A prospective longitudinal observation study was designed to examine temporal trends and changes in suicidal ideation over a 6-week period of time and their associations with selected stimulus and support variables from the quarantine to the post-quarantine period among university students during the COVID-19 surge in China.

This study was conducted during the COVID-19 lockdown period in Guangzhou, China, with a specific focus on university students who were under quarantine. These students were confined to their residential buildings and were not permitted to move freely in or out of these areas. Lockdown measures began on October 24, 2022, affecting all areas of Haizhu District and parts of Panyu and Tianhe Districts. Restrictions were gradually lifted starting on November 27, 2022, in Haizhu and Tianhe, with the entire city of Guangzhou fully released from lockdown by November 30. Notably, university campuses did not immediately lift restrictions but instead gradually eased them over 1 week.

Our observation started in later stages after the lockdown was implemented, which may capture the cumulative mental effects of the lockdown. This panel study included 6 waves of data, which were collected over 6 weeks. Wave 1 (November 19, 2022), Wave 2 (November 24, 2022), and Wave 3 (December 1, 2022) are quarantine periods; Wave 4 (December 8, 2022), Wave 5 (December 16, 2022), and Wave 6 (December 23, 2022) are post-quarantine periods.

Participants were recruited through a survey advertisement on the campus Bulletin Board System, a social media platform at the university used for both public announcements and internet-based social interaction. Inclusion criteria included the following: (1) quarantine university students, (2) having access to a smartphone, (3) knowing the Chinese language, and (4) being willing to participate in the panel study and provide follow-up information at 6 scheduled observation points. Participants were excluded if they refused to provide this information or had a medical condition that could limit or preclude their participation.

Within the registration system, potential participants were screened to ascertain eligibility. Upon consent, participants received an electronic questionnaire and instructions on how to proceed. After reading the instruction, they were asked to provide an electronic consent by tapping the “Confirmation and Authorization” button to opt in. After authorizing the consent, they were directed to the questionnaire and began answering the survey.

A special administrative WeChat (Tencent) group was established to manage the follow-up data collection, using a unique QR code for each respondent. The QR code was the vehicle, not only for identifying unique participants but also for prohibiting nonparticipants from taking the survey. After scanning the QR code, survey participants could enter the investigation group without further preconditions. All responses were anonymous. The questionnaire took 10 minutes to complete, and the same survey protocol was used for each wave of the survey to ensure homogeneity of data administration and collection.

All data were entered into a database using Microsoft Excel. They were then imported into SAS (version 9.4; SAS Institute Inc) for the statistical analysis. As most of the continuous variables included in this study were not normally distributed, we used nonparametric testing methods to examine the changing trends. The Mann–Kendall test was used to assess changing trends across the 6 observation points, with the parameters for statistical testing being the z value [32,33]. For categorical variables, the Cochran-Armitage test was used to examine their changing trends; the parameters for statistical testing were also the z value [34]. The generalized estimating equation was used to examine the association among confirmed new COVID-19 infected persons, behavioral beliefs, uncertainty stress, and suicidal ideation [35].

The study was conducted following the Declaration of Helsinki and was approved by the ethics committee of the Medical Center at Jinan University (JNUKY-2022‐047). All participants received an electronic instruction on how to proceed. After reading the instructions, they were asked to provide electronic consent by tapping the “Confirmation and Authorization” button and then were directed to the questionnaire. Participation in the study was entirely voluntary, with participants having the right to withdraw or discontinue at any time without any penalties or loss of benefits. All information was kept strictly confidential and anonymized to ensure participants could not be identified. As appropriate, a token of appreciation, 35 RMB (approximately US $5.00) were given to those participants who completed all 6 questionnaires.

There is little evidence on changing trends in the impact of quarantine measures on suicidal ideation. In this study, the decline in suicidal ideation did not occur immediately after the lifting of quarantine measures; instead, a delayed effect or “grace period” was observed. Uncertainty stress was consistently high throughout the observation period and positively associated with suicidal ideation.

This study aimed to fill a research gap by examining the changing trends of suicidal ideation and related factors during the transition from quarantine to post-quarantine periods in the COVID-19 surge in China. The prevalence of suicidal ideation was found to be 16.7%, 14.5%, and 14.5% during the quarantine period and 13.8%, 10.9%, and 10.0% during the post-quarantine period, showing a significant downward trend over the entire observation period. These findings indicate that quarantine has an impact on suicidal ideation. It is important to note that the decline in suicidal ideation did not occur immediately after the release of quarantine measures, but rather there was a grace period. Particularly in the early stages of release, the prevalence of suicidal ideation remained relatively high, with a larger decline observed in the later period. This suggests a time-lagged effect of lockdown on influencing suicidal ideation, a common phenomenon in public health policies. Understanding the duration of this lag can assist in predicting and addressing the consequences of implementing and lifting lockdown measures. These findings also highlight the necessity of expanding mental health services to individuals in quarantine.

Perceived risk and perceived severity showed a statistically significant upward trend across the total observation period. This means that the quarantine measure has decreased the role of perceived risk and perceived severity. This study adds new evidence on the influence of quarantine on the perceived risk of contracting COVID-19 and the perception of its severity. This phenomenon may be explained by SCM theory. Quarantine is a protective factor against stress stimuli, which can decrease high levels of perceived risk and perceived severity. When people are quarantined, they may think they are safe. However, when they are released, they may feel they have lost protection and perceive more risk and severity for COVID-19 infection.

