Edited by: Zhang Jinghuan, Shandong Normal University, China
Reviewed by: Lei Han, Shandong Normal University, China; Scott Meier, University at Buffalo, United States
This article was submitted to Cognition, a section of the journal Frontiers in Psychology
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
Quarantine and isolation at extended length, although considered as highly effective countermeasures for the novel coronavirus (COVID-19) which started at the end of 2019, can have great impact on individual's mental health, especially emotional state. The present research recruited 5,115 participants from the general public across 32 provinces and autonomous regions in China in an online survey study, about 20 days after the lockdown of the epicenter (Wuhan), to investigate the relationship between the length of the quarantine and negative affect (including depression and anxiety), as well as the mediating roles of negative cognition (including worry and anticipation), and the moderating roles of dispositional optimism, tolerance of uncertainty, social support, and healthy behavior. The results showed that: (1) Worry and anticipation mediated the relationship between quarantine length and depression and anxiety; (2) Dispositional optimism moderated the path coefficients of quarantine length to worry, worry to anxiety, and anticipation to depression; (3) Tolerance of uncertainty moderated the path coefficient of worry to anxiety; (4) Social support moderated the path coefficient of anticipation to anxiety. In conclusion, during quarantine, dispositional optimism, uncertainty tolerance, and social support can buffer the direct or indirect effects of quarantine length on depression and anxiety. These findings could have profound implications on the societal responses to COVID-19 and future pandemics.
Due to the high infectivity of the novel coronavirus disease (COVID-19), Wuhan was locked down on January 23, 2020 to curb the further deterioration and spread of COVID-19. On January 30, 2020, the World Health Organization (WHO) declared the outbreak of COVID-19 as an international public health emergency. On March 11, 2020, the WHO further declared COVID-19 as a pandemic largely due to the uprising COVID-19 cases in Europe, especially Italy, and the United States. As of today (mid-May), confirmed cases in the United States reached 1.5 million.
As the first country to fight the COVID-19 epidemic, China has accumulated valuable experience in detection, clinical diagnosis and treatment, epidemiological statistics, and transmission control. The sharp decrease in the number of confirmed cases within two months in China after the Wuhan lockdown has demonstrated that self-quarantine/isolation
However, little empirical research has assessed the potential mediating processes on the effects of quarantine on depression and anxiety, even though this knowledge could be important for the general public to cope with the potential mental health issues from extended isolation. The present study was thus carried out about 20 days after the lockdown of Wuhan, the epicenter of the coronavirus outbreak in China. At that time, as the number of confirmed cases in China was still rising, many cities across the country were in self quarantine at home. We assessed various individual difference factors, including worry, anticipation, dispositional optimism, tolerance of uncertainty, social support, and healthy behaviors, for the relationship between quarantine length and negative affect.
According to Schulz and Lazarus (
The extent to which the individuals are more less affected by quarantine and negative cognition (including worry and anticipation) may be closely related to the inner resources such as dispositional optimism. Dispositional optimism is defined as the belief that future events may have positive results (Lai and Yue,
In addition, given the extreme uncertainty regarding the quarantine length and whether the individual will be infected during the COVID-19 epidemic, another important individual factor will be tolerance of uncertainty. It is the set of negative and positive psychological response—cognitive, emotional, and behavioral—provoked by the conscious awareness of the lack knowledge about particular aspects of the world (Hillen et al.,
According to the buffer model of social support (Thoits,
As for the last factor, there is a large literature demonstrating the effects of healthy behaviors on mental health. For instance, exercise and healthy diet can reduce depression and anxiety (Byrne and Byrne,
In summary, the present study assessed several individual difference factors that may contribute to the effects of quarantine length on depression and anxiety in the general public under self-quarantine during the early phase of COVID-19 epidemic in China. It is expected that (a) quarantine length should predict depression and anxiety; (b) negative cognitive factors (including anticipation and worry) should mediate this relationship between them; (c) dispositional optimism, uncertainty tolerance, social support, and healthy behaviors should further moderate these relationships.
