Chen et al.

BMC Medical Education (2022) 22:627

https://doi.org/10.1186/s12909-022-03683-2

RESEARCH Open Access

Association of depression symptoms

and sleep quality with state‑trait anxiety
in medical university students in Anhui
Province, China: a mediation analysis
Jiangyun Chen1†, Yusupujiang Tuersun1†, Jiao Yang1, Man Xiong1, Yueying Wang1, Xinyi Rao1 and Shuai Jiang2,3*

Abstract
Background: The prevalence of depression symptoms among medical students is particularly high, and it has
increased during the COVID-19 epidemic. Sleep quality and state-trait anxiety are risk factors for depression, but no
study has yet investigated the mediating role of state-trait anxiety in the relationship between poor sleep quality and
depression symptoms in medical students. This study aims to investigate the relationship among depression symp-
toms, sleep quality and state-trait anxiety in medical university students in Anhui Province.
Methods: This was a cross-sectional survey of 1227 students’ online questionnaires collected from four medical
universities in Anhui Province using a convenience sampling method. We measured respondents’ sleep quality, state-
trait anxiety, and depression symptoms using three scales: the Pittsburgh Sleep Quality Index (PSQI), the State-Trait
Anxiety Inventory (STAI) and the Self-rating Depression Scale (SDS). We analysed the mediating role of STAI scores on
the association between PSQI scores and SDS scores through the Sobel-Goodman Mediation Test while controlling
for covariates. P < 0.05 was considered statistically significant.
Results: A total of 74.33% (912) and 41.40% (518) of the respondents reported suffering from poor sleep quality and
depression symptoms. Sleep quality, state-trait anxiety, and depression symptoms were positively associated with
each other (β = 0.381 ~ 0.775, P < 0.001). State-trait anxiety partially mediated the association between sleep quality
and depression symptoms (Sobel test Z = 15.090, P < 0.001), and this mediating variable accounted for 83.79% of the
association when adjusting for potential confounders. Subgroup analysis further revealed that STAI scores partially
mediated the association between PSQI scores and SDS scores in females and rural students and fully mediated the
association between PSQI scores and SDS scores in males and urban students.
Conclusions: This study found that sleep quality and state-trait anxiety have a significant predictive effect on depres-
sion symptoms. State-trait anxiety mediated the relationship between sleep quality and depression symptoms, with
a more complex mechanism observed among rural and female medical students. Multiple pathways of intervention
should be adopted, such as encouraging students to self-adjust, providing professional psychological intervention

†
Jiangyun Chen and Yusupujiang Tuersun contributed equally to this work.
*Correspondence: [email protected]
2
The First Affiliated Hospital of Zhengzhou University, Zhengzhou, China
Full list of author information is available at the end of the article

© The Author(s) 2022. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which
permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the
original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or
other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line
to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory
regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this
licence, visit http://​creat​iveco​mmons.​org/​licen​ses/​by/4.​0/. The Creative Commons Public Domain Dedication waiver (http://​creat​iveco​
mmons.​org/​publi​cdoma​in/​zero/1.​0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.
Chen et al. BMC Medical Education (2022) 22:627 Page 2 of 10

and timely monitoring, enriching extracurricular activities, and making changes in policies regarding long shifts and
working hours.
Keywords: Students, Medical, Sleep quality, Depression symptoms, Anxiety, China

Background In addition, the state-trait anxiety status of medical stu-

According to the World Health Organization, 264 million dents is strongly associated with depression symptoms.
people of all ages worldwide suffered from depression Depression symptoms and anxiety disorders coexist
symptoms in 2017, among which the lifetime prevalence almost simultaneously, and an article examining factors
of depression symptoms in China was 6.9%, and the associated with depression symptoms and anxiety in
proportion of Chinese undergraduates suffering from medical students showed a positive correlation between
depression symptoms was as high as 23.7% [1]. levels of state anxiety and trait anxiety and depression
Several studies have shown that medical students scores [20]. It has been shown that anxiety precedes
have a higher prevalence of depression than nonmedi- depression episodes for most older age groups [21]. A
cal students due to academic stress [2–4]. Additionally, previous study identified anxiety as a predictor of depres-
the prevalence of depression symptoms among medical sion symptoms in medical students [22]. High levels of
students has increased during the COVID-19 epidemic state-trait anxiety were a cause of medical students main-
[5–7]. According to Scott J Halperin, the prevalence of taining a depressive state [10].
depression symptoms was 70% higher in the COVID- Sleep quality affects the occurrence of state-trait anxi-
19 era than in previous studies of medical students [5]. ety. A previous survey showed that people with poor
A study by Franck Rolland showed a higher prevalence sleep reported a higher frequency of anxiety [23]. A
of 12-month major depressive episodes among medical 9-year follow-up survey showed that sleep difficulties
students one year after the start of the COVID-19 pan- significantly predicted self-reported anxiety six to nine
demic than in 2016 [6]. Yoshito Nishimura also reported years later, and as self-reported sleeping difficulty sever-
that online education during the COVID-19 pandemic ity increased, the risk of depression also increased [24].
aggravated the depression symptoms of medical stu- In summary, it has been proven that poor sleep qual-
dents [7]. Depression symptoms among medical stu- ity can lead to depression symptoms; that higher lev-
dents can lead to many negative consequences. First, the els of state-trait anxiety are risk factors for depression
development of suicidal ideation caused by depression symptoms; and that poor sleep quality can cause state-
symptoms in medical students is alarming, with studies trait anxiety. However, the relationship between the
showing that the incidence of depression or depression three is still unclear. Additionally, depression symptoms
symptoms was 28.0% and that the incidence of suicidal among medical students worsened due to the outbreak
ideation among medical students was 5.8% [8]. Second, of COVID-19, which affected their academic, health, and
depressive symptoms have a negative impact on the aca- interpersonal relationships. It is important to understand
demic performance of medical students [3]. Third, their the condition and the mechanisms associated with the
professional careers and personal lives will suffer nega- occurrence of depression symptoms in medical students
tive effects if depression symptoms are not treated [9]. in the context of COVID-19.
Fourth, students with depressive symptoms exhibit rela-
tionship problems, cynicism, and a decline in satisfac-
tion with academic activities [10]. Clearly, there is a need Present study
to pay attention to the depression symptoms of medical In this study, data on sleep quality, state-trait anxiety and
students. depression symptoms among medical students in four
Medical students’ sleep quality is poor due to late-night medical universities in Anhui Province were collected
studying [11, 12], a bad dormitory environment [13], using scales. The aim of this study was to investigate the
excessive use of electronic devices before bedtime [14, mediating role of state-trait anxiety in the relationship
15], and internship shifts [16]. The study showed that the between sleep quality and depression symptoms. Based
prevalence of sleep disturbance among medical students on a literature review, the following hypotheses were
is higher than that among nonmedical students and the formulated: quality of sleep directly affects depression
general population [17]. Additionally, sleep disturbance symptoms and indirectly affects depression symptoms
is a risk factor for depression symptoms [18]. The occur- through state-trait anxiety; that is, state-trait anxiety
rence of depression symptoms was related to lower sleep mediates the relationship between quality of sleep and
quality [11, 19]. depression symptoms.
Chen et al. BMC Medical Education (2022) 22:627 Page 3 of 10

