**Preface to "Inequalities in Health and Healthcare"**

Tackling inequalities in health and healthcare is more important than ever. The COVID-19 pandemic has already starkly illustrated the disproportional impact of the virus on those who already face disadvantage and discrimination (Bambra C et al. https://jech.bmj.com/content/74/11/964). Moreover, there is already evidence that the public health measures taken to contain the virus are likely to have longstanding differential impacts across populations.

Numerous studies have documented avoidable differences in health, within and between populations. Similarly, studies have consistently shown inequalities in access, use, experience and outcomes from many types of healthcare and public health programmes. The focus has often been on individual determinants, such as gender, age and ethnicity. Less attention has been paid to structural or contextual determinants, except for area-level socioeconomic conditions. In addition, to tackle inequalities, there is a need to move beyond measuring, in order to understand why these inequalities arise and how they can be addressed.

This Special Issue seeks to extend the parameters of inequalities research in health and healthcare beyond measuring and documenting inequalities. Reviews, observational studies, and quasi-experimental and other evaluation designs (using quantitative, qualitative or mixed methods) focusing on the following were welcomed for submission:


**Jessica Sheringham and Sarah Sowden**

*Editors*

### *Article* **COVID-19 at the Deep End: A Qualitative Interview Study of Primary Care Staff Working in the Most Deprived Areas of England during the COVID-19 Pandemic**

**Claire Norman \*, Josephine M. Wildman and Sarah Sowden**

Population Health Sciences Institute, Faculty of Medical Sciences, Newcastle University, Newcastle upon Tyne NE1 4LP, UK; josephine.wildman@newcastle.ac.uk (J.M.W.); sarah.sowden@newcastle.ac.uk (S.S.)

**\*** Correspondence: claire.norman7@nhs.net

**Abstract:** COVID-19 is disproportionately impacting people in low-income communities. Primary care staff in deprived areas have unique insights into the challenges posed by the pandemic. This study explores the impact of COVID-19 from the perspective of primary care practitioners in the most deprived region of England. Deep End general practices serve communities in the region's most socioeconomically disadvantaged areas. This study used semi-structured interviews followed by thematic analysis. In total, 15 participants were interviewed (11 General Practitioners (GPs), 2 social prescribing link workers and 2 nurses) with Deep End careers ranging from 3 months to 31 years. Participants were recruited via purposive and snowball sampling. Interviews were conducted using video-conferencing software. Data were analysed using thematic content analysis through a social determinants of health lens. Our results are categorised into four themes: the immediate health risks of COVID-19 on patients and practices; factors likely to exacerbate existing deprivation; the role of social prescribing during COVID-19; wider implications for remote consulting. We add qualitative understanding to existing quantitative data, showing patients from low socioeconomic backgrounds have worse outcomes from COVID-19. Deep End practitioners have valuable insights into the impact of social distancing restrictions and remote consulting on patients' health and wellbeing. Their experiences should guide future pandemic response measures and any move to "digital first" primary care to ensure that existing inequalities are not worsened.

**Keywords:** COVID-19; health inequalities; general practice; primary care; social determinants of health; social prescribing; remote consulting; marginalised communities; health care inequalities; health/healthcare inequity

#### **1. Introduction**

The COVID-19 pandemic has highlighted the impact of unfair and avoidable health inequalities, with death rates in deprived areas of the UK three times those of more affluent areas [1–3]. In this article, we make known the immediate and longer-term impacts of the pandemic from the perspective of primary care staff working in areas of blanket socioeconomic deprivation.

Governments around the world have used social distancing measures to slow the transmission of the virus. Social distancing aims to reduce virus transmission between households and includes physical distancing measures and the closure of sites of transmission, including schools and non-essential businesses. These measures are also referred to as "lockdowns". Although lockdowns have been shown to be effective at suppressing the number of cases (and therefore deaths) from COVID-19 [4,5], concerns have been raised that the social distancing measures themselves are not without harm and that these harms will fall disproportionately on those living in disadvantaged circumstances [6]. These harms range from early consequences of people delaying medical assessments for COVID

**Citation:** Norman, C.; Wildman, J.M.; Sowden, S. COVID-19 at the Deep End: A Qualitative Interview Study of Primary Care Staff Working in the Most Deprived Areas of England during the COVID-19 Pandemic. *Int. J. Environ. Res. Public Health* **2021**, *18*, 8689. https://doi.org/10.3390/ ijerph18168689

Academic Editor: Myriam Khlat

Received: 2 July 2021 Accepted: 13 August 2021 Published: 17 August 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/).

or non-COVID illness, to the longer-term effects of economic decline. In the UK, social distancing and lockdown policy has, at times, meant that schools have been closed to the majority of children and many services, including general practitioner (GP) surgeries, have attempted to reduce the amount of face-to-face contact they have with patients: appointments have become telephone or video by default, with 90% of consultations being via telemedicine [7]. The pandemic also necessitated the creation of "hot sites" for assessing patients with suspected COVID-19: these were often led by Primary Care Networks and staffed by local GPs [8]. Latterly, GPs have also co-ordinated community COVID-19 vaccine rollout.

North East England is the setting for this study. The North East of England has the lowest life expectancy in the country and in its most deprived areas, between the years 2010 and 2012 and the years 2016 and 2018, life expectancy has actually been decreasing [9]. The region performs the worst or second worst in the country for causes of death considered preventable: suicide and drug misuse at any age, as well as cancer, cardiovascular, respiratory, and liver diseases in those under 75. These differences can be attributed to a range of factors: high rates of poverty, poor housing, and the health and social sequelae of heavy industry and its more recent decline [10].

General Practitioners (GPs, also known as family practitioners or primary care practitioners), are the main providers of community care for acute and chronic illnesses in the UK. Everybody resident in or visiting the UK is entitled to access free medical care on the National Health Service (NHS) via their GP, although visitors may be charged for some hospital treatments or prescription medications. As one of a range of measures being implemented to address health inequalities, the North East and North Cumbria (NENC) is in the process of establishing a "Deep End" GP network of professionals working in practices in the region's most deprived areas. Deep End practices serve populations living in areas of blanket deprivation with high proportions of patients living in the 15% most deprived local areas, based on postcode data. The NENC Deep End network was inspired by the GPs at the Deep End network in Scotland: a forum for advocacy, sharing ideas and developing interventions to mitigate health inequalities [11,12]. Since the founding of the original Scottish network in 2009, Deep End GP networks have been founded in several other regions of the UK, plus Ireland, Australia and Canada.

Social prescribing is a relatively new intervention which acknowledges the impact of the social determinants of health on people's health and wellbeing. The aim is to use community organisations and other non-medical support services to address factors such as loneliness and poor housing, as an addition or alternative to offering clinical or pharmaceutical treatments to patients who may have multiple conditions or co-morbid mental and physical health problems [13]. The NHS England Long-Term Plan commits to providing social prescribing as part of its Universal Personalised Care model [14] and, typically, this is delivered by groups of GP practices via Primary Care Networks. The most common social prescribing model is for patients to be referred to a "link worker" (also known as care navigators or health trainers) who can identify the most appropriate service for their needs. Many of these link workers began working during the COVID-19 pandemic.

Deep End GPs and other primary care practitioners will have unique insights into the effect of COVID-19 on their communities, and the impact of public health measures designed to reduce viral transmission. In this qualitative study, conducted during the UK's second wave of COVID-19, we aim to explore experiences of delivering primary care in a pandemic among staff working in practices in areas of blanket deprivation in North East England.

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

Data were collected as part of a wider project on the co-design of a Deep End network for the North East and North Cumbria. There was no direct patient or public involvement in this co-design project; however, the work was informed by a multi-agency steering group of policy and practice partners and researchers from across the NENC region.

Ethical approval was granted by Newcastle University research ethics committee (ref: 4322/2020).

#### *2.1. Participants and Recruitment*

Practices included in the core Deep End NENC network were identified using the methodology applied in the Scottish Deep End project. This entails identifying the proportion of each practice population living in the 15% most deprived areas of England, based on the Index of Multiple Deprivation (2019) and NHS Digital Practice Populations by Lower layer Super Output Areas (LSOA) (January 2020). All practices were ranked, and 34 North East and North Cumbria practices were found to be amongst the 10% most deprived practices in England against this measure.

A purposive framework was used to sample within the 34 NENC Deep End practices, prioritising geographical representation from Deep End practices across the NENC region. We also aimed to speak to participants with different levels of Deep End experience. Invitations to participate were sent via email to all Deep End practices, in addition to convenience and snowball sampling of participants known to the research team and purposive sampling of staff from practices that were the only Deep End practice in their locality. All staff members in Deep End practices were invited to attend.

Participants were sent a participant information leaflet and consent was gained by electronic completion of a form or by recording verbal consent at the start of the interview.

#### *2.2. Data Collection*

Interviews were carried out between October 2020–March 2021. In the temporal context of the COVID-19 pandemic, most interviews were undertaken in the early stages of the UK second wave. Cases were once again starting to rise, and lockdown restrictions (either local or national) were being tightened. Schools were open at this time and all but two interviews were prior to the approval of any COVID-19 vaccines.

Interviews were conducted by C.N. and J.M.W. and recorded using the Zoom video conferencing platform (Zoom Client for Meetings, Version 5.7.5 (939), Zoom Video Communications, Inc., San Jose, CA, USA). A topic guide was developed by the research team and used to provide a semi-structured approach, and this guide was continuously updated in line with emerging themes and participant feedback. Interview data were stored on a secure password protected server, accessible only by the research team.

#### *2.3. Data Analysis*

Interviews were auto-transcribed using Zoom video conferencing software, with manual corrections by C.N. C.N. and J.M.W. engaged in ongoing constant comparison of the data, allowing concurrent collection and analysis. Interviews were double coded to enhance validity. Thematic content analysis was used to code the transcripts and categorise them into emerging inductive themes [15]. NVivo (version 12, QSR International (UK) Ltd., Cheshire, UK) was used for data management and to support data coding [16].

#### **3. Results**

#### *3.1. Participants*

Fifteen interviews were carried out with primary care practitioners (Table 1): eleven GPs, two social prescribing link workers (LW), one nurse practitioner (NP) and one district nurse (DN). We spoke to participants with a breadth of experience: their careers in Deep End practices ranged between 3 months and over 30 years. All participants worked in urban areas, reflective of the areas of high blanket deprivation in the region. A wide geographic coverage was achieved; all but one of the North East's Clinical Commissioning Group (CCG) areas were represented; at the time, CCGs were the organisations responsible for commissioning health services for individual geographical areas in England. A decision was taken to end recruitment as the emergence of recurrent themes suggested data saturation. We also found that the UK vaccine rollout was becoming a priority for primary care staff.



Our findings can be categorised into four overarching themes: (1) factors increasing the direct health risks of COVID-19 virus; (2) factors worsening pre-existing deprivation; (3) the role of social prescribing during COVID-19; (4) the benefits and costs of remote consulting.

#### *3.2. Factors That Increased the Health Risks of COVID-19*

Participants gave several reasons why COVID-19 cases and deaths might be higher in communities with high levels of deprivation. Multi-morbidity, rather than advanced age, was identified as the major risk factor for patients living in communities where *"getting to 55 would be pretty good"* (Interview 2, GP).

Concerns were raised about patients' low levels of health literacy, which reduced their understanding of health messaging around COVID symptoms. As one GP noted, even widely publicised symptoms were not triggering patients to seek testing:

*"I still find it amazing, a guy I spoke to last week, cough and breathlessness: "do you think it could be COVID?" "Oh, I don't know." "Have you had a COVID test?" "No." "Do you know how to get one?" "No, how do I do that doc?" And you just think, surely, with the last six months, the media, all the rest of it but it'd just not crossed his mind." Interview 8, GP*

Some participants noted that social distancing measures were not being adhered to in their communities. A lack of understanding, rather than deliberate rule flouting, was identified as a possible cause for non-observance:

*"I did a home visit yesterday, driving up the street, and we're meant to not be socialising with anyone out of the household and I drove past about 13 people in a garden sharing a fag over the fence* ... *You wonder how much of it is just like, I don't want to follow it and how much is actually understanding the impact of your potential action." Interview 3, NP*

Patients' ability to access to healthcare was identified as one of the most pressing challenges facing Deep End practices. One GP expressed concerns that the local "hot hub" facilities for assessing patients with suspected or confirmed COVID-19 in the community were inaccessible for those without cars.

*"The local hot site, say for patients in (local area) for COVID, you have to have a car to go* ... *That does not help our patients" Interview 13, GP*

Access issues created health risks for staff as well as patients. Lack of local testing and assessment services put pressure on GP surgeries to continue seeing symptomatic patients, potentially raising COVID case numbers among staff. Staff mentioned that their premises were smaller which made social distancing difficult.

*"The rooms are a lot smaller, patients are harder to manage on the phone because there's lots of digital poverty, for example, people turn up and ask to be seen who don't have a phone* ... *So, I see a lot more people face-to-face here and probably as a result, I got COVID a month ago" Interview 13, GP*

In addition to presenting a risk to staff health, exposure to the virus was creating staff shortages:

*"We had a big outbreak of staff, having it in lockdown one, there was an entire team went off with it" Interview 5, DN*

*"We had 11 members of staff off last week. Just complete carnage, trying to manage" Interview 11, GP*

District nursing staff felt that they were being asked to take on extra responsibilities because some GPs were, understandably, trying to reduce patient contact.

*"It has been really hard to get GPs out to see anyone and a lot of the time we find as nurses that we're telling GPs what we think over the phone, and they're saying, "Okay, yeah, we'll go with that" and not seeing the patient. So, even as far as palliative care—we're having patients that haven't been seen that are dying. And it's been quite tough for the families because, you know, they would quite like to see a doctor* ... *it does feel a bit like we are expected to diagnose someone so that they don't have to visit." Interview 5, DN*

#### *3.3. Factors Likely to Exacerbate the Effects of Deprivation in These Areas*

Although recognising the need for lockdown measures, Deep End staff found that the social distancing measures were having a huge impact on their communities. Social distancing meant that community initiatives that practices had put in place were on hold, including social groups for isolated patients and group consultations for chronic pain.

