Prevalence and associated factors of burnout among postgraduate trainee physicians in Bangladesh: a multicentre cross-sectional study

Abstract

Objectives

This study aimed to determine the prevalence of burnout and its associated factors among postgraduate trainee physicians in Bangladesh.

Design

A multicentre cross-sectional study.

Setting

Three public tertiary-care academic institutions across the country (Dhaka Medical College Hospital, Chittagong Medical College Hospital and Rangpur Medical College Hospital).

Participants

Study participants were 703 postgraduate trainee physicians from the Departments of Medicine, Surgery and Gynaecology & Obstetrics who were undergoing training programmes entitled Fellowship of the College of Physicians and Surgeons (FCPS), Doctor of Medicine/Master of Surgery (MD/MS) and Diploma/Master of Philosophy (MPhil).

Main outcome measures

Burnout was assessed using the Abbreviated Maslach Burnout Inventory, which evaluates burnout through three domains: emotional exhaustion (EE), depersonalisation (DP) and personal accomplishment (PA). In this study, weighted prevalence estimates were calculated where burnout was defined as a high EE score (≥9) combined with either a high DP score (≥6) or a low PA score (<12).

Results

The mean age was 30.9±3.6 (SD) years, with females comprising 56% of participants. Studied postgraduate trainees were enrolled in FCPS (62.2%), MD/MS (27.2%) and Diploma/MPhil (10.6%) courses. Overall, the prevalence of burnout was 70%, with high EE (68.3%), high DP (57.8%) and low PA (94.7%). Univariate analysis showed burnout was significantly associated with shorter sleep duration (<7 hours: 74.9% vs ≥7 hours: 61.4%, p<0.001) and lack of additional employment (73.5% vs 65.5%, p=0.036). However, multivariate logistic regression showed only shorter sleep duration was the independent predictor of burnout (OR 1.83, 95% CI 1.25 to 2.69, p=0.002).

Conclusions

This study demonstrates a high prevalence of burnout among postgraduate trainee physicians in Bangladesh. Targeted interventions addressing modifiable associated factors—particularly sleep deprivation, financial stress, and work structure—are urgently needed to promote mental well-being and sustain professional performance.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• This multicentre study recruited postgraduate trainee physicians from three tertiary-level teaching hospitals across different regions of Bangladesh, increasing the diversity of participants and enhancing the external validity of the findings.

• Inclusion of a large sample comprising trainees from medicine, surgery and gynaecology improved the statistical robustness and reliability of the study estimates.

• Burnout was assessed using domain-specific measures, allowing comprehensive evaluation of emotional exhaustion, depersonalisation and reduced personal accomplishment.

• As the study used a cross-sectional design, temporal relationships could not be established and causal interpretations of the observed associations should be made with caution.

Introduction

Burnout is a syndrome characterised by occupation-related physical, mental and emotional exhaustion (EE), which may culminate in psychological distress and illness.¹ It is widely recognised as a prevalent issue among healthcare professionals, with a reported prevalence ranging from 30% to 70% across different populations and settings.^2–5 The burden of burnout appears to be particularly high among physicians undergoing postgraduate training.

The postgraduate training period is one of the most critical and stressful phases in a physician’s career. During this period, trainees are exposed to intense academic and clinical responsibilities, including extended duty hours, patient care, academic presentations and examinations. Along with these burdens, social isolation further exacerbates the stress.⁶ Evidence indicates that globally 47.3% of postgraduate medical trainees experience burnout, with prevalence estimates varying between 43.1% and 51.5% across countries and specialties.⁷ In Tunisia, the prevalence of burnout among medical residents was reported as high as 79.3%.⁸ Burnout among physicians has far-reaching consequences, including impaired clinical judgement, diminished communication skills and reduced quality of patient care.^9–14 This problem is especially concerning among postgraduate trainees, where burnout may hinder the development of compassionate, patient-centred care and compromise professional growth.¹⁵