Methods 2 = without sleep disorder) and four levels of degree of

Subjects and procedure sleep quality (1 = fair sleep quality, 2 = fair sleep quality,
A cross-sectional survey of university students from four 3 = very poor sleep quality, 4 = very good sleep quality)
medical universities in Anhui Province was conducted by the scores of the collected data. In the current sample,
from September to December 2020 using a convenience Cronbach’s alpha was 0.796, whereas the Kaiser–Meyer–
sampling method. To better identify medical students, Olkin (KMO) was 0.746. The significance of Bartlett’s test
we included all medical universities in Anhui, leaving out of sphericity was P < 0.05. The scale exhibited superior
comprehensive universities. We included in this experi- reliability and validity.
ment all medical universities in Anhui Province (there
are four medical universities in total in Anhui Province). State‑Trait Anxiety Inventory (STAI)
These universities offer undergraduate degree education, The state-trait anxiety inventory (STAI) questionnaire
including Anhui Medical University, Anhui University of consists of 40 questions divided into two groups to
Traditional Chinese Medicine, Bengbu Medical College, assess anxiety as a transient state (state anxiety) and as
and Wannan Medical College (the number of under- an underlying trait (trait anxiety). State anxiety is consid-
graduate students at each university ranges from 12,000 ered to be a transient emotional state characterized by
to 16,000). We distributed online questionnaires to uni- subjective feelings, apprehension, and overactivity of the
versity students with the assistance of academic adminis- autonomic nervous system. Trait anxiety is a relatively
trators using the web-based Questionnaire Star platform stable personal state with a tendency to perceive situa-
(https://​www.​wjx.​cn/) and obtained informed consent tions as threatening. Both the state and trait scales con-
from each participating subject. In addition, because sist of 20 items, including directly and inversely worded
face-face surveys were difficult to conduct during the questions and punctuation. Scores range from 20 to 80,
COVID-19 outbreak and university students were the with higher scores indicating higher levels of anxiety [28,
main users of mobile devices, they could also complete 29]. The final STAI scores were obtained using an online
the questionnaire accurately on their own. The question- calculator (https://​www.​nsrusa.​org/​score.​php) to avoid
naire comprised four main sections. The first section confusion about the punctuation of the reverse wording.
obtained the sociodemographic characteristics of the The Cronbach’s α coefficient was 0.954, and the Kaiser–
participants (e.g., age, gender, grade). The second section Meyer–Olkin (KMO) was 0.971. The significance of Bar-
assessed the level of sleep quality of the medical students. tlett’s test of sphericity was P < 0.05. The scale exhibited
The third part assessed the level of state-trait anxiety of superior reliability and validity.
the medical students. In the fourth part, we assessed the
level of depression symptoms of the medical students Self‑rating Depression Scale (SDS)
(Supplementary Table 1). The questionnaire was com- The Self-rating Depression Scale (SDS) is a brief self-
pleted anonymously online, and subjects who provided rating scale that assesses the psychological and somatic
electronic informed consent and voluntarily participated symptoms of depression. It has been widely used in dif-
in the study were included in the study. The study proto- ferent age groups for screening purposes and to measure
col was approved by the Clinical Trials Ethics Committee depression [30]. It has been widely used to screen for and
of the First Affiliated Hospital of Zhengzhou University. measure the severity of depression [31]. It has good inter-
The ethical approval number is 2021-KY-0669. nal consistency test and retest reliability and has good
content validity and criterion validity [32]. The standard
Sleep quality (PSQI) converted score of the SDS is from 0 to 100 (raw score
Sleep quality was assessed using the Pittsburgh Sleep range from 20 to 80), and 50 and over suggests clinically
Quality Index (PSQI) [25]. The index consists of 19 self- significant symptoms [33]. The Cronbach’s α coefficient
rated items across seven components on a scale from 0 was 0.954, whereas the Kaiser–Meyer–Olkin (KMO)
to 3, where 0 indicates no difficulty and 3 indicates severe was 0.971. The significance of Bartlett’s test of sphericity
difficulty. The scores of these seven components are com- was P < 0.05. The scale exhibited superior reliability and
bined to provide a total score of 0–21, with higher scores validity.
indicating poorer sleep quality and scores greater than
5 distinguishing poor sleepers from good sleepers. The Covariates
PSQI is considered an appropriate tool for assessing sleep Age and sex were included as fixed covariates and were
quality in adults (18–80 years) [26], and Cronbach’s alpha adjusted for in the analyses. Other covariates were
among adolescents and young adults was 0.72 [27]. The included in the final model as potential confounders
PSQI in this study was divided into two levels of pres- if they altered PSQI/STAI estimates of SDS by >10% or
ence or absence of sleep disorder (1 = with sleep disorder, were significantly associated with SDS. The following
Chen et al. BMC Medical Education (2022) 22:627 Page 4 of 10

