The Institutional Review Board approval and subject informed consent were obtained before the study began [Tsinghua University, China (No. 20210140)]. This prospective survey study was conducted by the Chinese Association of Radiologists (CAR), between December 1, 2020 and April 30, 2021. The survey population comprised radiology residents during their standardized residency training across 31 provinces (data on Macao, Hong Kong and Taiwan are not available) in China. Data were collected using a self-administered and structured questionnaire, which was an anonymous web-based survey issued via a popular online survey platform called “Wenjuanxing” in China. At the start of the online link, consent was required to continue, and the respondents could withdraw their information at any time. A total of 3,666 out of 12,208 potentially eligible radiologists who were receiving standardized residency training during 2020 in China responded effectively to our survey. They came from 97 cities in 31 provinces of China, and 407 (73.1%) of 557 radiology programs were covered.

Descriptive analysis was calculated for continuous variables [mean, standard deviation (SD)] and categorical variables (n, %). The percentages of lifestyles among residents, distribution of competency-based performance and prevalence of burnout with confidence interval were calculated. Pearson’s chi-square test was performed to test associations between socio-demographics and lifestyle behaviors, perceived-competency and occupational burnout. Spearman’s correlation analysis was carried out among the lifestyle factors.

Multiple linear regression models were constructed to explore the relationship of lifestyles and perceived-competency, adjusted for socio-demographics. To further examine the impact of different lifestyle factors on competency, we constructed multivariable logistic regression models to obtain the odds ratios (ORs) and 95% confidence intervals (CIs) comparing lower and higher competency. Besides, to detect any discrepancy in the relationship between lifestyles and competency among residents with different characteristics, subgroup analyses were performed according to socio-demographic variables. To explore the relationship between lifestyles and occupational burnout, logistic regression models were used to identify predictive lifestyle factors to occupational burnout and each subscale (EE, DP). The association was presented as odds ratios (OR) and 95% confidence intervals (CIs).

Two-tailed p < 0.05 were considered statistically significant. Statistical analyses were performed using Stata 17.0 (Stata Corporation) and visualized using R 4.2.3 (R Foundation for Statistical Computing) for Windows.

A summary of the socio-demographic characteristics of the radiology residents is shown in Table 3. Of the 3,666 participants, most were female (58.02%), aged 25–30 (64.62%), and had a balanced regional distribution with fair representation from the Eastern (40.53%), Central (20.24%), and Western (33.28%) regions. Most held a bachelor’s degree (92.06%), were unmarried (77.63%), and childless (86.93%). Work experience was generally under 3 years (77.66%), with a majority working 8 h daily (72.15%) and not frequently on night duty (61.65%). Residents were evenly distributed across training years, with 96.84% trained in general tertiary A hospitals. Competency varied by several factors, while burnout was influenced by gender, location, family structure, and work patterns.

As shown in Table 4, the overall mean score for lifestyle behaviors was 2.79 (of 4), which indicates a moderately healthy lifestyle. Among the 4 scales, sleep, smoking, and alcohol consumption showed healthy behaviors of radiology residents, while they lacked physical activities. For perceived-competency, the overall mean score was 3.16 (of 9), indicating a low competence level, with the highest score reported in interpersonal and communication skills (3.51). Prevalence of occupational burnout in all participants was 25.78%, the proportions of radiology residents who scored high level in EE and DP were 16.83 and 23.19%, respectively.

The correlations among sleep, physical activity, smoking, and alcohol assumption were low (Supplementary Table 1), suggesting that there was a slight impact of confounders or collinearity. After adjusted for sociodemographic covariates (Supplementary Table 2 and Table 5), the radiology residents with high-level lifestyles had higher perceived-competency (β = 0.22, 95% CI = [0.08, 0.35]) compared with lifestyle behaviors scoring low, especially in medical knowledge (β = 0.16, 95% CI = [0.01, 0.30]), systems-based practice (β = 0.22, 95% CI = [0.07, 0.37]), practice-based learning and improvement (β = 0.24, 95% CI = [0.08, 0.40]), professionalism (β = 0.30, 95% CI = [0.13, 0.47]), and interpersonal and communication skills (β = 0.29, 95% CI = [0.10, 0.48]). Also, sleep was marginally associated with competency (β = 0.14, 95% CI = [0.02, 0.26]; β = 0.20, 95% CI = [0.09, 0.32]). Compared with residents who seldom exercise, those who had regular physical activities had higher competence (β = 0.25, 95% CI = [0.14, 0.37]). Among residents who consume alcohol, those consuming moderate and low alcohol had lower competence than those consuming high alcohol (β = −0.26, 95% CI = [−0.45, −0.07]; β = −0.35, 95% CI = [−0.58, −0.17]). Nevertheless, there was no significant association between smoke and competency-based performance among radiology residents.

