Prevalence and Associated Factors of Burnout Among Saudi Resident Doctors: A Multicenter Cross-Sectional Study

Abstract

Background:

Burnout is a prevalent psychological state among resident doctors. This study aimed at assessing the prevalence and associated factors of burnout among resident medical doctors in Medina, Saudi Arabia.

Methods:

This cross-sectional study was conducted among 426 resident doctors in Medina city, Saudi Arabia. We used the Maslach Burnout Inventory-Human Services Survey (MBI-HSS) to measure this phenomenon.

Results:

Of the participants, 81.22% scored high on at least one subscale of burnout and 18.31% scored high on all the subscales of burnout. Burnout was related to lack of physical exercise (P < .001), level of training (P < .001), number of on-call shifts per month (P = .020), number of weekends on-call per month (P < .050), number of patients seen per day (P = .002), number of clinics per week (P < .001), satisfaction with work–life balance (P < .001), and sources of stress in the workplace (P < .050).

Conclusion:

Burnout is present among resident doctors at a relatively high rate. Numerous factors associated with burnout were evident, particularly work-related factors and sources of stress in the workplace. Therefore, the Saudi Commission for Health Specialties and the residency program directors should act to improve working conditions and work–life balance, and minimize the impact of stressors in the workplace, to minimize the consequences of burnout among resident doctors. Provisions could be enacted to implement early comprehensive assessments of burnout syndrome among medical residents for early detection to curb the burnout phenomenon within healthcare systems.

Main Points

• This study explored burnout and its associated factors among resident medical doctors in Saudi Arabia.

• The Maslach Burnout Inventory-Human Services Survey (MBI-HSS) was used to measure burnout.

• Of the participants, 81.22% scored high on at least one subscale of burnout.

• Burnout was significantly associated with sources of stress in the workplace, lack of physical exercise, level of training, number of on-call shifts per month, number of weekends on-call per month, number of patients per day, number of clinics per week, and satisfaction with work–life balance.

Introduction

Burnout is a prevalent psychological state among resident doctors, and occurs due to continuous exposure to elevated workrelated stressors.¹ It is a syndrome consisting of emotional exhaustion (EE), depersonalization (DP), and reduced personal accomplishment (PA), which result from the accumulation of stress in the workplace.¹ Emotional exhaustion is a stress dimension of burnout, which is a feeling of fatigue, exhaustion of emotional and physical resources, and a lack of energy. Depersonalization relates to an impersonal response toward the recipients of one’s service. Reduced personal achievement or job satisfaction is described as dissatisfaction with one’s accomplishments, underestimating one’s own value, and seeing oneself as insufficient.^1,2

Work environments, especially medical settings, are intense, leading to stress among the employees. Notably, the general working population records 13-27% burnout rates compared to 70% among physicians.³ A recent systematic review of residents’ burnout found that the prevalence rate of burnout was 53.27% in surgical residents and 50.13% in medical residents.⁴ A study carried out in Riyadh found that the prevalence of overall burnout among resident doctors was 70%.⁵ In the same study, the emotional exhaustion subscale exhibited the highest scores (54%), followed by depersonalization (35%) and a low sense of personal accomplishment (33%).⁵

Residency is a demanding training program. Work demands, tight schedules, long hours of study and practice, and multiple evaluations and exams create an environment that makes resident doctors vulnerable to burnout.^2,3,6 This affects their perception of their careers as they become less satisfied with their service delivery and reduced effectiveness. The extreme exhaustion is especially threatening in the hospitals as it can adversely affect patient care and the related outcomes. The literature indicates that the work environment significantly affects the prevalence of burnout among resident doctors.⁶

The bulk of burnout literature to date explores the magnitude of the problem in Western settings.⁷ Although studies on burnout are emerging from around the globe, the apprehension of burnout would somewhat be influenced by the cultural context and organizational behavior across different settings and populations, based on Hofstede’s law on cultural dimensions affecting workplace burnout.⁸ We postulate that the prevalence of burnout may be higher among healthcare workers in Arab countries, in view of their health systems and financing models being either overburdened or rapidly developing and responding to the changing disease patterns of the population⁹ yet being correlated to a different cultural landscape of health system delivery as compared to the Western healthcare system. There is a scarcity of research concerning burnout among resident doctors in all specialty training programs in Saudi Arabia. By extension, there have been no such studies conducted in Medina city. Therefore, the prevalence of burnout among the resident doctors in Medina, Saudi Arabia, is relatively unknown and undocumented. This exploratory study aimed at assessing burnout while determining its prevalence and associated factors among medical resident doctors in Medina city, Saudi Arabia.

Methods

Study Setting and Sample

For this exploratory cross-sectional study, we invited all residents enrolled in the training programs run by the Saudi Commission for Health Specialties in Medina city, Saudi Arabia (n = 550). We collected the data during morning meetings and the academic half-day activities from all the training centers in the city. The study included all medical residents who joined the Saudi board program for more than 6 months. Those who joined the Saudi board program for less than 6 months or declined to participate were excluded. Sample size was calculated using G*Power software, version 3.1.9.7. With a confidence level of 95%, effect size of 0.15, and a power of 80%, the sample size was determined to be 343. An additional 15% was added to the calculated sample size to account for non-response, giving a finalized sample size of 394 respondents.

Study Instruments

For the collection of data, the study employed a printed self-administered structured questionnaire, which comprised 3 sections.

