Insomnia mediates the effect of perceived stress on emotional symptoms during the first wave of the COVID-19 pandemic in China

Abstract

The outbreak of the 2019 coronavirus disease (COVID-19) has significant effects on stress, emotion and sleep in the general public. The aim of this study was to explore the relationship between perceived stress and emotional symptoms during the first wave of the COVID-19 pandemic in China and to further determine whether insomnia could serve as a mediator in this relationship. A total of 1178 ordinary citizens living in mainland China conducted anonymous online surveys. The 10-item Perceived Stress Scale, the Insomnia Severity Index, the 9-item Patient Health Questionnaire and the 7-item Generalized Anxiety Disorder scale were used to estimate perceived stress, insomnia, depression and anxiety symptoms, respectively. Of the 1171 valid respondents from 132 cities in China, 46.6 % and 33.0 % showed symptoms of depression and anxiety, respectively. Perceived stress and insomnia independently predicted the prevalence of emotional symptoms and were positively correlated with the severity of these emotional symptoms. The mediation analyses further revealed a partial mediation effect of insomnia on the relationship between perceived stress and emotional symptoms during the first wave of the COVID-19 outbreak in China. Our findings can be used to formulate early psychological interventions to improve the mental health of vulnerable groups, specifically those with insomnia, during the COVID-19 pandemic.

1. Introduction

Since late December 2019, the 2019 coronavirus disease (COVID-19) was first reported in Wuhan, Hubei Province, China (Wang et al., 2020c) and then spread rapidly in China and throughout almost all countries in the world (Xu, 2020), becoming a public health emergency of international concern. A novel coronavirus, SARS-CoV-2, has been identified as the causative agent for COVID-19 (Zhou et al., 2020). Patients infected with SARS-CoV-2 may have mild to severe respiratory symptoms, such as fever, cough and shortness of breath, or even develop fatal respiratory diseases (Huang et al., 2020; Xu et al., 2020). Since SARS-CoV-2 is mainly transmitted between people through respiratory droplets and close contact (Chan et al., 2020) and can be contagious in asymptomatic and pre-symptomatic patients (Rothe et al., 2020), at 10 a.m. on January 23, one day before Chinese New Year’s Eve, Wuhan went into lockdown to stop the virus from spreading further across the country. Subsequently, other local governments adopted strict quarantine measures and mass home confinement to prevent people from being infected. Several factors related to the first wave of the COVID-19 pandemic in Wuhan have contributed to creating a continuously stressful environment for people, such as a growing number of suspected, confirmed and fatal cases, fears of falling sick and dying, and uncertainty about the outbreak.

At the peak of the first wave of the COVID-19 epidemic with strict lockdown measures, some special populations with certain occupations/diseases, such as acutely ill COVID-19 inpatients (Hao et al., 2020a), the workforce returning to work (Tan et al., 2020), and psychiatric patients (Hao et al., 2020b), showed high levels of depression, anxiety and stress, which is similar to previous studies indicating that perceived stress is related to the development of anxiety and depression symptoms (Mirón et al., 2019; Pun et al., 2018; Zhang et al., 2018). Some researchers also observed a wide range of emotional impacts on general population during the first wave of the COVID-19 outbreak. For example, a report from the initial stage of the COVID-19 outbreak in China showed that 16.5 % of participants had moderate to severe depressive symptoms and 28.8 % reported moderate to severe anxiety symptoms (Wang et al., 2020a). Another survey found that nearly half of the respondents felt helpless, horrified and apprehensive due to the pandemic (Zhang and Ma, 2020). These studies also mentioned the stress state during the COVID-19 pandemic, reporting moderate to severe stress levels of 8.1 % (Wang et al., 2020a) and 7.6 % (Zhang and Ma, 2020) in the local population in China. However, the possible mechanisms behind the relationship between perceived stress and emotional symptoms are not well described in current literature.

Existing literature indicates that sleep quality may play a significant mediating role in linking stress to emotional states (Lee and Hsu, 2012; Liu et al., 2017; Rusch et al., 2015). Liu et al. found that in the elderly, the influence of perceived stress on depression is partly mediated by poor sleep quality (Liu et al., 2017). Rusch et al. confirmed this finding and revealed that improving sleep quality is an effective intervention to reduce depression and posttraumatic arousal symptoms among deployed military personnel (Rusch et al., 2015).