Several factors have been observed to contribute to burnout, including long and irregular working hours, demanding shift patterns, insufficient recognition and limited time for personal and family life.^{6 16 17} Extended duty hours and night shifts are associated with reduced sleep and higher fatigue which may lead to high burnout.¹⁸ Financial strain, inadequate stipend and educational debt have also been associated with greater EE and perceived stress among trainees.¹⁹ Engagement in additional paid work or moonlighting is common in low-resource settings and may result in increased total workload, reduced sleep opportunity and worse well-being among trainees.²⁰ Maladaptive coping behaviours such as smoking and substance use also may correlate with higher burnout and poorer mental health among physicians and residents.²¹ Burnout during this formative period can impair learning, hinder the development of clinical skills and compromise the ability to establish rapport with patients and colleagues.²²

In Bangladesh, medical graduates complete a 5-year programme named Bachelor in Medicine and Bachelor in Surgery along with a 1-year internship. The primary postgraduate medical courses include Doctor of Medicine (MD), Master of Surgery (MS), Fellowship of the College of Physicians and Surgeons (FCPS), Diploma and Master of Philosophy (MPhil) programmes. Admission to these programmes is highly competitive with a limited number of available seats and high demand for specialisation. During postgraduate training, trainees assume significant responsibilities in both inpatient and outpatient care, often working long and irregular shifts while balancing academic commitments. This demanding environment is compounded by inadequate stipends and extensive work hours, which have led to repeated protests by trainees demanding better remuneration and improved working conditions.^{23 24}

Despite the global recognition of burnout as a significant issue among healthcare professionals, there is a paucity of research focusing specifically on postgraduate medical trainees in Bangladesh. Existing studies have concentrated on practicing physicians or frontline healthcare workers.^{9 25} A previous study in Bangladesh reported a high prevalence of burnout (21.3%) among physicians and nurses working in intensive care unit (ICUs).²⁶ During the COVID-19 pandemic, the overall prevalence of burnout syndrome among physicians in Bangladesh was 55.4%.⁹ However, the unique stressors faced by postgraduate trainees- balancing rigorous academic demands, clinical responsibilities and personal life-remain underexplored. Addressing these gaps is essential to identify modifiable risk factors and to inform interventions that protect the well-being and professional development of postgraduate trainees. Hence, given the local context and international evidence, we prespecified sleep duration, engagement in additional paid work, self-perceived financial status, and smoking status as factors which might have association with burnout. This study aims to determine the prevalence of burnout and its associated factors among postgraduate trainee physicians in Bangladesh.

Methods

Postgraduation training programme for physicians in Bangladesh

The postgraduate medical training system in public and private medical colleges and institutes in Bangladesh is primarily supervised by Bangladesh Medical University (BMU) and the Bangladesh College of Physicians and Surgeons (BCPS).^{27 28} The residency programmes (MD/MS); administered by BMU, offer 5-year clinical and academic training across various specialties in multiple medical colleges across the country. BMU also offers non-residency programmes through MPhil and Diploma courses, which usually require 2–3 years of training.²³ In contrast, the FCPS, conducted by BCPS, offers fellowship training in around 60 disciplines across different medical colleges and institutes nationwide.²⁴ Full programme details and seat distributions are provided in online supplemental table 1.

Study design and setting

This cross-sectional study was conducted over 6 months, from April to September 2024, at three medical colleges in Bangladesh: Dhaka Medical College Hospital, Chittagong Medical College Hospital, and Rangpur Medical College Hospital. These institutions were selected from three distinct administrative divisions of Bangladesh. Dhaka Medical College Hospital is situated in Dhaka, the capital of Bangladesh, and has 2600 beds and a large faculty. It accommodates around 750 postgraduate trainees across 42 disciplines, including MD/MS and MPhil/Diploma programmes under the coordination of BMU.²⁹ Chittagong Medical College Hospital, located in Chittagong, has over 1000 beds and offers postgraduate programmes such as MD, MS and Diploma in various specialties, enrolling approximately 130 postgraduate trainees each year.³⁰ Rangpur Medical College Hospital is also a 1000-bed hospital and provides postgraduate courses in medicine, surgery, paediatrics and gynaecology under BMU.³¹ All three institutes are affiliated with both BMU and BCPS for postgraduate training.

