key: cord-1049948-dm39opvg
authors: Frank, P.; Iob, E.; Steptoe, A.; Fancourt, D.
title: Trajectories of depressive symptoms among vulnerable groups in the UK during the COVID-19 pandemic
date: 2020-06-11
journal: nan
DOI: 10.1101/2020.06.09.20126300
sha: d1b4d7f62f86a311e8bb3f5ff1b35a6a84780b43
doc_id: 1049948
cord_uid: dm39opvg

Objective: The coronavirus disease 2019 (COVID-19) pandemic has affected many aspects of the human condition, including mental health and psychological wellbeing. This study examined trajectories of depressive symptoms (DST) over time among vulnerable individuals in the UK during the COVID-19 pandemic. Methods: The sample consisted of 51,417 adults recruited from the COVID-19 Social Study. Depressive symptoms were measured on seven occasions (21st March - 2nd April), using the Patient Health Questionnaire (PHQ-9). Sociodemographic vulnerabilities included non-white ethnic background, low socio-economic position (SEP), and type of work (keyworker versus no keyworker). Health-related and psychosocial vulnerabilities included pre-existing physical and mental health conditions, experience of psychological and/or physical abuse, and low social support. Group-based DST were derived using latent growth mixture modelling and multivariate logistic regression models were fitted to examine the association between these vulnerabilities and DSTs. Model estimates were adjusted for age, sex, and suspected COVID-19 diagnosis. Results: Three DSTs were identified: low [N=30,850 (60%)] moderate [N=14,911 (29%)], and severe [N=5,656 (11%)] depressive symptoms. DSTs were relatively stable across the first 6 weeks of lockdown. After adjusting for covariates, experiences of physical/psychological abuse (OR 13.16, 95% CI 12.95-13.37), pre-existing mental health conditions (OR 13.00 95% CI 12.87-13.109), pre-existing physical health conditions (OR 3.41, 95% CI 3.29-3.54), low social support (OR 12.72, 95% CI 12.57-12.86), and low SEP (OR 5.22, 95% CI 5.08-5.36) were significantly associated with the severe DST. No significant association was found for ethnicity (OR 1.07, 95% 0.85-1.28). Participants with key worker roles were less likely to experience severe depressive symptoms (OR 0.66, 95% 0.53-0.80). Similar but smaller patterns of associations were found for the moderate DST. Conclusions: People with psychosocial and health-related risk factors, as well as those with low SEP seem to be most vulnerable to experiencing moderate or severe depressive symptoms during the COVID-19 pandemic.

Due to the rapid spread of the coronavirus disease 2019 (COVID- 19) , large numbers of people across the UK have been urged to stay at home ("lockdown") for potentially significant periods of time. It is already evident that the COVID-19 pandemic and its related containment measures have profound implications for many aspects of society 1 . Up to now, much focus has been placed on investigating the incidence and mortality rates of COVID- 19 , the pathogenesis of the virus, its adverse effects on physical health, and its growing impact on the global economy. But there is growing awareness of the implications of the pandemic for mental health 2 . The consequences of the COVID-19 pandemic are likely to occur against the backcloth of elevated prevalence of mental health problems in the UK. The widespread experience of COVID-19-related stressors (e.g. loss of employment, illness or death of a relative), reduced access to mental health services, and frequent concerns about mental and physical health in the general population highlight the importance of identifying who is most at risk and how their experiences are evolving as the pandemic continues.

Previous studies on former epidemics and pandemics (e.g. the severe acute respiratory syndrome epidemic (SARS) in 2003 or the 2014 Ebola outbreak) suggest a plausible increase in mental health problems across the wider population, including symptoms of post-traumatic stress, confusion, anger and depression 2,3 . However, it is unclear how mental health problems are manifesting in more vulnerable groups during the current pandemic, such as in people with pre-existing physical and/or mental health conditions, people with experiences of physical and psychological abuse, and people of lower socio-economic position (SEP). Prior to the COVID-19 outbreak, the prevalence of depression was estimated at 4-5% in the general population 4,5 , with considerably higher rates observed in individuals with distinct health-related, sociodemographic and psychosocial vulnerabilities 6 . For example, the prevalence of depression in people with chronic obstructive pulmonary disease (COPD) has been placed at 27% 7 , type II diabetes 18-20% 8,9 , myocardial infarction 20% 10 , cancer 13-17% 11,12 , and stroke patients at 29-33% 13, 14 . Furthermore, meta-analytic evidence indicates significantly elevated rates of depression in women who experienced intimate partner violence (27%) 15 , and almost a two-fold increased risk of depression in people with lower SEP 16 .