Potential participants were recruited via online advertisements on social media. Using convenience sampling, 5,115 Chinese residents (72.75% females) were recruited in this online survey study between February 11, and February 19, 2020. The participants were from 32 provinces or autonomous regions including Hong Kong, Macao and Taiwan. Participants' age ranged from 15–71 years [mean = 21.27, standard deviation (
The survey study was conducted online in computer and smartphone friendly format. Participation of the study was anonymous and voluntary. The average completion time was 8.98 min. No monetary compensation was provided to the participants. This study was approved by the body for ethical evaluation of research projects at the Department of Psychology—part of the School for Social and Behavioral Sciences at Nanjing University, China. All procedures contributing to this work comply with the ethical standards of the relevant national and institutional committees on human experimentation and with the Helsinki Declaration of 1975, as revised in 2008. Informed consent was obtained from all individual participants included in the study.
The quarantine length was a self-report measure using a single survey question: “how many days have you been quarantined?” The score was as follows: 1 = 0 days, 2 = 1–7 days, 3 = 8–14 days, 4 = more than 15 days.
We used the depression subscale of Brief Symptom Inventory (Derogatis,
We used the anxiety subscale of Brief Symptom Inventory (Derogatis,
We used three items adapted from the McCaul Brief Worry Scale (McCaul and Goetz, n.d.) to assess worry (e.g., how worried are you about the coronavirus?). One item was rated on a 4-point scale, ranging from 1 (
We used two self-designed items to asses anticipation (i.e., “Do you think that you will contract coronavirus?” and “Do you think that your family will contract coronavirus?”). Items are rated on a 5-point scale, ranging from 1 (
We used the Life Orientation Test-Revised (Scheier et al.,
We used the Intolerance of Uncertainty–Short Form (Carleton et al.,
We used five items adapted from the Medical Outcomes Study Social Support Survey (Sherbourne and Stewart,
We used three self-designed items to assess how many days they engaged in three healthy behaviors over the past week (i.e., engaging in aerobic physical activity for at least 15 min, engaging in strengthening exercises, and eating fruits or vegetables). Items were rated on an 8-point scale, ranging from 1 (0 day) to 8 (7 days), to indicate the extent to which each statement applied to the participant. The scale had an acceptable Cronbach's alpha score of 0.65.
The Statistical Package for Social Sciences (SPSS) Version 22.0 and AMOS Version 22.0 were used for data analyses. Firstly, descriptive statistical analysis and statistical analysis of the differences in demographic variables were performed using independent
