Income, ethnic diversity and family life in East London during the first wave of the pandemic: An assets approach

Objective: This paper reports first results from a survey of 992 parents and parents to be living in an ethnically diverse and socio-economically unequal borough of East London during the coronavirus pandemic that reduced mobility, closed services and threatened public health. Background: Little is known about the place based impacts of the pandemic on families with young children. We describe the living circumstances of families with children under five or expecting a baby living in Tower Hamlets during the Coronavirus pandemic in 2020, and then examine the relative importance of household characteristics such as ethnicity and household income for adverse impacts on survey respondents, as seen in mental health outcomes. Method: a community survey sample recruited with support from the local council comprised 75% mothers/pregnant women, 25% fathers/partners of pregnant women. Reflecting the borough population, 35 percent were White British or Irish and 36 percent were Bangladeshi, and the remainder were from a wide range of ethnic backgrounds. Adopting an assets based approach, we describe material, familial and community assets using three household income bands and seven ethnic groups. We then use regressions to identify which assets were most important in mitigating adversity. Results: We find that material assets (income, employment, food insecurity, housing quality) were often insecure and in decline but familial assets (home caring practices, couple relationships) were largely sustained. Community assets (informal support, service provision) were less available or means of access had changed. Our analyses find that while descriptively ethnicity structured adverse impacts of the pandemic related changes to family life, income and couple relationships were the most important assets for mitigating adversity as seen in mental health status. Conclusion: Supporting family assets will require close attention to generating local and decent work as well as enhancing access to community assets.


Introduction
East London is a unique place, offering a very particular context for family life during the Covid-19 public health emergency. It consists of eight boroughs (of a total of 32) in London, itself an exceptionally diverse capital city of nearly 9 million residents (Sepulveda et al. 2011). London, and the UK as a whole, has had a difficult experience of the Covid-19 pandemic. By September 2021, more than 134,000 people had lost their lives, with the UK having had the highest number of confirmed Covid-19 deaths in Europe (Statista 2021). Comparatively late to begin, mobility restrictions to control the rate of transmission started in March 2020 and continued, with variations in intensity and location, until July 2021. London is the UK region with the highest Covid-19 mortality rate in the UK, at 263.8 deaths per 100,000 (Trust for London 2021). In East London virus transmission, and mortality, was intense in the early phase of the pandemic. Some residents were at more risk of contracting Covid-19 than others. Those working in key industries, where working at home was not possible, such as transport, health and social care, and food production, were particularly at risk, and in London, these workers were disproportionately likely to come from black and minority ethnic backgrounds (Platt & Warwick 2020).
Tower Hamlets, the focus of this paper, is an East London borough in which the twin hallmarks are a highly ethnically diverse population and high levels of inequality. Migration has long been a feature of Tower Hamlets, primarily due to its location next to the Thames and the river ports, with large scale arrival of Jewish emigres in the late 19C and those from the Indian subcontinent from the 1950s onwards. In the most recent national population estimates of 2016, there were 301,000 residents of Tower Hamlets, of whom 30 percent were of Bangladeshi origin, 30 percent White British, and the remaining 40 percent represented a very wide range of ethnicities (Greater London Authority 2021). This profile is quite different to England and Wales as a whole, where, in the national census of 2011, 81% of the population was White British.
Likewise, the pan-London historic pattern of rich and poor living side by side, has long been a feature of Tower Hamlets with its proximity to the business district of Canary Wharf and City of London wealth and linked jobs, despite deep levels of chronic poverty (Tower Hamlets Fairness Commission 2011). In 2019, median household income in Tower Hamlets was about average for London at £30,760 (Tower Hamlets 2019) but income inequality is profound. Child poverty rates are the highest in England (CPAG 2018), after housing costs are taken into account. Bangladeshi families in particular are highly vulnerable to poverty, and work patterns are often characterised by precarity for men and low employment rates for women (Dyson et al. 2009).
The borough's population is fast growing and youthful, mainly through international migration, including from continental Europe. In 2017, over 60 percent of the population was under the age of 30 (Tower Hamlets 2017). Ethnic diversity is particularly pronounced among those under 25. In 2021, just 34 percent of this age group were of any White background, while two thirds were Asian (48%), Black (8%) or mixed (8%) heritage (Greater London Authority 2021). In this context, the current paper seeks to address two questions: using an assets based framework (Kretzmann & McKnight 1993) i) what are the living circumstances of families with children under five or expecting a baby living in Tower Hamlets during the coronavirus pandemic in 2020 and ii) what is the relative importance of household characteristics such as ethnicity and household income for adverse impacts on survey respondents, as seen in mental health outcomes.
Existing literature on indirect impacts of health emergencies and associated economic shocks on families with young children in advanced, 'rich', societies were, at the outset of the study, limited (Richardson et al. 2020). Yavorksky et al.'s (2021) review likened the coronavirus pandemic to a 'disaster' in which pre-existing gender inequalities in work and care distribution were exacerbated: women were more at risk of unemployment, but also more likely to be working in jobs that demanded intensification in a health emergency, and to be affected by home working and home schooling. Moreover, there was only a slight change towards more egalitarian sharing of domestic labour. Deteriorating parental mental health related to living with pandemic restrictions have been noted in surveys in Canada, the US (Chicago), Italy and the UK (Gadermann et al. 2020;Kalil et al. 2020;Spinelli et al. 2020;Cheng et al. 2021). Income loss was more important than job loss (Kalil et al. 2020) and impacting both parental mental health and parent-child conflict. Protecting family income has wide ranging mental health and child wellbeing implications. Parental perception of ability to manage was significantly associated with stress and psychological symptoms, which in turn impacted the behavioural and emotional problems of children in Spinelli et al.'s (2020) online survey of parents in Italy. Fewer studies of ethnic diversity and family life in the pandemic have reported. Dickerson et al. (2020) found that Pakistani mothers in Bradford, a northern English city, were more likely than White British mothers to be financially insecure, live in poor quality housing, and be less able to work. But White British mothers were more likely to report moderate-severe depression, and poor mental health was associated with social disadvantage. No ethnic differences in couple relationship quality or confidence in supporting children were found. Across the UK, children from Pakistani and Bangladeshi backgrounds were particularly likely to be adversely impacted by school closures (Bayrakdar and Guveli 2020) and parents from Asian British backgrounds were more anxious than White British parents around the birth of children during lockdown (Babies in Lockdown 2020).

