The prevalence rate and demographic correlates of perceived stress in Alberta during the COVID-19 pandemic: A one-week cross-sectional study

Coronavirus Disease 2019 (COVID-19) continues to have an unprecedented global effect on health and daily life, with many countries struggling to adapt to the adverse pandemic impact. While strict public health measures are necessary to slow the virus’ spread, these measures may adversely affect individual mental health and wellbeing. Texting-based programs offer organizations a feasible and cost-effective option to deliver mental health supports and to collect population-level data. This study reports on the prevalence rate and demographic correlates of perceived stress on the one-week data obtained from Text4Hope enrollees during the COVID-19 pandemic. was a


Trail registration:
The study was approved by the University of Alberta Human Research Ethics Board (Pro00086163).

Background
Coronavirus Disease 2019 (COVID- 19) is an acute respiratory disease with severe cases resulting in potential death due to massive alveolar damage and progressive respiratory failure. [1][2] The rst case of the virus was reported in Wuhan, China in December 2019 and has spread globally. The World Health Organization (WHO) declared COVID-19 a pandemic on March 11, 2020, 3 and the virus continues to have unprecedented global impact on health and daily life, with many countries globally struggling to adapt to the adverse pandemic impact.
In early April 2020, Canada con rmed nearly 20,000 cases. 4 As COVID-19 spreads, Canadians continue to experience severe economic and societal impacts. Unemployment reached 7.8% in March 2020 with an unprecedented 3.18 million Canadians applying for Federal jobless bene ts and emergency income aid. 5,6 Closing of schools, small and large businesses, and the effect of self-isolation or self-quarantine impose increasing population challenges. The implementation of physical distancing as a public health measure has resulted in unavoidable increase in relative social isolation on a national scale. 7 The implementation of public health measures are necessary to slow the spread of COVID-19 and to " atten the curve" of infection and death rates. 8 Undoubtedly, these measures will have short-and longterm adverse consequences on the mental wellbeing of individuals. In a recent Canadian survey, the Mental Health Index dropped 12 points to 63, a score typically seen when people experience major life disruptions and mental health risks. 9 A survey recently completed in China during January 31 to February 2, 2020, also indicated adverse psychological impact of the virus. 10 Over half of respondents rated the psychological impact of COVID-19 as moderate or severe, with 29%, 17%, and 8%, respectively, reporting signi cant anxiety, depressive, and stress symptoms. This year, Nelson and colleagues reported that mental health symptoms for depression, anxiety, and stress were elevated across adults living in predominately English speaking countries. 11 They found that mental health symptoms were associated with loss of employment and being concerned about COVID-19. Another early pandemic response study examining stress levels of individuals in different countries around the world from March 17 to April 1, 2020, found that on average, individuals reported being moderately stressed. 12 Factors associated with signi cantly higher scores included identifying as female, young age, being a student, expressing concerns about virus contraction, and feeling susceptible to the coronavirus. Additionally, healthcare workers have reported more severe levels of anxiety, depression, insomnia, and somatization symptomologies during the COVID-19 pandemic. [13][14][15] The mental health effects to date from the COVID-19 pandemic appear to resemble those observed in other large-scale disasters. As such, using evidence from large-scale disaster research may help further our understanding of the mental health impact of the virus. Large-scale disasters are often accompanied by increases in depression, post-traumatic stress disorder (PTSD), substance-use disorder, a broad range of other mental and behavioural disorders, domestic violence, and child abuse. 16 These psychological impacts are often immediate and long-term. Based on interviews of residents affected by Hurricane Ike in 2008, two to ve months after the event, 6% and 5% met the criteria for PTSD and major depressive disorder, respectively. 17 Similarly, 8% of New Yorkers showed signs of PTSD and approximately 10% displayed symptoms of depression in the month following the 9/11 terrorist attackes. 18 Patients and clinicians during the Severe Acute Respiratory Syndrome (SARS) pandemic had reported signi cantly higher stress, depression, anxiety, and PTSD symptoms compared to controls, which were evident a year later. 19 The natural disaster literature also provides information on risk factors that contribute to developing psychological conditions after a major incident. Risk factors can include the degree of exposure, [20][21][22][23][24] gender, 10,25-28 social stressors such as unemployment status 25 or low socioeconomic status, 26 and preexisting physical and mental health conditions. 10,26,29,30 Early evidence from China revealed that healthcare workers exhibited signi cant symptoms of depression, with close to half of individuals experiencing signi cant anxiety symptoms and one-third experiencing sleep-related issues. 31 Correlates of symptomatology were related to exposure (e.g., working in Wuhan and working on the front line) and demographic factors, such as gender and occupation (i.e., identifying as female and being a nurse, respectively). Taken together, COVID-19 will likely have signi cant detrimental effects on mental wellness in the general population and may be more pronounced in certain groups (e.g., identifying as female, being socially stressed, being a frontline worker, or having a pre-existing mental disorder).
