Background
Measures taken to reduce the spread of COVID-19 in the United Kingdom (UK), including a national ‘lockdown’ implemented on 23rd March 2020 [
1], have had profound effects on everyday life. Under this lockdown, all non-essential shops and other premises were closed and people were only permitted to leave their home to shop for basic necessities, to exercise once a day, for medical needs or for work if absolutely necessary. Concerns that lockdown measures could worsen existing socioeconomic and health inequalities by disproportionately affecting disadvantaged populations were expressed early on in the pandemic [
2‐
5]. Since then, evidence suggests these concerns were well founded [
6]. Determining who is more likely to have suffered adverse effects will be critical in guiding comprehensive and targeted support for those most in need.
Several studies have highlighted consequences of lockdown on physical and mental health in the UK [
7‐
15]. During lockdown, health behaviours such as diet, alcohol consumption, smoking, sleep and physical activity were subject to rapid change, with both increases and decreases observed. Many health behaviours have complex relationships with health, such as sleep quantity, where too little or too much sleep can reflect worse mental and physical health [
16]. Similarly, an increase in meals may be associated with obesity, but a decrease in meals may reflect hunger due to financial difficulties, and could lead to under-nutrition. Since the lockdown is likely to have differentially affected people’s lifestyle, there will be variation in whether lockdown has improved or worsened their overall health. Analysis of five British cohort studies suggested lockdown widened socioeconomic inequalities in sleep but did not affect inequalities in other health behaviours [
7]. Another study found that, relative to pre-pandemic levels, participants consumed less fruit and vegetables, did less exercise and increased alcohol over a 3 month period from April 2020 [
17]. In the same study, worsening health behaviours were associated with being younger and female. A study of smartphone-tracked physical activity highlighted a general decline in physical activity from pre-lockdown to during lockdown, with a greater decline in young adults compare to those aged over 65 but no effect of socioeconomic group on level of decline [
18].
Financial and employment changes were also prevalent during lockdown, due in part to the effective shutdown of a number of sectors, including hospitality and travel [
19,
20]. Initial UK evidence suggested younger people were more affected by loss of employment and income during lockdown [
21]. There was a clear social gradient for loss of income and employment, and difficulties accessing food and medicines, with individuals of lower socioeconomic position (SEP) more negatively affected [
22,
23].
Impact on individuals of the UK lockdown may also differ by other factors, including a history of adverse childhood experiences (ACEs), that is, stressful events in childhood such as abuse, neglect, and family dysfunction. Exposure to ACEs has been shown to be associated with worse health and health-related behaviours across adolescence and adulthood [
24‐
27], as well as with poor adulthood socioeconomic outcomes [
28]. Therefore, individuals with a history of ACEs may be less resilient to the damaging effects of periods of adversity such as lockdown. Evidence from 12 longitudinal studies in the UK has shown that those who were experiencing psychological distress pre-pandemic were more likely to experience economic disruptions during the pandemic [
29]. Whether ACEs in particular predict increased hardship during lockdown remains to be determined [
30].
Using data from the Avon Longitudinal Study of Parents and Children (ALSPAC), a UK prospective cohort study, collected during the COVID-19 pandemic [
31], we investigated associations of SEP and ACEs with health and financial consequences of the first UK lockdown (March–June 2020). We aimed to elucidate whether young adults with low SEP or previous exposure childhood adversity were differently affected by lockdown measures, in ways which could exacerbate existing health and social inequalities experienced by this population. We hypothesized that those with lower SEP and more exposure to ACEs would experience more detrimental change in their health-related behaviours, such as increased smoking, increased alcohol use and decreased exercise. For some health-related behaviours, both increases and decreases could be adverse – for example increased meal/snack frequency could be a risk factor for increasing weight, but decreased meal/snack frequency could be a result of economic hardship and be a risk factor for underweight. We also hypothesized that those with lower SEP and greater exposure to ACEs would be more likely to lose their job or be furloughed during lockdown, and face greater financial adversity than others.
Methods
Study population
The study used data from the Avon Longitudinal Study of Parents and Children (ALSPAC), an ongoing birth cohort study that recruited 14,541 pregnant women in Avon, UK with expected delivery dates between 1st April 1991 and 31st December 1992 [
32‐
34]. Mothers, children, and mother’s partners have been followed up using clinics, questionnaires, and linkage to routine data. Additional eligible cases were recruited to the study when the oldest participants were approximately 7 years old. Including these, the offspring cohort consists of 14,901 participants that were alive at 1 year of age. The website contains details of available data through a fully searchable data dictionary:
http://www.bristol.ac.uk/alspac/researchers/our-data/.