An Assets based Framework
The Families in Tower Hamlets study sought to assess the economic, social and health impacts of the coronavirus pandemic on families with children under five. In Kretzmann & McKnight's (1993) framework, rather than assessing problems or deficiencies, assets are the central analytic unit. Assets are multi-level; they operate at individual, familial, community and institutional levels as resources or 'gifts, skills and capabilities' of individuals or residents in any given area, in addition to the associative and institutional fabric or infrastructure that serves the population. Assets offer opportunities for sharing experiences, skills and interpersonal resources (Kretzmann & McKnight 1993). Assets are 'positive individual factors such as academic competence, motivations and beliefs', while resources are external factors such as parental support for education (Sacker & Schoon 2007:874). Individual assets and family resources are 'key protective factors' in the development of young people (Sacker & Schoon 2007), and, we will argue, in families facing the coronavirus pandemic. An assets based approach starts with an ethic of valuing people's strengths, and recognising that assets are 'realised, expressed, mobilised and sustained through people's actions, connections and participation' (Rippon & Hopkins 2015:12). Communities, networks and places are a key site for participation, and generation of assets, for example through volunteering or sharing experiences, such that reciprocity can be a key health asset (Marmot 2020). Asset based approaches to reducing health and other inequalities 'sees citizens and communities as … co-producers [and] empowers communities to control their futures' (Foot & Hopkins 2010: 7). In this respect the assets based approach has common ground with Sen's Capability Approach in that it is a moral framework that values individual's agentic possibilities for freedom, and to be able to, and to take, action to support their goals (Alkire 2002). Moreover, both approaches foreground the means or resources individuals have to exercise their inherent agency, alongside recognition of conversion factors or contexts such as social or economic status.
The agentic positioning of parents in this study meant that we looked for evidence of parental capabilities to act to manage day to day life with young children under the pandemic, as seen in the ways they harnessed their material, familial and community assets and resources.
In this paper we examine the extent to which ethnicity and income were structural constraints on families' experiences of the pandemic, using mental health as an outcome indicator. First, we examine material assets such as income and housing, familial assets such as home learning and relationship quality, and community assets such as use of child -related services and informal support. Given the ethnic diversity of Tower Hamlets we foreground seven ethnic groups to describe sample characteristics, family life themes and mental health measures. Second, we examine how mental health patterns are influenced when controls for specific asset indicators are considered, namely household income, couple relationship quality and informal support, using regression models. We find that for all ethnic groups, income and couple relationship quality are largely driving mental health outcomes.

Measures to control Covid-19 transmission
Tower Hamlets, like everywhere in the UK, began its lockdown on 23 March 2020, with closure of early childhood education and care (ECEC) services, schools (except for children of essential workers), workplaces, non-essential shops and businesses and reduced health and social care provision, and restrictions placed on daily activities (House of Commons Library 2021). From 1 June, schools, ECEC services, and workplaces gradually reopened. Mobility restrictions were eased and replaced by localised restrictions at times of high rates of virus transmission.
By early September there were escalating concerns about rates of transmission, and new restrictions began to be introduced, notably the 'rule of six' on 14 September which legally limited associating to six people, whether in or out of a household. By this point, rules had diverged across the four nations of the UK. In England, in a further response to escalating Covid-19 cases, a three tiered approach to restrictions came into force on 14 October, and Tower Hamlets, along with the rest of London, entered Tier 2, defined as 'high alert' on 17 October. A second, four week, 'stay at home' lockdown ran from 5 November -2 December 2020 but schools stayed open. Escalating transmission during December paved the way for a third lockdown, when schools were open to children of a wide range of critical workers, but restricted to online learning for many, from 5 January to 15 March 2021.
The principal means to support income of those already in employment during the pandemic was the national Coronavirus Job Retention Scheme (furlough) scheme. In this scheme, 80 per cent income replacement (to a ceiling of £2,500 a month) was available for employees whose employers had to pause their active work. Conditions for claiming furlough payments meant it was less available to insecure workers on irregular schedules such as zero hours contract or those working in the gig economy. A separate scheme for those who were self-employed aimed to mirror the furlough payments but this similarly had gaps in eligibility.
The alternative national income support was 'Universal Credit', a welfare payment to those who are out of work or on very low income. Whereas income recovery through the furlough scheme gave individuals, of most income levels, a degree of financial continuity and security with its ceiling slightly higher than an average national wage (£30,000 p.a.), Universal Credit is significantly less generous with a ceiling of £20,000 per household. In part recognition, a Covid-19 supplement of £20 weekly per household was introduced nationally for new and existing claimants on 6 April 2020. For those living in a couple household, and aged 25 years or over, with two children, the maximum available from Universal Credit was £680.71 per month, excluding housing costs.

