*Review* **Ethnic Inequalities in Healthcare Use and Care Quality among People with Multiple Long-Term Health Conditions Living in the United Kingdom: A Systematic Review and Narrative Synthesis**

**Brenda Hayanga 1,\*, Mai Stafford <sup>2</sup> and Laia Bécares <sup>1</sup>**


**Abstract:** Indicative evidence suggests that the prevalence of multiple long-term conditions (i.e., conditions that cannot be cured but can be managed with medication and other treatments) may be higher in people from minoritised ethnic groups when compared to people from the White majority population. Some studies also suggest that there are ethnic inequalities in healthcare use and care quality among people with multiple long-term conditions (MLTCs). The aims of this review are to (1) identify and describe the literature that reports on ethnicity and healthcare use and care quality among people with MLTCs in the UK and (2) examine how healthcare use and/or care quality for people with MLTCs compares across ethnic groups. We registered the protocol on PROSPERO (CRD42020220702). We searched the following databases up to December 2020: ASSIA, Cochrane Library, EMBASE, MEDLINE, PsycINFO, PubMed, ScienceDirect, Scopus, and Web of Science core collection. Reference lists of key articles were also hand-searched for relevant studies. The outcomes of interest were patterns of healthcare use and care quality among people with MLTCs for at least one minoritised ethnic group, compared to the White majority population in the UK. Two reviewers, L.B. and B.H., screened and extracted data from a random sample of studies (10%). B.H. independently screened and extracted data from the remaining studies. Of the 718 studies identified, 14 were eligible for inclusion. There was evidence indicating ethnic inequalities in disease management and emergency admissions among people with MLTCs in the five studies that counted more than two long-term conditions. Compared to their White counterparts, Black and Asian children and young people had higher rates of emergency admissions. Black and South Asian people were found to have suboptimal disease management compared to other ethnic groups. The findings suggest that for some minoritised ethnic group people with MLTCs there may be inadequate initiatives for managing health conditions and/or a need for enhanced strategies to reduce ethnic inequalities in healthcare. However, the few studies identified focused on a variety of conditions across different domains of healthcare use, and many of these studies used broad ethnic group categories. As such, further research focusing on MLTCs and using expanded ethnic categories in data collection is needed.

**Keywords:** ethnic inequalities; healthcare use; care quality; multiple long-term conditions; UK

#### **1. Introduction**

Long-term conditions (also known as chronic conditions) are health conditions that are currently uncurable and consequently are managed with medication and other therapies (e.g., cardiovascular disease, diabetes and depression) [1,2]. In the UK, it is estimated that between 23% and 27% of the population live with two or more long-term conditions, and this number is expected to rise in the coming decades [2–4]. These trends present a challenge not only for individuals but also for society and entire healthcare systems [5,6]. People

**Citation:** Hayanga, B.; Stafford, M.; Bécares, L. Ethnic Inequalities in Healthcare Use and Care Quality among People with Multiple Long-Term Health Conditions Living in the United Kingdom: A Systematic Review and Narrative Synthesis. *Int. J. Environ. Res. Public Health* **2021**, *18*, 12599. https://doi.org/10.3390/ ijerph182312599

Academic Editor: Stuart Gilmour

Received: 27 August 2021 Accepted: 24 November 2021 Published: 29 November 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

with multiple long-term conditions (MLTCs) are more likely to have increased disability, poorer functioning, reduced well-being, lower quality of life and higher mortality [6,7]. The relationship between MLTCs and increased healthcare costs is well documented [8]. Further, the challenges in providing high quality care for people with MLTCs are recognized [9]. People with MLTCs have increased exposure to healthcare services and systems, which are often fragmented and/or tailored towards managing single health conditions, thereby hindering the holistic management of MLTCs [7]. This uncoordinated care may lead to extra obligations for patients and healthcare staff, threats to patient safety and an increase in patient-level frustration [10,11].

This study focuses on people from minoritised ethnic groups with MLTCs and how their patterns of healthcare use and care quality vary from their White counterparts. In line with other studies, we use the term minoritised ethnic group to refer to people who do not self-identify as belonging to the White majority ethnic group [12,13]. Commonly used acronyms such as BAME (Black, Asian and Minority Ethnic) can be exclusionary as they single out specific ethnic groups [14,15]. Other terms such as 'minority' can be associated with diminished status if we consider that, historically, the narrative of 'minorities' marked troubled histories of immigration control, policing, racial violence, inferiorisation and discrimination that were characteristic of daily life for early migrants to the UK from Africa, the Caribbean and Asia [16]. The term 'minoritised' places emphasis on how social positions are social constructions rather than practices and outcomes that are natural and inevitable [17].

There is some evidence to suggest that people from minoritised ethnic groups in the UK are at an increased risk of developing MLTCs when compared to the White majority population, and they are also more likely to develop MLTCs at an earlier age [18,19]. The findings from a recent review indicate a higher prevalence of MLTCs in some minoritised ethnic groups compared to their White counterparts [20]. These ethnic inequalities in MLTCs are likely to reflect broader economic and social inequalities, which in turn are driven by racism and racial discrimination [21,22]. These same mechanisms can lead to inequities in access and use of healthcare and care quality, which can lead to negative outcomes for people with MLTCs [23]. Studies of single conditions report that, in general, people from minoritised ethnic groups are less likely to access specialist services and less likely to report positive experiences of primary care when compared to their White counterparts [24–26]. It is possible that people with MLTCs from minoritised ethnic groups may face similar experiences when using healthcare services. Findings from a recent ethnographic study conducted by Revealing Reality for the Taskforce on Multiple Conditions give insight into how ethnic inequalities in healthcare use and care quality can arise [23]. The study explored the lives of people with MLTCs experiencing health inequity and disadvantage, living in some of the most deprived wards in the UK. This study illustrated how wider societal processes (e.g., deprivation and suboptimal healthcare provision) intersect with individual level processes (e.g., poor literacy skills, language difficulties, competing priorities) to negatively impact people's ability to access and utilise healthcare services, adhere to treatment regimens and ultimately manage their MLTCs [23].

Whilst the aforementioned study gives insight into the experiences of people with MLTCs, including those from minoritised ethnic groups, their focus was not on uncovering ethnic inequalities. It is important to examine ethnic variations in healthcare use and healthcare quality among people with MLTCs. Findings of such an exploration can illuminate ethnic inequalities and inform actions to redress the health disadvantage faced by particular populations [27], which, if ignored, can result in the widening of existent ethnic inequalities. Given the increasing ethnic diversity of the UK population [28], a detailed examination of the association between MLTCs, healthcare and ethnicity in the UK is warranted.

Past reviews of healthcare use and care quality, which have included studies reporting on differences across ethnic groups, have focused on a particular domain of healthcare (e.g., access to healthcare [29]) or health services for a particular group of conditions (e.g., somatic healthcare service related to screening, general practitioners, specialists, emergency rooms and hospital care [30]). In one review, the authors synthesised the best evidence for improving healthcare quality for people from minoritised ethnic groups [31]. However, the focus of these reviews was not on people with MLTCs [29–31]. To our knowledge, no review has synthesised evidence on ethnic inequalities in healthcare use and care quality among people with MLTCs living in the UK. Such an undertaking can highlight areas where inequalities are evident and inform discussions and efforts to address them. Therefore, the aims of this review are (1) to identify and describe the literature that reports on ethnicity and healthcare use and care quality among people with MLTCs living in the UK and (2) to examine how healthcare use and/or care quality for people with MLTCs compares across ethnic groups in studies counting more than two long-term conditions.

#### **2. Methods**

#### *2.1. Search Strategy*

In line with the Preferred Reporting Items for Systematic review and Meta-Analysis Protocols (PRISMA-P) [32], we registered the protocol for this review on PROSPERO (CRD42020220702). Between October and December 2020, we searched the following databases for studies that compared healthcare use and/or care quality across different ethnic groups of people with MLTCs living in the UK: ASSIA, Cochrane Library, EMBASE, MEDLINE, PsycINFO, PubMed, ScienceDirect, Scopus and Web of Science core collection. We also conducted a search on OpenGrey to ensure that relevant grey literature was not excluded. We supplemented the electronic search with a manual search of the key studies identified. We contacted relevant authors when full texts were not available.

We followed the conventions of each search engine and used search terms that denoted the key concepts in this review: Ethnicity (e.g., "Ethnic Groups" [Mesh] OR "BME" OR "BAME"), Multiple health conditions (e.g., "Multiple Chronic Conditions" OR Comorbid\* OR Multimorbidity), Health inequality (e.g., "Health Equity" [Mesh] OR "Healthcare disparit\*" [MeSH] OR Inequalit\*), Healthcare use (e.g., "Delivery of Healthcare" [Mesh] OR "Tertiary Healthcare" [Mesh]), Care quality (e.g., "Quality of Healthcare" [Mesh] OR "Patient Acceptance of Healthcare" [Mesh] OR "Patient Satisfaction" [Mesh]) and the geographical location (e.g., "United Kingdom" [MeSH Terms] OR "UK") (See Appendix A for a full list of search terms).

#### *2.2. Selection Criteria*

We did not restrict the start of the search to any particular period in time and included only UK studies, published in English, reporting on healthcare use and/or care quality among people with MLTCs, across different ethnic groups of people living in the UK [33]. Our justification for focusing on studies in the UK was driven by the recognition that the UK has a unique healthcare system that is publicly funded, with a range of comprehensive services that are (mostly) free at the point of use [34]. Further, it has a diverse minoritised ethnic group population [35]. These factors would complicate comparisons with other countries with different healthcare, political, and economic systems and population structures.

In the extant literature, MLTCs are defined and operationalised in different ways. Some use the term MLTCs synonymously with the term multimorbidity (here defined as the presence of two or more long-term health conditions [3,36]). Others also incorporate the term comorbidity (i.e., the presence of any distinct additional co-existing ailment in an individual with an index condition under investigation [37,38]). Given these definitions, we included studies that counted only two conditions (e.g., diabetes and depression) as well as those that counted two or more long-term conditions. However, to address the second aim we excluded studies that counted only two conditions and focused on those that also counted more than two long-term conditions as they are more likely to give insight into those with complex medical needs and greater use of healthcare [39,40].

Healthcare use and care quality are broad concepts that encapsulate different domains. Healthcare use can be defined as the quantification or description of the use of services by persons for the purpose of preventing and curing health problems, promoting maintenance of health and well-being, or obtaining information about one's health status and prognosis [41]. Indicators of healthcare use include GP consultations, hospital visits including inpatient, outpatient and day visits, hospital admissions, accidents and emergency department visits, diagnoses, prescriptions, referrals, immunisations and screening [29,42,43]. In contrast, healthcare quality has been defined as the degree to which healthcare services increase the chances of desired health outcomes for people and are aligned with current professional knowledge [44]. Indicators of care quality include effectiveness, patient-centeredness, efficiency, equity of care and principles such as acceptability, trust, responsiveness, safety, waiting times, patient experience, satisfaction with accessibility, humaneness of care, number of readmissions and cultural appropriateness [45,46]. We included studies regardless of the domain of healthcare use and care quality under investigation.

We imported the studies retrieved from the electronic search to Endnote X8. We first removed the duplicates. Following this, B.H. and L.B. screened a random sample (10%) of the titles and abstracts. Differences were resolved through discussion. B.H. proceeded to independently screen the remaining studies. The same process was repeated when screening the full texts.

#### *2.3. Data Extraction*

B.H. and L.B. extracted data from a random sample (10%) of the studies identified. Disagreements were settled by discussion. B.H. independently extracted data from the remaining studies. We extracted relevant information from the included studies using a structured form, which included the following items: study identifier, study design, geographical location, data source, sample size, population characteristics (e.g., age and gender profile, ethnic group categories), type and number of MLTCs, confounding variables and healthcare use and care quality domains and results.

#### *2.4. Outcomes*

The outcomes of interest were patterns of healthcare use and care quality among people with MLTCs for at least one minoritised ethnic group, compared to the White majority population.

#### *2.5. Data Analysis*

Owing to the lack of a common definition of healthcare use and care quality, the different domains of healthcare use and care quality assessed, the variety of conditions explored and the different ethnic group categories assessed in the included studies, we conducted a narrative synthesis of the findings. We present the findings of the synthesis in themes, and supplement the reporting with tables and figures. The findings are presented in two sections. First, we provide an overview of the studies that report healthcare use and care quality across ethnic groups of people with MLTCs, including the participant characteristics, domains of healthcare and care quality assessed, and types of health conditions under investigation. Second, we present the evidence of ethnic inequalities in healthcare use among people with MLTCs from the studies that went beyond counting only two long-term conditions. We use the terminology used by authors to describe ethnic categories in their studies.

#### **3. Results**

#### *3.1. Overview of Included Studies*

We identified 621 titles from the electronic search (See Figure 1, which is based on PRISMA guidelines [47]). After removal of duplicates and studies identified as ineligible from the title or abstract, 42 papers were eligible for further evaluation. A further 28 studies were excluded because, despite reporting on the key concepts of interest (i.e., MLTCS, ethnicity, healthcare use), some reported MLTCs and healthcare use separately (*n* = 21), others reported inequalities in healthcare for one health condition (*n* = 5) and others did not compare healthcare use across the different ethnic groups (*n* = 2). Consequently, 14 studies were included in the review, with five of these studies contributing to the evidence on ethnic inequalities in healthcare use in people with MLTCs living in the UK. These were studies in which the authors counted more than two long-term conditions and not just two conditions. The former are more likely to illuminate patterns of ethnic inequality among those with complex medical needs and greater use of healthcare [39,40].