Evidence already suggests that pre-existing health inequalities may be strongly reflected in the current COVID-19 pandemic. For example, recent mortality statistics indicate that the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), the virus that causes COVID-19, may be particularly detrimental to members of Black, Asian and minority ethnic (BAME) groups, with approximately 19% of UK COVID-19-related hospital deaths occurring in members of the BAME community 17 . In addition, key workers (e.g. frontline healthcare and . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted June 11, 2020. . https://doi.org/10.1101/2020.06.09.20126300 doi: medRxiv preprint social care staff) may be particularly vulnerable to developing symptoms of emotional distress. According to a study of the psychosocial effects associated with SARS in a sample of 510 hospital workers, 29% of health-care staff reported elevated symptoms of emotional distress 18 . The COVID-19 pandemic and related containment measures are also likely to accentuate social isolation and feelings of loneliness 1 . Notably, these factors are themselves strongly associated with the incidence, severity and progression of negative mental health outcomes, including depression, anxiety-related disorders, and suicide 19 . Other COVID-19 related stressors such as loss of employment, financial hardship, overcrowded households, and diminished access to social support networks may further contribute to disease burden, nationally and globally, by increasing levels of distress and reducing social opportunities relevant to both mental and physical wellbeing 20 .

Given the possibility that specific sociodemographic, psychosocial, and health-related characteristics may make some people particularly vulnerable to poorer mental health, an immediate research priority is to investigate the psychological wellbeing in these groups, and to understand how it evolves over time as the pandemic continues.

Therefore, the aim of the present study was to explore trajectories of depressive symptoms (DST) among vulnerable individuals in the UK during the COVID-19 pandemic.

The COVID-19 Social Study is an ongoing large panel study of adults aged 18 years and older residing in the UK.

The study was established on 21 st March 2020, using online weekly data collection to explore the psychological and social experiences of adults during the COVID-19 pandemic. The sample was recruited using three primary approaches. First, snowballing was used, including promoting the study through existing networks and mailing lists (including large databases of adults who had previously consented to be involved in health research across the UK), print and digital media coverage, and social media. Second, more targeted recruitment was undertaken focusing on (i) individuals from a low-income background, (ii) individuals with no or few educational qualifications, and (iii) individuals who were unemployed. Third, the study was promoted via partnerships with third sector organisations to vulnerable groups, including adults with pre-existing mental health conditions, older adults, carers, and people experiencing domestic violence or abuse. The study was approved by the UCL Research Ethics Committee [12467/005] and all participants gave informed consent.

The present analysis focused on participants recruited between 21 st March and 4 th May 2020. All duplicate email addresses were removed prior to the sample size being calculated and only individuals who provided fully completed interviews on at least one wave were included. This resulted in a final analytical sample of 51,417 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted June 11, 2020. . https://doi.org/10.1101/2020.06.09.20126300 doi: medRxiv preprint participants. Although participant selection was based on a multi-stage non-random sampling approach, the sample was well-stratified across sociodemographic characteristics and all data were weighted to the proportions of gender, age, ethnicity, education and country of living obtained from the Office for National Statistics (ONS, 2018).

Depressive symptoms were measured using the Patient Health Questionnaire (PHQ-9), a validated screening tool for diagnosing depression in primary care 21 . The questionnaire involves nine items asking about the frequency of experiencing common symptoms of major depressive disorder during the past week 22 . Each item was answered on a 4-point Likert scale, ranging from "not at all" to "nearly every day". Higher overall scores indicate more depressive symptoms, with scores of 0-4 suggesting minimal depression, 5-9 mild depression, 10-14 moderate depression, 15-19 moderately severe depression, and scores of 20-27 denoting severe depression 23 .

Pre-existing physical health conditions were measured by asking participants whether they had been clinically diagnosed with (i) diabetes, (ii) high blood pressure, (iii) heart disease, (iv) lung disease (e.g. asthma or COPD),

(v) cancer, and (vi) another clinically diagnosed chronic physical health condition. Each item was answered on a binary response scale (yes = 1, no = 0). For the purposes of the present analyses, a binary score was computed, distinguishing between the presence (at least one pre-existing physical health condition) and absence (no preexisting physical health condition) of a clinically diagnosed physical illness.