Economic, social and health impacts of the pandemic in the UK
During 2020, unemployment in the UK rose from 4 percent in April to 5 percent in December (Labour Force Survey 2021). Among men and women in the age bracket most likely to have young children (25-49 years), unemployment rose from 3.2 percent to 3.8 percent. About 21 percent of the adult population was economically inactive, for reasons of being a student, looking after family or long term sick. During 2020, income precarity soared with a 98 per cent uplift in Universal Credit claims between 13 March 2020 and 14 January 2021 (Official Statistics 2021). By this time there were 6 million people on Universal Credit with 4.5 million claims made between 13 March 2020 and 14 January 2021. In-work precarity meant that at the start of the pandemic 20-40 percent of low to middle income households were unable to manage for a month if they lost their main source of income (Blundell et al. 2020). Overall, women, young people, and those on a low income were particularly at risk of losing income but key workers, often those most at risk from Covid-19, were also often from ethnic minority backgrounds (Blundell et al. 2020). This meant the population of Tower Hamlets was both at particular risk of contracting the virus and of adverse economic impact.
There have also been profound impacts on mental health. O'Connor et al. (2020) found that women, those living in conditions of social disadvantage, and with pre-existing mental health conditions, experienced worsening mental health during the initial phases of lockdown. Henderson et al. (2020: 8) found that, among members of British cohort studies, in the early stages of the pandemic, and among those aged 30, '14% of males and 20% of females had high depressive symptoms, and 15% of men and 23% of women reported high levels of anxiety'. Similarly, 26% of males and 34% of females at age 30 reported loneliness. Age 30 most closely mapped onto the age group of cohort members likely to have young children. Henderson et al. (2020: 10) state that for those aged 30, there were 'higher levels of psychological distress among females during the Covid-19 lockdown compared to when previously measured at age 25'. This age group, and particularly women, was more likely to be adversely affected than older study participants.
Apart from impacts on parents, there are impacts on children. Concerns about the wellbeing of children were expressed by parents in the summer of 2020 (Children's Society 2020). Worldwide, studies have reported less time playing outside, increased sedentary behaviour and disturbed sleep patterns among all age groups including pre-school aged children (Lecuelle et al. 2020;Moore et al. 2020;Wang et al. 2020;Imran et al. 2020;Pecoraro et al. 2020;Dellagiulia et al. 2020;Rubén López-Bueno et al. 2020;Pisano et al. 2020;Di Gorgio et al. 2020;Alonso-Martínez et al. 2021;López-Gil et al. 2021) as part of more general concerns about mental health and wellbeing.
Early, locality led, scoping of the impacts of the pandemic on the borough of Tower Hamlets anticipated exacerbation of existing mental health difficulties including stress, anxiety, loneliness and grief through reduction in support services, economic shock on local businesses and concomitant employment. The borough also identified concerns about the consistency and quality of home learning while schools were closed, with potential for longer term impacts on children's wellbeing and attainment. Alongside these major concerns was recognition of some positive changes, such as community mobilisation and cohesion (Starkie 2020).

