The study was conducted at Victoria University (VU) in the western metropolitan region of Melbourne, Australia. VU has eight Victorian campuses, with 1994 academic (teaching and research) and professional staff, 988 sessional staff (academic and polytechnic teachers), and 598 casual professional staff. Among the total VU staff, 59.64% (2135/3580) identify as female.

All staff received an email invitation to participate in the online program and to opt-in to the research component. All staff were eligible to participate in the program, with no exclusion criteria or minimum attendance requirements. Using the general university mailing list, invitations to join the daily online sessions were issued over the first 2 weeks of the program, with staff asked to opt-in via email to receive ongoing meeting invitations. At the end of the program, all those who opted in to meeting invitations (17.93%, 642/3580) were invited by email to participate in the semistructured interviews.

All procedures involving human participants were approved by the Victoria University Human Research Ethics Committee (HRE20-054) and complied with the ethical standards of this institutional committee.

This study followed a mixed methods design. A repeated-measures within-subjects design was used to collect data on mental health outcomes and potential mediators of change from baseline to 32 weeks after program commencement. Initially, it was planned that follow-up assessments would occur every 3 weeks. However, given that the program ran longer than anticipated due to Melbourne’s extended lockdown period, the repeated-measures time points became longer as the program progressed to reduce participant burden. Additionally, brief weekly questionnaires were used to determine the acceptability of the program over the first 13 weeks of the program. At the end of the program, we also conducted in-depth semistructured interviews in a small sample of participants to explore their experiences with VU Elevenses.

The “VU Elevenses” program focused on mental health promotion, including universal, selected, and indicated prevention strategies [13]. Participants were provided information on how to access mental health treatment from existing internal (ie, Employee Assistance Program) and external (ie, Beyond Blue, headspace, Black Dog Institute, Lifeline, Relationships Australia, Family Relationships Advice Line) sources, as the “VU Elevenses” program was not designed to be a treatment for mental health conditions. The program was aligned with the VU Employee Wellbeing Policy [14].

The “VU Elevenses” program comprised brief (10-15 minutes), evidence-informed microinterventions and strategies to promote physical and mental well-being, utilizing an online meeting platform at 11 AM each weekday. The intervention had three main phases: (1) managing immediate concerns and stressors, (2) adjusting to working and studying remotely, and (3) preparing to return to work and study on campus. The design and implementation of the program were guided by the principles of inclusivity and accessibility, responsiveness, consistency, and connectedness. The sessions were live and unpolished to maintain authenticity and connection, and to allow for the rapid sharing of evidence-based clinical content for mental health and well-being support.

A collective of VU practitioners and researchers rotated and delivered interventions relevant to one of six identified lifestyle areas (healthy eating, physical activity, reducing alcohol intake, improving sleep, healthy relationships and social connection, and stress management) [15]. Each microintervention aimed to promote skill-building through simple mindfulness strategies, deep breathing exercises, relaxation exercises, time-management and routine-setting strategies, self-compassion strategies, physical activity guidance, sleep tips, nutrition advice, and fun activities for community connection (eg, quizzes and group singing sessions; see Table 1).

Members of the program team planned the following week of microintervention content, resources, and presenters. Resources for further information (ie, fact sheets, websites, apps, YouTube clips, online interventions) and links to recordings were provided to participants for each daily session via VU’s intranet. The program was evaluated for 32 weeks, from April 2020 to November 2020, with sessions each weekday for the first 24 weeks, followed by three times per week for the final 8 weeks of the program.

To determine acceptability of the program, percentages of participants who agreed or strongly agreed they were satisfied with the content of the session, felt the content helped with their mental health and well-being, and thought the program made them feel more connected with their colleagues during the first 13 weeks of the program were calculated. Additionally, the average number of live daily sessions attended by respondents in a week was calculated.

The semistructured interviews were digitally recorded and then transcribed verbatim. Transcripts were read and reread while listening to the recordings to ensure accuracy. Braun and Clarke’s [27] thematic analysis guide was used to examine the personal experiences and meanings of the program. Data were organized in a meaningful and systematic way, with coding used to reduce data into smaller chunks of meaning. Codes were examined and organized into broader themes and were reviewed and modified. The final themes were defined and named.

