Cardiac cost of occupational stress among obstetricians: a cross-sectional non-invasive haemodynamic study in a tertiary hospital Authors: Xiaoling Zhong, Dan Li, Shaoyang Lai, Robert A Phillips, Quanfeng Wu, Wei Wei, Meilin Yang Published: 2026 ABSTRACT Abstract AbstractObjectivesTo identify early cardiovascular changes in normotensive obstetricians, a high-stress group, using non-invasive haemodynamic monitoring, and to examine the association between burnout and haemodynamic parameters.DesignCross-sectional study.SettingA single tertiary hospital in China.ParticipantsA total of 120 healthy Han Chinese adults (aged 25–45 years, both sexes) were enrolled using stratified random sampling by age and categorised into three groups: obstetricians, clinical support staff and administrative personnel. Of these, 105 (87.5%) completed the study and entered the final analysis (obstetricians n=40; clinical support staff n=33; administrative personnel n=32); 15 were excluded due to incomplete questionnaire data. Key exclusion criteria were chronic medical conditions, medication use, acute illness, a clinical shift within 24 hours before measurement, pregnancy or lactation, body mass index extremes (≤18.5 or ≥ 28 kg/m2) and major life events within the past 6 months. Burnout was assessed using the Maslach Burnout Inventory-Human Services Survey; workload, lifestyle and family history were collected via questionnaire.Primary and secondary outcome measuresAdvanced haemodynamics were assessed via Ultrasonic Cardiac Output Monitor. The primary outcome was cardiac power output (CPO). Secondary outcomes included other non-invasive haemodynamic parameters, such as cardiac index (CI), systemic vascular resistance index, Smith–Madigan inotropy index and corrected flow time.ResultsSevere burnout was associated with reduced CPO and CI (−0.152 W and −0.403 L/min/m2, respectively; both p<0.05). Increased weekly exercise independently predicted higher CPO and CI (B=0.046 W and 0.155 L/min/m2, respectively; both p<0.001). This protective association remained significant in obstetricians and clinical support staff, and was stronger for CI in obstetricians (interaction p=0.03). The severe burnout–haemodynamics association was consistent across all groups.ConclusionsIn high-stress populations, advanced haemodynamic patterns may serve as an early-warning biomarker for burnout, guiding personalised exercise advice. Longitudinal studies are needed to confirm their predictive value.Keywords: Burnout, Cross-Sectional Studies, Physicians, Cardiology STRENGTHS AND LIMITATIONS OF THIS STUDY Haemodynamic assessment was conducted using non-invasive methods, which are easy to implement and well tolerated by participants. Haemodynamic assessment was conducted using non-invasive methods, which are easy to implement and well tolerated by participants. The cohort, free of major comorbidities, provided a clean sample for detecting subclinical changes. The cohort, free of major comorbidities, provided a clean sample for detecting subclinical changes. The single-centre design reduced population heterogeneity but may limit external validity. The single-centre design reduced population heterogeneity but may limit external validity. Analyses used multivariable linear regression with interaction testing to control for confounders. Analyses used multivariable linear regression with interaction testing to control for confounders. We did not measure sleep or detailed dietary intake, which may have led to residual confounding. We did not measure sleep or detailed dietary intake, which may have led to residual confounding. Introduction Physician burnout is a pervasive global crisis that compromises clinician well-being and patient safety.1,3 In China, fractious doctor–patient relationships and extreme work patterns intensify this issue.4 5 Evidence links prolonged work hours (≥ 55 per week) to elevated cardiovascular and metabolic risks,6 with excessive workload even implicated in sudden death among young physicians aged 30–39 in China.5 Mortality clusters in surgeons with regional variation in China, with Beijing reporting the highest incidence, followed by Fujian Province, which records a mortality rate of approximately 7.41%.7 8 While burnout has been studied among various clinical specialties, such as emergency medicine,9 anaesthesiology10 and paediatrics,11 obstetricians face particularly acute pressures. These include high patient expectations, critical emergencies (eg, shoulder dystocia), dual responsibility for maternal and fetal outcomes, frequent