Insomnia, as a common complaint of poor sleep quality, has been considered as an important psychological and physiological response to stressful events (Chen et al., 2017; Lee et al., 2019). According to the diathesis-stress model (also known as the “3P” model) proposed by Spielman (Spielman et al., 1987), acute insomnia is commonly caused by precipitating factors such as physiological, environmental, or psychological stressors that push individuals over the threshold of insomnia. COVID-19-related sleep disruption have been repeatedly reported and is common in the general population (Altena et al., 2020; Xiao et al., 2020). In the hardest-hit areas during the COVID-19 outbreak in China, participants with more frequent early awakenings or poor sleep quality reported higher posttraumatic stress symptoms (Liu et al., 2020). In addition, several studies have shown that adverse emotional states during the COVID-19 pandemic are associated with disturbed sleep (Huang and Zhao, 2020; Lai et al., 2020; Rajkumar, 2020). This is consistent with previous studies demonstrating that insomnia is a risk factor for depression and anxiety (Froese et al., 2008; Neckelmann et al., 2007; Roberts et al., 2000), as insomnia may increase the body’s susceptibility to adverse emotional responses while in a state of enhanced stress reactivity.

In light of the above considerations, this study was designed to investigate whether there was a direct and positive relationship between stress perception and emotional symptoms as well as whether insomnia played a mediating role in this relationship among the general population in mainland China during the first wave of the COVID-19 pandemic.

2. Methods

2.1. Participants

This cross-sectional study was conducted between February 14 and March 29, 2020, among the general public living in mainland China during the COVID-19 outbreak. The median number of days from the start date of Wuhan closure to the questionnaire survey date was 28 days (Interquartile Range (IR) = 19), ranging from 21 to 65 days. The study was approved by the ethics committee of the Institute of Psychology, Chinese Academy of Science prior to data collection and conformed to Declaration of Helsinki promulgated by the National Institute of Health.

A total of 1178 participants voluntarily participated and completed a series of electronic, anonymous questionnaires on socio-demographic characteristics, perceived stress, insomnia symptoms, depression symptoms and anxiety symptoms. Of these, six subjects were excluded because their IP addresses were outside China, and one was excluded for lack of certain socio-demographic characteristics. The final statistical analysis included 1171 valid respondents (age 22 ± 16 years, ranging from age 13 to 70 years, male/female = 359/812) from 132 cities in 31 provinces in China, with an effective response rate of 99.4 %.

2.2. Socio-demographic characteristics

The questionnaire included items covering socio-demographic information, such as age, sex, education, current residence, marital status, nationality, occupation, weight, height, smoking, drinking, annual family income and history of chronic disease. Economic losses during the COVID-19 outbreak were also collected. In addition, respondents were asked to state whether they had experienced the severe acute respiratory syndrome (SARS) epidemic in 2003 and whether any relatives or friends were infected with COVID-19.

2.3. Measures

2.3.1. Appraisal of perceived stress

Perception of stress was evaluated with the 10-item Perceived Stress Scale (PSS-10), a widely used psychological tool to measure the degree to which events are interpreted as stressful during the preceding month. The PSS-10 is rated on a 5-point Likert scale ranging from 0 (never) to 4 (very often). The four positively stated items (items 4, 5, 7, and 8) are reverse scored and the remaining six negatively stated items (items 1, 2, 3, 6, 9, and 10) are forward scored. The sum score of these items yields a global score ranging from to 0 to 40, with a higher score indicating greater perceived stress. Individuals were categorized as having high stress if the PSS-10 total score was >13. The Cronbach's α, a reliability coefficient, used in the present sample was 0.843.

2.3.2. Assessment of insomnia symptoms

Insomnia symptoms were assessed using a self-reported questionnaire, the 7-item Insomnia Severity Index (ISI), to index the diagnosis and severity of insomnia over the preceding two weeks. The ratings are summed, ranging from 0 to 28, with higher scores indicating more severe insomnia symptoms. Participants were categorized as having insomnia symptoms if the ISI score was 8 or higher (ISI ≥ 8). Our results yielded a reliability coefficient of 0.921.

2.3.3. Assessment of depression symptoms

Symptoms of depression were assessed by the 9-item Patient Health Questionnaire (PHQ-9), a self-assessment questionnaire based on the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV) diagnostic criteria for major depressive disorder within the previous two weeks. Each item uses a 4-point Likert scale ranging from 0 (not at all) to 3 (nearly every day). The total PHQ-9 score is calculated by the sum score of these items ranging from 0 to 27, and presumptive depressive symptoms were determined by a score >4. The Cronbach's α used in the present sample was 0.892.

2.3.4. Assessment of anxiety symptoms

Symptoms of anxiety were assessed by the 7-item Generalized Anxiety Disorder (GAD-7) scale, which has been widely used to measure the severity of anxiety based on DSM-IV diagnostic criteria for generalized anxiety disorder. Scoring of each item is based on a 4-point Likert scale ranging from 0 (not at all) to 3 (nearly every day), and the total GAD-7 score for the seven items ranges from 0 to 21. Presumptive anxiety symptoms were determined by a score >4. The Cronbach's α used in the present sample was 0.934.