*"Obviously, it's all on hold at the minute because of COVID, which is making us all feel very uncomfortable because it became a bit of a lifeline really for some of our more isolated patients" Interview 11, GP*

In addition to providing clinical care, supporting patients to address the social determinants of health formed a significant part of the workload in Deep End practices. The reduction in other services such as housing and social work was proving challenging for patients.

*"Housing is a recurring theme* ... *that's been really tricky recently, again because of COVID, it's just (they) basically aren't moving anybody. No matter what circumstances are, really they just will not move them." Interview 6, GP*

Concern around child safeguarding was a common theme, with participants reporting a reduction in family contact from health visitors and social workers. School closures also meant a reduction in safeguarding oversight, which was a source of deep concern to practice staff.

*"The child safeguarding situation fills me with dread* ... *throughout lockdown because we've been one of the services that's remained open and visible, we're being presented with a lot of this stuff which is difficult. And we're being presented with it without lots of the support that we normally have to manage it." Interview 1, GP*

Although telephone contact was happening, this was not felt to be adequate.

*"Health visitors are not doing a lot of face-to-face visits* ... *there's always the risk that we're missing things because they're not being seen face-to-face—it's just been telephone." Interview 2, GP*

The long-term effects of the pandemic were also a significant concern: communities that were already struggling economically may not be able to recover.

*"The legacy of this, the unemployment, the deprivation, that's just going to get worse for patients because as with all of these things, our communities will be the hardest hit going forward. They're not going to recover. They're not going to bounce back* ... *in the way that other areas may be able to. And so, yeah, it kinda depresses me really 'cos I just think this is just going to make things worse* ... *That's my concern from COVID, is just it's just going to push these communities further down." Interview 11, GP*

#### *3.4. Social Prescribing during COVID-19*

A lot of social prescribing services in the North East of England were commissioned just before or during the COVID-19 pandemic. Their roles were constantly changing.

*"For the most part we started in the pandemic. So, a lot of it was initially just COVID response stuff. Yeah, you know. Food parcels, medication deliveries, that kind of thing and like check in calls really for vulnerable patients. So initially we were getting sent a list of COVID patients or 80 plus (year old patients) for example and pretty much cold calling them. You know, we phoned on behalf of the GP practice. "Is there anything you need in this lockdown?" that kind of thing. But we've moved away from that." Interview 14, Link worker*

Lockdown periods saw an increase in referrals, particularly for mental health difficulties.

*"There's a huge amount people getting referred for support with losses and also just anxiety, generally coping with the lockdown, problems with isolation, no contact with families and I think people felt were moving beyond that in the autumn of last year and then to go into another lockdown. I think a lot of people really dipped during the winter with their mental health." Interview 15, LW*

Increased demand plus a reduction in other services due to the pandemic meant that waiting lists were often long, particularly for talking therapies. The social prescriber was seen as a stop gap for patients who were needing extra support *"We're doing a lot of more long-term handholding at the moment"* said one link worker (Interview 14, LW).

However, *"bearing in mind, we're not trained counsellors or therapists"* (Interview 14, LW), social prescribing link workers in the Deep End were proving a valuable resource for patients with non-clinical needs during the pandemic. One GP observed:

*"They've done a lot of very intense work with quite complex and risky people. So they've contained a lot of that complexity and risk, which I hadn't really appreciate that they'd be doing." Interview 1, GP*

However, the link workers themselves found that the provision of adapted services to refer patients to was very variable.

*"(North East (NE) town) seems to have a lot more going on if you like, a lot more of their groups seem to have adapted to the pandemic, so they're offering remote sessions or virtual sessions instead. Whereas in (another NE town) a lot of things seem to have just ceased" Interview 14, LW*

They also found that many referring practitioners and patients had great expectations of what the service could offer, which often did not reflect reality.

*"You know there's a lot of sort of magic wand expectations, and you know the GP saying "right, you're really lonely or isolated. You stuck at home. I'm gonna refer you social prescribing because they can help."* ... *We got to be able to do this, but those aren't* *options at the moment, so we need to look at something else. And I think a lot of the time that could be quite disappointing. Because they're like, "well, you know my GP said that you could help me and, you know, get me out and about and things like that."" Interview 14, LW*

#### *3.5. Benefits and Costs of Remote Provision*

Modifications to the way of working brought some positives and some participants felt that the pandemic acted as a catalyst for change.

*"COVID's been a shot in the arm to make changes that we, you know, have been considering for a while anyway, like changes to our access system" Interview 1, GP*

One participant felt strongly that the increased use of technology was a positive and that the pandemic had just brought forward an inevitable move away from face-to-face consulting.

*"Everybody's doing telephone consultation. You don't need to push anymore. Video. Yeah, everybody's trying to do video consultation, and certainly once they know how to do it. So, I don't want to go back to the old days." Interview 10, GP*

Text messaging was felt to be particularly useful, especially for reaching patients who were sometimes difficult to contact.

The social prescribing link workers found remote working meant they could support more people, because they were not spending so much time travelling; however, this came at the cost of fully assessing their circumstances.

*"I mean, you are not travelling, so you've got more time....So just concentrate on having that contact with people because you're literally picking up the phone and you know you're not having to drive to them, but I think it's really hard to get a picture of people's situations over the phone." Interview 15, LW*

While there was enthusiasm for the potential benefits of increased use of technology, this was tempered by an understanding of the risks of digital exclusion. Although participants understood that reducing physical contact was important during the pandemic, there was concern that the move to "remote by default" consulting may persist. Participants were keen to make sure that any change to consulting methods was not disadvantaging those who already struggled to access healthcare.

*"If nothing else, making sure that what, anything we introduce doesn't disadvantage those already disadvantaged." Interview 8, GP*

The drive for increased remote provision was felt to be a political decision, made without adequate knowledge of the challenges that deprived communities faced.

*"In terms of the technology that Matt Hancock seems to think is the way forward and just because him and all his peers, you know, have access to all the technology and it's very convenient for them to consult with their GP via zoom, that is not how it is for the people where I work." Interview 11, GP*

Online solutions were not felt to be accessible for many patients in the Deep End.

*"It's so hard with COVID, because I could say if there was more* ... *befriending schemes and things, but at the moment, it's all kind of zoom based and none of our patients could really do that* ... *I don't think many of them have smartphones or laptops." Interview 5, DN*

Digital poverty and lack of Information Technology (IT) literacy were often mentioned as concerns in the move to remote consulting. Participants reported that older patients in the communities often did not have internet-enabled devices at all and younger patients may have had devices, but Wi-Fi or data access was variable. Video consulting technology was available but rarely used.

*"I think IT literacy, access to technology, equipment, Wi-Fi, that's been a challenge for us because video consults are often not an option. Even sometimes people who don't even have phones or are unable to take photos, that's made it difficult in terms of COVID and remote consulting." Interview 7, GP*

While some initiatives have attempted to move online, this was felt to be problematic.

*"If you do things via Zoom you're then immediately removing a group of the population who can't be involved." Interview 8, GP*

Social prescribing link workers found that, although some services had moved online, there was mixed enthusiasm for this.

*"We've gone to a digital offer but then that's only accessible to so many people, so as much as we can encourage people to become like digitally active like, not everyone wants to. People want that face-to-face contact." Interview 15, Link worker*

Remote solutions were also problematic for patients with language barriers:

*"I find it really hard with people via an interpreter, trying sometimes to assess what's going on...as opposed to when you just see them face-to-face, it's a lot easier." Interview 13, GP*

Potentially most seriously, some felt the reduction in face-to-face contact was going to irreparably damage the relationship between primary care and the community.

*"I think barriers are going up because people don't have the technology. I think barriers are going up because people like to see their GP you know, and I suspect that's even more so in deprived areas* ... *I think for some people we are a bit of the centre of that community and I think, you know, you put barriers up in that we're saying to them all "well you probably shouldn't be going to the surgery, we'll try and do this over the phone or I'll send something..." you know it's not good." Interview 11, GP*

#### **4. Discussion**

This paper adds important context to the quantitative data on excess morbidity and mortality in deprived populations during the COVID-19 pandemic. We identify mechanisms through which socioeconomic deprivation exposes both patients and healthcare providers to an increased risk of COVID-19. We also add to the literature on the harms associated with some public health measures by highlighting the role of primary care in addressing the social determinants of health and the ways in which the pandemic is likely to worsen existing deprivation. We also highlight the work done by district nursing and social prescribing link workers during the pandemic. Finally, we contribute to the conversation around the move to remote consulting in primary care by identifying the potential risks that the drive to digital-first care poses to the provision of primary care in areas of deprivation.

Reasons for higher morbidity and mortality rates in deprived communities are multifactorial. People in deprived communities are more likely to be exposed to the virus, through overcrowded housing in built-up urban areas and through work in low-paid key worker roles [17]. As noted by the practitioners in our research, pre-existing vulnerability to the disease may be higher, due to increased rates of underlying health conditions such as smoking-related respiratory disease and hypertension [1]. The QCovid risk prediction model considers deprivation alongside other risk factors, such as ethnicity, body mass index and underlying health problems [18]. We welcome the UK government's decision to use the model to identify high-risk patients for shielding and vaccination: not only should this reduce COVID-19 morbidity and mortality in deprived areas, but it has set a precedent for publicly acknowledging the negative health impacts of deprivation. Low health literacy and ever-changing guidelines may lead to underestimation of the risks of the disease [19] which was also reflected in our findings: improving health literacy among the general population should be an urgent health and education priority.

We add that the location of COVID-19 assessment centres may make them inaccessible to low-income households without cars: this should also be considered for test centres and vaccination hubs. Without access to these sites, patients are more likely to visit their usual GP surgery and potentially increase transmission among staff and other patients. Risks to staff health are particularly important given that GP surgeries in deprived areas are likely to have fewer GPs per population, and that those who work there are more likely to be older [20] and, therefore, more vulnerable to becoming seriously unwell with COVID-19. Our participants also raise concerns that smaller premises made social distancing more challenging, potentially contributing to rates of COVID-19 transmission within the surgery. We highlight the need for healthcare spaces to have the space and ventilation required to minimise spread of diseases similar to COVID-19.

We add to the literature on the negative consequences of social distancing and remote working, highlighting the challenges that GPs faced as one of the few visible public services that remained functioning near-normally throughout the pandemic. Our findings highlight the vital role general practice plays in supporting patients with their social needs, but also the need for specific enhanced support in this area. Already vulnerable and socially isolated patients became increasingly so, with the cessation of many support services or organisations, or a move to online technologies that were not accessible for Deep End patients. However, social prescribing link workers were praised for their work, telephoning vulnerable people and navigating their non-medical needs. Concerns around child safeguarding were not unfounded, with a significant reduction in the number of children referred for Child Protection Medical Examinations during the period of the first lockdown [21,22]. A recent report highlighting Child Welfare Inequality shows the unequal distribution of safeguarding interventions across the socioeconomic spectrum and the need for enhanced safeguarding support in deprived communities during and after the pandemic [23]. As we increasingly move to remote and digital services, the value of home visiting and face-to-face encounters for picking up on signs of struggling families, or neglect and abuse, must not be forgotten.

Social prescribing is often seen as a panacea for many of the NHS's problems: supporting complex patients while increasing the amount of GP and nurse time for more traditional medical problems. Our link worker participants describe the challenges of starting new roles during a pandemic and the uncertainty and limited availability of onward and ancillary services, plus the expectation from professionals and patients that they were going to be able to wave a "magic wand". Their experiences of trying to carve out the role is similar to that described by Frostick and Bertotti [24], but likely made even more challenging by the nature of working with complex patients in the Deep End setting [25]. Primary care staff and commissioners should be clear with what they expect social prescribing to achieve in their local area and provide appropriate levels of oversight and supervision.

Our participants' experience of remote consulting matched some of those of Flemish GPs in the early part of the pandemic: fears around missed diagnoses and providing suboptimal care [26]. We add a Deep End perspective to the comprehensive review of Murphy et al., of the move to remote consulting in UK general practice [7]. Digital poverty and lack of IT literacy are major concerns for primary care staff in deprived communities and should be considered as a priority when taking forward the NHS long-term plan for a digital-first primary care [27]. Improving IT literacy among older or vulnerable populations should be considered a necessity to make sure that no-one is excluded as healthcare moves into the digital sphere; collaboration between adult education and health services may be required. Access to internet-enabled devices and affordable mobile data or Wi-Fi will also be vital.

Practitioners in these areas are aware that their surgery can often be a centre of the community. Whether this is an appropriate role for general practice, or possibly a symptom of the general decline of social capital, our participants worry that a move to remote consulting will damage this relationship. These concerns are reflected in research into the portrayal of remote consulting in the media during the pandemic, which showed a decline

in popularity and acceptance between the first and second waves of COVID-19 in 2020 [28]. A recent systematic review found that telephone consulting was favoured by certain groups: women, younger people, very old people and non-immigrants. Similarly, online consulting was weakly associated with younger, more affluent and educated populations [29]. It is vital that the concerns of practitioners in areas of deprivation are reflected in changes to consulting so that the nature of community medicine is not irreversibly altered.

Our district nurse participant describes her discomfort at being asked to act beyond her usual role. This contrasted with the experiences of our GP participants who felt that the reduction in the number of home visits was patient-led. Although inter-practice variability is likely, we suspect that the district nursing views had not been directly sought. Little has been written in the peer-reviewed literature about the experiences of community nurses during the pandemic, but the Royal College of Nurses surveyed its District and Community Nursing members and found similar themes to those raised by our participant [30]. Bowers et al. raise similar concerns, particularly around palliative care, adding that the work of community teams has often been overlooked amid widespread media coverage of those working in hospital [31]. Although a reduction in GP home visits was felt to be necessary to reduce virus transmission, it is important that relationships are not damaged between doctors and their district nursing colleagues, or patients and their relatives. There may also be implications for patient care and diagnosis if nurses and other allied health professionals are not adequately supported. Macdonald et al. argue that home visits for vulnerable and end-of-life patients must remain a priority for GPs as they are the experts in continuity and overseeing complexity [32]. These experiences are unlikely to be specific to deprived communities, but they highlight the need for multidisciplinary pandemic response planning.