covariates were selected based on established associa- the distribution of males versus females (P > 0.05) (see
tions and/or plausible biological relationships: major, Table 1 at the end of the article for details).
ethnicity, only child, birthplace, closest relationship, edu-
cation of closest relationship, education of mother, edu- Correlation analysis
cation of father, job of closest relationship, job of mother, SDS scores were positively correlated with PSQI
job of father. The relationship between each confounding scores (r = 0.381, p < 0.001) and STAI scores (r = 0.775,
factor and SDS is detailed in Supplementary Tables 2, 3 p < 0.001), and PSQI scores were positively correlated
and 4. with STAI scores (r = 0.428, p < 0.001) when controlling
for confounding factors (Table 2).
Statistical analysis
Relationship between PSQI and STAI/SDS
Continuous variables are reported as the mean ± SD, and
PSQI scores were significantly associated with adher-
categorical variables are reported as the frequency (%).
ence to STAI/SDS according to linear regression mod-
Characteristic differences were examined using Student’s
els before adjustment; after adjusting for age and gender
t test for continuous variables and the chi-squared test
(adjusted model 1); and after adjusting for age and gender
for categorical variables. Linear regression models were
as well as major, ethnicity, only child, birthplace, closest
used to measure the association between PSQI and SDS/
relationship, education of closest relationship, education
STAI scores before and after adjustment for covariates
of mother, education of father, job of closest relation-
and the association between STAI and SDS scores and
ship, job of mother and job of father (adjusted model 2)
are reported as coefficients and 95% confidence intervals
(P < 0.001) (Table 3).
(CIs). We analysed the mediating role of STAI scores on
the association of PSQI scores with SDS scores through
Relationship between STAI and SDS
the Sobel-Goodman Mediation Test [34] while control-
STAI scores were significantly associated with SDS scores
ling for all the selected covariates [35].
according to linear regression models before adjustment;
All p values were 2-sided, with an α ≤0.05 used to
after adjusting for age and gender (adjusted model 1);
define statistical significance. Data were analysed using
and after adjusting for age and gender as well as major,
Stata version 16 (2017, University Station, Texas 77,845
ethnicity, only child, birthplace, closest relationship, edu-
USA) and R version 3.6.3 (2018, R Foundation for Statis-
cation of closest relationship, education of mother, edu-
tical Computing, Vienna, Austria).
cation of father, job of closest relationship, job of mother
and job of father (adjusted model 2) (P < 0.001) (Table 4).
Results
General characteristics Mediation analysis
A total of 1300 questionnaires were collected, with 1227 PSQI scores were positively associated with SDS scores
valid questionnaires (389 samples were obtained from among medical university students. Mediation analy-
Anhui Medical University, 288 samples from Anhui Uni- sis including the STAI revealed that the association
versity of Traditional Chinese Medicine, 246 samples between PSQI and SDS scores was mediated via STAI
from Bengbu Medical University, and 304 samples from scores. STAI partially mediated the association between
Wannan Medical University). Thus, the effective rate PSQI and SDS in this study, and this mediating variable
was 94.4%. Among the respondents, 594 were males, accounted for 83.79% of the association when adjust-
accounting for 48.4%, and 633 were females, accounting ing for potential confounders. PSQI scores were related
for 51.6%. to STAI (β = 2.480, P < 0.001) and SDS scores (β = 0.225,
The demographics of the 1227 patients are presented P < 0.001). STAI scores were also related to SDS scores
in Table 1. The majority of students were in the first and (β = 0.470, P < 0.001). The final mediation models of the
fourth years, 23.5 and 55.1%, respectively, and 88.6% were independent variable (PSQI), the mediating variable
majoring in medicine. A total of 64.8% of students were (STAI) and the dependent variable (SDS) are shown in
from rural areas, 43.5% of medical students’ fathers were Fig. 1.
workers, and 35.1% of their mothers were workers. Poor
sleep quality was reported by 74.3% (912) of the respond- Subgroup analysis
ents, and 41.4% (518) reported suffering from depression Subgroup analyses of sex and birthplace are shown
symptoms. There was a statistically significant difference in Table 5. STAI partially mediated the associa-
in sleep quality scores between males and females, 27.3% tion between PSQI scores and SDS scores in females
(162) and 24.2% (153), respectively. State-trait anxiety (z = 10.313; indirect effect = 1.117, CI = 1.164 ~ 1.700,
and depression symptoms scores were nonsignificant in P < 0.001; direct effect = 0.315, CI = 1.976 ~ 2.818;
Chen et al. BMC Medical Education (2022) 22:627 Page 5 of 10

Table 1 Characteristics of respondents (N = 1227)

