Abstract
Objectives To determine whether clinically extremely vulnerable (CEV) children or children living with a CEV person in Wales were at greater risk of presenting with anxiety or depression in primary or secondary care during the COVID-19 pandemic compared with children in the general population, and to compare patterns of anxiety and depression during the pandemic (23rd March 2020–31st January 2021, referred to as 2020/21) and before the pandemic (March 23rd 2019–January 31st 2020, referred to as 2019/20), between CEV children and the general population.
Design Population-based cross-sectional cohort study using anonymised, linked, routinely collected health and administrative data held in the Secure Anonymised Information Linkage Databank. CEV individuals were identified using the COVID-19 Shielded Patient List.
Setting Primary and secondary healthcare settings covering 80% of the population of Wales.
Participants Children aged 2–17 in Wales: CEV (3,769); living with a CEV person (20,033); or neither (415,009).
Primary outcome measure First record of anxiety or depression in primary or secondary healthcare in 2019/20 and 2020/21, identified using Read and ICD-10 codes.
Results A Cox regression model adjusted for demographics and history of anxiety or depression revealed that only CEV children were at greater risk of presenting with anxiety or depression during the pandemic compared with the general population (Hazard Ratio=2.27, 95% Confidence Interval=1.94–2.66, p<0.001). Compared with the general population, the risk amongst CEV children was higher in 2020/21 (Risk Ratio 3.04) compared with 2019/20 (Risk Ratio 1.90). In 2020/21, the cumulative incidence of anxiety or depression increased slightly amongst CEV children, but declined amongst the general population.
Conclusions Differences in the cumulative incidences of recorded anxiety or depression in healthcare between CEV children and the general population were largely driven by a reduction in presentations to healthcare services by children in the general population during the pandemic.
Strengths and limitations of this study
Strengths of this study include its novelty, national focus and clinical relevance; to date this is the first population-based study examining the effects of the COVID-19 pandemic on healthcare use for anxiety or depression amongst clinically extremely vulnerable (CEV) children and children living with a CEV person in Wales
We compared 2020/21 data with pre-pandemic 2019/20 data for CEV children and children in the general population, to place the impact of the COVID-19 pandemic in the context of longer-term patterns of healthcare use
We used a novel approach and linked multiple datasets to identify a cohort of children living with a CEV person in Wales during the COVID-19 pandemic
There was heterogeneity within the Shielded Patient List that was used to create the cohorts of children identified as CEV or living with a CEV person, in terms of the type and severity of individuals’ underlying conditions; the manner in which people were added to the list; the time point that people were added to the list; and the extent to which people followed the shielding guidance
Routinely collected healthcare data does not capture self-reported health, and is likely to underestimate the burden of common mental disorders in the population
Introduction
In March 2020, Welsh Government and the Welsh National Health Service sought to protect people deemed clinically extremely vulnerable (CEV) to severe illness or death from COVID-19, advising them to ‘shield’ at home i.e. to remain indoors and minimise contact with others.1 To identify CEV people, a Shielded Patient List (SPL) was created, using an algorithm based on clinical code lists and applied centrally to patients’ electronic health records.2 Additionally, health professionals could add people to the list based on their clinical judgement. Shielding was in place from 23rd March 2020 until 16th August 2020 and re-introduced from 22nd December 2020 until 1st April 2021.1 CEV children were encouraged to return to school at the end of August 2020 taking into account the low rate of severe disease and death from COVID-19 amongst children, balanced against the harms of a lack of schooling and socialization.3
Studies have highlighted the detrimental impact of the COVID-19 pandemic on the mental health of CEV adults, with CEV individuals more likely to report increased depressive symptoms and anxiety4 and to have a clinical record of anxiety or depression during the pandemic compared with those who were not CEV.1 Meanwhile, studies have reported decreased diagnoses of mental health conditions in primary care across the population as a whole during the pandemic5,6 (attributed to a reluctance to seek healthcare, or reduced access to services, rather than a decrease in need). However, there is limited evidence on how children accessed healthcare during the pandemic for their mental health, and no studies focusing on CEV children, or children living with a CEV person.