#### *4.1. Strengths and Limitations*

Most interviews were carried out by C.N., a GP registrar who worked in a Deep End practice during the first wave. This had a positive impact on accessing participants and was felt that this resulted in interviews that were more candid, as demonstrated in the richness of the data. There was also increased understanding of clinical terminology and local systems. A topic guide was used to avoid shared conceptual blindness and reduce the risk of biasing the agenda with C.N.'s personal experience or opinions [33].

Conducting interviews via video proved acceptable and convenient, as well as COVIDsafe [34]. Participant recruitment was likely adversely affected due to the increasing clinical burden on primary care staff during the second wave of COVID-19 in late 2020 and early 2021, particularly during the vaccine rollout.

#### *4.2. Implications for Practice and Research*

Our findings are relevant to policymakers in both Primary Care and Public Health: it is vital that any public health intervention is ethically implemented, with consideration given to the most vulnerable in society [35]. Care should be taken not to increase existing inequality. They are also relevant to local authorities and adult education teams who contribute to the wider social determinants of health and the growing need for IT literacy.

Although we are pleased to have included the views of nursing staff and social prescribing link workers, future research should include other primary care practitioners, such a health visitors and midwives. Qualitative research with people who live in areas with high rates of COVID-19 infections and deaths would provide even more insight into the potential reasons behind the variation along socioeconomic lines. Further data should be collected on the benefits of social prescribing for patients and other NHS staff during the COVID-19 pandemic.

#### **5. Conclusions**

Deprived communities are facing the brunt of the COVID-19 pandemic. Through the eyes of primary care staff in these communities, we have shown that this goes beyond the impact of the disease itself, with social distancing measures and remote consulting exacerbating many existing inequalities. Deep End primary care practitioners are wellplaced to advocate for their patients and their views are crucial in ensuring that future Public Health measures and major systems changes are implemented in ways that reduce rather than maintain or even increase existing inequalities in health and healthcare.

**Author Contributions:** Conceptualization, J.M.W., S.S. and C.N.; methodology, J.M.W., S.S. and C.N.; investigation, C.N. and J.M.W.; formal analysis, C.N., J.M.W., S.S.; writing—original draft preparation, C.N.; writing—review and editing, J.M.W. and S.S.; supervision, J.M.W. and S.S. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research work is supported by the National Institute of Health Research (NIHR) Applied Research Collaboration (ARC) for the North East and North Cumbria (NENC). J.M.W. is a Research Fellow within the National Institute Health Research (NIHR) Applied Research Collaboration for the North East and North Cumbria. S.S. is supported by Health Education England (HEE) and the 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). C.N. is employed by Health Education England (HEE) as a GP registrar and Academic Extended Integrated Training Post Holder. The views expressed are those of the authors and not necessarily those of the NHS, the NIHR, or the Department of Health and Social Care.

**Institutional Review Board Statement:** Ethical approval was granted by Newcastle University research ethics committee (ref: 4322/2020).

**Informed Consent Statement:** Informed consent was obtained from all subjects involved in the study, including consent to publish.

**Data Availability Statement:** The data presented in this study are available on request from the corresponding author. The data are not publicly available due to containing potentially identifiable information about participants' places of work.

**Acknowledgments:** We would like to thank participants, the NENC Deep End Steering group and the Deep End network. We also thank Clare Bambra and the NENC ARC Inequalities and Marginalised Communities theme for prioritising this important workstream.

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

#### **References**


### *Review* **Relationships between Child Development at School Entry and Adolescent Health—A Participatory Systematic Review**

**Michelle Black 1,\*, Amy Barnes 1, Mark Strong 1, Anna Brook 1, Anna Ray 2, Ben Holden 1, Clare Foster <sup>1</sup> and David Taylor-Robinson <sup>3</sup>**


**Citation:** Black, M.; Barnes, A.; Strong, M.; Brook, A.; Ray, A.; Holden, B.; Foster, C.; Taylor-Robinson, D. Relationships between Child Development at School Entry and Adolescent Health—A Participatory Systematic Review. *Int. J. Environ. Res. Public Health* **2021**, *18*, 11613. https://doi.org/10.3390/ijerph 182111613

Academic Editors: Jessica Sheringham and Sarah Sowden

Received: 1 October 2021 Accepted: 25 October 2021 Published: 4 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/).

**Abstract:** The relationship between child development and adolescent health, and how this may be modified by socio-economic conditions, is poorly understood. This limits cross-sector interventions to address adolescent health inequality. This review summarises evidence on the associations between child development at school starting age and subsequent health in adolescence and identifies factors affecting associations. We undertook a participatory systematic review, searching electronic databases (MEDLINE, PsycINFO, ASSIA and ERIC) for articles published between November 1990 and November 2020. Observational, intervention and review studies reporting a measure of child development and subsequent health outcomes, specifically weight and mental health, were included. Studies were individually and collectively assessed for quality using a comparative rating system of stronger, weaker, inconsistent or limited evidence. Associations between child development and adolescent health outcomes were assessed and reported by four domains of child development (socio-emotional, cognitive, language and communication, and physical development). A conceptual diagram, produced with stakeholders at the outset of the study, acted as a framework for narrative synthesis of factors that modify or mediate associations. Thirty-four studies were included. Analysis indicated stronger evidence of associations between measures of socio-emotional development and subsequent mental health and weight outcomes; in particular, positive associations between early externalising behaviours and later internalising and externalising, and negative associations between emotional wellbeing and later internalising and unhealthy weight. For all other domains of child development, although associations with subsequent health were positive, the evidence was either weaker, inconsistent or limited. There was limited evidence on factors that altered associations. Positive socio-emotional development at school starting age appears particularly important for subsequent mental health and weight in adolescence. More collaborative research across health and education is needed on other domains of development and on the mechanisms that link development and later health, and on how any relationship is modified by socio-economic context.

**Keywords:** child development; childhood education; school; adolescent health; health inequality; adolescent mental health; adolescent weight

#### **1. Introduction**

Inequalities in many child health outcomes are increasing in the UK and the health of those living in its most disadvantaged areas are amongst the worst in the developed world [1]. Some of the roots of health inequality are thought to be in early childhood with socio-economically driven inequalities in child development persisting across the life course, negatively impacting people's future health, wellbeing and life chances, and perpetuating health inequalities into adulthood [2]. Evidence that the early years, or the first "1000 days", is a critical period of development [3,4] (together with health economics research in this field [5]) has meant that the early years have become a prime area for public policy and public health investment in many high-income countries including the UK [6].

All of the countries of the UK provide early childhood programmes, which aim to improve outcomes for children by supporting optimal health and development through access to services such as early education and care, between the ages of 0–4 years or pre-school [7]. There is evidence that programmes which support child development in readiness for school can improve cognitive and non-cognitive skills [8]. There is also evidence that positive cognitive development on starting school is associated with academic achievement by age 13 years [9] and positive socio-emotional development by age 10 years [10]. Non-cognitive skills, such as social skills and self-regulation on starting school, are also associated with later academic success and psychosocial outcomes in subsequent years of childhood and early adolescence [11]. There is less evidence for whether and how child development, or interventions to support child development, are related to subsequent health in childhood. For example, there is limited evidence on the effect of early child development programmes (such as attending pre-school, accessing health services and parenting programmes) on adolescent health, with one systematic review finding little to no effect of early childhood programmes on later child health, although with some evidence for obesity reduction, greater social competence, improved mental health and crime prevention [12]. A review of Sure Start (a UK early years programme from 1999 to 2017, for families with children under the age of four years and targeted in more disadvantaged areas) found that access to Sure Start was associated with fewer childhood hospitalisations for infections and injury [13]. Potential mechanisms proposed for this association were: the provision of information to parents and changing parents behaviour, leading to a safer and more nurturing home, and to reducing externalising behaviour in children, leading to less fights or dangerous activities [13].

To better understand whether and how child development at school starting age is associated with subsequent health in childhood requires a clear understanding of what is meant by "child development", reliable measures of child development, and also the development and testing of conceptual frameworks or theories regarding the relationships between child development and later adolescent health. In terms of defining what we mean by child development, this is contested academic and policy terrain and, as such, is difficult to define. For some, child development is understood through a narrow focus on cognitive education, whereas for others it is about broader life skills, including confidence and social competencies [14]. In English health and education policy, child development has tended to be defined in the former, relatively narrow manner, with, for example, child development at school starting age understood through a specific composite measure of a child's personal, social and emotional, physical, cognitive, and communication and language development, termed "school readiness" [15]. Internationally, school readiness, when considered more broadly, has been seen as a viable strategy to reduce inequalities in learning and development gaps at the start of formal education [16]. However, how it is defined and used in England has been criticised as reductionist, with school readiness used as a performance and accountability measure, resulting in a narrowing of the curriculum, marginalisation of children who fail to achieve required levels of development through grouping by ability, and subjugation of teachers and schools to meet targets [17]. Moving beyond targets to understanding child development more broadly, as an ongoing developmental process in a social context [18], is important if we are to develop interventions to support equitable health and development. Therefore, we consider "child development" in this review as any measure of child development which encapsulates a process of change in what a child is capable or able of doing, or in how they are feeling. There is no existing framework for characterising different aspects or measures of child development. Therefore, in this review we use four over-arching domains of child development: socio-emotional development, cognitive development, language and communication, and

physical development. These domains broadly encompass the areas of learning within the early years curriculum in England [15]. We see these categories as potentially useful despite the described shortcomings of England's composite measure, "school readiness". Conceptualising child development in this way provides a platform for learning about the relationships between specific domains of child development (using a range of child development measures) and subsequent health.

Understanding whether and how child development and adolescent health outcomes are related presents opportunities for interventions to improve health and reduce health inequalities at an important time in the life-course, adolescence. There is evidence that health in adolescence is on the causal pathway to socio-economic status (SES) in adulthood by enabling "selection" into education [19]. Therefore, focusing on health in this period is critical to enable children to optimise their subsequent educational outcomes for wellbeing and employment opportunities. Informing interventions requires evidence not just on associations between child development and adolescent health but also on the effect of socio-economic circumstances on any associations found. In our protocol we outlined pathways by which socio-economically driven health inequalities may manifest (family stress, material living circumstances and parental health behaviours) and also possible direct pathways (social and cognitive) between child development and subsequent health. This provides a conceptual framework for the review. To inform interventions on any of these pathways there is a need to identify factors which may explain, and the socioeconomic circumstances which may modify, the associations between child development and adolescent health. This requires a public health lens and, as far as we are aware, no review has analysed the evidence on relationships between different dimensions of child development and adolescent health outcomes or assessed the factors which may shape the relationships.

In summary, there is evidence that aspects of child development at school starting age are associated with later academic success, but less is known about whether and how particular dimensions of child development influence health outcomes in adolescence. This gap in understanding limits cross-sector interventions to improve adolescent health and reduce health inequality. This review addresses this gap by undertaking a participatory systematic review to: (1) synthesise evidence on the relationship between child development at school starting age (3–7 years) and subsequent health in adolescence (8–15 years) and (2) identify factors that shape the relationship.

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

#### *2.1. Protocol Registration*

The study protocol was registered with PROSPERO (CRD42020210011) and published [20]. The review is reported according to the Preferred Reporting Items for Reporting Systematic Reviews and Meta-Analyses (PRISMA) 2020 Statement [21,22]; the checklist available in additional file 1. Any deviations from protocol are stated and explained in the relevant sections.

#### *2.2. Review Questions*


#### *2.3. Definition of Terms*

Child development is defined as a developmental process incorporating measures of development that record changes within a child's cognitive or physical development, or language and communication, or socio-emotional development.

#### *2.4. Study Design*

The design for this study was a participatory systematic review, involving engagement with national and local stakeholders across health and education sectors.

#### Stakeholder Engagement to Design the Conceptual Diagram

The lead reviewer held discussions with stakeholders to develop a conceptual model of the relationship under study. This process is described in full in the study protocol [20]. Their views, together with a scoping review of the evidence, led to an initial conceptual diagram (available in additional file 2). This diagram highlights the main pathways by which socio-economically driven health inequalities manifest; family stress, material living circumstances and parental health behaviours [23]; and also illustrates possible direct pathways (knowledge/literacy and social/cognitive) between child development and education and subsequent health. The diagram acted as a framework for the review, providing initial categories for extracting and analysing evidence from published studies.

#### *2.5. Eligibility*

Studies needed to include children, some or all of whom were aged between 3 and 15 years, in high-income country settings defined as a member country of the Organisation for Economic and Co-operation and Development (OECD). Exposures were characteristics of child development at school starting age (3–7 years), defined as: cognitive or physical development, or communication and language, or socio-emotional development. Primary outcomes were health and wellbeing outcomes, reported between the ages of 8 and 15 years: specifically, weight, mental health and proxy measures such as dietary habits and behaviour and measures of wellbeing. Secondary outcomes were academic outcomes of academic tests and proxy measures such as executive function during the outcome age of interest. Secondary outcomes were only included if they were found in a study with a primary outcome of interest. Executive function was included as a secondary outcome of interest because it allows for the regulation, control and management of learning, and thus appears an important link between child development and academic outcomes. In addition, executive function is a good predictor of academic achievement [24]. Studies that provided data on associations between the exposures and outcomes in the age period of interest, and additionally those that provided evidence on mechanisms, were required. The population and context, exposure, outcomes and study designs are described in full in the published protocol [20] and summarised in relation to inclusion and exclusion criteria in Table 1.





