Sociodemographic characteristics
Among the 1131 nurses, the majority were female (1050/92.8%); the average age was 30.74 ± 6.34 years, and more than three- quarters were aged 25-44 (901/79.7%). Approximately three-quarters (847/74.9%) had a bachelor’s degree, about a quarter (268/23.7%) had not chosen their nursing specialty on their own, and nearly three-quarters (826/73.0%) were married. About two-fifths (471/41.6%) had worked for more than ten years, and a minority (16/1.4%) held the professional title of associate chief nurse or above. All of the participants were clinical nurses from tertiary hospitals, and their departments were mainly emergency (286/25.3%) and general wards (453/40.1%). Univariate analyses revealed that age, education level, self-directed choice of nursing, single status and work unit were significantly associated with turnover intention among nurses (P<0.05). More details are shown in Table 1.
Table 1 Sociodemographic characteristics of the sample (N = 1131)
Variables
|
N (%)
|
Turnover intention
|
F/t
|
P value
|
Gender
|
|
|
0.426
|
0.670
|
Male
|
81(7.2)
|
14.51±3.47
|
|
|
Female
|
1050(92.8)
|
14.32±3.77
|
|
|
Age (years)
|
|
|
13.180
|
<0.001
|
<25
|
189(16.7)
|
14.44±3.70
|
|
|
25-44
|
901(79.7)
|
14.45±3.73
|
|
|
>44
|
41(3.6)
|
11.41±3.36
|
|
|
Education level
|
|
|
8.900
|
<0.001
|
Junior college or below
|
223(19.7)
|
13.45±4.11
|
|
|
Bachelor degree
|
847(74.9)
|
14.49±3.63
|
|
|
Master and above
|
61(5.4)
|
15.46±3.50
|
|
|
Self-directed choice of nursing
|
|
|
-11.449
|
<0.001
|
Yes
|
863(76.3)
|
13.72±3.73
|
|
|
No
|
268(23.7)
|
16.31±3.06
|
|
|
Single status
|
|
|
3.643
|
<0.001
|
Yes
|
305(27.0)
|
15.00±3.65
|
|
|
No
|
826(73.0)
|
14.09±3.76
|
|
|
Years of work (years)
|
|
|
2.568
|
0.077
|
<5
|
313(27.7)
|
14.49±3.73
|
|
|
5-10
|
347(30.7)
|
14.60±3.64
|
|
|
>10
|
471(41.6)
|
14.04±3.83
|
|
|
Professional title
|
|
|
1.763
|
0.152
|
Nurse
|
184(16.3)
|
14.13±3.90
|
|
|
Nurse practitioner
|
455(40.2)
|
14.59±3.69
|
|
|
Nurse-in-charge
|
476(42.1)
|
14.22±3.74
|
|
|
Associate chief nurse or above
|
16(1.4)
|
12.94±3.57
|
|
|
Work unit
|
|
|
10.790
|
<0.001
|
Emergency department
|
286(25.3)
|
14.47±3.80
|
|
|
Intensive care unit
|
91(8.0)
|
15.68±3.16
|
|
|
General wards
|
453(40.1)
|
14.54±3.78
|
|
|
Operating room
|
138(12.2)
|
14.26±3.33
|
|
|
Others
|
163(14.4)
|
12.84±3.80
|
|
|
F, one-way analysis of variance, t, independent t test.
Depression, anxiety, emotional exhaustion and turnover intention of nurses
The prevalence of depression and anxiety among nurses was 25.9% and 22.3%, respectively. The average scores for depression, anxiety, emotional exhaustion and turnover intention were 1.92±1.59, 1.90±1.60, 11.55±4.84 and 14.34±3.75, respectively. The mean scores of each turnover intention dimension, from highest to lowest, were as follows: 5.50±1.27 (the possibility of employees having access to external work), 4.45±1.66 (the motivation for employees to find other jobs) and 4.38±1.61 (the possibility of an employee quitting the present job).
Associations among anxiety, depression, emotional exhaustion and turnover intention
The Pearson correlation analysis (Table 2) revealed a strong and significant positive association between depression and anxiety (r = 0.847). Emotional exhaustion showed a moderately positive correlation with both depression (r = 0.521) and anxiety (r = 0.538). Turnover intention showed a mildly significant positive correlation with both depression (r = 0.378) and anxiety (r = 0.391), as well as a moderately significant positive correlation with emotional exhaustion (r = 0.532).