Study participants and recruitment strategy

The study population comprised postgraduate trainee physicians enrolled in regional specialisation programmes (MD, MS, Diploma, MPhil or FCPS) within the faculties of Medicine, Surgery and Gynaecology & Obstetrics across the three participating public tertiary-care academic institutions. To be eligible for inclusion, trainees must have completed at least 6 months of training within their respective programmes to ensure adequate exposure to the occupational environment. A convenient census-based sampling strategy was employed, wherein all eligible trainee doctors actively rotating through the target departments during the 6-month study period were invited to participate. Recruitment was conducted on-site by trained research assistants who approached potential participants during departmental morning sessions, weekly academic seminars and clinical shift transitions to minimise disruption to patient care. After verifying eligibility criteria and explaining the study objectives, written informed consent was obtained, and the self-administered questionnaires were distributed for independent completion.

The study included 703 postgraduate trainee doctors enrolled in postgraduate programmes.

Sample size calculation

A total of 703 postgraduate trainee physicians from different specialties took part in this study. The minimum sample size was calculated using the formula n=z²pq/d²;³² where, n=sample size, z=standard score corresponding to a given confidence level, p=prevalence of condition and d=degree of precision. Since the true prevalence of burnout among resident physicians in Bangladesh is unknown, a prevalence of 50% (p=0.5) was assumed to maximise the sample size estimation. A 95% confidence level was selected (z=1.96), and a margin of error of 5% (d=0.05) was considered. Using the above-mentioned information, the estimated sample size is 384. This basic sample size was adjusted for design effect of 1.5 and then the required sample size was, n=576. To accommodate an anticipated 20% non-response rate, the final sample size was 720. However, 703 postgraduate trainee physicians were included in final analysis.

Data collection

Data collection was conducted using a structured approach. A semistructured questionnaire was developed with three parts: Part A covered sociodemographic information, Part B covered training characteristics and Part C covered burnout assessment using the aMBI. In addition, secondary exploratory outcomes regarding training satisfaction, further intentions regarding career, and mental health help-seeking approaches were captured. After obtaining consent, the questionnaire was distributed to the postgraduate trainee physicians, who completed it independently. Completeness and accuracy of responses were ensured through careful review.

Assessment of burnout

Burnout among enrolled postgraduate trainees was assessed using the aMBI, a nine-item questionnaire evaluating burnout through three subscales: EE, depersonalisation (DP) and personal accomplishment (PA).^{33 34} This tool was administered in its original English version in this study. Because the participants were postgraduate trainee physicians and the aMBI has previously shown valid and reliable performance as a short-form screening tool in medical professionals, the original English version was considered appropriate for this study. Exhaustion subscale describes feelings of being emotionally overextended and exhausted by one’s work while DP subscale describes an unfeeling and impersonal response towards recipients of one’s care or service. Besides, the subscale of PA describes feelings of competence and successful achievement in one’s work with people.³⁵ Each subscale contains three items scored on a 7-point Likert scale from 0 (never) to 6 (every day), with total subscale scores ranging from 0 to 18. The aMBI cut-off ranges for EE, DP and PA were applied according to the standard weighted conversion from the full Maslach Burnout Inventory-Human Services Survey (MBI-HSS), as shown in online supplemental table 2. A higher score on EE and DP indicates greater burnout, whereas a higher score on PA indicates lower PA.^{36 37} For this study, burnout was operationally defined as a high EE score (≥9) combined with either a high DP score (≥6) or a low PA score (<12), consistent with established criteria.

Statistical analysis

The statistical analysis was conducted using RStudio V.2025.09.2 for Windows. Descriptive statistics were used to summarise sociodemographic and training-related characteristics, with categorical variables presented as frequencies and percentages and continuous variables as means and SDs. Missing data for individual variables were as follows: gender (0.1%), marital status (0.1%), ability to manage expenses independently (0.3%), receipt of financial support (0.4%), postgraduate programme (0.7%), department of training (0.9%), tea/coffee consumption (0.9%), age (1.1%), other paid employment (1.6%), children (2.1%), smoking history (2.3%), average sleep duration (4.4%), duration of training (4.6%) and working hours per week (5.4%). As burnout status was a derived variable calculated from the component aMBI subscale items, missing item-level responses propagated to the derived classification; complete burnout classification was available for 596 of 703 participants (84.8%), with 107 participants (15.2%) missing. Complete-case analysis was performed for all outcomes, and no imputation was applied.