medicolegal disputes,12 unpredictable schedules due to the nature of childbirth13 and career uncertainties associated with China’s declining birth rate.14 Given the critical role of obstetrics, addressing their occupational well-being is critically important. Chronic stress may induce subclinical cardiovascular dysfunction. haemodynamic parameters such as systemic vascular resistance (SVR) and cardiac power output (CPO) are well-established risk markers15 16 ; notably, a 0.2 W increase in CPO correlates with a 45% mortality risk reduction in critical illness.16 Importantly, during early stages of stress-related pathophysiology,17 baroreceptor-mediated compensatory mechanisms act to preserve stable blood pressure (BP), while parameters like stroke volume (SV) and SVR may already show abnormalities.18 This suggests that haemodynamic profiling may detect burnout-related impairment early. The Ultrasonic Cardiac Output Monitor (USCOM) provides a non-invasive, accurate option for such assessment.19 20 Although widely used in clinical settings,21,23 its application to healthy individuals under occupational stress remains unexplored. Accordingly, the objectives of our study were (1) to identify early cardiovascular changes in normotensive obstetricians (a high-stress group) using non-invasive haemodynamic monitoring and (2) examine the association between burnout and haemodynamic parameters. Methods Study design A cross-sectional study was conducted at our institution from December 2024 to January 2025. As the regional highest-volume delivery centre and a municipal referral hub for critical obstetrics, our hospital performed 13 819 deliveries in 2024, attended by 45 obstetricians. Inclusion and exclusion criteria The inclusion criteria were as follows: (1) full-time and formally contracted staff; (2) age between 25 and 45 years and (3) voluntary participation with signed informed consent. The exclusion criteria were (1) a known history of hypertension, diabetes mellitus, dyslipidaemia, cardiovascular disease (CVD) or malignancy; (2) use of any prescription medication currently; (3) presence of acute infectious conditions (eg, common cold, diarrhoea); (4) having taken a clinical shift within the 24 hours prior to measurement to avoid acute fatigue; (5) current pregnancy or lactation; (6) body mass index (BMI)≥28 or ≤ 18.5 kg/m224 and (7) experiencing a major life event (eg, bereavement, divorce) within the past 6 months. Additionally, on test day, participants abstained from caffeinated beverages≥2 hours before haemodynamic assessment. Assessments were conducted daytime (9:00–18:00). Groups Participants were categorised into three groups based on their job function and level of patient contact: (1) obstetricians group: clinicians who provide direct and comprehensive obstetric care; (2) clinical support staff group: healthcare professionals who provide supportive medical services, including obstetric nurses, pharmacists and other allied healthcare providers; (3) administrative personnel group: non-clinical hospital staff, such as information department staff. Sample size calculation Given the lack of prior studies on haemodynamic parameters in healthcare workers, sample size was calculated from a pilot study (n=10 per group) based on CPO, a critical haemodynamic parameter reflecting myocardial contractility and systemic perfusion,16 using ANOVA with α=0.05 and power=80%. CPO measurements (mean±SD) for each group were as follows: obstetricians group: 0.79±0.08 W; clinical support staff group: 0.85±0.15 W; administrative personnel group: 0.91±0.11 W. Between-group variability was quantified using using η², calculated as where SSB (between-group sum of squares) and SST (total sum of squares) were derived from pilot data. For CPO, η² ≈ 0.16. Using the ANOVA sample size formula25: with Zα/2 = 1.96, Zβ = 0.84 and k=3 (groups), the total sample size was 88. Due to the unique nature of hospital operations (eg, conference attendance, visiting scholar), we anticipated 20% of participant dropout. The minimum required sample size was calculated as 106. This pilot study has not been published or deposited on a preprint server. Sampling A stratified random sampling method was employed to ensure representative sampling across age groups (25–30, 31–35, 36–40 and 41–45 years). To achieve balanced allocation across the three groups and four age strata (10 participants per stratum per group), the final target sample size was set to 120. Random selection within each stratum was performed using SPSS Statistics (V.27.0, IBM Corp, Armonk, New