Method
There were three study components: i) a community survey of households with children under five; ii) a repeat in-depth qualitative household interview panel with a subset of the survey; and iii) community assets mapping of changes to borough services for families and children under five. This paper draws on Wave 1 of the survey, which ran from mid-July to end November 2020. It included 126 items, designed in conjunction with national and European collaborators including: i) a parallel 'Born in Bradford' survey of job, housing and food insecurity, children's home learning, health and care service access, and mental health (Dickerson et al. 2020); ii) the 'Gender (In)equality in Times of COVID-19' survey using validated instruments to examine work-family conflict and parental coping strategies (Yerkes. et.al 2020), and iii) items from a national longitudinal household panel study called Understanding Society (IESR 2021).
Survey data was collected using Qualtrics, a multi-lingual online survey format, and exported to SPSS. Three open-ended questions were included: i) what are your three biggest worries right now?; ii) can you tell us about a challenge you have faced in the last two weeks?; and iii) can you tell us how lockdown has made any parts of your life easier or more enjoyable? These questions aimed to provide qualitative illustrations of respondents' experiences. Not all participants entered a response (491/992 answered Q2; 674 Q3. Since respondents could choose three 'worries' it is not possible to calculate an n for Q1). Response length varied from one word to several short sentences. Extracts were coded using Nvivo and followed the codebook developed by the Born in Bradford team, with some additional location specific codes. Frequencies of codes were totalled. For example, 766 respondents mentioned worries about mental health of parents and children as one of their three worries. However, in this paper, responses to open ended questions were selected for their illustrative expression of thematically presented survey findings and not subject to further qualitative analytic processes.
To address the first research question, descriptive statistics were used to assess the living circumstances of families in the borough. To address the second question, and disentangle the relative impacts of structural and relational assets with respect to mental health, binary and multiple logistic regression models were applied for two outcome variables, depression and anxiety. In line with national trends, we anticipated elevated mental health distress for the study population.
Mental health data was collected using standardised measures of depression (Kroenke et al. 2009) and anxiety (Spitzer et al. 2006). For depression, the PHQ-8 is an 8 item instrument with a 4 item scale (not at all, score=0, one or two days, score=1, more than half the days, score=2, nearly every day, score =3). A score of 0-4 = no depressive symptoms, 5 to 9 =mild depression, 10 thru to 14 =moderate depression, 15 -19= moderately severe depression and 20 to 24 =severe depression. The standard dichotomous depression PHQ-8 score ≥ 10 = a case, < 10 not a case (depression vs. no depression) was adopted for the regression analysis in line with identification of clinically symptomatic levels of depression. For anxiety, the GAD-7 is a 7 item instrument with a 4 item scale (not at all, score=0, one or two days, score=1, more than half the days, score=2, nearly every day, score =3). A score of 5=Mild anxiety, 10 =moderate anxiety, 15 or more =severe anxiety. and anxiety. The standard dichotomous anxiety score GAD-7 ≥ 10 = a case, < 10 not a case (anxiety vs. no anxiety) was adopted for the regression analysis in line with identification of clinically symptomatic levels of anxiety. The 5 point Likert scale measuring 'Quality of relationship with partner' was dichotomised to 'Very good/good' vs 'average/very poor', and the variable 'household income' was categorised as low income (<£20,799), mid income (£20,800-£51,999) and high income (£52,000 >).
Binary logistic regression was conducted for a subsample of respondents using a simplified version of the ethnicity profile (White (including White British, Irish, White Other), Asian (including Bangladeshi, Indian, Pakistani, Chinese, Vietnamese) and Black (including Somali) backgrounds). We explored the relative influence of: i) ethnicity (White, Asian and Black), ii) sex (male and female), iii) income (low, mid and high), iv) parental status (couple parents and one parent households), v) relationship quality with partner (very good/good and average/very poor), and vi) informal support received from friends and family outside of the household (received or not received) on depression and anxiety scores. For the regressions, the category "one parent" included all individuals living in one parent households (not cohabiting), combining the minority who had a non-residential partner (LATs) with the majority who were single. This recoding was done to enhance the sample size of the one parent category, recognising their heterogeneity (e.g. those with a partner living outside may have some differences).
Subsequently multiple logistic regression was carried out to identify factors associated with increase in depressive and anxious symptoms after controlling for all other confounding variables in the proposed model. Cases of missing data for some or all variables were deleted (known as listwise deletion).

Sample recruited
In the absence of an established sampling frame, recruitment of a community sample was via general and targeted communications carried out by local authority partners at the London Borough of Tower Hamlets. This included a borough wide social media campaign including residents' magazines and newsletters, websites, invitations to participate via the borough's child-facing services including early years, family support and health visiting teams, and civil society organisations such as community centres and faith-based organisations. Targeted communications included cooperation with a specialist voluntary organisation working with Somali women, whose inclusion depended on own language telephone interviews, and sending a postcard to 6585 families with young children who were registered on a borough held database of all those claiming a wide range of welfare benefits. Participants and volunteers were each thanked with a shopping voucher of £10.
The aim was to recruit 1,600 survey participants, men and women who were parents of children aged under five, or were pregnant. This target number represented approximately 8 percent of the age cohort and was considered, in consultation with the borough public health team, to be ambitious but achievable. The eventual sample consisted of 1,551 adults, of which 559 were excluded due to incomplete data making a sample size of 992. While participants were not randomly selected, the eventual sample broadly matches the borough population in terms of proportions drawn from the major ethnic groups (Bangladeshi and White British) (GLA 2021) and median salary as discussed below.

Ethnicity and gender
To make the complexity of ethnicity somewhat visible we allocated participants to one of seven ethnic categories that mirror distribution in the local population records. These are: i) White British/Irish; ii) Other White (all non-British/Irish White groups e.g. including continental European); iii) Asian: Bangladeshi; iv) Asian: Other (principally Indian, Pakistani, Chinese, Vietnamese); v) Somali; vi) Black: Other Black; and vii) Other ethnic group.
Table 1 describes the sample in terms of gender and ethnicity. Three quarters (n = 732) of the sample were female, while 25 percent (224) was male. Just over a third (35%) were White British or Irish, compared to 30% in the borough as a whole, while 36% were Bangladeshi, again a slight over-representation compared to borough figures (Greater London Authority 2021). Of 992 respondents, 88% had a child under five living with them at home, 6% said they were pregnant and 6% were both pregnant and had children under five.

Household and family relationship structure
One third (32.8%) were one parent households, half (46.8%) were couple households and in five percent there were three or more adults 1 . In the latter group of 52, 41 (79%) were of Bangladeshi origin. Among two adult households, half (51.5%) were White British/Irish and 27 percent were Bangladeshi families. A majority of parents (78%) lived in co-resident couple households (married/civil/non-formal partnership), 12.6 % were in non-partnered one parent households (single no current relationship) and 2.9% resided in a one parent household with a non-resident partner (LAT).