The results of the descriptive statistic of each demographic variable were shown in
Descriptive statistics of and difference in demographic variables of all study variables (
Total | 1.69 ± 1.20 | 3.42 ± 3.67 | 3.57 ± 4.02 | 6.78 ± 1.92 | 3.25 ± 1.42 | 22.77 ± 3.94 | 45.33 ± 7.97 | 13.23 ± 3.15 |
Male | 1.68 ± 1.18 | 3.44 ± 3.88 | 3.32 ± 4.18 | 6.45 ± 2.13 | 3.18 ± 1.45 | 21.92 ± 4.13 | 43.72 ± 8.57 | 12.75 ± 3.41 |
Female | 1.69 ± 1.20 | 3.41 ± 3.59 | 3.66 ± 3.95 | 6.90 ± 1.82 | 3.27 ± 1.41 | 23.09 ± 3.82 | 45.93 ± 7.65 | 13.41 ± 3.02 |
−0.43 | 0.26 | −2.65 |
−6.91 |
−2.02 | −9.20 |
−8.45 |
−6.35 |
|
— | — | −0.08 | −0.23 | — | −0.29 | −0.27 | −0.20 | |
Key regions | 2.60 ± 1.46 | 4.21 ± 3.88 | 5.27 ± 4.94 | 7.59 ± 1.96 | 4.17 ± 1.63 | 22.01 ± 3.79 | 45.73 ± 7.51 | 13.22 ± 3.34 |
Non-key regions | 1.68 ± 1.19 | 3.41 ± 3.66 | 3.54 ± 3.99 | 6.76 ± 1.92 | 3.23 ± 1.41 | 22.79 ± 3.94 | 45.32 ± 7.98 | 13.23 ± 3.14 |
5.69 |
1.95 | 3.14 |
3.85 |
5.92 |
−1.82 | 0.45 | −0.03 | |
0.69 | — | 0.39 | 0.43 | 0.62 | — | — | — | |
< ¥50,000 (1) | 1.62 ± 1.16 | 3.36 ± 3.63 | 3.50 ± 3.91 | 6.68 ± 1.89 | 3.16 ± 1.37 | 22.54 ± 3.97 | 45.59 ± 8.03 | 12.91 ± 3.19 |
¥50,000–¥100,000 (2) | 1.67 ± 1.18 | 3.38 ± 3.50 | 3.41 ± 3.80 | 6.79 ± 1.83 | 3.25 ± 1.40 | 23.20 ± 3.77 | 45.42 ± 7.61 | 13.51 ± 2.99 |
¥100,000–¥200,000 (3) | 1.83 ± 1.26 | 3.62 ± 3.86 | 3.91 ± 4.47 | 6.98 ± 2.03 | 3.38 ± 1.48 | 22.49 ± 4.04 | 43.91 ± 8.10 | 13.63 ± 3.17 |
¥200,000–¥300,000 (4) | 2.09 ± 1.37 | 3.68 ± 4.06 | 4.23 ± 4.38 | 7.31 ± 1.98 | 3.85 ± 1.64 | 22.96 ± 4.11 | 45.12 ± 8.37 | 13.87 ± 3.10 |
>¥300,000 (5) | 1.99 ± 1.32 | 3.7 ± 4.51 | 4.26 ± 5.10 | 6.90 ± 2.53 | 3.64 ± 1.74 | 23.07 ± 4.22 | 45.43 ± 9.04 | 13.63 ± 3.37 |
11.65 |
1.23 | 4.28 |
6.38 |
14.73 |
7.93 |
5.31 |
14.75 |
|
0.01 | — | 0.003 | 0.01 | 0.01 | 0.01 | 0.004 | 0.01 | |
3 > 1; 4 > 1; 5 > 1; 4 > 2; 5 > 2 | — | Ns | 3 > 1; 4 > 1; 4 > 2; | 3 > 1; 4 > 1; 4 > 2; 4 > 3; 5 > 1; 5 > 2 | 2 > 1; 2 > 3 | 1 > 3; 2 > 3 | 2 > 1; 3 > 1; 4 > 1; 5 > 1 | |
Below college degree (1) | 1.76 ± 1.21 | 4.00 ± 4.21 | 4.26 ± 4.64 | 6.48 ± 2.10 | 3.37 ± 1.49 | 22.25 ± 4.18 | 44.84 ± 8.60 | 12.62 ± 3.25 |
Undergraduate (2) | 1.67 ± 1.18 | 3.34 ± 3.58 | 3.44 ± 3.87 | 6.76 ± 1.89 | 3.20 ± 1.39 | 22.83 ± 3.93 | 45.34 ± 7.94 | 13.27 ± 3.15 |
postgraduate (3) | 2.25 ± 1.41 | 4.79 ± 4.83 | 6.24 ± 5.70 | 7.52 ± 2.31 | 4.36 ± 1.61 | 21.80 ± 3.91 | 45.66 ± 8.04 | 12.63 ± 2.93 |
21.18 |
15.99 |
45.78 |
15.73 |
59.77 |
7.62 |
0.50 | 7.33 |
|
0.01 | 0.01 | 0.02 | 0.01 | 0.02 | 0.003 | — | 0.003 | |
3 > 1; 3 > 2 | 3 > 2 | 1 > 2; 3 > 1; 3 > 2 | 3 > 1; 3 > 2 | 3 > 1; 3 > 2 | 2 > 3 | — | 2 > 3 |
As shown in
Intercorrelations between all study measures.
