Skip to main content Accessibility help
×
Home
Hostname: page-component-768ffcd9cc-s8fcc Total loading time: 1.006 Render date: 2022-12-04T13:36:35.902Z Has data issue: true Feature Flags: { "useRatesEcommerce": false } hasContentIssue true

Long-term psychological profile of general population following COVID-19 outbreak: symptom trajectories and evolution of psychopathological network

Published online by Cambridge University Press:  27 September 2022

Zheng-An Lu
Affiliation:
Peking University Sixth Hospital, Peking University Institute of Mental Health, NHC Key Laboratory of Mental Health (Peking University), National Clinical Research Center for Mental Disorders (Peking University Sixth Hospital), Chinese Academy of Medical Sciences Research Unit (No. 2018RU006), Peking University, Beijing, China
Le Shi
Affiliation:
Peking University Sixth Hospital, Peking University Institute of Mental Health, NHC Key Laboratory of Mental Health (Peking University), National Clinical Research Center for Mental Disorders (Peking University Sixth Hospital), Chinese Academy of Medical Sciences Research Unit (No. 2018RU006), Peking University, Beijing, China
Jian-Yu Que
Affiliation:
Peking University Sixth Hospital, Peking University Institute of Mental Health, NHC Key Laboratory of Mental Health (Peking University), National Clinical Research Center for Mental Disorders (Peking University Sixth Hospital), Chinese Academy of Medical Sciences Research Unit (No. 2018RU006), Peking University, Beijing, China
Yong-Bo Zheng
Affiliation:
Peking University Sixth Hospital, Peking University Institute of Mental Health, NHC Key Laboratory of Mental Health (Peking University), National Clinical Research Center for Mental Disorders (Peking University Sixth Hospital), Chinese Academy of Medical Sciences Research Unit (No. 2018RU006), Peking University, Beijing, China Peking-Tsinghua Center for Life Sciences and PKU-IDG/McGovern Institute for Brain Research, Beijing, China
Qian-Wen Wang
Affiliation:
Peking University Sixth Hospital, Peking University Institute of Mental Health, NHC Key Laboratory of Mental Health (Peking University), National Clinical Research Center for Mental Disorders (Peking University Sixth Hospital), Chinese Academy of Medical Sciences Research Unit (No. 2018RU006), Peking University, Beijing, China
Wei-Jian Liu
Affiliation:
Peking University Sixth Hospital, Peking University Institute of Mental Health, NHC Key Laboratory of Mental Health (Peking University), National Clinical Research Center for Mental Disorders (Peking University Sixth Hospital), Chinese Academy of Medical Sciences Research Unit (No. 2018RU006), Peking University, Beijing, China
Yue-Tong Huang
Affiliation:
Peking University Sixth Hospital, Peking University Institute of Mental Health, NHC Key Laboratory of Mental Health (Peking University), National Clinical Research Center for Mental Disorders (Peking University Sixth Hospital), Chinese Academy of Medical Sciences Research Unit (No. 2018RU006), Peking University, Beijing, China
Jie Shi
Affiliation:
National Institute on Drug Dependence and Beijing Key Laboratory of Drug Dependence, Peking University, Beijing, China
Yan-Ping Bao*
Affiliation:
National Institute on Drug Dependence and Beijing Key Laboratory of Drug Dependence, Peking University, Beijing, China
Lin Lu*
Affiliation:
Peking University Sixth Hospital, Peking University Institute of Mental Health, NHC Key Laboratory of Mental Health (Peking University), National Clinical Research Center for Mental Disorders (Peking University Sixth Hospital), Chinese Academy of Medical Sciences Research Unit (No. 2018RU006), Peking University, Beijing, China Peking-Tsinghua Center for Life Sciences and PKU-IDG/McGovern Institute for Brain Research, Beijing, China National Institute on Drug Dependence and Beijing Key Laboratory of Drug Dependence, Peking University, Beijing, China
*
Authors for correspondence: Yan-Ping Bao, E-mail: baoyp@bjmu.edu.cn; Lin Lu, E-mail: linlu@bjmu.edu.cn
Authors for correspondence: Yan-Ping Bao, E-mail: baoyp@bjmu.edu.cn; Lin Lu, E-mail: linlu@bjmu.edu.cn
Rights & Permissions[Opens in a new window]

Abstract

Aims

COVID-19 has long-term impacts on public mental health, while few research studies incorporate multidimensional methods to thoroughly characterise the psychological profile of general population and little detailed guidance exists for mental health management during the pandemic. This research aims to capture long-term psychological profile of general population following COVID-19 by integrating trajectory modelling approaches, latent trajectory pattern identification and network analyses.

Methods

Longitudinal data were collected from a nationwide sample of 18 804 adults in 12 months after COVID-19 outbreak in China. Patient Health Questionnaire-9, Generalised Anxiety Disorder-7 and Insomnia Severity Index were used to measure depression, anxiety and insomnia, respectively. The unconditional and conditional latent growth curve models were fitted to investigate trajectories and long-term predictors for psychological symptoms. We employed latent growth mixture model to identify the major psychological symptom trajectory patterns, and ran sparse Gaussian graphical models with graphical lasso to explore the evolution of psychopathological network.

Results

At 12 months after COVID-19 outbreak, psychological symptoms generally alleviated, and five psychological symptom trajectories with different demographics were identified: normal stable (63.4%), mild stable (15.3%), mild-increase to decrease (11.7%), mild-decrease to increase (4.0%) and moderate/severe stable (5.5%). The finding indicated that there were still about 5% individuals showing consistently severe distress and approximately 16% following fluctuating psychological trajectories, who should be continuously monitored. For individuals with persistently severe trajectories and those with fluctuating trajectories, central or bridge symptoms in the network were mainly ‘motor abnormality’ and ‘sad mood’, respectively. Compared with initial peak and late COVID-19 phase, aftermath of initial peak might be a psychologically vulnerable period with highest network connectivity. The central and bridge symptoms for aftermath of initial peak (‘appetite change’ and ‘trouble of relaxing’) were totally different from those at other pandemic phases (‘sad mood’).

Conclusions

This research identified the overall growing trend, long-term predictors, trajectory classes and evolutionary pattern of psychopathological network of psychological symptoms in 12 months after COVID-19 outbreak. It provides a multidimensional long-term psychological profile of the general population after COVID-19 outbreak, and accentuates the essentiality of continuous psychological monitoring, as well as population- and time-specific psychological management after COVID-19. We believe our findings can offer reference for long-term psychological management after pandemics.

Type
Original Article
Creative Commons
Creative Common License - CCCreative Common License - BY
This is an Open Access article, distributed under the terms of the Creative Commons Attribution licence (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted re-use, distribution and reproduction, provided the original article is properly cited.
Copyright
Copyright © The Author(s), 2022. Published by Cambridge University Press

Introduction

COVID-19, which caused over 500 million infections and 6 million deaths globally, has put the whole society under tremendous mental health strain (World Health Organization, WHO Coronavirus Dashboard, accessed on 5 August 2022). The continuous emergence of sporadic cases, frequent local resurgences, long-term pandemic control measures and socio-economic repercussions are all psychological stressors posing enduring threats to public mental health (Brooks et al., Reference Brooks, Webster, Smith, Woodland, Wessely, Greenberg and Rubin2020; Hiremath et al., Reference Hiremath, Suhas Kowshik, Manjunath and Shettar2020; Zhou et al., Reference Zhou, Snoswell, Harding, Bambling, Edirippulige, Bai and Smith2020; Wang et al., Reference Wang, Shi, Que, Lu, Liu, Lu, Xu, Liu, Sun, Meng, Yuan, Ran, Lu, Bao and Shi2021). Preliminary research conducted during COVID-19 outbreak indicated potential mental impacts of COVID-19, complying with findings from previous pandemics (Maunder et al., Reference Maunder, Lancee, Balderson, Bennett, Borgundvaag, Evans, Fernandes, Goldbloom, Gupta, Hunter, McGillis Hall, Nagle, Pain, Peczeniuk, Raymond, Read, Rourke, Steinberg, Stewart, VanDeVelde-Coke, Veldhorst and Wasylenki2006; Lee et al., Reference Lee, Wong, McAlonan, Cheung, Cheung, Sham, Chu, Wong, Tsang and Chua2007; Holmes et al., Reference Holmes, O'Connor, Perry, Tracey, Wessely, Arseneault, Ballard, Christensen, Cohen Silver, Everall, Ford, John, Kabir, King, Madan, Michie, Przybylski, Shafran, Sweeney, Worthman, Yardley, Cowan, Cope, Hotopf and Bullmore2020; Shi et al., Reference Shi, Lu, Que, Huang, Liu, Ran, Gong, Yuan, Yan, Sun, Shi, Bao and Lu2020, Reference Shi, Lu, Que, Huang, Lu, Liu, Zheng, Liu, Ran, Yuan, Yan, Sun, Sun, Shi, Kosten, Bao and Lu2021a, Reference Shi, Que, Lu, Gong, Liu, Wang, Ran, Ravindran, Ravindran, Fazel, Bao, Shi and Lu2021b; Zheng et al., Reference Zheng, Shi, Lu, Que, Yuan, Huang, Liu, Wang, Lu, Wang, Yan, Han, Sun, Bao, Shi and Lu2021). However, since psychological stressors may exert distinct influences on people across different pandemic phases, longitudinal studies are required to capture complete psychological trajectories after COVID-19 so as to provide reference for long-term psychological management (Fancourt et al., Reference Fancourt, Steptoe and Bu2021).