This study is based on 2710 offspring participants who responded between 26th May and 5th July 2020 to a questionnaire rapidly deployed early on during the COVID-19 pandemic [
31]. The questionnaire was developed and deployed using REDCap (Research Electronic Data CAPture tools), a secure web application for online data collection hosted at the University of Bristol [
35].
ACEs
ACE measures were derived from questions relating to multiple forms of ACEs reported by participants and their mothers at multiple timepoints from birth to 23 years of age via questionnaires. Full details are described elsewhere [
36]. Briefly, dichotomous indicators of exposure between 0 and 16 years were created for the ten ACEs included in the World Health Organization ACE international questionnaire [
37]. The ten ACEs we considered were:
1.
ever sexually abused or forced to perform sexual acts or touch someone in a sexual way (sexual abuse)
2.
adult in family was ever physically cruel towards or hurt the child (physical abuse)
3.
parent was ever emotionally cruel towards the child or often said hurtful/insulting things to the child (emotional abuse)
4.
child always felt excluded, misunderstood, or never important to family, parents never asked or never listened when child talked about their free time (emotional neglect)
5.
parent was a daily cannabis or any hard drug user or had an alcohol problem (parental substance abuse)
6.
parent was ever diagnosed with schizophrenia or hospitalized for a psychiatric problem or, during the first 18 years of the child’s life, parent had an eating disorder (bulimia or anorexia), used medication for depression or anxiety, attempted suicide, or scored above previously established cut-offs for depression (Edinburgh Postnatal Depression Scale (EPDS) > 12–13) (parental mental illness or suicide)
7.
parents were ever affected by physically cruel behaviour by partner or ever violent towards each other, including hitting, choking, strangling, beating, and shoving (violence between parents)
8.
parents separated or divorced (parental separation)
9.
child was a victim of bullying on a weekly basis (bullying)
10.
parent was convicted of a crime (parental criminal conviction)
Based on the sum of the dichotomous ACE constructs, each participant was given an ACE score (0–10 ACEs), which was categorized as 0, 1, 2–3, or 4+ ACEs for comparability with previous studies.
SEP
Socioeconomic position was indicated by occupational social class at age 23 (~ 4 years prior to the start of the pandemic), using the 3-class National Statistics Socio-economic classification (NS-SEC) [
38]:
1.
Higher managerial, administrative, and professional occupations (NS-SEC group 1)
2.
Intermediate occupations (NS-SEC group 2)
3.
Routine and manual occupations (NS-SEC group 3)
4.
Never worked and long-term unemployed (LTU)
This was derived from participant self-reports of occupation, business/industry and job responsibilities from a postal questionnaire administered at mean age 23 years. Responses were coded into eight NS-SEC classes, which were then collapsed into the above four classes (Supplementary Table
1). Participants were instructed to skip certain questions if they were “not engaged in any form of work”, so participants who skipped these questions only were assigned to the fourth class. Participants who indicated they were full-time students were excluded from analysis of SEP (Supplementary Fig.
1).
Outcomes
The term lockdown refers to the stay-at-home order made by the UK government on Monday 23rd March 2020. An online questionnaire deployed between 26th May and 5th July 2020 asked participants (mean age 27.8 years) to report their perceived change in several health-related behaviours compared to pre-lockdown. The question asked whether each of the following activities had decreased a lot, decreased a little, stayed the same, increased a little or increased a lot since lockdown: number of home-cooked meals eaten, number of meals eaten in a day, number of snacks eaten in a day, amount of physical activity/exercise, amount of sleep, alcohol and smoking/vaping. We combined the responses to give three levels: “decreased”, “stayed the same” and “increased”.
Participants could select “not applicable” if they didn’t do the activity before lockdown and weren’t doing it at the time of questionnaire completion. For the first five variables (number of home-cooked meals, number of meals, number of snacks, amount of physical activity, amount of sleep), the number of participants selecting “not applicable” was negligible (n = 6, 4, 8, 2, 4, respectively) so these responses were coded as missing. For alcohol and smoking, “not applicable” was coded as a fourth category, representing non-drinkers and non-smokers/vapers.