Section I included sociodemographic variables and work-related variables. The sociodemographic variables included variables like gender, age, marital status, history of chronic disease, past history of smoking tobacco, physical exercise, place of residence, depression medication, anxiety medication, etc. The work-related factors included variables such as specialty, level of training, years of experience, working hours per day, number of on-calls per month, number of clinics per week, number of patients seen per week, monthly income, etc.^10-13

Three validated questions from the Global Adult Tobacco Survey (GATS)¹⁴ assessed tobacco consumption, while the Godin leisure-time exercise questionnaire assessed the regularity of exercise.¹⁵

Section II employed the Maslach Burnout Inventory-Human Services Survey (MBI-HSS), which is a reliable, widely used validated tool for assessment of burnout. It addresses 3 dimensions of burnout: emotional exhaustion (EE), depersonalization (DP), and personal accomplishment (PA). It consists of 22 items within these 3 dimensions (each comprising 9, 5, and 8 items, respectively).²

We measured the responses on a 7-point Likert scale ranging from 0 (never) to 6 (every day). We then summed the scores and categorized them into “low,” “moderate,” and “high” in each subscale category. Lower scores regarding personal accomplishment predicted a greater likelihood of burnout.

We defined burnout as the presence of at least 1 of the following: (i) high score (27 and above) regarding EE; (ii) high score regarding depersonalization (13 and above); and (iii) low score regarding personal accomplishment (0-31).¹⁶

Among the dimensions, the scores for EE are the most crucial in evaluating burnout.¹⁶ All 3 subscales-EE, DP, and AP-showed high internal consistency, with Cronbach's alpha coefficient values of 0.837, 0.869, and 0.881, respectively.¹⁷

In section III, we assessed sources of stress with 19 items obtained from the literature. The following question headed these items: “To what extent do the following conditions cause stress to you?” Each item was scored from 1 (causing no stress) to 4 (causing severe stress).¹⁰ The Cronbach's alpha coefficient of these items in this study was 0.94.

Ethical Approval

We obtained ethical approval from the Institutional Review Board (IRB), General Directorate of Health Affairs in Madinah (IRB 393). We explained the objectives of the study verbally, accompanied by a written explanation attached to the questionnaires. Participant’s confidentiality and anonymity were assured. We obtained signed informed consent from the consenting participants.

Statistical Analysis

We performed descriptive and inferential data analyses using Statistical Package for the Social Sciences (SPSS) version 22.0 (IBM SPSS Corp.; Armonk, NY, USA). In the descriptive analysis, the mean±standard deviation (SD) was obtained for the continuous variables, while frequencies and percentages were obtained for the categorical variables. Age was categorized into 2 categories to determine which age group was more vulnerable to burnout. The independent samples t-test and analysis of variance (ANOVA) test were used to assess the differences between group means of the burnout subscales and the independent variables. Post hoc analysis was performed using the Bonferroni test if the ANOVA was significant. We performed a test of normality for each subscale of burnout. The Pearson correlation coefficients (r _p ) evaluated the associations between the burnout subscales and the sources of stress. We used multiple linear regression (backward method) to obtain the predictor model of burnout. Type I error rate was 0.05; P-values less than .05 (P < .05) were considered statistically significant.

Results

Sociodemographic and Work-Related Information of the Residents

A total of 426 resident doctors responded to the survey with completed questionnaires. The overall response rate was 77.45%. Slightly more than half of the participants were male (n = 231; 54.23%) and unmarried (n = 248; 58.22%). Those aged 27-30 made up the largest proportion (n = 258; 60.56%). Of the participants, 123 (28.87%) practiced family medicine, 87 (20.42%) pediatric medicine, and 63 (14.79%) practiced surgical specialties (general, orthopedic, urology, and neurosurgery) (Tables 1 and 2).

Table 1.

Sociodemographic Characteristics of the Residents (n = 426)

	n (%)
Age
≤30	380 (89.20)
>30	46 (10.80)
Gender
Male	231 (54.23)
Female	195 (45.77)
Marital status
Single/engaged/divorced	248 (58.22)
Married	178 (41.78)
Place of residence
Medina	393 (92.25)
Others	33 (7.75)
Chronic disease
Yes	59 (13.85)
No	367 (86.15)
Asthma
Yes	36 (8.45)
No	390 (91.55)
Diabetes Mellitus
Yes	7 (1.64)
No	419 (98.36)
Hypertension
Yes	6 (1.41)
No	420 (98.59)
Depression
Yes	20 (4.69)
No	406 (95.31)
Depression medication
Yes	19 (4.46)
No	407 (95.54)
Anxiety medication
Yes	23 (5.40)
No	403 (94.60)
Sleep disorder medication
Yes	25 (5.89)
No	401 (94.13)
Hours of sleep per day
<6 hours	106 (24.88)
6-8 hours	288 (67.61)
>8 hours	32 (7.51)
Smoking tobacco in the past
Yes	88 (20.66)
No	338 (79.34)
Smoking frequency (n = 88)
Daily	45 (51.14)
Not everyday	23 (26.14)
Not at all	20 (22.72)
Smoking years (n = 88)
<5 years	39 (44.32)
5-10 years	37 (42.05)
11-15 years	12 (13.63)
Physical exercise
Daily	54 (12.67)
Weekly	134 (31.45)
Never	238 (55.86)

Table 2.