2.4. Statistical analysis

IBM SPSS version 21.0 for Windows was used to perform all data processing (IBM SPSS Statistics, New York, USA). No missing data was included in final statistical analyses. The one-sample Kolmogorov-Smirnov test was used to detect distribution of the variables (normal distribution or non-normal distribution). The continuous variables were expressed as median and IR, and the categorical variables were expressed as frequency and proportion. For group comparisons in socio-demographic variables, PSS-10 and ISI scores were used with Mann-Whitney U test for continuous variables and Chi-Square test for categorical variables. Binary logistic regression analyses were performed to identify potential contributors for the occurrence of depression and anxiety, which remained significant (Ps < 0.05 in the group comparison analyses) after controlling for socio-demographic characteristics. Spearman's correlations were calculated to assess associations between PHQ-9, GAD-7, PSS-10, ISI and socio-demographic characteristics. Bonferroni corrections were used to adjust for multiple tests. In the hierarchical linear regression analyses, the PHQ-9 or GAD-7 score was used as the dependent variable, the PSS-10 and ISI scores were used as independent variables, and socio-demographic characteristics were used as covariates, which were Ps < 0.05 in the correlation analyses. Variance inflation factor values exceeding 5 were used to indicate multicollinearity. Mediation analysis was conducted using PROCESS V3.3 (model 4) in the SPSS software. The bootstrapping technique was used to estimate 5000 resamples of the data. Bias-corrected 95 % confidence intervals (CIs) did not include zero, indicating that the effects reached significant levels. Differences were considered to be statistically significant at P < 0.05.

3. Results

3.1. Socio-demographic characteristics of the study sample

Socio-demographic characteristics are presented in Table 1 . Of the 1171 valid respondents, the majority were female (69.3 %), age from 18 to 50 years (90.4 %), single (60.5 %), students (50.0 %), Han nationality (94.6 %), well educated (66.7 % ≥ bachelor's degree), normal BMI (55.5 %), and no history of medical diseases (86.9 %). Of all participants, 89.8 % did not currently smoke and 77.1 % did not currently drink alcohol; however, 30.1 % of smokers and 11.3 % of drinkers increased their use during the COVID-19 pandemic. Nearly half of the respondents had experienced SARS in 2003, and 0.7 % of those respondents had a relative or friend infected with COVID-19 during the pandemic. Most of the respondents (89.3 %) had an annual family income of <300,000 yuan per year, and 50 % of respondents clearly stated that they suffered economic losses ranging from a few hundred yuan to millions of yuan (median = 10,000 yuan).

Table 1.

Socio-demographic characteristics and the prevalence of depression and anxiety symptoms among the 1171 study participants.

Variables	Median or N	IR or %	Range
Age (years)	22	16	13–70
<18	78	6.7
18–25	569	48.6
26–50	490	41.8
>50	34	2.9
Sex
Male	359	30.7
Female	812	69.3
Education
No more than high school	96	8.2
Junior college	294	25.1
Undergraduate	674	57.6
Graduate	107	9.1
BMI (kg/m2)	21.48	4.45	13.93–36.33
Slim	118	10.1
Normal	650	55.5
Overweight	183	15.6
Obese	220	18.8
Marital status
Single	709	60.5
Married	412	35.2
Others	50	4.3
Nation
Han	1108	94.6
Others	63	5.4
Occupation
Students	586	50.0
Technicians	162	13.8
Managements	137	11.7
Services	108	9.2
Others	178	15.2
Annual family income (RMB, yuan/year)
<80,000	500	42.7
80,000–300,000	546	46.6
300,000–1,000,000	115	9.8
>1,000,000	10	0.9
Economic loss
No change	384	32.8
Loss	585	50.0
Unclear/unknown	202	17.2
Economic loss (RMB, yuan)	10,000	20,000	0–2,000,000
Current smoking
Non-smoker	1051	89.8
Smoker	83	7.1
Quit smoker	37	3.2
Smoking account (Cigarettes/day)
<5	40	48.2
5–15	34	41.0
>15	9	10.8
Change of smoking account
No change	58	69.9
Increased	25	30.1
Current drinking
Non-drinker	903	77.1
Drinker	213	18.2
Quit drinker	55	4.7
Drinking account (g/day)
<100	183	85.9
100–200	23	10.8
>200	7	3.3
Change of drinking account
No change	189	88.7
Increased	24	11.3
History of chronic diseases
No	1018	86.9
Yes	153	13.1
Cardio cerebrovascular disease	19	12.4
Liver diseases	6	3.9
Kidney disease	4	2.6
Diabetes mellitus	8	5.2
Other metabolic/endocrine diseases	31	20.3
Others	85	55.6
Ever experienced the SARS epidemic in 2003
No	614	52.4
Yes	557	47.6
Have any relatives or friends infected with COVID-19
No	1163	99.3
Yes	8	0.7
Days from the start date of Wuhan lockdown closure	28	19	21–65
Depression symptoms
PHQ-9 score	4	7	0–27
PHQ-9 ≤ 4	625	53.4
PHQ-9 > 4	546	46.6
Anxiety symptoms
GAD-7 score	2	6	0–21
GAD-7 ≤ 4	785	67.0
GAD-7 > 4	386	33.0
Perceived stress
PSS-10 score	18	7	0–40
PSS-10 ≤ 13	271	23.1
PSS-10 > 13	900	76.9
Insomnia symptoms
ISI score	3	7	0–28
ISI < 8	883	75.4
ISI ≥ 8	288	24.6