Overall (n = 1227) Male (n = 594) Female (n = 633) P value

Grade N (%) 0.133

1 288 (23.47) 121 (20.37) 167 (26.38)
2 75 (6.11) 39 (6.57) 36 (5.69)
3 91 (7.42) 43 (7.24) 48 (7.58)
4 676 (55.09) 345 (58.08) 331 (52.29)
5 97 (7.91) 46 (7.74) 51 (8.06)
Major 0.263
Medicine 1087 (88.59) 520 (87.54) 567 (89.57)
Other 140 (11.41) 74 (12.46) 66 (10.43)
Ethnicity 0.873
Han 1193 (97.23) 578 (97.31) 615 (97.16)
Minority 24 (1.96) 16 (2.69) 18 (2.84)
Only child <0.001
No 740 (60.31) 314 (52.86) 426 (67.30)
Yes 487 (36.69) 280 (47.14) 207 (32.70)
Birthplace 0.225
Urban 432 (35.21) 199 (33.50) 233 (36.81)
Rural 795 (64.79) 395 (66.50) 400 (63.19)
Closet relationship 0.192
Parents 1006 (81.99) 492 (82.83) 514 (81.20)
Grandparents 145 (11.82) 60 (10.10) 85 (13.43)
Siblings 51 (4.16) 27 (4.55) 24 (3.79)
Other 25 (2.04) 15 (2.53) 10 (1.58)
Education of closest relationship 0.001
Less than lower secondary education 640 (52.16) 301 (50.67) 339 (52.97)
Upper secondary & vocational 251 (20.46) 148 (24.92) 103 (16.27)
Tertiary education 336 (27.38) 145 (24.41) 191 (30.17)
Education of father 0.011
Less than lower secondary education 662 (53.95) 313 (52.69) 349 (55.13)
Upper secondary & vocational 290 (23.63) 157 (26.43) 133 (21.10)
Tertiary education 275 (22.41) 124 (20.88) 151 (23.85)
Education of mother 0.010
Less than lower secondary education 820 (66.83) 386 (64.98) 434 (68.56)
Upper secondary & vocational 235 (19.15) 136 (22.90) 99 (15.64)
Tertiary education 172 (14.02) 72 (12.12) 100 (15.80)
Job of closest relationship 0.003
Workers 396 (32.27) 226 (38.05) 170 (26.86)
Farmers 206 (16.79) 94 (15.82) 112 (17.69)
Civil servants, teachers and other intellectuals 264 (21.52) 110 (18.52) 154 (24.33)
Businessmen 172 (14.02) 82 (13.80) 90 (14.22)
Others 199 (16.22) 82 (13.80) 107 (16.90)
Job of father 0.411
Workers 534 (43.52) 275 (46.30) 259 (40.92)
Farmers 131 (10.68) 64 (10.77) 67 (10.58)
Civil servants, teachers and other intellectuals 237 (19.32) 108 (18.18) 129 (20.38)
Businessmen 197 (16.06) 94 (15.82) 103 (16.27)
Others 128 (10.43) 53 (8.92) 75 (11.85)
Job of mother 0.008
Workers 431 (35.13) 239 (40.24) 192 (30.33)
Chen et al. BMC Medical Education (2022) 22:627 Page 6 of 10

Table 1 (continued)
Overall (n = 1227) Male (n = 594) Female (n = 633) P value

Farmers 243 (19.80) 112 (18.86) 131 (20.70)

Civil servants, teachers and other intellectuals 185 (15.08) 86 (14.48) 99 (15.64)
Businessmen 154 (12.55) 72 (12.12) 82 (12.95)
Others 214 (17.44) 85 (14.31) 129 (20.38)
PSQI (Mean ± SD) 5.87 ± 2.94 5.75 ± 3.04 5.97 ± 2.86 0.010
STAI (Mean ± SD) 84.10 ± 17.28 84.23 ± 17.76 83.98 ± 16.82 0.278
SDS (Mean ± SD) 49.04 ± 10.85 48.87 ± 11.09 49.19 ± 10.62 0.228

Table 2 Partial correlations coefficients (r) among PSQI, STAI and Table 4 Linear aegression analysis for STAI associated with SDS
SDS in students, [β (95% CI)]
PSQI STAI SDS SDS

PSQI β(95% CI) P value

STAI 0.428*** STAI Unadjusted 0.49 (0.46,0.51) <0.001
SDS 0.381*** 0.775*** Adjusted 1 0.48 (0.44,0.49) <0.001
PSQI Pittsburgh Sleep Quality Index, STAI State-Trait Anxiety Inventory, SDS Self- Adjusted 2 0.48 (0.45,0.50) <0.001
rating Depression Scale
β beta coefficient, CI confidence interval, SDS Self-rating Depression Scale,
The model was adjusted for gender, birthplace, grade, major, ethnicity, only
STAI State-Trait Anxiety Inventory
child, close relationship, education of close relationship, education of father,
education of mother, job of close relationship, job of father, job of mother Adjusted 1: Adjusted for gender, age
Values are bolded if they achieved statistical significance at p ≤ 0.05 Adjusted 2: Adjusted for gender, age, major, ethnicity, only child, birthplace,
closest relationship, education of closest relationship, education of mother,
*** p < 0.001
education of father, job of closest relationship, job of mother, job of father

Table 3 Linear regression analysis for PSQI associated with STAI

and SDS in students, [β (95% CI)] Discussion
STAI SDS Previous studies have shown a bidirectional relationship
between state-strait anxiety (depression symptoms) and
β(95% CI) P value β(95% CI) P value
sleep quality [36]. Our study confirmed that depression
PSQI Unadjusted 2.51 (2.22,2.81) <0.001 1.40 (1.21,1.59) <0.001 symptoms can be predicted by sleep quality, that higher
Adjusted 1 1.51 (1.16,1.17) <0.001 0.88 (0.65,1.11) <0.001 levels of state-strait anxiety are a risk factor for depres-
Adjusted 2 1.48 (1.12,1.83) <0.001 0.89 (0.66,1.12) <0.001 sion symptoms and that state-strait anxiety can also be
β beta coefficient, CI confidence interval, PSQI Pittsburgh Sleep Quality Index, predicted by sleep quality. Moreover, our study further
STAI State-Trait Anxiety Inventory, SDS Self-rating Depression Scale revealed that state-strait anxiety mediated the effect of
Adjusted 1: Adjusted for gender, age sleep quality predicting on depression symptoms. In
Adjusted 2: Adjusted for gender, age, major, ethnicity, only child, birthplace, addition, the present study focused on the levels of sleep
closest relationship, education of closest relationship, education of mother,
education of father, job of closest relationship, job of mother, job of father quality, state-strait anxiety, and depression symptoms
among medical students during the COVID-19 pandemic
and analyzed the variability of subgroups by gender and
birthplace in this particular population.
P < 0.001) and rural students (z = 12.521; indirect The China National Mental Health Report (2019–2020)
effect = 1.213, CI = 1.222 ~ 1.699, P < 0.001; direct shows that university students have slightly higher rates
effect = 1.461, CI = 2.193 ~ 2.916; P < 0.001). STAI fully of depression symptoms risk than adolescents nationally
mediated the association between PSQI scores and and adults nationally [37]. Among university students,
SDS scores in males (z = 10.884; indirect effect = 1.217, medical majors have a higher prevalence rate of depres-
CI = 1.161 ~ 1.722, P < 0.001; direct effect = 0.0149, sion symptoms than nonmedical majors since medical
CI = 1.980 ~ 2.661; P = 0.128) and urban students students spend more time in university than students in
(z = 8.411; indirect effect = 1.104, CI = 1.213 ~ 1.687, other majors and face existential topics such as suffer-
P = 0.247; direct effect = 1.294, CI = 2.182 ~ 2.819; ing and death [38]. In our study, 41.4% (518) of medical
P < 0.001) (Table 5). students reported suffering from depression symptoms.
Chen et al. BMC Medical Education (2022) 22:627 Page 7 of 10