Y, USA). Haemodynamic parameters measurement Haemodynamic assessment was performed non-invasively using the USCOM (USCOM-1A, USCOM Ltd., Sydney, Australia) and anthropometrically calibrated Doppler cardiac output monitor with validation against surgically implanted transonic flow probes and pulmonary artery catheters across a sixfold range of pharmacotherapeutically altered outputs20 and in humans across 0.13 L/min to 18.7 L/min range of outputs.26 27 Measurements were conducted by three trained operators on supine participants following 5 min of rest. Parameters were measured three times consecutively with mean values were used for final analysis. Transaortic acoustic access was preferred to obtain the trans-aortic velocity–time integral (VTI) and flow time (FT). Using these primary measurements, contemporaneous BP and body surface area (BSA) were integrated using proprietary software to automatically generate advanced haemodynamic parameters. Parameters were categorised into three groups: fundamental flow volumes including SV, SVI, CO and cardiac index (CI); indices of afterload SVR and SVR indexed to BSA, SVRI; and sophisticated metrics of cardiovascular function and efficiency. The latter comprised the stroke work (SW), mean pressure gradient (Pmn), corrected flow time (FTc), potential kinetic energy ratio (PKR), myocardial contractility using the Smith–Madigan Inotropy Index (SMII) and CPO.28 The primary outcome was CPO. All other haemodynamic parameters listed above were secondary outcomes. Questionnaire administration The Maslach Burnout Inventory-Human Services Survey (MBI-HSS) and a self-designed questionnaire on family history, workload and lifestyle were completed using WenJuanXing after haemodynamic assessment (onlinesupplemental information S1 S2). Electronic informed consent was obtained prior to data collection. To maximise completion rates, incomplete questionnaires triggered two weekly reminders. Non-respondents were those failing to respond after the second reminder. Confidentiality was ensured by storing data in an anonymised online database without personal identifiers. Assessment of burnout Burnout was assessed using the Chinese version of the MBI-HSS, validated in Chinese healthcare workers.29 30 The MBI-HSS is a 22-item scale measuring burnout across three subscales: emotional exhaustion (EE, nine items), depersonalisation (DP, five items) and reduced personal accomplishment (PA, eight items).31 Participants rated the frequency of their feelings on a 7-point Likert scale ranging from 0 (never) to 6 (every day). Each subscale has its unique level categories sorted from ‘low’ (EE≤16; DP≤6; PA≤31), ‘moderate’ (EE=17–26; DP=7–12; PA=32–38) and ‘high’ (EE≥27; DP≥13; PA≥39). Participants were classified as burnout if they scored high on either the EE subscale and/or the DP subscale, as widely adopted in the literature.32 33 In our study, burnout was subdivided into two groups: (1) ‘severe burnout’ (high levels on both EE and DP), and (2) ‘moderate burnout’ (high level on either EE or DP). Those scoring below thresholds on both subscales were classified as non-burnout. Scores for the PA were reported separately. Exposure variables To identify potential co-factors associated with haemodynamic variables, we collected exposure variables through a self-reported questionnaire (online supplemental information S2): (1) family history of metabolic syndromes (CVD, diabetes and hyperlipemia) or malignancy was assessed for first-degree/second-degree relatives (binary yes/no response for each condition). (2) Workload was evaluated via two metrics: (a) weekly working days which was categorised into an extended workweek group (6–7 days per week) and a standard workweek group (4–5 days per week). (b) Night shift frequency: the number of night shifts per month was collected as a continuous variable. (3) Lifestyle: (a) dietary habits were assessed with a focus on meal regularity, a key concern among hospital staff due to unpredictable clinical demands. They were dichotomised into regular (consistent meal times) and irregular (frequently skipping or delaying meals). (b) Exercise was quantified as the total self-reported duration of moderate-to-vigorous physical activity (MVPA) per week.34 (c) Smoking status was assessed using two questions based on National Health and Nutrition Examination Survey (NHANES). Current smokers were defined as participants who reported having smoked≥100 cigarettes in their lifetime and currently smoking every day or some days. All others were classified as non-smokers. (d) Alcohol