Existing longitudinal studies based on large samples have unveiled different psychological symptom growth curves and trajectory patterns over the course of COVID-19, most of which indicated increase in distress at initial peak and symptom plateauing or alleviation during lockdown (Iob et al., Reference Iob, Frank, Steptoe and Fancourt2020; Planchuelo-Gómez et al., Reference Planchuelo-Gómez, Odriozola-González, Irurtia and de Luis-García2020; Salfi et al., Reference Salfi, Lauriola, Amicucci, Corigliano, Viselli, Tempesta and Ferrara2020; Fancourt et al., Reference Fancourt, Steptoe and Bu2021; Prati and Mancini, Reference Prati and Mancini2021; Ripoll et al., Reference Ripoll, Contreras-Martos, Esteva, Soler and Serrano-Ripoll2021; Saunders et al., Reference Saunders, Buckman, Leibowitz, Cape and Pilling2021a, Reference Saunders, Buckman, Fonagy and Fancourt2021b). However, all these studies employed traditional single-dimensional approaches that could not offer detailed information for psychological management at different pandemic phases. Another important question to answer is whether some populations suffer more adverse psychological impacts from COVID-19. Existing studies have identified some demographic predictors (e.g. young age, females, low socioeconomic status, poor physical or mental health status) and epidemic-related predictors (e.g. being infected, working as essential workers, quarantine, trauma history, exposure risk) for long-term psychological symptoms after pandemics (Maunder et al., Reference Maunder, Lancee, Balderson, Bennett, Borgundvaag, Evans, Fernandes, Goldbloom, Gupta, Hunter, McGillis Hall, Nagle, Pain, Peczeniuk, Raymond, Read, Rourke, Steinberg, Stewart, VanDeVelde-Coke, Veldhorst and Wasylenki2006; Lee et al., Reference Lee, Wong, McAlonan, Cheung, Cheung, Sham, Chu, Wong, Tsang and Chua2007; Liu et al., Reference Liu, Kakade, Fuller, Fan, Fang, Kong, Guan and Wu2012; Iob et al., Reference Iob, Frank, Steptoe and Fancourt2020; Saunders et al., Reference Saunders, Buckman, Fonagy and Fancourt2021b; Shi et al., Reference Shi, Lu, Que, Huang, Lu, Liu, Zheng, Liu, Ran, Yuan, Yan, Sun, Sun, Shi, Kosten, Bao and Lu2021a). However, these studies failed to quantify effect sizes of predictors at different pandemic phases, which limited their potentials in guiding time-specific interventions. Studies based on large representative samples, with longer observational periods, and incorporating multidimensional psychological modelling approaches are in urgent demand to offer reference for population- and time-specific mental health management.

Person-centred trajectory modelling approaches (i.e. latent growth curve models, LGCMs) that consider individual differences when estimating trajectories can reveal a more reliable developmental trend for psychological symptoms (Felt et al., Reference Felt, Depaoli and Tiemensma2017). Conditional LGCMs may quantify not only the effects of predictors on symptom baseline level, but also on symptom changing slopes (Lu et al., Reference Lu, Shi, Que, Zheng, Wang, Liu, Huang, Liu, Yuan, Yan, Shi, Bao and Lu2022). Further, some advanced models developed in recent years may offer in-depth information by incorporating novel perspectives of latent pattern identification and graph theory. Latent growth mixture model (LGMM), a person-centred clustering method, can extract the major psychological symptom trajectory patterns after COVID-19 from a heterogeneous population (Hannigan et al., Reference Hannigan, Pingault, Krapohl, McAdams, Rijsdijk and Eley2018). Characterising the demographics of these trajectory patterns can provide information for population-specific mental health management. In addition, the ‘psychopathological network’ analysis, which assumes mental disorders as networks based on causal interactions among individual symptoms, can provide insights into psychopathological structures and potential treatment target symptoms across different pandemic phases and populations in a trans-diagnostic perspective (Bringmann et al., Reference Bringmann, Albers, Bockting, Borsboom, Ceulemans, Cramer, Epskamp, Eronen, Hamaker, Kuppens, Lutz, McNally, Molenaar, Tio, Voelkle and Wichers2022). Specifically, the global network connectivity can reflect overall susceptibility to mental disorders of a population, and previous studies reveal that patients with densely connected network have worse prognosis (Borsboom and Cramer, Reference Borsboom and Cramer2013; O'Driscoll et al., Reference O'Driscoll, Buckman, Fried, Saunders, Cohen, Ambler, DeRubeis, Gilbody, Hollon, Kendrick, Kessler, Lewis, Watkins, Wiles and Pilling2021). The central symptom and the bridge symptom can be highly prospective treatment targets for emergence and comorbidity of mental disorders (Opsahl et al., Reference Opsahl, Agneessens and Skvoretz2010; Borsboom and Cramer, Reference Borsboom and Cramer2013; Hallquist et al., Reference Hallquist, Wright and Molenaar2021). Adopting an integrative approach by combining these multidimensional methods can greatly enrich our understanding of long-term psychopathology of the general population following pandemics so as to provide reference for psychological management.

The aim of the current study is to capture the multidimensional long-term psychological profile of the general population after COVID-19 outbreak by integrating symptom trajectory modelling, latent pattern identification and psychopathological network analyses based on repeatedly collected data from a nationwide sample of 18 804 adults in China, to provide population- and time-specific reference for mental health management after pandemics.

Methods

Procedures and participants

We conducted a longitudinal observational study via Joybuy, a large digital commercial clustering website with 0.44 billion active users from all 34 provincial regions in China, as detailed elsewhere (Shi et al., Reference Shi, Lu, Que, Huang, Lu, Liu, Zheng, Liu, Ran, Yuan, Yan, Sun, Sun, Shi, Kosten, Bao and Lu2021a; Wang et al., Reference Wang, Shi, Que, Lu, Liu, Lu, Xu, Liu, Sun, Meng, Yuan, Ran, Lu, Bao and Shi2021).

We fielded three surveys after COVID-19 outbreak. Survey 1 was fielded during initial COVID-19 peak and lockdown period (28 February 2020 to 11 March 2020). Survey 2 was fielded in the aftermath of initial COVID-19 peak (8 July 2020 to 8 August 2020), when main wave had been controlled and lockdown had ended, but sporadic cases and local resurgences were common, and pandemic control measures were changing frequently. Survey 3 was fielded during late COVID-19 phase (29 January 2021 to 26 April 2021), when daily new confirmed cases were few and most pandemic control measures were removed. During survey 1, we adopted an untargeted approach to recruit participants, by posting a survey link on Joybuy and allowing participants to voluntarily click on it until the sample represented all 34 provincial regions in China, as detailed elsewhere (Shi et al., Reference Shi, Lu, Que, Huang, Liu, Ran, Gong, Yuan, Yan, Sun, Shi, Bao and Lu2020). During surveys 2 and 3, we adopted both targeted and untargeted approaches to recruit participants. In the targeted approach, we sent survey links to all participants who had completed at least one previous survey via the message platform of Joybuy. In the untargeted approach, we followed the same recruiting procedure as survey 1.

All participants were registered members of Joybuy. During survey 1, 56 932 of 71 227 participants who clicked on survey links provided informed consent and completed the survey, with an effective response rate of 79.9%. After quality control, 56 679 adults with valid age information comprised the final survey 1 sample, as detailed elsewhere (Shi et al., Reference Shi, Lu, Que, Huang, Liu, Ran, Gong, Yuan, Yan, Sun, Shi, Bao and Lu2020). After excluding all participants under 18 or with invalid age information, the final survey 2 and 3 samples comprised of 27 961 and 34 041 adults, respectively. The final longitudinal sample comprised of 18 804 adults with valid data from at least two of the three surveys. Of the 18 804 participants, data were available for 16 508 in survey 1, 12 788 in survey 2 and 13 175 in survey 3. Detailed information is presented in Fig. 1 and in previous research (Lu et al., Reference Lu, Shi, Que, Zheng, Wang, Liu, Huang, Liu, Yuan, Yan, Shi, Bao and Lu2022).

Fig. 1. Flow graph for participants recruitment at three surveys.

