Document resource
WHAT IS ALREADY KNOWN ON THIS TOPIC Emerging evidence suggests that psychological stress may contribute to glucose dysregulation; however, findings are inconsistent, and the influence of demographic factors such as age and gender is unclear.WHAT THIS STUDY ADDS By examining perceived helplessness and perceived self-efficacy as distinct dimensions of perceived stress, this study demonstrates that gender significantly moderates their associations with prediabetes and glycemic measures.In men, higher levels of both perceived stress factors are associated with lower prediabetes prevalence and lower 2-hour postload glucose levels.Among women, age further modifies the impact of perceived helplessness: younger women with higher perceived helplessness have a higher prevalence of prediabetes and higher 2-hour postload glucose, associations not observed in older women.HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY These findings suggest that considering age and gender in psychosocial stress management strategies could improve prediabetes prevention efforts.Further research is warranted to evaluate the effectiveness of tailored interventions in reducing prediabetes across demographic groups.Introduction Diabetes is one of the leading causes of death and disability, currently affecting an estimated 6.1% of the global population, or approximately 529 million people, and posing a significant burden on healthcare systems worldwide. 1 One of the primary strategies for addressing the rapidly rising prevalence of diabetes is prevention and early identification to avoid disease progression. Prediabetes, characterized by glycemic levels above normal but below the diagnostic threshold for diabetes, strongly predicts progression to diabetes within 5 years.2 Targeting prediabetes therefore presents a critical opportunity for intervention, particularly through addressing modifiable behavioral and lifestyle risk factors, to reduce the incidence of diabetes.Several modifiable risk factors of prediabetes have been identified, including diet, physical activity, and smoking.3 A growing body of research suggests that psychological stress may also be a risk factor for dysregulation of glucose metabolism.4 Stress activates the hypothalamic-pituitary-adrenal (HPA) axis, leading to elevated cortisol levels, which promote energy mobilization through stimulating gluconeogenesis and lipolysis, resulting in increased circulating glucose and lipid levels. Additionally, cortisol impairs peripheral glucose uptake by reducing the translocation of glucose transporter type 4 to the cell membrane in muscle and adipose tissues, thereby contributing to insulin resistance. Over time, these metabolic alterations contribute to glucose dysregulation and insulin resistance, and the development of prediabetes and type 2 diabetes mellitus.5While some studies have demonstrated an association between increased psychological stress—including perceived stress and work-related stress—and the incidence of diabetes,4 6–10 the findings remain inconsistent.4 Notably, the relationship between stress and glucose metabolism appears to vary by sex, with some studies showing increased stress associated with increased abnormal glucose metabolism and risk of diabetes in women, but not men.11–13 Furthermore, the evidence for a stress-glucose metabolism association in prediabetic or healthy populations is limited and inconclusive.14–17 Understanding how stress affects glucose regulation during this potentially reversible state may offer opportunities to enhance early intervention and prevention strategies. Importantly, age may also play a moderating role in the stress-glucose relationship, particularly among women, as aging is associated with changes in HPA axis reactivity, cortisol regulation, and sex hormone levels.5Therefore, this study aimed to examine the association between perceived stress and glycemic measures in a diabetes-free population, including individuals with normoglycemia and prediabetes, and to assess whether this relationship is moderated by gender, followed by an examination of age as a moderator within the women and men subgroups, respectively.Research design and methods Study design and participants This study used cross-sectional data collected during the years 2020–2021 from a subset of participants in a broader investigation of diabetes risk factors in a healthy multiethnic population in Singapore, as described elsewhere. 