Material assets: Income and financial security
Using three household income bands, 39 percent of the sample were on a 'low' household income (designated as up to £20,799), 32 percent were on a 'medium' income, and 22 percent were on a high income (£52,000 or above) (Table2, remainder preferred not to say   Prior to March 2020, two-thirds (67%) of survey respondents were employed, or on leave from, employment with in-work benefits: 49% were employed, 8% were actively selfemployed and 10% were on maternity/parental leave. The remaining third (33%) were unemployed or not working despite a self-employed status. This was especially the case for Bangladeshi (43%) respondents who were unemployed and on unemployment benefits in contrast to 21% of White British/Irish respondents. By the time of the survey, reported employment had declined to 60% and unemployment had risen to 40%. Bangladeshi respondents were more likely to be unemployed (52%) than White British/Irish respondents (24%) but these are clearly much higher figures than nationally. So while the pandemic and economic shock had not led directly to major changes in employment rates, there was a deepening of pre-existing income precarity, as well as an influx of 'new poor', that affected some ethnic groups more than others.
6.1.1 Living with precarity Inadequate household income could be seen in the rates of foodbank usage and food insecurity. In the four weeks prior to survey, 22% of respondents had used a foodbank, including 4% reporting four times or more over this period. Furthermore, 45% reported that food did not last and they didn't have money to get more. One quarter (25%) of respondents reported having to skip meals because there was not enough money for food. Asked about challenges faced recently, money worries were the most frequently reported, such as: Not having enough money to even buy food for myself or my child and having to ask friends for money. (Mother, Black) Housing tenure and quality reflected an urban, low income sample. In a reversal of national trends, 56% of respondents rented their home, compared to 33 percent who owned or were buying their home with the help of a mortgage. Among Bangladeshi respondents, there was a higher than average proportion of renters (87%). By contrast, two thirds (65%) of White British/Irish respondents either owned their home outright or were buying with the help of a mortgage. Seventeen percent of respondents had a single bedroom, most frequently reported among Black (33%) and Asian Other (25%) respondents. These families are likely to be sharing bedrooms with one or more children and/or have parents sleeping in shared spaces such as the living room. Space constraints during lockdown meant multiple uses of all rooms with implications for mental health as this Somali mother explained: The hardest part is we are suffocated in our home we live in one bedroom flat with 6 of us there is nowhere to relax, play, sleep, eat, or  Moreover, only half (49%) of respondents had access to an outdoor space. The usual means of spending time with children and keeping them active had disappeared during the pandemic leading to isolation, and a perception of being in 'prison'. One respondent described her accommodation as a 'studio flat' (i.e., the bedroom and living room are one area) with 'hardly any space' for her two children 'to play'. With 'family and friends all across London', she had not 'seen much of them'. She concluded: 'I am used to being out and about with my kids, keeping them busy and staying active but that has been impossible' (Mother, Black). Another mother, this time from a mixed white and Black African background, reiterated the cramped conditions of lockdown: 'Living in a flat with no garden is difficult. Feels like prison due to limitations and restrictions'.

Familial assets: Home caring practices
Survey items about helping children learn at home and confidence in parental skills were summarised as 'home caring practices'. Nearly all (94%) parents who responded to the question said they were reading to their children (Table 3). Reading 'everyday' was most frequently reported among White British/Irish families (56%) followed by Other White (54%) while reading 'some days' was most common for Bangladeshi (32%) (reported levels were high among Somali parents (33%) but the overall number in this group was very low). This pattern of clear commitment amongst parents to support children's learning was also found when asked about helping children to learn the ABC (81%) and numbers and counting (85%). Although the numbers are low it is important to note that Somali and Black Other respondents were particularly likely to report helping develop their children's understanding of the alphabet and basic appreciation of mathematical concepts (96% and 89% respectively 2 ).
Parental deployment of home caring practices was accompanied by concerns about children's development. Nearly a quarter (23 percent) of respondents cited worries about children's learning and socialising at home. In one example, a 'other White' mother was worried about her son who was 'really struggling with reading and writing'. Moreover, she cited difficulties keeping 'my children occupied and safe when everything is closed'. Another, this time a Bangladeshi mother, was worried about 'children's learning and falling behind'. In addition, having children at home was giving rise to worries about behaviour and sibling dynamics. In a third illustration, this White British mother said her 'children [were] hyperactive, bored, uninterested in learning and missing their school friends. They are also constantly bickering'.  Parents had considerable familial assets to support children and most (73%) were confident in their skills, while 12 percent were not confident. There was ethnic patterning to this finding (White British/Irish respondents 82%; Bangladeshi 64%) and fathers were marginally more confident than mothers (75% vs 72%). But pandemic restrictions had resulted in over half (54%) of survey respondents having 'much' or 'slightly' less time for themselves, particularly among the Other White (66%) and Asian Other (63%) groups. However, a quarter (24%) had 'slightly' or 'much' more time at their disposal.