Measures

We designed questionnaires to collect information on demographics, epidemic-related information, quarantine conditions, as well as to measure symptoms of depression, anxiety and insomnia in three surveys. The detailed contents of questionnaires were provided in previous research (Shi et al., Reference Shi, Lu, Que, Huang, Liu, Ran, Gong, Yuan, Yan, Sun, Shi, Bao and Lu2020, Reference Shi, Lu, Que, Huang, Lu, Liu, Zheng, Liu, Ran, Yuan, Yan, Sun, Sun, Shi, Kosten, Bao and Lu2021a). Chinese versions of Patient Health Questionnaire-9 (PHQ-9), Generalised Anxiety Disorder-7 (GAD-7) and Insomnia Severity Index (ISI) were used to measure depression, anxiety and insomnia, respectively (Kroenke et al., Reference Kroenke, Spitzer and Williams2001; Löwe et al., Reference Löwe, Decker, Müller, Brähler, Schellberg, Herzog and Herzberg2008; Gagnon et al., Reference Gagnon, Bélanger, Ivers and Morin2013).

Variables

To explore long-term predictors for depression, anxiety and insomnia, we considered the following three categories of variables: (1) demographic variables: gender, age, living area, educational level, marital status, monthly family income, history of chronic diseases, history of psychiatric disorders, family history of psychiatric disorders, history of sleep disturbance, history of smoking and history of alcohol abuse; (2) variables related to COVID-19 infection: COVID-19 patients, family members of COVID-19 patients, close contacts of COVID-19 patients, engagement in COVID-19-related work, family members of workers directly engaging in COVID-19 control and occupational exposure risk to COVID-19; (3) variables related to post-pandemic repercussions: quarantine, living in places severely affected by initial peak, local resurgences, increases in workloads, unemployment due to COVID-19, seeking psychological consultation, wearing face masks and reducing social gatherings. Variables were dichotomised based on participants' responses to questionnaire questions, and variable constructing approaches were described elsewhere (Shi et al., Reference Shi, Lu, Que, Huang, Lu, Liu, Zheng, Liu, Ran, Yuan, Yan, Sun, Sun, Shi, Kosten, Bao and Lu2021a).

Statistical analyses

Descriptive statistics were used to present the demographic characteristics of the longitudinal sample. To explore trajectories, predictors, symptom trajectory patterns and evolution of psychopathological networks of depression, anxiety and insomnia, we conducted the following four analyses.

In the first step, to explore developmental trajectories of depression, anxiety and insomnia, we conducted LGCM analyses. Given the high correlation among three symptoms, we conducted a multi-process LGCM analysis, in which depression, anxiety and insomnia scores were simultaneously entered into one model and paths between symptoms were incorporated, so that the general symptom growing trend could be estimated (Hannigan et al., Reference Hannigan, Pingault, Krapohl, McAdams, Rijsdijk and Eley2018; Fancourt et al., Reference Fancourt, Steptoe and Bu2021). Next, we fitted three independent LGCMs, in which outcome variables were depression, anxiety and insomnia scores respectively, so that trajectories for the three single symptoms could be estimated. All variables considered were adjusted for in each model. For each model, we tested two types of growth factors: intercept-only and linear slope (Hannigan et al., Reference Hannigan, Pingault, Krapohl, McAdams, Rijsdijk and Eley2018). For the models with linear-slope growth factor, we further tested fixed (coded as 0, 5 and 12 in three surveys) and free slope factor loadings. We determined the optimal models based on comparative fit index (CFI), chi-square, standardised root-mean-square residual (SRMR) and the root mean square error of approximation (RMSEA) statistics (Hu and Bentler, Reference Hu and Bentler1999). Smaller values for chi-square, RMSEA, SRMR and larger values for CFI indicate a better fit. Following established recommendations, good fit was considered when CFI ⩾0.95, RMSEA ⩽0.05 and SRMR ⩽0.05 (Hannigan et al., Reference Hannigan, Pingault, Krapohl, McAdams, Rijsdijk and Eley2018; Pavlov et al., Reference Pavlov, Maydeu-Olivares and Shi2021). Mean for intercept reflects the average psychological symptom score at survey 1, and mean for slope reflects the average change in psychological symptom score from survey 1 to survey 3 (Hu and Bentler, Reference Hu and Bentler1999).

In the second step, to identify predictors of depression, anxiety and insomnia trajectories, we fitted three independent conditional linear slope LGCMs, in which outcome variables were depression scores, anxiety scores and insomnia scores, respectively. All variables considered were added into the model as independent variables, so that the growth factors (i.e. intercepts and slopes) could be regressed on these variables. In the conditional LGCM, the mean effect (B value) for intercept reflects the predicting effect of a variable on psychological symptom scores in survey 1, while the mean effect (B value) for slope reflects the predicting effect of a variable on longitudinal changes in scores from survey 1 to survey 3.

In the third step, to explore the psychological symptom trajectory patterns throughout COVID-19, we conducted a LGMM analysis. Considering high correlation among three symptoms, we employed a joint trajectory approach, which was previously used to identify co-developing trajectory patterns for several highly correlated symptoms (Hannigan et al., Reference Hannigan, Pingault, Krapohl, McAdams, Rijsdijk and Eley2018). Scores for depression, anxiety and insomnia were all entered into one single LGMM as outcome variables, so that co-developing patterns for the three symptoms could be unveiled. We gradually added number of latent classes from 2 to 11, and determined the optimal number of latent classes mainly based on Vuong–Lo–Mendell–Rubin likelihood ratio test (VLMR-LRT) (Lo et al., Reference Lo, Mendell and Rubin2001), with due consideration of parsimony, interpretability, Akaike information criterion (AIC) (Akaike, Reference Akaike1981), Bayesian information criterion (BIC) (Schwartz, Reference Schwartz1978), adjusted BIC (aBIC) (Sclove, Reference Sclove1987) and entropy (Lo et al., Reference Lo, Mendell and Rubin2001). VLMR-LRT was considered as an acceptable method to determine optimal number of classes (Lo et al., Reference Lo, Mendell and Rubin2001). A p value lower than 0.05 in VLMR-LRT suggests a better fit of model with k trajectory classes compared with model with k − 1 trajectory classes (Lo et al., Reference Lo, Mendell and Rubin2001). Lower BIC, aBIC and AIC values indicate a better fit (Schwartz, Reference Schwartz1978; Akaike, Reference Akaike1981; Sclove, Reference Sclove1987). Entropy characterises quality of classification on a 0–1 scale (Celeux and Soromenho, Reference Celeux and Soromenho1996). Entropy value closer to 1 indicates clear delineation of classes (Celeux and Soromenho, Reference Celeux and Soromenho1996) and 0.80 is often used as an acceptable level (Greenwood et al., Reference Greenwood, Youssef, Betts, Letcher, McIntosh, Spry, Hutchinson, Macdonald, Hagg, Sanson, Toumbourou and Olsson2019). After determining optimal number of latent trajectory classes, all individuals in the longitudinal sample acquired their class membership based on posterior probability. We conducted multinomial logistic regression of latent trajectory class membership on predictors using the three-step procedure (R3STEP) in Mplus software (Bakk et al., Reference Bakk, Tekle and Vermunt2013). In the three steps above, all missing data were handled with full information maximum likelihood estimation, which supposed that missingness was at random (Cham et al., Reference Cham, Reshetnyak, Rosenfeld and Breitbart2017). The significance level was set to two-sided p < 0.05. All statistical analyses were performed in SPSS 22 software (SPSS, Chicago, IL, USA), Mplus 8.3 and R version 4.0.3.

In the fourth step, to estimate the psychopathological network for depression, anxiety and insomnia in three surveys, we ran three sparse Gaussian graphical models with graphical lasso on eight items from Patient Health Questionnaire-9 (excluding sleep disturbances, overlapping with other scales), seven items from Generalised Anxiety Disorder-7 and three items form Insomnia Severity Index (excluding four non-symptom items) (Friedman et al., Reference Friedman, Hastie and Tibshirani2008). The tuning parameters were determined by extended BIC (van Borkulo et al., Reference van Borkulo, Borsboom, Epskamp, Blanken, Boschloo, Schoevers and Waldorp2014). In the psychopathological network, scores of items were considered as ‘nodes’, and pair-wise correlations between items were considered as ‘edges’ (Borsboom and Cramer, Reference Borsboom and Cramer2013). Global connectivity of network can be characterised by ‘global strength’ defined as sum of edge weights in the whole network. Higher global connectivity indicates greater susceptibility to mental disorders. The network structures can be characterised by network centrality and bridge centrality. Network centrality indices include strength, closeness, betweenness and expected influence, which can reflect the place of each node in the network. Expected influence is defined as sum of edge weights directly connected to a node, and is regarded as optimal indicator for centrality in a network with both positive and negative edges (Opsahl et al., Reference Opsahl, Agneessens and Skvoretz2010). Bridge centrality indices include bridge strength, bridge closeness, bridge betweenness and bridge expected influence, which can reflect the ability of each node in connecting different disorders. Bridge expected influence is defined as sum of edge weights connecting a node to all nodes from other disorders, and is commonly considered as an optimal indicator for the bridge centrality (Opsahl et al., Reference Opsahl, Agneessens and Skvoretz2010). The central symptom, defined as the symptom with greatest centrality, can be a potential treatment target for mental disorders (Opsahl et al., Reference Opsahl, Agneessens and Skvoretz2010; Borsboom and Cramer, Reference Borsboom and Cramer2013). The bridge symptom, defined as the symptom with greatest bridge centrality, can be a highly prospective treatment target for comorbidity of mental disorders. In this research, central symptom and bridge symptom were determined by values of expected influence and bridge expected influence, respectively (Opsahl et al., Reference Opsahl, Agneessens and Skvoretz2010; Borsboom and Cramer, Reference Borsboom and Cramer2013). Networks were also estimated in identified psychological symptom trajectory classes with the same procedures and parameters described above. The R package ‘bootnet’ was utilised to estimate the psychopathological network and evaluate the network stability (Epskamp et al., Reference Epskamp, Borsboom and Fried2018). The R package ‘NetworkComparisonTest’ was used to calculate and compare the global connectivity of networks in three pandemic phases. The R packages ‘qgraph’ and ‘networktools’ were utilised to calculate the network centrality indices and bridge centrality indices, respectively. The network analyses were performed with R version 4.0.3.

