Participants. Sociodemographic characteristics of the sample are summarized in Table 1.
HAWs from 52 aid organizations responded to the survey with less than 15% incomplete data (N=111). The sample included 64 men (57.7%) and 47 women (42.3%), with a mean age of 37.4 years old (SD = 10.7). Most of them were Bangladeshi HAWs (n = 99, 89.2%). Approximately two-fifths of the sample came from national NGOs (n = 47, 42.3%) and international NGOs (n = 55, 37.7%), while one-fifth came from UN or IFRC agencies (19.8%, n = 22). Participants had spent an average of 47 months (SD = 69.1) working in their current organization and 75 months (SD = 98.3) in the humanitarian sector. About half of the sample were middle-managers (n = 55, 49.5%), while approximately one-quarter were full-time staff (n = 26, 23.4%) or executive leaders (n = 30, 27.0%). Participants came from a variety of aid sectors, including: Emergency and Disaster Management and Response; Health; Monitoring, Evaluation, Accountability and Learning; Protection (Child Protection and Gender-Based Violence); and Water, Sanitation, and Hygiene sectors. Most of the participants worked directly with persons of concern (n = 80, 72.1%), and about half lived at the duty station (n = 51, 45.9%). Close to a third of the participants (n = 32, 28.8%) reported having a previous history of mental disorder.
Table 1
Sociodemographic Characteristics of Participants
Sociodemographic characteristic
|
Full sample (N = 111)
|
n
|
%
|
Sex
|
|
|
Male
|
64
|
57.7
|
Female
|
47
|
42.3
|
Nationality
|
|
|
Bangladeshi
|
99
|
89.2
|
International
|
12
|
10.8
|
Job Level
|
|
|
Full-time Aid Worker
|
26
|
23.4
|
Middle-manager
|
55
|
49.5
|
Executive Leadership
|
30
|
27.0
|
Organization Type
|
|
|
UN Agencies/IFRC Societies
|
22
|
19.8
|
International NGO
|
42
|
37.8
|
National NGO
|
47
|
42.3
|
Direct work with persons of concern a
Lives at duty station a
|
80
51
|
72.1
45.9
|
Previous history of mental disorder a,b
|
32
|
28.8
|
a Reflects participants answering “Yes” to this question.
b n = 7 preferred not to disclose their psychiatric history and were coded as missing values.
Burnout and Psychological Distress. Based on the cut-offs in the literature, 30.6% (n = 34), 16.4% (n = 18), 12.7% (n = 14), and 8.2% (n = 9) screened positive for moderate psychological distress, emotional exhaustion, depersonalization, and severe psychological distress, respectively. 8.2% (n=9) met the cut-offs for both emotional exhaustion and depersonalization. HAWs who reported a history of mental disorder reported higher scores of psychological distress (M = 9.1, SD = 5.2, N=32) than those who did not have a history of mental disorder (M = 5.7, SD = 3.9, N =71) with medium effect size, t (101) = -3.68, p < .001, d = .78. Other demographic variables did not have a significant effect on burnout and psychological distress.
Exposure to Adversity. The median number of potentially stressful life events experienced was 14 out of 28 (SD = 6.5). Specifically, the top five most frequently cited perceived adversities were: worrying about the well-being of family members (86.6%); being separated from family members due to work responsibilities (81.4%); threats of life-threatening or deadly diseases (e.g., COVID-19) (80.4%); travel difficulties (78.4%); and feeling isolated (74.2%) (Additional file 2). Staff with a history of mental disorder also endorsed higher frequency of exposure to common adversities overall (M = 52.5, SD = 11.2, n = 25) compared to staff without a mental disorder history (M = 45.9, SD = 10.2, n = 66), with medium effect size, t(89) = -2.67, p = .009, d = .63.
Workplace Psychosocial Stressors. The mean number of workplace stressors experienced was 17 out of 28 (SD = 6.5). In terms of offensive behaviors at work, 6.1%, 13.3%, and 34.7% reported experiencing some form of sexual harassment, violence, and bullying at work, respectively. Close to half of the sample (n = 47, 48.5%) reported experiencing discrimination at work, including discrimination due to gender (n = 19, 19.6%), age (n = 10, 10.3%), race/ethnicity (n = 6, 6.2%), nationality (n = 6, 6.2%), sexual orientation (n = 4, 4.1%), pregnancy/parenthood (n = 4, 4.1%), and mental health (n = 1, 1.0%). Exposure to workplace stressors did not differ across demographic groups.