Both unadjusted and adjusted models of multiple linear regression analysis showed significant associations between depression/anxiety, emotional exhaustion and turnover intention. Depression, anxiety, and emotional exhaustion were positive predictors of turnover intention among nurses. Unadjusted models showed that depression and emotional exhaustion explained 29.5% of the model variance (model 1a), while anxiety and emotional exhaustion explained 29.7% of the model variance (model 2a). Results are detailed in Table 3.
Table 2 Correlations between depression, anxiety, emotional exhaustion and turnover intention (N = 1131)
|
1
|
2
|
3
|
1 Depression
|
-
|
-
|
-
|
2 Anxiety
|
0.847***
|
-
|
-
|
3 Emotional exhaustion
|
0.521***
|
0.538***
|
-
|
4 Turnover intention
|
0.378***
|
0.391***
|
0.532***
|
***P<0.001
Table 3 Multiple linear regression analysis of turnover intention among nurses (N = 1131)
Independent variable
|
β (95% CI)
|
Model 1a
|
Model 1b
|
Model 2a
|
Model 2b
|
Emotional exhaustion
|
0.459 (0.311, 0.400) ***
|
0.402 (0.266, 0.356) ***
|
0.452 (0.305, 0.359) ***
|
0.395 (0.260, 0.352) ***
|
Depression/Anxiety
|
0.138 (0.192, 0.463) ***
|
0.139 (0.196, 0.462) ***
|
0.148 (0.210, 0.483) ***
|
0.147 (0.212, 0.479) ***
|
Adjusted R2
|
0.295
|
0.329
|
0.297
|
0.330
|
Models 1a and 1b: depression and emotional exhaustion were used as independent variables.
Models 2a and 2b: anxiety and emotional exhaustion were used as independent variables.
Models 1a and 2a, unadjusted models; Models 2b and 2b, adjusted for age, education level, self-directed choice of nursing, single status and work unit.
β, standardized regression coefficient; 95%CI, 95% confidence interval for regression coefficient.
***P<0.001.
The role of bidirectional associations of depression, anxiety and emotional exhaustion on turnover intention
To test the hypotheses, models 1-4 were constructed, as shown in Fig. 3 – 6. Each model's model fit indices demonstrated good fit (Appendix 1). Depression (β = 0.483, for the total effect) and anxiety (β = 0.483, for the total effect) were significantly and positively associated with turnover intention. Model 1 and Model 2 revealed that emotional exhaustion partially mediated the association between depression/anxiety and turnover intention, both with a mediation effect ratio of 60.7%. Meanwhile, emotional exhaustion (β = 0.620, for the total effect) was significantly and positively associated with turnover intention. Model 3 showed that the mediating ratio of depression was 17.6%, and Model 4 showed that the mediating ratio of anxiety was 16.5%. See more details in Table 4.
Table 4 The role of bidirectional associations of depression, anxiety and emotional exhaustion on turnover intention (N = 1131)
Estimate
|
B
|
β
|
95%CI
|
Total effect
|
|
|
|
Model 1: Depression→TI
|
0.365
|
0.483
|
(0.423, 0.540) ***
|
Model 2: Anxiety→TI
|
0.350
|
0.483
|
(0.423, 0.540) ***
|
Model 3: EE→TI a
|
0.209
|
0.620
|
(0.571, 0.664) ***
|
Model 4: EE→TI b
|
0.209
|
0.620
|
(0.571, 0.664) ***
|
Indirect effect
|
|
|
|
Model 1: Depression→EE→TI
|
0.222
|
0.293
|
(0.245, 0.346) ***
|
Model 2: Anxiety→EE→TI
|
0.212
|
0.293
|
(0.245, 0.346) ***
|
Model 3: EE→Depression→TI
|
0.037
|
0.109
|
(0.064, 0.155) ***
|
Model 4: EE→Anxiety→TI
|
0.050
|
0.102
|
(0.061, 0.141) ***
|
Direct effect
|
|
|
|
Model 1: Depression→TI
|
0.144
|
0.190
|
(0.109, 0.268) ***
|
Model 2: Anxiety→TI
|
0.137
|
0.190
|
(0.109, 0.268) ***
|
Model 3: EE→TI a
|
0.212
|
0.577
|
(0.441, 0.577) ***
|
Model 4: EE→TI b
|
0.216
|
0.518
|
(0.450, 0.582) ***
|
a depression as a mediating variable; b anxiety as a mediating variable; EE: emotional exhaustion; TI: turnover intention; B, path coefficient; β, standardized path coefficient; 95%CI, 95% confidence interval for the standardized path coefficient.
***P<0.001