18 19 Participants with a history or diagnosis of diabetes—defined as having at least two concurrent test results exceeding the American Diabetes Association (ADA)20 diagnostic cut-offs (ie, 2h-PG ≥11.1 mmol/L, fasting plasma glucose FPG) ≥7.0 mmol/L, and glycated hemoglobin (HbA1c) ≥6.5%)—were excluded. Thus, all our participants either had normoglycemia or prediabetes. Of the 472 participants who initially completed their perceived stress survey, two were excluded due to missing glycemic data for that visit, resulting in a final sample of 470 participants.Procedures Participants attended the study visit in the morning after fasting overnight for 10–12 hours. Their body weight and height were measured by trained research staff. Subsequently, fasting blood samples were collected to assess HbA1c and FPG. Next, participants underwent an oral glucose tolerance test (OGTT) by ingesting a 75 g glucose solution, with additional blood samples collected 2 hours later to evaluate postprandial glucose levels (ie, 2-hour post-OGTT plasma glucose (2h-PG)). On the same day, participants also completed an electronic questionnaire assessing their perceived stress.Perceived Stress Scale (perceived helplessness and perceived self-efficacy) Participants’ perceived stress over the past month was assessed using the 10-item Perceived Stress Scale (PSS) 21 22 through self-administered survey questionnaires. The PSS is widely used and demonstrates good validity and reliability across many countries,23 including Singapore.24 25 Example items include: “How often have you felt nervous and stressed?” and “How often have you been able to control irritations in your life?” Responses were recorded on a 5-point Likert scale ranging from 1=never to 5=very often.Several studies have demonstrated that PSS is best represented by a two-factor structure comprising perceived helplessness (ie, general negative responses to stress, six items) and perceived self-efficacy (ie, perceived ability to manage stress-inducing factors, four items).23 25–27 Therefore, we conducted a comparative confirmatory factor analysis (CFA) to evaluate the model fit using the recommended thresholds: comparative fit index (CFI) score ≥0.90 and a standardized root mean square residual (SRMR) score <0.10.28 The CFA supported the two-factor model (CFI=0.96, SRMR=0.05) rather than the one-factor model (CFI=0.58, SRMR=0.19), and a χ2 difference test further confirmed the superior fit of the two-factor model (χ2(1)=873.00, p<0.001). Internal consistency was high for both subscales (perceived helplessness: α=0.88; perceived self-efficacy: α=0.87). Accordingly, we used the average item scores of perceived helplessness and perceived self-efficacy subscales in subsequent analyses.Outcomes of glycemic measures and prediabetes FPG, 2h-PG and HbA1c were assessed using standard methods as previously described. 19 Prediabetes status was defined per the ADA criteria,20 as meeting any of the following: (i) 2h-PG: 7.8–11.0 mmol/L, (ii) FPG: 5.6–6.9 mmol/L, and (iii) HbA1c: 5.7%–6.4%.Covariates and moderator variables Body mass index (BMI) was calculated as weight in kilograms divided by height in meters squared (kg/m²), based on in-person physical assessments. Age was computed as the difference between each participant’s birth year and the visit year. Participants self-reported their ethnicity (Chinese, Malay, Indian, or others) and gender (male or female).Statistical analyses All analyses were conducted in STATA V.18. Prediabetes status and glycemic measures were regressed on continuous measures of the stress factor of perceived helplessness and perceived self-efficacy. Logistic regression was applied to categorical outcomes, and ordinary least squares regression was used for continuous outcomes. All models controlled for BMI, ethnicity, age, and gender. To test moderation by gender, interaction terms (gender×perceived helplessness and gender×perceived self-efficacy) were included. Moderation by age was examined by including the interaction terms (age×perceived helplessness and age×perceived self-efficacy) in gender-stratified models.Results Participant characteristics The sample consisted of 470 participants (57% women) with a mean age of 48.7 years (SD=9.9). Our study population was multiethnic, including Chinese (64%), Malay (14%), Indian (17%), and other ethnicities (6%). On average, participants reported a perceived helplessness score of 2.32 (0.70) and a perceived self-efficacy score of 3.45 (0.91). Based on the ADA diagnostic criteria, 51% had normoglycemia and 49% met the criteria for prediabetes (see online supplemental table 1 for details).SP110.1136/bmjdrc-2025-005368.supp1Supplementary dataAssociations between perceived stress factors and glycemic measures Table 1 presents the main effects of perceived helplessness and perceived self-efficacy on glycemic measures and prediabetes (models 1–4). Only higher perceived helplessness was significantly associated with lower FPG (model 3; B=−0.07, p=0.040). No other