Coping Styles. Participants endorsed more use of task-focused coping strategies (M = 24.93, SD = 5.30), followed by avoidance-focused (M = 19.73, SD = 5.17), and negative emotion-focused coping strategies (M = 18.92, SD = 6.26). Staff with psychiatric illness history reported greater use of negative emotion-focused coping strategies (M = 21.8, SD = 7.1, N = 24) than those who did not have psychiatric history (M = 17.8, SD = 5.7, N = 68), with medium effect size, t(88) = -2.76, p = .007, d = .66. Task- and avoidance-focused styles were not significantly associated with any demographic variables.
Psychosocial Model. Correlations between variables are reported in Table 2. As avoidance-focused coping and task-focused coping were not significantly associated with either psychological health outcomes in bivariate analyses, only negative emotion-focused coping was tested as an intervening variable in the subsequent path models. As bivariate analyses found that past psychiatric history was significantly associated with study variables, we controlled for history of mental illness in subsequent path models.
Table 2
Descriptive Statistics and Pearson’s Correlations for Main Variables
Scale
|
MBI-HSS
|
K-6
|
Adversity Exposure
|
COPSOQ-III
|
Avoidance-focused Coping
|
Emotion-focused Coping
|
Task-focused coping
|
MBI-HSS
|
–
|
|
|
|
|
|
|
K-6
|
.63**
|
–
|
|
|
|
|
|
Adversity Exposure
|
.42**
|
.39**
|
–
|
|
|
|
|
COPSOQ-III
|
.57**
|
.65**
|
.50**
|
–
|
|
|
|
Avoidance-focused coping
|
.15
|
.13
|
.17
|
.10
|
–
|
|
|
Emotion-focused coping
|
.46**
|
.60**
|
.48**
|
.54**
|
.22*
|
–
|
|
Task-focused coping
|
.06
|
.03
|
.13
|
.08
|
.40**
|
.25*
|
–
|
N
|
110
|
110
|
97
|
97
|
96
|
96
|
96
|
Mean
|
21.0
|
6.8
|
48.0
|
52.4
|
19.7
|
18.9
|
24.9
|
SD
|
15.6
|
4.6
|
11.1
|
12.8
|
5.2
|
6.3
|
5.3
|
Actual Range
|
0-78
|
0-24
|
27-81
|
30-93
|
7-31
|
7-35
|
7-34
|
α
|
.90
|
.85
|
.90
|
.92
|
.79
|
.88
|
.85
|
Note. MBI-HSS: Maslach Burnout Inventory - Human Services Scale. K-6: Kessler-6 scale. COPSOQ-III: Third Copenhagen Psychosocial Questionnaire (Adapted).
*p < .05, two-tailed. **p < .001, two-tailed.
After comparing model fit statistics (Table 3), the psychosocial model for burnout was preferred (Figure 2). The model with burnout as the outcome (i.e., distress as an intervening variable) had good exact fit, X2(2) = 0.733, p < .693, while the model with distress as the outcome had poor exact fit indicated by significant chi-square value, X2(3) = 13.423, p < .004. The model for burnout also had better approximate fit indices than the model for distress, as indicated by higher CFI (1.027 vs. .957), lower RMSEA (.000 vs. .178), and lower SRMR (.011 vs. .049). The decomposition of effects of the integrated model for burnout is summarized in Table 4 (see ‘Additional file 3’ for rejected model).
Table 3
Fit Statistics for Psychosocial Models for Burnout and Distress
Fit Statistic
|
Outcome
|
Burnout
|
Psychological Distress
|
χ²M
|
0.733
|
13.423
|
df M
|
2
|
3
|
p
|
.693
|
.004
|
CFI
|
1.027
|
.957
|
RMSEA [90% CI]
|
.000 [.000, .140]
|
.178 [.089, .279]
|
SRMR
|
.011
|
.049
|
Note. Fit statistics are reported for final models with non-significant paths removed. χ² M: likelihood ration chi-square. df M: model degrees of freedom. CFI: Comparative Fit Index. RMSEA: Root Mean Square Error of Approximation. CI: confidence interval. SRMR: Standardized Root Mean Residual.