consumption was assessed using questions adapted from NHANES. Non-drinkers were defined as participants who reported either fewer than 12 alcoholic drinks in their lifetime, or no alcohol consumption or drinking less than once per month in the past 12 months. All remaining participants were classified as current drinkers. Data analysis Prior to analyses, raw data were anonymised. Statistical processing used SPSS Statistics (V.27.0; IBM Corp, Armonk, New York, USA). Normality (Shapiro–Wilk test) and homogeneity of variance (Levene’s test) were assessed. Continuous data: normally distributed are expressed as mean with SD, non-normally distributed data are reported as median with an IQR. ANOVA was applied if both assumptions were satisfied; otherwise, the Kruskal–Wallis test was used. For post-hoc pairwise comparisons, Bonferroni correction was applied following significant ANOVA results, while Games–Howell test was applied in the instances of significant Kruskal–Wallis test. Categorical variables are summarised as frequencies/percentages. Group differences were determined by χ2 or Fisher’s testing. Significant overall findings were further explored with post-hoc analyses using Bonferroni-adjusted standardised residuals. To identify independent factors associated with haemodynamic parameters, multivariable linear regression (MLR) was used for dependent variables meeting normality and homoscedasticity assumptions. In the MLR models, the following covariates were simultaneously adjusted for: age (continuous), burnout status (moderate or severe, with non-burnout as reference), weekly exercise time (continuous), regular diet (yes/no), extended workweek (6–7 days/week vs 4–5 days/week), monthly night shift frequency (continuous), BMI (continuous), alcohol consumption (alcohol only vs none), smoking status (smoking only vs none), combined smoking and alcohol use (vs none) and family history (metabolic syndromes only; metabolic syndrome and malignancy). For non-normal variables despite log/reciprocal transformation, generalised linear models (GLMs) were applied. Standardised regression coefficients (β) with 95% CI were visualised using forest plots. In exploratory analyses, we performed stratified analyses by occupational group to assess associations within each subgroup and evaluated potential effect modification by adding interaction term (eg, exercise×group) to the models. Analyses used Python (V.3.12.3; Python Software Foundation, Wilmington, Deleware, USA). Patient and public involvement None. Results Participant enrolment Figure 1 shows recruitment profile of the Administrative personnel group from the different departments: information department (n=16), science and education department (n=7), medical affairs department (n=9), medical equipment department (n=15) and quality management department (n=13). The clinical support staff group participants were recruited from obstetric nurses (n=69), anaesthetists (n=40) and pharmacists (n=56). Based on a target sample size of 120, participants were selected via stratified random sampling (40 per group). From these 120 recruits, the final number of complete-case participants was 105 following exclusion of 15 non-respondents (12.5%) despite second reminders. Flowchart of participant selection. Characteristics of participants across groups Baseline characteristics were comparable across groups (table 1). All participants were of Han Chinese ethnicity. Comparisons revealed no intergroup differences in age, BSA, systolic blood pressure or diastolic blood pressure. However, haemodynamic assessments revealed key cardiac function indices of SMII, SW and CPO were significantly lower in both the clinical support staff (SMII: p<0.001; SW: p=0.002; CPO: p=0.007) and obstetricians (SMII: p<0.001; SW: p=0.011; CPO: p=0.039) groups compared with the administrative personnel group. No differences were observed between the clinical support staff and obstetricians groups. Table 1Baseline characteristics of participants across groupsVariableAdministrative personnel group (n=32)Clinical support staff group (n=33)Obstetricians group (n=40)P valueAge (years)*37.5 (11.0)36.0 (9.0)37.5 (10.0)0.834Male, n (%)5 (15.63)5 (15.15)4 (10.0)0.769BSA (m2)†1.7 (0.2)1.6 (0.1)1.7 (0.1)0.270BMI (kg/m²)*22.0 (3.6)23.3 (4.4)22.6 (4.4)0.504Systolic pressure (mm Hg)†119 (10)115 (12)114 (10)0.079Diastolic pressure (mm Hg)†75(8)72(8)72(8)0.175Haemodynamic parameters Pmn (mm Hg)*2.2 (1.0)a,b1.7 (0.8)a,c1.9 (0.9)a0.032 FTc (ms)*349.5 (58.0)365.0 (29.5)358.0 (30.0)0.398 SVR (dyne·s/cm5)*1594.5 (463.3)1875.0 (874)1706.0 (448.8)0.219 SVRI (dyne·s/cm5/m2)*2725.0 (1011.5)2955.0 (1479)2864.5 (666)0.400 PKR‡43.0 (29.8)53.0 (37.5)44.0 (22.8)0.190 SMII*1.6 (0.5)a1.1 (0.3)b1.1 (0.4)b<0.001 SW (mJ)†777.4 (175.2)a632.6 (177.1)b659.2 (155.0)b0.002 CPO (W)†0.89 (0.22)a0.73 (0.19)b0.77 (0.19)b0.006 VTI (cm)†22.2 (4.6)20.0 (4.9)21.4 (4.2)0.157 SV (mL)64.3 (13.6)a53.5 (14.4)b57.3 (13.0)a,b0.007 SVI (mL/m2)†38.2 (8.3)a32.6 (7.8)b33.8 (7.5)a,b0.012 CO (L/min)†4.43 (1.00)a3.73 (0.85)b4.05 (0.89)a,b0.010 CI (L/min/m2)2.63 (0.64)a2.26 (0.47)b2.42 (0.56)a,b0.031Exposure variablesBurnout (MBI-HSS), n (%)0.146 Severe2 (6.25)7 (21.21)8 (20.00) Moderate24 (75.00)20 (60.61)30 (75.00) Low6 (18.75)6 (18.18)2 (5.00)Family history‡, n (%)0.672 Both4 (12.50)5 (15.15)10 (25.00) Metabolic syndromes only25 (78.13)24 (72.73)27 (67.50) None3 (9.37)4 (12.12)3 (7.50)Daily smoking and weekly alcohol, n (%) Both3 (9.37)2 (6.06)0 (0.00)0.158 Smoking only2 (6.25)1 (3.03)0 (0.00)0.196 Alcohol only11 (34.38)4 (12.12)7 (17.50)0.069 None16 (50.00)a26 (78.79)b33 (82.50)b0.005Weekly exercise time (hours)*1.0 (1.5)1.0 (1.8)1.0 (2.0)0.793Regular diet, n (%) Yes22 (68.75)19 (57.58)24 (60.00)0.619 No10 (31.25)14 (42.42)16 (40.00)Monthly duty shifts (n)*1 (1)a5 (2)b6 (1)c<0.001Workdays per week, n (%) 4–5 days29 (90.63)a26 (78.79)a2 (5.00)b<0.001 6–7 days3 (9.37)7 (21.21)38 (95.00)Values sharing a common superscript letter are not significantly different (p>0.05).*Non-normal continuous variables: median (IQR).†Normal continuous variables: mean (SD)‡Family history includes metabolic syndrome and malignancy, with no cases of malignancy only.BMI, body mass index; BSA, body surface area; CI, cardiac index; CO, cardiac output; CPO, cardiac power output; DP, depersonalisation; EE, emotional exhaustion; FTc, flow time corrected; MBI-HSS, Maslach Burnout Inventory-Human Services Survey; PKR, potential kinetic ratio; Pmn, mean pressure gradient; SMII, Smith–Madigan inotropy index; SV, stroke volume; SVI, stroke volume index; SVR, systemic vascular resistance; SVRI, systemic vascular resistance index; SW, stroke work; VTI, velocity–time integral. Baseline characteristics of participants across groups Values sharing a common superscript letter are not significantly different (p>0.05). Non-normal continuous variables: median (IQR). Normal continuous variables: mean (SD) Family history includes metabolic syndrome and malignancy, with no cases of malignancy only. BMI, body mass index; BSA, body surface area; CI, cardiac index; CO, cardiac output; CPO, cardiac power output; DP, depersonalisation; EE, emotional exhaustion; FTc, flow time corrected; MBI-HSS, Maslach Burnout Inventory-Human Services Survey; PKR, potential kinetic ratio; Pmn, mean pressure gradient; SMII, Smith–Madigan inotropy index; SV, stroke volume; SVI, stroke volume index; SVR, systemic vascular resistance; SVRI, systemic vascular resistance index; SW, stroke work; VTI, velocity–time integral. Correspondingly, the clinical support staff group showed significantly lower cardiac stroke volume parameters SVI, CO and CI than the administrative personnel group (SVI: p=0.015; CO: p=0.007; CI: p=0.026), with no significant differences observed vs the obstetricians group. SVRI (reflecting afterload) and PKR (reflecting energy conversion efficiency) were comparable across the three groups. No significant differences were found among the three groups in family history or MBI-HSS scores. Regarding lifestyle, the administrative personnel group reported a significantly higher prevalence of tobacco and alcohol use than both the clinical support staff (p=0.020) and obstetricians (p=0.005) groups. Workload measures revealed significant differences: monthly night shifts differed across all groups (p<0.001), and 95% of the obstetricians group worked 6–7 days per week, a proportion significantly higher than that in both the administrative personnel and clinical support staff groups (both p<0.001). MLR analysis MLR was performed with CPO (primary outcome) and CI (representative secondary outcome), both meeting normality and homoscedasticity assumptions. Significant models were obtained for CPO (F=4.510, p<0.001; R2=39.2%, adjusted R2=0.305) and CI (F=4.590, p<0.001; R2=39.6%, adjusted R2=0.310). Diagnostic plots confirmed assumption validity (onlinesupplemental figures S3S6). Tolerance and variance inflation factors indicated no serious multicollinearity (tables2 3). Table 2Multiple linear regression analysis of the associations between exposure factors and cardiac power outputVariablesUnstandardised coefficientStandardised