#### *2.6. Search Strategy*

We searched four electronic databases for articles published from November 1990 to November 2020: MEDLINE (OVID), PsycINFO (OVID), ASSIA (ProQuest) and ERIC (EBSCO). We also searched the reference lists from all included articles for additional eligible articles. Further relevant literature was identified through stakeholder discussions. Grey literature searching was undertaken by searching relevant organisations' websites. The search strategy was informed by a scoping review of the literature and focused on terms relating to child development, school readiness and adolescent health. The search strategy is available in additional file 3.

#### *2.7. Study Selection and Data Extraction*

Retrieved citations were uploaded to EndNote and duplicates removed. Titles and abstracts were screened by five reviewers against the inclusion and exclusion criteria. A 10% sample of papers was independently checked by two reviewers and inter-rater reliability was 86%. Any disagreements were resolved by discussion between the reviewers, so that a consensus was reached. The full texts of papers were read in the second stage of the screening process, by five reviewers, to produce a final list of papers for full text review. The final list of papers included was exported to excel to be assessed for the data extraction process. The lead reviewer extracted data for those articles that met the inclusion criteria in full. Reasons for exclusion were recorded and a list of excluded papers, together with the reason, is available in additional file 4. Data extraction was undertaken solely by the lead reviewer using a bespoke form (additional file 5), which had been trialled on a sample of different sources, and a sample of 10% was second checked. The following data were extracted: author and year, study design, analysis method, country and setting, participants, exposure measure and age, exposure measurement instrument, outcome measure and age, outcome measurement instrument, association and effect size, mechanism (studied and proposed), and factors which moderate the association, strengths and weaknesses.

#### *2.8. Quality Assessment*

Our protocol stipulated the use of Liverpool Quality Assessment Tool (LQAT) [26]. However, it was found that LQAT was insufficiently detailed for this review. Therefore, in a deviation from protocol we adapted a tool appropriate for the study designs used in previous systematic reviews [27,28]. The methodological quality of each observational study was assessed for risk of bias and clarity of study description to assign studies to one of three categories of methodological quality: high, moderate or low, using the template in additional file 6. Specifically, studies were assessed against 12 criteria within the following categories: study population, study attrition, data collection and data analysis with each pertaining to validity, precision or informativeness. In line with the recommendations of Cochrane [29], studies were not scored, and instead a narrative indication of quality (using +, − and ? against each criteria) was made based on all criteria, with criteria pertaining to validity and precision carrying a greater weight in guiding overall quality. Quality assessments were undertaken by the main author and a 10% sample independently assessed by a member of the review team. In all cases the overall assessments of quality made by the reviewers were consistent.

In addition to assessing the quality of each individual paper, the overall strength of evidence for papers grouped by outcome and domain was assessed, e.g., mental health outcomes and the socio-emotional domain of child development. Within these groupings the overall findings were graded as providing either: stronger evidence (generally consistent findings in higher quality studies); weaker evidence (generally consistent findings in one higher quality study, or in multiple lower quality studies); inconsistent evidence (inconsistent findings across multiple studies); or very limited evidence (a single study). This method draws on techniques used by Hoogendoom [27] and Baxter [30,31].

#### *2.9. Data Synthesis*

As per the protocol, we undertook a narrative synthesis using the SwIM guidelines [32] (additional file 7) to guide reporting. This was in anticipation of heterogeneity in the variety of exposures, analysis methods and outcomes in the studies. Each study was assessed and associations between exposures and outcomes recorded as either "positive", "negative" or "no association". Studies were grouped by outcome and, within this, organised by

exposure domain and tabulated to illustrate both the associations and assigned quality. The groupings for the outcome measure were undertaken by allocating the measure into either mental health, obesity or academic outcomes. The grouping for exposure measures was an inductive process involving an interpretation of the way child development had been understood and measured in each included paper, and then classifying and allocating these into a particular domain of child development; namely, a socio-emotional domain, cognitive domain, language and communication domain, physical domain or multiple domains. This was a subjective process because, as indicated in the introduction, there is no existing framework for understanding child development and characterising measures of child development.

An overall rating on the strength of the evidence for each grouping (studies allocated within each domain of child development for each outcome: weight, mental health, academic) was derived as described in the quality assessment section. The results for factors which mediate or moderate associations between child development and subsequent health in adolescence (review question 2) was synthesised in relation to the conceptual diagram (additional file 2) of the relationship (produced with stakeholders at the outset of the review). Factors were classed as either mediators (those that explain associations) or moderators (those that alter the strength of associations) and assigned to a pathway (grouping of factors): family stress, knowledge/literacy, social/cognitive, material living and parent health behaviours. The overall ratings on the strength of the evidence for each domain and outcome, and stakeholder discussions, were used to inform a final diagram of the relationship between child development and adolescent health.

#### **3. Results**

#### *3.1. Literature Results*

Following the screening of 10,657 retrieved citations, 34 articles were included in the review. See Figure 1 for PRISMA diagram illustrating the study selection process. Fifty-two studies were excluded on full text review; the list of studies excluded, with the reason, is available in additional file 4.

#### *3.2. Study Design and Setting*

Of the 34 included studies there were 32 prospective longitudinal studies [33–64], one retrospective longitudinal study [65] and one meta-analysis [66]. Detailed descriptions of the included studies are available in additional file 8. Of the 34 studies, 14 were set in the United States [35,36,39,43,45,47,49–51,56–59,64], seven in Canada [38,46,52–55,63], five in Australia [34,41,44,48,62], three in the UK [40,42,61], three in The Netherlands [33,37,65], one in Denmark [60] and one in which the countries included in the analysis were not explicitly stated [66].

#### *3.3. Sample Size and Participant Characteristcs*

The total number of children in included studies in the review was 69,152 (48% female, in those where sex was reported). Participants were recruited from pre-birth (through mother's pregnancy) to age 12 years, with the majority recruited between the ages of 4–6 years, at pre-school or kindergarten. Across the studies recruitment took place between 1986 and 2009. The majority of the children were enrolled in existing longitudinal studies, were mainly Caucasian and from a mix of socioeconomic backgrounds. Six studies focused on socioeconomic disadvantage; three were of children from socio-economically disadvantaged families recruited from child care centres [50] or Head Start programmes (early years services to support low-income children and families in the US) [39,58], two studies oversampled for greater socioeconomic risk [51,61] and one oversampled for nonmarital status [47]. A further two studies had children from majority low income [42] and low to middle income families [49]. There were three studies in which children from socioeconomic disadvantage were less well represented [34,38,52]. Children were assessed either in their own homes, pre-school or school apart from in two studies where lab-based

assessments were made [46,51] and two where routinely collected healthcare data was used [61,65].

#### *3.4. Studies Identified across Different Domains of Child Development (Exposures) and Adolescent Outcomes*

Studies were found that focused on all domains of child development, namely: socio-emotional development, cognitive development, language and communication, and physical development. Table 2 illustrates the number of studies within each domain and the related adolescent outcome measure(s). Table 3 provides a summary of the main study characteristics and describes the exposures by domain of child development, outcomes and how they were measured. The main domain of child development studied in included papers was socio-emotional development with 24 studies [33–35,37–39,42,44–48,50,53,54,56–60,62,64–66]. Exposures included behaviours such as internalizing and externalizing behaviours, social competence, emotion knowledge, emotional wellbeing, emotional reactivity and peer relations. Exposures within the socioemotional domain were generally measured using the relevant sections of standardized instruments such as the Child Behaviour Checklist (CBCL), the Social Behaviour Questionnaire (SBQ) or the Strengths and Difficulties Questionnaire (SDQ), with a mixture of child report, teacher report and parent report across the studies.

**Figure 1.** PRISMA flowchart of study selection process.

Four studies [40,51,52,63] had an aspect of cognition as the exposure of interest, namely: mathematics skills, executive control, foundational cognitive ability, verbal ability/literacy and Intelligence Quotient (IQ). Executive control refers to a set of cognitive processes necessary for cognitive control of behaviour and was measured by observing tasks. Verbal ability was measured using literacy tests, mathematics skills by number knowledge tests or standardized assessments relating to the relevant country's curriculum, and foundational cognitive ability and IQ by standardized instruments.

Two studies [36,43] had language and communication as the main exposures and a further study [52] included language as one of multiple exposures. Exposures included receptive and expressive vocabulary. These were measured using the relevant sections of standardised assessments such as the Peabody Picture Vocabulary Test.

Two studies [49,55] incorporated exposures in the physical domain of child development. Exposures included fundamental movement skills (balance, agility, handeye co-ordination) and participation in structured and unstructured physical activity. These were measured by either parent report or assessment of skills by assessors in the child's home.

Two studies [41,61] measured across all domains of child development and education. One study assessed the component parts of teacher-rated school readiness in relation to the country's early development instrument and one focused on child development in all domains in a health visitor check as a composite measure. In the main, studies analysed the effect of the exposure at a certain time point on an outcome at one later time point. However, two studies repeated measures at subsequent ages to assign children to a trajectory for the exposure of interest [38,58] and four studies repeated measures to study trends over time [43,48,57,64].


**Table 2.** Studies by child development domain and adolescent outcome.

\* Includes one study which measured several outcomes. ˆ From a study centrally coded to a different domain due to multiple exposures studied.

#### *3.5. Quality Assessment*

Thirty-three of the 34 included studies were assessed using the methodological assessment tool for observational studies, available in additional file 6. One study, a meta-analysis, was assessed using AMSTAR (A MeaSurement Tool to Assess systematic Reviews). Results of the quality assessment process for all included studies is available in additional file 9. Ten were rated as low, 16 moderate and eight high in methodological quality. High implies a low risk of bias, moderate implies a moderate risk of bias and low quality implies a high risk of bias.

As outlined in quality assessment section of the methods, confidence in cumulative evidence was assessed within each grouping of papers, grouped by outcome and domain. This is referred to throughout the synthesis of the findings.




**Table 3.** *Cont.*

23





Nelson et


Pedersen et




#### *3.6. Narrative Synthesis*

There was a range of exposures and outcomes reported across the included literature. Studies were organised by outcomes and grouped as follows:


Within these above groupings, studies were subsequently organised by exposure and by each domain of child development as follows:

	- -Internalising—internally focused behaviour such as inhibition and withdrawal.
	- - Externalising—externally focused behaviour such as aggression, attention problems, hyperactivity and "delinquent" behaviour.
	- - Emotional—internal factors such as social competence, emotion knowledge, pro-social, co-operative and self-regulation skills. External factors such as peer relations, parent-child relationships, teacher-child relationships, socioemotional climate of school/pre-school setting.
	- -Temperament—negative emotionality, emotional reactivity and persistence.

A summary of the evidence on associations between exposures (domains of child development) and outcomes is presented in Figure 2. Each annotation does not always represent a study in its entirety as many studies analysed multiple exposures and outcomes.

#### *3.7. Primary Outcomes*

#### 3.7.1. Mental Health

Summary of Associations between Child Development and Mental Health

Positive development on starting school is associated with subsequent positive mental health. There is stronger evidence for associations between the socio-emotional domain of child development and later mental health, weaker evidence for the cognitive domain, inconsistent evidence for language and communication and limited evidence for physical development.

#### Summary of Associations between Socio-Emotional Development and Mental Health

Eighteen studies analysed associations between a socio-emotional exposure of child development and later mental health [33,35,38,39,42,44–47,50,53,56,57,59,60,64–66]. All associations highlighted that positive socio-emotional development is good for subsequent mental health, apart from five studies where no associations were found for some exposures and outcomes studied [39,42,50,60,65]. The evidence is stronger for exposures of externalising behaviour and emotional wellbeing at school entry, weaker for exposures of internalising behaviour and limited for exposures relating to temperament.


**Figure 2.** Evidence of associations between domains of "child development" (exposures) and outcomes of mental health symptoms, weight and academic.

Exposure of Internalising Behaviours at School Entry and Subsequent Mental Health

Eight studies analysed the relationship between early internalising behaviour and later mental health [33,39,47,50,53,60,65,66], highlighting weaker evidence for positive associations with internalising outcomes and limited evidence for positive associations with externalising outcomes. Of these, six studies analysed the association between early internalising and later internalising behaviours, with two studies of moderate quality showing positive associations [33,47], one high quality study where no association was found [39], and one low quality study, of 56 children, where no association was found with depression symptoms [50]. Specifically, anxious-fearful behaviour is associated with later emotional difficulties as reported by parents [60] in a study of moderate quality. Behavioural inhibition is associated with anxiety but this evidence was from a lower quality review [66]. Evidence on the relationship between early internalising and later externalising behaviours was scant; only limited evidence was provided, with two studies not studying that relationship specifically [47,53] and one study where no association was found between early emotional and behaviour problems and later externalising [65].

Exposure of Externalising Behaviours at School Entry and Subsequent Mental Health

Nine studies were found on the relationship between externalising behaviours and later mental health [33,38,39,45,47,50,53,60,65], highlighting stronger evidence for positive associations with both internalising and externalising outcomes. Of these, seven studies analysed the associations between early externalising and later internalising, with six studies showing positive associations and one study where no association was found. There was evidence of positive associations between general externalising behaviour problems [33,39] and later internalising symptoms, and specifically aggression [60] was associated with later internalising symptoms. However, although two studies showed no association between hyperactive behaviour [60] or inattention [65] and later general internalising symptoms, one high quality study of 2000 children did evidence an association between these behaviours and later anxiety symptoms [38]. One study, of lower quality, evidenced that disruptiveness was associated with later depression symptoms and loneliness [53].

Similar to internalising symptoms, whereby the continuity of association was found for early and later symptoms, the same is true for externalising symptoms, whereby early problems are associated with externalising at a later age. However, the evidence is stronger with two studies of moderate quality evidencing associations between general externalising [47], inattention and behaviour problems [65] and later general externalising symptoms, with a further study evidencing a relationship between poorer social competence with peers in mid-childhood and earlier behaviour problems [45]. Specifically, disruptiveness was associated with delinquency in one low quality study [53].