Results

Demographic characteristics of the longitudinal sample

The 18 804 participants had a mean (s.d.) age of 36.6 (8.2) years, among whom 8558 (45.4%) were males, 17 599 (93.6%) lived in urban areas, 15 489 (82.4%) had a college school or higher educational level, 14 783 (78.6%) were married and 4186 (22.3%) had a monthly family income lower than 5000 yuan, as is presented in online Supplementary Table S1.

Trajectories and predictors of psychological symptoms after COVID-19 outbreak

Modelling fitting statistics for LGCMs with different growth factors are presented in online Supplementary Table S2. The multi-process LGCM revealed that psychological symptoms generally decreased [depression, estimated mean (s.e.) for slope −0.09 (0.02), p < 0.001; anxiety, −0.39 (0.02), p < 0.001; insomnia, −0.11 (0.02), p < 0.001; Fig. 2a]. Specifically, the best-fitting model for depression indicated a mild worsening trend, in which the mean (s.e.) for slope was 0.43 (0.05, p < 0.001). However, the best-fitting models for anxiety and insomnia symptoms suggested slight alleviating trends. Mean (s.e.) for slope was −0.44 (0.02, p < 0.001) for anxiety and −0.49 (0.03, p < 0.001) for insomnia (Fig. 2b).

Fig. 2. Predicted trajectories of depression, anxiety and insomnia from the best fitting: (a) multi-process LGCM and (b) LGCMs for three single symptoms.

In addition, we identified demographic factors, infection-related factors and post-COVID-19 repercussions as long-term predictors for intercepts or slopes of psychological trajectories after COVID-19. Significant predictors are illustrated in online Supplementary Fig. S1 and Table 1.

Table 1. Predictors for intercepts and slopes of depression, anxiety and insomnia scores from the conditional LGCMs

Psychological symptom trajectory patterns among the general population after COVID-19 outbreak

Model fitting indicators for LGMMs with 2 to 11 latent classes are presented in online Supplementary Table S3. We selected 5-class LGMM as the optimal model mainly based on VLMR-LRT, with due consideration of parsimony, interpretability, entropy, AIC, BIC and aBIC.

We illustrated the five symptom trajectory classes from the optimal 5-class model in Fig. 3. The first class showed persistent moderate-to-severe symptoms throughout COVID-19 and accounted for 5.5% of the total sample, which was labelled as ‘moderate/severe stable’ class. In addition, there were three latent classes that demonstrated mild symptom level during initial peak, but followed distinct trajectories afterwards: the first class followed relatively stable trajectory and accounted for 15.3% of the total sample, which was labelled as ‘mild stable’ class; the second class experienced a dramatic deterioration in all three symptoms in the aftermath of initial COVID-19 peak, but came back to normal in the late COVID-19 phase, which was labelled as ‘mild-increase to decrease’ class and accounted for 11.7% of the total sample; the third class recovered to the normal level in the aftermath of initial peak, but experienced a substantial surge afterwards and ended with moderate-to-severe symptoms during late COVID-19 phase, which was labelled as ‘mild-decrease to increase’ class and accounted for 4.0% of the total sample. The last class had normal and stable psychological status throughout COVID-19 and accounted for the largest proportion (63.4%), which was labelled as ‘normal stable’ class (Fig. 3; online Supplementary Table S4).

Fig. 3. Predicted trajectories of depression, anxiety and insomnia across latent symptom trajectory classes from the best fitting 5-class LGMM. (a) Latent class 1: moderate/severe stable (n = 1044, 5.5%); (b) latent class 2: mild stable (n = 2885, 15.3%); (c) latent class 3: mild-increase to decrease (n = 2201, 11.7%); (d) latent class 4: mild-decrease to increase (n = 755, 4.0%) and (e) latent class 5: normal stable (n = 11 919, 63.4%).

Table 2 and online Supplementary Table S5 demonstrated the demographic characteristics of the five latent symptom trajectory classes. ‘Moderate/severe stable’ class featured highest proportion of males, youngsters, the poorly educated, the unmarried, the impoverished, individuals with history of chronic diseases or psychiatric disorders, as well as individuals severely affected by COVID-19 (i.e. quarantine, living in places severely affected by COVID-19 and COVID-19-related stressful life events). ‘Normal stable’ class distinguished itself from other four classes with its highest proportion of females, the mid-aged or elders and individuals with higher incomes. Furthermore, the two mild classes with fluctuating trajectories were more likely to report history of chronic disease or psychiatric disorders, quarantine experiences and poor education compared with ‘mild stable’ class.

Table 2. Multinomial logistic regression of psychological symptom trajectory class membership on predictors using a three-step approacha

a Odds ratios were derived from multinomial logistic regression analysis using a three-step approach in Mplus software. All the covariates presented in the table were entered into LGMM as auxiliary variables.

b Places severely affected by COVID-19 included places severely affected by initial peak and places with COVID-19 resurgences.

c COVID-19-related stressful life events included being COVID-19 patients, their family members or close contacts and being workers directly engaged in COVID-19 control or their family members.

Evolution of psychopathological network among the general population after COVID-19 outbreak

The global connectivity of psychopathological network significantly increased from initial COVID-19 peak to aftermath of initial peak (global strength at initial peak: 8.56; global strength in the aftermath of initial peak: 8.70; p < 0.001), then mildly decreased from aftermath of initial peak to late COVID-19 phase (global strength in the aftermath of initial peak: 8.70; global strength in the late COVID-19 phase: 8.61; p = 0.08). The global connectivity of psychological network at initial peak and late COVID-19 phase did not differ significantly (p = 0.19; Fig. 4; online Supplementary Fig. S2).

Fig. 4. Evolution of psychopathological networks for depression, anxiety and insomnia after COVID-19. Psychopathological networks at (a) initial peak, (b) aftermath of initial peak and (c) late COVID-19 phase were estimated by sparse Gaussian graphical models with graphical lasso based on 18 items from PHQ-9, GAD-7 and ISI. Blue edges indicate positive correlations. Red edges indicate negative correlations. Thicker edges indicate stronger regularised partial correlations. Green nodes indicate items from PHQ-9 (depression). Orange nodes indicate items from GAD-7 (anxiety). Blue nodes indicate items from ISI (insomnia). Central symptom and bridge symptom were determined based on values of expected influence and bridge expected influence, respectively. (a) Initial peak: global strength: 8.56; central symptom: D2-Sad mood; bridge symptom: D2-Sad mood; (b) aftermath of initial peak: global strength: 8.70; central symptom: D4-Appetite change; bridge symptom: A4-Trouble of relaxing and (c) late COVID-19 phase: global strength: 8.61; central symptom: D2-Sad mood; bridge symptom: D2-Sad mood.

In the overall population, central and bridge symptoms were both ‘D2-Sad mood’ for initial peak and late COVID-19 phase (online Supplementary Figs S3 and S4). However, for the aftermath of initial peak, central symptom was ‘D4-Appetite change’ (online Supplementary Fig. S3), and bridge symptom was ‘A4-Trouble of relaxing’ (online Supplementary Fig. S4).

For individuals in ‘moderate/severe stable’ class, central symptoms for initial peak, aftermath of initial peak and late COVID-19 phase were ‘D7-Motor abnormality’, ‘A4-Trouble of relaxing’ and ‘D7-Motor abnormality’, respectively, while bridge symptoms for the three stages were ‘D7-Motor abnormality’, ‘D4-Appetite change’ and ‘A7-Afraid’, respectively (online Supplementary Figs S5–S7). For individuals following fluctuating psychological trajectories (‘mild-increase to decrease’ class and ‘mild-decrease to increase’ class), central or bridge symptoms were mainly ‘D2-Sad mood’ at all three pandemic phases (online Supplementary Figs S8–S10). For individuals in ‘mild stable’ class, central or bridge symptoms were mainly ‘D8-Suicide and self-harm’ and ‘D2-Sad mood’ (online Supplementary Figs S11–S13). For individuals in ‘normal stable’ class, central or bridge symptoms were mainly ‘A5-Restlessness’ and ‘D2-Sad mood’ (online Supplementary Figs S14–S16).