Work-Related Factors of the Residents (n = 426)

Characteristics	n (%)
Specialty
Internal medicine	61 (14.32)
Surgical specialties (general, orthopedic, urology, neurosurgery)	63 (14.79)
Family medicine	123 (28.87)
Preventive medicine	24 (5.63)
Pediatric	87 (20.42)
Obstetrics and gynecology	30 (7.04)
Radiology	14 (3.29)
Ophthalmology	7 (1.64)
Other (ENT, forensic medicine, anesthesia, ICU, ER, psychiatry, neurology)	17 (3.99)
Level of training
Year1	133 (31.22)
Year 2	123 (28.87)
Year 3	94 (22.07)
Year 4	59 (13.85)
Year 5	17 (3.99)
Years of experience
≤2 years	229 (53.76)
2-5 years	165 (38.73)
>5 years	32 (7.51)
Working hours/day
≤8 hours	327 (76.76)
>8	99 (23.24)
Number of on-calls per month
≤4	99 (23.24)
>5	206 (48.36)
Not applicable	121 (28.40)
Number of weekends on-call per month
≤3	267 (62.68)
>4	22 (5.16)
Not applicable	137 (32.16)
Number of clinics per week
1-5	202 (47.42)
>5	110 (25.82)
Not applicable	114 (26.76)
Number of patients per week
≤30	272 (63.85)
>30	54 (12.68)
Not applicable	100 (23.47)
Monthly income
≤20 000	392 (92.02)
>20 000	34 (7.98)
Satisfaction with work–life balance
Yes	246 (57.75)
No	180 (42.25)
Ability to take annual occupational leave (30 days at one time)
Yes	290 (68.08)
No	136 (31.92)
If you were to go back in time, would you have chosen your specialty again?
Yes	287 (67.37)
No	139 (32.63)

Prevalence of Burnout Among the Residents

Among the resident doctors, 245 (57.51%) had high EE, 156 (36.62%) had high DP, and 55 (12.91) had low PA. Of them, 123 (28.87%) showed moderate EE, 139 (32.63%) showed moderate DP, and 143 (33.57%) showed moderate PA. While 78 participants (18.31%) scored high on all the 3 burnout subscales, 346 scored high on at least 1 subscale (81.22%), as shown in Table 3.

Table 3.

Prevalence of Burnout Among the Residents (n = 426)

	Low, n (%)	Moderate, n (%)	High, n (%)
EE	58 (13.62)	123 (28.87)	245 (57.51)
DP	131 (30.75)	139 (32.63)	156 (36.62)
PA	55 (12.91(	143 (33.57)	228 (53.52)

EE high: 27 or over, EE moderate: 17-26, EE low: 0-16; DP high: 13 or over, DP moderate: 7-12, DP low: 0-6; PA high: 0-31, PA moderate: 32-38, PA low: 39 or over (in the opposite direction from EE and DP).

Factors Associated with Burnout in the Univariate Analysis

To determine the factors associated with burnout in the univariate analysis, we used the total score of EE, DP, and PA as a continuous variable. The (mean (SD)) EE score was significantly higher among those aged more than 30 years (34.5 (10.5)), compared to those aged 30 years or less (28.1 (10.9)) (P < .001). Females also scored higher in this regard (30.7 (10.8)) compared to their male counterparts (27.2 (10.9)) (P < .001), while those who had diabetes mellitus (DM) (38.1 (7.9)) scored higher compared to those without it (28.7 (11.0)) (P = .023). Residents who suffered from depression (34.6 (11.1)) also scored higher, compared to those without it (28.4 (10.9)) (P = .007). Higher EE scores were recorded among those who took depression, anxiety, and sleep medication [(36.3 (19.5)), (34.6 (12.6)) and (33.8 (12.1)), respectively] compared to those who did not [(28.4 (10.9)), (28.4 (10.8)), and (28.4 (10.9)), with (P = .002), (P = .009), and (P = .019), respectively]. The number of hours of sleep significantly predicted EE (P < .001); on post hoc, those who slept less than 6 hours (32.8 (10.8)) reported higher rates of burnout compared to those who slept 6-8 hours (27.5 (10.4)) (P = .040) and those who slept more than 8 hours a day (26.6 (13.9)) (P = .044). Additionally, there was a significant difference in mean EE between those who exercised daily, weekly, or never exercised (P = .002); on the post hoc test, those who never engaged in physical exercise (30.4 (11.0)) suffered burnout at a higher rate compared to those who exercised weekly (25.8 (11.2)) (P = .041) and daily (27.0 (10.5)) (P = .043), as shown in Tables 4 and 5.

Table 4.

Sources of Stress (n = 426)

Item	Mean (SD)
COVID-19 pandemic	2.6 (1.0)
Too many tests/examinations	2.8 (0.9)
Large amount of content to learn	2.9 (0.8)
Time pressures and deadlines to meet	2.8 (0.9)
Too many promotion requirements	2.8 (0.7)
Work overload	2.8 (0.9)
Unfair assessment from superior	2.4 (1.0)
Fear of making mistakes	2.7 (0.9)
Work demands affecting personal/home life	2.6 (0.9)
Lack of time to review what has been learnt	2.7 (0.9)
Having difficulty grasping the content	2.3 (0.8)
Working with uncooperative colleagues	2.5 (0.9)
Inability to participate in decision-making	2.2 (0.9)
Unable to make full use of skills and ability	2.4 (0.9)
Life is too work-centered	2.5 (1.0)
Lack of authority to carry out job duties	2.2 (0.9)
Working with incompetent colleagues	2.3 (0.9)
Competition among colleagues	2.1 (1.0)
Difficulty in maintaining relationship with a superior	2.1 (0.9)

Table 5.