1. QL | 1 | |||||||
2. Depression | 0.14 |
1 | ||||||
3. Anxiety | 0.12 |
0.73 |
1 | |||||
4. Worry | 0.13 |
0.25 |
0.35 |
1 | ||||
5. Anticipation | 0.13 |
0.26 |
0.29 |
0.26 |
1 | |||
6. DO | −0.07 |
−0.36 |
−0.29 |
−0.09 |
−0.23 |
1 | ||
7. TU | −0.08 |
−0.31 |
−0.31 |
−0.24 |
−0.14 |
0.29 |
1 | |
8. SS | 0.02 | −0.20 |
−0.14 |
0.03 | −0.07 |
0.38 |
0.03 | 1 |
According to the results of the correlation analyses, quarantine length, worry, anticipation, depression, and anxiety were related to each other, which meets the requirements of the multiple mediation model (Marsh et al.,
a) As all variables are on single-dimension scales, they were treated as latent variables and their items were treated as explicit variables to better fit models.
b) Given the significant correlation between worry and anticipation, the correlation between worry and anticipation was established in order to avoid Type I error in expanding the model calculation results. More importantly, considering the often cooccurring and correlated depression and anxiety, we also established a correlation between depression and anxiety.
c) Because of the significant differences in several measures across gender, current residence, family income, and education level, these variables with significant differences were used as covariates to test the model among quarantine length, worry, anticipation, depression, and anxiety. However, for the sake of visual presentation of the results, they were not shown in the figures presented in this paper.
The SEM produced reasonable fits of the data (χ
The multiple mediation model among quarantine length, worry, anticipation, depression, and anxiety. ***
In order to investigate the moderated role of dispositional optimism, tolerance of uncertainty, and social support, SEM multiple-group analysis was carried out with these measures as grouping variables, respectively. Specifically, using the ±1 SD as the grouping standard, the high and low group were selected to fit the model in
Model 1(unconstrained model): The same model structure was defined for different groups, without any restrictions on each parameter in the model;
Model 2 (measurement model): Based on Model 1, the path coefficients of different groups of measurement models were limited to be equal;
Model 3 (structural model): Based on Model 2, the path coefficients of different groups of structural model parts were limited to be equal.
If there was a significant difference between the two groups in some path coefficients, a simple slope test (Preacher et al.,
The results showed that the model fit indices for Model 1, Model 2, and Model 3 were good (see
Models' fit indices.
Model 1 | 706.10 | 338 | 2.09 | 0.98 | 0.96 | 0.97 | 0.02 |
Model 2 | 959.69 | 351 | 2.73 | 0.96 | 0.94 | 0.95 | 0.03 |
Model 3 | 1013.27 | 359 | 2.82 | 0.96 | 0.94 | 0.95 | 0.03 |
Model 1 | 632.92 | 338 | 1.87 | 0.98 | 0.95 | 0.97 | 0.02 |
Model 2 | 697.29 | 351 | 1.99 | 0.97 | 0.95 | 0.97 | 0.03 |
Model 3 | 722.73 | 359 | 2.01 | 0.97 | 0.95 | 0.96 | 0.03 |
Model 1 | 788.156 | 338 | 2.33 | 0.98 | 0.96 | 0.97 | 0.03 |
Model 2 | 858.25 | 351 | 2.45 | 0.97 | 0.95 | 0.96 | 0.03 |
Model 3 | 876.59 | 359 | 2.44 | 0.97 | 0.5 | 0.96 | 0.03 |
The moderated role of dispositional optimism, tolerance of uncertainty, and social support.
Simple slope test results.