Exposures of Emotional Wellbeing at School Entry and Subsequent Mental Health

Nine studies were found on the associations between a child's emotional wellbeing and later mental health [35,38,39,42,44–46,57,64] with stronger evidence found for the association with internalising outcomes and weaker evidence for externalising outcomes. A child's emotional wellbeing, in terms of social competence, emotional knowledge (the ability to identify and label emotions), self-regulation and prosociality (behaviour intended to benefit others), appears beneficial to later health in adolescence. Negative associations were found between early social competence and internalising and externalising problems [35]. In two high quality studies, associations were found for emotional knowledge [39] and prosocial skills [38] and later anxiety, with increasing emotional knowledge and prosocial skills both associated with less anxiety symptoms. A child's ability to co-operate, a particular prosocial skill, highlighted mixed results in one study [42], with increasing co-operation associated with less externalising but no association found with internalising problems. Self-regulation problems, in terms of ability to control behaviours, attention, thinking, social interaction and emotions, were subsequently associated, in adolescence, with an increase in the risk of self-harm ideation and behaviour, suicidal ideation, school truancy,

mental health problems, smoking and alcohol use, and violent and property crime [44]. When self-regulation problems reduced, from age 4–5 to 6–7, the association between these adolescent outcomes and earlier self-regulation problems was no longer found [44].

In relation to the child's emotional wellbeing in the context of relationships or setting specific (external factors), studies were found on: mother–child attachment, relationship with parents, teachers, and peers (victimisation), and the socio-emotional climate in a pre-school setting, and all proved important for positive mental health in adolescence. A small study of 68 children, rated low quality, evidenced that disorganised maternal attachment at pre-school age was associated with greater depression and anxiety symptoms and lower self-esteem in early adolescence [46]. A positive relationship with parents, in terms of closeness, was associated with less loneliness, particularly for father–daughter relationships [64]. In one low quality study, a good quality relationship with teachers and a positive socio-emotional climate in a pre-school setting were both associated with improved social competence in mid-childhood [45]. With regard to relations with peers, one study evidenced that early and increasing peer victimisation was associated with depression symptoms and aggression [57].

Exposures of Temperament at School Entry and Subsequent Mental Health

Two studies were found on the association between temperament and later mental health highlighting limited evidence of a negative association, with higher levels of certain traits associated with worse outcomes. These studies investigated child temperament, in terms of negative emotionality and emotional reactivity (the former refers to the propensity to react with negative emotions and the latter relates to the intensity of emotion) [56,59] and both were of moderate quality. One showed an association between negative emotionality, emotional reactivity and depression symptoms [56] and one between emotional reactivity and internalising and externalising symptoms [59].

Summary of Associations between Language and Communication, Cognitive Development, Physical Development, and Multiple Domains and Mental Health

Eight studies analysed the associations between exposures relating to either language and communication, cognitive development, physical development or multiple domains of child development and later mental health. All associations highlighted that positive development across all of the domains of child development are good for subsequent mental health. There was weaker evidence for the effect of cognitive skills and the positive effect of cognitive development appears to alter with age. The evidence for associations between language and communication and later mental health outcomes was inconsistent in relation to internalising and externalising outcomes. There was limited evidence for both physical development and measures incorporating multiple domains.

Exposures within the Language and Communication Domain and Subsequent Mental Health

The results for the effect of language and communication skills on later mental health symptoms was inconsistent with two studies investigating these associations [36,43]. One study of 129 children evidenced that language skills at pre-school age predict internalising but not externalising behaviour problems in adolescence. Conversely, one low quality study of 74 children did find an association between good language skills (receptive and expressive language) and less externalising problems, namely, conduct problems but not hyperactivity.

Exposures within the Cognitive Domain and Subsequent Mental Health

Three studies analysed the effect of cognitive skills on later mental health symptoms, [40,51,63] with weaker evidence found. One study found that deficits in executive control predicted depression and anxiety symptoms and clinical level of depression [51]. The same study showed that foundational cognitive ability did not predict these outcomes. One high quality study showed an association between cognition, measured as IQ, and depression symptoms with an increased IQ in early childhood associated with less depression symptoms at age 11 [40]. However, by age 13–14 the association reversed. The loss of protective effect of cognition was also found in relation to the effect of cognitive skill (measured as mathematics skills and verbal ability) on internalising symptoms, whereby a protective effect seen at age 12-13 was reversed or had no associated effect at age 14–15 [63].

Exposures within Physical Development Domain and Subsequent Mental Health

There was limited evidence for the effect of physical development on later mental health related symptoms. One lower quality study, in which time between exposure and outcome was one year, found that structured physical activity was associated with less depression symptoms in boys, whereas unstructured physical activity was associated with more depression in girls [55].

#### Exposures Incorporating Multiple Domains and Subsequent Mental Health

Two studies provided evidence across multiple domains [41,61]. One study evidenced that all components of school readiness (as part of a model of early years data), measured by UK health visitors before starting school, predicted socio-emotional behaviour problems in early adolescence [61]. An Australian study which investigated the relationships between all domains of school readiness and wellbeing at the end of primary school found that all domains were negatively associated with internalising symptoms, whereas only physical and socio-emotional development were positively associated with overall wellbeing [41].

#### 3.7.2. Weight, Diet and Physical Activity Outcomes

#### Summary of Associations between Child Development and Weight

Positive development on starting school is associated with subsequent healthy weight related outcomes. There is stronger evidence for the socio-emotional domain of child development, and limited evidence for language and communication, cognitive and physical domains of child development.

#### Summary of Associations between Socio-Emotional Development and Weight

Five studies analysed the associations between a socio-emotional measure of child development and later weight diet or physical activity outcomes [37,44,48,54,62]. All associations highlighted that positive socio-emotional development is good for subsequent weight-related outcomes, apart from one study where mixed associations were found for exposures of certain temperamental traits and later weight related outcomes. The evidence is stronger for exposures within the emotional wellbeing domain, specifically self-regulation skills, with weaker evidence found for exposures of externalising behaviour and no evidence found for internalising behaviour.

Exposures of Externalising, Emotional Wellbeing and Temperament and Subsequent Weight

In relation to externalising, specifically aggressive behaviour, one higher quality study found a positive association with higher BMI [37] and one of moderate quality found an association with higher rate of change in BMI but in girls only [48]. In relation to selfregulation, one higher quality study [44] evidenced that early problems in self-regulation (ability to control attention, behaviour and emotion at age 4–5) were associated with being overweight or obese in adolescence but that a change in self-regulation (less problems) at a later age (age 6–7) had no effect on the association. Another study highlighted that increasing self-regulation skills (measured as class room engagement) were associated with lower BMI and increased sports participation [54]. Additionally, this study evidenced that emotional distress (a measure of self-regulation) was associated with less sports participation. In relation to temperament, one higher quality study [62] looked at the associations between the traits of persistence sociability and negative reactivity, and later BMI and weight attitudes and behaviours, and found that persistence decreased the risk of obesity and overweight, sociability increased the risk of overweight but not obesity, and negative reactivity was not associated with either. In relation to weight attitudes and behaviours, all three traits were associated with restrained eating habits in adolescence, with lower persistence and higher negative reactivity or sociability associated with restrained eating and use of unhealthy weight management strategies.

Summary of Associations between Domains of: Language and Communication, Cognitive, Physical Development, Multiple Domains and Weight

There was limited evidence on associations between the domains of language and communication, cognitive skills, and physical development, and later weight-related outcomes. One study, which looked at evidence on a range of school readiness skills and later wellbeing measures, evidenced that receptive vocabulary was associated with healthier dietary habits [52], with increasing receptive vocabulary predicting reduced sweet snack intake and increased dairy intake. The same study [52] evidenced that increasing mathematics skills predicted increasing involvement in physical activity, providing limited evidence for an association between cognitive skills and later weight-related outcomes. There was limited evidence on the association between physical development and weight-related outcomes, with one lower quality study finding no association between fundamental movement skills and later involvement in physical activity [49]. One study evidenced that all components of school readiness (as part of a model of early years data), as measured by UK health visitors before starting school, predicted overweight and obesity in early adolescence [61].

#### *3.8. Secondary Outcomes*

#### 3.8.1. Academic Tests and Executive Function

Summary of Associations between Child Development and Academic Outcomes

Four studies analysed associations between a domain of child development and later academic outcomes: three in relation to socio-emotional development and one in relation to cognitive development. All associations highlighted that positive development is good for subsequent academic outcomes. The evidence is stronger for exposures within the socio-emotional domain, specifically self-regulation skills and less behaviour problems, with weaker evidence found for exposures within the cognitive domain of child development. There were no studies found looking at the association between language and communication or physical development and academic outcomes.

Exposures within the Socio-Emotional and Cognition Domains and Subsequent Academic Outcomes

There were two studies found on associations between socio-emotional development and the secondary academic outcomes, both of higher quality [34,44]. One studied the effect of behaviour risk (a composite of poor sleep, emotions and inattention) on adolescent executive function and found that poorer behaviour is associated with lower executive function [34]. Another study highlighted that self-regulation problems are associated with reduced scores on numeracy and literacy tests in adolescence [44]. An additional study [58] investigated the effect of an intervention targeting socio-emotional functioning and language emergent literacy skills in pre-school, through comparing the impact of executive function trajectories on academic test results of children in the intervention group compared to those who were not. This study showed that socio-emotional and language programmes improved executive function and academic outcomes for children with the lowest executive function trajectory. There was limited evidence on associations between early cognitive skills and later academic outcomes, with one study showing a positive association between kindergarten mathematics skills and later academic outcomes [52].

#### *3.9. Factors Affecting Relationships*

Summary of findings on factors affecting associations (mediation and moderation) Limited evidence was found on factors affecting associations. Some evidence, however, was found on factors affecting associations between socio-emotional development and subsequent mental health and academic outcomes. The factors are discussed in relation to the pathways identified in the initial conceptual model devised with stakeholders. Factors were found in relation to family stress, knowledge/literacy and social/cognitive pathways. No factors were found that pertained to the material living or parent health behaviour pathways. All of the findings within this section fall into the category of limited evidence as all of the factors described were found in single studies only.

#### 3.9.1. Mediators

Six studies included data on mediating variables: five related to studies focusing on mental health outcomes [35,36,46,53,56] and one relating to academic outcomes [34]. None of the studies that focused on weight as an outcome included data on mediation. Factors mediating associations between socio-emotional development and mental health were self-esteem, type of internalising or externalising in mid-childhood, and relationships with teachers and friends. Factors mediating associations between socio-emotional development and academic outcomes were approaches to learning and attentional regulation.

#### 3.9.2. Moderators

Seven studies included data on variables to test for moderation effects on associations between exposure and outcome: five in relation to mental health [38,40,47,59,64], one in relation to weight [62] and one in relation to academic outcomes [58]. Factors moderating associations between socio-emotional development and mental health were household chaos and parenting. Household chaos had a negative effect, and aspects of parenting had a positive effect on the associations between the socio-emotional domain of child development and mental health outcomes. A factor found to moderate the association between socio-emotional development and academic outcomes was trajectory of executive function.

3.9.3. Factors Pertaining to the Family Stress Pathway—Moderating Associations between Child Development and Mental Health

This pathway incorporates factors related to stress in the home, which can affect parenting ability, parenting style and consequently child health and development. Household chaos and aspects of parenting were identified as moderators of the relationship between the socio-emotional domain of child development and later mental health symptoms. Household chaos was found to disproportionately affect children with higher emotional reactivity resulting in greater internalising problems [59]. This effect was not found for household income; that is, level of emotional reactivity made no difference to the impact of income on adolescent emotional and behaviour problems. This implies that the impact of low income on adolescent mental health is pervasive and not amenable to individual interventions promoting self-regulation (in terms of emotional response to events) but that interventions of this type might support how children respond to household chaos. Three studies analysed the moderating role of aspects of parenting on the relationships between socio-emotional measures and later mental health related symptoms [38,47,64], with all finding positive effects of aspects of child/parent relationships on adolescent outcomes. Two studies found a protective effect of relationships with fathers on continuity of behaviour problems. One found a protective effect for fathers' positive engagement on the continuity of earlier to later internalising and externalising behaviour problems, and for those in the greatest poverty, fathers' positive engagement was associated with a reduction in the continuity of internalising problems from age 5 to 9 years [47]. The authors hypothesize that this is due to development of secure attachment and the development of emotional and behavioural regulation skills. Another looked at the moderating role of parent–child closeness on the continuity of loneliness from age 6–11 years and found that as parent–child closeness increased, loneliness reduced and this relationship was particularly strong for girls and their fathers [64]. Another study looked at the moderating role of maternal parenting practices, warmth and discipline, on the relationship between behavioural characteristics of; inattention, hyperactivity, aggressiveness and low prosociality and trajectory of anxiety in children between the ages of 6 and 12 years [38]. It found that a lack of maternal warmth increased the association between hyperactivity and anxiety. It also found that high level of maternal discipline (rules and controlling child's behaviour) increased the probability of belonging to the high anxiety group.

3.9.4. Factors Pertaining to the Knowledge/Literacy Pathway—Moderating and Mediating Associations between Child Development and Academic Outcomes

This pathway relates to the way knowledge and literacy can lead to behaviours that can be positive for wellbeing [67]. Two studies analysed factors within this category, both on the relationship between the socio-emotional domain and later academic outcomes. One highlighted a moderating role of executive function on the relationship between an intervention to improve socio-emotional and language emergent skills and later academic outcomes, and found that the effect of the intervention was higher in children with low executive function in the intervention group resulting in better academic outcomes compared to controls [58]. Another highlighted attentional regulation and approaches to learning as mediators, partially explaining the relationship between child behaviour (composite of sleep, emotional dysregulation and inattention/hyperactivity) and later executive functioning.