Quantitative data analysis was performed using SPSS (v.25) and the R package. Trajectories of change in symptoms of depression, anxiety, and stress were estimated using multilevel growth curve models. First, an unconditional model with a random intercept was run for each outcome (null model). Subsequently, a series of models were run with a fixed linear effect of time (Model A), quadratic effect of time (Model B), and cubic effect of time (Model C) added. Because the time between follow-ups differed, time was entered as the number of weeks from baseline centered on the date of baseline collection. All models were fitted with a random intercept. Additionally, models were run with random slopes. However, the inclusion of random slopes did not significantly improve the fit of any model; thus, the results are reported for models with a random intercept only. The likelihood ratio test was used to compare model fits and determine which model fit the data best. Before analysis, models were checked for normality. The results indicated that depression and anxiety were not normally distributed, and therefore these outcomes were log-transformed before analysis. Because logs of zeroes cannot be computed, 1 was added to all scores of depression and anxiety before transformations were made.

To determine mediators of changes in depression symptoms from baseline to the end of the program, 1-1-1 multilevel multiple mediation models were estimated using the MLmed macro tool [28]. Models were estimated with random intercepts, fixed slopes, and uncorrelated residuals. The significance of indirect effects was estimated via Monte Carlo 95% CIs based on 10,000 resamples. Variables for each of the six identified lifestyle areas (healthy eating, physical activity, reducing alcohol intake, improving sleep, healthy relationships and social connection, and stress management) were tested as possible mediators. Additionally, to account for potential changes due to fluctuating responses to COVID-19, COVID-19–related distress was also tested as a potential mediator. To keep models parsimonious, the three variables with the strongest within-person correlations [29] with each of the outcomes were included in a single multiple mediation model. Within-person correlations between variables are provided in Multimedia Appendix 2.

There were large amounts of missing data. Only six participants provided data at the fourth follow-up, and therefore data from this time point were excluded from analyses. Of the remaining five assessments, 43 participants only completed the baseline assessment, 18 participants completed two assessments, 8 participants completed three assessments, 15 participants completed four assessments, and 20 participants completed all five assessments. Data were available for 106 participants at baseline, 51 at the first follow-up, 42 at the second follow-up, 36 at the third follow-up, and 31 at the fifth follow-up. One-way analysis of variance and Kruskal–Wallis tests were used to determine if there was a significant difference in study variables at baseline between participants with complete data and those missing data at follow-up. Missing value analysis indicated no significant differences in symptoms of depression, anxiety, or stress, or in demographics between participants with different amounts of missing data, indicating that the data were missing completely at random. Missing data for symptoms of depression, anxiety, and stress were handled using the full information maximum-likelihood estimation method. This has been shown to reduce parameter errors compared to ad hoc approaches to handling missing data, and provides appropriate standard errors of estimations, even with substantial amounts of missing data and small sample sizes [30,31].

For the mediation analysis, 31 participants completed both the baseline and follow-up assessments. Results indicated that participants who completed both baseline and follow-up assessments reported significantly less COVID-19–related distress at baseline than participants who did not complete the follow-up assessment, indicating that the data were not missing completely at random. Because multilevel models cannot handle missing data for independent variables using maximum likelihood, missing values for mediator variables were imputed using multiple imputation. A total of 10 imputed data sets were created using Markov Chain Monte Carlo with predictive mean matching. Imputation models included all mediators and outcomes as predictors of missing values. Results from the multiple imputed data sets were combined using Rubin’s Rule [32].

Attendance at the daily sessions ranged from 174 to 240 participants in the first week of the program, 63-84 at week 16, and 21-24 in the final week of the program. The number of responses to weekly feedback ranged from 16 to 90 participants. The average weekly attendance for participants who responded to the weekly feedback questionnaires ranged between 2.38 and 3.22 sessions. Participants reported high levels of satisfaction with the sessions during the first 13 weeks of the study (81.0% and 97.7% of weekly respondents agreeing or strongly agreeing that they were satisfied with the sessions they attended that week), and found the sessions they attended helpful for their mental health and well-being (80.9% and 94.2% of weekly respondents agreeing or strongly agreeing). Increased connection with colleagues was moderately endorsed (46.1% and 94.1% of weekly respondents agreed or strongly agreed).

A total of 106 participants completed the baseline questionnaire. At baseline, the average age of the 106 participants was 47.34 (SD 11.82) years and 92 (86.8%) identified as female. Overall, 79 (74.5%) participants were administrative or executive staff and 26 (24.5%) were teaching or research staff. The majority (70/106, 66.0%) were full-time staff. At baseline, 22.9%, 15.2%, and 25.2% of participants reported moderate-to-severe symptoms of depression, anxiety, and stress, respectively.

Results from the multilevel linear growth models are shown in Table 2. The results showed that symptoms of depression, anxiety, and stress all followed a cubic trajectory decreasing from baseline to June 8, increasing from June 8 to July 31 (corresponding with a return to strict lockdown), and decreasing from July 31 to November 27 as lockdown restrictions were eased (Figures 2-4).