Differences in Mean Burnout Scores Between Groups of Sociodemographic Characteristics

	Burnout
EE		DP		PA
Variables	Mean (SD)	P	Mean (SD)	P	Mean (SD)	P
Age
≤30	28.1 (10.9)	<.001	10.2 (6.1)	.037	29.9 (8.1)	.349
>30	34.5 (10.5)	12.3 (7.2)	31.7 (7.4)
Gender		.001		.603		.194
Male	27.2 (10.9)		10.5 (6.1)		30.5 (7.9)
Female	30.7 (10.8)	10.2 (6.5)	29.5 (8.1)
Marital status		.989		.173		.965
Not married	28.8 (10.4)		10.7 (6.0)		30.1 (7.9)
Married	28.7 (11.8)	9.9 (6.6)	30.1 (8.2)
Place of residence
Medina	28.6 (10.9)	.273	10.4 (6.2)	.662	29.7 (7.8)	.002
Others	30.8 (11.3)	9.9 (7.3)	34.3 (9.1)
Chronic disease
Yes	29.1 (11.3)	.802	10.4 (6.9)	.974	30.6 (7.6)	.582
No	28.7 (10.9)	10.4 (6.2)	29.9 (8.1)
Asthma
Yes	27.2 (11.6)	.389	9.7 (6.5)	.465	28.9 (6.8)	.378
No	28.9 (10.9)	10.5 (6.2)	30.2 (8.1)
Diabetes Mellitus
Yes	38.1 (7.9)	.023	16.8 (8.6)	.006	33.1 (7.4)	.309
No	28.2 (11.0)	10.3 (6.2)	30.0 (8.0)
Hypertension
Yes	32.2 (12.1)	.447	17.3 (9.0)	.005	28.8 (5.4)	.704
No	28.7 (11.0)	10.3 (6.2)	30.1 (8.1)
Anxiety
Yes	34.6 (11.1)	.064	13.7 (7.6)	.060	34.3 (7.7)	.068
No	28.6 (10.9)	10.3 (6.2)	30.0 (8.0)
Depression
Yes	35.2 (10.5)	.007	12.5 (7.6)	.116	29.2 (8.8)	.639
No	28. (11.0)	10.3 (6.2)	30.1 (8.0)
Hypothyroidism
Yes	26.0 (0.0)	.722	10.0 (0.0)	.190	38.0 (0.0)	.163
No	28.8 (11.0)	10.4 (6.28)	30.0 (8.0)
Depression medication
Yes	36.3 (9.5)	.002	13.2 (7.2)	.045	28.5 (8.4)	.391
No	28.4 (10.9)	10.3 (6.2)	30.1 (8.0)
Anxiety medication
Yes	34.6 (12.6)	.009	12.2 (6.8)	.153	29.6 (10.0)	.836
No	28.4 (10.8)	10.3 (6.2)	30.1 (7.9)
Sleep disorder medication
Yes	33.8 (12.1)	.019	10.9 (5.8)	.669	29.7 (9.6)	.832
No	28.4 (10.9)	10.4 (6.3)	30.1 (7.9)
Hours of sleep per day
<6 hours	32.8 (10.8)	<.001	11.5 (6.6)	.107	30.4 (8.2)	.060
6-8 hours	27.5 (10.4)		9.9 (5.9)		30.3 (7.8)
>8 hours	26.6 (13.9)	10.4 (7.2)	26.8 (9.3)
Smoking frequency
Daily	27.3 (12.1)	.551	11.7 (5.8)	.053	29.3 (9.1)	.501
Not everyday	29.6 (7.7)		12.2 (4.7)		31.5 (9.3)
Not at all	28.9 (11.1)	10.0 (6.4)	30.1 (7.7)
Physical exercise
Daily	25.8 (11.2)		7.9 (4.9)		32.4 (8.9)
Weekly	27.0 (10.5)	.002	10.1 (6.0)	.004	31.1 (8.4)	.004
Never	30.4 (11.0)		11.1 (6.5)		28.9 (7.4)

EE, Emotional exhaustion; D, Depersonalization; PA, Personal accomplishment.

A significant difference in mean EE was found between the different groups of specialties (P < .001); post hoc, residents from the fields of internal medicine and pediatrics suffered from burnout at a higher rate than those from family medicine and preventive medicine (P = .044). Residents from the field of surgical specialties (general, orthopedic, urology, and neurosurgery), obstetrics and gynecology, and ophthalmology also reported higher rates of burnout compared to those from preventive medicine (P < .001), whereas those from radiology and other specialties (ENT, forensic medicine, anesthesia, ICU, ER, psychiatry, and neurology) showed no significant difference (P > .05). There was an overall significant difference in mean EE between the year 1, year 2, year 3, and years 4-5 level of training (P < .001); on post hoc, residents at the year 4 and year 5 levels of training (32.9 (11.5)) reported burnout at a higher rate compared to their counterparts in year 1 (25.1 (11.3)) (P < .001), year 2 (29.0 (10.1)) (P = .062), and year 3 (30.3 (9.8)) (P = .066). There was a significant difference in mean EE between those who had less than 2 years, 2-5 years, and more than 5 years of experience (P = .015). Multi-comparison analysis on post hoc showed that participants with less than 2 years of experience reported lower EE burnout compared to those with 2-5 years of experience (P = .044) and those with more than 5 years of experience (P = .058). The EE score was significantly higher among residents who worked more than 8 hours (30.9 (11.9)) compared to those who worked less than 8 hours (28.1 (10.7)) (P = .037). The same was evident among residents who treated more than 30 patients per day (32.2 (12.4)) as compared to their counterparts who treated less than 30 patients per day (28.2 (10.4)) (P = .044). Finally, participants who were not satisfied with their work–life balance (34.8 (9.8)) reported higher EE scores than those who were (23.4 (9.7)) (P < .001).