Children are particularly vulnerable to indirect impacts of the pandemic.7 Drawing on evidence from longitudinal surveys,8–10 the Department for Education concluded that children’s mental health declined during the pandemic, reporting that rates of probable mental health disorders were higher from 2020 than before.11 The data also highlighted variation in mental health trajectories; children with long-term health conditions were more likely to experience poor mental health during the pandemic.11 However, the extent to which these trends are attributable to the pandemic, or are a continuation of pre-existing upward trends is unclear.
Almost 5,000 CEV children were living in Wales by July 2020; approximately 3.9% of the Welsh CEV population.12 Children with chronic illnesses are at increased risk of behavioural and emotional problems and psychiatric disorders compared with their peers,13 but CEV children may be particularly susceptible to mental health difficulties relative to non-CEV children since the pandemic, due to additional restrictions imposed by shielding guidance, potentially exacerbating loneliness and isolation.14 CEV children may have also experienced heightened health anxiety due to their potential higher risk of severe illness from COVID-19. Additionally, there were almost 14,400 school-aged children living with a CEV person in June 202015 who may be at greater risk of mental health difficulties due to both restrictions to protect the vulnerable members of their household, and fears of causing harm.16
We investigated the impact of the COVID-19 pandemic on use of healthcare services for anxiety or depression in Wales, for CEV children, children living with a CEV person, and children in the general population, using routinely collected population-level linked data. The primary aim was to determine whether CEV children or children living with a CEV person were more likely to have a record for anxiety or depression in primary or secondary care during the pandemic compared with children in the general population. The secondary aim was to compare patterns of anxiety or depression in 2019/2020 and 2020/2021 between CEV children and the general population, to place the impact of COVID-19 and the shielding guidance in the context of longer-term patterns of healthcare use.
Methods
Study design and data source
This is a population-based cross-sectional cohort study using anonymised, linked, routinely collected health and administrative data for the population of Wales, UK, held in the Secure Anonymised Information Linkage (SAIL) databank (www.saildatabank.com). Within the SAIL databank, encrypted linkage fields are used to link data anonymously from various sources at individual and household level (appendix pp 1–2); known as Anonymised Linking Fields (ALFs) for individuals and Residential Anonymised Linking Fields (RALFs) for residences.17,18 We used these to link multiple datasets in this study (Table 1). General Practices (GPs) opt-in to providing data to SAIL; currently, SAIL contains primary care data for around 80% of the Welsh population, and the available data are representative of the entire Welsh population with respect to age, sex, and deprivation.19 The SAIL databank was interrogated using DB2 structured query language (SQL).
Ethical Approval
This study used anonymised data and therefore did not require National Research Ethics Committee approval. Approval to access and link the data within SAIL was granted by the Information Governance Review Panel under project number 1265 (appendix p 3).
Patient and Public Involvement
The study protocol was presented at the SAIL consumer panel meeting prior to study commencement. This panel consists of members of the public with an interest in data and its uses to improve services and healthcare. The panel provided advice and feedback on the study design from a public perspective.
Data Access and Cleaning Methods
All authors had full access to all the data in the study. Data cleaning included deduplication and restructuring of the SPL prior to cohort creation and analysis, and was undertaken by LEC.
Role of the funding source
The funders had no role in the study design, data collection, data analysis, interpretation, writing of the report, or the decision to submit the paper for publication.
Study population and setting
We created three study cohorts for 2020: 1) CEV children 2) children living with at least one CEV person and 3) a general population group of children who were not identified as CEV or living with a CEV person, along with two further cohorts for 2019 for comparison purposes. Figure 1 shows a flow diagram of the inclusion criteria for each cohort. We included all children aged 2–17 years who were alive, living in Wales and registered with a SAIL-supplying GP on 23rd March 2020; and who had either an exact match on NHS number or demographics (name, date of birth, gender code, and postcode), or a probabilistic match of 90% or greater.17 We excluded those for whom full demographic or residence data were not available.