significant associations were observed between the stress factors and glycemic outcomes (all ps>0.05).Table 1Associations between perceived stress and glycemic measures (models 1–4) and gender as a moderator (models 5–8)All study participantsDependent variable predictorsModel 1PrediabetesModel 22h-PGModel 3FPGModel 4HbA1cMain effects Perceived helplessness−0.07−0.09−0.07*−0.03 Perceived self-efficacy−0.12−0.18−0.020.00 R20.110.140.14 Pseudo-R20.06 F6.55***9.23***9.18*** χ237.20***Dependent variable predictorsModel 5PrediabetesModel 62h-PGModel 7FPGModel 8HbA1cInteraction terms Perceived helplessness×gender0.72*0.62*0.060.02 Perceived self-efficacy×gender0.44*0.50**0.000.01 R20.130.140.14 Pseudo-R20.07 F6.63***7.45***7.36*** χ247.38****p<0.05, **p<0.01, ***p<0.001 (two-tailed). The table indicates unstandardized regression coefficients. Model 1 or 5 presents the results of logistic regression, whereas models 2–4 or 6–8 present the results of ordinary least squares regression. For brevity, additional independent variables were not reported in models 1–4 (ie, BMI, age, ethnicity, gender) or in models 5–8 (ie, BMI, age, ethnicity, gender, perceived helplessness, and perceived self-efficacy).BMI, body mass index; FPG, fasting plasma glucose; HbA1c, glycated hemoglobin; 2h-PG, 2-hour plasma glucose.Gender as a moderator of stress-glycemic associations In table 1, we found that gender significantly moderated the associations between the stress factors (perceived helplessness and perceived self-efficacy) and both prediabetes status and 2h-PG levels. Significant interaction terms were observed for prediabetes (model 5: perceived helplessness, B=0.72, p=0.014; perceived self-efficacy, B=0.44, p=0.047) and 2h-PG (model 6: perceived helplessness, B=0.62, p=0.016; perceived self-efficacy, B=0.50, p=0.009). No significant moderating effects of gender were observed for FPG or HbA1c.To further interpret these interactions, we stratified the sample by gender and plotted predicted values at 1 SD above and below the mean of the perceived stress factors, following Aiken and West.29 As shown in figure 1, the direction and strength of the associations between the perceived stress factors and the glycemic measures differed by gender. Moreover, simple slope analyses based on marginal estimated values30 indicated that among men, higher perceived helplessness was significantly associated with a lower prevalence of prediabetes (B=−0.59, p=0.012) and lower 2h-PG levels (B=−0.46, p=0.027). In contrast, these associations were non-significant for women (prediabetes: B=0.25, p=0.181; 2h-PG, B=0.14, p=0.400). Similarly, higher perceived self-efficacy was significantly associated with a lower prevalence of prediabetes (B=−0.36, p=0.037) and lower 2h-PG levels (B=−0.47, p=0.002) for men. However, these associations were not significant for women (prediabetes: B=0.05, p=0.723; 2h-PG, B=0.06, p=0.666). Taken together, these results suggest that both perceived helplessness and perceived self-efficacy were significantly associated with lower prediabetes prevalence and 2h-PG levels among men, but not among women.Figure 1Gender as a moderator for the association between perceived stress and glycemic measures. An asterisk (*) indicates a statistically significant association (p<0.05). The straight (or dashed) line indicates the strength and direction of the association between perceived helplessness/self-efficacy and a dependent variable for men or women. The steeper slope of the line indicates a more significant association. The figures indicate that higher perceived helpfulness and control were significantly associated with a lower prevalence of prediabetes and lower levels of 2h-PG following OGTT for men, but these associations were non-significant for women. 2h-PG, 2-hour plasma glucose; OGTT, oral glucose tolerance test.Age as a moderator of stress-glycemic associations within gender We examined the moderating effects of age separately for women and men. The corresponding results are presented in table 2 and online supplemental table 2, respectively. For women (table 2), age significantly moderated the association between perceived helplessness and two glycemic outcomes. Specifically, the interaction between age and perceived helplessness was significant for prediabetes status (model 1: B=−0.04, p=0.030) and 2h-PG levels (model 2: B=−0.06, p<0.001). Moreover, age did not significantly moderate the associations between perceived self-efficacy and all the glycemic outcomes for women (interaction terms all ps>0.05).Table 2Age as a moderator of perceived stress-glycemic associations among womenDependent variablepredictorsModel 1PrediabetesModel 22h-PGModel 3FPGModel 4HbA1cInteraction terms Perceived