The EE score was significantly lower among residents who were able to take annual occupational leave (30 days at one time) (27.9 (10.8)) compared to residents who were not (P = .030). The same was evident among participants who said that they would choose the same specialty again if they could go back in time (26.1 (10.3)), in contrast to those who said that they would not (P < .001), as shown in Table 6.

Table 6.

Differences in Mean Burnout Scores Between Groups of Work-Related Factors

	Burnout
EE		DP		PA
Variables	Mean (SD)	P	Mean (SD)	P	Mean (SD)	P
Specialty
Internal medicine	31.1 (9.2)	<.001	11.7 (5.2)	<.001	29.2 (9.1)	.045
Surgical specialties (general, orthopedic, urology, neurosurgery)	29.4 (11.4)		11.9 (5.9)		32.6 (7.7)
Family medicine	25.7 (10.0)		9.8 (5.2)		28.9 (7.5)
Preventive medicine	18.4 (11.0)		4.8 (5.4)		29.4 (8.8)
Pediatrics	32.3 (9.8)		10.8 (7.1)		30.4 (7.4)
Obstetrics and gynecology	31.9 (12.4)		12.1 (5.3)		31.2 (8.7)
Radiology	26.6 (8.6)		8.2 (6.2)		25.6 (6.4)
Ophthalmology	35.8 (10.3)		12.0 (9.2)		31.6 (10.2)
Other (ENT, forensic medicine, anesthesia, ICU, ER, psychiatry, neurology)	29.8 (14.1)		8.2 (9.0)		31.9 (7.8)
Level of training
Year 1	25.1 (11.3)	<.001	9.2 (5.8)	.029	28.5 (8.9)	.033
Year 2	29.0 (10.1)		10.5 (5.9)		30.1 (7.2)
Year 3	30.3 (9.8)		10.9 (6.5)		31.1 (7.5)
Year 4 and Year 5	32.9 (11.5)		11.7 (7.0)		31.5 (7.9)
Years of experience
<2 years	27.3 (10.9)	.015	10.0 (5.7)	.452	29.3 (8.1)	.097
2-5 years	30.3 (10.9)		10.8 (6.7)		30.9 (7.9)
>5 years	31.0 (10.8)		10.7 (7.5)		31.1 (7.7)
Working hours per day
≤8 hours	28.1 (10.7)	.037	10.1 (6.0)	.075	29.7 (7.8)	.073
>8 hours	30.9 (11.9)		11.4 (6.9)		31.3 (8.6)
Number of on-calls per month
≤4	28.7 (9.5)	<.001	10.5 (6.1)	<.001	30.7 (7.9)	.641
>5	31.5 (11.0)		11.6 (6.5)		30.0 (8.3)
No on-call	24.2 (10.7)		8.2 (5.3)		29.7 (7.5)
Number of weekends on-call per month
≤3	30.7 (10.8)	<.001	11.3 (6.4)	<.001	30.6 (8.1)	.027
>4	30.0 (10.2)		11.4 (5.3)		26.0 (9.4)
No weekend on-call	24.8 (10.6)		8.4 (5.7)		29.6 (7.6)
Number of clinics per week
1-5	29.2 (11.2)	.229	11.3 (6.1)	.022	30.9 (7.9)	.077
>5	27.2 (10.2)		9.5 (6.1)		30.0 (7.6)
No clinic	29.4 (11.3)		9.7 (6.6)		28.7 (8.5)
Number of patients per day
≤30	28.2 (10.4)	.044	10.8 (6.2)	.077	31.1 (7.5)	.001
>30	32.2 (12.4)		10.6 (6.3)		29.7 (8.6)
Not applicable	28.5 (11.5)		9.2 (6.3)		27.6 (8.6)
Monthly income (Saudi Riyal)
≤20 000	28.5 (11.0)	.132	10.2 (6.3)	.096	30.0 (8.0)	.542
>20 000	31.5 (11.0)		12.1 (6.0)		30.9 (8.1)
Satisfaction with work–life balance
Yes	23.4 (9.7)	<.001	8.9 (5.4)	<.001	30.6 (8.0)	.114
No	34.8 (9.8)		12.4 (6.8)		29.3 (8.0)
Ability to take annual occupational leave (30 days at one time)
Yes	27.9 (10.8)	.030	9.9 (5.8)	.068	30.2 (7.7)	.728
No	30.4 (11.4)		11.3 (7.1)		29.9 (8.7)
If you were to go back in time, would you have chosen your specialty again?
Yes	26.1 (10.3)	<.001	9.4 (5.5)	<.001	30.5 (8.2)	.099
No	34.3 (10.3)		12.5 (7.1)		29.1 (7.6)

EE, Emotional exhaustion; D, Depersonalization; PA, Personal accomplishment.

All 19 sources of stress in this study correlated positively and significantly with EE burnout, with the r _p ranging from 0.20 to 0.39 (P < .001), as shown in Table 7.

Table 7.