Note: N = number. IR = Interquartile Range; BMI = Body Mass Index; ISI = Insomnia Severity Index; PSS-10 = the 10-item Perceived Stress Scale; PHQ-9 = the 9-item Patient Health Questionnaire; GAD-7 = the 7-item Generalized Anxiety Disorder Scale; COVID-19 = the 2019 Coronavirus Disease; SARS = Severe Acute Respiratory Syndrome.

3.2. The prevalence of emotional symptoms among the study sample

The overall median score of PHQ-9 of respondents was 4 (IR = 7) and GAD-7 of respondents was 2 (IR = 6), as reflected in Table 1. Based on PHQ-9 and GAD-7 scores, 46.6 % of participants (n = 546) were considered to have depressive symptoms and 33.0 % (n = 386) were considered to have anxiety symptoms (Table 1). The respondents' median score of perceived stress shown on PSS-10 was 18 (IR = 7). Of all respondents, 900 (76.9 %) were considered to suffer from perceived stress (Table 1). Also of all respondents, 288 (24.6 %) were considered to have insomnia symptoms (Table 1).

3.3. Comparison of socio-demographic characteristics, perceived stress and insomnia in participants with and without emotional symptoms

As shown in Table 2 , there were significant differences in age, education, marital status, occupation, annual family income, economic losses during the COVID-19 outbreak, history of medical diseases, and days from the start date of Wuhan lockdown closure between depressed and non-depressed participants (Ps < 0.05). Table 2 also shows that compared to non-anxiety respondents, respondents with anxiety symptoms had significantly higher economic losses during the COVID-19 outbreak, shorter days from the start date of Wuhan lockdown, and a history of medical diseases (Ps < 0.05). Thus, these demographic characteristics were controlled as covariates in the following logistic regression analyses. In addition, respondents with depression and anxiety symptoms had significantly higher PSS-10 and ISI scores than those without depression and anxiety symptoms (Ps < 0.001) (Table 2).

Table 2.

Comparison of socio-demographic characteristics, perceived stress and insomnia among participants with and without emotional symptoms.

Variables	Depression		Anxiety
	U/χ2	P	U/χ2	P
Age	17.900	<0.001	2.611	0.456
Sex	1.817	0.178	0.244	0.622
Education	15.158	0.002	7.530	0.057
BMI	1.092	0.779	0.256	0.968
Marital status	16.049	<0.001	0.869	0.647
Nationality	0.380	0.537	0.237	0.626
Occupation	18.587	0.001	3.362	0.499
Annual family income	10.638	0.014	4.573	0.206
Economic loss	8.822	0.012	13.993	0.001
Current smoking	0.570	0.752	0.971	0.615
Current drinking	2.371	0.306	5.414	0.067
History of chronic diseases	14.175	<0.001	17.326	<0.001
Ever experienced the SARS epidemic in 2003	2.223	0.136	0.089	0.766
Have any relatives or friends infected with COVID-19	0.000	1.000	0.424	0.515
Days from the start date of Wuhan lockdown closure	153,157.500	0.002	140,408.000	0.035
PSS-10 score	233,108.000	<0.001	216,996.000	<0.001
ISI score	268,359.500	<0.001	229,507.000	<0.001

Note: BMI = Body Mass Index; ISI = Insomnia Severity Index; PSS-10 = the 10-item Perceived Stress Scale; COVID-19 = the 2019 Coronavirus Disease; SARS = Severe Acute Respiratory Syndrome. The bold values in the table represent P < 0.05, indicating that statistical significance has been achieved.

