We conducted two population-based telephone surveys of Hong Kong citizens between 17 January to 26 February 2018 and between 15 July to 7 August 2020. Both surveys were conducted using the Web-based Computer Assisted Telephone Interview (Web-CATI) system (supplementary material 1). They followed the American Association for Public Opinion Research (AAPOR) reporting guidelines. All respondents provided verbal consent as approved by the Institutional Review Board and Ethics Committees.
All interviews were conducted anonymously. Inclusion criteria of both surveys were adult Hong Kong residents (18 years or older), able to speak Cantonese, and able to give verbal consent. The target sample was divided into 10 gender-age strata to ensure the sample characteristics were similar to those of the general population .
Survey instrument and Measures
Both surveys included questions on the respondent’s demographic characteristics, including age, gender, and education level. Questions on marital status, employment status, and income were included in survey 2020 only. We use a validated single-item life satisfaction ranging from 0 – 10, with 10 indicating the most excellent satisfaction .
We used six items extracted and regrouped from the List of Threatening Experiences  to assess the respondent’s experience of stressful life events in the past year. The six items captured the following sub-types of SLEs: (1) family/relationship issues, (2) legal issues, (3) work/school stress, (4) financial difficulties, (5) health/injury (self), and (6) health/injury (family/close people). To further capture the impact of social unrest and COVID-19 on the population in the 2020 survey, for each of the SLEs endorsed, we asked the respondents whether this event was related to social unrest only, related to COVID-19 only, or related to both social unrest and COVID-19. Those reporting the SLE as not related to either social unrest or COVID-19 social unrest were considered as personal SLEs.
In survey 2020, we used a 4-item Patient Health Questionnaire (PHQ-4) to assess the respondent’s depressive (2 items) and anxiety (2 items) symptoms in the past two weeks . Respondents rated the frequency of each symptom on a 4-point Likert scale from 0 (not at all) to 3 (nearly every day). The summary score of depressive and anxiety symptoms was calculated by summing up the two corresponding items (0 – 6, 6 indicating the highest level of depressive or anxiety symptoms). The PHQ-4 demonstrated high internal reliability, with a Cronbach’s α of 0.83 in this study. We also assessed the respondent’s experience of rumination using two items on the brooding of the Rumination Response Scale . We also used a validated 2-item Connor-Davidson Resilience Scale . Respondents rated each item of both scales on a 4-point Likert scale, ranging from 0 (almost never) to 4 (almost always). The summary score of each scale was calculated by summing up the responses (0 – 8, 8 indicating the highest level of rumination and resilience). In this study, the Cronbach’s α of the scale of rumination and resilience was 0.77 and 0.65, respectively.
The response rate was calculated using the American Association for Public Opinion’s Research Response Rate definition . Two-sided P < .05 was considered statistically significant. All analyses were performed with Statistical Package for the Social Sciences (SPSS), version 26.0. Weighting was applied to the dataset to adjust for any differences between the sample and the general population.
To test life satisfaction as a good indicator for general mental health, we examined the partial correlation between life satisfaction and depression and anxiety symptoms using the data of survey 2020, with age, gender, and education as controlling variables. Life satisfaction and experience of all sub-types of SLEs were compared between 2018 and 2020 using the Mann-Whitney U test (as the assumption of data normality was not met) and chi-square test, respectively. To compare the relationship between personal SLEs and life satisfaction in the 2018 and 2020 samples, a moderation analysis, with the year of the survey as a dichotomous moderator variable, was performed with PROCESS in SPSS. The model was adjusted for age, gender, and education level. A significant interaction effect would suggest that the relationship between personal SLEs and life satisfaction varied across the two years.
To assess SLEs and psychological factors associated with life satisfaction, we built a 6-step hierarchical multiple regression using the data from survey 2020. The outcome was life satisfaction, and covariates included gender, age, education, marital status, employment status, and income. Variables were added to the regression in the following order: (1) demographic variables (as covariates); (2) number of personal SLEs (those not associated with either social unrest or COVID-19); (3) number of SLEs related to social unrest only; (4) number of SLEs related to COVID-19 only; (5) number of SLEs related to both social unrest and COVID-19, and (6) psychological factors, including rumination and resilience. The SLEs variables were added according to the chronological order of events. This procedure allowed for determining additional variance in life satisfaction separately explained by the different types of SLEs on top of personal SLEs and psychological responses. The multicollinearity in the regression models was assessed using the variance inflation factor (VIF). Values of VIF exceeding 2.5 would be regarded as indicating multicollinearity.