The Impact of Informal and Formal Care Disruption on Older Adults’ Psychological Distress During the COVID-19 Pandemic in UK


 This paper investigates how formal and informal caregiving disruptions—due to the U.K. government’s non-pharmaceutical interventions (NPIs) aimed at reducing transmission of the SARS-CoV-2 virus—may have affected the likelihood of psychological distress among older individuals. We model the association between disruption of formal and informal care and mental health of the elderly during the first wave of the COVID-19 pandemic using a recursive simultaneous - equation model for binary variables. Our findings reveal that public interventions, which are most essential for reducing the pandemic spread, influenced the provision of formal and informal care. The lack of adequate long-term care following the COVID-19 outbreak has also had negative repercussions on the psychological well-being of these adults.



Introduction
The first national lockdown to mitigate the transmission of COVID-19 in the U.K. was introduced on March 23, 2020 and remained in place until July 4, 2020. During the lockdown the government imposed national restrictions and required all those who could to work from home, closed all but essential shops, and advised the population to stay at home and limit contact with other people outside of their households. Moreover, the U.K.'s National Health Service (NHS) identified specific -clinically vulnerable‖ individuals thought to be at higher risk of severe COVID-19 complications and related deaths, and strongly advised them to stay home and avoid all face-toface contact. The entire elderly population, regardless of individual medical conditions, was also considered clinically vulnerable and advised to stay home as much as possible (Public Health England, 2020; Cabinet Office, 2020).
Although effective in preventing a further dissemination of COVID-19, these interventions were immensely disruptive to people's social connections and had potential repercussions on sectors with high direct face-to-face contact-e.g., the healthcare industry and social services (Bu et al., 2020). Vulnerable groups such as older people encountered unique and remarkable challenges in coping with their care needs without leaving their homes (Age U.K., 2020).
In the U.K., elderly support is dependent upon a combination of informal and formal care: statutory-source community care and social services, privately paid care workers, neighbours, friends, and family members (Vlachantoni et al., 2015;Maplethorpe et al., 2015). The strict restrictions introduced by the U.K. government, together with the reorganization of the healthcare system at all levels, produced a disruption in both types of caregiver activities (Topriceanu, et al. 2021).  1 However, an investigation into the effects of COVID- 19 and its accompanying control measures on formal and informal care disruptions, on elderly unmet care needs, and health-related outcomes (i.e., physical, and mental health) has remained relatively scant.
Relying on data from the Understanding Society COVID-19 Survey (April 2020) during first the COVID-19 wave across the U.K, Evandrou et al. (2020) provided the first descriptive evidence on informal care disruptions affecting the elderly during this time. The authors investigated the extent of support received by older people from family, friends, and neighbours in the first period of the lockdown. According to their findings, a significant proportion of older people received an increased level of help (ranging from shopping, dressing, meal preparation, assisting with online or internet access, gardening, or house repairs) from those who had provided care to them before the outbreak or from new caregivers. This was especially the case among those living alone or with a partner aged 70 and over. However, Evandrou et al. (2020) also showed that a smaller group of frail elderly people with difficulties in performing key activities of daily living suffered from an informal care disruption and received less care and support during the lockdown compared to the pre-COVID-19 outbreak period. This evidence raised the spectre that a group of older vulnerable individuals might not have received an adequate level of social care during the lockdown.
Tur-Sinai et al. (2021) investigated how the initial outbreak influenced the supply of formal and informal care among the elderly in need in 23 European countries and Israel by using data from the Survey of Health, Ageing and Retirement in Europe (SHARE Corona Survey), again adopting a descriptive approach. According to their findings, in the first months of the outbreak, informal care appeared to be more resilient than formal care services; indeed, a significant proportion of older adults in European countries continued to receive informal help, enjoying an increase in the amount of care from children, neighbours, friends, and colleagues, while informal help from other relatives decreased. Alternatively, older adults encountered great difficulty in obtaining formal help from professional caregivers. Brugiavini et al. (2022) investigated whether the disruption of elderly parent-adult child contacts due to social distancing restrictions, which characterized European countries during the first wave of the pandemic, increased symptoms of depression in the elderly, using the eighth wave of the SHARE and the SHARE Corona Survey. They adopted a joint model of parent-child contact disruption and mental health issues, estimated by using a recursive bivariate probit model. Their findings showed that interventions deemed essential to reduce the spread of the pandemic, including physical distancing and other epidemiological control measures (e.g., stay-at-home orders, travel restrictions, and so forth), disrupted some personal parent-child contacts, with negative consequences on the elderly parents' mental health.
To the best of our knowledge, no studies have been conducted on the connection between disruption of formal care and its potential impact on the elderly population's mental health, nor on J o u r n a l P r e -p r o o f the inter-relationship between formal and informal care disruptions due to lockdown restrictions and older adults' mental-health deterioration in the U.K context. This paper aims to fill this gap by providing additional insights regarding the short-term consequences of mental health care disruptions to the elderly during the COVID-19 outbreak on the elderly. The empirical evidence provided by this paper may shed light on the importance of designing public policies to contain pandemic crises with the realization that some population groups are more affected than others.
Hence, these groups need different social restrictions from those imposed on the general population since they may suffer more from the consequences of isolation and reduction in social contacts (Gulland, 2020;Carers UK, 2020 Our findings show that the disruption of informal and formal support represents a significant risk factor for psychological well-being in older adults and increases their risk of depression.