**Figure 1.** PRISMA flowchart [47].

The 14 studies included in this review were published between 2001 and 2021. There were three national studies [48–50] and 11 local studies conducted in Birmingham [51], Leicester [52] and London [53–60]. The number of participants in the included studies ranged from 45 to nearly 61.5 million. The majority of studies used patient records. In eight of the 14 studies, data from primary care records were analysed [48,54,56–61]. The remaining studies used hospital records (*n* = 2) [49,50] and records from specialist services such as Diabetes Outpatient Clinics (*n* = 2) [52,55]. One study used data from the Comorbidity Dual Diagnosis Study [53], and another used data from a community-based Mental Health and Substance Misuse services survey [51].

#### *3.2. Participant Characteristics*

#### 3.2.1. Ethnic Group Identification

Nine of the 14 included studies explicitly reported how ethnicity was identified (64%). Of these, participants self-reported their ethnic identity in seven studies [53–57,60,61]. In one study, ethnicity was assigned by keyworkers [51], and in another study, computerised name recognition software was used to identify South Asian people [52].

#### 3.2.2. Ethnic Group Categorisation

Of the 14 included studies, two compared ethnic variations in healthcare use among people with MLTCs between two ethnic group categories. Of these studies, White people were compared to Black [58] and South Asian people [52]. Three studies categorised their participants into three ethnic group categories [54–56], and two studies compared outcomes across four ethnic groups [53,57]. The remaining studies grouped their participants into five or more ethnic group categories (*n* = 7) [48–51,59–61].

#### 3.2.3. Missing Ethnicity Data

Information concerning missing ethnicity data was available in nine of the 14 included studies (64%). In two of these studies, those with missing ethnicity data were labelled as missing/unknown and included in the analyses [48,49]. In the remaining seven studies, participants with missing ethnicity data were excluded from the analyses [51,54–57,59,60]. One study excluded participants who were of 'Other' ethnicity due to the heterogeneity by ethnicity within the group [60]. Only one study conducted sensitivity analyses to ascertain if the results would differ if those with missing ethnicity data were excluded [48].

#### 3.2.4. Gender and Age

There were 11 studies that reported the gender profile of the participants. One study included only female participants [61], and the remaining ten studies included both male and female participants [48,51–56,58–60]. Of the 14 included studies, six reported the mean age and standard deviation (SD). The average age of participants in these studies ranged from 26.8 (SD = 5.9) years to 66 (SD = 8.5) years [48,52,53,55,60,61]. Four studies included participants aged 18 years and above [51,54,57,59]. In one study, participants were aged 25 years and above [56], and in another, they were aged 16 years and above. [58]. The focus of one study was on children and young people aged between 10 years and 24 years [49], while another study included participants aged 10 years and over [50].

#### *3.3. Domains of Healthcare Use and Care Quality Assessed in Included Studies*

Table 1 below lists the domains and sub-domains of healthcare use and care quality assessed in the included studies. The most frequently assessed domain was disease management/monitoring (*n* = 6). Of these studies, the authors examined ethnic differences in diabetes management and cardiovascular risk factors monitoring among people with MLTCs. These were measured by assessing HbA1c levels, cholesterol levels, smoking status, protein urea levels and Body Mass Index, [52,54,55,57,58,60]. One study assessed ethnic differences in health screening, including mammography and cervical smears, among people with psychosis and comorbidities [58]. There were three studies that reported on ethnic differences in prescriptions among people with MLTCs [55,57,61]. Another three studies reported on the use of hospital services, including admission and length of hospital stay [49,50,53]. Few studies looked at disease progression (*n* = 2), mortality/risk of mortality (*n* = 2) and quality of treatment (*n* = 2). One study assessed the use of Mental Health and Substance Misuse services among people with severe mental health problems who use substances problematically [51].


**Table 1.** Domains and sub-domains of healthcare use and care quality assessed in included studies.

#### *3.4. Studies Reporting on Ethnic Differences in Patterns of Healthcare Use and Care Quality among People with Multiple Long-Term Conditions Living in the UK*

Of the 14 included studies, 12 studies (86%) specified an index condition when reporting on ethnic differences in healthcare use and care quality among people with MLTCs (Table 2). The most frequently cited index conditions were diabetes (*n* = 6) [48,52,54–56,60] and mental health conditions (*n* = 4) [51,53,58,61]. One study focused on people with hypertension [59], and another assessed alcohol-related conditions as a comorbidity [50].

Two studies (14%) did not specify an index condition when examining ethnic inequalities in patterns of healthcare use and care quality among people with MLTCs. Of these studies, one assessed risk factor management among people with cardiovascular multimorbidity [57], while the other assessed emergency admissions and long-term conditions in children and young people [49].

#### *3.5. Evidence of Ethnic Inequalities in Healthcare Use among People with Multiple Long-Term Conditions*

In this review, five studies also counted more than two long-term conditions and are likely to give us insight into people with complex healthcare needs and greater use of healthcare [39,40]. Four studies focused on disease management, and one study focused on use of hospital services, in particular, emergency admissions. It would be inappropriate to combine their results because the studies represent different domains of healthcare use. Consequently, we discuss these two domains separately in the following section.


**Table 2.**

Characteristics

 of included studies.


*Int. J. Environ. Res. Public Health* **2021**, *18*, 12599


HbA1c: Haemoglobin

 A1c; IMD: Index of Multiple Deprivation;

 NICE: National Institute for Health and Care Excellence; NR: Not reported; SD: Standard

Deviation.

3.5.1. Ethnic Inequalities in Disease Management among People with Multiple Long-Term Conditions

The four studies that suggest that there are ethnic inequalities across different domains of disease management among people with MLTCs are local studies that analysed data from primary care records using a cross-sectional study design, where the authors assessed the outcomes at a single point in time [52,54,57,58]. The sample sizes ranged from 1090 participants to 6690 participants, and comparisons were made between White participants and Black [54,57,58], South Asian [52,54,57], Asian [49] and those who self-identified as belonging to Mixed [49,57] and 'Other' ethnic groups [57]. Three of these studies specified an index condition: diabetes [52,54] and psychosis [58]. Mehta and colleagues (2011) assessed the relationship between glycaemic control, chronic disease comorbidity and ethnicity in people with diabetes. They found that among patients with Type 2 diabetes mellitus, the excess odds of having suboptimal glycaemic control (HbA1c ≥ 7%) was 1.86 (95% CI: 1.49 to 2.32) for South Asians, with a comorbidity relative to White Europeans. Taking into consideration cardiac disease comorbidity and non-cardiac disease comorbidity, South Asians (compared to White Europeans) with Type 2 diabetes had an excess risk of having suboptimal glycaemic control, with odds ratios of 1.91 (95% CI: 1.49 to 2.44) and 2.27 (95% CI: 1.50 to 3.43), respectively.

Alshamsan and colleagues (2011) set out to examine ethnic inequalities in diabetes management among people with and without comorbid health conditions after a period of sustained investment in quality improvement in the UK [54]. After adjusting for age, sex, diabetes duration, BMI, socioeconomic status and practice level clustering, they found that the presence of two or more cardiovascular comorbidities was associated with similar blood pressure control among White people and South Asian patients when compared with White people without comorbidity [54]. The mean difference in systolic blood pressure was +1.5 mmHg (95% Confidence Interval (CI): −0.3–3.3) and +1.4 mmHg (95% CI: −0.8–3.6), respectively [54]. In contrast, the presence of two or more cardiovascular comorbidities was associated with worse blood pressure control among Black patients, with a mean difference in systolic blood pressure of +6.2 mmHg (95% CI: 3.5–8.5) [54].

Similarly, Mathur and colleagues (2011) investigated the likelihood of reaching clinical targets for blood pressure, total serum cholesterol and glycated haemoglobin by ethnic group for patients with MLTCs [57]. Their results show that after adjusting for age, sex and clustering by general practice, among those with three to five cardiovascular morbidities, Black patients were less likely to meet their blood pressure target, with adjusted odds ratios (AORs) of 0.63 (95% CI: 0.53 to 0.75) [57]. However, there were no differences apparent between White and South Asian patients [57]. Among those with three to five morbidities, both South Asian and Black patients were less likely to reach an HbA1c target of ≤7.5% compared to White patients, with adjusted odds ratios of 0.69 (95% CI: 0.60 to 0.79) and 0.79 (95% CI: 0.67 to 0.93), respectively [57]. For total serum cholesterol in patients with three to five morbidities, South Asian patients were consistently more likely to reach the target of ≤4 mmol/L than patients of White ethnicity, with adjusted odds ratios of 1.65 (95% CI: 1.49 to 1.83), but Black patients were less likely to meet the cholesterol target (AOR: 0.83 (95% CI: 0.71 to 0.97)) [57]. Patterns in statin prescribing mirrored those for control of total cholesterol; compared to White patients, South Asian patients were more likely to be prescribed statin, but Black patients were less likely to be prescribed statin [57].

The findings from Pinto and colleagues (2010) also point to ethnic inequalities in disease management in people with MLTCs. They investigated ethnic differences in the primary care management of patients with psychosis and analysed health screening and monitoring rates according to the presence of comorbidity [58]. After adjusting for age and area-level deprivation, no significant differences were evident between White and Black patients in relation to cholesterol tests, blood pressure reading, BMI, smoking status and mammogram screening rates [58]. However, they found lower cervical smear rates in Black women with previously abnormal cervical smears, with an odds ratio of 0.22 (95% CI: 0.07–0.69) [58].

3.5.2. Ethnic Inequalities in Emergency Admission among People with Multiple Long-Term Conditions

The findings of one study are suggestive of ethnic inequalities in hospital admissions, in particular, emergency admissions, in people with MLTCs [49]. The study conducted by Wijlaars and colleagues (2018) was a national cross-sectional study that used hospital records. The 763,199 children and young people who took part in this study were categorised into the following ethnic groups: White, Black, Asian, Mixed and Unknown [49]. The authors set out to explore whether changes in emergency admission rates during transition from paediatric to adult hospital services differed in children and young people (aged between 10 and 24 years) with and without underlying long-term conditions [49]. They considered emergency admission to be a clinically important indicator of poor health, which might be affected by the quality of healthcare received from the community during transition [49]. They excluded pregnancy-related admissions and injury-related admissions, with the exception of intentional self-harm, which could signify an underlying mental health condition [49]. After adjusting for age, sex, deprivation and transition, Black and Asian ethnicity were associated with an increase in emergency admission rates for children and young people with LTCs (Incidence Rate Ratio (IRR): 2.49, 99% CI: 2.39 to 2.60)) and Asian ethnicity (IRR: 1.13, 99% CI: 1.08 to 1.19) [49]. This study also found that across the whole sample, the rates of emergency admission increased at the age when young people transition from paediatric care to adult healthcare [49].

#### **4. Discussion**

#### *4.1. Summary of Findings*

Of the studies that counted more than two long-term conditions, there were no studies that reported on care quality and few explored ethnic inequalities in healthcare use among people with MLTCs. The findings from these few studies indicate that there are ethnic inequalities in emergency admission and some aspects of disease management among people with MLTCs. Both Asian and Black children and young people with MLTCs were more likely to have higher rates of emergency admissions when compared to their White counterparts [49]. The findings also suggest that some minoritised ethnic groups with MLTCs are at particular risk of suboptimal disease management. In particular, Black people with MLTCs were found to be less likely to be prescribed statins and to reach set targets for blood pressure, HbA1c levels and total serum cholesterol levels when compared to other ethnic groups [54,57]. In addition, Black women with MLTCs and previously abnormal smears had lower cervical smear rates compared to White women [58]. In contrast, South Asian patients with MLTCs were more likely to have better control of their blood pressure and total serum cholesterol, but less likely to meet targets for HbA1c levels when compared to patients with MLTCs from other ethnic groups [52,54,57]. However, given the few studies identified, the different domains of healthcare use under investigation and the different health conditions explored, our conclusions are tentative.

#### *4.2. Comparison with Other Reviews*

To our knowledge, this is the first review of studies reporting on ethnic inequalities in these domains of healthcare use among people with MLTCs in the UK. Therefore, it is difficult to make comparisons with other reviews that focus on different populations or particular dimensions of healthcare use. However, some of the findings of this review complement those of other reviews of ethnic inequalities in healthcare use that have not focused on MLTCs. For example, the evidence from a review conducted by Dixon-Woods and colleagues (2005) found that utilisation of primary care was generally high among most minoritised ethnic group populations, though there were important exceptions [29]. Just as in this review, they found that uptake of some preventative services (e.g., breast and cervical screening) was relatively lower for minoritised ethnic group people [29]. Their findings also suggest that there are important variations within and between minoritised ethnic groups in their utilisation of healthcare [29]. This variation was also evident in our

review, as South Asian patients with MLTCs had better blood pressure and cholesterol control compared to Black patients with MLTCs [57].

#### *4.3. Mechanisms*

The association between MLTCs, socioeconomic status and healthcare use has been reported; people with MLTCs living in poverty have been found to be less likely to use health services than those with financial resources [62]. Given the close link between ethnicity and socioeconomic disadvantage [29], it is important to consider socio-economic disadvantage when interpreting ethnic inequalities in healthcare. Of the five studies that also counted more than two long-term health conditions, two adjusted for area-level deprivation and one adjusted for socioeconomic status (and other factors, e.g., age, sex and cardiovascular risk) [49,54,58]. Ethnic inequalities in disease management were still evident after adjustment of socio-economic deprivation (and other factors on the explanatory pathway), with Black people reported to have poorer disease management [54,58], and Black and Asian children more likely to have increased rates of emergency admission [49]. While Mathur and colleagues did not adjust for individual level deprivation, their analysis focused on populations living in the eight most socially deprived localities in Britain [57].