helplessness×age−0.04*−0.06***−0.010.00 Perceived self-efficacy×age0.010.010.000.00 R20.110.150.23 Pseudo-R20.10 F3.25***5.24***8.25*** χ235.40****p<0.05, ***p<0.001 (two-tailed). The table indicates unstandardised regression coefficients. Model 1 presents the results of logistic regression whereas models 2–4 present the results of ordinary least squares regression. For brevity, additional independent variables were not reported in models 1–4 (ie, body mass index, ethnicity, perceived helplessness, perceived self-efficacy, and age).FPG, fasting plasma glucose; HbA1c, glycated hemoglobin; 2h-PG, 2-hour plasma glucose.Figure 2 illustrates the conditional associations between perceived helplessness and both prediabetes status and 2h-PG levels, using the margins method30 with 95% CIs. These estimates are based on the Johnson-Neyman technique31 and span the full age range of female participants in our sample (32–74 years; see also Piszczek and Pimputkar32). We found that perceived helplessness was significantly associated with a higher prevalence of prediabetes among women aged 32–47 years (ie, all 95% CIs excluded zero or ps<0.05), but this association became non-significant for women aged 48–74 years. Similarly, perceived helplessness was significantly associated with higher 2h OGTT PG levels among women aged 32–46 years (all 95% CIs excluded zero or ps<0.05), was non-significantly associated with 2h-PG levels among women aged 47–61 years, and was significantly associated with lower 2h-PG levels among women aged 62–74 years. Jointly, these results supported that higher perceived helplessness was more associated with a higher prevalence of prediabetes and higher 2h-PG levels for younger women.Figure 2Age as a moderator for women. The solid line represents changes in the outcome variables based on a one-unit increase in perceived helplessness across different ages, whereas the two dotted lines indicate the 95% CIs for these effects. When the 95% CIs did not include zero at certain ages, the associations between perceived helplessness and its outcome variables were significant (p<0.05). Thus, specific age ranges were identified where these significant associations occurred. The figures indicate that higher perceived helplessness was significantly associated with higher prediabetes and higher levels of 2h-PG post-OGTT for younger women. Specifically, higher perceived helplessness was significantly associated with a higher prevalence of prediabetes for women aged 32–47 years and higher levels of 2h-PG post-OGTT for women aged 32–46 years. Moreover, higher perceived helplessness was significantly associated with lower levels of 2h-PG post-OGTT for women aged 62–74 years. 2h-PG, 2-hour plasma glucose; OGTT, oral glucose tolerance test.By contrast, age did not significantly moderate the associations between perceived stress factors and glycemic outcomes among men (see online supplemental table 2); all interaction terms involving age were non-significant (all ps>0.05).Discussion Understanding the associations between perceived stress and glycemic measures—and how these associations vary by gender and age—is important for developing targeted strategies to prevent the onset and progression of prediabetes. In our study, higher levels of both perceived helplessness and perceived self-efficacy were associated with a lower prevalence of prediabetes and lower 2h-PG levels among men. A similar pattern was observed among older women, with higher perceived helplessness linked to lower 2h-PG levels. In contrast, among younger women, higher perceived helplessness was associated with a higher prevalence of prediabetes and elevated 2h-PG levels.Although previous studies have highlighted the negative consequences of perceived helplessness, such as poorer health outcomes,25 online game addiction,33 and heightened anxiety and depression,34 our findings suggest that perceived helplessness is associated with lower prediabetes prevalence and reduced 2h-PG levels. This finding aligns with prior research showing that higher work stress is linked to a reduced risk of type 2 diabetes in men, but not in women.12 A similar pattern emerged among older women (aged 62–74 years), for whom higher perceived helplessness was also associated with lower 2h-PG levels. One possible explanation is that perceived helplessness may prompt greater engagement in adaptive coping behaviors, such as physical activity. Supporting this notion, prior research has found that among older adults, the number of stressors is positively associated with engagement in physical activity.35 Thus, in older individuals, perceived helplessness may act as a motivational cue, regardless of gender, encouraging health-promoting behaviors that help regulate glucose