Community assets: Access to health services
Survey respondents reported that accessing community assets such as primary care services (midwifery, health visiting) was less easy than usual during the pandemic. The mode of delivery was largely online or telephone and there were restrictions on partners' accompanying women during hospital admissions and antenatal appointments. While official figures reported that contact was maintained with around 90 percent of mothers and pregnant women in the borough (Gilmour, p.c.) only 75 percent of pregnant survey respondents had access to routine midwifery appointments when needed. Bangladeshi (69%) and Asian Other (60%) women were less likely than other groups to report access to midwifery. Moreover, 25 percent of Bangladeshi respondents said they did not have access to maternity scans, many more than White British/Irish (14%) women.
Some respondents referred to the worry caused by changes to mode of antenatal care delivery, such as this White British pregnant woman in a response to an open ended question: 'Not being able to have a face to face meeting with the midwife'. Others talked about the impact of changes on pregnancy services in terms of isolation from support. One Bangladeshi mother reported 'Sleepless nights and anxiety' as a worry. This stemmed from 'going through pregnancy appointments alone as no one is allowed to accompany me'.
Another pregnant woman, whose ethnicity was classified as 'Asian Other', found 'not having health care professionals to speak about certain worries' and the isolation of going to appointments alone difficult. She said 'My partner not being able to attend hospital appointments i.e., scans' was a worry and combined with another: 'taking public transport … as people don't comply with wearing masks and authorities not taking action'. A fourth mother, this time White British, was 'worried about giving birth during Covid and not being able to have visits from my husband after baby arrives and being alone in hospital during recovery'. The separation of fathers from antenatal and postnatal care was also a worry for some, as one White British father said: 'Wife have birth! Getting to the hospital, having to leave after the birth with baby in ICU. Not being able to see them for days'.
Universal child health screening is a key community asset and underpins infant health care but only 30 percent of respondents had had access to routine checks such as newborn hearing, blood spot, new baby checks and health checks at 6-8 weeks among those who had given birth since March 2020. Similarly, just 24 percent had had access to routine immunisations for their newborn children. In a contrast to other findings, White British/Irish respondents were the ethnic group least likely to access routine new baby appointments. Again, respondents drew attention to the links with maternal wellbeing of not accessing community assets as they had expected to. For example, an 'Asian Other' mother talked about her anxiety being 'exacerbated by lockdown because the regular baby clinics are now unavailable. Thus I do not have regular interactions with health providers to ask questions on a casual basis as they come up, the anxieties build and I doubt myself'. Another mother, who was White British, referred to feelings of vulnerability that arose without standard health checks for babies: 'Delayed access to standard baby check ups has made me feel more vulnerable. No 6 week appointment and 8 week delayed until 11 weeks'.
Lack of access to health visitors to address child-rearing issues was also reported although highly valued when it happened. Of those who responded to a question about accessing a health visitor, 59 percent had been able to do so (24 percent had not tried) and almost all who accessed a health visitor had received the support they needed. But being without health visitor or other peer support made parenting 'lonely' during the pandemic, as this White British mother explained:

My daughter is turning two and her behaviour is difficult at the moment. I don't know if this is … to be expected -or because she is struggling with our circumstances. It is hard for me to get support or advice from others which is how I would normally cope with this type of situation. Parenting is a much more lonely task at the moment.
The absence of support from any source, whether formal health services or informal friendships or parent groups is specific to the pandemic for many parents, and the consequent loneliness may have implications for parental health and wellbeing (El Osta et al. 2021).

Deploying assets to mitigate adverse impacts
Survey respondents' material assets (income, housing) were becoming more precarious and there was a dramatic change in community assets (service accessibility, facilities available) during the 2020 Coronavirus pandemic. Familial assets (home caring) were sustained and widespread, although considerable concerns about children's wellbeing and development were expressed as well as a reported sense of depleting time for self. Descriptively, it appeared that certain ethnicities were more likely to experience adverse impacts from the pandemic than others.
To investigate what was helping parents through this time we discuss our findings in relation to three indicators of potential protective assets: household income, relationship quality and informal help from outside the household. These three selected protective assets represent multiple levels of assets on which families draw. We use mental health as a proxy outcome indicator. In this section we first descriptively discuss household income, relationship quality and informal support. Then we employ regression models to try to account for the relative impacts of income and ethnicity on mitigating adverse impacts of living through the pandemic in urban London.

Household income and health
Seventy six percent of respondents said their health was good, very good or excellent while 24 percent said their health was 'fair' or 'poor'. Among ethnic groups, Bangladeshi and Black Other respondents were most likely to report 'fair' or 'poor' health (32% and 49% respectively).
However, mental health difficulties were common. While 39 percent of respondents had no symptoms of depression, 31 percent had moderate to severe depressive symptoms, compared to 19 percent during this period nationally (ONS 2021). Depressive symptoms were more commonly found among Black and Black Other respondents (42% reported experiencing symptoms of moderate depression) compared to other groups (e.g.: 9%, Somali: 13%, White other: 14%, Asian other: 15%, White British/Irish: 18%, Bangladeshi 21%). Similarly, only 43% of respondents had no symptoms of anxiety. Moderate anxiety was more common among White British/Irish (20%), Bangladeshi (18%) and Black (18%) respondents compared to other groups. While we do not know for how long these mental health symptoms had lasted, material assets were an important protective factor. More low income respondents experienced symptoms of moderately severe depression (60.6%) compared to medium (23.4%) and high income (7.4%) households (Table 4).