Pre-existing mental health conditions were assessed by asking participants whether they had a clinical diagnosis of (i) depression, (2) anxiety, and/or (3) other mental health condition. Items were ranked on a binary response scale (yes = 1, no = 0). A summary score was computed, categorising responses into the presence (at least one clinically diagnosed health condition) versus absence (no pre-existing mental health conditions) of a pre-existing mental health condition.

. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 11, 2020. . https://doi.org/10.1101/2020.06.09.20126300 doi: medRxiv preprint

Participants were asked to provide information about their ethnic backgrounds. Ethnicity was categorised into two 

First, a continuous latent socio-economic position (SEP) index was computed using five indicators of SEP: (i) household income, (ii) employment status, (iii) education, (iv) household tenure, and (v) household overcrowding.

Income was measured by self-reported annual household income. Responses were categorised as "< £16,000 a

year", "£16,000-£29,999 a year", "£30,000-£59,999 a year", "£60,000-89,999 a year", and "£90,000 or more per year". Employment was measured by asking participants to provide information about their employment status.

Responses were grouped as "employed", "inactive", and "unemployed". Education was assessed in reference to participants' highest educational qualification (postgraduate, undergraduate, A-level or vocational, and GCSE or lower). Household tenure was categorised as "owned outright" versus "owned with mortgage / shared ownership" versus "rented". Overcrowded households were defined as households with < 1 room per occupant. Next, we computed a binary SEP index variable by first dividing the continuous latent SEP variable into quartiles, and then categorising the highest quartile as "low SEP".

Participants were asked whether they had currently fulfilled any of the government's identified key worker roles (yes = 1, no = 0). Key workers included people with jobs that were deemed essential during the pandemic and included those in health and social care, education and childcare, key public services, local and national government, public safety and national security, transport, as well as utility workers. People were only classed as keyworkers if their role involved them leaving the home to carry out this work during the lockdown.

Participants indicated whether during the last week they had been "physically harmed or hurt by someone else" or "bullied, controlled, intimidated, or psychologically hurt by someone else". Responses were rated on a 4-point Likert scale ranging from "not at all" to "nearly every day". A binary variable was created focusing on any response on either item that indicated any experience of abuse on at least one occasion (yes = 1, no = 0).

(II) Social support . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 11, 2020. . https://doi.org/10.1101/2020.06.09.20126300 doi: medRxiv preprint Social support was measured using an adapted version of the six-item short form of Perceived Social Support Questionnaire (F-SozU K-6) 24 . This includes six questions asking participants whether, during the past week, they had "experienced a lot of understanding and support from others", "a very close person whose help they can always count on", "people with whom they can spend time and do things together", "friends and family who will take care of them if they get sick", "people they can talk to without hesitation if they feel down", and whether they could "easily borrow something from neighbours or friends if needed". Each item is rated on a 5-point scale from "not true at all" to "very true", with higher scores indicating higher levels of perceived social support. Minor adaptations were made to the language in the scale to make it relevant to experiences during COVID-19 (see sTable 1 for a comparison of changes, Supplement). A sum score was computed by adding up the mean scores (across all available waves) of each social support question per individual and dividing it into quartiles. We subsequently derived a binary social support variable, defining "low social support" as scores in the lowest quartile.

Covariates included age, gender, and the presence/absence of a (suspected) COVID-19 diagnosis.

Group-based trajectories of depressive symptoms were estimated using latent growth mixture (LGM) modelling . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 11, 2020. . https://doi.org/10.1101/2020.06.09.20126300 doi: medRxiv preprint

The characteristics of the study participants at the first assessment are reported in Table 1 (see sTable 3 and sTable   4 for the weighted and unweighted descriptive statistics at each wave, Supplement). The weighted sample was 51% female, and 12% of participants had a non-white ethnic background. There was a higher proportion of participants in the oldest age groups (32%) compared with the youngest groups (18%). 60% of participants were employed and 22% were undertaking key worker roles. In contrast, around 40% of the sample was inactive or unemployed. The majority of participants had GCSE or A-level qualifications, and around 30% also had degree qualifications. There was a higher proportion of participants in the low-and medium-income groups compared with the highest ones. A total of 8% of participants lived in overcrowded households. Almost 40% of participants had a pre-existing physical illness, 20% reported having at least one mental health problem, and almost 1 in 3 participants had moderate or severe symptoms of depression at the first assessment [see sFigure 1 (Supplement) for the average levels of depressive symptoms throughout the study]. 11% had experienced psychological or physical abuse on at least one occasion since lockdown began.