Flow diagram of the inclusion criteria for the creation of three study cohorts: clinically extremely vulnerable children, children living with a clinically extremely vulnerable person, and children in the general population. CEV, clinically extremely vulnerable; CVSP, COVID-19 Shielded People List; SAIL, Secure Anonymised Information Linkage; GP, general practice; RALF, residential anonymised linking field; NHS, national health service.
Cohort 1: Clinically extremely vulnerable children
Cohort 1 consisted of all children who were identified as CEV (either by algorithm2 or health professionals) between 23rd March 2020 and 16th August 2020 (N=3,769).
Cohort 2: Children living with at least one clinically extremely vulnerable person
To identify children living with a CEV person, we first identified the RALF for all CEV people in Wales as of 23rd March 2020, including dates of residence. To minimize bias, we then adopted a conservative approach and included 1) children who were recorded as residents at the same RALF on 23rd March 2020 and had an entry date of residence within 6 months of the entry date of the CEV person 2) children born to mothers who were recorded as resident at the same RALF as the CEV person, or 3) children who shared the same maternal ALF as a younger CEV child. We excluded 1) care homes20 and 2) RALFs containing more than ten people21, with a final 20,033 children in cohort 2.
Cohort 3: Children in the general population
Cohort 3 consisted of all other children who were alive, living in Wales and registered with a SAIL-supplying GP on 23rd March 2020 (N=415,009).
Clinically extremely vulnerable children in Wales in 2019 (pre-COVID-19 CEV children)
To explore longer-term patterns of anxiety or depression amongst children with the conditions included within the shielding guidance, we created a cohort of pre-COVID-19 “CEV” children who had similar health concerns to CEV children in a period prior to 2020. We included children who were alive, living in Wales, and registered with a SAIL-supplying GP one year prior to the introduction of shielding (23rd March 2019) and who had one or more of three of the health condition categories warranting inclusion on the SPL (respiratory illnesses, blood or bone cancer, and immunosuppression therapy). These categories were chosen as they only required the patient to have one of the listed codes within a given time period and therefore children with these conditions could be identified with a high degree of certainty. We used International Classification of Diseases (ICD) version 10 diagnostic codes and OPCS Classification of Interventions and Procedures version 4 codes for these categories, which were taken from the SPL documentation and provided in the appendix (pp 4–5). For comparison purposes we also created a general population cohort of children aged 2–17 who were alive, living in Wales, and registered with a SAIL-supplying GP as of 23rd March 2019. A flow diagram of inclusion criteria for these cohorts is provided in the appendix (p 6). We performed a sensitivity analysis to confirm the validity of this approach (appendix pp 7–9).
Measures
Outcome of interest: Anxiety or depression
The outcome of interest was the first record of anxiety or depression in primary or secondary healthcare data during the COVID-19 pandemic (i.e. 23rd March 2020–31st January 2021, referred to as 2020/21), and pre-pandemic (March 23rd 2019–January 31st 2020, referred to as 2019/20). We used validated Read v2 codes to identify children with primary care records for anxiety22 or depression23 (including diagnoses, symptoms, and prescriptions) in the WLGP dataset, and ICD-10 diagnostic codes to identify children with hospital admissions or outpatient appointments for anxiety or depression in the PEDW and OPD datasets, respectively. Code lists were reviewed by a clinician with expertise in child psychiatry (JT) and are provided in the appendix (pp 10–14).
Covariates: History of anxiety or depression
We used the same process to identify children in the study population with a ‘recent’ or ‘past’ history of anxiety or depression, defined as any record in the year prior to the pandemic (23rd March 2019–23rd March 2020), and any record occurring any time before 23rd March 2019 respectively.1
Covariates: Demographics
We calculated age and determined Lower-layer Super Output Areas (LSOA) as of 23rd March 2020. LSOA codes were derived from the WDS dataset based on the child’s RALF, and used to ascertain deprivation quintiles and urban/rural classification by linking to the WIMD 201924 and RUC datasets.
Statistical analysis
We used R version 4.1.1 for statistical analyses. P-values of <0.05 were considered statistically significant.