Results from the repeated-measures correlation (Multimedia Appendix 2) showed that COVID-19–related distress, self-compassion, and social connectedness had the strongest within-person correlations with symptoms of depression. The results from the mediation model for depression are presented in Table 3, showing a significant negative indirect effect of time through COVID-19–related distress on depression. There were no significant indirect effects through self-compassion or social connectedness. When controlling for mediators, there was no direct effect of time on symptoms of depression, indicating that changes in symptoms of depression were mediated by changes in COVID-19–related distress.

Repeated-measures correlation showed that COVID-19–related distress, self-compassion, and sleep quality had the strongest within-person correlations with symptoms of anxiety. Results from the mediation analysis for anxiety (Table 4) showed that there was also a significant negative indirect effect of time through COVID-19–related distress. There were no significant indirect effects through self-compassion or sleep quality. When controlling for mediators, there was no direct effect of time on symptoms of anxiety, indicating that changes in symptoms of anxiety were mediated by changes in COVID-19–related distress.

Repeated-measures correlation showed that COVID-19–related distress, self-compassion, and diet quality had the strongest within-person correlations with stress. Results for the mediation analysis for stress (Table 5) showed that there was a significant indirect effect of time through self-compassion. There were no significant indirect effects through COVID-19–related distress or diet quality. When controlling for mediators, there was no direct effect of time, indicating that changes in stress were mediated by changes in self-compassion.

This study examined the acceptability, outcomes, and potential mediators of daily, evidence-informed, online microinterventions to promote the mental health of staff of a tertiary education institution during the COVID-19 pandemic. Regarding program acceptability, results from weekly questionnaires showed that the vast majority of participants were satisfied with the program content, and believed that the content was helpful for their mental health and well-being. Additionally, qualitative interviews showed that participants had positive perceptions about the program’s structure and content, believed it supported their mental health and well-being, helped them implement positive behavioral change, and made them feel more socially connected and supported by the university. Results from the longitudinal analysis of mental health symptoms showed that participants had moderate-to-severe symptoms of depression, anxiety, and stress at baseline, and that symptoms declined over the first 7 weeks of the intervention. Additionally, although participants’ symptoms of depression, anxiety, and stress increased with the commencement of a second, stricter lockdown period, symptoms did not return to the levels reported at the beginning of the initial COVID-19 lockdown. This suggests that participation in the “VU Elevenses” program may have protected against worsening symptoms of depression, anxiety, and stress associated with mass lockdown, fears of an outbreak, and economic recession [33]. Additionally, results from the multilevel mediation analysis suggested that reductions in depression and anxiety were explained by reductions in COVID-19–related distress, and that participants who experienced greater increases in self-compassion over the study experienced greater reductions in symptoms of stress.

The moderate to severe symptoms of depression, anxiety, and stress reported by approximately 15%-25% of our participants at baseline are expected responses to the unknown and stressful circumstances of a health epidemic [2,34] and mandated self-isolation [35,36]. As stated by international mental health experts, a rise in these symptoms was predicted to occur during the COVID-19 pandemic, as well as an increase in the prevalence of diagnosed depression and anxiety disorders, self-harm behaviors, and suicide [34]. Indeed, the mental health impacts of the COVID-19 pandemic are apparent, with reported increases in the experience of psychological distress, symptoms of depression and anxiety, and incidence of mental disorders [37-39], and highlight the need for effective mental health supports during this time of crisis.

Results from this study indicate that a brief daily online intervention may be an acceptable way to provide mental health support during a time of crisis. Staff experienced the VU Elevenses program as a positive program during the COVID-19 pandemic, and the vast majority of staff agreed that they found the content helpful for their mental health and well-being. Similarly, high levels of satisfaction have been reported in multiple other online interventions during the COVID-19 pandemic [40,41], indicating that it may be viable to deliver mental health interventions online. Additionally, many participants agreed that they felt more connected with their peers and the university due to the VU Elevenses program. It is promising that our program, delivered during the period of mandated self-isolation, increased the sense of connection with colleagues. An adverse consequence of COVID-19 restrictions is the impact of social isolation and loneliness on mental health and well-being [35,36]. This is likely a consequence of the intervention being delivered in a group setting and allowing participants to connect with their peers daily, which can increase perceptions of social support during social isolation [42]. Additionally, by delivering the intervention in groups, the program may have normalized any difficulties staff may have been experiencing during the lockdown.

