Study design and participants
This study employed a cross sectional survey design and adult Singapore residents aged 21 years and above were recruited using the stratified sampling method. The sampling process involved stratifying by housing type (80% public and 20% private), region (5 sectors consisting of 28 postal districts that covered the whole of Singapore), age group (equal proportion of older and younger adults in each of the 5 sectors) and gender (equal proportion). Simple random sampling was employed to select 5 residents from each selected building. In line with age classification standards by World Health Organisation, older adults were defined as those 60 years and above. Those 59 years and below were defined as younger adults.
Door-to-door survey was conducted between 17 October 2020 and 27 November 2020. The questionnaire containing all the study measures including the validated DASS scale was in the English language. To be included for the study, participants must be able to speak English, which is the first language of Singapore, residing in the country during the CB period (7 April to 1 June 2020) and 21 years old and above of age. Only one resident of each household unit was approached for survey interview using computer-assisted personal interviewing (CAPI) technique. Participants were excluded if they exhibited signs of cognitive impairment. All interviewers were trained by the Geriatrician in the study team to screen for cognitive impairment. For individuals between age of 21–69 years old, they were excluded during the introduction stage and consent taking process if they exhibited signs of (i) memory loss, (ii) delirium (e.g., drowsy, sleepy, agitation) and (iii) language problems (e.g., repeating sentences that don't make sense). Additionally, individuals above 70 years old and above were excluded if they failed any one of three items from the Abbreviated Mental Test Score [21]. The purpose of the test was to rapidly screen the older adults for the possibility of dementia, mental confusion and other cognitive impairments. All interviewers involved were also trained to administer the survey questionnaire stipulated by the research team. To ensure the safety of participants and interviewers, data collection procedures complied with existing safe distancing measures during the study period. This study received ethics approval from National Healthcare Group Domain Specific Review Board (2020/00973) and all participants gave written informed consent.
Study measures
Mental health
Mental health status of participants was assessed using the shortened version of Depression, Anxiety, and Stress Scale (DASS-21) [22]. DASS-21 consists of three 7-item subscales designed to measure levels of depression, anxiety, and stress. A sample item for Depression included “I couldn't seem to experience any positive feeling at all”. A sample item for Anxiety included “I was worried about situations in which I might panic and make a fool of myself”. A sample item for Stress included “I found it difficult to relax”. Internal consistencies of all three subscales were found to be good (Depression, α = 0.87; Anxiety, α = 0.76; Stress, α = 0.87). Participants indicated the extent each item statement applied to them during the CB period on a 4-point Likert scale ranging from 0 (did not apply to me at all) to 4 (applied to me very much or most of the time). Each subscale was scored by summing up the respective item scores and multiplied by two [22] and categorized into either high or low value group using the median as a cut-off. This approach of categorizing variables is justified on account of the highly skewed distributions [23, 24] with most cases obtaining the lowest possible score for each of the subscales.
Adaptive behaviours
Participants were asked about their adaptability in various psycho-social domains (e.g., “I was able to adjust my regular social activities to my satisfaction”, “I was able to adjust the way I interact with those I lived with to my satisfaction” and “I was able to adjust to how I spend my free time [e.g., hobbies, entertainment] to my satisfaction”) and ability to run essential activities (e.g., “I was able to physically run essential errands that I needed to do”, “I was able to use online services to settle what I needed to do [e.g. online banking, fill application forms]” and “I was able to buy takeaway food by myself if there was a need to do so”) during the lockdown period. Participants responded to a 5-point Likert scale ranging from 1 (Strongly disagree) to 5 (Strongly agree). Both scales showed good internal consistency (Adaptability, α = 0.82; Running essential activities, α = 0.81).
Social support
A 3-item measure adapted from the social support subscale of the Resilience Scale for Adults [25] was employed to assess the social support of participants during the lockdown period. Sample items included “I have some close friends/family members who really care about me”, “I always have someone who can help me when needed” and “I can discuss personal matter with friends/family members”. Participants responded to a 5-point Likert scale ranging from 1 (Strongly disagree) to 5 (Strongly agree). The scale was found to have good internal consistency (α = 0.86).
Perceived stress concerns
Participants were asked to rate their perceived stress in relation to physical health concerns, finance related concerns, emotion related concerns, supplies related concerns, news and information related concerns, as well as change of routine. Sample items in response to the following question “Thinking about the circuit breaker period please rate to what extent you agree that the following are concerns that generally affect the stress levels of yourself? included “Physical health concerns (e.g. contracting COVID-19/ deterioration of existing health/not being able to visit doctor for check-up, etc.)” and “Finance related concerns (e.g. / losing income / not paying bills/ not being able to pay rent / losing job / impact on my business, etc.)”. Participants responded to a 5-point Likert scale ranging from 1 (Strongly disagree) to 5 (Strongly agree). The 6-item scale was found to have good internal consistency (α = 0.88).
Digital media usage
Participants self-reported their daily average time spent on the internet during the CB period. In addition, they were asked to rate their ability in using digital platforms to get updates on COVID-19 situation, and to meet their personal needs during the CB period. Sample items included “I was able to keep up to date on all the changing measures and regulations announced by the government related to the COVID-19 situation through the various online platforms (Such as Channel News Asia, Gov.sg Telegram and Whatsapp Groups)” and “ Overall, I was able to use digital platforms to meet my needs during the circuit breaker (e.g. to buy supplies, to run essential services, entertainment such as music, video or gaming, for work, socialising etc.)”. Participants responded to a 5-point Likert scale ranging from 1 (Strongly disagree) to 5 (Strongly agree).
Change in circumstances
Participants were asked to rate their perceived change in circumstances in relation to their “state of health”, “financial circumstances”, “stress levels”, and “general living circumstances” during the CB period as compared to 6 months prior to the pandemic outbreak. Participants responded to a 5-point Likert scale ranging from 1 (a lot worse now) to 5 (a lot better now).
Other measures
Demographic data were collected on age, gender, marital status, ethnicity, religion, education levels, and occupation.
Sample size calculation
Based on a priori power analysis (G*Power 3.1.9.7) using a power of 0.80 and error probability of 0.05, a sample size of 300 participants is required for each group to detect a between-group difference of small effect size.
Statistical analysis
Independent samples t-tests were performed to examine differences in mental health, adaptive behaviours, social support, perceived stress, change in circumstances, and digital media usage between younger and older adults. Bivariate Pearson’s correlations were conducted to examine the relationship between age and other continuous measures. Logistic regressions were performed to ascertain the effects of age group, adaptability, social support, and self-perceived health status on the likelihood of poor mental health after adjusting for education status, gender, employment status, and digital media timespan. All analyses were conducted using Stata version 14.0 (StataCorp, Texas).