The Government of Alberta (GOA) released COVID-19 modelling on April 8, 2020. 32 In three scenarios ranging from the probable, elevated, and extreme spread of the virus, Alberta is expecting between 800,000 to 1.6 million cases and 400 to 32,000 deaths in probable and extreme scenarios, respectively.
Recent hospitalization and intensive care unit data released from the GOA on April 28, 2020, suggest that social distancing and other public health measures implemented by the province have had a positive impact of attening the curve with current numbers being far lower than expected. 33 To maintain the province's positive momentum, mental health services can be provided to Albertans during the COVID-19 pandemic through mobile-based interventions. These interventions can be implemented quickly, are simple for people to join, cost-effective, and accessible. Texting-based programs are especially promising given that the vast majority of Canadians own a smartphone 34 and textmessaging is often free or low-cost to end-users, does not require technical skill for use, and does not require expensive data plans. This intervention option is also cost-effective to providers, costing pennies per message to deliver.
Several randomized control studies (RCTs) have demonstrated effectiveness of supportive text messages in reducing depressive symptoms, increasing abstinence duration in alcohol use disorder, and promoting high user satisfaction. In two RCTs, patients with depression showed reduced symptoms on standardized self-report compared to patients not receiving messages (with large effect sizes: Cohen's d=0.85, Cohen's d=0.67). 35,36 Small and moderate effects were found in another RCT evaluating an addiction-related supportive messaging mobile intervention. 37 The authors found that cumulative abstinence duration and the average number of days until the rst drink was greater than twice the length for those receiving supportive addiction-related messages compared to controls (approximately 60 days versus 26 days, respectively). In two user satisfaction surveys, over 80% of subscribers reported that the texting-based program improved their mental health. 36,37 Subscribers reported that text messages made them feel more hopeful about managing issues (82%), in charge of managing depression and anxiety (77%), connected to a support system (75%), and improved their overall mental wellbeing (83%).
The aim of this study is to demonstrate how mobile based interventions can be used to quickly and safely collect population-level data by reporting on the one-week prevalence rate and demographic correlates of perceived stress in Alberta during the COVID-19 pandemic. Quick access to data can help policy makers assess the healthcare needs of a population, rapidly disseminate information to the public, implement public health measures in a timely manner, and properly allocate resources during a pandemic.

Design
We conducted a cross-sectional study examining baseline survey data collected one-week after the Text4Hope program launch (March 23 to March 30, 2020). Text4Hope is a text-messaging program aimed at providing Albertans support during the COVID-19 pandemic. The program was endorsed by provincial government o cials, including Alberta's Chief Medical O cer of Health and the Alberta Premier. Additionally, funders (see acknowledgements) actively promoted the program on social media platforms, websites, newsletters, and posters. Individuals self-subscribed to the program by texting "COVID19HOPE" to a short-code number. Once subscribed, individuals received one daily text message, at the same time each day, for 12 weeks. Messages were aligned with a cognitive behavioural framework, with content written by mental health therapists, psychologists, and psychiatrists.