coefficient (β)B (95% CI)P valueConstant0.533 (0.125 to 0.941)0.011–Age (years)0.000 (-0.006 to 0.007)0.8860.013Moderate burnout−0.073 (-0.185 to 0.040)0.202−0.159Severe burnout−0.152 (-0.295 to −0.008)0.039−0.269Weekly exercise time (hours)0.046 (0.020 to 0.072)<0.0010.335Regular diet0.028 (-0.056 to 0.112)0.5090.0656–7 workdays per week0.005 (-0.080 to 0.089)0.9090.012Monthly duty shifts−0.014 (-0.032 to 0.004)0.125−0.157BMI (kg/m²)0.017 (0.001 to 0.032)0.0340.206Alcohol only0.071 (-0.022 to 0.164)0.1320.140Smoking only−0.188 (-0.425 to 0.049)0.119−0.150Smoking and alcohol−0.154 (-0.330 to 0.022)0.086−0.158Family history(metabolic syndromes only)−0.101 (-0.227 to 0.026)0.118−0.216Family history (metabolic syndrome and malignancy)−0.088 (-0.236 to 0.061)0.243−0.163Tolerance values for all independent variables ranged from 0.382 to 0.888, and variance inflation factors ranged from 1.126 to 2.617, indicating no serious multicollinearity.BMI, body mass index. Multiple linear regression analysis of the associations between exposure factors and cardiac power output Tolerance values for all independent variables ranged from 0.382 to 0.888, and variance inflation factors ranged from 1.126 to 2.617, indicating no serious multicollinearity. BMI, body mass index. Table 3Multiple linear regression analysis of the associations between exposure factors and cardiac indexVariablesUnstandardised coefficientStandardised coefficient (β)B (95% CI)PConstant3.116 (2.051 to 4.180)<0.001–Age (years)0.004 (-0.012 to 0.021)0.6270.043Moderate burnout−0.174 (-0.467 to 0.119)0.242−0.139Severe burnout−0.403 (–0.777 to −0.028)0.036−0.260Weekly exercise time (hours)0.155 (0.087 to 0.223)<0.0010.413Regular diet−0.057 (-0.276 to 0.162)0.604−0.0496–7 workdays per week0.112 (–0.109 to 0.332)0.3180.098Monthly duty shifts−0.024 (–0.071 to 0.022)0.302−0.100BMI (kg/m²)−0.016 (–0.056 to 0.024)0.430−0.072Alcohol only0.182 (–0.061 to 0.425)0.1400.130Smoking only−0.417 (–1.035 to 0.202)0.185−0.122Smoking and alcohol−0.299 (–0.758 to 0.160)0.199−0.112Family history (metabolic syndromes only)−0.465 (–0.795 to −0.136)0.006−0.365Family history metabolic syndrome and malignancy)−0.469 (–0.856 to −0.082)0.018−0.317Tolerance values for all independent variables ranged from 0.382 to 0.888, and variance inflation factors ranged from 1.126 to 2.617, indicating no serious multicollinearity.BMI, body mass index. Multiple linear regression analysis of the associations between exposure factors and cardiac index Tolerance values for all independent variables ranged from 0.382 to 0.888, and variance inflation factors ranged from 1.126 to 2.617, indicating no serious multicollinearity. BMI, body mass index. After adjusting for other confounders, weekly exercise time was independently associated with both CPO (B=0.046 W, 95% CI 0.020 to 0.072, p<0.001; table 2) and CI (B=0.155 L/min/m², 95% CI 0.087 to 0.223, p<0.001; table 3). Conversely, severe burnout (both EE and DP at high levels) was the strongest predictor of significantly reduces CPO (B=−0.152 W, 95% CI −0.295 to −0.008, p=0.039; table 2) and CI (B=−0.403 L/min/m², 95% CI −0.777 to −0.028, p=0.036; table 3). Additionally, elevated BMI was significantly associated with a reduction in CPO but not with CI. Conversely, a positive family history of metabolic syndromes independently associated with reduced CI but not CPO (tables2 3). The forest plot of β (figure 2) indicated that weekly exercise time was the strongest independent predictor, with significant positive associations with both CPO (β=0.335, p<0.001) and CI (β=0.413, p<0.001). Standardised coefficient forest plot for CPO (A) and CI (B). BMI, body mass index; CI, cardiac index; CPO, cardiac power output. Subgroups and interactions Subgroup analyses indicated that weekly exercise time remained a significant protective factor for both CPO and CI outcomes in the obstetricians and clinical support staff groups (all p<0.05), whereas severe burnout did not show a statistically significant association within any of the three subgroups (online supplemental tables S7 and S8). To further explore whether occupational group modified these associations, interaction analyses were performed (online supplemental table S9). Results confirmed that the beneficial effect of weekly exercise time on CI was significantly more pronounced in the obstetricians group (interaction p=0.03). Additionally, interaction analyses identified alcohol consumption as a detrimental factor, with its negative association with both CPO and CI being significantly stronger in the