QL → Worry | 0.15 | 7.72 |
0.07 | 3.64 |
— | — | — | — | — | — | — | — |
QL → Anticipation | — | — | — | — | — | — | — | — | — | — | — | — |
Worry → Depression | — | — | — | — | — | — | — | — | — | — | — | — |
Worry → Anxiety | 0.38 | 23.44 |
0.22 | 12.40 |
0.33 | 20.44 |
0.21 | 10.89 |
— | — | — | — |
Anticipation → Depression | 0.25 | 15.06 |
0.09 | 4.25 |
— | — | — | — | — | — | — | — |
Anticipation → Anxiety | — | — | — | — | — | — | — | — | 0.32 | 17.57 |
0.22 | 11.01 |
QL → Depression | — | — | — | — | — | — | — | — | — | — | — | — |
QL → Anxiety | — | — | — | — | — | — | — | — | — | — | — | — |
The results showed that the model fit indices of Model 1, Model 2, and Model 3 were good (see
The results showed that the model fit indices of Model 1, Model 2, and Model 3 were good (see
In this study, SEM was used to investigate the relationship between quarantine length and negative affect, as well as the roles of negative cognitions and several protective factors. The results showed that anticipation and worry partially mediated the relationship between quarantine length and negative affect. Dispositional optimism, tolerance of uncertainty, and social support moderated one or more paths of the relationship among quarantine length, anticipation, worry, depression, and anxiety. High dispositional optimism, high tolerance of uncertainty, and good social support can decrease the prediction effects of some paths.
Consistent with longer quarantine in the key regions, anxiety, worry, and anticipation of infection were higher in residents from the key regions than those in residents from non-key areas. Interestingly, women experienced worse anxiety, and worry of infection than men did on the one hand, but better dispositional optimism, social support, and tolerance of uncertainty on the other hand, directly replicating a previous finding of increased experience in both negative and positive affect (Yue et al.,
Another demographic factor, family annual income, also had similar impacts on most measures. Specifically, worry, anticipation, and tolerance of uncertainty of individuals with family annual income of more than ¥200,000 were worse than those with a family annual income of < ¥200,000. However, this high-income group also showed better experience in dispositional optimism and social support at the same time. These seemingly contradictory findings may result from the association between the level of annual family income and the source of the family income. On the one hand, some industries have high income but low stability (e.g., self-employed households), while others have low to medium income but high stability (e.g., civil servants). It is possible that individuals with annual household income higher than ¥200,000 had less stability in maintaining their income. Consequently, worry, anticipation and intolerance of uncertainty of individuals were in these participants, which is consistent with a recent finding that the level of mental health of individuals with unstable family income was low in the epidemic (Cao et al.,
It is worth noting that quarantine length, depression, anxiety, worry, anticipation, dispositional optimism and social support were worse in postgraduates than participants with junior college degree or below. The worse emotional well-being in postgraduates in the present study is consistent with the overall findings of the potential mental health issues in postgraduates (e.g., Hou et al.,
One of the key finding of the present study is that quarantine length, anticipation, worry, depression, and anxiety are correlated with each other. Removal of the social environment during the quarantine can be an important source of psychological stress. In the literature, lots of research on quarantine are based on analog simulation of space environment. For example, participants in bed rest for 60 days exhibited fluctuations (high-low-high-low) in depression and anxiety over time (Qin et al.,
On the basis of correlation analysis, we further use SEM to investigate the mediating effect of anticipation and worry on quarantine length and negative affect. Consistent with the hypothesis, worry and anticipation about the COVID-19 mediate the relationship between quarantine length and negative affect. Specifically, the longer the durations of quarantine are, the greater anticipation and worry of individuals and their families to contract the virus are, which further worsens depression and anxiety symptoms. This result is in line with Schulz and Lazarus (
The results of multiple-group analyses highlighted three factors. First, dispositional optimism moderated the path coefficients of quarantine length to worry, worry to anxiety, and anticipation to depression. These novel findings are consistent with some previous evidence that dispositional optimism can act as a buffer. For instance, dispositional optimism can alleviate the relationship between stress and mental health (Chang,
Second, tolerance of uncertainty moderated the path coefficient of worry to anxiety. Tolerance of uncertainty has recently gained research interests in the health care context, given the various sources of uncertainties in clinical setting (Hillen et al.,
Third, social support moderated the path coefficient of anticipation to anxiety. At present, the relationship between social support and physical and mental health has reached a general consensus, but the mechanisms of social support are still controversial (Cohen and Wills,