Discussion

In this research, we found an overall ameliorating trend in mental health symptoms after COVID-19 outbreak. Five psychological symptom trajectory classes with different demographic and psychopathological characteristics were identified: normal stable (63.4%), mild stable (15.3%), mild-increase to decrease (11.7%), mild-decrease to increase (4.0%) and moderate/severe stable (5.5%). General population showed distinct psychopathological networks at different pandemic phases. Aftermath of initial peak may be a psychologically vulnerable period with specific psychopathological structures. To the best of our knowledge, this research is the first to provide a comprehensive long-term psychological profile of the general population after COVID-19 from a multidimensional perspective. We believe our findings are valuable for long-term population- and time-specific mental health management following COVID-19 and future pandemics.

The overall improving trend of mental health status after COVID-19 complies with findings from other countries, including the United Kingdom, the United States, Germany, Spain and India (Daly et al., Reference Daly, Sutin and Robinson2020; Gopal et al., Reference Gopal, Sharma and Subramanyam2020; Bendau et al., Reference Bendau, Kunas, Wyka, Petzold, Plag, Asselmann and Ströhle2021; Daly and Robinson, Reference Daly and Robinson2021). Improvement in pandemic control, lift of lockdown, removal of imperative measures and economic recovery may all contribute to the relief of mental strain (Daly et al., Reference Daly, Sutin and Robinson2020; Gopal et al., Reference Gopal, Sharma and Subramanyam2020; Bendau et al., Reference Bendau, Kunas, Wyka, Petzold, Plag, Asselmann and Ströhle2021; Daly and Robinson, Reference Daly and Robinson2021).

Despite the overall ameliorating trend in psychological symptoms, we found that there were still about 5% individuals showing persistently moderate-to-severe distress and approximately 16% following fluctuating psychological trajectories. The psychological symptom trajectory patterns identified in our research comply with findings from other countries, including the United Kingdom and Australia (Batterham et al., Reference Batterham, Calear, McCallum, Morse, Banfield, Farrer, Gulliver, Cherbuin, Rodney Harris, Shou and Dawel2021; Pierce et al., Reference Pierce, McManus, Hope, Hotopf, Ford, Hatch, John, Kontopantelis, Webb, Wessely and Abel2021). We all found that although the majority of individuals showed consistently good mental health status, there were a small fraction of individuals showing persistently severe symptoms, and a considerable proportion of individuals beginning with mild symptoms, but following diverse trajectories afterwards. These findings suggest that apart from individuals showing severe distress, special attention should also be paid to those showing mild symptoms at initial peak, since their symptoms may fluctuate as the pandemic evolves.

In this research, some demographic factors, infection-related factors and post-COVID-19 repercussions were found to predict persistently severe psychological symptoms after COVID-19 outbreak. These predictors largely overlapped with those identified in other relevant studies (Daly et al., Reference Daly, Sutin and Robinson2020; Fancourt et al., Reference Fancourt, Steptoe and Bu2021). Young adults' long-term distress might be attributable to their higher exposure to social media and misinformation, more significant increases in work burden as well as greater insecurity in jobs and finance (Ganson et al., Reference Ganson, Tsai, Weiser, Benabou and Nagata2021). Individuals with low incomes or unmarried status might be less capable of coping with the financial adversities due to COVID-19 (Fancourt et al., Reference Fancourt, Steptoe and Bu2021). COVID-19 patients, their family members and high-risk workers might experience substantial fears, witness traumatic events and bear overwhelming workloads, which could trigger enduring stress responses, emotional exhaustion and burn-outs (Lee et al., Reference Lee, Wong, McAlonan, Cheung, Cheung, Sham, Chu, Wong, Tsang and Chua2007; Iob et al., Reference Iob, Frank, Steptoe and Fancourt2020; Mazza et al., Reference Mazza, De Lorenzo, Conte, Poletti, Vai, Bollettini, Melloni, Furlan, Ciceri, Rovere-Querini and Benedetti2020; Saghafi et al., Reference Saghafi, West, Lopez and Cleary2021). The negative mental health impacts of local resurgences might be partially explained by the ‘double blows’ effects (Spittlehouse et al., Reference Spittlehouse, Joyce, Vierck, Schluter and Pearson2014; Tang et al., Reference Tang, Zhao, Lu, Yan, Wang, Zhang and Xu2017). Overwhelming workloads and unemployment, as common psychological stressors, might have magnified psychological effects in the background of pandemics due to the large-scale economic recession and life disturbances (McKee-Ryan et al., Reference McKee-Ryan, Song, Wanberg and Kinicki2005; Maslach and Leiter, Reference Maslach and Leiter2016; Shaw et al., Reference Shaw, Main, Findley, Collie, Kristman and Gross2020). Moreover, we identified that motor abnormality might be a potential treatment target symptom among individuals with persistently severe symptoms, complying with other research conducted during COVID-19 and indicating psychological benefits of physical exercises during lockdown (Wang et al., Reference Wang, Hu, Feng, Wilson and Chen2020).

Furthermore, we found that individuals following fluctuating psychological trajectories were more likely to report lower educational level and history of chronic diseases or psychiatric disorders. Low educational level and poor health status may impair psychological resilience, contributing to greater vulnerability to psychological stressors and instability of mental health status (Davydov et al., Reference Davydov, Stewart, Ritchie and Chaudieu2010; Wu et al., Reference Wu, Liu, Li and Li2016; Xiao et al., Reference Xiao, Wang, Chang, Chen, Yu and Risch2019). Additionally, we identified ‘sad mood’ as the potential treatment target symptom among these individuals. We believe these findings can provide valuable information for psychological management during pandemics.

It is noteworthy that our study population showed highest network connectivity in the aftermath of initial peak among the three pandemic phases, indicating greatest susceptibility to mental disorders in the special phase. The findings suggest that aftermath of initial peak may be a psychologically vulnerable period, which is consistent with previous studies identifying increase in psychological symptoms at the end of lockdown (Saunders et al., Reference Saunders, Buckman, Leibowitz, Cape and Pilling2021a, Reference Saunders, Buckman, Fonagy and Fancourt2021b). One possible explanation is that during this period, most pandemic restrictions were removed and life largely came back to normal (e.g. large-scale return to work, economic recovery, increases in workloads), while the pandemic control situation was still unstable with emergence of sporadic cases and local resurgences (Shi et al., Reference Shi, Lu, Que, Huang, Lu, Liu, Zheng, Liu, Ran, Yuan, Yan, Sun, Sun, Shi, Kosten, Bao and Lu2021a). The instability might arouse senses of uncertainty and insecurity, which in turn enhanced psychological vulnerability (Mohamed et al., Reference Mohamed, Ismail, Shuib and Ali2021).

Additionally, central symptom (‘appetite change’) and bridge symptom (‘trouble of relaxing’) for aftermath of initial peak were totally different from the potential target symptoms (‘sad mood’) at the other two pandemic phases. The findings indicated that in the aftermath of initial peak, general population not only showed greater psychological vulnerability, but also had special psychopathological structures warranting different intervention strategies. The greater centrality of anxiety and somatic symptoms during this phase might be attributable to senses of uncertainty caused by social instability (Mohamed et al., Reference Mohamed, Ismail, Shuib and Ali2021). Our findings indicate that special attention should be paid to this specific period. Intermittent emergence of sporadic cases, local resurgences and frequently changing pandemic control measures can lead to social instability witnessed by aftermath of initial peak, thus causing persistence of distress (Shi et al., Reference Shi, Lu, Que, Huang, Lu, Liu, Zheng, Liu, Ran, Yuan, Yan, Sun, Sun, Shi, Kosten, Bao and Lu2021a). We believe the findings can provide implications for psychological management in future COVID-19 waves.

Strengths and limitations

This research is the first to provide a comprehensive long-term psychological profile of general population following COVID-19 outbreak from a multidimensional perspective. The major strengths of this research included large and nationwide sample, as well as a relatively long observational period. Moreover, we provided in-depth information for population- and time-specific mental health management by incorporating multidimensional psychological modelling approaches.