3.9.5. Factors Pertaining to the Social/Cognitive Pathway—Mediating Associations between Child Development and Mental Health

This pathway relates to the influence of individual experiences, the actions of others and environmental factors that provide the social context for learning to influence health behaviours [68]. Five studies analysed mediators pertaining to this pathway and all in relation to mental health related symptoms: four in relation to the socio-emotional domain [35,46,53,56] and one from the language and communication domain [36]. In relation to the socio-emotional domain of child development and later mental health there

is limited evidence that self-esteem, type of internalising or externalising in mid-childhood, and relationships with teachers and friends all play a role in explaining the relationship. One study evidenced the role of self-esteem, which partially mediated the relationship between emotional wellbeing (measured as parental attachment) and depression but not anxiety [46]. Two studies highlighted the role of relationships: one with peers [53] and one with teachers [56]. In relation to peers, peer rejection and number of friends in midchildhood mediated the relationship between disruptiveness at age 6 and depression at age 13 years, with peer rejection also mediating the relationship between disruptiveness and loneliness. Relationships with teachers (closeness and conflict) also appear important, with one study analysing the effect of child–teacher relationships on the relationship between negative emotionality at age 4, emotional reactivity at age 7 and depression symptoms at age 11–12 years [56]. This study found that teacher-child conflict mediated the relationship between emotional reactivity and depression symptoms, with children higher in emotional reactivity having more depression symptoms, and this was partially explained by conflict with teachers (teacher reported).

3.9.6. Factors Pertaining to Child Characteristics—Moderating Associations between Child Development and Weight and Mental Health

Sex moderated the association between socio-emotional domain of child development and weight outcomes, with worse outcomes for girls. Age moderated the association between the cognitive domain of child development and subsequent mental health, with a protective role of positive cognitive development on mental health in early adolescence reversing in mid-adolescence.

Two studies analysed the effect of age on the relationship between cognition and later depression symptoms, and found that age reversed the protective effect of cognition on early adolescent (age 11) mental health by age 13–15 years [40,63] but that this reversed again at age 17 for females [40]. This study also found that pubertal status mimicked the relationship by age but that this was stronger for females. The loss of the health protective effect of cognition may be due to exam pressures at certain time points or, for females, biological hormonal changes. Sex was the only factor studied in relation to weight. One study evidenced that girls with higher aggression cores throughout childhood had a higher rate of change in their BMI [48]. Another study highlighted that girls higher in sociability in early childhood had a greater fear of weight gain at age 14–15 years [62].

#### *3.10. Conceptual Model/Diagram Development*

A summary of findings and revised conceptual model was discussed with stakeholders and a final diagram of the relationships, as informed by this systematic review, was discussed and agreed—see Figure 3. Factors which were not found in the review but were deemed important by stakeholders are highlighted on the diagram. These included neighbourhood factors, such as community engagement and community environment, which stakeholders felt could create conditions conducive for optimal health and development. Political and system factors were also identified, such as short political cycles not giving policy sufficient time to embed and effect change, and regulators focusing narrowly on academic outcomes rather than broader social and emotional wellbeing, which dictates the focus of a school. The stakeholder group identified these factors as potential moderators of the relationships between child development and health.

#### **4. Discussion**

This review asked the questions: what are the associations between child development and adolescent health, and what factors explain or alter the associations. The review clearly shows that positive development on starting school is good for later health outcomes, but that the evidence is stronger for relationships between some domains of child development than others, with gaps in the evidence base across domains. In relation to mental health outcomes, there is stronger evidence for associations between socio-emotional development and later mental health, weaker evidence for associations with cognitive development, inconsistent evidence for language and communication, and limited evidence for physical development. In relation to adolescent weight, there is stronger evidence for associations with children's socio-emotional development and limited evidence of relationships with language and communication, cognitive development or physical development. In relation to secondary (academic) outcomes, there is stronger evidence for associations with socioemotional development, limited evidence for an association with cognitive development and no evidence found for an association with language and communication or physical development. In relation to what factors explain or moderate the associations, the evidence identified in this review is largely limited to factors shaping the relationship between socioemotional development and mental health and academic outcomes with factors pertaining to the pathways of family stress, knowledge/literacy and social/cognitive.

Our findings build upon the existing limited evidence that attendance at pre-school (a proxy for good child development) is associated with positive mental wellbeing and healthy weight in adolescence [12], and provides detail on which domains of child development are associated with these positive health outcomes. Supporting the existing literature, we found positive relationships between both cognitive and socio-emotional development (such as self-regulation and social competence) on mental health and academic outcomes in early adolescence [9–11], with stronger evidence found for socio-emotional development. Additionally, we identified evidence of a negative relationship between socio-emotional wellbeing and unhealthy weight. The review provided a test of our conceptual model and we found that the evidence base was lacking for some of the proposed pathways between child development and later health.

Undertaking this review highlighted a complexity in classifying the rich and broad literature that exists on child development. In consequence, the review had to embed an inductive process of interpretation of how included studies had understood and measured child development. Using a classification system of four domains of child development to aid analysis of a very broad concept enabled us to categorise a multitude of measures of child development within this system. This complexity and the need for interpretation is perhaps unsurprising given that the field of child development spans the disciplines of psychology, sociology education, biology, genetics and public health. That a classification system for understanding child development does not exist is in itself a finding of our review. Our classification of four domains of child development adds to the literature, providing a framework for other researchers to use and to critique.

Our findings show that conceptualising child development into domains of development matters because different aspects of development seem to have different impacts on later health outcomes. Understanding this can help to inform public health interventions in childhood. For example, in our review, we found that socio-emotional development when children start school has the most evidence for subsequent impact on adolescent health, in terms of mental health and healthy weight, and as such could be a focus for intervention. The findings in relation to mental health are to be expected, with much literature highlighting the continuity of early problems with socio-emotional functioning and later onset of mental health conditions [69]. The evidence is stronger for early externalising behaviours and their impact on both internalising and externalising behaviours in adolescence, and this is supported by wider literature [70,71]. A finding of this review is that there is more evidence that behaviours such as aggression and hyperactivity pose a risk to future mental health than anxious/withdrawn behaviour, particularly for externalising

outcomes. This finding should be interpreted cautiously because it may be that early internalising behaviours, compared to externalising, are more likely to resolve by early adolescence [72], or it may be that internalising is harder to identify, whereas externalising behaviours are more obvious and easier for parents and teachers to report, which could lead to less associations being observed for internalising behaviours and consequently less associations found [39,70]. The finding that emotional wellbeing was more closely associated with later internalising may be because emotional stability promotes regulation and mood stability leading to less internalising [73], but other studies have found that emotional wellbeing (in terms of regulation skills) is associated quite strongly with both internalising and externalising, but particularly so for internalising after the early years [74].

The findings in relation to stronger evidence on the associations between socioemotional development and weight add to a growing field of evidence exploring this relationship, with evidence of co-development and temporal associations in mid-childhood [75], evidence of obesity having a detrimental impact on socio-emotional behaviour [76], and evidence on associations between social competence and weight with social competence reducing the odds of later overweight [77]. From this review, emotional wellbeing and, in particular, self-regulation skills, appear to be important factors to study in this complex relationship between socio-emotional development and weight. However, other developmental pathways in the development of obesity, such as physical activity and cultural and social factors, are important to consider alongside self-regulation development [44].

More evidence is needed on how adolescent health outcomes are shaped by other domains of child development, particularly the impact of language and communication, and cognitive and physical development at school starting age. Evidence on these domains are important for engaging health and education sectors to work together because education and health services share a common goal for optimal developmental potential of children [78]. This evidence would help the development of a shared understanding and provide a platform from which to develop the context and settings that may work best for optimal health and development of children, regardless of their stage of development when starting school. Including executive function as an exposure and outcome in this review allowed for inclusion of any evidence on the bi-directional relationship between executive function and health [79]. The analysis of secondary outcomes of academic tests and executive function highlight the importance of socio-emotional development on these outcomes (health improves executive function). Conversely, the protective effect of cognitive skills (measured as executive control) on adolescent mental health highlights that executive function improves health. However, age appears to be an important factor in this latter relationship, with the protective effects of cognition on mental health being reversed or no associations found in mid-adolescence [40,63], and this warrants further research.

Understanding the impact of domains of child development on later health has important policy implications in relation to reducing inequalities, and in relation to a policy extension beyond the first 1000 days. In relation to reducing inequalities, our review highlights a strong relationship between socio-emotional development and later health. Applying a public health lens to "child development" helps to understand exactly what it is that pre-school provision or early years centres may need to focus on if we are to improve adolescent health and wellbeing. Re-invigoration of early childhood programmes such as Sure Start, with a renewed focus on socio-emotional development, may be one area of policy improvement, particularly if we are to focus on their longer-term potential to reduce inequalities [80].

In addition, arguably a policy shift is required, which extends beyond the first 1000 days, to understand and support optimal development throughout childhood and into adolescence [81], to address the consequences of inequalities in child development as children age and because adolescence is a significant period of development and an important period in the life course [82]. If we are to maximise the opportunities conferred by education as a platform to improve public health and reduce health inequalities [83], policy that incorporates a life course approach to healthy development is needed, and this requires cross-sector collaboration.

Fostering collaboration to inform policy on reducing child and adolescent health inequalities beyond the first 1000 days requires more research on how development and education translates into health throughout childhood, and on the effect of socio-economic circumstances on this relationship. Findings in relation to the factors that explain or alter associations between child development and subsequent health were limited in this review, with all findings pertaining to single studies. The most evidence was within the social-cognitive pathway with self-esteem, relations with peers and teachers all providing some explanation for the relationship between socio-emotional development and subsequent mental health outcomes, and this can inform interventions for optimal health and development through the primary school years. Surprisingly, in relation to the original conceptual diagram designed with stakeholders, there was no literature found on material living circumstances, parent health behaviours, community factors, and political and system factors. To some extent this was because the studies controlled for the effect of income, housing, parental education and parent health behaviours. However, this presents a problem because we need to know more about how these elements of socio-economic circumstance affect the associations under study in this review. For example, we know that children from more deprived backgrounds experience poorer health and development than their more affluent peers [7]. If we are to pragmatically intervene to improve the health and development trajectory of children in more deprived circumstances, and reduce the attainment and health gap, we need to understand exactly how poverty, household circumstances and home environments affect learning and the co-development of education and health. This requires the design of public health research that respects agency but more clearly theorizes children within their social and economic context [84], so as to encapsulate socio-political cultural and familial environments because, in many ways, this is what "defines" child development in practice, over and above genetic make-up.

In addition to collaborating to produce more evidence on individual, home or schoollevel interventions to mitigate against poor development, interventions at system level are required to tackle prevention earlier so that children reach a good stage of development, and to reduce inequalities in development measures upon starting school. This requires evidence in relation to macro-determinants such as political and system factors, e.g., addressing poverty, the role of regulators in generating a target-driven culture that focuses attention on academic achievement, a political system not conducive to cross-departmental perspectives or action, and a system which stifles innovation and creativity. Larger studies, such as natural experiments or evaluations of existing policies, are needed that perhaps compare areas with different working systems or policies to identify any particular cultures or practices that are conducive to promoting positive trajectories for the health and development of children. The participatory element of this research identified some of these macro-determinants, identifying a gap between research and practice. This finding highlights the need to bring research and practice closer together [85] through listening to views and experiences of those working in service roles, at the system level and/or with children. It is hoped that this method can help to inform future research by highlighting what is evidenced in the literature, but also bringing to light views from lived experience, which may not be captured in any published evidence but could steer future research.

The strengths of this review are in its systematic design aiming to incorporate all relevant studies to answer the research question and in its engagement with a stakeholder group to steer the review and engage research with practice. The involvement of a sample of stakeholders raises the potential for biases to be introduced by selection of stakeholders with particular views, opinions or experiences. The use of replicable and transparent systematic review methods helps to minimise this risk. The research question was broad. This limited the search strategy in incorporating all possible terms to address the breadth of the research question and this may mean that some evidence was not found. Another limitation is that the secondary outcomes were only included where they were found in

papers that also encompassed the primary outcome. This likely means that the associations found are underestimated. Including any paper with the secondary outcomes would have led to an unmanageable number of papers and, given that the focus of the review was health outcomes, it steered the decision on only including secondary outcomes where relevant in understanding any temporal dynamics to the relationship under study. The grouping of child development measures, used for data synthesis, could be seen as both a strength and a weakness: a strength in that it allowed for the classification of a range of child development measures into developmental domains, and a limitation in that it was a subjective process and, as such, is open to critique.

#### **5. Conclusions**

Positive socio-emotional development at school starting age appears particularly important for subsequent mental health and weight in adolescence. There are gaps in the evidence about what factors affect the relationships between child development and subsequent health, in particular, the effect of socio-economic factors. More collaborative research across health and education is needed to develop and define appropriate measures of child development across key domains of child development, and also on the relationships and mechanisms between domains of development, particularly cognitive, language and communication, and physical development, and later health, in the context of socio-economic inequality. This requires the design of public health research that respects agency but more clearly theorizes children within their social and economic context [84], so as to encapsulate socio-political, cultural and familial environments. Research designed using longitudinal cohorts could be one way forward here and be considered in future work on this topic. This theoretically informed research and knowledge is imperative to inform interventions to address health inequalities in mid-childhood and adolescence.

**Supplementary Materials:** The following are available online at https://www.mdpi.com/article/ 10.3390/ijerph182111613/s1, Additional file 1: PRISMA Checklist, Additional file 2: Conceptual diagram, Additional file 3: Search Strategy, Additional file 4: Excluded papers, Additional file 5: Data extraction form, Additional file 6: Quality assessment form, Additional file 7: SWIM Checklist, Additional file 8: Table of study characteristics, Additional file 9: Results of quality assessment process.

**Author Contributions:** M.B. identified the topic and led the study design process with contributions from A.B. (Amy Barnes), M.S. and D.T.-R.; M.B., A.B. (Anna Brook), A.R., B.H. and C.F. screened the search results; M.B. extracted data and undertook the quality assessments, with input from A.B. (Amy Barnes), A.B. (Anna Brook), B.H. and C.F.; M.B. drafted the manuscript with input from all authors. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the National Institute for Health Research (NIHR Doctoral Fellowship, Mrs Michelle Black, NIHR300689). The views expressed are those of the authors and not necessarily those of the NHS, the National Institute for Health Research or the Department of Health and Social Care. DTR is funded by the MRC on a Clinician Scientist Fellowship (MR/P008577/1).

