The COVID-19 pandemic continues to affect communities across the world in different ways. Globally, mental health deteriorated in early 2020, when countries across the world were implementing COVID-19 restrictions (e.g., lockdowns) to stop or slow virus transmission and infections. A meta-analysis encompassing 65 longitudinal studies, predominantly from Europe and North America, reported small but significant increases in mental health-related symptoms among the general population and for people with pre-existing physical health conditions (1). The increase in mental health symptoms in relation to pre-pandemic levels was higher in the early stages of the pandemic (March–April 2020) but has been reported to have almost returned to pre-pandemic levels in May–July 2020 (1).

Countries such as Australia that imposed hard lockdown procedures (e.g., movement restrictions, curfews, and school closures) were in a unique position compared to other countries by having low infection and death rates at the start of the pandemic (2). A report from the Australian Institute of Health and Welfare (AIHW) examined the indirect health effects of COVID-19 (2). Similar to the meta-analysis (1), this report (2) also identified an increase in psychological distress during the peak of the pandemic that reverted to pre-pandemic levels by April 2021, except for younger people aged 18–44 years and those experiencing severe psychological distress. Similar results were also reported for loneliness, with almost half of participants reporting feeling lonely for at least some of the time in April 2020 but the level dropping to ~10% by the end of May (2).

However, there is growing evidence that not everyone was affected the same way during the pandemic, with some people improving, some worsening, but most reporting no change in their behavior during the start of the pandemic. For example, individual differences in behavioral changes (e.g., physical activity, fast food consumption, alcohol intake, and screen use) during the pandemic were observed in the AIHW report (2). Investigating these changes at a group level during COVID-19 lockdowns can provide important insights for developing targeted health campaigns for specific cohorts. Two studies (3, 4) examined the different health profiles (e.g., mental health and substance use) in tertiary students before and during COVID-19 by using latent class analysis (LCA) and latent transition analysis (LTA). LCA is a statistical technique that helps establish unknown subgroups within a population, whereas LTA is the longitudinal version of LCA that allows individuals to transit between the groups (5, 6). For example, the USA study that examined student behavior found that mental health problems increased before and during COVID-19 lockdowns, whereas substance use, sexual behavior, physical inactivity, and food insecurity decreased during COVID-19 lockdowns (4). Similarly, a German study revealed five behavioral classes encompassing substance use (3). Students who occasionally used different substances seemed to change their behavior during the pandemic, with most of them stopping substance use, whereas non-consumers or regular smokers did not change their behavior (3).

Furthermore, two studies from Australia examined subjective change in people's behavior (e.g., physical activity) during the lockdown period in relation to both before (7) and after (8) the lockdown by using retrospective recall. For example, a study from Queensland reported a significant association between negative change in behavior (physical activity, sleep quality, smoking, and alcohol intake) and mental health symptoms (depression, anxiety, and stress) before and during the lockdown (7). Furthermore, the study by Bhoyroo et al. (8) revealed similar polarizing effects of COVID-19 during and 2 months after the initial lockdown in Western Australia in 2020. Similar to the AIHW report (2), these results suggest that while some individuals struggled to adapt to the lockdown, others indicated an improvement in their health during the lockdown period, while the majority of participants reported no change in their behavior during the lockdown period.

Another important, but far less studied, research area is the impact of health promotion campaigns and the public's health literacy skills during the time of COVID-19. According to the World Health Organization (WHO), health promotion is a process that enables individuals to improve and to be more in control over their own health (9). Health literacy, on the other hand, is often defined as individuals' ability to find, understand, and use health-related information to guide their decisions and actions regarding their health (10). Both are important skills for successful public health communication. During COVID-19, the number of health promotion campaigns, especially in health-related (e.g., hand washing, hygiene, exercise, and mental health), was increased in Australia (11) and around the world (12–15). While some studies examined the cost-effectiveness of government actions during the pandemic (16, 17) or explored the impact of strategies and policies implemented in different countries (16, 18, 19), or mitigation versus containment policies within a country (20) on the spread of COVID-19, very few studies have investigated the relationship between health promotion campaign recall and the change in people's behavior during the COVID-19 pandemic (8, 21). For example, Bhoyroo et al. (8) investigated how well residents of Western Australia were able to recall different health campaigns during the COVID-19 pandemic. Most commonly recalled health promotion messaging was related to washing hands or social distancing, whereas health promotion campaigns encompassing physical activity, nutrition, and mental wellbeing were less recalled but identified as required by respondents (8). Another study from the US reported a positive relationship between increased vaccine confidence and the participant's ability to recall health promotion campaigns, emphasizing the importance of memorizing the health promotion campaigns to change people's behavior (21).