Data
This study uses individual-level data from the U.K. Household Longitudinal Study (U.K. HLS), Understanding Society, a nationally representative panel study of the British population. For the HLS, sample members living in the U.K. were interviewed annually since 2009 with the aim of recruiting over 100,000 individuals in 40,000 households. The first wave of the study and data collection period spanned two years and thus wave #1 ran from 2009 to 2011, wave #2 from 2010 to 2012, and so on. Since April 2020, a subsample of participants from the U.K. HLS survey have been interviewed each month, and they completed short web surveys that focused on the impact of the COVID-19 pandemic. The short web surveys covered the changing impact of the pandemic on the welfare of individuals and households. Each month, participants completed one survey that included core content designed to track changes alongside variable updated content as the coronavirus situation developed. Core modules included detailed information on household composition, coronavirus illness, long-term health conditions management, mental health measures, loneliness, and employment. Individuals were identified by a personal unique identifier that J o u r n a l P r e -p r o o f remained for all waves and could be used to link respondents' information across different waves (Institute for Social and Economic Research, 2020).
The integrated data set used for this analysis is the result of matching wave #9 (2017-2019) and #10 (2018-2020) of the main survey and the first month of the COVID-19 wave (April 2020).
This data set provided us the opportunity of gathering information related to the COVID-19 outbreak and the years before it.
After correcting for missing values, the sample included 3,721 individuals. In this paper, we focused specifically on individuals aged 65 and over and found that the COVID-19 pandemic took a heavy toll on their physical as well as mental health. The measures adopted by the U.K. government regarding social distancing and isolation to protect the elderly from risk of infection often resulted in social isolation and loneliness (to which older adults are more vulnerable because of their functional dependency) that in turn might have increased their likelihood of depression (Banerjee, 2020).

Dependent Variables
As previously discussed, the main aim of this study was to investigate the potential effects of informal and formal care disruptions on the mental health deterioration of older people in the U.K during lockdown restrictions intended to curb the COVID-19 spread.
The first step toward a full understanding of this effect required a complex model that considered the simultaneous relationships between informal and formal care disruption and older individuals' psychological well-being. Following Brugiavini et al. (2022), we employed a simultaneous equation model for binary variables. We constructed a joint model of informal and formal care disruption and mental health outcomes that we estimated through a recursive multivariate probit model that considers individuals' unobserved heterogeneity that may characterize these relationships (see Subsection 3.2). 2 Thus, we identified two classes of dependent variables: informal and formal care reception and mental health outcomes-i.e., older individuals' psychological distress. To measure individuals' psychological distress, we used the 12-item Generalised Health Questionnaire (GHQ-12), which is one of the most widely used screening tools for psychological distress that has been validated for epidemiological studies (Goldberg et al., 2 A recursive model is a special case of a system of equations in which the endogenous variables are determined in sequence. Thus, the right-hand side of the reduced-form equations for the endogenous variables include exogenous variables only. The right-hand side of the structural equation includes the exogenous variables and the endogenous variables estimated by the reduced-form equations. The model's development may be traced back to the pioneering work of Heckman (1978), and it is a common approach to deal with the endogeneity of binary dependent variables.
J o u r n a l P r e -p r o o f 1997). The GHQ-12 was collected in all waves of the U.K. HLS Understanding Society to date and included in the Understanding Society COVID-19 Survey. Each one of its 12 items regarding symptoms, feelings, or behaviours is answered on a four-category Likert scale ranging from -not at all‖ to -much more than usual‖: categories 1 and 2 (-not at all,‖ -no more than usual‖) were scored as 0, and categories 3 and 4 (-rather more than usual,‖ and -much more than usual‖) were scored as 1. 3 Finally, the scores from the 12 items were added to obtain an overall score. The measure attained in this way is called GHQ-12 Caseness and respondents scoring 3 or more (out of a possible total of 12) are likely to be experiencing anxiety and/or depression (Cox et al., 1987). In How has the help and support you receive from family, friends or neighbours who do not live in the same house/flat as you changed?‖ (Response options included: -1. There has been no change; 2. I receive more help from some people who previously helped me; 3. I receive less help from some people who previously helped me; 4. I currently receive help from family, friends or neighbours who did not previously help me‖). To capture a potential disruption in informal care, we constructed a binary variable that takes the value of 1 if respondents reported they had not received informal care in the last 4 weeks before the interview (from non-cohabiting family members, neighbours, or friends), but they had received help before the outbreak, or if they had received less help from certain people who previously helped them, and 0 otherwise (if they had received support in the last 4 weeks before the interview, or if they had not received support in the last 4 weeks before the interview, but there has been no change with respect to the pre-outbreak period).
In reference to formal care (i.e., community health and social care services), the Understanding Society COVID-19 Survey asked respondents -in need‖ of formal care to report whether they had received help with personal care/medications/shopping/cooking/cleaning/wound dressing/injections from someone visiting them at home regularly before the pandemic restrictions. 5 The answers ranged from 1 to 4, specifically: -1. Yes, as before; 2. Yes, but with reduced support; 3. Yes, with increased support; 4. No.‖ We constructed a binary indicator that takes a value 1 if respondents, who needed formal care, reported they had experienced a reduction in community health and social care services in 2020, or they did not receive any services compared to the prepandemic period, and 0 otherwise.
According to Evandrou et al., (2020) a relatively low proportion of the elderly reported a disruption in informal care and formal care received during the first COVID-19 wave. Indeed, about 4% of the elderly in our sample experienced a disruption in informal care received, while about 3% reported a disruption in formal care.