Fig. 1 Mediation Analysis. Note: STAI = State-Trait Anxiety Inventory; PSQI = Pittsburgh Sleep Quality Index; SDS = Self-rating Depression Scale.
The Sobel test was used to test the hypothesis that the indirect role was equal to 0, adjusting for potential confounders (gender, birthplace, grade,
major, ethnicity, only child, close relationship, education of close relationship, education of father, education of mother, job of close relationship, job
of father, job of mother). Values are bolded if they reached statistical significance at p ≤ 0.05

Table 5 Subgroup analysis of mediation models for SDS associated with PSQI mediated by STAI in medical university students
Indirect effect Direct effect Total effect Z Sobel p value Proportion of
total effect that is
mediated

Gender
Male 1.217*** 0.149 1.366*** 10.884 <0.001 89.12
Female 1.117*** 0.315** 1.432*** 10.313 <0.001 78.01
Birthplace
Urban 1.104*** 0.189 1.294*** 8.411 <0.001 85.37
Rural 1.213*** 0.247** 1.461*** 12.521 <0.001 83.07
STAI State-Trait Anxiety Inventory, PSQI Pittsburgh Sleep Quality Index, SDS Self-rating depression scale. Sobel-Goodman Mediation Test in adjusted models for gender,
birthplace, grade, major, ethnicity, only child, close relationship, education of close relationship, education of father, education of mother, job of close relationship, job
of father, job of mother. Values are bolded if they achieved statistical significance at P ≤ 0.05
** p < 0.01, *** p < 0.001

This figure is higher than the 28.9% combined prevalence Among the results of the characteristics of the
rate for Chinese medical students that was reported in respondents, we found a significant difference in sleep
2019 [4]. Chinese medical students also show a higher quality between men and women, which is similar to the
prevalence of depression symptoms than foreign medi- results of previous studies [42], but found no difference
cal students. The positive rates of depressive symptoms in the distribution of depressive symptoms and anxiety
in medicine students reported in our study are higher state traits by gender, which is unlike other studies that
than the 19.2% rate reported in a German study in 2018 reported a higher prevalence of depression symptoms
[39] and higher than those of medical students in Middle and state-trait anxiety in women than in men [43, 44].
Eastern countries (41.1%) who had the highest positive This may be related to the subject of our study. Accord-
screening rate for depressive symptoms in a 2015 study ing to the Chinese educational system, the age of Chinese
across three countries [40]. These inconsistent findings medical undergraduates is 18–23 years. More than half of
may be related to the outbreak of the COVID-19 epi- the medical student population in our study was in their
demic, during which medical students had more severe fourth year of college, which means that the average age
depressive symptoms [41]. was 22 years. Previous research examined the analysis
Chen et al. BMC Medical Education (2022) 22:627 Page 8 of 10