The present study extends the study by Bhoyroo et al. (8) that explored behavioral changes (physical activity, diet, alcohol intake, and mental health symptoms) during and after the COVID-19 lockdown among the general population in Western Australia. However, far less is known about how these behavioral changes are linked to each other, or how people recovered from COVID-19 lockdowns. The present study aimed to examine these differences using an LTA approach to form behavioral profiles and investigate the transitions between these LTA profiles. Open-ended responses were used to enhance the understanding of these behavioral changes related to different LTA profiles. The objectives of this study were as follows:

This study was based on a survey research methodology design and reports on a subset of participants (n = 313) who completed all questions in the Health and Wellbeing study (n = 547) by Bhoyroo et al. (8) and therefore followed the same study procedures. Ethics approval was obtained from the institution's Human Research Ethics Committee (REF 2020-133F). Western Australian residents (aged 18 years and above) completed a cross-sectional online survey using Qualtrics (22), ~2 months after the lockdown (from mid-August to October 2020). During this single survey, participants were asked to retrospectively recall and assess their behavior during the COVID-19 lockdown (2 months earlier) and at present (after COVID-19 lockdown restrictions had been lifted). The priori sample size calculation using G^*Power (23) considering the group and timepoint comparisons (medium effect size = 0.31, α = 0.05, power = 0.80) yielded a minimum of 174 respondents. For a more detailed description of the study procedures, see Bhoyroo et al. (8).

The study by Bhoyroo et al. (8) revealed several individual changes in behavior and mental health during and after the COVID-19 lockdown. Similarly, drawing from the same sample as Bhoyroo et al. (8), Piggott et al. (24) more deeply investigated the association between physical activity, sedentary behavior, and mental health symptoms during the COVID-19 lockdown. Since the LTA offers a unique approach to explore the hidden subgroups that the previous conventional methodological approach (e.g., regression, within-subject analysis, or group comparisons) cannot detect, the present study expands the earlier research by Bhoyroo et al. (8) and Piggott et al. (24) by focusing on group behavior rather than individual- or population-level changes.

The pre-analysis revealed a large number of records of systematic missing values. In our subsample, participants who had more than 50% values missing were excluded, yielding a total of 313 responses.

Comparative analysis revealed that the subsample used in this study (n = 313) was statistically similar to the original sample (8) across all of the dimensions measured (Supplementary Table S1). Of note, those who were excluded from this study were younger [t (467.63) = −3.45, p < 0.001], more likely to be born in Australia [χ2(1) = 6.038, p = 0.014], and consumed more fast food [χ2(1) = 8.125, p = 0.004] and alcohol [χ2(1) = 5.106, p = 0.024] after lockdown finished (Supplementary Table S1).

Most respondents in this study were female (76.1%) and middle-aged (M = 50.1 years, SD = 15.7 years) and had a university degree (69.7%). Approximately half of the participants had an annual household income of at least AU$100,000 (50.6%) or were born in Australia (61.0%). Less than two percent identified themselves as indigenous.

The behavioral characteristics during and after the initial lockdown period are presented in Table 2. People reported being significantly more inactive, more time spent in screen time activity, drank more alcohol, had more severe mental health symptoms, and felt lonelier during the lockdown. There was no significant change in eating fast food (p = 0.668) with ~40% consuming fast food at least once a week at both timepoints. Since no significant change was detected, fast food consumption was not included as a variable in the LTA models.