Estimation Method
Identifying an association between formal and informal care disruption and the mental health of the elderly may be complicated by the presence of endogeneity. Older individuals' isolation, resulting from the U.K. government restrictions to contain the virus, might have increased the risk of depression while simultaneously influencing access to formal and informal support Hence, we identified two classes of dependent variables: care disruption-namely, formal, and informal care-and health outcome (i.e., the dummy indicator for individuals' mental health as measured by the GHQ-12 Caseness score). In the structural equation for mental health, formal and informal care disruption are included as regressors.
chronic bronchitis, chronic obstructive pulmonary disease, cystic fibrosis, hypothyroidism or an under-active thyroid, cancer, diabetes, epilepsy, high blood pressure/hypertension, emotional, nervous or psychiatric problem, multiple sclerosis, H.I.V., chronic kidney disease, conditions affecting the brain and nerves, motor neurone disease, learning disability or cerebral palsy, problems with spleen, obesity, other long standing/chronic condition), or were having/waiting for treatment at the time of the interview (such as an operation or procedure planned, targeted therapy, tests/consultations). In the abovementioned setting, the exogeneity condition is stated in terms of the correlation coefficients, which can be interpreted as the correlation between the unobservable explanatory variables of the different equations. All equations in system (1) can be estimated separately as single probit models only in the case of independent error terms (i.e., the coefficient is not significantly different from zero).
Conventionally, the identification of a recursive multivariate probit model has been based on exclusion restrictions to obtain a more robust identification of the parameters. Maddala (1983) proposed that at least one of exogenous variables (i.e., in the vectors x 1i and x 2i ) of the reduced-form

J o u r n a l P r e -p r o o f
This subsection describes the exclusion restrictions that we adopted for both reduced-form equations.  6 We also include in the reduced-form equation for informal care disruption a binary variable that takes the value of 1 if none of the respondent's friends live in his or her local area.

a. Disruption of informal care equation
We gathered this information from wave #9 in the -Family Networks‖ and -Social Network‖ modules, respectively (that were not included in the most recent waves #10 and the COVID-19 Survey), by assuming that non-proximity with children and friends remained broadly constant over time.

b. Disruption of formal care equation
While the U.K.'s NHS provides universal healthcare, the provision of publicly funded formal long-term care (LTC) services is based on a needs assessment (i.e., whether the potential care recipient can eat, wash, or dress without help) and means assessment (i.e., income that includes pensions, benefits, and assets), and it is a statutory responsibility of local authorities. In cases where care needs do not meet the criteria or financial means are above the threshold, formal care services should be privately purchased: individuals being cared for (or their family) pay all or most of the costs for their care.
In the last decade, the means test has become meaner, and the usage rate of social services has declined. Among those who must pay for themselves, cost was often cited as a reason for not seeking help (AgeUK, 2022). The pandemic further exacerbated this affordability challenge for many older households, and thereby increased their risk of care disruption (Phillipson et al., 2021).
The Social Care Module of the wave #9 of the Understanding Society Survey includes information about who usually manages payment for the care provider. We created a binary variable that takes the value of 1 if the respondents themselves paid for all formal pre-pandemic care services without any support from family, friends, or local authorities. We expect that those who did not receive any support in paying for the costs of services might have significantly suffered from worse care access and a higher probability of formal care disruption. Table 1 shows the other independent variables in the three equations model of (1), grouped into listed categories.