In the same questionnaire, participants reported their employment situation before lockdown. We categorized responses into working and not working pre-lockdown and combined these with employment status at questionnaire completion to derive change in employment status during lockdown. This had five categories: employed with no change, employed with reduced hours, employed and on furlough or paid or unpaid leave, no longer working, and not working pre-lockdown.
Participants also reported how their financial situation compared to before the COVID-19 pandemic. Possible responses were: “much worse off”, “a little worse off”, “about the same”, “a little better off”, or “much better off”; we grouped the first two into ‘worse off’ and the last two into ‘better off’. Participants reported whether they or their partner had made any new claims for benefits since the pandemic, or had used rent or mortgage or other debt deferral since the pandemic.
Missing data
ACE measures were derived from > 500 questions answered between birth and 23 years of age, and no participant had data on all these questions. We therefore used multivariate multiple imputation to estimate missing values, that were assumed to be missing at random. This avoids exclusion of participants whilst reducing selection bias. Participants were only excluded from analyses if they responded to < 10% of ACE questions (Supplementary Fig.
1). For participants who responded to ≥ 50% of questions for an ACE, these questions were used to create a binary indicator of presence/absence of that ACE. For participants who responded to < 50% of questions for an ACE, presence/absence of the ACE was set to missing and was imputed. The ACE score was derived after imputing presence/absence of individual ACEs. Given known sex differences in ACE prevalence, missing data for males and females were imputed separately, and the datasets re-combined before analysis. The imputation model included all outcomes, exposures and covariates included in analysis, and 24 auxiliary variables likely to predict missingness or ACE exposure (details in Additional File
2). Some participants had missing SEP data, so we included in the imputation model 8-class NS-SEC [
38], from which the 4-group version described above was derived. Since only participants who completed the COVID-19 questionnaire were included in analysis, most of the analytic sample had complete outcome data. For those that had not responded to certain sections or questions, these were imputed in the same model. Using the mice package [
39] in R version 4.0.2, we created 50 imputed datasets for both males and females, with 10 iterations per dataset. Imputed values were combined using Rubin’s rules [
40] and trace plots used to check convergence of estimates.
Statistical analyses
All analysis was conducted in R version 4.0.2. To explore how participants in our analysis differed from the full cohort, we compared the distribution of maternal education between included participants and those excluded due to missing data. We used multinomial logistic regression to examine associations of ACE score, individual ACEs, and SEP with health behaviour and employment outcomes during lockdown. We assessed unadjusted associations, and associations adjusted for covariates. These included participant’s ethnicity, age in years at questionnaire completion, and home ownership. We also adjusted for their mother’s marital status, parity, and age at the participant’s birth, and educational qualifications of the mother and their partner. We tested for an interaction between each exposure and sex on the outcome.
Discussion
In a population-based longitudinal cohort study we identified changes in multiple health-related behaviours in both directions during lockdown, as well as changes to employment status and financial situation. The evidence of associations between SEP and health-related behaviour changes during lockdown was limited, but a higher overall ACE score and various individual ACEs were found to be associated with some behaviours e.g., decreased sleep and increased smoking. There was, however, clear evidence that adverse employment and financial changes were more likely to be experienced by people with low SEP or a history of ACE exposure. In this cohort of young adults, people in routine or manual occupations or long-term unemployment 4 years prior to the pandemic were more likely to experience adverse changes to their employment status during lockdown compared to participants in higher managerial, administrative, or professional occupations at that time. Those with a higher ACE score were more likely to be put on furlough or other leave or stop working entirely during lockdown. In addition, we found some evidence that the group with the greatest ACE exposure were more likely to experience adverse financial outcomes during lockdown.
There is a growing body of evidence of the indirect effects of lockdown on health-related behaviours. A study analyzing five British cohorts of different ages investigated whether behaviour change during lockdown differed by SEP and, like the present study, found that adverse effects on sleep during lockdown were more frequent amongst socioeconomically disadvantaged groups [
7]. The same study, which had a large combined sample size, found that lower SEP was associated with lower exercise quantity during lockdown, something that wasn’t detected in our sample, or in a study that measured smart-phone tracked physical activity [
18]. Analysis of a large UK cohort found that overall smoking declined during lockdown and that there was no interaction of smoking behaviour with education level [
11]. There is evidence of socioeconomic disparities in drinking behaviours during lockdown [
13,
14], which we did not detect in our sample.
