Improved lifestyle is associated with improved depression, anxiety and well-being over time in UK healthcare professionals during the COVID-19 pandemic: insights from the CoPE-HCP cohort study

Patient and public involvement

A multidisciplinary team of HCPs (including mental health experts, behavioural change experts and those involved with well-being support) was consulted, and feedback was sought on the survey design. The input was also received from the Barts Education Academy on the design and delivery of the surveys.

Statistical analysis

Statistical analysis was performed using Stata V.17.0. At each time point, we calculated the prevalence of the mental health and well-being parameters using the validated cut-off and the proportions showing improvement in the different aspects of lifestyle (physical activity, cigarette smoking, alcohol consumption, vaping and healthy eating).

To estimate the overall change in lifestyle separately at baseline and follow-up, we developed a novel score that considered each behaviour and estimated the score based on the following: a positive score was given for a healthier lifestyle change and a negative score was given for a worsening of lifestyle. For example, physical activity was coded as +1.0, –1.0 and 0, corresponding to increased activity, decreased activity and no change, since the beginning of the pandemic. Similarly, a scoring of −1.0, +1.0 and 0 was assigned for increased, decreased and no change, respectively, for cigarette smoking, alcohol consumption and vaping. Participants who reported never smoking/drinking/vaping were given a score of zero to indicate no change. Healthy eating was assigned 0 for never being a healthy eater and no change, +1.0 for being able to eat healthy meals, −0.5 for trying to but not always managing to eat a healthy diet and −1.0 for not managing to eat a healthy diet. ‘Prefer not to say’ was classed as missing.

Scores were aggregated to produce separate lifestyle change scores at baseline and follow-up. A score was also calculated to assess lifestyle change from baseline to follow-up by subtracting the baseline score from the follow-up score.

Validated cut-offs were used for the mental health assessments. A score of 10 or higher on the GAD-7 and PHQ-9 indicates the presence of GAD and MDD, respectively.22 23 A score of 18 or below on the SWEMWBS indicates low mental well-being.24

Separate logistic regression models were conducted to relate the novel lifestyle score to the presence of MDD, GAD and low mental well-being at baseline and follow-up. Each model was adjusted for age, gender, education, relationship status, number of people living in the household, and currently diagnosed mental health and physical health condition.

Finally, separate linear regression models were then conducted to examine whether improved lifestyle scores (from baseline to follow-up) are associated with improved scores on the respective mental health measures (PHQ-9, GAD-7 and SWEMWBS). These models were adjusted for age, gender identity, education, relationship status, number of people living in the household, as well as current mental health and physical health conditions. The models were then adjusted for the aforementioned variables and baseline PHQ-9, GAD-7 or SWEMWBS score.

To help interpret the regression outputs, further plots were developed based on separate linear regression models where tertiles of lifestyle score were calculated to predict the change in PHQ-9, GAD-7 and SWEMWBS scores. These separate plots were adjusted for baseline PHQ-9, GAD-7 and SWEMWBS, respectively.

Demographic characteristics for cohort of HCPs (N=613)

A total of 613 HCPs with both baseline and follow-up data were included in the analysis (see figure 1 for flowchart). Online supplemental table 1 shows the baseline demographic characteristics of the sample. The median age group was 36–50 years. Most participants were of white ethnicity (80.9%) and self-identified as female gender (76.7%). Most HCPs were also married or living with a partner in the family (65.6%), and most lived with at least one other person in their household (84.7%). Lastly, the majority of HCPs had high educational attainment (86.8% had a bachelor’s degree, master’s degree or doctor of philosophy degree).

Request permission

Lifestyle change and prevalence of mental health outcomes in HCPs

At baseline, UK HCPs generally reported worsened health-risk behaviours compared with before the pandemic (table 1). A higher proportion became physically less active since the beginning of the pandemic (43.2% reported decreased activity vs 22.3% reported increased activity vs 32.6% reported no change), while the majority reported the same levels of alcohol consumption (35.4%). Most participants were never smokers (72.6%) and never vapers (94.9%). Out of those who consumed alcohol, relatively higher proportions reported consuming more (29.9%) rather than less (18.3%). In smokers and vapers, relatively more participants reported smoking more (4.9% smoking more vs 1.6% smoking less) and vaping more (1.5% vaping more vs 0.5% vaping less). At baseline, 23.2% of UK HCPs reported trying but not always managing to eat a healthy diet.