To the best of our knowledge, this is the first study that uses LTA to investigate the behavioral changes in the general population during and after COVID-19 lockdown, whereas previous LTA studies have investigated the change either before and during (3, 4), or before and after (47) COVID-19 lockdown. Using validated and widely utilized psycho-social measures, LTA analysis identified five latent classes with unique combinations of physical activity, leisure screen time use, alcohol consumption, loneliness, and mental distress during COVID lockdown. Two classes were characterized by inactivity, and three classes were characterized as active. The transition between LTA classes post-COVID-19 lockdown provided important insights for protective and risk factors for maintaining health and wellbeing during pandemic conditions, extending the research by Bhoyroo et al. (8) and Piggott et al. (24).

Since the LTA can identify subpopulations based on the behavioral profiles, it can be useful for informing diverse targeted strategies to effectively plan future health promotion campaigns and procedures in future. For example, some of the topics that were identified in the studies by Bhoyroo et al. (8) and Piggott et al. (24) also emerged in the present study as being associated with specific subgroups. For instance, Piggott et al. (24) reported that some people were “challenged to stay motivated and to find new ways to exercise in general or lacked motivation overall,” which is most likely be associated with inactive groups from the present study. Similarly, the present study revealed detailed information of the different characteristics of individuals that reported “no change” in Bhoyroo et al. (8) study—that is, individuals who “remain” in their group in present study. Group-level analysis, such as LTA, reveals different type of trajectories of wellbeing and health behavior that conventional sample- or population-level studies fail to detect. By expanding the methodological approach in COVID-19 studies, we gained valuable information of how different groups within the population are experiencing the health and social impacts of COVID-19. This evidence can inform the reduction in health inequality within society by creating more efficient health responses to future health crises.

Confirming our aim of the study, parts of the community reacted differently in how they managed with the lockdown restrictions but also in their recovery. Being active and not feeling lonely helped people be resilient to the stressful event of COVID-19 lockdown. Active and Happy and Inactive, Distressed and Lonely classes were distinctly different, emphasizing the importance of physical activity and social support for recovering from COVID-19 lockdown. For example, people who were Active and Happy reported higher resilience, sense of control, and reported less existing mental health issues, and fewer changes in their social relations during the lockdown than people who were inactive or lonely. This finding is supported by a large longitudinal study conducted in UK that found loneliness and decreased physical activity were risk factors for worsening mental health during the pandemic among people over 50 (48). Similarly, lower resilience, along with COVID-19-related worries, were reported to moderate the relationship between COVID-19-related loneliness and sleep problems among older population in Iran (49). These findings are highlighting the different needs and behaviors within the population, and therefore the need of more diversified ways to communicate with different subgroups of the public. In fact, Hyland-Wood et al. (50) highlighted the importance of tailored public health messaging for diverse audience in their article encompassing key characteristics and recommendations for effective governmental communication during crisis management, such as COVID-19. As an example, Hyland-Wood et al. (50) listed, e.g., people with disabilities, language barriers, or cultural differences, or age-sensitive public health messaging. Failing to identify some of these subgroups may lead to increased health risks within the community, like spreading the COVID-19 (50).