In keeping with our findings, adverse effects on employment seem to be more common for younger people and those in lower paid occupations [
6,
21,
22]. One study exploring adversities during the first 3 weeks of lockdown by SEP found that people of low SEP were more likely to lose work, experience a cut in household income and be unable to pay bills during lockdown than people of higher SEP [
22]. Similarly, a Welsh study found that unemployment and furlough during lockdown disproportionately affected those experiencing financial difficulties and living in deprived communities [
23].
To our knowledge, ours is the first study to assess ACEs as a risk factor for adverse employment and financial changes during lockdown. There is evidence that childhood maltreatment is associated with poor adulthood socioeconomic outcomes, but the mechanisms underlying this relationship are largely unclear, with mental health and adolescent cognition potentially playing a role [
28]. Further study into the causal pathway from ACEs to adulthood adversity is warranted so that interventions can better support those with a history of ACEs.
Strengths and limitations
A key strength of this study is the detailed data on a range of ACEs (captured at multiple time points throughout childhood and adolescence) combined with data collected during the COVID-19 pandemic, which gave us the unique opportunity to examine the effect of the pandemic on people who have experienced childhood adversity. This study has limitations. The sample size and thus statistical power was restricted as the questionnaire was online only - invites were only sent to those participants for whom the study had a valid email address, impacting participants who prefer to complete questionnaires on paper. Our results therefore require replication in larger sample sizes, however this is a challenge given lack of cohorts with data on both ACEs and the pandemic. We have not corrected for multiple testing due to the correlated nature of our exposures and outcomes, meaning that the large number of statistical tests carried out are not independent. We acknowledge that type I error may be present. Some of the associations found have large confidence intervals. For these findings, had multiple testing corrected
p values been presented (i.e., if the α level for individual tests had been adjusted downward), many would not pass the arbitrary threshold for ‘statistical significance’. We do not consider these estimates to be definitive evidence against the null hypothesis of no association; replication in larger samples is recommended. Self-reported measures were used, so associations may have been biased by measurement error and reporting biases. During the pandemic, the ALSPAC cohort administered questionnaires that asked participants about changes in behaviour/socioeconomic circumstances. The questions used capture subjective changes in behaviour/circumstances since the start of lockdown, and not absolute levels of those behaviours/circumstances. This precludes analysis of how trajectories of absolute levels of a given behaviour/circumstances changed during the pandemic. Furthermore, the changes within each category of the ‘change variable’ are likely to be heterogeneous. ALSPAC participants are more socioeconomically advantaged and less ethnically diverse than the national average, and mothers of participants included in this study had higher educational attainment compared to the rest of the ALSPAC cohort. Results therefore may not be generalizable to the UK population. Although ALSPAC participants are in the age group most likely to be affected by lockdown in terms of employment [
21], their similar ages meant we could not study the effect of age on the outcomes. Finally, our SEP measure was solely based on participants’ occupations, but individual’s income, education, or residential area may also determine negative impacts of lockdown.
Conclusions
The results of this study support findings from routine data that the financial and employment situation of adults with lower occupational social class are more likely to have been adversely affected by the COVID-19 non-pharmaceutical interventions [
41]. Our results also suggest that exposure to multiple ACEs was associated with decreased sleep quantity and increased smoking/vaping, but we found little evidence that SEP or ACEs influenced changes in other health-related behaviours during the pandemic in this study of young adults, providing some reassurance that, at least in the short-term, the non-pharmaceutical interventions used against COVID-19 have not led to increased inequalities in health-related behaviours in this age group.
However, importantly, our findings highlight the economic adversity experienced by adults who have been exposed to ACEs. Despite the long-reaching consequences of ACEs persisting into adulthood, there are few interventions and services targeted at this group (with most services for adults targeting adult risk factors rather than previous childhood risk factors), and they are largely hidden from view in routine data. Our findings demonstrate the need for continued support of people who experience ACEs into adulthood and demonstrate that this need may have increased during the COVID-19 pandemic.
Acknowledgements
We are extremely grateful to all the families who took part in this study, the midwives for their help in recruiting them, and the whole ALSPAC team, which includes interviewers, computer and laboratory technicians, clerical workers, research scientists, volunteers, managers, receptionists and nurses.
Open AccessThis article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit
http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (
http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.