group [14].

Previous research on insulin resistance in shift workers has revealed multiple potential risk factors. Demographic factors such as age and sex have shown varying associations, with some cohorts reporting an elevated insulin resistance risk among women shift workers, whereas others have suggested context-dependent patterns [2,15]. In addition, lifestyle factors, including poor dietary habits, physical inactivity, and irregular meal timing, are associated with a high prevalence of insulin resistance [16]. Metabolic parameters, particularly obesity and metabolic syndrome, demonstrate robust associations with insulin resistance; however, the temporal relationships remain debated [17]. Emerging evidence suggests that psychological stress contributes to the development of insulin resistance [18] and depressive symptoms have also been positively associated with increased insulin resistance [19]. Furthermore, quality of life, as a multidimensional indicator of physical and psychological well-being [20], was adjusted for to minimize confounding from general health status.

Although shift work is an independent risk factor for insulin resistance [16], previous studies on the relationship between shift work and metabolic disorders had several limitations [21,22]. Previous studies conducted primarily in the nursing and transportation sectors have largely focused on single occupations and predominantly single-sex samples [21,22]. These underscore the need for analyses that leverage nationally representative sampling frameworks and consider complex survey methods to generate robust population-level estimates [21,22]. This study extends this research trajectory by considering nationally representative data from Korean shift workers across diverse occupations.

This study aimed to identify significant predictors associated with the likelihood of having insulin resistance among Korean shift workers from the Korea National Health and Nutrition Examination Survey (KNHANES) [23]. To achieve this, we first characterized the study population by comparing clinical and lifestyle factors between shift workers with and without insulin resistance, followed by multivariable logistic regression analysis to determine independent associations.

We investigated the prevalence of insulin resistance and its predictors among Korean shift workers using nationally representative KNHANES data and a complex sampling methodology [23]. Weighted logistic regression models were used to identify independent predictors while adjusting for relevant covariates. The prevalence of insulin resistance among shift workers in this study was 28.5%, higher than the global and Korean adult averages of 26.5% and 23.3%, respectively [2]. While these findings should be interpreted cautiously, considering the variability in insulin resistance cut-offs, sex distribution, and menopausal status across studies [2], and the absence of a non-shift worker comparison group in our study, they may reflect potential metabolic vulnerabilities in this occupational population. Three independent predictors emerged: higher BMI, metabolic syndrome, and elevated perceived stress were associated with increased insulin resistance risk.

BMI remained a significant predictor after adjusting for metabolic syndrome, indicating that total adiposity contributes to insulin resistance through pathways not fully captured by metabolic syndrome criteria. While metabolic syndrome defines abdominal obesity by waist circumference thresholds, BMI reflects total fat mass including subcutaneous depots that secrete adipokines and inflammatory mediators [1,16]. In shift workers, this relationship may be influenced by circadian and behavioral factors associated with shift work, including disrupted sleep-wake cycles and altered meal timing patterns [10,27]. However, our study did not assess specific shift work characteristics (such as shift type, rotation patterns, or night shift frequency), precluding definitive conclusions about how particular shift work schedules might modify the BMI-insulin resistance relationship. BMI's practical advantage lies in its simplicity, enabling self-monitoring without laboratory resources despite healthcare access barriers faced by shift workers.

Metabolic syndrome demonstrated the strongest association (OR = 2.64), remaining significant after BMI adjustment. This pronounced effect likely reflects synergistic interactions among syndrome components rather than simple additive effects. Each component—hypertension, dyslipidemia, hyperglycemia, and abdominal obesity—not only indicates existing dysfunction but also actively promotes insulin resistance through distinct pathways: endothelial dysfunction impairing insulin-mediated glucose delivery, ectopic lipid deposition in muscle and liver, and glucotoxicity further impairing insulin signaling [16,24]. Meta-analyses demonstrate that shift work increases metabolic syndrome prevalence [24], though the specific mechanisms linking shift work schedules to metabolic syndrome development require further investigation with detailed exposure characterization. However, metabolic syndrome assessment requires comprehensive clinical and laboratory evaluations, limiting feasibility for frequent monitoring.