Correlation Between Burnout and Sources of Stress

Item	EE		DP		PA
r p	P	r p	P	r p	P
COVID-19 pandemic	0.21	<.001	0.09	.079	0.02	.661
Too many Tests/examinations	0.26	<.001	0.08	.089	-0.05	.315
Large amount of content to learn	0.24	<.001	0.01	.831	-0.07	.124
Time pressures and deadlines to meet	0.23	<.001	0.01	.811	0.01	.979
Too many promotion requirements	0.29	<.001	0.12	.017	-0.09	.043
Work overload	0.39	<.001	0.13	.010	-0.02	.731
Unfair assessment from superior	0.31	<.001	0.17	<.001	-0.04	.425
Fear of making mistakes	0.28	<.001	0.09	.055	-0.06	.218
Work demands affecting personal/home life	0.30	<.001	0.18	<.001	-0.11	.028
Lack of time to review what has been learnt	0.29	<.001	0.09	.077	-0.05	.280
Having difficulty grasping the content	0.27	<.001	0.15	.002	-0.12	.012
Working with uncooperative colleagues	0.24	<.001	0.06	.261	0.05	.345
Cannot participate in decision making	0.24	<.001	0.11	.026	-0.02	.730
Unable to make full use of skills and ability	0.21	<.001	0.13	<.007	-0.14	.004
Life is too work-centered	0.32	<.001	0.18	<.001	-0.06	.241
Lack of authority to carry out job duties	0.32	<.001	0.22	<.001	-0.07	.075
Working with incompetent colleagues	0.25	<.001	0.19	<.001	0.01	.781
Competition among colleagues	0.24	<.001	0.24	<.001	-0.08	.101
Difficulty in maintaining relationship with a superior	0.35	<.001	0.27	<.001	-0.08	.120

r, correlation coefficient; EE, Emotional exhaustion; D, Depersonalization; PA, Personal accomplishment.

Depersonalization was higher among those more than 30 years of age (12.3 (7.2)) compared to those aged less than 30) 10.2 (6.1)) (P = .037). Residents living with DM (16.8 (8.6)) and those with hypertension (HTN) (17.3 (9.0)) also reported DP at a higher rate compared to those with neither DM (10.3 (6.2)) nor HTN (10.3 (6.2)) (P = .006), (P = .005), respectively. This was also true among residents who took depression medication (13.2 (7.2)) compared to those who did not (10.3 (6.2)) (P = .045). Physical exercise was significantly associated with DP (P = .004). Multi-comparison analysis post hoc showed those who never performed physical exercise (11.1 (6.5)) reported DP at a higher rate, in contrast to those who engaged in daily exercise (7.9 (4.9)) (P = .039) and weekly exercise (10.1 (6.0)) (P = .054), as shown in Table 5.

There was a significant difference in mean DP between the different groups of specialties (P < .001); post hoc analysis showed that residents from the preventive medicine specialty recorded lower rates of burnout compared to those from internal medicine, surgical subspecialties, pediatrics, and obstetrics and gynecology (P = .038). Residents from the field of family medicine, radiology, ophthalmology, and other specialties (ENT, forensic medicine, anesthesia, ICU, ER, psychiatry, and neurology) also reported higher rates of burnout compared to those from preventive medicine, but no significant differences were observed (P > .05). The participants’ academic year significantly predicted burnout (P = .029); post hoc showed that residents at the year 4 and year 5 levels of training suffered burnout at a higher rate compared to residents in year 1 (P = .042), year 2 (P = .056), and year 3 (P = .058). The DP score among those who were not satisfied with their work–life balance (12.4 (6.8)) significantly exceeded that of their satisfied counterparts (8.9 (5.4)) (P < .001). The DP score was significantly lower among residents who said that they would choose the same specialty again if they could go back in time (9.4 (5.5)), compared to those who said that they would not (12.5 (7.1)) (P < .001), as shown in Table 6.

A significant positive correlation was evident between DP and 13 out of the 19 sources of stress (r _p ranged from 0.10 to 0.26) (P < .001), as shown in Table 7.

There was an overall significant difference in mean personal accomplishment (PA) between those who exercised daily, weekly, or never exercised (P = .004); the post hoc test showed that those who never performed physical exercise (28.9 (7.4)) had a lower PA score compared to those who exercised daily (32.4 (8.9)) (P = 0.044) and weekly (31.1 (8.4)) (P = .044), as shown in Table 5.

There was a significant difference in mean PA between the different groups of specialties (P = .040). However, multi‑comparison analysis, post hoc, showed negligible disparity in the rate of burnout among the specialties (P > .05). There was an overall significant difference in mean PA between the year 1, year 2, year 3, and years 4-5 levels of training (P = .033); but multi‑comparison analysis, post hoc, showed negligible disparity in the rate of PA burnout among the training levels (P > .05). Residents who took more than 4 weekends on-call per month (26.0 (9.4)) reported lower PA compared to those who had less than 3 weekends on-call per month (30.6 (8.1)) and those had no on-call shifts at all (29.6 (7.6)) (P = .027), as shown in Table 6.

A statistically significant negative correlation was found between PA and 4 out of the 19 sources of stress. These sources included arduous promotion requirements, work demands affecting personal/home life, difficulty in grasping the content, and the inability to make full use of their skills and ability (r _p : −0.09, P = .043; r _p : −0.11, P = .028; r _p : −0.12, P = .012; and r _p : −0.14, P =.004, respectively), as shown in Table 7.

Factors Associated with Burnout in the Multivariate Analysis

In the multivariate analysis, all variables that showed a statistically significant relationship with burnout in the univariate analysis were included in the regression analysis. We used all the 3 subscales of burnout as continuous variables, and 3 regression models were obtained.