However, this research has several limitations. First, due to online recruitment, our sample had a bias towards youngsters and highly educated people, and the follow-up rate was relatively low. The report rate for history of mental disorders was also low compared with other studies (Fancourt et al., Reference Fancourt, Steptoe and Bu2021). It might be attributable to stigma towards mental illnesses and low recognition rate of mental health symptoms in China (Wang et al., Reference Wang, Murray, Toussaint, Chen, Guo, He, Luo, Yu, Liu, Huang, Dong and Zhang2019; Zhang et al., Reference Zhang, Sun, Jatchavala, Koh, Chia, Bose, Li, Tan, Wang, Chu, Wang, Tran and Ho2019; Fancourt et al., Reference Fancourt, Steptoe and Bu2021), and bring bias to our results. Generalisation of our results to other populations should be made with caution. However, since the demographics did not differ much between our longitudinal sample and the baseline full sample, we believe the bias brought by drop-outs can be partially avoided (online Supplementary Table S6). Second, the observational duration might not be long enough to capture the complete trajectories. Third, mental health symptoms were evaluated with self-reported standardised questionnaires instead of clinical diagnoses. Fourth, we only considered three mental symptoms when exploring psychological symptom trajectory patterns and psychopathological networks, and the results have not been validated in other samples. Fifth, we only conducted three surveys after COVID-19 outbreak to capture psychological symptom development. Although the time points chosen were highly representative of different pandemic stages, limited number of time points might still preclude detailed depiction of trajectories. Finally, network analysis is rooted in the theory of ‘psychopathological network’ assuming mental disorders as networks based on causal interactions among individual symptoms. Since the theory lacks strong supporting evidence, our findings should be interpreted with caution, especially when guiding clinical practice (Bringmann et al., Reference Bringmann, Albers, Bockting, Borsboom, Ceulemans, Cramer, Epskamp, Eronen, Hamaker, Kuppens, Lutz, McNally, Molenaar, Tio, Voelkle and Wichers2022). Therefore, future relevant research studies based on larger and more representative samples, adopting more objective measurements, involving multiple psychological symptoms and time points and conducted in other countries are called for to further validate our findings.

Conclusions

This research provides a comprehensive and multidimensional long-term psychological profile of general population after COVID-19 outbreak. We drew three main conclusions: (1) mental health status generally improves among general population in 12 months after COVID-19; (2) there are still about 5% individuals showing persistently severe distress and approximately 16% following fluctuating psychological trajectories who demonstrate distinctive demographic characteristics and psychopathological structures. For individuals with persistently severe distress, the potential target symptom is ‘motor abnormality’, while for individuals following fluctuating trajectories, the potential target symptom is ‘sad mood’; (3) aftermath of initial COVID-19 peak may be a psychologically vulnerable period deserving special attention. ‘Sad mood’ emerges as a potential treatment target for initial peak and late COVID-19 phase. However, ‘appetite change’ and ‘trouble of relaxing’ emerge as potential treatment targets for aftermath of initial peak. We believe our findings can offer population- and time-specific reference for long-term mental health management after pandemics.

Supplementary material

The supplementary material for this article can be found at https://doi.org/10.1017/S2045796022000518.

Data

The corresponding authors have full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analyses.

Acknowledgements

We thank all the participants who took part in this study.

Author contributions

Z. L., L. S. and L. L. proposed the topic and main idea; Z. L. analysed the data; Z. L. wrote the initial draft of the manuscript; L. S., Y. Z., J. Q., W. L., Q. W., Y. H., J. S., Y. B. and L. L. commented on and revised the manuscript; L. L. finalised the manuscript, with contributions from all authors. All authors have read and agreed to the published version of the manuscript.

Financial support

This study was supported by grants from the National Key Research and Development Program of China (2021YFC0863700, 2020YFC2003600, and 2019YFA0706200), National Programs for Brain Science and Brain-like Intelligence Technology of China (2021ZD0200800, 2021ZD0200700), National Natural Science Foundation of China (81761128036, 81821092, 82171514 and 31900805), Young Elite Scientists Sponsorship Program by CAST (2019QNRC001) and PKU-Baidu Fund (2020BD011).

Conflict of interest

The authors declare no conflict of interest.

Ethical standards

The study was approved by the ethics committee of Peking University Sixth Hospital (Institute of Mental Health). Informed consent was received online before the respondents began the surveys. This study follows the American Association for Public Opinion Research (AAPOR) reporting guideline.

Footnotes

*

These authors contributed equally to this work.