To assess burnout, a weighted conversion of the 9-item aMBI subscale scores was applied to map them onto the standard MBI-HSS cut-off ranges. To enable a direct comparison between the 22-item MBI-HSS and the abbreviated version, we adopted the proportional weighting procedure described by Lim et al.³⁶ Weights (W) for each subscale were calculated as the ratio of the number of items in the abbreviated subscale (n_aMBI) to those in the standard MBI-HSS subscale (n_MBI) using the following formula: W=(n_aMBI)/(n_MBI). This approach yielded specific weight factors of 0.333 for EE (3/9 items), 0.600 for DP (3/5 items) and 0.375 for PA (3/8 items). The baseline MBI-HSS risk thresholds established by Gabbe et al were then multiplied by these W and rounded to the nearest whole integer.³⁷ This adjustment provided equivalent, clinically distinct cut-off boundaries tailored specifically for the abbreviated subscales, as detailed in online supplemental table 2.

The EE and DP subscales showed acceptable internal consistency (Cronbach’s α=0.77 and 0.70, respectively), and the PA subscale yielded a poor score (α=0.54).

Associations between categorical variables and the individual burnout dimensions were evaluated using χ² tests or Fisher’s exact tests, as appropriate. Multivariable logistic regression analysis was performed to identify factors independently associated with overall burnout, while multiple linear regression analysis was used to model continuous EE scores. Candidate predictor variables for the multivariable models were prespecified based on prior literature and clinical relevance.^{19 25} Model stability was verified by assessing multicollinearity using variance inflation factors (VIF); only variables with a VIF <5 were retained in the final models. All statistical tests were two-sided, and a p<0.05 was considered statistically significant.

Result

The mean age of the studied trainee physicians was 30.9±3.6 (SD) years and the females represented slightly higher proportion (55.7%). The majority of trainees lived with family (80.3%), and 51.9% were able to manage personal expenses independently. About 73.2% were married and 48.1% had at least one child. About 62.2% were enrolled in the FCPS programme, 27.2% in residency programme (MD/MS) and 10.6% were in non-residency programmes (MPhil/diploma). About 48.6% were in Medicine, 26.1% were in Surgery and 25.3% were in Gynaecology & Obstetrics. A total of 75.9% trainees received training for 2 years or less. Approximately 67% of the trainees worked more than 48 hours a week and mean working hour/week was 56.9±12.6 (SD). Approximately 47.1% reported secondary employment, such as clinical duties in external clinics or hospitals, private chamber practice or tutoring. The mean sleep duration in a 24-hour window was 6.3±1.1 (SD) hours, while 63.8% of the doctors sleep less than 7 hours a day. Only 5.4% were reported as smokers and 54.8% reported consuming more than two cups of tea a day (table 1).

Table 1. Sociodemographic characteristics and details of training among studied trainee physicians (n=703).

Characteristics	N=703*
Age (years)
<32	430 (61.9)
≥32	265 (38.1)
Mean±SD	30.9±3.6
Gender
Male	311 (44.3)
Female	391 (55.7)
Living arrangements
Living with family	560 (80.5)
Living alone	136 (19.5)
Marital status
Married	514 (73.3)
Unmarried	182 (25.9)
Widow/separated/divorced	6 (0.8)
Children
No	357 (51.9)
Yes	331 (48.1)
Postgraduate programme
FCPS	434 (62.2)
MD/MS	190 (27.2)
Diploma/MPhil	74 (10.6)
Department of training
Medicine	339 (48.6)
Surgery	176 (26.1)
Gynaecology and obstetrics	182 (25.3)
Duration of received training (current programme)
≤2 years	509 (75.9)
3–5 years	112 (16.7)
>5 years	50 (7.5)
Working hours per week
≤48	218 (32.8)
>48	447 (67.2)
Mean±SD	56.9±12.6
Able to manage personal expenses independently
Yes	364 (51.9)
Can manage some of the expenses independently but not fully	186 (26.4)
No	151 (21.5)
Receive financial support from family
Receive full financial support	188 (26.9)
Receive partial financial support	119 (16.9)
No	393 (56.2)
Have other paid employment
Yes	326 (47.1)
No	366 (52.9)
Average sleep duration (hours)
<7 hours	429 (63.8)
≥7 hours	243 (36.2)
Mean±SD	6.3±1.1
Smoking history
Currently smoker	37 (5.4)
Currently non-smoker	650 (94.6)
Consumption of tea
Not frequent (≤2 cups per day)	315 (45.2)
Frequent (>2 cups per day)	382 (54.8)

^*Mean (SD); n (%). Percentages were calculated using available cases; denominators vary because of missing data.