At follow-up assessment (approximately 4 months later), a relatively larger proportion compared with baseline reported being physically less active (55.6% at follow-up vs 43.2% at baseline) as compared with the beginning of the pandemic. A larger proportion at follow-up (25.4% at follow-up vs 18.3% at baseline) reported drinking less compared with the beginning of the pandemic, and there was relatively little change in smoking or vaping status. The percentage of UK HCPs trying but not always managing to eat a healthy diet increased at follow-up (28.4% at follow-up vs 23.2% at baseline).

The mean total lifestyle scores in the 613 UK HCPs were −0.11 (range=−4 to 3) at baseline and −0.20 (range=−4 to 3) at follow-up (see online supplemental table 2), indicating that, at both baseline and follow-up, the overall lifestyle of UK HCPs worsened as compared with the beginning of the pandemic. The decreased mean score at follow-up indicated that lifestyle worsened over the 4-month period. Regarding mental health, the prevalences of all outcomes increased from baseline to follow-up (table 1).

Cross-sectional logistic regression models relating the novel lifestyle score to the presence of mental health and well-being at baseline and follow-up in UK HCPs

There were consistent associations between the novel lifestyle score and the presence of MDD, GAD and low mental well-being (SWEMWBS) (table 2).

A whole unit increase in total lifestyle score was associated with a 40% reduction in the risk of MDD at baseline (adjusted (adj.) OR: 0.60, 95% CI: 0.51 to 0.70, p<0.001) and a 24% reduction at follow-up (adj. OR: 0.76, 95% CI: 0.66 to 0.87, p<0.001), a 26% reduction in the risk of GAD at baseline (adj. OR: 0.74, 95% CI: 0.62 to 0.87, p<0.001) and a 17% reduction at follow-up (adj. OR: 0.83, 95% CI: 0.71 to 0.97, p=0.017), and a 36% reduction in the risk of low mental well-being at baseline (adj. OR: 0.64, 95% CI: 0.53 to 0.77, p<0.001) and a 31% reduction at follow-up (adj. OR: 0.69, 95% CI: 0.58 to 0.83, p<0.001) (table 2).

Linear regression predicting the change in mental health scores with the change in lifestyle score from baseline to follow-up in UK HCPs

Separate linear regression models showed that a whole unit increase in lifestyle change was inversely associated with the change in PHQ-9 (adj. coefficient: −0.51, 95% CI: -0.73 to −0.30, p<0.001) and GAD-7 scores (adj. coefficient: −0.32, 95% CI: −0.53 to −0.10, p=0.004), and associated with increased SWEMWBS scores (adj. coefficient: 0.37, 95% CI: 0.18 to 0.55, p<0.001) over the 4-month period (table 3). Collectively, this demonstrated that improved lifestyle over time is associated with improved mental health symptoms and well-being during the same period.

When categorised into tertiles of lifestyle change score, significant trends were observed between tertile and change in PHQ-9, GAD-7 and SWEMWBS scores (figure 2) (adjusted for baseline PHQ-9, GAD-7 and SWEMWBS scores, respectively; for unadjusted plots, see online supplemental figure 1).

From baseline to follow-up, for PHQ-9, there was a significant reduction in the change of PHQ-9 scores in those in the highest tertile of lifestyle change (adj. coefficient: −1.69, 95% CI: −2.47 to −0.92, p<0.001) but not in the middle tertile, compared with participants in the lowest tertile. For GAD-7, there was a significant reduction in the change of GAD-7 scores in those in both the middle (adj. coefficient: −0.76, 95% CI: −1.52 to −0.00, p=0.049) and highest (adj. coefficient: −1.00, 95% CI: −1.76 to −0.24, p=0.010) tertile, compared with participants in the lowest tertile. For SWEMWBS, there was a significant increase in the change of SWEMWBS scores in those in the highest tertile (adj. coefficient: 0.89, 95% CI: 0.36 to 1.41, p=0.001) but not in the middle tertile, compared with participants in the lowest tertile.

Request permission

Main findings

This cohort study indicated that the overall lifestyle in HCPs, consisting of changes to physical activity, cigarette smoking, alcohol consumption, vaping and healthy eating, had deteriorated compared with the beginning of the pandemic and worsened at follow-up. It is also concerning to note that the prevalence of mental health outcomes increased during this period. The study showed reliable associations between improved lifestyle and reduced risk of mental health issues in UK HCPs during the COVID-19 pandemic: a whole unit increase in lifestyle change was significantly associated with reductions in the odds of having depression, anxiety and low well-being at baseline and follow-up. More importantly, and to our knowledge for the first time, we demonstrated that improved lifestyle over the 4-month period is significantly associated with improved symptoms of validated depression, anxiety and well-being measures in HCPs during the COVID-19 pandemic (figure 3), regardless of their respective baseline mental health score. This is valuable evidence showing that lifestyle, as a modifiable factor, can improve mental health in UK HCPs and could be a recommended intervention, in addition to dedicated psychological support, during the current and future pandemics.