The LGM analysis resulted in three distinct DSTs ( is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 11, 2020. 

This is the first study to investigate trajectories of depressive symptoms among vulnerable groups in the UK during the COVID-19 pandemic using data from a large longitudinal study. Our analysis revealed a number of key findings. First, trajectories of depressive symptoms in the sample were relatively constant across the first 6 weeks of lockdown. However, this does not imply that mental health has not been affected by the pandemic. Data from various surveys suggest that mental health worsened in the lead up to lockdown being introduced in the UK, leading to higher than usual levels of anxiety and depression 25-29 . Although our analysis does not claim to show prevalence due to the non-random nature of the sample, it is notable that the average scores presented here were substantially higher than PHQ-9 averages previously reported in population estimates in other high income countries such as the US 30 . Further, our results suggest that there was little improvement in depression in the first few weeks of lockdown.

Our findings also showed that individuals facing certain sociodemographic, psychosocial and health-related vulnerabilities were at heightened risk of experiencing moderate and severe depression during lockdown. The risk of both moderate and severe depressive symptoms was considerably higher among people experiencing abuse, low social support, in individuals with low SEP, and in those with pre-existing mental and physical health conditions. Notably, these associations were larger than might be expected from previous literature, particularly for psychosocial vulnerabilities. For example, meta-analytical results suggest that the odds of depression are 1.87(1.42;2.46) higher in people exposed to intimate partner violence as compared to non-exposed people 15 , and 0.74 (0.72;0.76) lower in people with high levels of social support compared with those with poor social support 31 . The relationships of abuse and low social support with severe depressive symptoms were more than 5 times larger in our study, with odds ratios of 13.2 for abuse and 12.7 for low social support. Likewise, previous metaanalyses have shown that low SEP and chronic physical health conditions such as cardiovascular disease increase the odds of depression by 1.81(1.57;2.10) and 1.75(1.36;2.26) respectively. Our data suggest that the odds of . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 11, 2020. . https://doi.org/10.1101/2020.06.09.20126300 doi: medRxiv preprint severe depressive symptoms were 3.41 times higher for people with chronic physical illnesses and 5.22 times higher in those facing socioeconomic disadvantage. These figures are particularly worrying considering likely increases in rates of abuse, domestic violence, and unemployment and diminished access to social networks during the current COVID-19 pandemic 3,32 . As such, the results presented here suggest that certain groups who usually experience higher odds of depression are at even greater risk during the current pandemic.

Notably, non-white ethnicity was related to higher depressive symptoms, but these results were explained by other sociodemographic characteristics, abuse and social support, as well as pre-existing physical or mental illnesses.

Thus, it appears that the higher prevalence of other socio-economic and psychosocial vulnerabilities amongst ethnic minority groups is a greater risk factor than ethnicity itself for depression during the pandemic.

Additionally, despite evidence suggesting that people with key worker roles such as healthcare workers are particularly susceptible to poor mental health outcomes during epidemics 18 , our results show that the risk of elevated depressive symptoms was similar in participants with and without key worker roles. However, our key worker measure was not limited to healthcare professions but also included other keyworker roles such as teachers and transport workers who might be experiencing different levels of work-related stress during the current pandemic. Further, it is possible that self-selection bias determined these findings, with only key workers who are psychologically coping with the current demands of their roles taking the time to participate in the research.

This study has a number of strengths, including the large sample size, the longitudinal study design, and the statistical methods employed to explore the link between specific vulnerabilities and DSTs. However, our findings need to be interpreted in light of several limitations. First, our sample, though well-stratified across sociodemographic characteristics and weighted to population proportions, was not random, and hence is not nationally representative. It is possible that the study inadvertently attracted individuals experiencing greater psychological distress during the pandemic, or individuals who are more engaged or interested in mental health. Hence, the results shown here are not presented as prevalence figures but are instead used to understand risk factors. Second, the data used in the present analyses are based on self-report measures, bearing the risk of self-report bias. Lastly, causality cannot be assumed since the study is observational and only provides information about the severity of depressive symptoms among vulnerable groups over time. As we lack data on individuals from prior to lockdown being brought in the UK, we are unable to identify if and how patterns of depressive symptoms during lockdown vary relative to individuals' usual mental health.