obstetricians group (interaction p < 0.01). In contrast, the combined habit of smoking and alcohol exhibited a specifically harmful interaction in the clinical support staff group for the CI outcome (interaction p=0.03). Discussion Among healthy hospital employees, both the obstetricians and clinical support staff groups showed significantly reduced cardiac function parameters (CPO, SMII, SW) compared with the administrative personnel group (all p<0.05). Additionally, the clinical support staff group had lower stroke volume parameters (SVI, CO, CI) than the administrative personnel group, with no significant difference versus the obstetricians group. MLR analysis confirmed that each additional hour of weekly exercise was associated with significantly higher CPO (B=0.046 W) and CI (B=0.155 L/min/m²), with both p<0.001. Significantly lower values of both CPO (B=−0.152 W) and CI (B=−0.403 L/min/m²), p<0.05, were the most predictive indicators of severe burnout. In contrast to prior studies linking workload and lifestyle to disease,35 36 we aimed to identify subclinical status via haemodynamic monitoring, which detects preclinical disturbances before BP changes. While invasive techniques provide precision, non-invasive monitoring is advantageous for early risk stratification due to its practicality and reproducibility.37 For instance, in pregnancy, SVR physiologically decreases to support uteroplacental circulation, while elevated SVR predicts adverse obstetric outcomes.22 38 The principle that haemodynamic dysregulation precedes clinical disease inspired our study and may be used for early detection, monitoring and therapy. Contrary to our hypothesis, Obstetricians and Clinical support staff showed no significant haemodynamic differences, but both clinical groups differed from Administrative personnel. To our knowledge, this is the first study to explore the predictive value of these non-invasive parameters in occupationally stressed individuals, despite established reference ranges in healthy populations.28 Notably, we assessed correlates of these haemodynamic differences. Prior studies consistently show physical activity is inversely associates with incident CVD risk in both the general population and those genetically predisposed to CVD.39 40 Consistently, our study identified weekly exercise, MVPA time, as the strongest independent correlate of haemodynamic parameters. This association remained significant in subgroup analyses, where weekly exercise time was a protective factor for both CPO and CI outcomes in both the obstetricians and clinical support staff groups (all p<0.05). Furthermore, interaction analysis revealed that the protective effect of weekly exercise on CI was significantly more pronounced in the obstetricians group (interaction p = 0.03). This suggests that the cardioprotective benefit of exercise may be particularly enhanced among obstetricians, a group potentially exposed to unique occupational stressors. Regarding the levels of physical activity, it should be noted that our recorded MVPA fell below WHO recommendations which advise adults accumulate 150–300 min of moderate-intensity or 75–150 min of vigorous-intensity aerobic activity, or an equivalent combination weekly.34 Our assessment of exercise explicitly excluded occupational and household activities, even if those activities met WHO criteria for moderate-intensity physical activity. Additionally, severe burnout (high EE and DP) was present in about 20% of participants, within the reported range (6.3–52.7%).10 11 Critically, severe burnout was independently associated with adverse haemodynamics. A recent meta-analysis (n=26 916) reported a 21% higher CVD risk with burnout (fully adjusted OR=1.21, 95% CI 1.03 to 1.39).41 Chronic psychological stress contributes to disease via neurological, immune and hormonal pathways.17 42 Given burnout is a form of chronic stress, we are the first to demonstrate that altered haemodynamics in severe burnout may serve as an intermediate process linking this form of chronic stress to CVD. However, no statistically significant interaction or subgroup differences were found for severe burnout, indicating a consistent association with adverse haemodynamics irrespective of occupational group. Moreover, CPO and CI were positively correlated with BMI and family history of metabolic syndrome, respectively. Families with metabolic diseases are known to predispose offspring to metabolic disorders.43 44 The BMI-CPO association in our obesity-free cohort may suggest a physiological adaptation, wherein the