As shown in Table 5, after transforming the perceived-competency into the categorical variable, it appears that the relationship between lifestyle behaviors and overall high performance was not noticeable (OR = 1.08, 95% CI = [0.91, 1.30]; OR = 1.21, 95% CI = [0.98, 1.50]). However, residents with normal sleep were 1.24 times (95% CI = [1.03, 1.50]) more likely to develop high performance than those who had insomnia, and residents with regular physical activities were 1.40 (95% CI = [1.17, 1.67]) times than residents who exercise infrequently. Residents who consumed moderate and low alcohol were 0.64 (95% CI = [0.48, 0.85]) and 0.53 (95% CI = [0.39, 0.72]) times, respectively, than those who drank more.

In subgroup analyses, there were obvious differences in the correlation between lifestyle and perceived-competency in terms of work hours per day and night shifts (Figure 1 and Supplementary Table 3). These associations were stronger among residents working for over 8 h per day (p = 0.003) and rotating more night shifts (p = 0.011). Specifically, as shown in Supplementary Table 4, radiology residents with more work hours per day and more night shifts were more likely to suffer from sleep disorders (p < 0.001; p < 0.001), and had fewer physical activities (p < 0.001; p = 0.009), likewise, radiology residents with more night shifts were more likely to drink more alcohol (p < 0.001), which led to a closer association between lifestyle and competency.

Table 6 reveals the results of the analysis of lifestyle factors associated with occupational burnout by a multivariate logistic regression. Smoking, sleep, and physical activity were independently associated with occupational burnout, while the effect of alcohol consumption was non-significant. The OR of smoking, alcohol consumption, sleep and physical activity factors were 0.60 (95% CI = [0.40, 0.89], 1.23) (95% CI = [0.09, 17.59]), 7.69 (95% CI = [4.23, 14.67]), and 6.24 (95% CI = [1.33, 29.37]), respectively, comparing the worst to the healthiest state. This demonstrated that participants who suffer insomnia almost every day and never exercise had higher risk for occupational burnout when compared with those with healthy lifestyles, while participants who smoke show the lower risk.

Secondary analysis was conducted similarly for emotional exhaustion (EE) and depersonalization (DP) subscale to identify the specific association between lifestyles and occupational burnout. As shown in Supplementary Table 5, 6, the logistic regression analysis revealed that high EE was predicted by sleep (insomnia almost every day versus never; OR = 15.08, 95% CI = [7.16, 31.73]) and physical activity (never exercise versus almost every day, OR = 10.06; 95% CI = [1.23, 82.51]), high DP was predicted by sleep (insomnia almost every day versus never, OR = 11.74; 95% CI = [5.41, 25.49]) and smoking (smoking versus no smoking; OR = 0.60, 95% CI = [0.38, 0.92]), which indicated that the effects of lifestyle on EE and DP are slightly different.