Our results also examined the impacts of health campaigns during COVID-19. First, the community education programs about the importance of hand hygiene and social distancing were the most recalled and effective health campaigns across all the classes which suggests that this multi-pronged strategy was well-received and successful. However, our study also revealed an overall increase in alcohol intake, loneliness, and mental distress along with decrease in physical activity, yet participants failed to recall these health campaigns. In general, most individuals reported being worried or scared of the unknown circumstances related to COVID-19 or being negatively affected by how the media was reporting COVID-19-related news suggesting there is the need for a more optimistic outlook to the public when advocating the public health messages. This is supported by many people across the groups requesting better access to the local state COVID-19 information and other studies that have discovered that trust in government or other institutes during COVID-19 are associated with improved mental health, sense of control, or lower COVID-related stressors (51–59). For example, Roccatto et al. (56) discovered that participants' perceived control over their lives mediated the association between political trust, mental health, and collective angst and anger. Similarly, Hyland-Wood et al. (50) emphasized the importance of developing and maintaining the public trust by providing clear and transparent public health messaging via trusted platforms (e.g., governmental website) to achieve a long-term success in crisis communication. In future, governments should pay attention to their public health messaging, so it is comprehensive and up to date. It is highly encouraged that the content should also convey positive and assuring, yet firm, information to enhance public's mental health and sense of control during uncertain times. The public health message should focus on emphasizing the sense of togetherness, along with practical examples of how to cope during COVID-19 lockdown (58, 60).

Second, the need of public health messaging was more essential among people who were either inactive or were in the Transition class. For example, people who were Inactive, Distressed and Lonely seemed to struggle the most but rarely managed to recall any health promotion campaigns nor made changes in their behavior, indicating they are a very challenging group to engage with. On the other hand, people who remained Inactive and Lonely were able to recall most health promotion campaigns, yet they were the most fearful class. Both classes that remained inactive were also the most frustrated toward the governmental restrictions, and more prone to raise mental health concerns, feelings of lack of control, and COVID-19-related fears, signifying the need of specific and targeted health promotion messaging. These class behaviors confirm the need for public messaging to be clear and coherent across the governmental and health policies since conflicting communication between these providers has been shown to lead to greater confusion, uncertainty, and fear among the public (61). Additionally, weaker trust in government's response to pandemic has been linked to greater anxiety and lack of control (62), which also enhances the likelihood of conspiracy theories (58, 62, 63).

Many in the inactive and Transition classes had the lowest level of feeling in control during the lockdown in comparison with active classes during the lockdown. This difference also remained in Inactive, Distressed and Lonely class after lockdown finished, raising concerns since the contemporary health campaigns did not seem to reach nor work within this class. Similarly, as many people who transitioned between the classes after lockdown were generally young, the public health campaigns should consider social-demographic differences when designing the content and distribution of the public health messaging. By investigating and understanding the differences between these class characteristics, we can inform public health messaging to be more aligned with how our target audience consumes such information, and from sources they are more likely to engage with. Future studies will be required to help tease out how best to provide these different sections of the community with the knowledge, skills, and support they need to adapt during crisis and return to a balanced level of mental wellbeing post-crisis. Given the strong association between physical activity and psychological wellbeing (64, 65) and the way this relationship has been observed during COVID lockdown (24), the government and health agencies should provide clear and specific messages to the public about the benefits of physical activity and the types of exercises they can do at home or outside when access to the normal gym and other sport activities are restricted.

The COVID-19 lockdown, which dramatically altered the community's normal social function, resulted in significant changes in people's behavior and mental wellbeing, both during and after the lockdown period. While many in the community demonstrated a resilience to the restrictions and maintained a physically active and healthy lifestyle, many reported more negative behaviors. LTA identified five different classes within the community based on their self-reported physical activity levels, leisure screen time, loneliness, and psychological distress. While some people transitioned from poorer states of health and wellbeing following the lifting of restrictions, others remained inactive, lonely, and, in some cases, distressed. In future, governments need to focus attention to their media content and public health messaging by offering comprehensive and up-to-date information and make sure that the information is accessible across the population. Targeted health campaigns tailored to people who are most at risks are recommended. More positive and supportive public health policies are encouraged to help the community maintain their everyday routine and promote their mental health during possible future lockdowns.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by HREC at the University of Notre Dame Australia (Ref: 2020-133F). The patients/participants provided their written informed consent to participate in this study.

JC led the study and oversaw the funding arrangements, ethics application, and questionnaire design and data collection with PC. RB helped with ethics application, questionnaire design, data collection, undertook the original data cleaning, and scoring of instruments. KS undertook the analysis and wrote the first manuscript draft. All authors participated in the manuscript review process and approved the final manuscript.