That ethnic inequalities for some groups still persisted after adjustment of deprivation (and other factors) in some of these studies suggests that the observed inequalities are likely to be driven by other factors. Given the complex, intersecting processes that shape the development of MLTCs and determine the use of healthcare and care quality [18,27], the mechanisms underlying the observed ethnic inequalities are likely to be the result of the interplay of several processes. Individual-level factors, such as poor management among some people [54] and cultural barriers to effective self-management [52], have been proposed as reasons underlying observed ethnic differences. However, we argue that understanding ethnic inequalities in healthcare use requires an appreciation of the ways in which individual-level processes (e.g., ethnicity and class) intersect with macrolevel processes (e.g., racism and discrimination) to produce inequalities [63]. International studies have illustrated how racism and negative discriminatory practices can result in mistrust of healthcare professionals and create barriers to compliance with treatment, timely diagnoses and treatment and healthcare use [64–66]. These processes can impact efforts to manage MLTCs among minoritised ethnic group populations, thereby resulting in ethnic inequalities. Further evidence is provided by Ben and colleagues (2017), who conducted a systematic review and meta-analysis of quantitative studies reporting on the associations between self-reported racism and different dimensions of healthcare service utilisation [67]. They found that people experiencing racism were approximately two to three times more likely to report reduced trust in healthcare systems and professionals, lower satisfaction with health services and perceived care quality, and compromised communication and relationships with healthcare providers [67]. As such, the influences of racism and discrimination cannot be ignored, as they directly and indirectly create conditions that disadvantage many from minoritised ethnic groups, which in turn can result in ethnic inequalities in healthcare use.

#### *4.4. Strengths and Limitations*

A limitation of this review is that a single reviewer initially screened the titles and abstracts and excluded irrelevant studies, which might have introduced a level of reviewer bias. It is therefore possible that we may have missed relevant studies [68]. However, a manual search of the reference list of key studies was conducted to increase the likelihood of identifying as many relevant studies as possible. In addition, a subset of studies (10%) were double-screened and extracted prior to the independent screening and extraction to reduce reviewer bias. While the interest in MLTCs and associations with healthcare utilisation, costs and healthcare systems has grown over the last decade [33], the guidelines to optimise care for people with MLTCs are fairly recent. For example, in 2016, the National Institute for Health and Care Excellence published guidance for healthcare professionals, people with MLTCs and their families/carers [69]. Thus, there has not been much time to assess care quality among people with MLTCs, and thus studies in this area are sparse. Those that have done so have not explored ethnic inequalities in care quality [70,71]. As such, they were not included in this review. Relatedly, there were no qualitative studies that met the inclusion criteria; therefore, the findings of this review are based on the evidence from quantitative studies. It is important to remember that evidence from qualitative studies is equally important as it gives us an in-depth understanding of the experiences of people with MLTCs while illuminating the processes that can lead to inequalities in healthcare use and care quality as reported above [23]. The findings from these studies can help healthcare systems adapt to the needs of people with MLTCs, thereby improving their health [72].

Despite these limitations, this review has several strengths. First, the review was informed by the PRISMA guidelines to facilitate the transparent reporting of the review process [47,73]. Second, we conducted the electronic search across a range of databases to locate (un)published studies and hand-searched the reference lists of relevant studies and systematic reviews to reduce the likelihood of missing key studies. Third, when synthesising the results of studies that contributed to the evidence of ethnic inequalities in healthcare use and care quality among people with MLTCs, we only included studies that also counted more than two long-term conditions to give us insight into ethnic inequalities in healthcare use among people with complex healthcare needs [40].

This review also highlighted the limitations of the studies conducted in this area. For example, the review has illuminated the limited range of long-term conditions considered. The majority of studies included in this review focused on index conditions, particularly diabetes [52,54,56], mental health conditions [51,53,58,61] and cardiovascular disease [59]. As such, we have a partial understanding of ethnic inequalities in healthcare use among people with MLTCs. In addition, many of these studies categorised their participants into broad ethnic categories. In the five studies that contributed to the evidence of ethnic inequalities in healthcare use among people with MLTCs, minoritised ethnic group people were often clustered into Black [49,54,57,58], South Asian [52,54,57], Asian [49], Mixed [49,57] and Other [57] ethnic categories. It is important to note that in certain circumstances, combining individual ethnic groups into larger categories can facilitate the identification of broad patterns, given that some may have shared experiences of racism, discrimination, marginalisation and social exclusion [53]. However, these broad ethnic categories may mask the extent of intra-ethnic inequalities. For example, as reported above, Black people with MLTCs may be at particular risk of poor disease management [54,57,58]. However, the Black ethnic group population is diverse, and healthcare use and care quality might vary among the different subgroups. Findings from Afuwape and colleagues (2006) exemplify this notion [53]. They examined the characteristics of a community cohort with psychosis and comorbid substance misuse by ethnic group and found that Black Caribbean people had the longest mean contact with mental health services compared to Black African, Black Other and White patients [53]. This study highlights the value of disaggregating broad ethnic group categories. This nuanced approach is more likely to lead to the identification of those who are most vulnerable to developing MLTCs and in greatest need of intervention, and moves away from essentialising minoritised populations.

It is likely that reported ethnic inequalities are underestimated. The studies that contributed to evidence of ethnic inequalities in disease management and emergency admission all analysed data from patient records from primary and secondary care. Ethnicity recording across the National Health Service has improved markedly over the past decade [74]. However, there is evidence that ethnicity coding for patients who self-identify as White British is recorded correctly, but there are higher levels of incorrect coding of the ethnicity of patients from minoritised ethnic groups [75]. Others have also found that in most cases, hospital records over-represent 'Other' ethnic group categories while under-representing 'Mixed' ethnic groups and some specific ethnic groups [76]. Incomplete or inaccurate recording of ethnicity data makes it difficult to reliably assess health needs, access and outcomes across different ethnic groups [76]. Many of the studies included in this review excluded people in the 'Other' ethnic group. It is therefore possible that these studies underestimate the true extent of ethnic inequalities in emergency admission and disease management among people with MLTCs.

#### *4.5. Implications*

The observed inequalities in disease management across ethnic groups suggest that universal coverage and investment in quality initiatives may not be adequate and that enhanced strategies or targeted interventions are needed to improve equity of disease management across populations [52,54]. It is possible that the observed ethnic inequalities in emergency admission among children and young people with MLTCs from minoritised ethnic groups might not only be due to a higher level of ill health but also the poor management of health conditions in primary care. This finding also suggests that ethnic inequalities in healthcare use and care quality start early in the life course. However, further research is required to unpack these findings.

As mentioned previously, the minoritised ethnic group population in the UK is diverse and consists of those born outside the UK and those born in the UK [28]. With different migration histories, the length of residence in the UK among those born outside the UK will vary and may impact healthcare utilisation. Interestingly, studies exploring the association between healthcare use and the number of years spent in the UK have found mixed evidence [77–79]. One study found no differences in healthcare use between non-UK-born migrants and the UK-born population [79]. Another reported that international migrants were less likely to have used secondary care than established residents and within-England migrants [77]. These findings mirror those of Saunders and colleagues (2021), who found that newly arrived migrants have lower healthcare utilisation levels than the UK-born population, a pattern partially explained by younger age and lower levels of ill health [78]. However, these studies do not explicitly focus on populations with MLTCs. Given that none of the studies included in the review considered length of residence in the UK, further research is required to ascertain whether there is an association between length of residence, healthcare use among people with MLTCs and observed ethnic inequalities reported in this review.

The limitations of the studies identified in this review reflect the methodological challenges of investigating ethnic inequalities in healthcare use and care quality among people with MLTCs [29]. Evidently, more work is required to develop a comprehensive understanding of the extent of ethnic inequalities in healthcare use and care quality among people with MLTCs living in the UK. Future studies would need to consider how best to address the challenge of varying definitions for healthcare use and care quality. They would need to include people with a range of MLTCs and include more ethnic group categories, including marginalised White populations (e.g., Gypsy, Roma and Traveller communities), who have been reported to have poor health outcomes when compared to people from other communities [80,81]. They would also need to assess ethnic variations in other domains of healthcare and account for both individuallevel and area-level deprivation and how they intersect with other factors. Such studies would add to the sparse evidence base in this area and allow for national and international comparisons.

In this review, studies that counted more than two long-term conditions that reported on care quality were lacking. If we consider that the assessment of care quality among people with MLTCs is in its infancy, this finding is not surprising. However, future studies should also aim to explore ethnic inequalities in care quality. Studies that adopt a longitudinal approach to analysing ethnic inequalities in healthcare use and care quality are required. These studies would give insight into the longitudinal association of MLTCs, healthcare use and care quality delivered with health outcomes across different ethnic groups [7,27]. Future studies would also benefit from conceptualising and analysing ethnic inequalities in healthcare use and care quality in people with MLTCs through an

intersectional lens that considers the complex, multifaceted processes [63] that lead to the development of MLTCs and influence healthcare use and care quality. Such work could illuminate the extent to which key explanatory pathways, including racism and discrimination, contribute to the development of ethnic inequalities. The findings of such analyses could inform discussions on how ethnic inequalities in healthcare use and care quality among people with MLTCs can be effectively addressed.

#### **5. Conclusions**

This review identified few studies reporting on ethnic inequalities in healthcare use among people with MLTCs living in the UK. It illustrates a sparse evidence base, characterised by studies focusing on different health conditions and different domains of healthcare, which precludes us from drawing any firm conclusions. Indeed, the few studies identified are suggestive of ethnic inequalities in emergency admissions and particular domains of disease management among people with MLTCs. However, the methodological limitations of the studies identified in this review hamper our understanding of the full extent of ethnic inequalities in healthcare use and care quality among people with MLTCs. Based on these limitations, we call for action and have provided directions for future studies that we hope will provide evidence that can inform targeted prevention and management strategies to reduce inequalities in healthcare use and care quality among people with MLTCs.

**Author Contributions:** L.B. and M.S. formulated the overarching research goals and aims of the systematic review. L.B., M.S. and B.H. planned the methodological approach and developed the protocol. B.H. formulated the search terms in discussion with L.B. and M.S. B.H. conducted the search, imported the results and removed the duplicate studies. B.H. and L.B. screened and extracted data from a random sample of studies (10%). B.H. screened and extracted data from the remaining studies. B.H. conducted the narrative synthesis with substantial methodological and intellectual input from L.B. and M.S. B.H. prepared the manuscript and wrote the initial draft. L.B. and M.S. critically reviewed and commented on the initial and subsequent drafts. When reviewing the manuscripts, both L.B. and M.S. verified the data from the studies that contributed to the evidence of ethnic inequalities in healthcare use and care quality among people with multiple long-term conditions. All authors had full access to the included studies. B.H. submitted the manuscript for publication. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work is funded by The Health Foundation [AIMS 1874695].

**Institutional Review Board Statement:** Not applicable, this study did not involve humans.

**Informed Consent Statement:** Not applicable, this study did not involve humans.

**Data Availability Statement:** This study did not report any supporting data.

**Conflicts of Interest:** M.S. is employed by The Health Foundation. The authors declare no conflict of interest.

#### **Appendix A**

**Table A1.** Search terms used when searching Applied Social Sciences Index and Abstracts.





#### **References**


### *Protocol* **Which Non-Pharmaceutical Primary Care Interventions Reduce Inequalities in Common Mental Health Disorders? A Protocol for a Systematic Review of Quantitative and Qualitative Studies**

**Louise Tanner 1,\*, Sarah Sowden 1, Madeleine Still 1, Katie Thomson 1,2, Clare Bambra 1,2 and Josephine Wildman 1,2**


**Abstract:** Common mental health disorders (CMDs) represent a major public health concern and are particularly prevalent in people experiencing disadvantage or marginalisation. Primary care is the first point of contact for people with CMDs. Pharmaceutical interventions, such as antidepressants, are commonly used in the treatment of CMDs; however, there is concern that these treatments are over-prescribed and ineffective for treating mental distress related to social conditions. Nonpharmaceutical primary care interventions, such as psychological therapies and "social prescribing", provide alternatives for CMDs. Little is known, however, about which such interventions reduce social inequalities in CMD-related outcomes, and which may, unintentionally, increase them. The aim of this protocol (PROSPERO registration number CRD42021281166) is to describe how we will undertake a systematic review to assess the effects of non-pharmaceutical primary care interventions on CMD-related outcomes and social inequalities. A systematic review of quantitative, qualitative and mixed-methods primary studies will be undertaken and reported according to the PRISMA-Equity guidance. The following databases will be searched: Assia, CINAHL, Embase, Medline, PsycInfo and Scopus. Retrieved records will be screened according to pre-defined eligibility criteria and synthesised using a narrative approach, with meta-analysis if feasible. The findings of this review will guide efforts to commission more equitable mental health services.

**Keywords:** mental disorders; healthcare disparities; primary health care; systematic review; health inequalities; PROGRESS-Plus

#### **1. Introduction**

Common mental health disorders (CMDs), such as depressive disorders and anxiety disorders, are a major global healthcare problem, causing a large amount of suffering and imposing huge economic costs; for example, mental health problems are estimated to cost the global economy around GBP 105 billion a year [1]. In many countries, including the United States, Canada, Australia and European countries such as France and the UK, primary care is usually the first point of contact for people with mental health problems. Most patients with a mental health problem are seen only in primary care [2–4], and in the UK, mental ill health comprises a third of GP appointments [5]. Pharmaceutical interventions, such as antidepressants, are commonly used, and are frequently effective in the treatment of CMDs. However, there is concern amongst healthcare professionals that pharmaceutical treatments are over-prescribed or inappropriately used, resulting in the medicalising of everyday stresses and distress caused by socioeconomic deprivation [6–8].