Consistent with previous studies, radiology residents having adequate sleep demonstrate higher competency and lower occupational burnout in our analyses. Healthcare professionals, including radiologists, often suffer from sleep deprivation and most of them reported that they were not receiving the amount of sleep needed (23, 24). This is particularly true for residents due to their demanding work schedules, including numerous shifts and on-call responsibilities. Our results align with previous studies suggesting that adequate sleep was one supporting factor of the high perceived-competency of radiology residents, notably in systems-based practice, practice-based learning and improvement, professionalism, and interpersonal and communication skills (25–27). It is probable that sleep deprivation affects the residents’ cognitive function (28, 29), such as visual and perceptual expertise required in substantial report interpretation of daily work in radiology, thereby leading to attentional failure (30). It also has been reported that higher emotional exhaustion and depersonalization are related to sleep quality directly through insomnia, or indirectly through addictive behaviors such as bedtime smartphone use (31, 32). Complicated and critical working environment and high level of workforce stress can easily induce negative psychological (33), while sleep deprivation or circadian disorders may partly aggravate feelings of burnout and the chronic depletion of energy stores (34, 35). A study conducted in medical students in Hong Kong had shown that high EE was predicted by sleep quality, also high DP was predicted by sleep quality (36). Moreover, in the current studies, sleeping habits were associated with all subscales of burnout, which was consistent with our findings (36).

Similarly, radiology residents engaging in regular physical activities experienced lower risk of occupational burnout and better competency-based performance in our analyses, which is consistent with previous studies (7, 37–39). Physical activity, enhancing an improved sense of self-control and greater social interaction, has been confirmed of antidepressant and anxiolytic effects and protecting against harmful consequences of stress, which may have positive implications for mental health (40) and decrease risk of burnout, especially emotional exhaustion (41). Moreover, radiology residents spend most of their time working in a seated position in daily work, taking an average of only 294 steps per hour (12). Such sedentary behavior increases the risk of musculoskeletal discomfort (42) and eyestrain (43), causing metabolic disorders (44), and weakening the benefit of exercise on health and fatigue relief (45, 46). Conversely, the immense stress of residents due to heavy workload could be alleviated by regular physical activities (47).

Interestingly, residents with high alcohol consumption reported higher competency than those who consumed little alcohol, and smokers reported lower risk of burnout than non-smokers. Although this may seem counterintuitive, several potential explanations exist for these findings. Firstly, in Chinese culture, social events often involve drinking, potentially enhancing interpersonal and communication skills (48, 49). Alternatively, alcohol consumption and smoking could be a compensatory behavior to cope with stress and depression; when the residents have no choice but to compromise their personal health for better psychology and better performance at work, they might turn to these substances for immediate relief (50). Furthermore, previous study has also highlighted that stress increases cigarette cravings (51), suggesting that satisfying these desire may momentarily alleviate stress and, to some extent, decrease burnout risk. Burnout often leads to a negative approach to work and poor lifestyle choices including smoking (52). Additionally, it was also reported that smoking was associated with burnout (53).

Chinese standardized residency training aims to improve residents’ core competencies (54). It is equally essential to pay attention to occupational psychology to sustain radiologists’ performance. Residents, burdened with responding to patients and senior medical staff, suffer from circadian clock disruptions and exhaustion. Moreover, increased screening programs and related responsibilities have intensified the workload for radiologists, making a healthy lifestyle more challenging. Still, the residents had a higher rate of attentional failures during prolonged work hours and on-call shifts (30, 55), leading to energy depletion, decreased interest and further deterioration of their lifestyles, which are the maladaptation to their busy schedules and stress coping mechanisms (53), and further affect residents’ perceived-competency and increasing burnout risk, potentially reinforcing a vicious cycle (56). Therefore, it is necessary and urgent to focus on the modification of lifestyle behaviors among radiology residents, especially their sleep, smoking, physical activities and drinking habits. Reasonable strategies could include incorporating non-exercise activity thermogenesis (NEAT) into work routines—ensuring adequate hydration, taking regular breaks, performing basic exercises, and monitoring caloric intake (23, 44)—and scheduling night shifts to allow for sufficient sleep. It is not ethical nor sustainable for residents to sacrifice personal health for professionalism and performance. Our findings also indicated that the changes in lifestyles among this vulnerable group may contribute to enhancement in competency. In addition, systemic/organizational interventions are needed as well as individual management. Human factors and ergonomics (HFE) in healthcare delivery is helpful to optimize clinician wellbeing and clinical outcomes (57). Previous studies have shown that the system performance and safety will be improved by reducing distractions in workplace, operating the standard procedure for patient identification (58), and taking the organization level safety workshops (59).