References

Akaike, H (1981) Likelihood of a model and information criteria. Journal of Econometrics 16, 314.CrossRefGoogle Scholar
Bakk, Z, Tekle, FB and Vermunt, JK (2013) Estimating the association between latent class membership and external variables using bias-adjusted three-step approaches. Sociological Methodology 43, 272311.CrossRefGoogle Scholar
Batterham, PJ, Calear, AL, McCallum, SM, Morse, AR, Banfield, M, Farrer, LM, Gulliver, A, Cherbuin, N, Rodney Harris, RM, Shou, Y and Dawel, A (2021) Trajectories of depression and anxiety symptoms during the COVID-19 pandemic in a representative Australian adult cohort. Medical Journal of Australia 214, 462468.CrossRefGoogle Scholar
Bendau, A, Kunas, SL, Wyka, S, Petzold, MB, Plag, J, Asselmann, E and Ströhle, A (2021) Longitudinal changes of anxiety and depressive symptoms during the COVID-19 pandemic in Germany: the role of pre-existing anxiety, depressive, and other mental disorders. Journal of Anxiety Disorders 79, 102377.CrossRefGoogle ScholarPubMed
Borsboom, D and Cramer, AO (2013) Network analysis: an integrative approach to the structure of psychopathology. Annual Review of Clinical Psychology 9, 91121.CrossRefGoogle Scholar
Bringmann, LF, Albers, C, Bockting, C, Borsboom, D, Ceulemans, E, Cramer, A, Epskamp, S, Eronen, MI, Hamaker, E, Kuppens, P, Lutz, W, McNally, RJ, Molenaar, P, Tio, P, Voelkle, MC and Wichers, M (2022) Psychopathological networks: theory, methods and practice. Behaviour Research and Therapy 149, 104011.CrossRefGoogle ScholarPubMed
Brooks, SK, Webster, RK, Smith, LE, Woodland, L, Wessely, S, Greenberg, N and Rubin, GJ (2020) The psychological impact of quarantine and how to reduce it: rapid review of the evidence. The Lancet 395, 912920.CrossRefGoogle Scholar
Celeux, G and Soromenho, G (1996) An entropy criterion for assessing the number of clusters in a mixture model. Journal of Classification 13, 195212.CrossRefGoogle Scholar
Cham, H, Reshetnyak, E, Rosenfeld, B and Breitbart, W (2017) Full information maximum likelihood estimation for latent variable interactions with incomplete indicators. Multivariate Behavioral Research 52, 1230.CrossRefGoogle ScholarPubMed
Daly, M and Robinson, E (2021) Psychological distress and adaptation to the COVID-19 crisis in the United States. Journal of Psychiatric Research 136, 603609.CrossRefGoogle Scholar
Daly, M, Sutin, AR and Robinson, E (2020) Longitudinal changes in mental health and the COVID-19 pandemic: evidence from the UK Household Longitudinal Study. Psychological Medicine 13, 110.CrossRefGoogle Scholar
Davydov, DM, Stewart, R, Ritchie, K and Chaudieu, I (2010) Resilience and mental health. Clinical Psychology Review 30, 479495.CrossRefGoogle ScholarPubMed
Epskamp, S, Borsboom, D and Fried, EI (2018) Estimating psychological networks and their accuracy: a tutorial paper. Behavior Research Methods 50, 195212.CrossRefGoogle ScholarPubMed
Fancourt, D, Steptoe, A and Bu, F (2021) Trajectories of anxiety and depressive symptoms during enforced isolation due to COVID-19 in England: a longitudinal observational study. The Lancet. Psychiatry 8, 141149.CrossRefGoogle ScholarPubMed
Felt, JM, Depaoli, S and Tiemensma, J (2017) Latent growth curve models for biomarkers of the stress response. Frontiers in Neuroscience 11, 315.CrossRefGoogle ScholarPubMed
Friedman, J, Hastie, T and Tibshirani, R (2008) Sparse inverse covariance estimation with the graphical lasso. Biostatistics (Oxford, England) 9, 432441.CrossRefGoogle ScholarPubMed
Gagnon, C, Bélanger, L, Ivers, H and Morin, CM (2013) Validation of the Insomnia Severity Index in primary care. Journal of the American Board of Family Medicine 26, 701710.CrossRefGoogle ScholarPubMed
Ganson, KT, Tsai, AC, Weiser, SD, Benabou, SE and Nagata, JM (2021) Job insecurity and symptoms of anxiety and depression among U.S. young adults during COVID-19. Journal of Adolescent Health 68, 5356.CrossRefGoogle ScholarPubMed
Gopal, A, Sharma, AJ and Subramanyam, MA (2020) Dynamics of psychological responses to COVID-19 in India: a longitudinal study. PLoS One 15, e0240650.CrossRefGoogle ScholarPubMed
Greenwood, CJ, Youssef, GJ, Betts, KS, Letcher, P, McIntosh, J, Spry, E, Hutchinson, DM, Macdonald, JA, Hagg, LJ, Sanson, A, Toumbourou, JW and Olsson, CA (2019) A comparison of longitudinal modelling approaches: alcohol and cannabis use from adolescence to young adulthood. Drug and Alcohol Dependence 201, 5864.CrossRefGoogle ScholarPubMed
Hallquist, MN, Wright, AGC and Molenaar, PCM (2021) Problems with centrality measures in psychopathology symptom networks: why network psychometrics cannot escape psychometric theory. Multivariate Behavioral Research 56, 199223.CrossRefGoogle ScholarPubMed
Hannigan, LJ, Pingault, JB, Krapohl, E, McAdams, TA, Rijsdijk, FV and Eley, TC (2018) Genetics of co-developing conduct and emotional problems during childhood and adolescence. Nature Human Behaviour 2, 514521.CrossRefGoogle ScholarPubMed
Hiremath, P, Suhas Kowshik, CS, Manjunath, M and Shettar, M (2020) COVID 19: impact of lock-down on mental health and tips to overcome. Asian Journal of Psychiatry 51, 102088.CrossRefGoogle Scholar
Holmes, EA, O'Connor, RC, Perry, VH, Tracey, I, Wessely, S, Arseneault, L, Ballard, C, Christensen, H, Cohen Silver, R, Everall, I, Ford, T, John, A, Kabir, T, King, K, Madan, I, Michie, S, Przybylski, AK, Shafran, R, Sweeney, A, Worthman, CM, Yardley, L, Cowan, K, Cope, C, Hotopf, M and Bullmore, E (2020) Multidisciplinary research priorities for the COVID-19 pandemic: a call for action for mental health science. The Lancet. Psychiatry 7, 547560.CrossRefGoogle Scholar
Hu, LT and Bentler, PM (1999) Cutoff criteria for fit indexes in covariance structure analysis: conventional criteria versus new alternatives. Structural Equation Modeling: A Multidisciplinary Journal 6, 155.CrossRefGoogle Scholar
Iob, E, Frank, P, Steptoe, A and Fancourt, D (2020) Levels of severity of depressive symptoms among at-risk groups in the UK during the COVID-19 pandemic. JAMA Network Open 3, e2026064.CrossRefGoogle ScholarPubMed
Kroenke, K, Spitzer, RL and Williams, JB (2001) The PHQ-9: validity of a brief depression severity measure. Journal of General Internal Medicine 16, 606613.CrossRefGoogle ScholarPubMed
Lee, AM, Wong, JG, McAlonan, GM, Cheung, V, Cheung, C, Sham, PC, Chu, CM, Wong, PC, Tsang, KW and Chua, SE (2007) Stress and psychological distress among SARS survivors 1 year after the outbreak. Canadian Journal of Psychiatry. Revue Canadienne de Psychiatrie 52, 233240.CrossRefGoogle ScholarPubMed
Liu, X, Kakade, M, Fuller, CJ, Fan, B, Fang, Y, Kong, J, Guan, Z and Wu, P (2012) Depression after exposure to stressful events: lessons learned from the severe acute respiratory syndrome epidemic. Comprehensive Psychiatry 53, 1523.CrossRefGoogle ScholarPubMed
Lo, Y, Mendell, NR and Rubin, DB (2001) Testing the number of components in a normal mixture. Biometrika 3, 767778.CrossRefGoogle Scholar
Löwe, B, Decker, O, Müller, S, Brähler, E, Schellberg, D, Herzog, W and Herzberg, PY (2008) Validation and standardization of the generalized anxiety disorder screener (GAD-7) in the general population. Medical Care 46, 266274.CrossRefGoogle ScholarPubMed
Lu, ZA, Shi, L, Que, JY, Zheng, YB, Wang, QW, Liu, WJ, Huang, YT, Liu, XX, Yuan, K, Yan, W, Shi, J, Bao, YP and Lu, L (2022) Accessibility to digital mental health services among the general public throughout COVID-19: trajectories, influencing factors and association with long-term mental health symptoms. International Journal of Environmental Research and Public Health 19, 3593.CrossRefGoogle ScholarPubMed
Maslach, C and Leiter, MP (2016) Understanding the burnout experience: recent research and its implications for psychiatry. World Psychiatry 15, 103111.CrossRefGoogle ScholarPubMed
Maunder, RG, Lancee, WJ, Balderson, KE, Bennett, JP, Borgundvaag, B, Evans, S, Fernandes, CM, Goldbloom, DS, Gupta, M, Hunter, JJ, McGillis Hall, L, Nagle, LM, Pain, C, Peczeniuk, SS, Raymond, G, Read, N, Rourke, SB, Steinberg, RJ, Stewart, TE, VanDeVelde-Coke, S, Veldhorst, GG and Wasylenki, DA (2006) Long-term psychological and occupational effects of providing hospital healthcare during SARS outbreak. Emerging Infectious Diseases 12, 19241932.CrossRefGoogle ScholarPubMed
Mazza, MG, De Lorenzo, R, Conte, C, Poletti, S, Vai, B, Bollettini, I, Melloni, EMT, Furlan, R, Ciceri, F, Rovere-Querini, P and Benedetti, F (2020) Anxiety and depression in COVID-19 survivors: role of inflammatory and clinical predictors. Brain, Behavior, and Immunity 89, 594600.CrossRefGoogle ScholarPubMed
McKee-Ryan, F, Song, Z, Wanberg, CR and Kinicki, AJ (2005) Psychological and physical well-being during unemployment: a meta-analytic study. Journal of Applied Psychology 90, 5376.CrossRefGoogle ScholarPubMed
Mohamed, S, Ismail, Z, Shuib, N and Ali, NF (2021) Peri-traumatic distress and its relationship to resilience and coping among medical students in Malaysia during COVID-19 lockdown. Frontiers in Psychiatry 12, 789554.CrossRefGoogle ScholarPubMed
O'Driscoll, C, Buckman, JEJ, Fried, EI, Saunders, R, Cohen, ZD, Ambler, G, DeRubeis, RJ, Gilbody, S, Hollon, SD, Kendrick, T, Kessler, D, Lewis, G, Watkins, E, Wiles, N and Pilling, S (2021) The importance of transdiagnostic symptom level assessment to understanding prognosis for depressed adults: analysis of data from six randomised control trials. BMC Medicine 19, 109.CrossRefGoogle ScholarPubMed
Opsahl, T, Agneessens, F and Skvoretz, J (2010) Node centrality in weighted networks: generalizing degree and shortest paths. Social Networks 32, 245251.CrossRefGoogle Scholar
Pavlov, G, Maydeu-Olivares, A and Shi, D (2021) Using the standardized root mean squared residual (SRMR) to assess exact fit in structural equation models. Educational and Psychological Measurement 81, 110130.CrossRefGoogle ScholarPubMed
Pierce, M, McManus, S, Hope, H, Hotopf, M, Ford, T, Hatch, SL, John, A, Kontopantelis, E, Webb, RT, Wessely, S and Abel, KM (2021) Mental health responses to the COVID-19 pandemic: a latent class trajectory analysis using longitudinal UK data. The Lancet. Psychiatry 8, 610619.CrossRefGoogle Scholar
Planchuelo-Gómez, Á, Odriozola-González, P, Irurtia, MJ and de Luis-García, R (2020) Longitudinal evaluation of the psychological impact of the COVID-19 crisis in Spain. Journal of Affective Disorders 277, 842849.CrossRefGoogle ScholarPubMed
Prati, G and Mancini, AD (2021) The psychological impact of COVID-19 pandemic lockdowns: a review and meta-analysis of longitudinal studies and natural experiments. Psychological Medicine 51, 201211.CrossRefGoogle ScholarPubMed
Ripoll, J, Contreras-Martos, S, Esteva, M, Soler, A and Serrano-Ripoll, MJ (2021) Mental health and psychological wellbeing during the COVID-19 lockdown: a longitudinal study in the Balearic islands (Spain). Journal of Clinical Medicine 10, 3191.CrossRefGoogle Scholar
Saghafi, F, West, S, Lopez, V and Cleary, M (2021) Mental health impacts of family members isolated from patients in the ICU during the coronavirus disease (COVID-19) pandemic. Issues in Mental Health Nursing 43, 8790.CrossRefGoogle ScholarPubMed
Salfi, F, Lauriola, M, Amicucci, G, Corigliano, D, Viselli, L, Tempesta, D and Ferrara, M (2020) Gender-related time course of sleep disturbances and psychological symptoms during the COVID-19 lockdown: a longitudinal study on the Italian population. Neurobiology of Stress 13, 100259.CrossRefGoogle ScholarPubMed
Saunders, R, Buckman, JEJ, Leibowitz, J, Cape, J and Pilling, S (2021 a) Trends in depression and anxiety symptom severity among mental health service attendees during the COVID-19 pandemic. Journal of Affective Disorders 289, 105109.CrossRefGoogle ScholarPubMed
Saunders, R, Buckman, JEJ, Fonagy, P and Fancourt, D (2021 b) Understanding different trajectories of mental health across the general population during the COVID-19 pandemic. Psychological Medicine 3, 19.CrossRefGoogle Scholar
Schwartz, G (1978) Estimating the dimension of a model. The Annals of Statistics 6, 3138.Google Scholar
Sclove, SL (1987) Application of model-selection criteria to some problems in multivariate analysis. Psychometrika 52, 333343.CrossRefGoogle Scholar
Shaw, WS, Main, CJ, Findley, PA, Collie, A, Kristman, VL and Gross, DP (2020) Opening the workplace after COVID-19: what lessons can be learned from return-to-work research? Journal of Occupational Rehabilitation 30, 299302.CrossRefGoogle ScholarPubMed
Shi, L, Lu, ZA, Que, JY, Huang, XL, Liu, L, Ran, MS, Gong, YM, Yuan, K, Yan, W, Sun, YK, Shi, J, Bao, YP and Lu, L (2020) Prevalence of and risk factors associated with mental health symptoms among the general population in China during the coronavirus disease 2019 pandemic. JAMA Network Open 3, e2014053.CrossRefGoogle ScholarPubMed
Shi, L, Lu, ZA, Que, JY, Huang, XL, Lu, QD, Liu, L, Zheng, YB, Liu, WJ, Ran, MS, Yuan, K, Yan, W, Sun, YK, Sun, SW, Shi, J, Kosten, T, Bao, YP and Lu, L (2021 a) Long-term impact of COVID-19 on mental health among the general public: a nationwide longitudinal study in China. International Journal of Environmental Research and Public Health 18, 8790.CrossRefGoogle ScholarPubMed
Shi, L, Que, JY, Lu, ZA, Gong, YM, Liu, L, Wang, YH, Ran, MS, Ravindran, N, Ravindran, AV, Fazel, S, Bao, YP, Shi, J and Lu, L (2021 b) Prevalence and correlates of suicidal ideation among the general population in China during the COVID-19 pandemic. European Psychiatry 64, e18.CrossRefGoogle ScholarPubMed
Spittlehouse, JK, Joyce, PR, Vierck, E, Schluter, PJ and Pearson, JF (2014) Ongoing adverse mental health impact of the earthquake sequence in Christchurch, New Zealand. Australian and New Zealand Journal of Psychiatry 48, 756763.CrossRefGoogle ScholarPubMed
Tang, W, Zhao, J, Lu, Y, Yan, T, Wang, L, Zhang, J and Xu, J (2017) Mental health problems among children and adolescents experiencing two major earthquakes in remote mountainous regions: a longitudinal study. Comprehensive Psychiatry 72, 6673.CrossRefGoogle ScholarPubMed
van Borkulo, CD, Borsboom, D, Epskamp, S, Blanken, TF, Boschloo, L, Schoevers, RA and Waldorp, LJ (2014) A new method for constructing networks from binary data. Scientific Reports 4, 5918.CrossRefGoogle ScholarPubMed
Wang, Y, Murray, AM, Toussaint, AK, Chen, L, Guo, WJ, He, N, Luo, SX, Yu, JY, Liu, Y, Huang, MJ, Dong, ZQ and Zhang, L (2019) Why is the recognition rate of psychological distress under-estimated in general hospitals? A cross-sectional observational study in China. Medicine (Baltimore) 98, e16274.Google ScholarPubMed
Wang, Y, Hu, Z, Feng, Y, Wilson, A and Chen, R (2020) Changes in network centrality of psychopathology symptoms between the COVID-19 outbreak and after peak. Molecular Psychiatry 25, 31403149.CrossRefGoogle ScholarPubMed
Wang, YH, Shi, L, Que, JY, Lu, QD, Liu, L, Lu, ZA, Xu, YY, Liu, JJ, Sun, YK, Meng, SQ, Yuan, K, Ran, MS, Lu, L, Bao, YP and Shi, J (2021) The impact of quarantine on mental health status among general population in China during the COVID-19 pandemic. Molecular Psychiatry 26, 48134822.CrossRefGoogle ScholarPubMed
Wu, Z, Liu, Y, Li, X and Li, X (2016) Resilience and associated factors among mainland Chinese women newly diagnosed with breast cancer. PLoS One 11, e0167976.CrossRefGoogle ScholarPubMed
Xiao, Y, Wang, Y, Chang, W, Chen, Y, Yu, Z and Risch, HA (2019) Factors associated with psychological resilience in left-behind children in southwest China. Asian Journal of Psychiatry 46, 15.CrossRefGoogle ScholarPubMed
Zhang, Z, Sun, K, Jatchavala, C, Koh, J, Chia, Y, Bose, J, Li, Z, Tan, W, Wang, S, Chu, W, Wang, J, Tran, B and Ho, R (2019) Overview of stigma against psychiatric illnesses and advancements of anti-stigma activities in six Asian societies. International Journal of Environmental Research and Public Health 17, 280.CrossRefGoogle ScholarPubMed
Zheng, YB, Shi, L, Lu, ZA, Que, JY, Yuan, K, Huang, XL, Liu, L, Wang, YH, Lu, QD, Wang, Z, Yan, W, Han, Y, Sun, XY, Bao, YP, Shi, J and Lu, L (2021) Mental health status of late-middle-aged adults in China during the coronavirus disease 2019 pandemic. Frontiers in Public Health 9, 643988.CrossRefGoogle ScholarPubMed
Zhou, X, Snoswell, CL, Harding, LE, Bambling, M, Edirippulige, S, Bai, X and Smith, AC (2020) The role of telehealth in reducing the mental health burden from COVID-19. Telemedicine Journal and e-Health 26, 377379.CrossRefGoogle ScholarPubMed
Figure 0