FCPS, Fellowship of the College of Physicians and Surgeons; MD, Doctor of Medicine; MS, Master of Surgery.

About 70% (n=417) of the study patients reported burnout as per our operational definition. Continuous score distributions for the burnout domains are presented in figure 1. The median score for EE was 10 (IQR: 6–14; mean: 10.1±4.7), while the median DP score was 3 (IQR: 0–6; mean: 4±4). For PA, the median score was 13 (IQR: 11–16; mean: 13±3.8).

Figure 1. Distribution of Maslach Burnout Inventory Subscales Scores among participants (n=703).

Trainees who slept less than 7 hours per day showed a significantly higher proportion of burnout compared with those sleeping 7 hours or more (74.9% vs 61.4%; p<0.001). Similarly, respondents lack of another paid employment reported a higher rate of burnout than those engaged in additional paid work (73.5% vs 65.5%; p=0.036) (table 2).

Table 2. Burnout status among trainee physicians by participant characteristics (n=703).

	Burnout
Characteristic	Absent N=179*	Present N=417*	P value
Age (years)			0.175†
<32	120 (32.2)	253 (67.8)
≥32	58 (26.8)	158 (73.1)
Sex			0.407†
Female	98 (28.7)	243 (71.3)
Male	81 (31.9)	173 (68.1)
Marital status			0.215†
Married	124 (28.5)	311 (71.5)
Not married	54 (33.7)	106 (66.2)
Living arrangement			0.636†
Living with family	139 (29.4)	334 (70.6)
Living alone	37 (31.6)	80 (68.4)
Financially self-sufficient			0.086†
Yes	100 (33.3)	200 (66.7)
No	79 (26.9)	215 (73.1)
Have another paid employment			0.036†
No	85 (26.5)	236 (73.5)
Yes	92 (34.5)	175 (65.5)
Sleep duration (hours)			<0.001†
<7	94 (25.1)	281 (74.9)
≥7	78 (38.6)	124 (61.4)
Smoking			0.175†
Non-smoker	166 (29.8)	391 (70.2)
Smoker	11 (42.3)	15 (57.7)
Consumption of tea/coffee			0.327†
Frequent (>2 cups per day)	91 (28.3)	231 (71.7)
Not frequent (≤2 cups per day)	86 (31.2)	183 (68.0)
Current postgraduate programme			0.248†
FCPS	116 (32.2)	244 (67.8)
MD/MS	45 (27.3)	120 (72.7)
Diploma/M.Phil	16 (23.5)	52 (76.5)
Department			0.593†
Medicine	94 (32.1)	199 (67.9)
Gynaecology	44 (28.8)	109 (71.2)
Surgery	40 (27.8)	104 (72.2)
Duration of training (years)			0.247†
≤2	139 (31.8)	298 (68.2)
>2	35 (26.5)	97 (73.5)
Weekly working hour			0.300†
>48	111 (29.2)	269 (70.8)
≤48	61 (33.5)	121 (66.5)

Bold text indicates that the findings are statistically significant.

^*n (%).

^†Pearson’s χ2 test.

FCPS, Fellowship of the College of Physicians and Surgeons; MD, Doctor of Medicine; MS, Master of Surgery.

Univariate logistic regression analysis revealed, shorter sleep duration (<7 hours) (OR 1.88, 95% CI 1.30 to 2.72; p<0.001) and lack of additional employment (OR 1.88, 95% CI 1.03 to 2.08; p=0.036) were found to be associated with burnout. However, in multivariate analysis, only shorter sleep duration (OR 1.75, 95% CI 1.25 to 2.45); p=0.001) was observed to be independently associated with high burnout (table 3).

Table 3. Factors associated with burnout among participants (n=703).