3.4. Independent factors predicted the prevalence of emotional symptoms

The results of binary logistic regression analyses showed that perceived stress (OR = 1.118, 95 % CI = 1.090–1.146, P < 0.001 for depression and OR = 1.134, 95 % CI = 1.102–1.166, P < 0.001 for anxiety) and insomnia (OR = 1.272, 95 % CI = 1.227–1.318, P < 0.001 for depression and OR = 1.178, 95 % CI = 1.145–1.212, P < 0.001 for anxiety) increased the risk of having depression and anxiety symptoms (Table 3 ). As shown in Table 3, older age (OR = 0.456, 95 % CI = 0.357–0.583, P < 0.001) and higher annual family income (OR = 0.774, 95 % CI = 0.621–0.965, P = 0.023) were protective factors for the development of depressive symptoms during the COVID-19 outbreak (Table 3). Moreover, longer days from the start date of Wuhan lockdown was a protective factor for suffering from anxiety symptoms during the COVID-19 pandemic (OR = 0.977, 95 % CI = 0.964–0.991, P = 0.001) (Table 3).

Table 3.

Logistic regression analyses predicting the prevalence of depression and anxiety symptoms.

Factors	Depression				Anxiety
	B	OR	95 % CI	P	B	OR	95 % CI	P
PSS-10 score	0.111	1.118	1.090–1.146	<0.001	0.125	1.134	1.102–1.166	<0.001
ISI score	0.240	1.272	1.227–1.318	<0.001	0.164	1.178	1.145–1.212	<0.001
Age	−0.785	0.456	0.357–0.583	<0.001	–	–	–	–
Annual family income	−0.256	0.774	0.621–0.965	0.023	–	–	–	–
Days from the start date of Wuhan lockdown closure	–	–	–	–	−0.023	0.977	0.964–0.991	0.001

Note: CI = Confidence Interval, OR = Odds Ratio; ISI = Insomnia Severity Index; PSS-10 = the 10-item Perceived Stress Scale. The bold values in the table represent P < 0.05, indicating that statistical significance has been achieved.

3.5. Correlations between emotional symptoms and perceived stress or insomnia

As shown in Table 4 , correlation analyses demonstrated that PHQ-9 and GAD-7 scores were positively correlated with the PSS-10 score (r = 0.410 and r = 0.464, Ps < 0.001) and ISI score (r = 0.549 and r = 0.476, Ps < 0.001), which remained significant after Bonferroni corrections (Ps < 0.0029). Younger age, single, student status, low annual family income, high economic loss during the outbreak of COVID-19, current drinking status and history of medical diseases were significantly positively correlated with PHQ-9 score (Ps < 0.05) (Table 4). After Bonferroni corrections, the associations of PHQ-9 score with age, marital status, occupation, and history of medical diseases remained significant (Ps < 0.001). Chinese minority, student status, high economic loss during the outbreak of COVID-19, and history of medical diseases were significantly associated with higher GAD-7 score (Ps < 0.05) (Table 4). After Bonferroni corrections, only the association between the GAD-7 score and history of medical diseases remained significant (P < 0.001). In addition, the closer the questionnaire survey date was to the Wuhan lockdown start date, the higher the PHQ-9 and GAD-7 scores (r = −0.107 and r = −0.072, Ps < 0.05) (Table 4). However, only the correlation between PHQ-9 and number of days from the survey date and the Wuhan lockdown start date remained significant after Bonferroni corrections (P < 0.001).

Table 4.

Correlations between emotional symptoms and perceived stress or insomnia.

Variables	PHQ-9		GAD-7
	rs	P	rs	P
Age	−0.113	<0.001	−0.035	0.236
Sex	−0.022	0.443	0.016	0.587
Education	0.003	0.908	0.049	0.097
BMI	0.002	0.936	0.012	0.679
Marital status	−0.123	<0.001	−0.054	0.067
Nationality	−0.035	0.236	−0.064	0.028
Occupation	−0.111	<0.001	−0.058	0.049
Annual family income	−0.082	0.005	−0.010	0.742
Economic loss	0.076	0.009	0.083	0.004
Current smoking	0.039	0.185	0.004	0.887
Current drinking	0.070	0.017	0.055	0.058
History of chronic diseases	0.147	<0.001	0.165	<0.001
Ever experienced the SARS epidemic in 2003	−0.058	0.046	0.005	0.868
Have any relatives or friends infected with COVID-19	0.019	0.515	0.013	0.659
Days from the start date of Wuhan lockdown closure	−0.107	<0.001	−0.072	0.013
PSS-10 score	0.410	<0.001	0.464	<0.001
ISI score	0.549	<0.001	0.476	<0.001

Note: BMI = Body Mass Index; ISI = Insomnia Severity Index; PSS-10 = the 10-item Perceived Stress Scale; PHQ-9 = the 9-item Patient Health Questionnaire; GAD-7 = the 7-item Generalized Anxiety Disorder Scale; COVID-19 = the 2019 Coronavirus Disease; SARS = Severe Acute Respiratory Syndrome. The bold values in the table represent P < 0.05, indicating that statistical significance has been achieved.