**Citation:** Tanner, L.; Sowden, S.; Still, M.; Thomson, K.; Bambra, C.; Wildman, J. Which Non-Pharmaceutical Primary Care Interventions Reduce Inequalities in Common Mental Health Disorders? A Protocol for a Systematic Review of Quantitative and Qualitative Studies. *Int. J. Environ. Res. Public Health* **2021**, *18*, 12978. https://doi.org/10.3390/ ijerph182412978

Academic Editor: Paul B. Tchounwou

Received: 23 September 2021 Accepted: 25 November 2021 Published: 9 December 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

Antidepressant prescriptions show an increasing trend [9] that has outpaced the rise in the prevalence of CMDs [10]. There is also evidence that some indicators of disadvantage and marginalisation, such as unemployment, are associated with increased antidepressant use independent of a diagnosis of depression [11].

Non-pharmaceutical interventions provide alternative treatment options for mental distress. In England, for example, the Improving Access to Psychological Treatment (IAPT) service was developed by the National Health Service with the aim of integrating psychological therapies into primary care [12,13]. More recently, "social prescribing" is being formally embedded into primary care to provide an alternative for patients with mental health disorders (and other chronic health conditions) [14]. Social prescribing aims to improve patients' health and wellbeing by offering them support and linkage to community-based services that provide support with social needs and health behaviours [15,16].

With the aim of applying an equity lens to healthcare interventions, the Cochrane and Campbell Equity Methods group developed the PROGRESS-Plus framework to identify a range of sociodemographic characteristics that stratify health outcomes [17]. A number of PROGRESS-Plus domains have been found to be associated with the prevalence of mental health problems. There are sex differences in rates of mental ill health, with women having higher rates of anxiety and depression [18] and higher rates of substance-abuse-related mental health problems in men [19]. Mental health outcomes have also been found to be associated with one's place of residence, including in terms of access to green space [20] and living in areas of socioeconomic disadvantage [21]. Differential rates of mental health problems have also been found to be associated with race and ethnicity [22–24], occupation [25–27], religious identity [28–30], social capital [31], educational attainment [27,30], age [30], disability status [32] and sexual orientation [33]. Mental ill health is a particular problem in areas of socioeconomic deprivation, where mental health problems can be both a cause and effect of poverty and of social problems such as unemployment, homelessness, debt and violence [30].

In addition to experiencing higher rates of CMDs, people living with disadvantage and marginalisation are less able to access and benefit from treatments for conditions such as anxiety and depression [16,34]. In the UK, as in other high-income countries [35], addressing both mental ill health and health inequalities are key policy objectives, as evidenced in the NHS Long-term Plan [36] and the narrative surrounding the NHS response to the COVID-19 pandemic [37]. For these reasons, there is a pressing need for evidence about which interventions will reduce inequalities in treatment outcomes and which may, unintentionally, increase them. This systematic review will examine evidence on primary care interventions that are likely to decrease, or potentially increase, health inequalities in treatment access and outcomes for patients experiencing CMDs. Findings will guide the commissioning of more equitable mental health services.

In line with PRISMA-E guidelines [38–40], as the first stage of this equity-focused review, a framework for conceptualising primary care interventions and mental health inequalities was developed (Figure 1). The framework outlines the three types of nonpharmaceutical interventions considered in this review:

1. Social prescribing (for example, arts activities, healthy eating, housing and financial advice);

2. New models of care (for example, integration of primary and secondary healthcare, or the integration of health and social care);

3. New methods of clinical practice (for example, clinical psychologists integrated with general practice teams or extended consultation times).

The framework considers the potential domains of inequality addressed by an intervention, the approach taken, and factors that may impact the effectiveness of any given intervention in reducing inequalities in mental health outcomes. It was developed based on an existing framework used in a previous equity-focused review [41] and will be revised iteratively [42] as evidence from the systematic review emerges.

**Figure 1.** Framework for addressing inequalities in CMD-related health outcomes in relation to the PROGRESS-Plus domains [17,43,44], adapted from Sowden et al. [41].

#### **2. Materials and Methods**

The review will be carried out following established criteria for the good conduct and reporting of equity-focused systematic reviews using PRISMA-E guidelines [38–40] and reporting here conforms to the standards of the Preferred Reporting Items for Systematic Reviews and Meta-analysis Protocols (PRISMA-P) (see Supplementary File S1) [45]. The protocol for this systematic review was registered on the PROSPERO database on 23rd September 2021 (registration number: CRD42021281166) [46].

#### *2.1. Research Questions*

The main research question to be addressed in this review is:

Which non-pharmaceutical primary care interventions reduce inequalities in CMDrelated adverse health outcomes?

Review sub-questions are:


#### *2.2. Objectives*

The objectives of this systematic review are to:


#### *2.3. Inclusion Criteria*

The following criteria (summarised in Table 1) will be applied to each full text in order to assess their eligibility for inclusion in the review.

**Table 1.** Summary of inclusion and exclusion criteria.


<sup>a</sup> CMDs of interest in this review are: anxiety, depression, somatoform disorders, post-traumatic stress disorder or post-natal depression.

#### 2.3.1. Population

The population of interest consists of people who are being treated in primary care in any high-income country defined by The Organisation for Economic Co-operation and Development (OECD) [47] whose characteristics in relation to one or more of the PROGRESS-Plus factors [48] are reported. Included studies must indicate that all or some participants have a CMD, which must be one of the following disorders, defined by Lund (2020): anxiety, depression, somatoform disorder, post-traumatic stress disorder or postnatal depression [49]. Participants are not required to have received a CMD diagnosis; presence of a CMD may be indicated using data from mental health screening tools (e.g., the Self-Reporting Questionnaire (SRQ-20)). The mental health status of participants may be reported narratively (e.g., in the title, participant characteristics or inclusion criteria) or in the baseline characteristics table. We will exclude studies exclusively involving participants with the following more severe and less common conditions, defined by Lund (2020): psychosis, dementias, child and adolescent mental disorders, conversion disorders, body dysmorphic disorders, personality disorders, eating disorders, suicide, self-harm, substance use disorders, intellectual disability, epilepsy and developmental disorders [49].

We will include studies reporting interventions that have been delivered exclusively to a disadvantaged population subgroup covered by the PROGRESS-Plus criteria (e.g., where participants are all from an ethnic minority group). We will also include studies involving participants with other specific characteristics (e.g., persons with specific exposures such as victims of abuse) if one or more of their PROGRESS-Plus characteristics are also reported. Furthermore, studies that report interventions that have been delivered universally to people from disadvantaged and non-disadvantaged backgrounds (e.g., older versus younger persons; individuals from low- versus high-income households) will be included if the authors report a sub-group analysis of the differential effectiveness of the intervention between population sub-groups (e.g., persons with and without disabilities).

#### 2.3.2. Intervention

Interventions delivered by or referred to from primary care teams (including GPs and allied health professionals based in GP practices and community pharmacies) will be included.

For the purpose of this review, a broad definition of non-pharmaceutical primary care interventions will be used, including referral of individuals to activities and support services provided by the voluntary sector as well as new models of care or methods of clinical practice in relation to patient care. We will include psychological interventions, such as Cognitive Behavioural Therapy.

Studies exclusively investigating the effects of pharmaceutical interventions will be excluded. However, where a pharmaceutical intervention is one component of a multifaceted, integrative, holistic approach to treatment and care, the overall intervention will be included.

The intervention must have been either delivered exclusively to a disadvantaged population sub-group covered by the PROGRESS-PLUS criteria (e.g., older persons; individuals with disabilities) or reported a sub-group analysis of the differential effectiveness of the intervention between population sub-groups (e.g., older versus younger adults; persons with and without disabilities).

#### 2.3.3. Comparators

#### Population

We will include studies reporting data on CMD-related health outcomes in relation to at least one type of inequality from the PROGRESS-PLUS criteria [48], i.e., place of residence (e.g., rural/urban location); race/ethnicity/culture/language; occupation; gender/sex; religion; education; socioeconomic status including social capital [48].

Eligible studies may present data enabling between-group comparisons (e.g., between persons who received the intervention from low- versus high-income households) or within-group comparisons (e.g., before and after the intervention, amongst individuals who are all from low-income households).

#### Intervention

Data from included studies comparing the effectiveness of intervention versus no or alternative intervention (including alternative similar interventions and variations in the format, duration and intensity of the intervention as well as usual treatment vs. novel one) will also be extracted where available and included in the synthesis.

#### 2.3.4. Outcomes

Relevant outcomes from quantitative studies will include measures of morbidity which are directly related to CMDs (e.g., the number of health care consultations and measures of medication usage for CMDs) in addition to assessments from validated mental health screening tools, including but not limited to the State–Trait Anxiety Inventory (STAI) [50], the Perceived Stress Scale (PSS) [51], the Positive and Negative Affect Schedule (PANAS) [52], the Warwick–Edinburgh Mental Well-being Scale (WEMWBS) [53], the Self-Reporting Questionnaire (SRQ-20) [54] or the General Health Questionnaire (GHQ) [55]. We will also include adapted versions of these mental health screening tools that have been validated in other languages (e.g., the Spanish and Dutch versions of the PANAS) [56,57], as well as validated tools that have been developed for use in high-income countries outside of the UK, if the study is reported in English.

The overall impact of the interventions of interest will be assessed by comparing the occurrence of CMD-related adverse health outcomes before and after the intervention amongst health-disadvantaged population subgroups. Health inequalities will be assessed by comparing CMD-related adverse health outcomes between the most and least healthdisadvantaged groups.

From qualitative studies, we will extract information providing insights on the mechanisms by which non-pharmaceutical primary care interventions could impact CMD outcomes and inequalities, in addition to barriers and facilitators to the successful implementation of these programmes.

#### 2.3.5. Study Design

Quantitative primary studies, including randomised controlled trials (RCTs), other intervention studies (e.g., quasi-experimental), longitudinal studies (e.g., cohort and panel studies), repeated cross-sectional studies and ecological studies [58], will be included in the review in addition to qualitative [59] and mixed-methods primary studies.

#### 2.3.6. Context

In order to be included in the review, studies must be written in English language and have been published in an OECD high-income county (studies that do not meet these criteria will be excluded during screening) [47].

#### *2.4. Search Strategy*

The following databases will be searched from their start until 1 June 2021 (host platforms in brackets): Applied Social Sciences Index and Abstracts (ASSIA; ProQuest)(Ann Arbor MI, USA); Cumulative Index to Nursing and Allied Health Literature (CINAHL) (EBSCO, USA); Embase (Ovid, London, UK); PsycInfo (EbscoHost, Ipswich MA, USA); Scopus (Elsevier, Amsterdam, The Netherlands). The draft search strategy for Medline is shown in Supplementary File S2.

The useful resource list of the Social Prescribing Network [60], the Social Interventions Research and Evaluations Network (SIREN) [61] and relevant charity websites will also be purposefully searched for relevant articles. Citing references will also be identified using Google Scholar's "cited by" feature. We will also screen the reference lists of reviews located during the searches which are deemed relevant to the research question, as well as any primary studies which are included in the review, to identify further potentially relevant studies. No limits on date or language will be placed on the searches.

Search strings for the relevant databases were built from existing search filters for PROGRESS-Plus [43] and mental health components [62]. Primary health care elements of the search were taken from that used in a Cochrane review of primary care treatment for alcohol and drugs [63]. The search strings for each database will be peer-reviewed by an experienced information specialist, using the PRESS checklist [64] prior to implementation.

#### *2.5. Screening and Selection*

Records located in the searches will be downloaded into an Endnote [65] library and de-duplicated. Rayyan software (Qatar Computing Research Institute, HBKU, Doha, Qatar)will be used to screen studies retrieved from the literature searches [66]. A twostage process will be used to identify studies for inclusion in the review. First, titles and abstracts will be screened to identify studies relevant to the review topic. The full texts of potentially relevant studies will be sourced and assessed for eligibility in relation to pre-defined inclusion and exclusion criteria. One reviewer (MS) will screen each record, and a second member of the review team (LT) will check a random 10% sample at both stages of the screening process. Screening conflicts will be resolved via discussion and adjudication by a third reviewer (JMW, KT or SS) where necessary.

#### *2.6. Data Extraction*

Separate data-extraction forms will be created for quantitative, qualitative and mixedmethods studies. These will be based on existing tools and pre-piloted on a sample of studies deemed eligible for inclusion in the review, with modifications to ensure that all relevant information is captured. Extracted information from quantitative studies will include citation details (first author name and publication date), study characteristics (study aims, design, country and setting), population characteristics (PROGRESS-Plus and other reported characteristics), intervention details (type of intervention, mode and duration of delivery), comparators (pertaining to the population and intervention), outcomes and results (mean and SD values for each comparison group continuous data; number of events and sample size for each comparison group for categorical outcomes). We will extract additional data reported in the included studies which quantify the association between the exposures and outcomes of interest (e.g., results from correlational, regression and modelling studies). The qualitative-data-extraction form will include bibliographic information, methods (e.g., number of participants, data collection method), relevant findings, illustrations from the paper (e.g., participant quotes) and a suggested category or code for that finding. Data from each study will be extracted by one person and checked by a second person.

#### *2.7. Quality Appraisal*

Appropriate tools will be selected based on the study designs identified for inclusion in the review. The CASP tools [67] will be used to assess the quality of quantitative studies; a modified version of the relevant CASP tool will be used to quality appraise qualitative studies [68]. If eligible mixed-methods studies are identified, these will be critically appraised using the Mixed Methods Appraisal Tool (MMAT) [69]. Repeated cross-sectional studies will be assessed using the Appraisal tool for Cross-Sectional Studies (AXIS) [70]. All studies will be synthesised regardless of quality. A nominal scoring system will be devised to enable comparability of the overall quality between studies collecting the same type of data (i.e., quantitative, qualitative and mixed-methods). Based on the scoring system developed by the Cochrane Collaboration, studies will be rated as low quality, some concerns or high quality based on domain scores [71].