The baseline survey was administered via Select Survey, 38 an online survey software, by Alberta Health Services (AHS). All Text4Hope subscribers were invited to complete the survey. Individuals were asked demographic questions about their identi ed gender, age, identi ed ethnicity, highest level of education completed, current employment status, current relationship status, and current housing status.
Individuals were also asked to complete the Perceive Stress Scale (PSS). 39 The PSS is a validated, 10item scale that measures the degree to which situations in one's life are appraised as stressful. PSS scores between 0 and 13 were classi ed as "low stress," scores between 14 and 26 were classi ed as "moderate stress," and scores between 27 and 40 were classi ed as "high stress." Individuals classi ed as moderate stress or high stress were collapsed together as moderate/high stress for analysis.
The study was approved by the University of Alberta Human Research Ethics Board (Pro00086163).
Before starting the survey, participants had to read brief informative statements which included: "Completing the survey means that you are agreeing to take part in the evaluation." Consent was therefore implied if the participants completed the 10 minute survey and submitted their responses. Since the population of Alberta is approximately 4.3 million people, the sample size needed to estimate the prevalence of stress levels with a con dence level of 99% and a 2% margin of error was 4,200 individuals.
The expected response rate was 20%. 36 Data Analysis Data were analyzed using the Statistical Package for Social Sciences (SPSS) version 26. 40 Descriptive statistics for demographic and clinical characteristics are described in number and percentages. Crosstabular univariate analyses with Chi-square or Fisher's exact tests were used to explore the relationship between each categorical variable in the study and moderate/high stress. A total of 513 responses (8.5%) were excluded from analysis due to the PSS variable being incomplete (e.g., a participant left the PSS blank) or invalid (e.g., there were more than 1 PSS item responses missing). Variables with a statistically signi cant relationship (p<0.05, two-tailed) and variables that trended toward signi cance (0.05 ≤ p ≤ 0.10, two-tailed) with the likelihood of moderate/high stress on univariate analysis were analysed using logistic regression modelling. Before performing the logistic regression analysis, correlation diagnostics were performed to ensure that high inter-correlations among the predictor variables were avoided. Odds ratios from the binary logistic regression analysis were examined to determine the association between each of the variables in the model and the likelihood of individuals reporting moderate/high stress, controlling for the other variables in the model.

Results
There were 32,805 individuals enrolled in Text4Hope one week after the program launched. The online survey received 6,041 responses (18.4% response rate). The prevalence rate for moderate/high stress in our sample was 84.7% (n=5,528). Table 1 provides summaries for demographic and clinical characteristics of respondents. The most common respondent demographic characteristics reported were: 86.6% identi ed as female (n=5,986); 43.3% were between 41 and 60 years of age (n=5,870); 82.2% identi ed as Caucasian (n=5,970); 85.6% had post-secondary education (n=5,983); 72.1% were currently employed (n=5,166); 71.6% were married, common-law, or partnered (n=5,889); and 66.5% owned their own home (n=5,982).   Individuals who identi ed as female were 1.5 times more likely to report moderate/high stress than individuals who identi ed as male (95% CI=1.2 -1.9). Individuals who identi ed as other gender were 2.1 times more likely to report moderate/high stress compared to those who identi ed as male (95% CI=0.58 -7.21). However, this was not signi cant.
Those who were unemployed were 2.5 times more likely to report moderate/high stress than those who were employed (95% CI=1.8 -3.6). Those who were retired were 1.6 times less likely to report moderate/high stress compared to those who were employed (95% CI=1.1 -2.2). While students were 1.2 times more likely than those who were employed to report moderate/high stress (95% CI=0.7 -2.1), this difference was not signi cant.