Examining risk of anxiety or depression between different populations in 2020/21
We tested the hypothesis that there was no difference in the risk of having a record of anxiety or depression in 2020/21 between the three cohorts (CEV children, children living with a CEV person, and children in the general population). We plotted Kaplan-Meier survival curves for each cohort. We used Cox regression to calculate unadjusted and adjusted hazard ratios (HRs) with 95% confidence intervals (CI). We report three models examining the risk of anxiety or depression compared with the general population; (1) unadjusted, (2) adjusted for demographic factors (age group, sex, deprivation and rurality) and (3) adjusted for demographic factors and previous history of anxiety or depression (no history, recent history, past history, or both recent and past history).
Examining risk of anxiety or depression between different populations in 2019/20
We tested the hypothesis that there was no difference in the risk of having a record of anxiety or depression in 2019/20 between two cohorts (“CEV” children and all other children living in Wales in 2019). As above, we calculated unadjusted and adjusted HRs, and calculated the change in cumulative incidences of depression and anxiety for these two groups between 2019 and 2020.
Study reporting
This study is reported in accordance with the Reporting of Studies conducted using Observational Routinely-collected data (RECORD) guidelines25 (appendix pp 15–23).
Results
Descriptive statistics and demographic characteristics of the study population
Demographic characteristics of the 2020/21 study population are presented in Table 2, and for the 2019/20 study population in the appendix (p 24). For both years, there were greater proportions of boys, older children (aged 13–17), children living in the least and most deprived quintiles and children with a history of anxiety or depression in the CEV children than for the general population. In 2020/21, a greater proportion of children living with a CEV person were older (aged 13–17), and had a history of anxiety or depression, compared with the general population (Table 2). The conditions leading to children being identified as CEV in 2020/21 are shown in Table 3.
In children whose reasons for shielding were known (3,165/3,769) 6.3% (198/3,165) had more than one condition
Risk of anxiety or depression in children in 2020/21
Numbers of censored children in each group (due to death, migration or registration with a non-SAIL supplying GP) and numbers with a record for anxiety or depression are presented in Table 4. Of those with a record, 5,768/6,251 (92.3%) presented to primary care while 483/6,251 (7.7%) presented to secondary care.
Unadjusted and adjusted hazard ratios are presented in Table 5. In the unadjusted model (1), both CEV children and children living with a CEV person were at significantly greater risk of anxiety or depression compared with the general population (HR=3·09, 95% CI=2·64–3·61, p<0·001 and HR=1·16, 95% CI=1·04–1·30, p<0·05, respectively). For CEV children, the hazard ratio remained significant in both model (2) and (3). However, for children living with a CEV person, the hazard ratio was no longer significant in either model (2) or (3). Full results of the adjusted models are included in the appendix, pp. 25–26. The unadjusted survival curves for each cohort are shown in Figure 2.
Kaplan-Meier survival curves for each cohort, showing the time to first record of anxiety or depression between 23rd March 2020 and 31st January 2021 (2020/21)
Risk of anxiety or depression in children in 2019/2020
Unadjusted and adjusted hazard ratios for anxiety or depression among the “CEV” children as compared with the general population between March 23rd 2019–January 31st 2020 are presented in Table 6. In 2019/20, in the unadjusted model (1) there was an increased risk of anxiety or depression amongst the “CEV” children compared with the general population (HR=1·94, 95% CI=1·31–2·87, p<0·001). This remained evident in the adjusted models (2) and (3). Full results of the adjusted models are included in the appendix, pp. 27–28. The unadjusted survival curves for each cohort are shown in Figure 3.
Kaplan-Meier survival curves for each cohort, showing the time to first record of anxiety or depression between 23rd March 2019 and 31st January 2020 (2019/20)
Difference in the risk of records of anxiety or depression between 2019/20 and 2020/21
In 2019/20, “CEV” children had increased risk of having recorded anxiety or depression compared with children in the general population, and in 2020/21 the risk ratio increased to 3.04 (Table 7). This reflects a marked decline in presentation amongst children in the general population over this period (from 2·19% to 1·39%), alongside a small increase for CEV children (from 4·17% to 4·22%).