#### **3. Results**

#### *Synthesis*

The review will be reported according to the PRISMA-Equity guidance [40]. We will include a paragraph summarising the overall characteristics of included studies, including the number and percentage of studies with different characteristics (e.g., types of study, country of publication and participant characteristics). It is anticipated that heterogeneity will prevent the implementation of meta-analysis. However, if the quantitative data allow, pairwise meta-analysis will be performed using RevMan 5 [72]. For continuous outcomes, we will compare mean values, whereas the number of events and sample size will be extracted for binary outcomes to determine outcome rates. For studies presenting within-group comparisons (where an intervention has been delivered exclusively to a disadvantaged subgroup), we will compare mean values for continuous outcomes (e.g., mean anxiety scores) before versus after the intervention. For studies presenting between-group comparisons, we will compare changes from baseline values (where the data are available) for continuous variables and post-intervention data for binary variables (e.g., GP consultation rates) between population sub-groups. For the between-group comparisons, where there are data for >2 categories for a particular domain of inequality (e.g., low-, middle- and high-income households), we will compare the most and least disadvantaged population subgroups (e.g., highest- versus lowest-income households). Statistical heterogeneity will be assessed using the Chi2 and Higgins I2 statistics. Heterogeneity will be deemed to be present if the *p* value for the Chi2 test is <0.10 or the Higgins I<sup>2</sup> statistic is >50%. A random-effects model will be implemented. If it is not feasible to undertake meta-analysis of the quantitative data, a vote-counting approach, such as a Harvest plot, will be used to graphically present the quantitative data. [73]

Thomas and Harden's (2008) three-stage approach to qualitative synthesis [74] will be used to thematically synthesise qualitative data. This will involve (1) coding the data line by line according to content and meaning; (2) grouping codes according to similarities and differences to produce descriptive themes; (3) generating analytical themes according to the reviewer's interpretation of the data in relation to the review question.

The results will be presented in relation to each domain of inequality for which relevant data were identified. The domain of inequality we are primarily interested in in this systematic review is socioeconomic status. Data in relation to this characteristic will be included in the synthesis preferentially over other PROGRESS-plus factors, if including evidence in relation to other social factors would be unmanageable within the time constraints of this project.

The Economic and Social Research Council's framework [75] will be used to write the narrative synthesis, combining the quantitative (effectiveness) results with the qualitative (mechanisms of action; barriers and facilitators) themes. This includes the following components:


A Best Available Evidence approach [76] will be used to synthesise evidence from the quantitative studies. This will involve reporting evidence from studies with the most robust designs assessing the clinical effectiveness of interventions, prior to evidence from lessrobust study designs. The following hierarchy will be used: (1) RCTs, other intervention studies (e.g., quasi-experimental); (2) individual-level longitudinal studies; (3) repeated cross-sectional studies; (4) ecological studies.

#### **4. Discussion**

CMDs are prevalent globally. In England, for example, 1 in 6 people report experiencing a common mental health problem (such as anxiety and depression) in any given week

(data from 2014) [77]. Non-pharmaceutical primary care interventions provide alternative therapeutic options for mental distress to drug treatments. In order to support the use of such treatments for people with CMDs, their effectiveness must be assessed, as well as their impacts on health inequalities.

This systematic review will provide evidence regarding the effectiveness of nonpharmaceutical primary care interventions at reducing inequalities in CMD-related outcomes in relation to the PROGRESS-PLUS domains [48]. The mechanisms by which these interventions impact the outcomes of interest and barriers and facilitators to implementation will be explored. This will enable policy makers to identify non-pharmaceutical primary care interventions that are most effective at reducing inequalities in mental health outcomes and to determine which aspects of these interventions increase or decrease their effectiveness in different populations.

Limitations of the methods include potential language bias from excluding studies not published in the English language. Additionally, we will deviate from gold-standard methods by only having one reviewer screen, extract data and quality-appraise all of the articles and a second reviewer perform a 10% check.

Dissemination of findings will take place through a written report for stakeholders. The report findings will also be shared in two half-day workshops: one with practitioner stakeholders, including from the Clinical Commissioning Groups, Primary Care Networks and the Integrated Care Systems, and the other with members of the public. Dissemination workshops will also seek input from practitioners and public stakeholders into a bid for further funding. Two articles will be published in peer-reviewed journals.

#### **5. Conclusions**

Globally, CMDs create significant health and economic burdens. In many countries, patients with CMDs are predominantly treated in primary care. While many CMDs are treated with pharmaceutical interventions, non-pharmaceutical interventions, such as psychological therapies, are available as alternative treatments for people with anxiety and depressive disorders. Within primary care, new models of care and new methods of clinical practice, such as social prescribing, are also being developed to provide nonpharmaceutical options for patients with CMDs. However, the effects of these interventions on social inequalities in CMD-related health outcomes is unknown. This review will assess the impacts of non-pharmaceutical primary care interventions on social inequalities in CMD-related health outcomes, based on evidence from quantitative, qualitative and mixed-methods studies. The results will provide evidence to support the delivery of more equitable mental health services.

**Supplementary Materials:** The following are available online at https://www.mdpi.com/article/10 .3390/ijerph182412978/s1, file S1: PRISMA-P 2015 checklist, file S2: draft search strategy for Medline.

**Author Contributions:** Conceptualization: J.W., C.B., K.T. and S.S.; Methodology: J.W., K.T., L.T., M.S., C.B. and S.S.; Writing (Original Draft Preparation): J.W., K.T., L.T., M.S. and S.S.; Writing (Review and Editing): J.W., K.T., L.T., M.S., C.B. and S.S. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research is funded by NIHR Research Capability Funding (RCF) from NHS North of England Care System Support (NECS). S.S. is supported by Health Education England (HEE) and the National Institute for Health Research (NIHR) through an Integrated Clinical Academic Lecturer Fellowship (Ref CA-CL-2018-04-ST2-010) and RCF funding, NHS North of England Care System Support (NECS). This project is supported by the National Institute of Health Research (NIHR) Applied Research Collaboration (ARC) for the North East and North Cumbria (NENC). K.T. and J.W. are members of the NIHR ARC NENC. The views expressed are those of the authors and not necessarily those of the funders.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


### *Article* **'It All Kind of Links Really': Young People's Perspectives on the Relationship between Socioeconomic Circumstances and Health**

**Hannah Fairbrother 1,\*, Nicholas Woodrow 2, Mary Crowder 2, Eleanor Holding 2, Naomi Griffin 3, Vanessa Er 4, Caroline Dodd-Reynolds 3, Matt Egan 4, Karen Lock 4, Steph Scott 5, Carolyn Summerbell 3, Rachael McKeown 6, Emma Rigby 6, Phillippa Kyle <sup>3</sup> and Elizabeth Goyder <sup>2</sup>**


**Abstract:** Meaningful inclusion of young people's perceptions and experiences of inequalities is argued to be critical in the development of pro-equity policies. Our study explored young people's perceptions of what influences their opportunities to be healthy within their local area and their understandings of health inequalities. Three interlinked qualitative focus group discussions, each lasting 90 to 100 min, with the same six groups of young people (*n* = 42) aged 13–21, were conducted between February and June 2021. Participants were recruited from six youth groups in areas of high deprivation across three geographical locations in England (South Yorkshire, the North East and London). Our study demonstrates that young people understand that health inequalities are generated by social determinants of health, which in turn influence behaviours. They highlight a complex interweaving of pathways between social determinants and health outcomes. However, they do not tend to think in terms of the social determinants and their distribution as resulting from the power and influence of those who create and benefit from health and social inequalities. An informed understanding of the causes of health inequalities, influenced by their own unique generational experiences, is important to help young people contribute to the development of pro-equity policies of the future.

**Keywords:** health inequalities; social inequalities; social determinants of health; young people; qualitative

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

#### **1. Introduction**

There is a well-established relationship between socioeconomic position and health [1,2]. Health follows a socioeconomic gradient, where each step up the socioeconomic ladder is associated with better outcomes [3,4]. This patterning is longstanding and evident throughout the life course across a range of different outcomes at both micro and macro geographical levels [5–7]. In the UK, set against a backdrop of rising levels of poverty and the fallout of government austerity policies following the 2008 recession [8], the past decade has seen socioeconomically patterned health inequalities widen for both adults and children

**Citation:** Fairbrother, H.; Woodrow, N.; Crowder, M.; Holding, E.; Griffin, N.; Er, V.; Dodd-Reynolds, C.; Egan, M.; Lock, K.; Scott, S.; et al. 'It All Kind of Links Really': Young People's Perspectives on the Relationship between Socioeconomic Circumstances and Health. *Int. J. Environ. Res. Public Health* **2022**, *19*, 3679. https://doi.org/10.3390/ ijerph19063679

Academic Editors: Jessica Sheringham and Sarah Sowden

Received: 19 January 2022 Accepted: 12 March 2022 Published: 19 March 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

and young people [9,10]. More recently, the COVID-19 pandemic has exacerbated existing inequalities, with those lower down the socioeconomic ladder being disproportionately affected both in economic and health terms [11].

Central to many contemporary explanations for socioeconomically patterned health inequalities is the concept of Social Determinants of Health (SDH). The World Health Organisation's Commission on the Social Determinants of Health described the SDH as 'the conditions in which people are born, grow, live, work and age' and argued that 'the marked health inequities between countries are caused by the unequal distribution of power, income, goods, and services, globally and nationally' [7] (p. 1). However, while there is broad consensus as to the importance of the SDH, there is much less consistency in the way in which the concept is mobilised [12–14]. What we see is a range of discourses that draw upon the SDH but differ significantly in the way they explain *how* societal factors result in differences in health [15]. These differences in interpretation, Raphael (2011) argues, are not just about 'intellectual world views' but fundamentally affect how we seek to approach and redress health inequalities [12] (p. 223). Raphael (2011) proposes a spectrum of seven discourses to encapsulate the different ways of understanding (and responding to) the SDH (see Table 1) [12].

**Table 1.** Summary of Raphael's (2011) seven discourses of the social determinants of health (SDH).


According to Raphael (2011) and other key researchers in the field (such as Scott-Samuel and Smith (2015), a real reduction in apparently intractable health inequalities will only be possible by tackling inequitable political structures and the power and influence of the people that shape them (Discourse Level Six and Seven) [12,16]. To create a step-change, Raphael argues, we need to 'educate [ ... ] the public that deteriorating quality SDH and inequitable SDH distributions result from the undue influence upon public policymaking of those creating and profiting from social and health inequalities' [12] (p. 230). The argument that we need to change public understandings is widespread [17] and reinforced by a recent collaboration between the Health Foundation and the Frameworks Institute, which sought to 'develop a deeper appreciation of the ways in which people understand and think about health in order to develop more effective approaches to communicating the evidence' [18]. Improving public awareness of health inequalities and the social determinants of health is argued to be vital for galvanizing support for change to the political status quo and the development of pro-equity policies [18].

Studies exploring public perceptions of the link between socioeconomic circumstance and health, however, are limited [17,19]. There is broad agreement regarding the importance of the SDH among the research community, with many narratives echoing the higher-level discourse of Raphael's (2011) typology through repeated critiques of a focus on lifestyle behaviours and neglect of the causal pathways of health inequalities and economic and environmental factors [12,20,21]; though see Dijkstra and Horstman's (2021) critique of social epidemiological research which constructs low socioeconomic status populations as 'inherently unhealthy and problematic' [22] (p. 6). However, public understanding of the factors shaping health has been argued to be limited [12,17,23]. This is supported by recent research by the Frameworks Institute which found that 'public discourse and policy action is limited in acknowledging the role that societal factors such as housing, education, welfare and work play in shaping people's long-term health' [18] (p. 1). Drawing on the Frameworks Institute's findings on young people's views, Marmot et al. (2020) characterised public understandings as individualistic, fatalistic and prone to divisive 'them and us' thinking [1] (p. 145). In contrast, in their review of the admittedly limited evidence base (a meta-ethnography of 17 qualitative studies), Smith and Anderson (2018) argue that people experiencing socioeconomic disadvantage do display an awareness of how socioeconomic hardship can lead to ill health [17]. The picture is thus mixed with contradictory findings regarding the perceived chasm between research consensus and public understanding. In the context of increasing socioeconomic and health inequalities over recent decades and particularly recently due to the COVID-19 pandemic (which has exacerbated existing, socially patterned inequalities through its interaction with inequalities in chronic disease and the social determinants of health including poor quality housing and lower access to healthcare in disadvantaged communities) [1,11,24], it is an opportune time to revisit public perceptions of how socioeconomic circumstances shape health. Further, Smith and Anderson (2018) highlight a dearth of studies exploring the views and experiences of young people [17] (see also Woodgate and Leach's (2010) study and Backett-Milburn et al.'s 2003 study [25,26]). This is an important gap in the evidence base [17,22,27,28]. Youth activism in other spheres such as climate change teaches us that young people have the potential to galvanize support for and contribute to significant policy change [29].

#### *Study Aim*

The aim of our research project was to explore young people's perceptions of what influences their opportunities to be healthy within their local area and their understandings of health inequalities. This paper presents key findings on young people's perspectives on the relationship between socioeconomic circumstances and health.