Renters were 1.7 times more likely to report moderate/high stress than homeowners (95% CI=1.3 -2.1). Those who were living with family or identi ed their current housing status as "other" were more likely to report moderate/high stress than those who owned their own home; 1.3 (95% CI=0.8 -2.2) and 3.5 times more likely (95% CI=0.8 -15.1), respectively, but these differences were not signi cant. reported feeling "quite a bit" or "extremely stressed." 41 While the surveys are not directly comparable (i.e., individuals are asked to re ect on different timeframes), the prevalence rate from this study indicates that stress levels may be elevated compared relatively recent Canadian data. Our ndings align with the literature, which suggest that elevated psychological stress is typical of epidemic and disease outbreaks. [42][43][44] They also align with other evidence emerging from the COVID-19 pandemic. [10][11][12] Our prevalence rate is substantially higher than that reported by Wang et al. at 8.1% in the general Chinese population. 10 We speculate that the discrepancy between prevalence rates is possibly attributable to differing survey times. In the Chinese study, data was collected from January 31 to February 2, 2020, a day after the WHO declared COVID-19 a public health emergency of international concern on January 30, 2020. At this point, there were 7,818 total con rmed cases worldwide. 45 Our study collected data from March 23 to March 30, 2020, which was shortly after the WHO had declared COVID-19 a pandemic on March 11, 2020. At that time, there were 118,319 con rmed cases globally, 46 a 15 time increase compared to January 30, 2020. Nevertheless, there are cultural and contextual differences that may also explain the difference.
In relation to correlates of perceived stress in the present sample, identifying as "other gender," being ≤ 25 years of age, having attained less than a high school diploma, being unemployed, being single, and living with family were associated with signi cant stress. Several of these factors have been reported in association with higher rates of psychological stress during COVID-19 and natural disasters. 10,25,26,31 Our data indicate that speci c groups are at risk for greater levels of stress in response to public health emergencies and this represents an important focus for interventions during times of crisis. Overall, our ndings are consistent with other studies examining the demographic correlates of stress. [10][11][12] Our model identi ed gender as a predictor of moderate/high stress. Speci cally, female identifying individuals were 1.5 times more likely to report moderate/high stress than male identifying individuals.
This is comparable to prior results for the general population 10,12 and for healthcare workers 15 during the current pandemic. As proposed by others, 12 the greater likelihood of reporting moderate/high stress in female identifying individuals may be attributed to a number of factors. Several studies have suggested that females generally score higher on the PSS, 47,48 report higher intensity of psychological symptoms, 49 and are exposed to a greater number of stressful life events compared to those identifying as male, 50 which may contribute to the higher prevalence estimate with female identifying participants.
The association between age and psychological stress in the literature is mixed. Some reports indicate an absence of an association while others observe a signi cant positive or negative correlations between age and stress. 10,11,12 In agreement with others, 12,51 we found a negative correlation between age category and moderate/high stress in our model. Speci cally, individuals in the ≤ 25 year category were more likely to report on moderate/high stress levels compared to all other age categories. While COVID-19 symptoms are more severe and mortality rates are higher for older individuals, 52 there is evidence to suggest that older individuals may generally score lower on the PSS compared to younger individuals, suggesting that stress perception naturally declines with age. 48 Additionally, older individuals have been found to have lower reactivity to stressful events, which may be related to greater resiliency or more effective emotional regulation abilities. 51 Another explanation, based on research conducted during the SARS outbreak, is that younger individuals tend to have signi cant psychiatric morbidities, which has been shown to correlate with higher levels of stress. 12,53 More research is needed to elucidate the effects of age on perceived stress for COVID-19.
Employment status contributed signi cantly to our model, with unemployed individuals having more than 2 times the likelihood to report moderate/high stress compared to those who were employed. Similarly, Nelson et al. reported that the loss of employment due to COVID-19 was related to increased psychological symptoms. 11 Periods of economic recession have been linked with degradation of mental health, negative coping strategies (e.g., alcohol and drug use), and suicide rates. 54,55 These detriments may persist for long periods during nancial recession. Given the unprecedented number of Canadians applying for Federal jobless bene ts and emergency income aid, our ndings are aligned with the reporting of current trends in the economy. Since we found age to be negatively correlated with moderate/high stress, it is not surprising that retirees were 1.6 times less likely to report moderate/high stress compared to those who were employed. Generally, older individuals are more likely to be retired. We found no statistical association between being a student and reporting moderate/high stress. This contrasts a positive association with being a student and reporting higher stress scores found by Wang et al. 10 A reason for these differences could be related to how participants were sampled. In the Chinese study, the online survey link was initially disseminated to university students who were then encouraged to share the link with others. In this study, Text4Hope was endorsed by provincial government o cials and presented as a service offered to the general public. As such, the sampling method in our study may be more re ective of a general population.