Table 4
Decomposition of Effects from Psychosocial Model for Burnout
Effect
|
β
|
B
|
SE
|
95% CI
[LL, UL]
|
p
|
Direct Effects
|
Adversity Exposure à EmoCope
|
.28
|
0.18
|
0.06
|
[0.05, 0.29]
|
.002
|
Workplace Stressors à EmoCope
|
.41
|
0.20
|
0.05
|
[0.12, 0.30]
|
< .001
|
Adversity Exposure à Distress
|
-.01
|
-0.01
|
0.05
|
[-0.09, 0.09]
|
.927
|
Workplace Stressors à Distress
|
.45
|
0.16
|
0.04
|
[0.09, 0.24]
|
< .001
|
EmoCope à Distress
|
.33
|
0.24
|
0.06
|
[0.13, 0.37]
|
< .001
|
Adversity Exposure àBurnout
|
.17
|
0.27
|
0.13
|
[0.01, 0.53]
|
.039
|
Workplace Stressors à Burnout
|
.31
|
0.38
|
0.15
|
[0.10, 0.69]
|
.007
|
Distress à Burnout
|
.36
|
1.23
|
0.47
|
[0.30, 2.14]
|
.010
|
Indirect Effects
|
Adversity Exposure à Burnout via EmoCope and Distress
|
.05
|
0.05
|
0.03
|
[0.01, 0.14]
|
.006
|
Adversity Exposure à Burnout via
Distress
|
-.01
|
-0.01
|
0.06
|
[-0.13, 0.12]
|
.871
|
Workplace Stressors à Burnout via EmoCope and Distress
|
.06
|
0.05
|
0.04
|
[0.01, 0.17]
|
.005
|
Workplace Stressors à Burnout via
Distress
|
.20
|
0.19
|
0.10
|
[0.05, 0.43]
|
.006
|
EmoCope à Burnout via Distress
|
.12
|
0.30
|
0.15
|
[0.06, 0.66]
|
.007
|
Adversity Exposure à Distress via
EmoCope
|
.09
|
0.04
|
0.02
|
[0.01, 0.09]
|
.004
|
Workplace Stressors à Distress via
EmoCope
|
.14
|
0.05
|
0.02
|
[0.02, 0.09]
|
< .001
|
Total Effects
|
Adversity Exposure à Burnout
|
.20
|
0.31
|
0.15
|
[0.03, 0.59]
|
.032
|
Workplace Stressors à Burnout
|
.52
|
0.64
|
0.12
|
[0.42, 0.90]
|
< .001
|
EmoCope à Burnout
|
.12
|
0.30
|
0.15
|
[0.06, 0.66]
|
.007
|
Distress à Burnout
|
.34
|
1.23
|
0.47
|
[0.30, 2.14]
|
.010
|
Outcome Variable
|
R2
|
SE
|
95% CI [LL, UL]
|
p
|
Negative Emotion-Focused Coping
|
.38
|
.07
|
[0.22, 0.51]
|
.001
|
Psychological Distress
|
.46
|
.10
|
[0.26, 0.63]
|
.001
|
Burnout
|
.50
|
.08
|
[0.31, 0.64]
|
.002
|
Note. N = 111, n = 5,000 bootstrap replications. SE: bootstrap standard error. EmoCope: negative emotion-focused coping. CI: bias-corrected bootstrap confidence interval. LL: lower limit. UL: upper limit.
Adversity exposure and workplace stressors had overlapping but distinct pathways to burnout. Both types of stressors affected burnout directly, and indirectly through negative emotion-focused coping, while controlling for psychiatric history. The total indirect effect of both stressors on burnout through negative emotion-focused coping was β = .12, 95% CI [.06, .66], p = .007. However, workplace stressors but not adversity exposure had a direct effect on psychological distress (β = .45, 95% CI [0.09, 0.24], p = < .001 vs. β = -.01, 95% CI [-0.09, 0.09], p = .927).
In addition to having more direct and indirect pathways to burnout, the effects of workplace stressors also had the greatest magnitude. The total effect from workplace stressors to burnout was β = .52, SE = 0.12, 95% CI [0.42, 0.90], p = < .001, which was of greater than the total effects from psychological distress to burnout, β = .34, SE = 0.47, 95% CI [0.30, 2.14], p = .010, and adversity exposure to burnout, β = .20, SE = 0.15, 95% CI [0.03, 0.59], p = .032.
The model explained 38% of the variance in emotion-focused coping (SE = 0.07, 95% CI [0.22, 0.51], p = .001); 46% of the variance in psychological distress (SE = 0.10, 95% CI [0.26, 0.63], p = .001); and 50% of the variance in burnout (SE = 0.08, 95% CI [0.31, 0.64], p = .002).