#### **2. Materials and Methods**

#### *2.1. Overview*

We undertook a series of three interlinked qualitative focus group discussions with six groups of young people (*n* = 42) aged 13–21, resulting in 18 focus group discussions in total. Participants were recruited from six youth groups across three geographical locations in England (South Yorkshire (SY), the North East (NE) and London (L)). All three locations fell within the most deprived quintile based on the 2019 English indices of multiple deprivation (IMD). Data generation took place between February and June 2021, during the COVID-19 pandemic. Due to the UK's lockdown and social distancing restrictions [30], the majority of focus groups were conducted online (*n* = 15). However, focus group discussions (*n* = 3) with one youth group in the North East were conducted face-to-face, once social distancing restrictions permitted, since the youth group did not have facilities to support online data generation in their building (e.g., computers, Wi-Fi) and not all the young people had the technologies to participate from home. Focus group discussions lasted between 90 and 100 min. Further details on the methodological and ethical challenges of this study are described elsewhere [31]. Ethical approval for the study was granted by the School of Health and Related Research (ScHARR) Ethics Committee at the University of Sheffield.

Throughout our project, we actively engaged with Mason's (2018) 'difficult questions' for qualitative research to help ensure the quality, rigour and methodological integrity of our study [32]. In relation to reliability, we seek to provide a detailed and transparent account of our sampling, recruitment, data generation and analysis. Our concerns for the validity of our method and interpretation focus on ensuring the fit between our method and 'tracing the route' (albeit a messy and non-linear one) of our interpretations.

#### *2.2. Sampling and Recruitment*

The focus groups involved young people from pre-existing youth organisations, and our sampling was shaped by each group's demographics. Given our focus on socioeconomic circumstances, we initially sought to work with youth groups in socioeconomically contrasting areas. Recognising that socioeconomic position permeates and intersects with other axes of inequality, we also sought to ensure that we worked with young people of different genders and ethnicities in both urban and rural areas (including coastal areas), and we approached youth groups that we thought would enable this. However, due to challenges of recruitment during a pandemic (with youth groups pausing and/or moving online) we had to take a pragmatic approach and work with youth groups with whom we already had established working relationships, all of which were in areas of high deprivation. The youth group workers we approached saw issues around health inequalities as pertinent in their areas and thus important to engage with. Further, while we initially aimed to work with young people aged 13–17, we took a flexible, inclusive approach as some of our youth groups also included young people over 18. We did not want to exclude young people outside of this range if they were keen to participate, particularly since the focus groups replaced their usual weekly meetings. Our inclusive approach also recognises that young people's transitions to adulthood in the UK have become increasingly elongated and less linear [33]. It is important to understand the concepts of 'youth' and 'adulthood' as not being simply a feature of age but also encompassing a variety of different experiences and understandings within this life phase [34]. The young people we worked with were all members of youth organisations, which, for us, was a primary criterion for participating in this study.

In this way, we adopted a purposive sampling strategy, designed to encapsulate a relevant range of perspectives [32]. Drawing on Braun and Clarke (2021), our sample was guided by the breadth and focus of the research question(s); the demands placed on participants; the depth of data likely to be generated; pragmatic constraints; and the analytic goals and purpose of the overall project [35]. Our approach coheres with Braun and Clarke's (2021) description of qualitative research as a 'situated, reflexive and theoretically embedded practice of knowledge generation' [35] (p. 210). This focus on the active construction of meaning opens up the potential to keep working towards new understandings. Our final sample consisted of 42 young people aged 13–21 and included young people of different genders and ethnicities in both urban and rural areas (including coastal areas) (see Table 2).

Youth workers invited group members to participate and shared an information video and project overview, and researchers attended sessions with the youth groups to talk through the study and build rapport. Any young person interested in taking part was given a more detailed information sheet. For potential participants under the age of 16, opt-in consent from parents/guardians was gained. Written consent was then gathered for all participating young people (either on paper or electronic). Participants were asked to provide basic demographic information, including their postcode, which we used to capture an overall average deprivation rank measure (average position out of the 32,844 small areas in England, with closer to 1 being more deprived) (see Table 2). Despite all the field sites falling in the most deprived quintile, the average participant position across the groups ranged from quintile 1 to 3.


#### **Table 2.** Sample demographics.

#### *2.3. Data Generation*

The stigmas around topics of health and inequality (where practices and situations are individualised and equated with deficit, passivity and irrational choice) make discussion of such topics challenging [36,37]. We employed focus group discussions to generate data and we gave careful consideration to the topic guides (activities and language used), as well as how support could be provided during and after the sessions [31]. While focus groups may prevent people sharing information due to concerns around privacy and stigma [38], they can help to reduce potential power differentials between researchers and participants and provide a space where people can discuss challenging topics with the support of others [39]. We ensured that we framed our questions so that participants could talk generally about young people in their area rather than feeling pressured into discussing their own personal experiences (e.g., 'What kind of things where you live support young people to be 'healthy'?'). Youth workers helped to facilitate the discussions alongside the research team. As well as having at least one youth worker involved in each session, we had four members of the research team in each online session and two members of the research team in each face-to-face session. There was at least one week between each of the three sessions for each group, which helped to avoid fatigue and to provide the opportunity for participants to reflect on and discuss the sessions with youth workers and peers.

Topic guides were piloted and revised as part of our Public Involvement and Engagement work with partner youth organisations (see Supplementary File S1: Topic Guides). Both online and face-to-face focus groups followed the same format (introductions, warmup activity, main activity (in smaller breakout groups) and close and cool-down activity). The first focus group used a participatory concept mapping activity (for example see Jessiman et al. 2021 [40]) to explore perceptions of what influences young people's opportunities to be healthy in their local area (see Supplementary File S2 for an example of a map developed from participants' discussions). The second looked at understandings of inequalities in health. Participants were asked to discuss what they understood by the term 'health inequality' and asked to select and share their ideas about contemporary news articles

relevant to health inequalities (e.g., free school meals, the uneven impact of COVID-19). The third focus group involved a discussion of the young people's key priorities for change in improving health in their area.

#### *2.4. Data Analysis*

In keeping with our open question framing (designed to avoid participants feeling pressured into disclosing personal stories), we employed thematic analysis, drawing on Braun and Clarke's (2006) framework [41]. In particular, our approach was guided by an emphasis on analysis as 'creative and active' [42] (p. 343) and an inherently 'interpretive, reflexive process' [42] (p. 332). The qualitative data management software system NVivo-12 was used to support data management. A coding frame (see Supplementary File S3: Coding Framework) was developed through HF, MC, VE, NG, EH and NW independently reading and adding descriptors to a selection of transcripts. Key codes and overarching themes were then discussed and agreed upon. The development of the coding framework was largely inductive, but an initial scaffolding was provided by key concepts in the literature and the research questions [17,43–46]. Following independent double coding of a selection of the transcripts (*n* = 6, two from each geographical area), we refined the framework before finally coding all transcripts. The development of a coding framework enabled multiple researchers in different locations to contribute to coding. The framework was used as a flexible starting point for analysis for this paper, which was carried out by the two lead authors (H.F. and N.W.). While we did not originally code our transcripts in relation to Raphael's (2011) SDH discourse framework [12], we mobilise this framework in the discussion of our findings as a helpful tool for illuminating *how* young people understand the relationship between socioeconomic circumstances and health.

We use verbatim extracts from the focus groups to illustrate the key findings. While we collected demographic data, this was anonymised at the point of collection to protect participant confidentiality. This means only the field site location and focus group session for each quote are provided (e.g., SY1.1 = South Yorkshire Group 1, session 1). Thus we are unable to identify individual quotes from participants but have endeavoured to present a range of young people's voices across and within all participant groups.

#### **3. Results**

#### *3.1. Perceptions of Factors Linking Socioeconomic Position and Health*

Participants in our study identified a number of different factors that they perceived to impact upon young people's opportunities to enjoy good health in their local area. Through the course of their discussions, they described how abilities to eat healthily, access health-promoting spaces and activities and housing conditions all influenced health and were all shaped by socioeconomic position. The exacerbating impact of COVID-19 upon these abilities was also discussed. These themes were salient across all youth groups.

#### 3.1.1. Eating Healthily in Contexts of Deprivation

Young people described a range of barriers to eating healthily in contexts of deprivation: the cost of and access to 'healthy' food, the apparent ubiquity of 'unhealthy' food, time pressure and competing priorities for limited financial resources. There was a general consensus among participants that 'healthy food' (particularly fresh fruit and vegetables) was more expensive than 'unhealthy food' (particularly processed foods) and that this was a key source of inequality:

*'you can get chocolate bars for £1, you can get KFC for £1, £2 for a whole meal [* ... *] and then you go for the healthy meals and it's like £3, £4 for no reason. And then they ask, oh why is everyone not eating healthy food instead? How can we eat healthy food if the area doesn't even have any healthy food, it's, our environment is just full of unhealthy food.'* (L1.1)

Many young people described poor access to healthy food within their local areas and contrasted this with the ubiquity of 'fast-food' take-away outlets. Young people from London in particular linked fast-food density to the socioeconomic context of the area:

*'when I go to richer parts of London, for example, like, when I go to the City where my university is, the [Name of university], I don't see that many fast food shops around me but I see, like, when I'm in my own local area, [Name of location], there's so many fast food shops.'* (L2.2)

They also described how this situation had worsened due to the COVID-19 lockdown measures, with spaces seen as 'unhealthy' (takeaways) noted to be quicker to open up than health-supporting spaces (youth clubs, gyms). Generally, limited financial resources alongside limited access created a context of *unaffordability*, which constrained consumption choices. A number of participants challenged the notion that eating healthily is necessarily expensive and described how buying takeaways would be costly when looking at the perspective of buying every day and for a whole family. However, whilst some participants foregrounded the importance of behaviours such as cooking skills and planning meals ahead, others, through the group discussions, positioned individualised arguments like this against working families' busy lives. They thought that people on a low income were also likely to be 'time poor' and that this would push them towards quicker and easier, but not healthier, 'choices': '*people on low incomes often, a lot of the time they work more hours and they can't afford childcare and stuff so they don't have the time to like prepare meals, which are like really healthy*' (SY2.2). One participant eloquently explained how a lack of time 'forced' parents '*to do—in a way—irrational things, such as constantly sending a fast food order*' (NE2.2).

While in one group some young people initially found it difficult to understand higher rates of obesity among lower socioeconomic groups, through the course of their discussion, they made sense of the apparently counterintuitive link:

*Participant A You associate free school meals with poorer families who don't necessarily have obesity, if you get me. So the people who have got the money buy the food and then eat it and then get obese. But for me it's quite interesting that obviously obesity is associated with poorer families.*

*Participant B Healthier foods tend to be more expensive, like you can get one thing which is like full fat and it'll be like £2 and if you want to get the fat-free version it's like £3.50 or something.*

*Participant A Yeah, yeah, I was thinking the same as well. Obviously the cheaper stuff's worse, if you get me, and more unhealthy.'* (SY1.2)

Young people also highlighted competing priorities for people on a low income (e.g., household bills, activities, clothes), which meant that they could not always 'choose' the healthy option. A salient theme within many food narratives was the shame associated with the inability, or bounded ability, to 'consume correctly': '*When people have to buy cheaper options, sometimes they get ashamed quite a lot, people saying that they're being right cheap or it's bad things or they're being lazy* [ ... ]' (SY1.1). Indeed, some participants highlighted how food banks, designed to attenuate the impacts of poverty, could represent a source of embarrassment and shame for those who used them:

*'there's more food banks and stuff opening, which is a good thing, especially in this area but some people might be embarrassed to go to one because they don't want to show that they're in poverty* ... *people might shame them for it, definitely* ... *There's like the ideal, you can provide for your family without any help or charity help, and people want to show to be like that and they don't want people to see them as like not working and being lazy, which obviously is not going to be good on the mental health.'* (SY2.1)

Such quotes highlight the importance young people attached to the shame associated with poverty, and the use of the word 'lazy' hints at their awareness of deficit discourses of the 'undeserving poor' [47]. In relation to food then, young people demonstrated nuanced understandings of the everyday challenges of life on a low income and described how different factors compounded each other.

#### 3.1.2. Health-Promoting Spaces and Activities

Health-promoting spaces were generally described as places where young people could exercise and/or socialise, and participants related them to both physical and mental health. Opportunities to access and participate in health-promoting spaces and activities were often perceived to be strongly shaped by socioeconomic position. Many young people emphasised the high cost of access to activities and spaces (e.g., gyms, sports clubs), and participants in South Yorkshire and the North East also talked about the prohibitive cost of public transport to different leisure spaces (London participants highlighted that transport was free for young people). Personal accounts described how this played out: '*we did badminton for a while but then they made it £3 a night and barely anyone could afford it [* ... *] the whole club fell in on itself and stopped because people couldn't pay to attend it*' (SY1.1). Demonstrating a clear sense of injustice, one participant, talking about meeting at the local snooker hall, argued, '*some places can be unnecessarily expensive, and it's not really fair on them though because like all we's want to do is hang out with friends and we's can't get to it*' (NE2.1). The move from 'them' to 'we' in the course of the short narrative serves to convey how this personally affects the participant and their friends and perhaps hints at how acutely young people experienced the unfairness here. Young people consistently contrasted expensive or inaccessible activities with their local youth groups which were perceived as providing a nearby safe, welcoming and affordable space to relax and socialise: '*[the youth group] is good for our health 'cos we get to hang out with our friends and play out back*' (NE2.1). Youth groups then were depicted as attenuating the impacts of poverty and socioeconomic disadvantage.