The longitudinal changes in mental health reported are similar to those reported in previous studies. Several studies reported worse mental health status at the beginning of the COVID-19 pandemic compared to prepandemic levels; however, mental health improved during the early months of the pandemic [43-45]. This may be because the greatest burden of lockdowns occurs early, and people may be able to adapt to their new situation over time [46]. Other studies have also reported a nonlinear trajectory in mental health, with mental health problems reemerging after the initial improvement [44]. This is likely to follow increases in case numbers and reintroductions of local lockdown measures, contributing to worsening mental health [44]. Although previous online mental health interventions reported short-term improvements in mental health during the COVID-19 pandemic [47-49], the results from this study indicate that these improvements may not be sustained long-term, especially if local cases began to rise and stricter restrictions are reintroduced. Therefore, rapid-response health promotion interventions need to be adapted to changing participant needs throughout a crisis period to be able to provide the most appropriate supports when they are required the most.

Results from the mediation analysis indicated that preintervention to postintervention changes in depression and anxiety were explained by changes in COVID-19–related distress. This suggests that the program may have provided participants with the strategies and skills to adapt their behaviors and regulate their emotions to reduce the distress associated with the pandemic. By contrast, changes in stress were explained by changes in self-compassion over time. The Elevenses intervention used several strategies to improve self-compassion, including self-compassion meditation, resetting expectations, and simple mindfulness strategies (Table 1). Self-compassion may promote a healthier and more adaptive style of thinking and decrease rumination [50], which may explain why it was associated with changes with stress over time.

In line with recommendations to share information on managing stress, depression, and anxiety using telehealth [2] or digital responses [34], our program delivered psychological and lifestyle strategies and was regularly accessed by a group of VU staff. The weekly evaluation demonstrated a high level of program acceptability and indicated that participants who engaged in the program perceived real benefits to their mental health and well-being. The uncontrolled intervention was offered each workday, and there was no requirement to attend each day, providing flexibility for individual circumstances.

The content of our program was evidence-informed and was grounded in the six identified lifestyle intervention areas for well-being [11], including stress management, physical activity, nutrition, reducing substance use, sleep, and relationships. These influence coping, resilience, and mental health generally as well as in the context of stressful experiences [34]. The program’s content can be delivered in any workplace that has practitioners or clinicians on staff, can facilitate access to allied health professionals, or source and share appropriate existing online resources and clinical content. Despite the strength of having a flexible and responsive approach to content, aspects of the program that were based on general clinical content such as self-compassion, relaxation, and mindfulness practices could readily be reproduced. In addition, our findings demonstrate that it is possible to develop and deliver an acceptable program of online microinterventions in a timely manner, utilizing simple and accessible low-cost online meeting platforms, offered to an entire workplace at one time point. Until May 15, 2020, Australia’s federal government response was focused on managing health and financial needs rather than on the mental health implications of the COVID-19 crisis [2,51]. Our program was able to address this gap within our workplace, as it was developed and launched within 4 weeks from when staff were required to work remotely, and delivered wholly online to adhere to physical distancing and self-isolation requirements. This program allowed for the rapid collection of feedback from participants as well as timely sharing among the research team to adapt to anticipated changes in social regulations and restrictions. To the best of our knowledge, this is the first live, brief, weekday mental health promotion program in a workplace during a crisis.

Several limitations should be considered in the interpretation of these results. First, participants who opted into the program and the evaluation, including the semistructured interviews, may be subject to self-selection bias. Participants who value mental health promotion initiatives may be more likely to participate. Despite our intention for this program to be a universal mental health promotion and prevention intervention, only 18% of invited staff participated in the program, indicating modest reach into the target population [52]. Indeed, we do not have data to determine if our sample represents the larger staff body. The gender disparity in participation in the research component (females: 80% research opt-in vs 60% overall VU staff) may reflect the well-documented gender differences in help-seeking behaviors related to mental health, or reflect the attitudinal or societal barriers reported by men in accessing mental health support [53]. This should be considered for future program marketing and content development. As this program was developed as a rapid mental health promotion response, we could not develop an exhaustive internal promotion and engagement strategy. Inviting staff to participate via a general announcement email list can be problematic as many staff may not read these emails in detail. Follow-up responses to the opt-in research component were lower than expected and it is possible that the length of the survey may have contributed to this low response rate. As an uncontrolled intervention without a comparison group, it is possible that the changes in psychological symptoms reflect natural adaptation to new routines and life circumstances over time. This is reflected in the results of the mediation models, which showed that changes in COVID-19–related distress explained changes in depression and anxiety from baseline to the study endpoint.

This study demonstrates the acceptability of a rapidly developed program of microinterventions, delivered in a timely manner utilizing a simple and accessible low-cost online meeting platform during a time of crisis when mental health is highly likely to deteriorate. Participation in the program may have protected against worsening symptoms of depression, anxiety, and stress associated with extended COVID-19 lockdown periods, and improved symptoms of stress by increasing self-compassion.