Relationship quality and wellbeing
In terms of familial assets, a second potential protective factor is relationship quality. With restrictions on mobility and enforced working from home where possible, potentially the pandemic has helped families have more time with each other and as parents. Threequarters (77%) of respondents said they had a good or excellent relationship with their spouse while almost a quarter (23%) described the relationship as poor-to-average. Forty percent of those who responded to a question about aspects of the pandemic that had made life more enjoyable referred to spending time as a family or improved relationship with partner and five percent said that their partner did more to help now. Working from home or being on furlough was the main driver for this as some respondents explained. Additional time as a family afforded by fathers and mothers working from home 'has strengthened the[ir] bonds [between father and child]' (Mother, Other White), 'makes life a lot easier with two small children' (Mother, White British) and enabled 'more job sharing and understanding of each other's roles' (Mother, prefer not to say ethnicity). Furthermore, the children, said one Bangladeshi mother, 'really loved' their father being at home when on furlough. Another mother, this time White British, stressed the benefit of having time to pause, appreciate family and rethink work-family relationship: 'We have a lot more family time together and are closer than ever. As much as I find it hard to look after myself when working at home … it's great to be at home all together. My husband has been able to rethink his career too'. However, the impacts of the pandemic were highly variable. Nearly as many (38%) of those responding to the question of enjoyability said 'nothing'. Two navigational difficulties between couples were allocation of time and sharing care of children. One White British mother expressed this as 'buying time from my partner so that I can do more work while he looks after the children. This is a constant struggle' while another, from a Bangladeshi background, talked about the challenge of 'coping with a newborn, toddler & teens and trying to explain to my husband I need help'. When asked what they did to cope with the current coronavirus situation just over half (51%) of survey respondents tried to find practical solutions together. Just under half (41%) were affectionate to each other and coped emotionally together and 8% tried something else. Nine percent of respondents did not let their partner know that they felt stressed by the situation. Navigational difficulties echo the findings of Yerkes et al. (2020) on the magnification of some gender inequalities during the first lockdown.

Informal support
For community assets, our indicator was giving and receiving support from outside the household and community engagement. Just over half (52%) of respondents said they had had help from someone outside the household and 57 percent gave support to others. There was little difference in giving and receiving support across the income bands, but was more common among those on a low income (56.5% were receiving and 59.2% were giving support). Likewise giving and receiving support was distributed across ethnic groups. Somali respondents were most likely to be receiving support (70%) (but again the total number for this group was small). White British/Irish were most likely to be giving support (57%). Support was mostly in relation to childcare and shopping. The scale of informal support was smaller than expected, given the vibrant local voluntary sector and celebrated 'community spirit' of the area, where 'in every corner of our wonderful borough, we have seen people selflessly putting their hands up to help others' … 'it has been incredibly heartening to see so many volunteers, community groups, partner organisations, local businesses, council staff and residents working together from across our diverse communities to tackle coronavirus' (Tower Hamlets 2021). Perhaps unsurprisingly, only a few survey respondents referred to informal community support as an aspect of life that had become more enjoyable during lockdown. However, one 'Asian Other' mother talked about the park as a community asset that became a place to meet new friends during lockdown: 'Through the daily exercise time … I have met many mothers of young babies and toddlers at our local park'. She went on to link lockdown restrictions to stimulating new friendships and supporting mental health and wellbeing: 'I don't think everyone would have been out and so talkative if it weren't for the lockdown. These women have provided much needed friendship, practical and mental health support for me during this time'. Another mother, White British, also reported 'talk[ing] to anyone in the park'… 'people have been more friendly' and parks have been 'cleaner and nicer environments for children'.
But conversely, being without friends and family also meant less practical support, particularly around the birth of a child, which prompted some anxiety for one White British mother who expressed concern about how she would cope without 'the practical support that comes with seeing close friends and family' when she had her second child, and if 'friends and family cannot come around and help out', again underscoring respondents worry about the impact of social isolation on parenting (El Osta et al. 2021).
Overall, harnessing community assets in the form of informal support during lockdowns appeared to be shaped by existing cultural practices and habits among kin, and to an extent by taking the initiative to be out and about in local areas.

Assets as predictors of mental health outcomes
To address our second question about the relative importance of six assets on health outcomes, we turn to our regression models. When examining the relative influence of ethnicity, gender, income, parental status (couple or single parent), relationship quality with partner, informal support from outside the household, on mental health, we found that predictors associated with a 'clinically important' increase in depression were largely material assets (income) and familial assets (relationship quality). These findings mirror those of Dickerson et al. (2020) which found financial insecurity and relationship quality were important predictors of depression.  Table 5 shows the bivariate and simultaneous logistic regression models for depression. The simultaneous model suggests that respondents from medium and high-income households and those reporting very good or good quality couple relationships have a decreased likelihood of experiencing depressive symptoms, adjusted for ethnic group, sex, informal support received and parent household status. This finding indicates that parents who have a supportive relationship with their partner (familial assets) and higher incomes (material assets), above £20,800, are more protected from depression symptoms once ethnicity, sex, informal support received and parent household status are adjusted for. Tables 5 and 6 include p-values for the regressions, even though this is not a probability sample, as information (for a discussion on the use of p-values in non-probability samples please see Hirschauer et al. 2020). A similar pattern was found for the anxiety simultaneous model. Respondents from medium and high income households and those reporting very good or good quality partner relationships have a decreased odds of experiencing symptoms of anxiety, adjusted for ethnic group, sex, informal support received and parent household status (Table 6). As with depression, material assets and familial assets protect parents from symptoms of anxiety.
The pattern of these regression results suggests that, for this sample, optimal mental health during a pandemic, requires both an adequate income and a "good enough" couple relationship. Survey respondents lived in different couple contexts but mainly as co-resident parents (78%) with a minority (2.9%) residing in a one parent household with a nonresident partner (LAT). In addition, six per cent of the sample were in couple relationships awaiting the arrival of their first child. Moreover, the pattern held in families with diverse ethnic backgrounds including among those living in multi-generational households.