Other Independent Variables
[ Table 1] For our study, we considered the following categories: demographics, socioeconomic variables, and health conditions that existed before the COVID-19 outbreak. Among demographics, we included the respondent's gender (1: male; 0: female), age, rural living (1: rural area; 0: urban area), area-level context captured with regional fixed effects (i.e., Wales, Scotland, Northern Ireland, and English region), and type of household categorized into single-household living vs.
living with a partner. We also included an indicator of social capital and two COVID-related variables: one in the NHS Shielding category, and the other related to changes of individuals' mobility due to COVID-19.
Among the socioeconomic variables, we included an indicator of respondents' living standards that may influence the probability of psychological distress, the probability of accessing (2) medium education (level 3 in the U.K. education system or equivalent qualifications); and (3) higher education (i.e., levels 4-7 in the U.K. education system).
To account for respondents' -needs‖ unrelated to the pandemic itself and the associated lockdown, we also included information on their health status before the outbreak (U.K. HLS wave #10      7 Arguably, individuals with pre-existing poor health conditions were more likely to suffer from care disruptions because they had been using care before the pandemic.) 8 The inclusion restrictions required that the indicator variables of non-proximity with children and friends, and the variable that measures the absence of any support in affording the costs of social care, should be correlated with informal care and formal care disruptions respectively at the individual level. Our estimation results of the formal and informal care disruption equations (see Table 3) confirm that the estimated marginal effects are all statistically significant at the 5% level, indicating that the inclusion restrictions are clearly met (Li et al., 2021).

Results and Discussion
In terms of socioeconomic status, perceived lower financial stability is associated with disruption in both informal and formal care even though the marginal effects are relatively low; moreover, according to our results, a higher education level positively influences informal care disruption only, with a marginal effect of about 2%. Arguably, a higher level of education raises awareness of the virus and may be positively associated with engagement in all types of preventive behaviours-including complying with stay-at-home rules. This implies a higher probability of inperson contact disruption and consequently the informal care provision particularly among the oldest population that is more vulnerable to COVID-19 infections (Li et al., 2020).
In reference to the structural equation (Column 3 in Table 3 Respondents' altruistic attitude, proxied by charitable donations in our study, contributes negatively to older adults' psychological wellbeing This is consistent with recent research on altruism and mental health during the outbreak, suggesting that altruism does not serve as a protective mental health factor against the threat of COVID-19, as highly altruistic individuals are more likely to feel anxious and depressed due to their empathy towards infected people, and to the impossibility of helping others due to self-isolation regulations (Feng et al., 2020).
We estimate that a reduction of one standard deviation in the combined Google mobility index is associated with an increase of 1.4% in the probability of suffering from depression, which suggests that mobility limitations, as reflected by a decrease of movements, increases the likelihood of suffering from psychological distress. 9 Finally, there exists a positive correlation between pre-existing health conditions, psychological distress (as measured by the SAH and GHQ-12 in 2019, respectively), and worsening mental health.
As previously discussed, we constructed a simultaneous equation model for three binary variables. The multivariate probit estimation allowed us to test for unobserved heterogeneity that may characterize the relationship between informal and formal care disruption and individuals' psychological distress. The unobserved heterogeneity is captured by the correlation between the error terms from the single equation models. Table 4 shows the correlation coefficients for the full recursive model. The null hypothesis of exogeneity is rejected in only one case. According to our results, there exists a negative statistically significant correlation between the disturbance of the formal care disruption equation and the structural equation for individuals' psychological distressi.e., unobservable variables that increase the likelihood of depression and decrease the probability of disruption in formal care provisions. Arguably, the inability to access social support services due to COVID-19 contributes to worsening anxiety and depressive symptoms especially among the elderly affected by pre-existing mental health conditions. As such, the virus increases their demand of formal care support that in turn decreases the likelihood of formal care disruption.

Conclusions
In this paper, we investigated how informal and formal care disruption due to the COVID As lesson for future pandemics, the potential impact of the disruption of long-term care on older individuals' mental health should be considered. Indeed, the possible benefits of mandatory lockdown in curbing the virus spread need to be carefully weighed against the potential psychological health costs. Successful use of isolation as a public health measure requires a realistic reduction in the negative effects associated with it, especially among more vulnerable groups.
One limitation of our data set is that it did not allow us to study possible differences of the disruption impacts related to territories, age groups, and gender. The sample size must be larger to implement heterogeneity tests. This is left for future research.  J o u r n a l P r e -p r o o f   Legend: * = 10% significance level, ** = 5% significance level, *** = 1% significance level