3.6. Independent relationships between emotional symptoms and perceived stress or insomnia

Hierarchical linear regression analyses were further used to determine independent relationships between emotional symptoms, perceived stress and insomnia. As shown in Table 5 , step 1 included the socio-demographic variables of Ps < 0.05 in the correlation analyses, which could explain the 6.1 % variation of depression (F_(9,1161) = 8.322, P < 0.001, R² = 0.061) and 3.6 % variation of anxiety (F_(5,1165) = 8.758, P < 0.001, R² = 0.036). In step 2, when perceived stress was added as an independent variable, this increased to 15.1 % variance of depression (F_(10,1160) = 31.195, P < 0.001, R² = 0.212) and anxiety (F_(6,1164) = 44.597, P < 0.001, R² = 0.187). In step 3, adding insomnia to the model accounted for 44.5 % variance of depression (F_(11,1159) = 84.623, P < 0.001, R² = 0.445) and 35.3 % variance of anxiety (F_(7,1163) = 90.754, P < 0.001, R² = 0.353), a total increase of 23.3 % and 16.6 %, respectively, of which was explained by the additional variable (ΔR² = 0.233 for depression and ΔR² = 0.166 for anxiety). The significant socio-demographic predictors in step 1 and 2 were no longer significant (Ps > 0.05) (Table 5).

Table 5.

Hierarchical regression analyses exploring the effects of perceived stress and insomnia on the severity of emotional symptoms.

Factors	PHQ-9					GAD-7
	B	t	P	F	R2	B	t	P	F	R2
Step 1				8.322	0.061				8.758	0.036
History of chronic diseases	2.949	6.313	<0.001			2.022	5.343	<0.001
Economic loss	0.696	3.397	0.001			0.512	3.041	0.002
Current drinking	0.615	2.201	0.028
Step 2				31.195	0.212				44.597	0.187
History of chronic diseases	2.018	4.665	<0.001			1.199	3.402	0.001
Economic loss	0.508	2.700	0.007			0.338	2.174	0.030
Current drinking	0.607	2.372	0.018
PSS-10 score	0.293	14.925	<0.001			0.237	14.687	<0.001
Step 3				84.623	0.445				90.754	0.353
History of chronic diseases	0.504	1.363	0.173			0.142	0.444	0.657
Economic loss	0.096	0.605	0.545			0.068	0.489	0.625
Current drinking	0.105	0.488	0.626
PSS-10 score	0.202	11.927	<0.001			0.177	11.962	<0.001
ISI score	0.494	22.090	<0.001			0.339	17.296	<0.001

Note: ISI = Insomnia Severity Index; PSS-10 = the 10-item Perceived Stress Scale; PHQ-9 = the 9-item Patient Health Questionnaire; GAD-7 = the 7-item Generalized Anxiety Disorder Scale. The bold values in the table represent P < 0.05, indicating that statistical significance has been achieved.

3.7. The mediating role of insomnia in the relationship between perceived stress and emotional symptoms

Fig. 1 shows the model of the relationship between perceived stress and emotional symptoms mediated by insomnia. After controlling for demographic variables, the bootstrap 95 % CI was above zero, demonstrating that insomnia mediated the relationship between perceived stress and emotional symptoms (Bias-corrected 95 % CI: 0.065–0.119 for depression and Bias-corrected 95 % CI: 0.042–0.080 for anxiety) (Table 6 ). The mediation models of the relationship between perceived stress and emotional symptoms mediated by insomnia were significant (F_(11,1159) = 84.623 for depression and F_(7,1163) = 90.754 for anxiety, both P < 0.001), accounting for 31 % and 25 %, respectively, of the mediating effect. This suggested that the 31 % of the variance in depression and the 25 % of the variance in anxiety were produced by insomnia acting as a mediator.

Fig. 1.

The estimated coefficients of mediation effects of insomnia on perceived stress and emotional symptoms. ***P < 0.001.

Table 6.

Mediation analyses of insomnia between perceived stress and emotional symptoms.

Paths	B	SE	Bias-corrected 95 % CI		Ratio
			Lower	Upper
PSS-10 → PHQ-9
Indirect effect	0.090	0.014	0.065	0.119	31 %
Direct effect	0.202	0.017	0.169	0.236	69 %
Total effect	0.293	0.020	0.254	0.331
PSS-10 → GAD-7
Indirect effect	0.060	0.010	0.042	0.080	25 %
Direct effect	0.177	0.015	0.148	0.206	75 %
Total effect	0.237	0.016	0.205	0.268

Note: CI = Confidence Interval; PSS-10 = the 10-item Perceived Stress Scale; PHQ-9 = the 9-item Patient Health Questionnaire; GAD-7 = the 7-item Generalized Anxiety Disorder Scale.