Conclusions
This paper addresses two questions about families' lives in a highly unequal and diverse population in one part of London during the 2020 coronavirus pandemic. The first was to describe the living circumstances of families with young children in terms of material, familial and community assets, and the second was to assess the relative influence of household characteristics and assets in predicting ways of mitigating adverse impacts of the pandemic with mental health as our proxy indicator. A striking strength of the data is the high level of representation of ethnic groups other than White British/Irish, who are often not included in similar studies.
Material assets were descriptively structured by ethnicity in that income precarity was more likely among some ethnic groups. This predated the pandemic but there were signs of deepening income precarity, particularly in food precarity. Housing quality was poor, with frequent reports of compressed living in small spaces, exacerbated by a government mandate to stay at home. For those unemployed and on furlough the experience was quite different to those working from (or away from) home in terms of time available for children and family life but in both cases already squashed accommodation became multi-use sites for daily life. Moreover, deployment of community assets was less than expected, with changes to mode of service delivery that was experienced as loss of support at critical family moments, such as giving birth alone. But familial assets were reportedly largely sustained, through couple relationships and home caring practices with children. These findings concur with those at a national level, where adverse pandemic impacts on income, health and community support have been felt disproportionately by families with children and especially those on a low income (Collard et al. 2021).
In relation to the second question, using regression analyses to focus on identifying the most influential assets on mental health outcomes, we found that income as a material asset and couple relationships as a familial asset were key factors in predicting depression and anxiety. Ethnicity was descriptively important but in the regression not as significant as income. Data from the second wave of our survey (in early 2021) will indicate whether unemployment, low income and associated food insecurity, and familial practices, worsened as the pandemic progressed.
Our initial findings, based on original models in a community setting, suggest that recovery policies should focus on income and family support. To date, the generation of secure, well paid and local employment as the best protection against poverty (EU 2015) has been argued for at a national level (Michie 2021) as well as supporting pandemic related uplifts in the social security system (Waters & Wernham 2021). Post pandemic policies that support family life with specifically young children have been less evident.
Of particular concern is the finding that community assets such as health and care services, and informal support were not reliably available, and concurs with the findings of a national survey of parents (Babies in Lockdown 2020). The disruption to accustomed practices in relation to family support when giving birth may have altered the meaning of 'contact' with health professionals and account for the discrepancy between the borough's delivery data and the experience of parents. Informal support was difficult to enact when living in isolation, particularly when living in a very densely populated urban environment where virus transmission risk was high. But across ethnic groups, familial assets in the form of home caring and learning practices and couple relationships were strong despite adversities.
Clearly, addressing material assets is fundamental. Decent work, with its four pillars of more employment, for mothers as well as fathers, in a policy climate of social protection for families, rights at work and social dialogue that offer freedom to act is a first step to improve mental health (ILO 2017;Marmot 2020). Moreover, improving the quality of housing can improve mental health and wellbeing (Evans et al. 2000) and a policy of shaping urban housing design to recognise children's needs is now adopted in Tower Hamlets (Tower Hamlets 2020).
Supporting familial assets will require attention to the temporality of family-work lives as well as the availability of work and education, health and care services in modes that families find helpful. Families in our study found the pandemic generated more time spent as a family, with the potential to develop closer relationships, and this could be particularly beneficial when babies were born and fathers could spend more time at home. Localised employment that does not require long journeys to work would help support familial assets. Responsive community assets such as local health services, as well as availability of places to go with children such as parks are clearly crucial to mothers' and fathers' parenting and family wellbeing.
There are several limitations of the study. First, as a volunteer sample, and in the absence of up to date demographic information, we cannot be sure it is entirely representative of the borough population. It may also have under-represented families with severe couple dissatisfaction. Second, low numbers mean we cannot confidently represent the perspectives of some ethnic groups in a detailed way. Third, missing data for some items is a limitation but to be expected for survey completion during a global pandemic. We have tried to point out where missing data may have impacted on findings and in particular this may have had an impact on the regression models which depend on full completion of the relevant standardised measures.
Further work is required to better understand some aspects of the data and analyses are ongoing, particularly to identify changes in families' circumstances between survey waves and to understand the extent of multiple and intersecting social and environmental disadvantages on, for example, health and wellbeing outcomes.
Our overall conclusion is that while families from all backgrounds and household structures are generally managing as best they can, higher household incomes and lower housing costs are needed to improve families' conditions of living and particularly their mental health. Local policy actors in the borough, and those with responsibility for areas with similar characteristics, have to look "upstream", including to national government, at how they can create decent employment for mothers as well as fathers and responsive social protection policies to support young children through maternal, child health and early childhood education services through what will be a challenging period ahead. by the UK Economic and Social Research Council (ES/V004891/1). We would like to thank the many organisations and individuals that supported the study in the London Borough of Tower Hamlets and Charlie Owen and Francisco Zamorano Figueroa for their assistance with analysis.
Ethical approval for the study was awarded by the UCL Institute of Education Research Ethics Committee (REC1366) and the NHS Health Research Authority (20/LO/1039).