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Lists of included and excluded articles are available in the published article and the associated supplementary files.

**Acknowledgments:** We would like to thank Susan Baxter, Andrew Booth and Mark Clowes for their helpful comments on the methods for the review. We are also very grateful to the stakeholders who contributed to the discussion in designing the conceptual model.

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

#### **References**


### *Review* **Outcomes and Critical Factors for Successful Implementation of Organizational Health Literacy Interventions: A Scoping Review**

**Marise S. Kaper 1,2,\*, Jane Sixsmith 3, Sijmen A. Reijneveld <sup>1</sup> and Andrea F. de Winter <sup>1</sup>**


**Abstract:** Organizational health literacy (OHL)-interventions can reduce inequality and demands in health care encountered by patients. However, an overview of their impact and critical factors for organization-wide implementation is lacking. The aim of this scoping review is to summarize the evidence on: (1) the outcomes of OHL-interventions at patient, professional and organizational levels; and (2) the factors and strategies that affect implementation and outcomes of OHL-interventions. We reviewed empirical studies following the five-stage framework of Arksey and O'Malley. The databases Scopus, PubMed, PsychInfo and CINAHL were searched from 1 January 2010 to 31 December 2019, focusing on OHL-interventions using terms related to "health literacy", "health care organization" and "intervention characteristics". After a full-text review, we selected 24 descriptive stu-dies. Of these, 23 studies reported health literacy problems in relation to OHL-assessment tools. Nine out of thirteen studies reported that the use of interventions resulted in positive changes on OHL-domains regarding comprehensible communication, professionals' competencies and practices, and strategic organizational changes. Organization-wide OHL-interventions resulted in some improvement of patient outcomes but evidence was scarce. Critical factors for organization-wide implementation of OHL-interventions were leadership support, top-down and bottom-up approaches, a change champion, and staff commitment. Organization-wide interventions lead to more positive change on OHL-domains, but evidence regarding OHL-outcomes needs strengthening.

**Keywords:** health literacy; organization and administration; health care settings; organizational innovation; culture; program development

#### **1. Introduction**

Almost one in every two people in Europe encounter problems handling health issues because of limited health literacy skills [1]. These problems are more prominent among people of a higher age and lower educational level [1]. Health literacy is defined as 'the degree to which people are able to access, understand, appraise and communicate information to engage with the demands of different health contexts' [2]. As Rudd et al. consistently point out [3–5], a health literacy gap is emerging between the abilities of patients and the demands placed by increasingly complex health services. This gap can contribute to a range of negative consequences for people with limited health literacy [1,6], who find it difficult to access and navigate health care organizations, communicate with health professionals, understand information, and engage in decision making and self-management [3,6–10]. These consequences can have a profound impact on patients, affecting their safety, quality of care, and health outcomes [1,6]. In order to reduce and prevent these problems, it has been recommended to reduce the complex demands in health care organizations [5,11–13].

**Citation:** Kaper, M.S.; Sixsmith, J.; Reijneveld, S.A.; de Winter, A.F. Outcomes and Critical Factors for Successful Implementation of Organizational Health Literacy Interventions: A Scoping Review. *Int. J. Environ. Res. Public Health* **2021**, *18*, 11906. https://doi.org/10.3390/ ijerph182211906

Academic Editors: Jessica Sheringham and Sarah Sowden

Received: 3 October 2021 Accepted: 10 November 2021 Published: 12 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/).

Health care organizations can reduce these demands and "make it easier for people to navigate, understand, and use information and services to take care of their health" [13,14], which is the definition of the concept of organizational health literacy (OHL). Contextualising health literacy to health care organizations involves the design of accessible and easy to use health services, including health promotion and ill health prevention, fostering equality and a responsive health system, supporting people to navigate that system, and engaging them in making informed health related decisions [15–17]. Palumbo [15] conducted a literature review with a preventive medicine orientation, and distinguished five themes from the identified OHL literature: (1) understanding OHL as a preventive health policy issue and promoting the integration of health literacy into organizations po-licies and daily activities; (2) contextualizing OHL in a patient-centred care perspective, building on a combination of formal (top-down) and informal (bottom-up) approaches to improve the accessibility of health services and engagement of patients; (3) raising awareness and strengthening commitment for achieving OHL, (4) preparing a health literate workforce using tailored training and capacity building, and (5) measuring efforts and outcomes related to OHL by using a systematic approach. Measurement should focus on the ability of health organizations to engage patients in a co-creation relationship, the quality of communication, supportive services and technologies.

Reducing the organizational demands for people with limited health literacy requires a combination of approaches targeted at the level of patients, professionals and the organization [13,18], denoted as organizational health literacy (OHL)-interventions. At the patient level, interventions can improve oral, written, and digital communication, and accessibility of services and physical navigation, as well as involve patients more actively in improving health information and services. At the professional level, OHL-interventions can improve capacity building and promotion of health literacy friendly communication practices. OHL improvement at the organizational level involves domains such as leadership and culture, organizational policies, systems processes, and structures. Over the last decade, a number of such OHL-interventions have been developed [19,20]. These interventions usually involve two phases: (1) assessment of health literacy problems from the perspectives of patients, professionals and independent observers; and (2) planning and application of interventions aimed at reducing demands in healthcare organizations.

Two reviews concluded that evidence on the planning, application, and outcomes of OHL-interventions was limited [19,20]. Until recently, these interventions focused mostly on the assessment of health literacy problems at the patient and professional levels, including physical navigation, and written-, digital-, and spoken communication, but with limited attention to an organization-wide approach [19,20]. The available studies of applied interventions reported a number of facilitators and barriers that influenced OHLinterventions, such as lack of health literacy awareness, staff commitment, and leadership support [19,20]. The evidence on outcomes indicated that implementation periods were brief and improvement of OHL-outcomes limited.

Since the publication of these reviews, new insight has been gained regarding outcomes and implementation of OHL-interventions, and on how organizational transformation may improve patient outcomes. Current research on health literate organizations focuses more on facilitating sustainable transformation to improve OHL outcomes at patient, professional and organizational levels [14]. This scoping review summarizes the evidence regarding: (1) outcomes of OHL-interventions at patient, professional and organizational levels; and (2) factors and strategies that influence implementation and outcomes of these interventions.

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

To guide this scoping review we used the five-stage framework for scoping reviews developed by Arksey and O'Malley (2005) [21]. The five stages are: (1) Identify the research questions, (2) Identify and retrieve relevant articles, (3) Select articles, (4) Chart the data, (5) Collate, summarize and report. We structured the methods section in line with these stages.

#### *2.1. Stage 1. Identify the Research Questions*

Before conducting the review, within the group of authors we defined two preliminary research objectives and discussed the concepts to guide the literature search. We aimed at a sensitive search to catch all potentially relevant studies regarding the domains of OHL interventions, and criteria to specify the interventions regarding the phases of assessment and application of OHL interventions.

#### *2.2. Stage 2. Identify and Retrieve Relevant Articles*

First, to identify and retrieve relevant articles, we set up a literature search strategy based on search terms and inclusion criteria used in two previous reviews of OHL interventions [19,20]. Second, with the help of a librarian (TvI), we refined the research objectives and search strategy, and developed a protocol, all of which we discussed among the co-authors (MK, JS, SAR, AFdW). This was to ensure that methods and search strategies were consistent and comprehensive. We applied the final search strategy to the MEDLINE/PubMed databases and then adapted it for the other databases, covering all publications up to 31 December 2019. We searched the databases PubMed, Scopus, Psych-Info and CINAHL. In the literature search we included keywords and MESH terms related to the concept of "health literacy"; we combined these with Boolean operator AND search terms related to the health care setting, and Boolean operator OR search terms involving intervention characteristics. The complete search string is provided in Table S1 as supplementary material.

To ensure inclusion of all relevant studies in the review we used reference searches of retrieved articles to complement the electronic searches. Inclusion criteria were: (1) publication between January 2010 and December 2019; (2) inclusion of an abstract written in English; (3) an OECD country as geographical setting; (4) a study setting involving a health care setting in primary or secondary care; (5) a study aimed at assessment of organizational barriers and improvement of outcomes for adults with limited health lite-racy; (6) a study design involving an intervention, evaluation of a program, a pilot-study or needs assessment; (7) an intervention focused on assessing problems or changing two or more domains of organizational health literacy: changes at patient level (oral, written and digital communication and health literacy levels); changes at professional level (health literacy capacities and communication practices); or changes at organizational level (leadership and culture, organizational policies, systems processes, and structures).

#### *2.3. Stage 3. Selection of Articles*

After removing duplicate articles, we reviewed the title and abstract of identified articles against the following exclusion criteria: (1) health literacy was assessed or addressed only at the individual or family level (e.g., validation of screening tools or educational interventions for patients); (2) the only focus was to investigate determinants associated with health literacy and health outcomes; (3) the aim was to develop and validate instruments to measure organizational health literacy without investigating their implementation in organizations.

One investigator (MK) did the initial screening. In cases of uncertainty, a second investigator (AFdeW) reviewed the abstract or full text of an article; together consensus was reached on inclusion or exclusion in the review. Articles identified for inclusion underwent full text screening and two investigators screened a sub-section to ensure fit to criteria and consistency.

#### *2.4. Stage 4. Charting the Data*

In three steps we extracted the data from the selected studies in Excel, sorted them in tables, and analysed them based on the study purpose. First we extracted descriptive data: author, year and country, design and evaluation method, aim, setting, sample, and OHL-intervention components. Second, we extracted data on outcomes of OHLinterventions at patient, professional and organizational levels. Third, we extracted data

on whether critical factors and strategies were considered to be facilitators or barriers to implementation processes.

#### *2.5. Stage 5. Collate, Summarize and Report*

In three steps we extracted the data from the selected studies, sorted them in tables and analysed them based on key themes informed by the study purpose, to: (1) assess the outcomes of OHL-interventions, and (2) to unravel the factors and strategies affecting the implementation and outcomes of OHL-interventions. First, we tabulated the selected studies by author, year and country, research design, setting, sample, OHL domains addressed, and focus of the study, i.e., assessment or application of OHL-interventions. Second, we summarized and reported the outcomes of OHL-interventions following their assessment or application, and the level to which the outcome applied: patient, professional, and/or organization. Third, we summarized and reported factors and strategies which influenced the assessment and application of OHL-interventions, and analysed whether these were facilitators or barriers at patient, professional, and/or organizational level.

#### **3. Results**

We identified 5420 records from the literature search and one record through reference searching (we retrieved 1511 records from Pubmed; 1351 from Scopus; 1750 from Cinahl; and 808 from Psychinfo). After removing 2223 duplicates, we screened 3197 titles and abstracts and included 82 articles for full-text review. After reading the full text, we selected and excluded articles based on the criteria specified above. We included twentyfour articles in the data extraction. This results section presents: (1) description of the studies, (2) outcomes of OHL-interventions, and (3) strategies and factors that influence the implementation of OHL-interventions. Figure 1 presents the results of the literature search and study selection.

#### *3.1. Description of Studies*

The 24 selected articles involved 17 original research projects (Table 1); several studies were part of larger research projects (these were: Grabeel [22], Grabeel [23], and Tester [24]; Beauchamp [25], Goeman [26], and Jessup [27]; Mabachi [28] and Brega [29]; Vellar [30] and Mastroianni; Weaver [31] and Wray [32]). We included some articles because, although they reported on a single domain, they were connected with other articles reporting different domains of the same study. We sorted the studies according to the results of the assessment of OHL-domains, and the planning and delivery of interventions aimed at improvement of health literacy related problems. Unlike the study conducted by Cawthon et al. [33], the remaining 23 studies conducted an OHL-assessment. Thirteen of these studies focused solely on assessment of health literacy related problems [3,17,22–24,34–40]. Together with the assessment, these studies also often evaluated the feasibility of the OHL-instrument. Eleven studies reported on both the assessment and on findings regarding the planning and delivery of interventions [22,25–33,41–43]. Fourteen studies were conducted in the United States [22–24,28,29,31–34,36,38–40,43]; other studies were conducted in Australia [25–27,30,41], New Zealand & Canada [37], and several European countries including Austria [35], Italy [17], Ireland and the Netherlands [42], and Spain [3]. Study settings involved hospitals, as well as general health care settings like community and primary care practices, pharmacies and dental clinics.

The majority of the studies used a mixed-method approach (n = 16), or qualitative (n = 4) or quantitative approaches (n = 4). Multiple informants and methods were used to report on the assessment and application of OHL-interventions; these included managers, professionals, patients and observers who had taken part in surveys, interviews, focus group discussions, and observation and review of documents. The interventions targeted a variety of OHL domains using different tools and approaches. Domains most frequently addressed were the comprehensibility of written patient information materials, digital communication, oral communication, and navigation. Fewer studies targeted OHL as a

strategic priority, health literacy policies, and capacity building of staff [17,25,30–32,35]. A number of studies [3,17,31,36,40,42] used or adapted the toolkit "The Health Literacy Environment of Hospitals and Health Centers. Partners for Action: Making Your Healthcare Facility Literacy-Friendly" (HLEHHC Toolkit) developed by Rudd and Anderson [44]. Other studies used, e.g., the HLUP toolkit [28,29,34] or the Agency for Healthcare Research and Quality (AHRQ) Health Literacy Assessment Tool [39,43].

**Figure 1.** Flow of studies through the review.


**Table 1.** Descriptive results of OHL-interventions regarding research design, aim, setting, sample, and OHL-intervention.