This study examined the effects of quarantine length on negative affect during COVID-19 and the potential mediating and moderating factors. The moderating effects have identified dispositional optimism, uncertainty tolerance, and social support as potential psychological buffers for coping with the negative affect experienced during COVID-19. These protective factors are supplementary to those reported in a recent study (Wang et al.,
Nonetheless, this study has some caveats. Firstly, with a cross-sectional design (i.e., quarantine length is measured across respondents) and the regression approach, the present findings do not provide any evidence for a causal relationship between quarantine length and negative affect. Second, with the development of epidemic situation over time, the observed relationships among the various factors and measures may be dynamic and different at different key timepoints. As a result, it is unclear whether our current findings can be generalized from the study period (20 days after the outbreak) to other time points of this epidemic. Thirdly, to facilitate participant recruitment and to ensure the data quality for this online survey study at a particularly stressful time, we tried to limit the study length to roughly 10 min. Consequently, most measures in the present study used short scales instead of the complete scales. Although these short scales have been well-established in the literature, future research needs to implement the full scales to get a more systematical assessment of the various outcome measures. Fourthly, the short quarantine and measurement of quarantine length as Likert-5 scale rather than accurate days limits the statistical power, which affected the effect sizes, especially the relationship involving quarantine length. Therefore, caution is still advised before drawing conclusions. Fifthly, the method of multiple-group analysis in moderated effect tests leads to waste of participants, so the results of the moderated variable as continuous variables as shown in the Supplements can also be considered. Sixthly, due to the number of non-key regions was significantly larger than the number of non-key regions (27 vs. 636 cities), 1.59% of participants were from the key regions, whereas approximately 98.41% of participants were from the non-key regions in this study, so the present findings may mainly reflect the epidemic related mental health issues in these non-key regions. Seventhly, the method of convenient sampling limits the generalization of our conclusion. On the one hand, according to study demographics, the samples essentially represent Chinese female college students, as indicated by gender ratio (72.75% females) and average age (21.27 ± 4.40); on the other hand, based on the significant negative correlations between completion time and quarantine length (
This study found that quarantine length could predict depression and anxiety. This relationship is further mediated by worry and anticipation about the COVID-19, and moderated by several protective factors, including dispositional optimism, uncertainty tolerance, and social support.
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.
The studies involving human participants were reviewed and approved by the body for ethical evaluation of research projects at the Department of Psychology—part of the School for Social and Behavioral Sciences at Nanjing University, China. The patients/participants provided their written informed consent to participate in this study.
LH, FL, and YM collected the data. LH and RZ analyzed and interpreted the data. LH, XC, and WZ wrote the current version of the manuscript. All authors have made a significant contribution to this work.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
We would like to express our gratitude for the support of the projects. Also, we are really grateful to Prof. Kate Sweeny, Dr. Campbell, and Dr. Da Jiang for providing help in the experimental design, language check and edit, and giving suggestions, respectively.
The Supplementary Material for this article can be found online at:
1According to Centers for Disease Control and Prevention (
2We adopted the classification launched by the Shanghai Government, China. Key regions refer to the regions with more confirmed cases and were perceived to have a higher level of risk of infection. Non-key regions refer to the regions with no or fewer confirmed cases and were perceived to have a lower level of risk of infection. According to the classification of Shanghai City, 27 cities in China were classified as key regions, whereas all other cities in China were classified as non-key regions. Different home quarantine policies were applied to individuals returning from key vs. non-key regions. For instance, people in the key regions need to be quarantined at for a longer period of time than those who were in the non-key regions (14 vs. 7 days).
3The mediating effect value was calculated by the sum of the path coefficient from the independent variable to the mediated variable multiplied by the path coefficient from the mediated variable to the dependent variable. And the proportion of the mediating effect was calculated by the sum of the mediating effect divided by the mediating effect plus the direct effect.