	Univariate			Multivariate
Characteristic	OR	95% CI	P value	OR	95% CI	P value
Age	1.02	0.97 to 1.07	0.509	1.01	0.95 to 1.09	0.734
Sex
Female	—	—		—	—
Male	0.86	0.61 to 1.23	0.407	1.06	0.65 to 1.73	0.816
Marital status
Not married	—	—		—	—
Married	1.28	0.86 to 1.88	0.216	1.05	0.61 to 1.80	0.867
Living arrangement
Living with family	—	—		—	—
Living alone (hostel)	0.90	0.58 to 1.40	0.636	0.97	0.54 to 1.76	0.918
Financial self-sufficiency
Yes	—	—		—	—
No	1.36	0.96 to 1.94	0.087	1.18	0.76 to 1.84	0.468
Additional employment
Yes	—	—		—	—
No	1.46	1.03 to 2.08	0.036	1.28	0.85 to 1.93	0.240
Sleep duration
≥7 hours	—	—		—	—
<7 hours	1.88	1.30 to 2.72	<0.001	1.70	1.13 to 2.57	0.011
Smoking habit
Non-smoker	—	—		—	—
Smoker	0.58	0.26 to 1.32	0.180	0.43	0.16 to 1.22	0.103
Consumption of tea/coffee
Frequent (>2 cups per day)	—	—		—	—
Not frequent (≤2 cups per day)	0.84	0.59 to 1.19	0.327	0.74	0.49 to 1.11	0.145
Postgraduate programme
Diploma/M. Phil	—	—		—	—
FCPS	0.65	0.34 to 1.16	0.157	0.57	0.26 to 1.16	0.132
MD/MS	0.82	0.42 to 1.56	0.555	0.66	0.29 to 1.42	0.301
Department
Gynaecology	—	—		—	—
Medicine (+paediatrics)	0.85	0.55 to 1.31	0.471	1.06	0.60 to 1.88	0.840
Surgery	1.05	0.63 to 1.74	0.851	1.31	0.67 to 2.57	0.427
Duration of training (year)
≤2	—	—		—	—
>2	1.29	0.84 to 2.02	0.248	1.37	0.83 to 2.30	0.222
Weekly duty hours
≤48	—	—		—	—
>48	1.22	0.83 to 1.78	0.300	1.07	0.69 to 1.66	0.747

Bold text indicates that the findings are statistically significant.

FCPS, Fellowship of the College of Physicians and Surgeons; MD, Doctor of Medicine; MS, Master of Surgery.

Linear regression analysis was conducted to identify factors associated with EE. Univariate analysis revealed, male sex (β=−0.87, p=0.016), financial self-sufficiency (β=−1.30, p<0.001), additional employment (β=−1.00, p=0.004), ≥7 hours of sleep (β=−1.20, p=0.002) and FCPS training (β=−1.40, p=0.017) were significantly associated with lower EE. In multivariate analysis, financial self-sufficiency (β=−1.08, p=0.014), ≥7 hours of sleep (β=−1.07, p=0.013) and FCPS training (β=−1.66, p=0.021) remained significant. Tea/coffee consumption showed a borderline association (β=−0.80, p=0.055) (table 4).

Table 4. Factors associated with emotional exhaustion among participants (n=703).

Characteristic	Univariate			Multivariate
	Beta	95% CI	P value	Beta	95% CI	P value
Age	−0.04	−0.14 to 0.05	0.361	−0.02	−0.15 to 0.10	0.704
Sex
Female	—	—		—	—
Male	−0.87	−1.6 to −0.16	0.016	−0.43	−1.41 to 0.55	0.391
Marital status
Not married	—	—		—	—
Married	0.42	−0.37 to 1.2	0.297	0.16	−0.95 to 1.27	0.774
Living arrangement
Living alone (hostel)	—	—		—	—
Living with family	0.33	−0.56 to 1.2	0.465	0.26	−0.94 to 1.45	0.674
Financial self-sufficiency
No	—	—		—	—
Yes	−1.3	−2.0 to −0.61	<0.001	−1.08	−1.94 to −0.21	0.014
Additional employment
No	—	—		—	—
Yes	−1.0	−1.8 to −0.34	0.004	−0.38	−1.22 to 0.45	0.364
Sleep duration
<7 hours	—	—		—	—
≥7 hours	−1.2	−1.9 to −0.41	0.002	−1.07	−1.91 to −0.23	0.013
Smoking habit
Non-smoker	—	—		—	—
Smoker	−0.95	−2.5 to 0.65	0.244	−0.77	−2.72 to 1.18	0.437
Consumption of tea/coffee
Frequent (>2 cups per day)	—	—		—	—
Not frequent (≤2 cups per day)	−0.41	−1.1 to 0.31	0.263	−0.80	−1.61 to 0.02	0.055
Postgraduate programme
Diploma/M. Phil	—	—		—	—
FCPS	−1.4	−2.6 to −0.26	0.017	−1.66	−3.06 to −0.25	0.021
MD/MS	−0.84	−2.1 to 0.45	0.203	−1.01	−2.52 to 0.51	0.191
Department
Gynaecology	—	—		—	—
Medicine	−0.62	−1.5 to 0.26	0.165	0.27	−0.91 to 1.44	0.656
Surgery	−0.64	−1.6 to 0.36	0.211	0.32	−1.02 to 1.66	0.638
Duration of training (years)
≤2	—	—		—	—
>2	−0.11	−0.95 to 0.73	0.793	0.10	−0.85 to 1.05	0.833
Weekly working hours
≥48	—	—		—	—
>48	0.69	−0.08 to 1.5	0.079	0.66	−0.22 to 1.54	0.140