Finally, our model suggests that current housing status is a predictor for moderate/high stress. Similar to unemployment status, this nding may be linked to economic instability. There is evidence supporting the association of housing foreclosures with unemployment, poverty, and suicide rates. 55 In our study, renters were 1.7 times more likely to report moderate/high stress than home-owners. Although federal emergency funding and nancial aids have been put in place throughout Canada, there are several stressors directly impacting renters. In a recent survey of 1,100 renters across Canada, around 70% of respondents reported that COVID-19 had impacted their nancial situation and 42% stated that they did not qualify for employment insurance or federal emergency funding. 56 The report also indicated that 35% of respondents did not have enough money to pay rent on May 1, 2020. In Alberta, this is the date where rent non-payment eviction can be enforced, which may have contributed to higher reporting of moderate/high stress in renters.
There are several limitations in this study. This is a report of a single cross-sectional sample, which limits the ability to track changes in perceived stress over time. Additionally, Text4Hope is an intervention to help reduce stress and anxiety due to the COVID-19 pandemic. It is possible that voluntary subscribers differ in marked ways from the general population of Alberta, which could systematically in uenced the perceived stress prevalence rate. Speci cally, individuals who have higher stress levels may be more likely to enrol in the Text4Hope program. Additionally, the current study was limited to a Canadian setting and was overwhelmingly represented by individuals who identi ed as Caucasian (82.2%), which may limit generalizability to other non-Western countries. Finally, this survey is unable to measure the direct effect of COVID-19 on persons with any con rmed mental disorder.
Despite the above limitations, our ndings have clinical and policy implications. First, the results provide important information on demographic correlates of perceived stress in Alberta during the COVID-19 pandemic. This can help health authorities identify high-risk groups based on sociodemographic information for early psychological interventions. The results can also be used as a historical reference for future studies of mental health impacts of pandemics and other natural disasters. Notably, our ndings may inform aspects of the development of mobile-based interventions intended to provide support and minimize negative psychological impact during pandemics and other disaster settings. In particular, we provide a framework on how mobile-based interventions can be a useful tool to provide mental health support and can be used to collect population-level data in a short timeframe. This approach may be a key strategic contribution to inform stakeholders for time sensitive policy decisions during pandemics and other disasters. The current results support the case for mobile-based interventions in planning how to allocate resources to meet the mental health needs of at risk populations in a disaster context.
The near-and longer-term psychological impacts of COVID-19 remain unknown. Longitudinal studies will be necessary for our attempts to understand any temporal effects occurring due to this pandemic. In particular, time series analysis should be conducted to determine the changes in psychological impacts in a population in this context. The Text4Hope study team intends to report on the 6-and 12-week Text4Hope data once they become available. The upcoming dataset will provide information on the effectiveness of daily supportive text messages during a pandemic. Additionally, we will be able to report on user satisfaction, changes in the severity of psychological stress, and self-reported changes in health utilization patterns against the current baseline reference data for prevalence estimation from this rsttime sample.

Conclusion
The COVID-19 pandemic has brought economic and societal changes on a global scale. These changes are likely to contribute to detrimental short-and long-term psychological impacts. Results from this paper indicate that the virus has contributed to elevated levels of stress in Alberta, with 84.7% of respondents indicating moderate/high stress levels. Those identifying as female, are 25 years or younger, currently unemployed, and are currently renting are populations that are at heightened risk of psychological stress.
The implementation of mobile-based interventions is potentially an effective and feasible solution to provide psychological support. Furthermore, the information gathered from the prevalence rate and demographic correlates of perceived stress can help policy makers identify and properly allocate resources to support the mental health needs of populations who are at higher risk of being negatively affected by COVID-19.