The independent contributions of BMI and metabolic syndrome suggest they capture different dimensions of metabolic risk. BMI primarily reflects quantitative adipose burden and inflammatory state, while metabolic syndrome indicates qualitative dysregulation across multiple organ systems. For practical surveillance, BMI monitoring can facilitate frequent self-assessment, while periodic comprehensive metabolic syndrome evaluation should be conducted regardless of BMI status, as metabolic abnormalities may occur even with normal weight. Individuals with either elevated BMI or any metabolic syndrome components warrant intensified monitoring. Longitudinal study demonstrates bidirectional relationships, in which insulin resistance can promote subsequent weight gain and metabolic dysfunction [28]. However, our study lacks detailed shift work exposure data (shift patterns, duration, night shift frequency) and a non-shift worker comparison group, precluding definitive conclusions about shift work's specific contribution to the observed associations.

Elevated perceived stress was independently associated with insulin resistance, with a 49% increase in the odds among shift workers reporting higher stress levels. This is consistent with previous evidence demonstrating an association between perceived stress and metabolic dysfunction [19,21]. A meta-analysis of 17 studies showed that self-reported perceived stress was significantly correlated with visceral obesity and lipid parameters of metabolic syndrome [29]. Among shift workers, elevated perceived stress is associated with metabolic syndrome and altered cortisol patterns [30,31]. Notably, perceived stress represents an individual’s subjective appraisal of stress rather than objective stressors or biological markers [32]. While chronic stress induces insulin resistance through hypothalamic-pituitary-adrenal axis activation [13,19] and inflammatory pathways [16,19], perceived stress reflects complex interactions between external demands, coping resources, and psychological appraisal [32].

Sex was not a significant predictor of insulin resistance in the multivariate model (OR = 0.61, p = .208), contrasting with univariate findings where men showed higher insulin resistance prevalence than women (p = .034). This change from bivariate significance to multivariable non-significance occurred after adjusting for BMI, suggesting that sex differences in insulin resistance may be largely mediated by body composition rather than representing an independent sex effect. When BMI was included in the model, the apparent protective effect of women was attenuated, indicating that lower insulin resistance prevalence among women in our sample may primarily reflect differences in adiposity rather than sex-specific metabolic mechanisms.

This finding appears to contradict previous evidence suggesting metabolic vulnerabilities in women shift workers [13,14]. However, most meta-analyses have focused on diabetes incidence rather than insulin resistance [13,21,22], and the progression from insulin resistance to overt diabetes may be influenced by sex-specific factors including beta cell function, hormonal fluctuations, and compensatory mechanisms [33,34]. Several contextual factors warrant consideration. First, occupational sampling characteristics differ substantially across studies: previous research has predominantly examined women-dominated nursing cohorts [22,35] characterized by moderate-to-low-intensity physical activity combined with high cognitive demands, which may create distinct metabolic risk profiles compared with men-dominated industries such as transportation and manufacturing [36]. Our sample (78.1% men) likely reflects different occupational exposures. Second, cross-sectional analyses may underrepresent metabolically vulnerable workers who opt out of shift work schedules due to health challenges [21,22]. Third, the predominantly men composition may have limited statistical power to detect sex-specific interactions. These findings suggest that the relationship between sex and insulin resistance in shift workers is highly context-dependent, potentially reflecting unique occupational, body composition, and methodological factors in our Korean sample.

Physical activity did not emerge as a significant predictor of insulin resistance in our analysis, despite well-established evidence that regular physical activity improves insulin sensitivity through multiple mechanisms including enhanced glucose uptake, improved mitochondrial function, and reduced systemic inflammation [37,38]. This null finding likely reflects our study population characteristics and cross-sectional design features. Over half of our sample (51.7%) worked in manual labor occupations, whose high baseline occupational physical activity may have created a ceiling effect, diminishing the measurable additional metabolic benefit of leisure-time exercise. Additionally, our binary assessment of leisure-time physical activity did not capture total physical activity exposure (occupational plus leisure-time), which may be particularly important in populations with high occupational activity. This measure also did not differentiate between occupational and leisure-time physical activity—which have distinct metabolic effects [39]—nor did it capture variations in exercise intensity, duration, type, or consistency that influence insulin sensitivity. The cross-sectional design cannot assess temporal patterns such as exercise consistency or duration of physical activity habits, both critical for metabolic adaptations. We initially considered converting physical activity data to metabolic equivalents (METs) for more precise quantification, but substantial missing data and logical inconsistencies in the activity frequency and duration variables would have compromised the validity of such calculations. Therefore, we maintained the binary physical activity variable despite its limitations. Future research employing objective physical activity measures (such as accelerometry) and detailed assessments distinguishing occupational from leisure-time activity may better elucidate the relationship between physical activity patterns and insulin resistance in shift workers.