levels.Our findings contribute to the literature on gender differences in stress-related health outcomes by clarifying how specific stress components interact with age to influence glycemic regulation in women. Previous studies have reported that increased work stress is associated with a higher risk of type 2 diabetes in women but not in men,11 12 with explanations pointing to women’s greater physiological reactivity to stress, such as elevated cortisol levels.12 Extending this work, we found that perceived helplessness, but not perceived self-efficacy, was significantly associated with higher prediabetes prevalence and elevated 2h-PG levels among younger women. This suggests that younger women may be more susceptible to the detrimental effects of perceived helplessness than to the protective effects of perceived self-efficacy. One possible explanation involves hormonal fluctuations across the menstrual cycle, which have been shown to increase stress sensitivity. For instance, stress reactivity intensifies during the luteal phase,36 and younger women exhibit stronger negative responses to a stressful task assignment during the premenstrual phase.37 Taken together, these findings highlight the importance of accounting for both age and the nature of stress perceptions when investigating gender-specific pathways to glycemic dysregulation, and they point to potential boundary conditions for the adverse effects of stress observed in prior research on women’s health.One explanation for the stronger adverse associations between perceived helplessness and glycemic outcomes among younger women, as shown in our research, is that their stress may be less effectively buffered by social support. Relatedly, previous research has demonstrated a significant association between poor social support and 2h post-OGTT glucose levels among women.38 Although social support was not directly assessed in our study, both gendered roles and life stages are likely to influence access to support networks. In Singapore, cultural norms place a disproportionate caregiving burden on women.39 Younger women may therefore receive inadequate support while simultaneously managing the demands of an early career and caregiving responsibilities, such as childcare and eldercare. By contrast, older women often have more established social networks and have fewer childcare concerns as they age. Consequently, younger women may be more vulnerable to the adverse physiological consequences of stress than men or older women.We also found significant associations between perceived stress and glycemic outcomes when using prediabetes status and 2h-PG levels as outcome variables, but not when using FPG or HbA1c. These measures capture different aspects of glucose metabolism and may reflect gender differences in metabolic responses to stress. Research indicates that men, under stress, use blood glucose more efficiently than women, achieving a higher glucose influx rate—the rate at which glucose is transported from the bloodstream into cells for energy use.40 This dynamic process is better captured by the 2h-PG, which reflects the body’s capacity to manage a glucose load and how effectively glucose is taken up by peripheral tissues such as muscle. In contrast, FPG and HbA1c reflect more static or chronic aspects of glucose regulation: FPG is influenced by hepatic glucose output and insulin function during fasting, while HbA1c indicates average glucose levels over the past 2–3 months. The OGTT, from which 2h-PG is derived, provides a more sensitive and dynamic assessment of glucose regulation and is particularly useful for detecting prediabetes and predicting type 2 diabetes risk.41 Notably, few psychological stress studies incorporate OGTT due to its time and cost demands, making its inclusion a strength of our study. This is especially relevant in a diabetes-free population, where OGTT can reveal early dysregulation not detectable by FPG or HbA1c, helping to uncover stress-glucose associations that may be overlooked in prior research.Moreover, our findings suggest that gender-specific interventions incorporating psychosocial strategies and lifestyle modifications for diabetes prevention may be beneficial. Prior research highlights gender differences in stress-related behaviors: men are more likely to engage in physical activity to cope with stress,42 whereas women tend to consume more unhealthy food,43 and poor diet quality is more strongly linked to prediabetes risk in women than in men.44 Future studies could include behavioral measures, such as exercise frequency and unhealthy food intake, to examine how these behaviors mediate the relationship between perceived stress and glycemic regulation, thereby informing more effective