of gender differences in depression in several countries work an average of more than 90 hours per week during
around the world, including China, and noted that the their inpatient rotations, sleep an average of 2 hours per
significant decrease in gender differences in depressive night in an on-call status, and sleep an average of 4 hours
symptoms from adolescence to the early 20s, while differ- less per night than they do at home [60], resulting in a
ences between the ages of 20–29 and later years were not significant reduction in sleep quality. Therefore, there is
significant [45]. a need to make changes in policies regarding long shifts
Our findings showed that state-strait anxiety medi- and working hours. Hospital administrators and policy-
ated the effect of sleep quality on depression. This is makers should limit shift work to 12–16 hours, schedule
demonstrated by the fact that poor sleep quality causes at least 10 hours of rest between shifts [61] and should
the medial prefrontal cortex, which mediates the brain’s not work more than 80 hours per week [62], thus ensur-
emotions, to be in a state of deactivation, which leads to ing quality sleep and more efficient work for medical
increased anxiety [46]. Additionally, sleep deprivation students.
also amplifies basic emotional responses and increases
negative emotional states such as anxiety [47], and these Strengths and limitations
emotional dysfunctions, such as anxiety, affect their nor- This study makes an important contribution to the litera-
mal interpersonal interactions [48]. Anxiety disorders ture by assessing the associations among the PSQI, STAI
eventually develop into depression symptoms due to and SDS through data reported by medical students from
interpersonal dysfunction [49]. four medical universities in Anhui Province. The present
Additionally, the results of the subgroup analysis fur- study confirms our research hypothesis that state-trait
ther showed that STAI scores partially mediated the asso- anxiety moderates the relationship between poor sleep
ciation between PSQI and SDS scores among females and quality and depression symptoms in university students.
rural students and fully mediated the association among Additionally, we conducted subgroup analyses of gen-
males and urban students, which was also consistent with der and birthplace and showed that the extent to which
our study hypothesis. For the gender subgroup, poor poor sleep quality affects depression symptoms through
sleep quality in women induces more complex mecha- state-trait anxiety was more pronounced in female medi-
nisms affecting depression, which may be due to the cal students and in rural-born medical students. These
fact that women report both more intense positive and findings all enrich the theory of depression symptoms-
more intense negative emotions in their daily lives [50, related research among medical students.
51]. For the birthplace subgroup, Chinese universities This article also has some limitations. First, our study
are established in cities, and there is a large gap between is a cross-sectional study that can only account for the
urban and rural areas in China. Moreover, there is a cer- associations among the PSQI, STAI and SDS and cannot
tain degree of migration discrimination and self-induced explain their causal relationship. Second, we used a con-
psychological distress for groups that come to live in venience sampling method to collect questionnaire infor-
urban from rural areas [39]. Therefore, for students who mation from medical students within the four medical
were born and had always lived in the city, the change in universities in Anhui Province, which may have affected
environment weakly impacts their emotions [37, 38, 40]. the sample representativeness. However, we collected a
However, for those who were born in rural areas, uni- sample of 1227 respondents, and this large sample was
versity life was a huge change to the living environment, able to compensate for this shortcoming to some extent.
which could increase the complexity of emotional coping. Third, there was a possibility of recall bias, as all data
Several measures should be considered for improv- were self-reported.
ing both sleep quality and anxiety to relieve depression
symptoms among university students. First, students Conclusions
should be encouraged to self-adjust through cognitive- In conclusion, this study confirms the importance of the
behavioural therapy [52, 53] and comprehensive sleep association among depression symptoms, sleep qual-
management programs such as sleep hygiene education, ity, and state-trait anxiety. Sleep quality and state-trait
relaxation training, and music therapy [54]. Second, con- anxiety have a significant predictive effect on depres-
sulting centres staffed with psychotherapists or trained sion symptoms. State-trait anxiety mediated the effect
counsellors should be established to provide professional of sleep quality on depression symptoms, and a more
psychological intervention and monitor students’ anxiety complex mechanism was seen in rural and female med-
levels and psychological states [55]. Third, extracurricular ical students. These findings suggest that improving
activities should be enriched by offering certain physical sleep quality and state-trait anxiety can meaningfully
exercises [56], courses such as tai chi [57, 58], yoga, and improve depression symptoms. Depression symptoms
mindfulness training [59]. Moreover, medical students could be prevented and improved by encouraging
Chen et al. BMC Medical Education (2022) 22:627 Page 9 of 10

students to self-adjust, providing professional psycho- Consent for publication

Not applicable.
logical intervention and timely monitoring, enriching
extracurricular activities, and taking changes in policies Competing interests
regarding long shifts and working hours to improve the The authors declare that the research was conducted in the absence of any
commercial or financial relationships that could be construed as a potential
quality of sleep and state characteristic anxiety in medi- conflict of interest.
cal students.
Author details
1
School of Health Management, Southern Medical University, Guangzhou,
Abbreviations China. 2 The First Affiliated Hospital of Zhengzhou University, Zhengzhou,
PSQI: Pittsburgh sleep quality index; S-AI: State anxiety; T-AI: Trait anxiety; STAI: China. 3 Institute for Hospital Management of Henan Province, Zhengzhou,
State-Trait Anxiety Inventory; SDS: Depression Self-Rating Scale. China.

Received: 16 February 2022 Accepted: 5 August 2022

Supplementary Information
The online version contains supplementary material available at https://​doi.​
org/​10.​1186/​s12909-​022-​03683-2.