#### *3.2. Outcomes of OHL-Interventions*

In this section, we present first the outcomes of the OHL-assessments, and second the impact after the delivery of interventions (Table 2). Findings are the result of descriptive studies. Most studies (n = 23) assessed and identified OHL-related problems at the levels of patients, professionals and organizations. Patients encountered problems relating to navigation, spoken communication, and understanding and acting upon written and digital information [3,17,23,25,28–31,34,36–43], although they also reported positive experiences [3,31,36,40]. Professionals reported limited understanding of health literacy, a lack of training, and infrequent use of recommended health literacy practices, such as use of plain language and the teach-back method [3,17,24–27,30,31,34–40,42,45]. Other studies reported that professionals had a patient-centred attitude and applied health literacy practices, but on an informal basis [17,36]. However, the assessment itself often increased awareness of health literacy problems among professionals [3,30,31,34,42]. At the organizational level, OHL was rarely considered a strategic priority, and strategic plans, policies, and routine procedures were often considered insufficient to address pro-blems related to OHL [17,25,30,31,35,38,42,43]. For example, the concept of patient-centred care was not translated into a concrete plan, and procedures to improve coordination of care were lacking [3,17,37,38].

The application of organization-wide OHL-interventions resulted in some improvement of patient outcomes [25–27,30,41], and greater changes in intermediate outcomes at professional and organizational levels [25–27,30,32,33,41,42]. Despite relatively small sample sizes, two research projects reported some improvement in patient-related outcomes [25–27,30,41], such as increased health literacy skills, participation in health care, and increased self-management abilities following interventions involving peer community members. Although not evaluated by patients, independent assessors reported both improved comprehensibility related to patient information materials [30,41], and some li-mited changes in the complexity of materials [29]. Improved health outcomes were not reported. Studies which reported greater change on intermediate outcomes at professional and organizational levels [25–27,30,32,33,41,42] used an organization-wide and long-term approach to deliver OHL-interventions. After training, (health) professionals in these studies reported increased competency to address health literacy and application of recommended practices [25–27,30,32,41,42]. Intermediate outcomes at the organizational level included integration of OHL into policies and systems, redesign of services, organization-wide programs to promote staff capacity building, and promotion of health literacy strategies by professionals in written, digital, and spoken communication [25–27,30,32,41]. Limited impact was reported regarding routine organizationwide application of practices [25,32,42], navigation, and distal outcomes such as health indicators, quality of care, patient safety, and cost-effectiveness [25,28,29,32,42,43]. A few studies with only brief implementation periods struggled with defining priorities and action plans, and reported limited changes among professionals and organizations [28,29,43], although they undertook preliminary attempts to improve written communication and train staff.

#### *3.3. Factors and Strategies Influencing the Application of OHL-Interventions*

Reported facilitators of a comprehensive OHL-assessment were: patient engagement, a change champion, commitment and capacity of staff, support from leadership and researchers, and an innovation culture, see Table 3 [3,28,31,34,39,42,43]. Patient engagement was found to be crucial for identifying health literacy problems from their perspective [25,30,42]. Health professionals needed to perceive the OHL-assessment as relevant and feasible, be committed to its implementation, and have knowledge of quality improvement [3,28,34,42,43]. Clear introduction meetings were found to increase HL awareness and staff buy-in [34,39,42,43]. Support from researchers added credibility to the intervention and promoted its quality of implementation [3,28,34,42,43]. Facilitators at the orga-nizational level were: an innovation culture focused on quality improvement, leadership support, and coordination by a change champion [28,30,32,34,43].







 related to navigation guidelines

Critical facilitators regarding the delivery of OHL-interventions were reported to be: leadership support, an organization-wide approach, a change champion and project committee, sufficient resources, professional commitment and competencies, and patient engagement, in order to achieve improvement at professional and organizational levels, see Table 3 [25–27,30–33,41,42]. An organization-wide approach, supported by senior management, was reported to stimulate the development of program logic models, strategic prioritization, and planning of OHL improvement [25,30,32]. These organizations often reported having simultaneously used top-down and bottom-up strategies to increase staff commitment to and knowledge of change strategies and quality improvement [25,30,32]. Co-design strategies and PDCA cycles were applied to develop, refine, and test interventions [25,30,32]. In contrast to the assessment phase, patients seemed to be less engaged in the application of interventions [25,42]. Only in the studies of Vellar et al. (2017) and Mastroianni et al. (2019) [30,41] were patients systematically involved in processes to improve navigation and patient-information materials. In the research project of Beauchamp et al. (2018) [25], small samples of patients were involved in the development and testing of interventions. Studies that found OHL-interventions to have only a limited impact reported that their implementation periods were brief, and affected by barriers such as lack of a change champion and coordinated planning processes [29,43], as well as limited time, resources and leadership support [22,28,29,43].

#### **4. Discussion**

The aim of this scoping review was to summarize the evidence regarding: (1) outcomes of OHL-interventions at patient, professional and organizational levels; (2) factors and strategies that influence the implementation and outcomes of OHL-interventions. We selected 24 articles, which included 17 original research projects (fully) based on qualitative and quantitative descriptive studies. With regard to the outcomes we: (a) identified OHL-related problems across patient-, professional- and organizational levels [3,25,32,34,36–38,42]; and (b) found that application of organization-wide OHL-interventions resulted in some improvement of patient outcomes [25–27,30–32,41], and greater change in intermediate outcomes at professional and organizational levels [25–27,30,32,33,41,42]. However, some studies reported only limited change [28,29,43], and no studies reported improvement on more distal outcomes. We found that several critical factors and strategies facilitated organizationwide outcomes of OHL [25–27,30–33,41,42]: leadership support, an organization-wide approach, an innovation culture, a change champion, commitment and adequate capacity of staff, and patient engagement.

Compared with the earlier reviews of Farmanova et al. [19] and Lloyd et al. (2018) [20], our findings confirmed the evidence regarding identified OHL-related problems, and we observed greater progress on the impact of organization-wide OHL interventions [25–27,30–33,41,42]. A first point regarding our evidence is that the number of OHL-pro-blems identified across a variety of countries underlines the need to use comprehensive frameworks to improve organizational health literacy in health care settings [14,35,46–50].The progress we observed related particularly to recent studies, which showed how a single health literacy project led to development of a health literate organization by employing a systematic and organization-wide approach. These studies strengthened the evidence particularly on three points: (1) patient outcomes showed some evidence of increased health literacy, understanding of information, and participation in health care [25–27,30,41]; (2) outcomes among health professionals showed evidence of improved competencies and practices to address health literacy [25,30,32,42]; (3) intermediate organizational outcomes showed evidence of embedding of OHL into policies and structures, staff training, and interventions to improve screening, communication and patient engagement [25–27,30–33,41,42]. This review thus indicates a growing awareness of how to achieve sustainable improvement on various OHL-domains, and supports the findings in recent reviews by Zanobini et al. (2020) [18] and Meggetto et al. (2020) [51].

Our review points to several critical facilitators and strategies that can promote health literacy friendly organizations in the long term: leadership support, an organization-wide approach, an innovation culture, a change champion, commitment and capacity of staff, and patient engagement [25–27,30–33,41,42]. These facilitators correspond with findings reported in other studies on innovation in health care settings [52–56] and universal processes for organizational change [19,20]. In our review, some studies reported limited outcomes because they had a shorter duration (six months) [28,29,43], struggled with coordination, staff turnover, and a lack of a change champion as well as leadership support and resources [28,29,43]. Other studies in our review suggest that a systematic organization-wide approach is more promising [25–27,30–33,41,42]. These implementation strategies involved simultaneous use of top-down and bottom-up strategies to engage staff and patients; such strategies have been widely used in the field of health promotion [32,57]. This observation underlines the frameworks of Trezona (2017) [47] and Zanobini (2020) [18] in the sense that various OHL-domains are interconnected and need to be targeted simultaneously in order to initiate a cyclical and widening process of improving the quality of health care by making organizations responsive to health literacy [51]. These findings have thus strengthened the evidence base for implementation of OHL-interventions.

However, our review also shows the evidence for OHL-interventions still to be generally weak, particularly regarding their effects on more distal outcomes like improved health or cost-effectiveness [18,20]. The first, general, issue regards the total lack of studies with an experimental design: studies conducted only baseline measurements, or had small samples when investigating change over time, and did not compare outcomes with control settings. Second, the instruments for measuring OHL outcomes did not include information on reliability and validity, although some instruments [34,44] indicated having face validity, and were used in different settings and countries [20]. Recently, several instruments were designed to assess a wide spectrum of OHL-domains [34,44,46,47], and one of these was reported to have satisfactory reliability and validity [49,58]. Although these instruments did not evaluate the outcomes of interventions, they may have the potential to be used for benchmarking and for investigating change over time [49].

The particular weakness of the evidence for OHL-interventions is that their impact is still unclear regarding more distal outcomes like patient health outcomes, quality of care, and cost reduction. This may be explained by several factors. First, in our review, mea-surement of more distal outcomes among larger samples of patients was lacking. However, we noted that, in some studies, small groups of patients were engaged in the development and evaluation of interventions [25,30,41], which resulted in improvement of health literacy levels, and in understanding and self-management of patients. Second, it seems plausible that the impact of organization-wide OHL interventions results first in intermediate outcomes among professionals and organizations, outcomes which may be influenced by many factors [14]. Zanobini [18] for example reports that (single) interventions directly targeted at patients result in improved outcomes in patient satisfaction, knowledge, and skills. In sum, promising outcomes may result from studies that combine patient-targeted interventions with systematic approaches directed at professional and organizational levels, and include measurement of distant patient outcomes, quality of care, and cost-effectiveness.

#### *4.1. Strengths and Limitations*

Several strengths of this study can be noted. We conducted a comprehensive search strategy and selection procedure to include relevant studies in the review. The fact that the selected studies were conducted in various health care organizations and countries is promising for the generalizability of the results. However, several limitations should be mentioned. First, the approach of a scoping review did not include a quality assessment of the selected studies; this limited the potential to connect content and quality. Second, we focused on peer-reviewed articles which had abstracts in English; this may have led to missing relevant studies from the grey literature or studies published in other languages. We are, however, confident that we have selected the most relevant ones. A final limitation is that publication bias may have influenced this review: studies reporting negative results

could be difficult to get published. However, we identified several studies which explicitly reported the problems encountered, and consider the influence of publication bias to be limited.

#### *4.2. Implications*

Organization-wide implementation of OHL-interventions can improve intermediate outcomes among professionals and organizations, and has the potential to mitigate health literacy problems among patients. We recommend: (1) assessing OHL problems using a comprehensive and valid instrument; (2) starting with implementation of easy-toachieve interventions; (3) using a systematic approach to achieve greater organizational change, simultaneously applying bottom-up and top-down approaches; (4) taking into account the critical facilitators of implementation: a change champion vs a project committee, lea-dership support, sufficient resources, patient involvement, and competent and committed staff.

In order to strengthen evidence on OHL-interventions, we need studies with a more rigorous design to evaluate their effectiveness, and which use OHL-instruments that have adequate reliability and validity and are suitable for the European context [14,18,20]. Furthermore, more distal patient-related outcomes like quality of care, safety, and costeffectiveness should be evaluated.

Health care organizations have primarily focused on treatment, but there is an increasing recognition of their role in health promotion and prevention in order to address health inequalities in the broader social context [14,15,25]. OHL-interventions are one approach to improve outcomes for individuals with limited health literacy. Other effective strategies may be school-based health literacy education, mass-media communication or empowering individual people as well as communities, and building health literacy competencies of (future) health professionals [59]. As such, OHL-interventions are probably most effective in combination with these other approaches, but this evidently requires further study.

A contextual factor that must be acknowledged in relation to this scoping review is that the period of the literature search preceded the start of the COVID-19 pandemic. The importance of health literacy came to the fore during the COVID-19 pandemic, as the resilience of communities and the relationship of citizens to health care providers depend on it, particularly in crisis situations. This underlines the relevance of this scoping review on OHL-interventions. The COVID-19 pandemic is likely to have influenced the field of OHL-intervention research as health care organizations have, to a greater or lesser extent, faced several periods of crisis due to exceptional service demands. The nature of this influence is unknown. Therefore, we recommend that future studies investigate the influence of the COVID-19 pandemic on the research related to organizational health literacy. Organization-wide OHL-interventions have previously required longer time periods, of several years, for changes to be implemented successfully and sustained. Since the onset of the pandemic in March 2020, health care organizations may have responded in one of two ways: putting the implementation of OHL-interventions on hold or embracing OHL quickly in response to the situation. The COVID-19 pandemic has shown that health settings can accelerate innovation, but whether this holds for OHL-interventions is to be determined.

#### **5. Conclusions**

Delivery of organization-wide OHL-interventions resulted in some improvement in patient-related outcomes and changes at the professional and organizational levels and may be a promising approach to mitigate health literacy problems. Critical success factors for organization-wide implementation are leadership support, simultaneous topdown and bottom-up approaches, a change champion and project committee, and staff commitment. Efforts to implement organization-wide OHL-interventions should take into account these critical success factors. Organization-wide interventions were reported to

achieve more positive change on OHL-domains, but evidence regarding OHL-outcomes needs strengthening.

**Supplementary Materials:** The following are available online at https://www.mdpi.com/article/10 .3390/ijerph182211906/s1, Table S1: Overview detailed search strategy.

**Author Contributions:** Conceptualization, M.S.K., S.A.R. and A.F.d.W.; methodology, M.S.K., S.A.R., J.S. and A.F.d.W.; software, M.S.K.; validation, A.F.d.W., M.S.K., J.S.; formal analysis, M.S.K. and A.F.d.W.; investigation, M.S.K.; resources, M.S.K.; data curation, M.S.K.; writing—original draft preparation, M.S.K.; writing—review and editing, S.A.R., J.S. and A.F.d.W.; visualization, M.S.K.; supervision, S.A.R.; project administration, M.S.K.; funding acquisition, A.F.d.W. All authors have read and agreed to the published version of the manuscript.

**Funding:** This study immediately builds upon the work of the IROHLA project, 2013–2016, which was coordinated by the University Medical Center Groningen and has received external funding from the European Union's Seventh Framework Programme (FP7/2007-2013) under grant agreement #305831.

**Acknowledgments:** The authors would like to thank Truus van Ittersum for her contribution to the search strategy.

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

#### **References**