Fig. 1. Flow graph for participants recruitment at three surveys.

Figure 1

Fig. 2. Predicted trajectories of depression, anxiety and insomnia from the best fitting: (a) multi-process LGCM and (b) LGCMs for three single symptoms.

Figure 2

Table 1. Predictors for intercepts and slopes of depression, anxiety and insomnia scores from the conditional LGCMs

Figure 3

Fig. 3. Predicted trajectories of depression, anxiety and insomnia across latent symptom trajectory classes from the best fitting 5-class LGMM. (a) Latent class 1: moderate/severe stable (n = 1044, 5.5%); (b) latent class 2: mild stable (n = 2885, 15.3%); (c) latent class 3: mild-increase to decrease (n = 2201, 11.7%); (d) latent class 4: mild-decrease to increase (n = 755, 4.0%) and (e) latent class 5: normal stable (n = 11 919, 63.4%).

Figure 4

Table 2. Multinomial logistic regression of psychological symptom trajectory class membership on predictors using a three-step approacha

Figure 5

Fig. 4. Evolution of psychopathological networks for depression, anxiety and insomnia after COVID-19. Psychopathological networks at (a) initial peak, (b) aftermath of initial peak and (c) late COVID-19 phase were estimated by sparse Gaussian graphical models with graphical lasso based on 18 items from PHQ-9, GAD-7 and ISI. Blue edges indicate positive correlations. Red edges indicate negative correlations. Thicker edges indicate stronger regularised partial correlations. Green nodes indicate items from PHQ-9 (depression). Orange nodes indicate items from GAD-7 (anxiety). Blue nodes indicate items from ISI (insomnia). Central symptom and bridge symptom were determined based on values of expected influence and bridge expected influence, respectively. (a) Initial peak: global strength: 8.56; central symptom: D2-Sad mood; bridge symptom: D2-Sad mood; (b) aftermath of initial peak: global strength: 8.70; central symptom: D4-Appetite change; bridge symptom: A4-Trouble of relaxing and (c) late COVID-19 phase: global strength: 8.61; central symptom: D2-Sad mood; bridge symptom: D2-Sad mood.

Supplementary material: File

Lu et al. supplementary material

Tables S1-S6

Download Lu et al. supplementary material(File)
File 35 KB
Supplementary material: PDF

Lu et al. supplementary material

Figures S1-S16

Download Lu et al. supplementary material(PDF)
PDF 1 MB
You have Access Open access

Save article to Kindle

To save this article to your Kindle, first ensure coreplatform@cambridge.org is added to your Approved Personal Document E-mail List under your Personal Document Settings on the Manage Your Content and Devices page of your Amazon account. Then enter the ‘name’ part of your Kindle email address below. Find out more about saving to your Kindle.

Note you can select to save to either the @free.kindle.com or @kindle.com variations. ‘@free.kindle.com’ emails are free but can only be saved to your device when it is connected to wi-fi. ‘@kindle.com’ emails can be delivered even when you are not connected to wi-fi, but note that service fees apply.

Find out more about the Kindle Personal Document Service.

Long-term psychological profile of general population following COVID-19 outbreak: symptom trajectories and evolution of psychopathological network
Available formats
×

Save article to Dropbox

To save this article to your Dropbox account, please select one or more formats and confirm that you agree to abide by our usage policies. If this is the first time you used this feature, you will be asked to authorise Cambridge Core to connect with your Dropbox account. Find out more about saving content to Dropbox.

Long-term psychological profile of general population following COVID-19 outbreak: symptom trajectories and evolution of psychopathological network
Available formats
×

Save article to Google Drive

To save this article to your Google Drive account, please select one or more formats and confirm that you agree to abide by our usage policies. If this is the first time you used this feature, you will be asked to authorise Cambridge Core to connect with your Google Drive account. Find out more about saving content to Google Drive.

Long-term psychological profile of general population following COVID-19 outbreak: symptom trajectories and evolution of psychopathological network
Available formats
×
×

Reply to: Submit a response

Please enter your response.

Your details

Please enter a valid email address.

Conflicting interests

Do you have any conflicting interests? *