4. Discussion

To the best of our knowledge, this large-scale study investigated for the first time the relationship between perceived stress and emotional symptoms, and estimated the mediating effect of insomnia on this relationship during the peak of the first wave of the COVID-19 pandemic among the general public in mainland China. The results showed that 46.6 % and 33.0 % of the participants experienced depression and anxiety symptoms during the first wave of the COVID-19 pandemic in Wuhan, respectively. Participants with emotional symptoms exhibited more stress and insomnia symptoms than those without emotional symptoms. Perceived stress and insomnia independently predicted the occurrence of depression and anxiety symptoms and were positively correlated with the severity of these emotional symptoms during the first wave of the COVID-19 outbreak. The study also revealed that perceived stress contributed to depression and anxiety partially due to its effect on insomnia.

4.1. The prevalence of emotional symptoms during the first wave of the COVID-19 pandemic

In this study, the prevalence of depression symptoms during the first wave of the COVID-19 pandemic was 46.6 %, which is significantly higher than that in cross-sectional studies where the prevalence of depression in the Chinese public ranged from 16.5 % to 20.1 % (Huang and Zhao, 2020; Wang et al., 2020a). However, this finding is similar to previous observations showing that about 50 % of participants felt helpless, horrified and apprehensive in the general public (Zhang and Ma, 2020), and also showing that 50.4 % of health care workers exposed to COVID-19 reported depressive symptoms (Lai et al., 2020). Our study additionally found the prevalence of anxiety symptoms during the COVID-19 pandemic to be 33.0 %, which is again similar to previous studies that showed the prevalence of anxiety symptoms ranging from 28.8 % to 44.6 % (Huang and Zhao, 2020; Lai et al., 2020; Wang et al., 2020a; Wang et al., 2020b). Together, these findings suggest that the COVID-19 outbreak has had significant impacts on the emotion of the general population in mainland China.

4.2. Factors associated with the prevalence of emotional symptoms during the first wave of the COVID-19 pandemic

The prevalence rates of depression and anxiety symptoms during the COVID-19 pandemic were associated with several factors. Previous studies have revealed that higher levels of depression and anxiety symptoms are significantly associated with female gender, younger age, single status, student status, low income, specific physical symptoms (e.g., myalgia, dizziness, coryza) and a poor self-rated health status (Losada-Baltar et al., 2020; Wang et al., 2020a). In the present study, age and annual family income were significant predictors of depressive symptoms. In addition, in logistic regression analyses, the number of days from the Wuhan lockdown start date was a predictor of anxiety symptoms.

In term of socio-demographic characteristics, our results were not much different from other studies. However, our study revealed that perceived stress and insomnia independently predicted the occurrence of depression and anxiety symptoms during the first wave of the COVID-19 outbreak. Additionally, these findings are consistent with previous correlation analyses showing that emotional symptoms were significantly associated with perceived stress and/or insomnia during the COVID-19 outbreak (Huang and Zhao, 2020; Lai et al., 2020; Rajkumar, 2020; Wang et al., 2020a; Zhang and Ma, 2020). Our correlation analyses also confirmed these observations, indicating that perceived stress and insomnia were positively correlated with depression and anxiety symptoms. These associations may be directly associated with people’s persistent stress due to fears of the COVID-19 pandemic and massive negative news or with sleep disturbances caused by continued isolation at home (Altena et al., 2020; Xiao et al., 2020).

4.3. Insomnia partly mediated the relationship between perceived stress and emotional symptoms during the first wave of the COVID-19 pandemic

More importantly, this study specifically explored the mediating role of insomnia in the relationship between perceived stress and emotional symptoms during the first wave of the COVID-19 pandemic. This finding is consistent with previous studies, showing that insomnia acted as an important mediator between stress and emotional states (Lee et al., 2019; Liu et al., 2017; Rusch et al., 2015). For example, Liu et al. found that shortened sleep duration and daytime dysfunction partly mediated the impact of perceived stress on depression among the elderly in urban communities (Liu et al., 2017). Similarly, Lee et al. revealed an interactive effect of insomnia and recent stressful life events on emotional symptoms in a community dwelling population, suggesting that insomnia may be an indicator of emotional symptoms after stressful life events (Lee et al., 2019). Furthermore, a study of military personnel showed that improved sleep quality had a positive effect on the association between post-traumatic arousal symptoms and depression (Rusch et al., 2015). These findings support our findings by demonstrating the partially mediating effect of insomnia on the association between stress and mental health.

Mechanisms related to the “3P” behavioral model of insomnia and neurobiological pathways may help explain the role of insomnia on the association between perceived stress and emotional symptoms. According to the conceptualization of the “3P” model of insomnia, the precipitating factor of stress, promoted the development of acute insomnia (Spielman et al., 1987). During the COVID-19 pandemic, many studies have repeatedly confirmed the coexistence of stress and insomnia (Liu et al., 2020), which is consistent with previous studies that insomnia has been identified as an important psychological and physiological response to stress events (Chen et al., 2017; Lee et al., 2019).