Additional file 1: Supplementary Table 1. Variable Description. References

1. 2019 China Depression Field Blue Book_Depression Institute. https://​
Additional file 2: Supplementary Table 2. The Selection Process of www.​medsci.​cn/​artic​le/​show_​artic​le.​do?​id=​1db71​861e8​f6. Accessed 14
Covariates: Step 1 -Analyzing The Relationship Between the Covariate and Apr 2022.
Y (Y = SDS) One by One. 2. Fawzy M, Hamed SA. Prevalence of psychological stress, depres-
Additional file 3: Supplementary Table 3. The Selection Process of sion and anxiety among medical students in Egypt. Psychiatry Res.
Covariates: Step 2 -Covariates Were Introduced into The Basic Model 2017;255:186–94.
and Removed from The Complete Model to Observe the Change of the 3. Jin T, et al. Prevalence of depression among Chinese medical students: A
Regression Coefficient of X (X = PSQI-level). systematic review and meta-analysis. Psychol Health Med. 2021. p. 1–17.
https://​doi.​org/​10.​1080/​13548​506.​2021.​19507​85.
Additional file 4: Supplementary Table 4. The Selection Process of 4. Zeng W, et al. Prevalence of mental health problems among
Covariates: Step 2 -Covariates Were Introduced into The Basic Model medical students in China: A meta-analysis. Medicine (Baltimore).
and Removed from The Complete Model to Observe the Change of the 2019;98(18):e15337.
Regression Coefficient of X (X = STAI score). 5. Halperin SJ, et al. Prevalence of Anxiety and Depression Among Medical
Students During the Covid-19 Pandemic: A Cross-Sectional Study. J Med
Educ Curric Dev. 2021;8:2382120521991150.
Acknowledgments 6. Rolland F, et al. Mental health and working conditions among French
We thank all participants and staff at the study sites for their cooperation. medical students: A nationwide study. J Affect Disord. 2022;306:124–30.
7. Nishimura Y, et al. Impact of the COVID-19 Pandemic on the Psychologi-
Authors’ contributions cal Distress of Medical Students in Japan: Cross-sectional Survey Study. J
Jiangyun Chen: Conceptualization, Writing- Original draft preparation; Yusupu- Med Internet Res. 2021;23(2):e25232.
jiang Tuersun: Conceptualization, Writing- Original draft preparation. Jiao Yang: 8. Puthran R, et al. Prevalence of depression amongst medical students: a
Data curation; Man Xiong: Data curation; Yueying Wang: Data curation; Xinyi meta-analysis. Med Educ. 2016;50(4):456–68.
Rao: Data curation; Shuai Jiang: Writing – Review & Editing. The authors read 9. Moreira DSJ, Moreira CA, Telles-Correia D. Anxiety, Depression and
and approved the final manuscript. Academic Performance: A Study Amongst Portuguese Medical Students
Versus Non-Medical Students. Acta Med Port. 2018;31(9):454–62.
Funding 10. Azad N, et al. Anxiety And Depression In Medical Students Of A Private
This work was supported by the Joint Co-Construction Program of Medical Medical College. J Ayub Med Coll Abbottabad. 2017;29(1):123–7.
Science and Technology of Henan Province (LHGJ20200347), Key Laboratory 11. Cvejic E, Huang S, Vollmer-Conna U. Can you snooze your way to an ’A’?
of Philosophy and Social Sciences of Colleges and Universities in Guangdong Exploring the complex relationship between sleep, autonomic activity,
Province: Public Health Policy Research &Evaluation (2015WSYS0010), and wellbeing and performance in medical students. Aust N Z J Psychiatry.
Guangzhou Public Health Service System Construction Research Foundation 2018;52(1):39–46.
(2021–2023). 12. Almojali AI, et al. The prevalence and association of stress with sleep qual-
ity among medical students. J Epidemiol Glob Health. 2017;7(3):169–74.
Availability of data and materials 13. Zheng B, et al. Associations between dormitory environment/other fac-
All data generated or analyzed during this study are included in this published tors and sleep quality of medical students. Zhonghua Liu Xing Bing Xue
article [and its supplementary information files]. Za Zhi. 2016;37(3):348–52.
14. Zhong Y, et al. Prevalence of smartphone addiction among Asian medical
Declarations students: A meta-analysis of multinational observational studies. Int J
Soc Psychiatry. 2022:207640221089535. https://​doi.​org/​10.​1177/​00207​
Ethics approval and consent to participate 64022​10895​35.
Informed consent was obtained from all subjects and/or their legal guardians 15. Pham HT, Chuang HL, Kuo CP, Yeh TP, Liao WC. Electronic Device Use
for our study. All the subjects well-understood that their participation was before Bedtime and Sleep Quality among University Students. Healthcare
completely voluntary, and they were informed that they could withdraw from (Basel). 2021;9(9):1091.
the study at any time. The study protocol was approved by the Clinical Trials 16. CS, Braganza D, Edwin N. Quality of life among interns at a south-
Ethics Committee of the First Affiliated Hospital of Zhengzhou University. ern Indian tertiary care hospital: a cohort study. Natl Med J India.
And all methods were carried out in accordance with relevant guidelines and 2014;27(4):214–6.
regulations. The ethical approval number is 2021-KY-0669. 17. Azad MC, et al. Sleep disturbances among medical students: a global
All methods were carried out in accordance with relevant guidelines and perspective. J Clin Sleep Med. 2015;11(1):69–74.
regulations. 18. Plante DT. The Evolving Nexus of Sleep and Depression. Am J Psychiatry.
2021;178(10):896–902.
Chen et al. BMC Medical Education (2022) 22:627 Page 10 of 10