A small number of participants voiced a belief that, irrespective of income levels, young people could always use outdoor spaces for exercise. However, across all groups participants spoke of how perceptions of safety in their local areas were key inhibitors to accessing public spaces. Participants frequently described high levels of crime within their local areas compared to other places:

*'So, when there are a lot of like stabbings going on in the area, people, like their parents won't let them go outside [* ... *] so I think if crime could reduce in the area maybe people would have more access to these mental health spaces that are available.'* (L1.1)

The phrase 'these mental health spaces' highlights young people's emphasis on the potential for social spaces to positively impact upon their health and wellbeing. Parks were particularly singled out as places that young people could not enjoy to their full potential due to safety concerns: '*I live near a skate park and sometimes I get intimidated when I'm walking past because a lot of the time they're doing like drugs, drinking. On a night time I wouldn't want to be like round there*' (SY2.2). Narratives about risk (crime and safety) were particularly common among female and LGBTQ+ participants across the different areas. While official supervision (e.g., security, police) was noted in some cases to help young people feel safer and support the use of such spaces, such supervision seemed to be rare. Indeed, for many participants, concerns about public anti-social behaviour, and especially the substance use of other people, was noted to shape perceptions and use of space. There was, however, an acknowledgement from some that 'risky behaviours' were also related to exclusion or a lack of activities for young people to engage in: '*if there's nothing to do then we're going to get ourselves into trouble*' (SY1.1). The movement between describing 'others' in narratives about drug use in the local area and the 'we' and 'ourselves' in this extract affords a pertinent example of how participants moved between individualising, othering narratives to a more collective sense of the importance of socioeconomic circumstances in limiting opportunities.

#### 3.1.3. The Relationship between Poor Housing and Poor Health

Young people highlighted the relationship between poor housing and poor health, particularly mental health:

*'I feel like the housing is very cramped in the area, like it's very cramped, like it's very overcrowded and I feel like that also does have a big impact on mental health as well. Because it's so overcrowded you don't have any time to yourself, any time to think, literally with people around.'* (L1.1)

Echoing the perceived shame associated with not eating 'correctly', participants described the shame related to living in the 'wrong' kind of housing: '*I've seen people who've felt embarrassed over it, not wanting to like invite friends over and then they're just kind of feeling alone*' (SY1.1). In this way, young people emphasised how poor housing had a significant impact on their social and emotional wellbeing—through everyday stress, embarrassment and reduced opportunities to socialise in one's own home. Although much less salient in their narratives, they also discussed the relationship between housing and physical health. This was articulated particularly in relation to the COVID-19 lockdown measures, which meant people had to stay at home more than usual. Some described how 'richer' families could afford to purchase home exercise machines and contrasted this with poor people who had neither the financial resources nor space to do so: *'So obviously some people might be in a small flat or whatever, no garden, they might not have the space to exercise either indoors or outdoors as such*' (SY1.2).

#### *3.2. Patterning and Pathways in Socioeconomic Inequalities in Health*

As well as highlighting specific factors linking socioeconomic position and health, young people voiced their understandings of how inequalities were patterned. They described both geographical (regional and localised) and intergenerational patterns of socioeconomic inequality. They also directly and indirectly emphasised the interrelationships between factors affecting health and the complexity of pathways between socioeconomic position and health outcomes.

#### 3.2.1. Regional and Localised Inequalities

Regional inequalities were seen by the participants as underlying socioeconomic inequalities in health. Young people across all groups described a North–South divide in terms of wealth. The government was perceived to be responsible for creating and perpetuating this inequality through uneven investment, as articulated here by one of the London participants:

*'I know that in north England [people] are not as wealthy as the south of England, kind of thing. Because obviously, like, the government, well, over the recent years the government's basically just been focusing on the south of England because of, yeah, that's where the capital is and it's a bit more, the economy in the south of England's a lot better than the north. So I guess, the pandemic has highlighted the fact that they've been, the government has, kind of, been putting the north on the side and just, like, yeah, not paying attention to their needs as much* ... *I feel like as, like, as, like, as a whole that the south of England has just got more investment than the north of England.'* (L2.2)

Young people vividly articulated how differences in local economies and labour markets between the North and South created tangible differences in everyday working and living conditions: '*Well there's obviously more technical industries in London, so like engineering or ICT work. There isn't those jobs in [South Yorkshire town]*' (SY2.2). Local labour markets in the North were perceived to revolve around hospitality and service sector employment, which many associated with low pay, insecurity and low job satisfaction: '*the more like boring [jobs]*' (SY2.1).

Focus group discussions often contained references to much more local-level inequalities too. In the following narrative, reference is made to a 'clear split' in wealth distribution between different areas:

*'There's certain parts of town where you can, they're just known for people being either real poor there or they're barely scraping by and there's also bits where basically people who are wealthy live and it's like quite a clear split. So all those people who live in the, I* *wouldn't say dodgy areas but like with poorer people, they haven't got as good quality of diet and stuff because they're probably living off more cheaper meals that are just packed full of like chemicals or sugars and stuff.'* (SY2.2)

The participant appears to show some awareness that people and places can be stigmatised and that lack of money and place-level disadvantage are barriers to healthy lifestyles. Hence the participant expresses unease about using the word 'dodgy' and appears to be avoiding individualistic, victim-blaming discourses. Some of the young people in London travelled to schools outside their local area, and this seemed to heighten their awareness of localised inequalities:

*'The schools I have been to have generally been in wealthier areas than the area I live in and I've noticed that they definitely have a lot more green space and like generally just a lot more space within school to do sports and stuff as well, yeah* ... *in like less like affluent places like there's more like residential spaces and that's, like people would say like [Name of location] has like an overcrowding like housing issue. And like I think like the main reason why is because in wealthier places like people are more spread out like the sort of like, on, like well people who are more affluent tend to have less children, people who are more affluent tend to like live out, more spaced out from each other* ... *you can get stuck in a cycle because it's so expensive in Central London so then because it's so expensive you're spending your money on other stuff you won't be able to afford to move out to a wealthier area, where there's like potentially, I don't know more green space and less air pollution. So you can, yeah, you can just kind of get stuck in the, that cycle yeah.'* (L2.1)

Here, however, the narrative moves from emphasizing environmental factors (access to green space and better housing) to behavioural factors (affluent families have fewer children) and then back to environmental factors (expensive housing, green space, air pollution). The narrative echoes the interplay and pull between different factors and exemplifies young people's willingness to engage in complex understandings of causal pathways. Further, through their discussions, young people demonstrated an awareness of individualised discourses around blaming. They also consistently highlighted the injustice of the inequality that they perceived: '*It's actually unfair. The facts are right there in front of your eyes, because if you're born quite a poor person, then most people would expect you to stay poor and vulnerable to a lot of diseases*' (NE2.2).

Many participants also discussed how substance use (tobacco smoking and drugs) was more prevalent in their area than other, more affluent areas. There was a suggestion that 'other' young people surrounded by drug taking and drinking would go on to engage in these behaviours themselves: '*Like round my area it's quite bad for drugs and stuff like that* ... *they see other people doing it, it'll make them want to try it and then they'll probably end up getting addicted and stuff like that*' (SY1.1). However, the participants positioned themselves as avoiding the inevitability of this. Thus, they acknowledged structural issues and suggested deterministic outcomes for 'other' young people due to place-based disadvantage but discussed exercising their own agency to avoid this: '*Around my area it's like the teens who are similar to my age have all gone mad with nights out and like drugs and that, so I won't walk out. I see gangs and I'm like no, you're not getting me*' (SY1.1).

#### 3.2.2. Intergenerational Patterns of Inequality

Young people repeatedly articulated the interactions between regional and intergenerational patterns of inequality and frequently commented on the presence and transfer of health-damaging practices through families and within communities. In the following narrative, one young person eloquently describes intergenerational continuity in practices and intergenerational cycles of poverty but also the inextricable link between the two:

*'If they're in a poor area, it's much worse because their mum and dad might just give them a quid and tell them to go and buy their tea, instead of having like a home-cooked meal that's full of good stuff. If that happens in one place, then it'll start spreading in a* *way so more people will be getting poor because, like—let's say one family, if they have two sons and those two sons have sons and they're all like staying in the same area, it multiplies and then there's like these areas where there's shops and stuff and it's all corner shops where they sell like ready meals and stuff and they all live off that and then they don't have as good a diet, which isn't their fault in the first place, it's just where they were born and put into the world.'* (SY2.2)

Highlighting the permeation of adverse health practices, there was an explanation of how health practices, such as diet, were shaped by experiences and exposure to parents' and peers' behaviours in a sociocultural context, which were described as 'normalising' such practices:

*'Like we were discussing earlier about, your personal life, your friends and family, and you might adapt to how they are. So if a parent is eating fast food almost every day, then the child might say, "Actually, do you know what? That's OK because my dad is doing it.'* (NE2.2)

However, again this needs to be understood alongside young people's foregrounding of the influence of economic and environmental factors on health practices, particularly in relation to food.

3.2.3. Interrelationships between Factors Linking Socioeconomic Position and Health

Young people's discussions consistently highlighted the interrelationships between factors linking socioeconomic circumstance and health. The complex aetiology of health inequalities was both directly and indirectly acknowledged with understandings rooted in experience. Highlighting their focus on the interrelationship between different factors and the pathways through which inequalities were created and perpetuated, they articulated pathways between root causes (such as local labour market precarity) and secondary factors (such as not being able to afford to eat healthily):

*'I think obviously because there's high rates of unemployment and that links to not having money and then not like spending loads, that all links into like buying the cheapest food, which is not naturally healthy. So it all kind of links really.'* (SY1.1)

The phrase 'it all kind of links really' encapsulates young people's emphasis on the interwoven nature of inequalities. However, they consistently foregrounded poverty as the root cause of socioeconomic patterned inequalities in health: *'if you don't have a very good income then you can't really live in a very good house. It can affect your health as it is and can cause like, it can cause stress which can cause other things*' (SY1.2). The bounding and constraining impacts of stretched financial resources upon health practices and outcomes were clearly highlighted: '*I feel like money is one of the biggest factors for nearly everything, diet, mental health*' (SY1.3). However, while in general young people's narratives demonstrated their awareness of the socioeconomically disadvantaged nature of their local area, at times their discussions hinted that they associated poverty with others rather than themselves. In particular, one participant from one of the North East groups, the most socioeconomically disadvantaged area we worked in, noted: '*The less fortunate could actually find it harder* ... *they're not as privileged as we are in terms of money and wealth*.' (NE2.3).

Whether they explicitly made the link or not themselves, young people's narratives illuminated the importance they attached to the impact of poverty on mental health, typically the everyday, chronic stress and strain of living in poverty. They highlighted particular pinch points where limited financial resources were acutely stressful: '*I think there's a certain level of stress if you go knowing that you've maybe not got as much money and there's going to be certain times of the month where you have to really mind what you're spending*' (NE1.2). Mental health was perceived to be a consistent 'link' within a causal cycle of inequality—linked both to a decreased likelihood of engaging in healthful behaviours (such as eating well, engaging in exercise and labour market engagement) and an increased likelihood of engaging in risky behaviours (such as drinking and taking drugs):

*'mental health is, like, connected to so many other things* ... *like, reducing physical activities, and diet, and stuff like that. So mental health is, kind of, like, it could be, like, a major cause for the other things to happen so, like, comfort foods, for example, eating when you're, like, depressed or something, or not getting out of bed due to, like, lack of food, like, due to depression so physical activity is just lower.'* (L2.3)

#### **4. Discussion**

#### *4.1. Social Determinants Shaping Health: Interacting Factors and Complex Pathways*

Through the course of their discussions, participants in our study often demonstrated nuanced understandings of how socioeconomic circumstances shaped health outcomes. Young people's narratives showed how they were making sense of inequalities in health as they talked—at times echoing but, crucially, moving away from more populist individualised, neoliberal explanations for inequalities. Understandings of the relationship between socioeconomic circumstance and health, then, were not fixed and static but rather malleable and dynamic [17]. Overall, their narratives demonstrated a subtle appreciation of the ways in which the SDH get 'under the skin to shape health' through 'interacting material, psychological and behavioural pathways' (Raphael's (2011) Discourse Level Three) [12] (p. 226).

#### 4.1.1. Material Pathways

Young people demonstrated an acute awareness of how differential access to material resources shaped opportunities to eat healthily, access health-promoting spaces and activities and enjoy good housing. Poverty was perceived to be all-pervasive, and young people consistently emphasised limited financial resources as a major barrier to health [48]. However, they also emphasised how this was exacerbated by other factors such as local infrastructures and perceived safety [49]. They also highlighted the uneven socioeconomic patterning of time—a finding not foregrounded by previous research exploring public perspectives of socioeconomic circumstances and health [17]. Their descriptions of the everyday stresses for low-income parents managing unsociable hours and caring responsibilities, particularly in relation to providing healthy food, resonate strongly with Strazdin et al.'s (2016) call to consider time as a social determinant of health as it has the potential to affect so many opportunities for good health—including time to engage in health-promoting activities, rest and care for each other [50]. The participants' emphasis on time perhaps also hints at a weakness in the SDH framework which deals with social 'domains' and determining factors, rather than the mechanisms through which inequalities are sustained.

#### 4.1.2. Psychosocial Pathways

Young people consistently highlighted the importance of psychosocial mechanisms linking socioeconomic circumstance and health inequalities. They discussed the importance of mental health as a critical element in understanding the pervasive, complex influence of socioeconomic circumstance on health behaviours, experiences and outcomes (Discourse Level Three) [12]. Echoing previous studies with mostly adult participants this was particularly poignant in relation to housing [39,51–54]. Young people described both acute and chronic stress of living in inadequate housing, including the associated shame and stigma [39], offering poor housing as an important reason for higher rates of mental ill health among lower socioeconomic groups (Discourse Level Three) [12]. Participants' discussions regarding socioeconomic inequalities in access to safe, green spaces and the 'complex mix of spatial and social intertwinings' also highlighted the impact on mental wellbeing [55] (p. 8). Such understandings contrast with findings from the recent Frameworks Institute project where participants foregrounded a 'mentalism' model in which 'mental health issues such as depression and anxiety [ ... ] were seen as being determined by an individual's mindset' (their self-discipline and willpower) [18] (p. 7).