Studies have shown that stress-related insomnia may be caused by the activity of the hypothalamo-pituitary-adrenal (HPA) system (Han et al., 2012). Due to the weakening of the inhibition of the HPA axis by slow wave sleep, HPA axis dysregulation was commonly seen in patients with insomnia (Hirotsu et al., 2015). Studies have also indicated that patients with depression and anxiety showed hyperactivity of the HPA axis, which is thought to be central to the genesis of these illnesses (Chen et al., 2015). Thus, the HPA axis has been considered as a shared biological substrate of both insomnia and emotional disorders (Asarnow, 2019). In summary, HPA axis hyperactivity is an important physiological link between stress, insomnia and emotional disorders. Stress may activate the HPA system and induce insomnia, which may further affect HPA functions, leading to an increased risk of depression and anxiety symptoms. This plausible explanation supports insomnia having mediator effects on the relationship between perceived stress and emotional disturbances.

4.4. Limitations

This study had some limitations. First, we adopted a cross-sectional design in this study. Thus, any causal relationship could not be drawn between perceived stress, insomnia and emotional symptoms. Second, due to selection bias caused by oversampling of students, young people accounted for a relatively large proportion in the study population. As a result, the conclusion could not be generalized to the entire population, particularly to the elderly and those with a lower education level. Third, most of respondents might be those who had the financial, emotional, and mental latitudes to appropriately complete the survey, and the number of people who had confirmed or suspected cases of COVID-19 was very small, also limiting the generalization of our findings. Fourth, the measurements of insomnia, depression and anxiety symptoms by the self-reported questionnaire alone are not equivalent to the clinical diagnosis based on the “gold standard” psychiatric interview. Finally, the rating scales (especially the anxiety scale) are not specific to viral epidemics, which means that it is impossible to determine whether the measured anxiety levels are from SARS-CoV-2 viral anxiety.

5. Conclusion

In summary, this study indicated that during the first wave of the COVID-19 outbreak in China, nearly half of the respondents suffered from depressive symptoms, and more than one-third reported anxiety symptoms. Perceived stress and insomnia were associated with the levels of these emotional symptoms. There may be partially mediating effects of insomnia contributing to the impact of perceived stress on depression and anxiety symptoms. These results suggest the importance of early intervention in treating insomnia to mitigate the severity of depression and anxiety in stressful situations. Cognitive behavioral therapy (CBT), based on the 3P model for insomnia, has strong evidence of being an effective intervention for insomnia symptoms (Rossman, 2019). CBT, specifically Internet CBT (I-CBT), has been found to be an effective treatment for insomnia during the COVID-19 pandemic (Soh et al., 2020). This therapy has also been shown to be effective for psychiatric symptoms during the COVID-19 pandemic (Ho et al., 2020). Therefore, the use of low-cost I-CBT (e.g. Moodle) to intervene in insomnia to reduce emotional symptoms under stress may be effective and worth promoting (Zhang and Ho, 2017). Importantly, as the COVID-19 pandemic continues, the pandemic-related burnout/fatigue is increasing significantly, although the continued adherence to a dynamic zero-COVID strategy has been effective in curbing the spread of COVID-19 (Lau et al., 2022). Many individuals are experiencing increased levels of stress, anxiety, depression, fear, post-traumatic stress and burnout. Future research is needed to confirm the current findings under the dynamic zero-COVID strategy and to further evaluate the impact of the COVID-19 pandemic in the larger population.

Role of the funding source

This research work was supported by the CAS Pioneer Hundred Talents Program (Xiang-Yang Zhang), Sanming Project of Medicine in Shenzhen (No. SZSM202011014), Shenzhen Fund for Guangdong Provincial High-level Clinical Key Specialties (No. SZGSP013), Shenzhen Key Medical Discipline Construction Fund (No. SZXK072), and Shenzhen Science and Technology Research and Development Fund for Sustainable Development Project (No. KCXFZ20201221173613036). This source had no further role in this study design, in the data collection and analysis, in the writing of the report, and in the decision to submit the paper for publication.

CRediT authorship contribution statement

Y. Z., Y. W., and Y. W. were responsible for collecting the samples. L. M. and X. Z. were responsible for statistical analysis and manuscript preparation. G. J. and H. W. were involved in evolving the ideas and editing the manuscript. X. Z. was responsible for study design. X. Z. and J. W. were responsible for providing the funding for the study. All authors have contributed to and have approved the final manuscript.

Conflict of interest

All authors have indicated no financial conflicts of interest.