Cancer and arthritis showed no significant association with insulin resistance, likely reflecting the subclinical nature of insulin resistance in our relatively young working population and insufficient exposure duration for chronic disease development [4,6,40]. Insulin resistance progresses through an extended asymptomatic phase, during which affected individuals may report good health despite objective metabolic dysfunction [4]. Studies have demonstrated that insulin resistance is associated with subclinical vascular changes and precedes symptomatic disease by years, dissociating laboratory-detected abnormalities from perceived health status [40]. These findings underscore that the absence of health complaints does not exclude insulin resistance, supporting the rationale for objective metabolic screening in high-risk occupational populations.

This study has several notable strengths, including the use of the nationally representative KNHANES data with a complex sampling methodology, which ensures generalizability to Korean shift workers across diverse occupations [23]. We included multiple occupational sectors and used weighted multivariable logistic regression analyses to account for the complexity of the survey design. Most notably, the inclusion of metabolic syndrome as a comprehensive covariate allowed us to effectively control for the intricate and overlapping influences of individual metabolic components on insulin resistance. The use of HOMA-IR has enabled practical assessment in a large epidemiological study [1].

Nevertheless, this study has some significant limitations. The cross-sectional design precludes causal inferences and establishment of temporal relationship establishment between predictors and insulin resistance. The absence of a non-shift worker comparison group limits our ability to determine whether the observed 28.5% prevalence—higher than Korean adult averages—specifically results from shift work. Future comparative studies with detailed shift work exposure characterization (shift type, duration, rotation patterns, cumulative exposure) are needed to clarify independent metabolic effects. Our assessment of perceived stress relied on a single-item self-report measure that may not capture multidimensional stress constructs and is subject to recall bias. Quality-of-life instruments may have insufficient sensitivity for detecting subclinical metabolic dysfunction. Additionally, self-reported cancer and arthritis data may have underestimated the true prevalence. Notably, KNHANES lacked information on chronic liver and kidney diseases, both of which can significantly influence metabolic status and insulin resistance [41]. Future studies should incorporate comprehensive liver and kidney function screening. Physical activity assessment was limited to binary self-report measures that inadequately capture activity complexity, intensity variations, or occupational versus leisure-time distinctions. Data quality issues prevented sophisticated MET-based quantification. Residual confounders from unmeasured variables—detailed dietary patterns, sleep quality/duration, specific shift schedules, and shift work duration—may exist. Important metabolic risk factors including nutritional intake and sleep duration were not assessed due to data availability constraints in KNHANES: sleep assessment instruments differed between 2019 and 2021, and nutritional data lacked consistency across survey years, preventing meaningful cross-year integration. This reflects inherent characteristics of secondary data analysis using previously collected survey data, where variable selection is bound by available measures rather than prospective study design. Finally, findings may have limited generalizability beyond Korean populations, and the predominantly men sample may limit sex-specific insights.

This nationally representative study identified insulin resistance prevalence of 28.5% among Korean shift workers, with higher BMI, metabolic syndrome, and elevated perceived stress as independent predictors. The independent effects of BMI and metabolic syndrome underscore that both general adiposity and metabolic dysregulation warrant clinical attention, suggesting practical surveillance strategies: routine BMI monitoring for accessible self-assessment and periodic comprehensive metabolic syndrome evaluation regardless of BMI status. The significant association between perceived stress and insulin resistance highlights the importance of addressing psychological factors alongside metabolic risk. However, important limitations must be acknowledged: the absence of detailed shift work exposure data and a non-shift worker comparison group preclude definitive conclusions about shift work's specific contribution, while the cross-sectional design limits causal inferences. Future longitudinal studies with detailed shift work characterization, comparison groups, objective physical activity measurement, and comprehensive health screening are needed to clarify causal pathways and develop targeted interventions for this occupational population.