#### 4.1.3. Behavioural Pathways

At times, and particularly early on in discussions, young people emphasised the uneven patterning of risky health behaviours between socioeconomic groups, particularly in relation to substance use, smoking and alcohol (Discourse Level Two) [12]. This resonates with survey-based data with adults [18,53,56], and some qualitative work with both adults and young people [18,26]. Importantly, however, in the context of their discussions, young people's narratives in our study frequently shifted towards a subtler appreciation of the role of the SDH, emphasising material and environmental factors as underpinning health behaviours. This finding differs markedly from recent UK-based research which characterises public understandings as focusing on personal responsibility and contrasts this with expert opinion (among those working in the field of social determinants) that behaviours are the very 'endpoint in a long chain of causes and consequences that produce health outcomes' [18] (p. 7). The young people's accounts in our study resonate much more closely with the 'expert' understandings. They also echo the 'integrated explanations for socioeconomically patterned inequalities' evident among the mainly adult participants in Smith and Anderson's (2018) meta-ethnography [17]. Importantly, however, both the participants in our study and the majority of those in the studies in Smith and Anderson's (2018) review lived in socioeconomically disadvantaged areas and thus had personal experience of how inequalities played out in everyday life [17]. This may well explain the divergence.

#### *4.2. The Role of Public Policy Decisions (and Their Underlying Ideologies)*

While young people did not explicitly discuss the intersection between material living circumstances and gender or race (and only rarely referred directly to 'class' (Discourse Level Four) [12]), young people's narratives sometimes demonstrated a critical consciousness of the role of public policy decisions and their underlying political philosophies in creating and sustaining inequity (Discourse Levels Five and Six) [12]. However, this was only really evident in their discussions regarding uneven geographic labour market precarity and the absence of regeneration investment [57]. A lack of political will to invest in the North (and vested interests in ensuring the success of the South) and underinvestment in certain local areas were directly blamed for reducing opportunities for good work and living conditions and, ultimately, good health [58,59]. The narratives echo previous research in which adult participants perceived some policies to be more favourable to some groups than others [14,51,60]. Our participants were also acutely aware of the unequal impact of the economic fallout of the COVID-19 pandemic on already disadvantaged young people [61], echoing research showing that young people suffer disproportionate impacts upon their employment trajectories and wages when exposed to economic uncertainty [62]. Young people's emphasis on the unacceptability of poverty and scale of inequality contrasts with earlier studies (e.g., Shildrick and MacDonald's 2013 study [47]). But reflects broader shifts in societal attitudes with 'both phenomena being more widely regarded as prevalent and unacceptable than in the past' [63] (p. 164). However, in general, there was much less evidence that participants spoke to Raphael's Discourse Level Seven—about the power imbalances that underpin the uneven distribution of the SDH [12]. Health inequalities were described in relation to slightly abstract or faceless phenomena such as unemployment, poverty and regional inequality, but there was very little discussion about who has the power and how it is used to privilege some and marginalise others.

#### Determinants of Health Inequalities?

While young people's narratives offered apparently little space to disrupt the pathways between socioeconomic insecurity and health inequality, somewhat paradoxically, young people at times positioned themselves as avoiding the inevitability of this. Area fatalism and individual agency to resist risky health behaviours, for example, sat side by side. This was particularly evident in relation to (avoiding) substance use. Their emphasis on 'room for agency' to some extent echoes concerns about the language of social 'determinants'.

McMahon (2021) highlights that such a framing can perpetuate a reductionist approach to health inequalities [64]. Taken to its logical endpoint, this reduces individual people to 'puppets on a string' [65] (p. 475) and loses sight of the interaction between individuals, services, materiality and health [22]. The tension, however, was much less evident in relation to eating healthily and engaging in health-promoting activities where young people were more likely to share personal stories of the barriers they themselves faced [66]. This perhaps links to a greater acknowledgment of the bounding influence on poverty in relation to the food and exercise within public discourses more broadly. Indeed, at the time of the focus groups, a campaign for free school meals, led by Marcus Rashford, a prominent English football player, was the centre of much media attention [67], and the unequal impact of COVID-19 on people's everyday living and working situations was very much in the spotlight [68].

Further, our analysis of young people's emphasis on the interrelationships between pathways to inequalities also supports calls to move away from depicting discrete categories of determinants in relation to health inequalities [69]. Indeed, Dahlgren and Whitehead (2021) highlight that their rainbow model was only ever meant to depict determinants of health, not determinants of health *inequalities* [70]. To fully understand the root causes of health inequalities, they argue, we need to 'take a further conceptual leap and focus on the pathways and mechanisms by which [ ... ] determinants [ ... ] bring about social gradients in health' [70] (p. 22). Focusing on pathways and mechanisms in this way may also help to address the thorny issue of adequately articulating how health-relevant practices are constrained by people's social and economic environment without inadvertently disempowering and further stigmatising underserved communities [64].

#### *4.3. Study Limitations and Strengths*

Our sample of young people from socioeconomically deprived areas may limit the relevance of our findings for young people from more affluent areas. It also plays into a wider critique that by focusing on areas of socioeconomic deprivation such areas are perceived as the only communities in which inequality matters [17]. Further, while our sample as a whole is ethnically diverse, all participants in our North East and South Yorkshire groups were White British.

Our decision to prioritise participant confidentiality also means that we have not provided individual participant demographic information alongside quotes. While this limits our ability to explore the extent to which individual participants held different views and the ways in which their understandings may have developed in the course of the discussions, we believe our commitment to confidentiality helped to facilitate young people's engagement and openness during data generation. We were guided by a desire to ensure young people felt able to talk as freely as possible in the focus group setting. Indeed, we appreciated the limits of confidentiality in group discussions and therefore framed our questions in ways that ensured participants did not have to disclose personal information if they did not wish to and encouraged them to talk generally about people in their areas in light of this. This often resulted in discussions about their experiences and perspectives framed around '(some) young people'.

It is also important to acknowledge the potential limitations of recruitment through existing youth organisations. Many youth organisations undertake work around health; therefore participants may have had more awareness about health inequalities than other groups of young people. Nevertheless, working closely with youth groups afforded many benefits. Youth workers helped to refine our topic guides and facilitate participant engagement, and they provided an invaluable source of trusted support for participants (see Woodrow et al., 2021 [31]).

Our approach of using three interlinked focus groups provided an opportunity to develop rapport, sense check and build on ideas over the sessions. The supportive atmosphere of the focus group in which young people were surrounded by peers and youth workers they knew, as well as research team members experienced in working with young

people, perhaps helped to foster a more critical take and to enable participants to challenge each other. The context afforded young people a forum in which to develop understandings rather than being solely a means of extracting ideas. This highlights the importance of giving young people time and space to discuss and reflect on their perspectives on health inequalities [71,72]. Perhaps most importantly, we received consistently positive feedback from both participants and youth leaders across the three areas. Indeed, the retention of our participants over the series of three focus groups, which involved young people actively joining to participate in their free time (both whilst at home and during their youth groups sessions), demonstrates their engagement with and commitment to the project.

Generating data during the COVID-19 pandemic also afforded a unique lens through which the young people viewed and subsequently discussed inequalities in health. Indeed, many young people recognised the unequal impact of the pandemic on health and were, to some extent, aware of the way existing inequalities have been exposed by the pandemic [11]. Therefore, this may help explain some of our findings around young people's nuanced appreciations of the links between socioeconomic position and health.

#### *4.4. Priorities for Future Research*

More research exploring young people's perspectives on the relationships between socioeconomic circumstances, inequality and health is needed to address the current paucity. In particular, work with marginalised groups (such as looked-after children, care leavers, homeless young people, young people not in education, employment or training) who may be more likely to experience adverse social determinants of health would be beneficial [73]. Conversely, work with young people from more affluent contexts would provide interesting comparison and help counter a more general focus in the literature on areas of socioeconomic deprivation [17]. Further, research with groups not recruited through youth organisations would help explore if the perspectives found in our work were shaped by the participants' involvement in youth organisations. Finally, it would be beneficial to explore ways to more effectively discuss, describe and teach topics of health inequality and look at ways to explore such topics in ways that are not stigmatising or fatalistic but that encourage positive social change [71].

#### *4.5. Policy and Practice Implications*

Our study highlights an ongoing need for policies that address young people's everyday socioeconomic realities and experiences. First and foremost, young people's emphasis on the all-pervasive impact of poverty on their opportunities to enjoy good health underscores the importance of pro-equity policies to end poverty. Their foregrounding of the uneven socioeconomic patterning of time and its impact on health and wellbeing highlights a need to tackle long (and often unsociable) working hours for people living in the most deprived neighbourhoods [74]. Further, there is an ongoing need for policies that address the conditions and impacts of unsuitable housing and that make it easier for young people, particularly those in socioeconomically disadvantaged areas, to eat more healthily and access health-promoting activities and spaces.

While local authorities have responsibility to implement important practical changes here (e.g., enhancing green spaces and parks, making streets safer and establishing cycle lanes), this needs to be enabled by funding. The public health grant awarded to local authorities is currently one billion pounds lower (in real terms per capita) than it was in 2015/16 [24], and reductions in funding allocations have been higher in the poorest areas of the country [75]. In particular, young people in this study highlighted that youth clubs afford a safe space to socialise with peers, access information and advice and form trusting relationships with professionals. Yet, policy decisions have resulted in significant drops in funding for youth services with, for example, 750 youth centres forced to close between 2010/11 and 2018/19 [76]. This worrying trend has been exacerbated by increased funding pressures during COVID-19 [77]. Further, while on the one hand our study points to the importance of cross-sectoral action across a range of policy areas [46,78], we are wary here of falling into the trap of 'shifting from a social inequality to a health inequality frame' [79] (p. 653), and focusing our attention on the lower rather than the higher levels of Raphael's (2011) seven discourses [12]. Such a framing, Lynch (2017) argues, can serve to make tackling inequalities seem like an insurmountable problem and divert attention away from policies (such as taxation, redistribution and labour market regulation) that we know will impact upon socioeconomic inequalities and, in doing so, health inequalities [79,80].

#### **5. Conclusions**

Our study affords an important contribution to the dearth of exploration around young people's perspectives on inequalities in health [17,27,28]. Our focus on areas of high deprivation provides important insights and contributes to the limited body of work exploring the perspectives of people living on a low income in socio-epidemiological research more broadly [22,81] and calls for policy to tackle inequalities to be 'grounded in the realities of people living in poverty' [82] (para.2). Our study demonstrates that young people understand that health inequalities are generated by social determinants of health, which in turn influence behaviours. They highlight a complex interweaving of pathways between social determinants and health outcomes. However, they do not tend to think in terms of the SDH and their distribution as resulting from the power and influence of those who create and benefit from health and social inequalities. It may be that they are unused to thinking in this way or that they have understandings that we have not fully appreciated. An informed understanding of the causes of health inequalities, influenced by their own unique generational experiences, is important to help young people achieve greater equity in the future than they perceive at the present.

**Supplementary Materials:** The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/ijerph19063679/s1, Supplementary File S1: Topic guides; Supplementary File S2: Participatory map; Supplementary File S3: Coding framework.

**Author Contributions:** Conceptualisation, H.F., C.D.-R., M.E., K.L., S.S., C.S. and E.G.; Data curation, H.F., N.W., M.C., E.H., N.G., V.E., P.K. and C.D.-R.; Formal analysis, H.F., N.W., M.C., E.H., N.G., V.E. and P.K.; Investigation, H.F., N.W., M.C., E.H., N.G. and V.E.; Methodology, H.F., N.W., M.C., E.H., N.G., V.E., M.E., K.L., E.R., S.S., C.S. and E.G.; Project administration, H.F., N.W., M.C., E.H., N.G., V.E., P.K., C.D.-R., E.R., C.S. and E.G.; Resources, H.F., C.S. and E.G.; Supervision, H.F., C.D.-R., M.E., K.L., S.S., C.S. and E.G.; Writing—original draft, H.F., N.W., M.E., R.M. and E.R.; Writing—review and editing, M.C., E.H., N.G., V.E., P.K., C.D.-R., M.E., K.L., R.M., E.R., S.S., C.S. and E.G. All authors have read and agreed to the published version of the manuscript.

**Funding:** This project is funded by the National Institute for Health Research (NIHR) School for Public Health Research (SPHR) (grant reference PD-SPH-2015). The views expressed are those of the authors and not necessarily those of the NIHR or the Department of Health and Social Care.

**Institutional Review Board Statement:** The study was approved by the School of Health and Related Research (ScHARR) ethics committee at the University of Sheffield. Date of approval: 25 November 2020. Ethics form reference number: 037145. All methods were carried out in accordance with relevant guidelines and regulations.

**Informed Consent Statement:** Informed consent was obtained from all subjects involved in the study. Written informed consent included consent for publication of the findings and the use of anonymised quotations in publications.

**Data Availability Statement:** Data available on request due to restrictions. The data presented in this study are available on reasonable request from the corresponding author. The data are not publicly available due to privacy reasons.

**Acknowledgments:** The authors would like to thank the members of our stakeholder steering group for their support and input throughout the project. We thank members of the youth organisations who piloted and provided feedback on our data generation tools and methods and Naoimh McMahon for commenting on an earlier draft of the manuscript. Finally, we thank the young people and youth organisations that took part in the research for their contributions, insights and enthusiasm.

**Conflicts of Interest:** The authors do not have any conflict of interest. The funders had no role in the design of the study; in the collection, analyses or interpretation of data; in the writing of the manuscript or in the decision to publish the results.

#### **References**