Uncertainty stress is an important observational variable in this study. This study showed that uncertainty stress did not significantly change, remaining at a high level throughout the observation period. Extreme uncertainty is a key issue in the COVID-19 epidemic, which includes the uncertainty of COVID-19 itself and the suddenness of the lockdown and release policy; these factors may lead people to a high uncertainty stress level. Many studies found that uncertainty stress was positively associated with mental and behavioral problems during the COVID-19 period [18-21]. Due to uncertainty stress closely associated with social trust, it is possible that it influences the implementation of the policy during the COVID-19 pandemic [36-38]. We believe that sustainable high-level uncertainty stress poses a large challenge for people’s compliance with COVID-19 control policies.

The Health Belief Model argues that individual behavior is determined by perceived risk and perceived severity [11,12]. The risk of disease and the severity of outcomes are crucial predictors of mental problems [11,12]. This study found that perceived risk and perceived severity of COVID-19 were positively associated with suicidal ideation. Many studies found there is a close association between strong risk and threat perceptions of COVID-19 and mental and behavioral outcomes [13-16]. At the same time, some studies also indicate that suicide is a prominent manifestation of mental problems [26-28]. Therefore, it is rational that perceived risk and perceived severity are related to suicidal ideation.

This study found that uncertainty stress was positively associated with suicidal ideation. A host of evidence supports the assertion that uncertainty constitutes a powerful stressor [12,39,40], significantly associated with many mental and behavioral problems related to COVID-19 [18,21]. An outstanding characteristic of the COVID-19 epidemic is uncertainty. It is a new virus, and much of the information about it is unknown or incomplete, with considerable ambiguity surrounding the disease, and there are frequent changes in pathogens. It should be emphasized that this characteristic has persisted so far. It should be emphasized that uncertainty is also closely related to the government’s way of disease control, such as ambiguity in providing information, the suddenness of implementing lockdown and opening up, inconsistencies in policies issued, low professionalism and authority of experts invited, and other factors. A basic strategy to decrease uncertainty is to implement scientific decisions. All policies and interventions need to be based on scientific evidence when dealing with the COVID-19 epidemic. However, in some areas or regions, policies and interventions were not based on scientific evidence. Many of these policies and interventions were arbitrary and ineffective. Ineffective policies and interventions can cause people to question their government, which ultimately may be detrimental to disease control.

Suicidal ideation decreased over the study period, despite upward trends in perceived risk, perceived severity, and the number of new patients. However, perceived risk was positively associated with suicidal ideation, and higher uncertainty stress contributed to higher suicidal ideation. This phenomenon may be difficult for some people to understand. It should be noted that trend analysis provides a description of the changes in variables during the observation period, whereas correlation analysis of the observed research variables examines the interactions between multiple factors, considering individual differences and temporal changes. Therefore, the results from trend analysis and correlation analysis should not be expected to align perfectly. This study also performed subgroup analysis based on different sample characteristics, which yielded results consistent with the previous ones. This phenomenon is observed in many other studies [41-43].

This study found a significant increase in the number of new infectious individuals throughout the observation period, which aligns with real-world observations. When the lockdown was lifted, people regained access to the outside world, increasing opportunities for infection, which led to a rise in the number of patients [8]. While quarantine measures were linked to suicidal ideation, the increase in new infections did not show a similar association with suicidal ideation. This highlights the importance of carefully considering the mental health impacts of quarantine when designing lockdown policies.

There are several limitations to this study. First, our sample size appears to be small. Nevertheless, this is a prospective longitudinal panel study, and the variables included were repeatedly measured for each participant, which ensured statistical power for the tests used in this study was high. We estimated the power for repeated measures for each variable [44]. This analysis calculated statistical power (1-β) at given sample size (N=221), 5% significance level, and effect size used as a parameter in the model. A statistical power of 0.8 and higher was considered acceptable [45]. Since the effect size of various variables may differ, their statistical powers are also different. The results showed that statistical powers for all variables achieved accepted level (1-β=0.8 and higher). This indicates that the sample size in this study was large enough to make appropriate inferences. Second, sample attrition may introduce “cluster” bias because many longitudinal studies likely overrepresent some of these characteristics. A more sophisticated design and more representative sample would be necessary to resolve this problem. Another important limitation is that our participants were restricted to university students, who were selected due to their good cooperation in follow-up for a panel study. Thus, our results cannot be generalized to the wider Chinese population.

This study found that suicidal ideation among Chinese university students declined over time as quarantine measures were lifted, but the decrease was delayed, indicating a lagged mental health effect of the lockdown. Perceived risk and perceived severity of COVID-19 increased, suggesting that sense of safety during lockdown may have reduced risk perception, which then rebounds once restrictions are lifted. Uncertainty stress remained consistently high throughout, highlighting a persistent mental health risk burden during the COVID-19 surge. Moreover, higher perceived risk, higher perceived severity, and higher uncertainty stress were related to a higher likelihood of reporting suicidal ideation. However, the proportion of infected individuals was not linked to suicidal ideation. University students are more sensitive to perceptions influenced by the COVID-19 context rather than direct medical facts. This may primarily drive mental health problems. This study provides new information for the governments and policy makers to design effective intervention strategies targeted at reducing the COVID-19 and other infectious diseases.