19. Dudo K, et al. The association of sleep patterns and depressive symp- 45. Salk RH, Hyde JS, Abramson LY. Gender differences in depression in rep-
toms in medical students: a cross-sectional study. BMC Res Notes. resentative national samples: Meta-analyses of diagnoses and symptoms.
2022;15(1):109. Psychol Bull. 2017;143(8):783–822.
20. Brenneisen MF, et al. Factors associated to depression and anxiety in 46. Deng Z, Law YW. Rural-to-urban migration, discrimination experience,
medical students: a multicenter study. BMC Med Educ. 2016;16(1):282. and health in China: Evidence from propensity score analysis. PLoS One.
21. King-Kallimanis B, Gum AM, Kohn R. Comorbidity of depressive and 2020;15(12):e0244441.
anxiety disorders for older Americans in the national comorbidity survey- 47. Herrmann-Lingen C, Melzer J, von Boetticher D. Emotions, Bodily
replication. Am J Geriatr Psychiatry. 2009;17(9):782–92. Changes, and the Social Environment: How Did Early Psychosomatic
22. Guerrero LJ, et al. Factors that predict depression in medical students. Medicine Consider the Social Dimension in Health and Disease? Psycho-
Gac Med Mex. 2013;149(6):598–604. som Med. 2019;81(8):694–703.
23. Varma P, Burge M, Meaklim H, Junge M, Jackson ML. Poor Sleep Quality 48. Shin YB, et al. The effect on emotions and brain activity by the direct/
and Its Relationship with Individual Characteristics, Personal Experiences indirect lighting in the residential environment. Neurosci Lett.
and Mental Health during the COVID-19 Pandemic. Int J Environ Res 2015;584:28–32.
Public Health. 2021;18(11):6030. 49. Doell KC, Conte B, Brosch T. Interindividual differences in environmentally
24. Silva V, et al. Depression in medical students: insights from a longitudinal relevant positive trait affect impacts sustainable behavior in everyday life.
study. BMC Med Educ. 2017;17(1):184. Sci Rep. 2021;11(1):20423.
25. Buysse DJ, et al. The Pittsburgh Sleep Quality Index: a new instrument for 50. Fujita F, Diener E, Sandvik E. Gender differences in negative affect
psychiatric practice and research. Psychiatry Res. 1989;28(2):193–213. and well-being: the case for emotional intensity. J Pers Soc Psychol.
26. Hinz A, et al. Sleep quality in the general population: psychometric 1991;61(3):427–34.
properties of the Pittsburgh Sleep Quality Index, derived from a German 51. Gohm CL. Mood regulation and emotional intelligence: individual differ-
community sample of 9284 people. Sleep Med. 2017;30:57–63. ences. J Pers Soc Psychol. 2003;84(3):594–607.
27. de la Vega R, et al. The Pittsburgh Sleep Quality Index: Validity and factor 52. Friedrich A, Schlarb AA. Let’s talk about sleep: a systematic review of
structure in young people. Psychol Assess. 2015;27(4):e22–7. psychological interventions to improve sleep in college students. J Sleep
28. Lin CS, Wu SY, Yi CA. Association between Anxiety and Pain in Den- Res. 2018;27(1):4–22.
tal Treatment: A Systematic Review and Meta-analysis. J Dent Res. 53. Taylor DJ, et al. A pilot randomized controlled trial of the effects of cogni-
2017;96(2):153–62. tive-behavioral therapy for insomnia on sleep and daytime functioning in
29. Sirin Y, Yildirimturk S, Ay N. Do state-trait anxiety and previous unpleasant college students. Behav Ther. 2014;45(3):376–89.
dental experiences predict the need for sedation in women having third 54. Gao R, et al. Effects of comprehensive sleep management on sleep
molar surgery? Br J Oral Maxillofac Surg. 2020;58(5):530–4. quality in university students in mainland China. Sleep Biol Rhythms.
30. Schrag A, et al. Depression rating scales in Parkinson’s disease: critique 2014;12(3):194–202.
and recommendations. Mov Disord. 2007;22(8):1077–92. 55. Bantjes J, et al. A Web-Based Group Cognitive Behavioral Therapy Inter-
31. Jokelainen J, et al. Validation of the Zung self-rating depression scale vention for Symptoms of Anxiety and Depression Among University Stu-
(SDS) in older adults. Scand J Prim Health Care. 2019;37(3):353–7. dents: Open-Label, Pragmatic Trial. JMIR Ment Health. 2021;8(5):e27400.
32. Jegede RO. Psychometric properties of the Self-Rating Depression Scale 56. Ji C, Yang J, Lin L, Chen S. Physical Exercise Ameliorates Anxiety, Depres-
(SDS). J Psychol. 1976;93(1st Half ):27–30. sion and Sleep Quality in College Students: Experimental Evidence from
33. Zung WW. From art to science. The diagnosis and treatment of depres- Exercise Intensity and Frequency. Behav Sci (Basel). 2022;12(3):61.
sion. Arch Gen Psychiatry. 1973;29(3):328–37. 57. Caldwell K, et al. Changes in mindfulness, well-being, and sleep quality
34. Baron RM, Kenny DA. The moderator-mediator variable distinction in in college students through taijiquan courses: a cohort control study. J
social psychological research: conceptual, strategic, and statistical consid- Altern Complement Med. 2011;17(10):931–8.
erations. J Pers Soc Psychol. 1986;51(6):1173–82. 58. Caldwell KL, et al. Effects of tai chi chuan on anxiety and sleep quality in
35. Liu Z, et al. Association between perceived stress and depression among young adults: lessons from a randomized controlled feasibility study. Nat
medical students during the outbreak of COVID-19: The mediating role Sci Sleep. 2016;8:305–14.
of insomnia. J Affect Disord. 2021;292:89-94. https://​www.​scien​cedir​ect.​ 59. Ding X, et al. Relationship Between Trait Mindfulness and Sleep Qual-
com/​scien​ce/​artic​le/​pii/​S0165​03272​10047​29?​via%​3Dihub. ity in College Students: A Conditional Process Model. Front Psychol.
36. Alvaro PK, Roberts RM, Harris JK. A Systematic Review Assessing Bidirec- 2020;11:576319.
tionality between Sleep Disturbances, Anxiety, and Depression. Sleep. 60. Chen J, et al. How will changes in resident work hours affect medical
2013;36(7):1059–68. students? Curr Surg. 2003;60(3):268–70.
37. Mental Health Blue Book: China’s National Mental Health Development 61. Blum AB, et al. Implementing the 2009 Institute of Medicine recommen-
Report (2019-2020)_ Xiaolan Fu, Kan Zhang, Xuefeng Chen, and Zhiyan dations on resident physician work hours, supervision, and safety. Nat Sci
Chen. https://​www.​pishu.​com.​cn/​skwx_​ps/​bookd​etail?​SiteID=​14&​ID=​ Sleep. 2011;3:47–85.
12369​491. Accessed 14 Apr 2022. 62. Friedman E, Karani R, Fallar R. Regulation of Medical Student Work Hours:
38. Jurkat HB, et al. Depression and stress management in medical students. A National Survey of Deans. Acad Med. 2011;86(1):30–3.
A comparative study between freshman and advanced medical students.
Nervenarzt. 2011;82(5):646–52.
39. Pukas L, et al. Prevalence and predictive factors for depressive symptoms Publisher’s Note
among medical students in Germany - a cross-sectional study. GMS. J Springer Nature remains neutral with regard to jurisdictional claims in pub-
Med Educ. 2022;39(1):Doc13. lished maps and institutional affiliations.
40. Gold JA, et al. Medical student depression and its correlates across three
international medical schools. World J Psychiatry. 2019;9(4):65–77.
41. Moayed MS, et al. Coronavirus (COVID-19)-Associated Psycho-
logical Distress Among Medical Students in Iran. Adv Exp Med Biol.
2021;1321:245–51.
42. Ford DE, Cooper-Patrick L. Sleep disturbances and mood disorders: an
epidemiologic perspective. Depress Anxiety. 2001;14(1):3–6.
43. Gao W, Ping S, Liu X. Gender differences in depression, anxiety, and stress
among college students: A longitudinal study from China. J Affect Disord.
2020;263:292–300.
44. Sun X, et al. Gender, negative life events and coping on different stages
of depression severity: A cross-sectional study among Chinese university
students. J Affect Disord. 2017;209:177–81.