Request permission

Our findings are consistent with previous work in the general population (and more so in HCPs) during the pandemic.17 In UK health and social care workers, substance use was found to increase and exercise was found to decrease approximately 6 months after baseline assessment during the pandemic.19 McFadden and colleagues also found that well-being was consistently associated with substance use and exercise at both time points. Elsewhere, Vietnamese HCPs whose smoking and alcohol consumption was unchanged or increased had significantly increased risk of having depression or anxiety, compared with HCPs whose smoking and alcohol consumption stopped, decreased or never consumed.5 Similar observations were made for physical activity and a healthy diet but in the reverse direction.5 Our study extends this work considerably by demonstrating associations between improved lifestyle and improved mental health and well-being scores over time.

There are several strengths to this study. First, we examined the associations between lifestyle and various domains of mental health, namely, depression, anxiety and a global indicator of well-being. The value of this is that we can differentiate the relevance of improved lifestyle to specific domains of mental health. Second, the inclusion of a relatively large sample size obtained at two separate points during the pandemic provides reliability in the observed associations. Finally, we obtained changes in mental health over a 4-month period, as opposed to snapshot associations.

Limitations

Despite these strengths, there are some limitations that must be acknowledged. First, most participants were white and female; therefore, our findings could be less applicable to those identifying as male and of other ethnicities. Second, we treated 'prefer not to say' lifestyle responses as missing, but this may underestimate the potential change in behaviours not disclosed by HCPs due to their occupation. Nevertheless, the number of responses for 'prefer not to say' are minimal relative to the sample size. Third, since we used a global score, the extent that improved mental health scores can be attributed to specific behaviours is uncertain and, unlike validated lifestyle measures such as the SMILE-C (The Short Multidimensional Inventory Lifestyle Evaluation during Confinement),25 we have not accounted for lifestyle domains such as sleep pattern and social support. We expect that either physical activity or alcohol consumption is driving the associations because these behaviours showed a greater number of participants reporting change compared with before the pandemic. Finally, since our lifestyle measures are self-reported, the perceptions of lifestyle change could be influenced by the adverse mental health brought on by the pandemic or otherwise.

Implications

By demonstrating robust evidence for the association between improved lifestyle and improved depression, anxiety and well-being symptoms over time, we can set the groundwork for a lifestyle-based intervention to help mitigate the mental health impact of the pandemic in HCPs.

It would be beneficial to understand the impact of specific lifestyle changes on different domains of mental health in HCPs stratified by their current lifestyle. For example, while increased physical activity (or sustained physical activity in those who are physically active) is likely to improve (or maintain) mental health in HCPs, there will be other lifestyle aspects relevant only to those individuals engaging in that specific lifestyle (e.g. an intervention focusing on reducing alcohol consumption may only be relevant to those who are heavy drinkers). Drawing on this, there are underlying groups that require further study. In this analysis, those who never drank alcohol and those who consume the same as before are categorised as ‘no change’. Moreover, we expect that those who drink less than before the pandemic could represent those who continue to be heavy drinkers.

It is also unclear what the relevance of different aspects of lifestyle change is to different domains of mental health. It would be valuable to understand whether improving diet, for example, is as effective as physical activity in mitigating anxiety, or whether improved diet is more relevant to domains such as depression. Future research implementing randomised controlled trials are required to test the efficacy of specified and tailored lifestyle-based interventions to different subgroups of HCPs. A randomised controlled trial will also determine the extent of a bidirectional relationship between lifestyle change and mental health. We hope future researchers can draw on this study to improve the design of future studies evaluating the extent that improvements to lifestyle mitigate adverse mental health in pandemic settings.

Separately, a potential link between pandemic-influenced lifestyle changes and physical health conditions may exist within HCPs, in addition to the potential impact of chronic psychological stress on physical health. This has been recently hypothesised within the general population regarding lockdown measures.26 Since the lifestyle and mental health of UK HCPs appears to have worsened during this 4-month period of the pandemic, further research is needed to examine the longer-term impact on physical health.