heart increases output to meet the metabolic demands of a larger body size. This contrasts with the pathological cardiac overload in obesity, which is driven by chronic inflammation and neurohormonal activation and ultimately impairs cardiac function.45 While we used BMI normalised values, BMI is limited in its accuracy for measuring adipose distribution or reflecting general health status; waist-to-hip ratio and waist-to-height ratio were recommended to complement it.46 47 Future studies could integrate these measures to determine if this adaptation generalises across diverse populations. In contrast to earlier studies linking workload to diabetes, infections and CVD,6 we found no such associations. High variability in post-nightshift rest (range: none to 2 days) and overtime duration precluded precise total working hour calculation. We therefore operationalised workload using workdays and duty shifts as independent variables, an approach which potentially attenuated effect sizes, but which was more feasible. While this study is thorough in design and execution, and promising in its results a number of limitations may impact its general adoption. First, the results is from a single tertiary hospital and may not necessarily provide for universally applicable results, despite findings that are consistent with emerging physiologic concepts and prior allied studies. Second, while sufficiently powered for primary analyses, the study had limited statistical power for exploratory subgroup and interaction analyses. Non-significant interaction findings should therefore be interpreted with caution. Third, we balanced age across groups, but could not match on other confounders (eg, sex) due to occupational demographic differences, which may affect group comparability. Fourth, non-response was uneven across groups. This imbalance may have introduced selection bias. Fifth, although our participant questionnaire captured common conditions and behaviours, it was not comprehensive, so some potentially significant variables may have been overlooked. This included the need for specialised instruments to properly assess factors such as sleep, nutrition and other specific lifestyle details. Finally, while the MBI-HSS was selected for its established authority in measuring burnout, the assessment of complex mental states relies on self-reporting, which suggests a potential for bias. Conclusions Our study demonstrates that healthcare workers from different occupational backgrounds exhibit distinct haemodynamic profiles, independent of BP. While this study was conducted in China’s healthcare environment our findings are more far reaching. The novelty of applying non-invasive haemodynamic monitoring to screen for sub-groups in high stress medical environments may have expanded implications in many other high-stress occupations. Future longitudinal studies are needed to confirm the causal impact of these findings and their effect on occupational health outcomes. Supplementary material Footnotes Funding: This work was supported by the Xiamen Health Care projects grant number 3502Z20244ZD1218 and the Natural Science Foundation of Fujian Province grant number 2024J08308. Prepublication history and additional supplemental material for this paper are available online. To view these files, please visit the journal online (https://doi.org/10.1136/bmjopen-2026-117014). Provenance and peer review: Not commissioned; externally peer reviewed. Patient consent for publication: Not applicable. Ethics approval: This cross-sectional study was approved by the Ethics Committee of Women and Children’s Hospital, Xiamen University (KY-2024-148-H01). The study was performed in accordance with the ethical standards of the Declaration of Helsinki. Electronic informed consent was obtained from all participants prior to data collection. Data availability free text: The datasets generated and analysed during the current study are not publicly available due to the sensitive nature of the psychological questionnaire data involved. De-identified data are available from the corresponding author, MY, upon reasonable request. Requests will be reviewed by the study’s Ethics Committee to ensure they align with the original ethical approval. Requestors may be required to sign a data use agreement. Patient and public involvement: Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research. Data availability statement Data are available upon reasonable request. References Supplementary Materials Data Availability Statement Data are available upon reasonable request.