Bold text indicates that the findings are statistically significant.

FCPS, Fellowship of the College of Physicians and Surgeons; MD, Doctor of Medicine; MS, Master of Surgery.

Detailed bivariate cross-tabulations of participant characteristics against individual burnout subscales including EE, DP and PA are provided in online supplemental table 3-5. These supplementary subscale analyses are intended to provide comprehensive baseline data and should be interpreted as exploratory. Overall, 52.9% trainees reported dissatisfaction with their training. About 40.7% considered changing specialties, 37.8% switching professions and 41.7% migrating internationally. Only 11.9% consulted mental health providers and 6.1% used medication (online supplemental table 6).

Discussion

This study explored burnout and its associated factors among trainee physicians enrolled in postgraduate programmes at three medical colleges across Bangladesh. A high prevalence of burnout was observed, with 70% of the surveyed doctors experiencing overall burnout. This finding indicates a substantial mental health burden and a serious occupational challenge within this population.

Globally, nearly half of postgraduate medical trainees experience burnout,⁷ though the rates of burnout among healthcare professionals vary widely by country and system.³⁸ A study conducted in Saudi Arabia among 6606 trainee physicians reported the prevalence of burnout as 56%.³⁹ Another study from Thailand reported the rate as 46.3%.⁴⁰

Burnout was assessed through three subscales-EE, DP and PA using the aMBI. Among the studied physicians, 68.3% reported high EE, 59.3% reported high DP and nearly all trainees (94.7%) reported low PA. The cultural and systemic context of the healthcare system of Bangladesh played a critical role in shaping burnout patterns among these postgraduate trainees. Being a densely populated country with very limited resources, Bangladesh faces a significant healthcare workforce shortage, often placing excessive clinical burdens on trainees. Similar to this study, EE has been reported as a prominent burnout dimension in low- and middle-income countries including: Pakistan (47%), Nepal (24.5%) and Sudan (70.7%).^41–43 The study from Sudan also reported DP and low PA in 44.2% and 73.1% trainees accordingly,⁴³ which aligned with the present study where a concerning number of doctors specialising in their postgraduation reported low PA. The 70% prevalence observed in our cohort is notably higher and consistent with broader regional data from low- and middle-income countries due to shared systemic stressors like overwhelming patient volumes and prolonged shifts.

Among the factors associated with burnout, shorter sleep duration was the only factor independently associated with overall burnout. The average sleep duration among studied trainees was approximately 6 hours in a 24-hour window, which is below the 7–9 hours recommended for adults.⁴⁴ Chronic sleep deprivation can impair cognitive function, emotional regulation and decision-making abilities, all of which are critical for both patient care and physician well-being. The finding of this study mirrors this theory. Previous studies have also reported that inadequate sleep is linked to increased burnout among healthcare professionals.⁴⁵ The role of sleep in emotional regulation and stress recovery remains insufficient and lack of rest further exacerbates stress and diminishes coping abilities, creating a cycle that intensifies burnout symptoms.

In unadjusted analysis, absence of additional employment was associated with a higher prevalence of overall burnout, but this association did not remain statistically significant in the multivariate logistic regression. The relationship between additional employment and burnout may be multidimensional. Hence, additional employment was associated with higher DP, the crude association with overall burnout appeared to move in the opposite direction. One possible explanation is that additional employment may provide financial relief for some trainees while simultaneously increasing workload and emotional detachment, resulting in contrasting patterns across burnout dimensions rather than a uniform effect on overall burnout. Global evidence reported a link between economic stability with lower burnout levels among healthcare workers.^{46 47} In this study, only half of the trainees were financially self-dependent and nearly half of them were engaged in additional employment.