Discussion
In Wales, CEV children and children living with a CEV person were more likely to access health services for anxiety or depression during the pandemic than children in the general population. For CEV children, this pattern remained evident after adjusting for demographic differences and the likelihood of having a previous history of anxiety or depression. Although a small increase in risk was found for children living with a CEV person, after adjusting for demographic characteristics and previous history of anxiety and depression, this was no longer significant.
Given the detailed methodology used to identify CEV individuals in Wales,2 we were able to develop a comparable cohort of children with a subset of health conditions before the pandemic, using routine healthcare data. This enabled us to examine patterns of presentation for anxiety and depression amongst CEV children outside of the context of the pandemic. We found this group were at greater risk of having a record for anxiety or depression compared with children in the general population in 2019/20, before COVID-19. In 2020/21, CEV children remained at higher risk, and the difference was greater, although this is explained by a marked decline amongst children in the general population presenting to healthcare services with anxiety or depression during this time.
The reduction in presentations for anxiety and depression amongst children in the general population most likely reflects reduced access to NHS services during the pandemic. Other evidence suggests increased demand and unmet need for mental health support in the UK, for children with and without pre-existing mental health problems, since 2020.26,27 Findings from self-report surveys8–10 and the current study suggest that the pandemic has widened the gap between need and access to mental healthcare for the general population of children in Wales, but additional data are required to unpack the relationship between self-reported mental health needs and presentation to healthcare services.
Meanwhile, the relatively stable cumulative incidences of anxiety and depression for CEV children in 2019/20 and 2020/21 could indicate that this group did not experience an increase in mental health needs during the pandemic over and above past years, and that they had access to mental health support through existing care pathways for underlying conditions.
Alternatively, if CEV children experienced the same increase in mental health needs as reported elsewhere amongst the general population, then these figures may mask unmet demand for mental health support among this group. A survey of adults supports the latter explanation, reporting increased anxiety and depressive symptoms amongst shielding individuals compared with non-shielding individuals.4 However, we have found no UK surveys focusing on the mental health of CEV children.
Our finding that children living with a CEV person were at no greater risk of presenting with anxiety or depression during the pandemic compared with the general population (after controlling for other factors), could be interpreted in two ways. It is possible that the impact of the pandemic on the mental health of children living with a CEV person was not as great as for CEV children, but this seems unlikely given that research has suggested increased anxiety amongst children who were shielding their siblings.16 Another explanation is that children living with a CEV person suffered from similar barriers to access to mental healthcare services to the general population, and did not have the same routes to access that CEV children did. This explanation is supported by research with shielding families, which suggests that they have felt left behind, and that children living with a CEV person may have ‘fallen under the radar of educational and healthcare professionals’.16
Strengths and limitations
To our knowledge, this is the first population-based study examining the effects of the COVID-19 pandemic on healthcare use for anxiety or depression amongst CEV children and children living with a CEV person in Wales. Linkage of population-based routinely collected data is a valuable method for generating evidence with a high level of external validity and applicability for policy making. A strength of this study is the comparison of 2020/21 data with pre-pandemic 2019/20 data for CEV children and children in the general population.
Another strength is the use of a novel approach using multiple linked datasets to identify a cohort of children living with a CEV person during the pandemic.
This study used the SPL to create cohorts of children identified as CEV and a cohort of children living with a CEV person. There was heterogeneity within the SPL in terms of the type and severity of individuals’ underlying conditions; the manner in which people were added to the list (via the algorithm or clinical judgement); the time point that people were added to the list; and the extent to which people followed the shielding guidance. In addition, the impact of following shielding guidance is likely to have varied due to individual circumstances and the level of support received. The 2019 “CEV” cohort was a relatively small sample and for pragmatic reasons, only included children with a subset of the conditions included in the shielding guidance. To identify children living with a CEV person, we adopted a conservative approach and included children only if they were living with their mother. We took this approach in order to minimize bias and increase the generalisability of the findings, however this approach is likely to have underestimated the number of children living with a CEV person. Finally, this study focused on healthcare use using clinical codes. Routinely collected healthcare data does not capture self-reported health, and is likely to underestimate the burden of common mental disorders in the population.28 Focusing on healthcare use with routine data alone cannot tell us about the underlying reasons for changes in utilisation, or the scale of mental health need.