The significant predictors of EE included anxiety medication (P = .001), lack of physical exercise (P = .019), level of training (P < .001), having less than 3 weekends on-call per month (P < .001), and dissatisfaction with work–life balance (P < .001). Time pressures and deadlines (P = .026) and work overload (P < .001) also predicted EE, as well as the inability to participate in decision-making (P = .030). Finally, the inability to make full use of their skills and abilities (P = .042), a work-centered life (P = .026), and difficulty in maintaining relationship with their superiors (P < .001) significantly predicted EE. The total model was significant (P = .003) and explained 25% of the variance.

Significant predictors of DP included weekly physical exercise (P = .013), having more than 5 on-call shifts per month (P = .016), having 1 to 5 clinics per week (P < .001), and dissatisfaction with work–life balance (P < .001). Suffering from HTN (P = .007) and difficulty in maintaining relationship with superiors (P < .001) also predicted DP. The total model was significant (P = .005) and explained 25.6% of the variance.

Lack of physical exercise (P < .001), number of weekends on-call per month (P = .025), and the treatment of more than 30 patients per day (P = .002) significantly predicted low PA. The same was true for the participants’ level of training (P < .001) and their inability to make full use of skills and abilities (P = .008), as shown in Table 8. The total model was significant (P = .023), and explained 21.30% of the variance.

Table 8.

Factors Associated with Burnout in Multivariate Analysis

	B	SE	P	95% CI
Emotional exhaustion
Medication last year for anxiety	5.98	1.84	.001	2.36-9.59
Physical exercise
Never	2.01	0.86	.019	0.33-3.69
Daily (reference)
Level of training
Year 2	3.85	1.08	<.001	1.72-5.97
Year 3	4.36	1.17	<.001	2.06-0.66
Year 4 and Year 5	5.74	1.26	<.001	3.25-8.23
Year 1 (reference)
Number of weekends on-call per month
≤3	3.53	0.89	<.001	1.79-5.27
No weekend on-call (reference)
Satisfaction with work–life balance	6.96	0.92	<.001	5.16-8.77
Time pressures and deadlines to meet	1.31	0.59	.026	0.58-2.465
Work overload	3.41	0.59	<.001	2.23-4.58
Cannot participate in decision-making	1.34	0.61	.030	0.13-2.54
Unable to make full use of skills and ability	1.34	0.65	.042	0.05-2.63
Life is too work-centered	1.21	0.54	.026	0.15-2.27
Difficulty in maintaining relationship with a superior	2.28	0.55	<.001	1.20-3.35
Depersonalization
Physical exercise
Never	2.56	0.87	.003	0.85- 4.27
Weekly	2.30	0.93	.013	0.48-4.12
Daily (reference)
Number of on-calls per month
≤4	1.99	0.83	.020	0.36-3.62
>5	1.86	0.77	.020	0.35-3.38
No on-call (reference)
Are you satisfied with work–life balance?	2.47	0.64	<.001	1.22-3.72
HTN	6.45	2.36	.007	11.09-1.81
Difficulty in maintaining relationship with a superior	1.43	0.31	<.001	0.83-2.03
Number of clinics per week
1-5	2.61	0.69	<.001	1.26-3.97
No clinic (reference)
Personal accomplishment
Physical exercise
Never	-2.427	0.75	.001	-3.9-0.94
Daily (reference)
Level of training
Year 2	2.31	0.96	.016	0.43-4.20
Year 3	3.06	1.03	.003	1.03-5.09
Year 4 and Year 5	3.46	1.10	.002	1.29-5.63
Year 1 (reference)
Number of weekends on-call per month
≤3	1.73	0.77	.025	0.22-3.25
No weekend on-call (reference)
Unable to make full use of skills and ability	-1.14	0.43	.008	-1.99-0.29
Number of patients per day
>30	2.40	0.78	.002	0.87-3.93
Not applicable (reference)

B, Regression Coefficient; SE, Stander Error; 95% CI, 95% Confidence Interval.

Discussion

The primary aim of this study was to assess the prevalence and associated factors of burnout among resident doctors in Medina city, Saudi Arabia. This study found that 81.22% of the residents reported a high score on at least 1 burnout subscale and 18.31% scored high on all the subscales. Of the participants, 57.51% and 36.62% reported high EE and DP, respectively. Meanwhile, 53.52% reported low PA. Studies have shown marked discrepancies in the prevalence of burnout across countries and regions worldwide. The prevalence rates in Western countries ranged between 27% and 74%,^18-20 while studies from Saudi Arabia reported burnout prevalence to range between 25.2% and 70%.^5,13,21,22 Specifically, a study from Riyadh found high prevalence of EE and DP, at 54% and 35% respectively, while low PA remained at 33%.⁵ A study conducted in Malaysia revealed an EE rate of 36.6% among resident doctors.¹¹

A plausible explanation of such inconsistencies on the prevalence rates across different studies could be the variations in patient or organizational workplace culture in different populations.¹³ These variations could also be attributed to the utilization of different measurement scales or tools in different populations or settings, causing threats to measurement validity to over- or underestimate the true prevalence of burnout.¹³ Most tools that measured burnout were only validated or had a suitable content validity to the Western population. Operationalization of burnout “case” adopted in different studies would definitely compromise the true estimate of burnout prevalence⁷ if adopted across different cultures or populations.