In conclusion, our analysis suggests that certain vulnerable groups are at particular risk of experiencing elevated depressive symptoms during the current COVID-19 pandemic, including people with pre-existing mental and . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 11, 2020. . physical health conditions, experience of physical/psychological abuse, low social support, and those with low SEP. These groups may be experiencing even greater risk than in ordinary circumstances. In contrast, key workers and individuals from ethnic minority groups were not more likely to report depressive symptomatology when other vulnerabilities were taken into account. These differential associations highlight the importance of developing strategies to identify vulnerable individuals, reallocate mental health services to those in need, and provide evidence-based treatments to alleviate depressive symptoms.

. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 11, 2020. . https://doi.org/10.1101/2020.06.09.20126300 doi: medRxiv preprint Note. Weighted descriptive statistics of the sample at the first assessment. a Quartiles were computed prior to weighting so actual percentages vary from 25% in the weighted sample.

. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted June 11, 2020. . https://doi.org/10.1101/2020.06.09.20126300 doi: medRxiv preprint . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted June 11, 2020. . https://doi.org/10.1101/2020.06.09.20126300 doi: medRxiv preprint Note. The odds ratios represent the risk of belonging to the Moderate or Severe Depressive symptom trajectory compared with the Low trajectory. All models were adjusted for sex, age, and COVID-19 diagnosis and weighted using survey weights.

. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted June 11, 2020. . https://doi.org/10.1101/2020.06.09.20126300 doi: medRxiv preprint

The Academy of Medical Sciences. Survey results: Understanding people's concerns about the mental health impacts of the COVID-19 pandemic

Multidisciplinary research priorities for the COVID-19 pandemic: a call for action for mental health science. The Lancet Psychiatry

The psychological impact of quarantine and how to reduce it: rapid review of the evidence

The prevalence and distribution of major depression in a national community sample: the National Comorbidity Survey. The American journal of psychiatry

Prevalence and incidence studies of mood disorders: a systematic review of the literature

The Routledge international handbook of psychosocial epidemiology: Routledge

Prevalence of depression in COPD: a systematic review and meta-analysis of controlled studies

The prevalence of co-morbid depression in adults with Type 2 diabetes: a systematic review and meta-analysis

The association between diabetes and an episode of depressive symptoms in the 2002 World Health Survey: an analysis of 231 797 individuals from 47 countries

Natural history, predictors and outcomes of depression after stroke: systematic review and meta-analysis

Frequency of depression after stroke: a systematic review of observational studies

Intimate partner violence against adult women and its association with major depressive disorder, depressive symptoms and postpartum depression: a systematic review and meta-analysis

Socioeconomic inequalities in depression: a meta-analysis

Office for National Statistics. Deaths involving COVID-19, England and Wales: deaths occurring in

Psychosocial effects of SARS on hospital staff: survey of a large tertiary care institution

Contribution of risk factors to excess mortality in isolated and lonely individuals: an analysis of data from the UK Biobank cohort study. The Lancet Public Health

Are we all in this together? Longitudinal assessment of cumulative adversities by socio-economic position in the first 3 weeks of lockdown in the UK

Measuring depression outcome with a brief self-report instrument: sensitivity to change of the Patient Health Questionnaire (PHQ-9)

American Psychiatric Association. Diagnostic and statistical manual of mental disorders (DSM-5®)

The PHQ-9: a new depression diagnostic and severity measure

When to Release the Lockdown? A Wellbeing Framework for Analysing Costs and Benefits

Survey results: Understanding people's concerns about the mental health impacts of the COVID-19 pandemic. 2020. MQ: Transforming Mental Health and the Academy of Medical Sciences

Britain's mood, measured weekly

Millions of UK adults have felt panicked, afraid and unprepared as a result of the coronavirus pandemic -new poll data reveal impact on mental health

Depression and anxiety spiked after lockdown announcement, coronavirus mental health study shows

Stability of the distribution of Patient Health Questionnaire-9 scores against age in the general population: data from the National Health and Nutrition Examination Survey