Similarly, trainees in non-residency programmes showed higher EE, and surgical trainees reported lower PA. These findings may reflect differences in mentorship structure, institutional support, workload culture, or perceived career progression across programmes and specialties, but they were more evident at the subscale level than in the adjusted analysis of overall burnout.

The demanding nature of postgraduate training was evident in this study characterised by long weekly working hours with a mean of 56 hours. This schedule is notably above the recommended limits for safe and healthy work practices,⁴⁸ increasing the risk of fatigue and stress-related mental and physical issues. Unlike previous reports,^{12 49} the present study didn’t find any association of working hours and presence of burnout likely due to uniform overwork among the study participants.

Some differences were also observed across demographic and training-related subgroups, particularly for the subscales. Male trainees showed higher DP, whereas female trainees reported lower PA, although overall burnout was not significantly associated with gender. These patterns may reflect differences in workplace expectations, social roles, and perceived recognition during training, but the adjusted model did not identify gender as associated with burnout. Globally, higher rates of burnout associated with DP and low PA tend to be reported in female physicians, reflecting a feeling of less efficacy and professional growth.^{1 7 50 51}

To evaluate the absolute robustness of these associations, a sensitivity analysis was conducted treating individual scores of the EE subscale as continuous outcomes. This multiple linear regression model scores beautifully mirrors the findings of our primary logistic model. Specifically, both adequate sleep duration and financial self-sufficiency retained independent, highly significant protective associations against continuous exhaustion levels. Conversely, variables like male gender and additional employment, which appeared influential during univariate testing, lost their significance in the adjusted linear model.

A particularly concerning finding is that 40.7% of the study participants have considered changing their specialty, and 37.8% have contemplated leaving the profession due to stress. This high rate of potential attrition poses a serious threat to the healthcare workforce in Bangladesh and underscores the urgent need for interventions to address burnout among postgraduate trainee physicians. More than half of them were dissatisfied with their training. The financial strain, coupled with the demanding nature of training, may have led to this high dissatisfaction.

These findings present clear implications for promoting high performance in postgraduate medical training. Since sleep deprivation was the primary predictor of high burnout, there is an urgent need to restructure training and duty schedules. These findings present potential implications for reforming the structure of postgraduate medical training. As sleep deprivation showed the strongest association with burnout in this study, restructuring the duty schedules might be considered. Uninterrupted longer duty shifts could be restructured with structured rosters and mandatory postcall rest which might protect both the well-being of the trainee and the safety of the patient. Additionally, discussion and dialogue should be promoted to the relevant stakeholders to implement proactive support strategies for the well-being of all trainee doctors. In this study, the low PA scores are obvious and hereby signify the requirement of further training, monitoring and feedback mechanisms to improve the situation. It is expected that such supportive measures could help reduce professional isolation and contribute to a more sustainable learning environment.

While this study provides critical evidence regarding the high prevalence and associated factors of burnout among postgraduate trainee physicians in Bangladesh, several limitations must be considered. First, the cross-sectional design constrains our ability to establish causal pathways between the identified risk factors and burnout. Second, due to the unavailability of a well-designed previous study on this topic in our country, we assumed a prevalence of 50% to obtain the maximum possible sample size and calculated the required sample accordingly. However, we acknowledge that regression modelling was not considered during the sample size estimation process. Third, the tool used in this study to assess burnout lacks a dedicated cross-cultural validation and structural optimisation study within the Bangladeshi healthcare context. Although the applied proportional linear weighting matrix mathematically bridges the abbreviated tool to international ranges, future longitudinal research using localised, cross-culturally validated instruments is warranted to further confirm these structural patterns among Bangladeshi postgraduate trainees.

Conclusions

This study reported a 70% prevalence of burnout among Bangladeshi trainee physicians with a concerning proportion of low PA and high EE. Addressing burnout among postgraduate physician trainees requires a comprehensive approach involving individual, institutional and peer involving strategies.

Data availability statement

Data are available on reasonable request.