Regarding contingent factors, this study found that high EE was significantly associated with the older age group, female gender, DM, and depression. Demographic attributes such as respondents’ age or gender have shown inconsistent associations with burnout phenomenon in previous studies.^23,24 These variations could somewhat be influenced by cultural contexts within the population being studied or the workplace organization structure across different settings.²³ It is also associated with medications for anxiety, depression, and sleep disorders, as well as insufficient sleep and lack of physical exercise. Therefore, those who have underlying medical conditions appear to be associated with a statistically significant higher rate of burnout. However, the cause of this needs to be further elucidated in future studies.

The association between HTN, DM, and burnout may be due to the emotional burden of managing these diseases.

In the study conducted in Riyadh, they found significant associations between EE, job satisfaction, and old age, while DP was associated with taking psychotropic drugs. In the same study, PA was significantly associated with marital status and smoking tobacco or water pipe.²⁵ However, another study found that older physicians were less likely to report EE than young physicians. Additionally, older physicians were less likely to have low PA scores than those who were under 29 years of age.²⁶ Thus, there is a disagreement regarding the association between older age and burnout.

This study found that low PA was associated with less physical exercise. Two previous studies from Iran and the United States found that residents who were physically active were less likely to experience burnout.^27,28 Another study conducted in Riyadh found that participants who experienced burnout suffered significantly from sleep deprivation,⁵ while another study in the same city found no such association.²⁹ In multivariate analysis, we found that EE was associated significantly with the level of training, time pressures, working hours, the number of on-call shifts, difficulty in maintaining relationship with a superior, satisfaction with work, and annual occupational leave. In contrast, DP was associated with the number of on-call shifts, difficulty in maintaining relationship with a superior, and satisfaction with work. Low PA was associated with level of training, the number of weekends on-call, and the inability to make full use of skills and ability. Previous studies have reported similar results.^21,25

Regarding specialty, we found that residents from preventive medicine reported burnout at a lower rate compared to those from internal medicine, the surgical subspecialties, pediatrics, obstetrics, gynecology, and ophthalmology. A meta-analysis study showed that radiology, neurology, and general surgery were the top 3 specialties with the highest prevalence of burnout, while psychiatry, oncology, and family medicine had the lowest prevalence.⁴ However, a previous study from Riyadh found that the prevalence of burnout showed no statistical disparity across specialties.²⁹ Despite such studies suggesting the absence of specialty differences in burnout prevalence,²⁹ the present study would recommend that this postulation may be more of an anomaly, and that the specialty differences in prevalence reported in the meta-analysis⁴ do appear to align with the findings of the current study.

Regarding sources of stress, a meta-analysis study found that workplace-related factors, such as poor work environment, excessive work demands, and poor work–life balance were statistically significantly associated with burnout.³⁰

Moreover, many studies have found a positive and significant correlation between sources of stress and burnout.^16,10,25,29

Consequently, there is consensus among studies regarding the association between burnout and sources of stress in the workplace. This emphasizes the importance of stress management in the workplace. The current pandemic has overwhelmed healthcare systems. However, it has catalyzed most countries worldwide to implement strict mitigation and suppression measures to flatten the epidemic curve.^31-34 Medical residents, being at the frontline, were tasked for deployment to complement shortage of healthcare resources. These circumstances may pose escalated burden to medical residents, being emotionally fatigued and facing burnout,³⁵ and thus has the tendency to cause serious mental health repercussions. Although our exploration on a real-time attribute, the COVID-19 pandemic, showed statistical significance with burnout in the univariate level was somewhat eliminated at the multivariate model. A recent study from Egypt found burnout impact higher in healthcare professionals managing COVID-19.³⁵

Residents in training usually have many responsibilities. Over time, their progression through the healthcare system from learners to providers, and their patient load and responsibilities increase. These changes contribute to the high prevalence of burnout.

Study Strengths and Limitations

This study has several strengths. First, it had a good response rate (77.45%). Second, our study assessed burnout among medical resident doctors in all specialties in Medina city, contrary to other studies that focused on one or a few single specialties. Third, this was a multicenter study, which made it possible to generalize our results to resident doctors elsewhere in Saudi Arabia. Fourth, we studied many sources of stress and many work-related factors.

Our study is limited by participant location (one city), and was a cross-sectional study using self-reported data. Because of the cross-sectional nature of our study, we cannot predict causation from our data and can only study associations between variables. The self-reported data may be subject to social desirability or recall bias. We note that, as residency trainees are assigned to different rotations throughout their program during the conduct of this study, they may have experienced different levels of clinical workload, which would influence their attitude toward work and questionnaire responses. The wide 95% CI gaps on certain variables could principally be due to lack of sample size, as the sample size was calculated for prevalence as a whole and not for individualized hypothesized attributes. Finally, we used many independent variables adapted from previous literature, with Likert responses, thus increasing the variation of apprehension among the respondents and causing wider CI gaps on certain correlated variables.

In conclusion, our study found that the prevalence of burnout among residents was relatively high (81.22%). We identified numerous factors associated with burnout, particularly sources of stress in the work environment and dissatisfaction with work–life balance. Therefore, the Saudi Health Commission for Health Specialties and the program directors should act to improve working conditions, the work environment, and work–life balance, to help minimize burnout among resident doctors. There is a need for comprehensive assessment of burnout and stress in medical residents at different levels and specialties, which may help diagnose such problems earlier and trace them to better solutions. To guarantee personal well-being, which is critical for the successful education of the next generation of physicians, a priority focus should be placed on resident physicians’ well-being to provide optimal care for others.