Implications for policy and practice
Our findings have implications for recovery planning to prevent, mitigate and respond to the mental health impacts of the pandemic. We have shown changes in presentation to primary and secondary healthcare services with anxiety and depression for CEV children and children in the general population during the pandemic, and there are concerns regarding potential increases in unmet mental health needs over time. As highlighted by UK organisations such as the Centre for Mental Health,29 services face challenges in tackling this demand. This has been recognised by Welsh Government, who invested an additional £9.4m in children’s mental health services in 2021.30
This novel linked data study contributes to our understanding of the direct and indirect impact of shielding on children’s mental health in Wales during the COVID-19 pandemic. This evidence should be considered in light of additional, more detailed routine healthcare linkage studies, and national surveys, to provide a comprehensive understanding of the relationship between mental health support needs, expressed demands and care provision to better target services to those who need them the most.
Beyond the indirect impacts of the pandemic, our findings highlight the increased mental health needs of children with serious medical conditions. Given that these children are likely to have greater contact with healthcare services, signposting across services including mental health services, is likely to be beneficial.
Funding statement
This work was supported by the Adolescent Mental Health Data Platform (ADP). The ADP is funded by MQ Mental Health Research Charity (Grant Reference MQBF/3 ADP). This work was also supported by the National Centre for Population Health and Wellbeing Research (NCPHWR) which is funded by Health and Care Research Wales.
Author contributions
LEC: study design, literature search, data curation, data analysis, figures, data interpretation, writing – original draft, writing – review and editing; KH: data analysis, data interpretation, supervision, writing – review and editing; JS: conceptualisation, study design, methodology, data curation, data analysis, data interpretation, supervision, writing – review and editing; TW: methodology, writing – review and editing; JT: validation, writing – review and editing; AJ: methodology, funding acquisition, writing – review and editing; AB: methodology, writing – review and editing; ARD: conceptualisation, study design, data analysis, data interpretation, supervision, funding acquisition, writing – review and editing. LEC and JS verified the underlying data.
Competing interests’ statement
Ann John is a member of the Welsh Government COVID-19 Technical Advisory Group and is also co-chair of the Scientific Pandemic Insights Group on Behaviours (SPI-B) which is a subgroup of the Scientific Advisory Group for Emergencies (SAGE) advising the UK government. None of the other authors have any competing interests.
Data sharing statement
The data used in this study are available in the SAIL Databank at Swansea University (Swansea, UK) via the Adolescent Mental Health Data Platform, but, as restrictions apply, they are not publicly available. All proposals to use SAIL data are subject to review by an independent Information Governance Review Panel (IGRP). Before any data can be accessed, approval must be given by the IGRP. The IGRP carefully considers each project to ensure proper and appropriate use of SAIL data. When access has been granted, it is gained through a privacy-protecting safe haven and remote access system referred to as the SAIL Gateway. SAIL have established an application process to be followed by anyone who would like to access data via SAIL, details of which can be found at https://saildatabank.com/data/apply-to-work-with-the-data/
Acknowledgements
ADP and the authors would like to acknowledge the data providers who supplied the datasets enabling this research study. The views expressed are entirely those of the authors and should not be assumed to be the same as those of ADP or MQ Mental Health Research Charity. This study makes use of anonymised data held in the Secure Anonymised Information Linkage (SAIL) Databank. We would like to acknowledge all the data providers who make anonymised data available for research.
Footnotes
Author email addresses: l.e.cowley{at}swansea.ac.uk; Karen.Hodgson{at}wales.nhs.uk; Jiao.Song{at}wales.nhs.uk; Anthony.Whiffen{at}gov.wales; Jacinta.Tan{at}oxfordhealth.nhs.uk; A.John{at}swansea.ac.uk; A.Bandyopadhyay{at}swansea.ac.uk; Alisha.davies{at}wales.nhs.uk