Research context
Given the rapid expansion of the epidemic, many provincial- and municipal-level Chinese governments initiated the highest-level of response to a major public health emergency since the end of January 2020 [12]. These responses included strict measures to conduct comprehensive investigation of people arriving from cities with a higher burden of disease. Shenzhen in Guangdong Province is one of the cities that implemented strict response measures. Shenzhen, a city with 4 million permanent residents and 8.5 million migrant people, required all residents who had visited or stayed in key epidemic areas (i.e., Hubei Province) in the past 14 days to socially isolate for 14 days, and all people who had close contact with infected COVID–19 patients were placed in centralized isolation locations. Despite of encouraging citizens to adopt self-isolation like many other countries, China organized a mandatory social isolation for these groups from late January to late April. On February 21, 2020, the number of people who were under home or centralized social isolation in Shenzhen was 25,000.
An online cross-sectional survey was conducted on people in mandatory home or centralized social isolation in Shenzhen from February 28 to March 6 in 2020. As of February 28, the confirmed cases of COVID–19 patients in Shenzhen was 418. Among them, 141 were imported from Hubei Province, the epicenter of disease, and 261 were local cases [13].
Sample collection
Eligible participants were required to be older than 16 years old and were currently under or had completed mandatory home or centralized social isolation during the survey period. People under mandatory home or centralized social isolation can only be accessed by public health service providers. We randomly selected two districts of Shenzhen (Luohu and Longgang) and partnered with their local public health workers in communities to distribute the survey. The local public health workers sent the survey links or QR codes containing the survey link to the participants. Participants completed the survey voluntarily and anonymously.
The sample size was calculated based on the assumption that the people under home or centralized social isolation would present 10% higher depressive syndrome than the known normal people under COVID–19 [14]. Therefore, a sample of size n = 335 would ensure that a statistical analysis with α = 0.05 has 80% power to examine the differences of people under mandatory social isolation in depressive syndromes.
This study has received approval from the institutional review board of Shenzhen Center for Disease Control and Prevention.
Measurements
Depression
Depressionrefers to a variety of negative psychological symptoms including in depressive mood, loss of interest, fatigue, difficulty with attention, and suicidal ideation. Depression was measured by the Patient Health Questionnaire Depression scale (PHQ–9) [15], which was previously validated for use among Chinese adults. The participants were asked to report the extent to which they had the nine psychological symptoms on a four-point scale, from “not at all” (0) to “nearly every day” (3). Its reliability in this study was acceptable, with a Cronbach’s alpha coefficient of 0.89. The scores of the nine items were averaged for analysis.
Perceived risk
Perceived risk indicates individuals’ subjective perception to certain risks, specifically represented by perceived severity and perceived susceptibility. Perceived severity in the present study was assessed by the perception of how long the epidemic would last. Participants were asked to rate on a five-point scale, ranging from “Less than one week” (1) to “More than six months” (5). Perceived susceptibility was measured by the perception of possibility that one could get infected during the social isolation. Participants rated the probabilities from “highly unlikely” (1) to “highly likely” (5).
Perceived tone of media coverage
Perceived tone of media coverage showcases individuals’ perception about the tone or emotions of media coverage regarding COVID–19. The perceived tone of media coverage was evaluated by seven items on a bipolar semantic scale. A wide array of opposite adjectives was presented, including negative vs positive, critical vs encouraging, complaining vs forgiving, non-reflective vs reflective, worried vs composed, indifferent vs touching, timid vs brave. Participants were required to select on a seven-point scale, ranging to –3 to 3. Cronbach’s alpha coefficient in this study was 0.91, indicting excellent reliability. The scores of the seven items were averaged for analysis.
People-oriented public health services
People-oriented public health services measured whether public health officials and workers who were tasked with supporting the daily life routine of those people who were in mandatory home or centralized social isolation were well understood, seen as caring, and trusted. People-oriented public health services were measured by three statements, such that “Public health service (PHS) workers responded to my question in the ways that I can understand”, “PHS workers cared about my feelings and emotions”, and “I perceived the PHS workers trustworthy”. Participants rated on a five-point scale, from “Strongly disagree” (1) to “Strongly agree” (5). The reliability was excellent, with Cronbach’s alpha coefficient at 0.95. We averaged the scores of each item and further categorized the people-oriented public health services into high quality (averaged scores higher than 4), and medium quality (averaged score between 3 and 4) and low quality (averaged scores lower than 3).
Control variables
Demographic variables, including age (continues variable), gender (female = 0, male = 1), education (categorical variable from primary education or below to master’s degree or above), and monthly income (categorical variable from no income to higher than 30,000 RMB), and venue of social isolation (home or centralized) were controlled for in this study. In addition, prior studies have found that media exposure and online social support could affect depressive symptoms [16, 17]. Therefore, factors such as participants’ time on COVID-related news and social support received online were also controlled in analysis. Time on COVID-related news was measured by how much time the participants spent paying attention to COVID-related news, ranging from few times (less than one hour) to many times per day (more than seven hours). Online social support was measured through how participants reported receiving information, emotional, instrumental and esteem support from others online. Participants rated on a five-point scale whether the statements such as “When I felt scared, I turned online to my relatives and/or friends to talk about my feelings” was similar to their experiences or not.
Statistical analysis
Demographic characteristics and perceptions of participants were first described. Then a multiple regression was conducted to examine the main effects of perceived susceptibility, perceived severity, perceived tone of media coverage, and people-oriented public health services on depression. These variables were mean centered to avoid multicollinearity before conducting the moderation analyses. Finally, three stepwise multiple regressions were performed to examine the moderating effects of perceived susceptibility, perceived severity, and perceived tone of media coverage with people-oriented public health services, respectively. Age, gender, education, monthly income, venue of social isolation, time spent on COVID-related news, and online social support were controlled in all regression models. The data analyses were conducted using R programming language, and p-values less than or equal to 0.05 were considered as statistically significant.