interventions.Our age-specific findings add nuance to the stress-glycemia link. That only perceived helplessness relates to glycemic outcomes in younger women suggests emotional distress may be more influential than self-efficacy in this group. A recent survey found that stress levels have risen, particularly among younger women,45 underscoring the need to target helplessness in stress-reduction efforts. Future research could explore strategies such as reframing distress as a sign of passion.46Limitations and future research directions Several limitations of our study point to avenues for future research. First, the cross-sectional design limits causal inference and may inflate associations due to the temporal proximity of measures. However, we mitigated this concern by using distinct types of measures—self-reported stress as predictors and biological indices as outcomes. 47 Future studies could adopt longitudinal designs to clarify the directionality of the observed associations.Second, participants’ perceived helplessness levels were mostly within the low-to-moderate range on a 5-point scale (1=never, 2=almost never, 3=sometimes; 25th percentile=1.83, median=2.33, 75th percentile=2.83). Prior research suggests that moderate stress may lower blood glucose, whereas extremely high stress may elevate it.48 This may explain the inverse association between perceived helplessness and glycemic measures observed in men. Future research could examine whether moderate and high levels of perceived helplessness differentially affect glycemic outcomes.Third, we assessed participants’ stress using self-reported ratings, which may be subject to bias. For instance, participants might under-report their stress levels to maintain a positive self-image or to conform to social desirability norms. Our findings may also reflect reference bias,49 in which individuals evaluate the same item against differing implicit standards. For example, males and females may adopt different benchmarks for what qualifies as ‘stressful’, potentially leading to divergent associations between self-reported stress and specific outcome variables. To strengthen validity, future studies could incorporate biological measures of stress (eg, cortisol tests) to examine the replicability of the present findings.Fourth, our data were collected during the COVID-19 pandemic and exclusively from a Singaporean sample. Stress and social dynamics were atypical in 2020–2021 owing to lockdowns, remote working, and other pandemic-related disruptions. As such, the present findings might reflect the characteristics of this sample, including its demographic composition, cultural context, and the unique timing of data collection. These factors might have influenced the observed associations between stress and glycemia. To strengthen the generalizability of these results, future research could examine postpandemic populations and include samples from diverse cultural settings.Finally, our findings underscore the importance of identifying potential mediators linking perceived stress to glycemic regulation. For instance, gender differences in metabolic responses to stress or hormonal fluctuations across phases of the menstrual cycle in women warrant further investigation. Although both perceived helplessness and perceived self-efficacy were associated with lower 2h-PG levels in men, these associations may operate through distinct mechanisms. For example, perceived helplessness might reduce appetite, whereas perceived self-efficacy may promote engagement in physical activity. Future research should examine these behavioral and physiological pathways to clarify how different components of perceived stress influence glucose metabolism.Conclusion Our study offers new insights into the relationship between stress and blood glucose by examining specific stress factors, glycemic measures, and the moderating roles of gender and age. The findings suggest that moderate levels of perceived stress may benefit men’s health by reducing the risk of prediabetes and lowering 2h-PG levels. While prior research highlights the adverse effects of stress on women, our results add nuance—showing that such effects, particularly those linked to perceived helplessness, are more pronounced in younger women. Future research should explore age-specific and gender-specific interventions for diabetes and prediabetes prevention. In doing so, it will be important to use longitudinal designs, assess a broader range of stress levels, and examine behavioral and physiological mediators—such as physical activity, dietary patterns, and hormonal responses—to better understand how different components of perceived stress influence glucose regulation across demographic groups.