Article Text

Download PDFPDF

Small numbers, big impact: making a utilitarian case for the contribution of inclusion health to population health in England
  1. Claire X Zhang1,2,
  2. Dan Lewer3,4,5,
  3. Robert W Aldridge2,
  4. Andrew C Hayward3,4,
  5. Carlotta Cornaglia1,
  6. Peta Trussell1,
  7. Charlotte Lillford-Wildman1,
  8. Joanna Castle1,
  9. Jake Gommon1,
  10. Ines Campos-Matos1
  1. 1 Addictions & Inclusion Directorate, Office for Health Improvement and Disparities, Department of Health and Social Care, London, UK
  2. 2 Institute of Health Informatics, University College London, London, UK
  3. 3 Institute of Epidemiology & Health Care, University College London, London, UK
  4. 4 UCL Collaborative Centre for Inclusion Health, University College London, London, UK
  5. 5 Bradford Institute for Health Research, Bradford, UK
  1. Correspondence to Claire X Zhang, UCL Institute of Health Informatics, London, NW1 2DA, UK; claire.zhang.19{at}ucl.ac.uk

Abstract

Inclusion health groups make up a small proportion of the general population, so despite the extreme social exclusion and poor health outcomes that these groups experience, they are often overlooked in public health investment and policy development. In this paper, we demonstrate that a utilitarian argument can be made for investment in better support for inclusion health groups despite their small size. That is, by preventing social exclusion, there is the potential for large aggregate health benefits to the whole population. We illustrate this by reframing existing published mortality estimates into population attributable fractions to show that 12% of all-cause premature deaths (95% CI 10.03% to 14.29%) are attributable to the circumstances of people who experience homelessness, use drugs and/or have been in prison. We also show that a large proportion of cause-specific premature deaths in the general population can be attributed to specific inclusion health groups, such as 43% of deaths due to viral hepatitis (95% CI 30.35% to 56.61%) and nearly 4000 deaths due to cancer (3844, 95% CI 3438 to 4285) between 2013 and 2021 attributed to individuals who use illicit opioids. Considering the complexity of the inclusion health policy context and the sparseness of evidence, we discuss how a shift in policy framing from ‘inclusion health vs the rest of the population’ to ‘the impact of social exclusion on broader population health’ makes a better case for increased policy attention and investment in inclusion health. We discuss the strengths and limitations of this approach and how it can be applied to public health policy, resource prioritisation and future research.

  • Health inequalities
  • HEALTH POLICY
  • MORTALITY

Data availability statement

Data are available in a public, open access repository.

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Inclusion health as a complex policy area

Inclusion health is an umbrella term for approaches to reduce extreme health inequalities experienced by socially excluded and stigmatised populations.1 It can refer to migrants in vulnerable circumstances, survivors of human trafficking and modern slavery, Gypsy, Roma and Traveller communities, people who experience homelessness, people who experience drug and alcohol dependence, sex workers and people in contact with the justice system.2 These groups of individuals experience a multitude of intersecting risk factors for poor health. While health outcomes among the general population are distributed across a social gradient, inclusion health groups experience a ‘cliff-edge’,3 with much worse health outcomes than groups typically considered as socioeconomically deprived in the general population, such as those living in poor neighbourhoods and those with low income.4 . In England, for example, chronic obstructive pulmonary disease prevalence was 1% and 2% for housed individuals in the least and most deprived quintiles respectively, but 14% in people experiencing homelessness.3 An international review of mortality rates among inclusion health groups in high-income countries found rates were approximately 12 times higher than in the general population for women, and 8 times higher for men.5 Despite these striking statistics, there are several challenges to health policy and services for inclusion health groups in England.

First, the legislative and policy context concerning inclusion health is complex. The Equality Act 2010 makes it illegal to discriminate against people based on several factors including age, sex, ethnicity and disability.6 Health inequalities related to characteristics such as homelessness do not fall under the protected characteristics outlined in this Act. Therefore, despite mounting evidence of the stark health inequalities experienced by inclusion health groups, there is no legal and public sector duty to address discrimination or exclusion on the basis of characteristics such as housing status, sex work and drug use. Progress on policy development for some inclusion health groups has been faster than for others. In 2019, the UK government committed to end rough sleeping by 2024 and recently published an ambitious Rough Sleeping Strategy, also extending the Rough Sleeping Drug and Alcohol Treatment Grant.7 However, whether these policies have led to quantifiable changes in health outcomes is yet to be determined. Inclusion health is also a politicised topic, with controversial policies like offshore processing,8 data sharing agreements between the National Health Service and immigration enforcement,9 opposition to supervised injection facilities10 and criminalisation of various activities surrounding sex work.11

Second, evidence concerning inclusion health outcomes is sparse in comparison to evidence related to broader health inequalities and protected characteristics. There are several reasons for this. Data on indicators for social exclusion and health outcomes for inclusion health groups are not consistently collected in routine administrative records.2 Challenges in identifying inclusion health groups in these datasets are compounded by poor data quality as well as disjointed data systems. There is also a lack of harmonisation in defining inclusion health and difficulties with estimating the extent of overlap between various dimensions of social exclusion.12 The various methods that researchers have used to estimate population sizes and health outcomes has resulted in limited comparability between studies, particularly for some inclusion health groups. Undertaking primary data collection is also a challenge due to the mobile and excluded nature of inclusion health groups. In some circumstances, it may also be unethical or inappropriate to ask about inclusion health characteristics.

Additionally, real-terms government spending on public health (COVID-19 aside) has markedly reduced since 2015.13 The relatively small size of inclusion health groups means that it can be challenging to make a strong policy and economic case for increased or even sustained funding despite the deontological arguments in some of the literature to date, that is, the moral and ethical imperative for doing so. As such, there are policy benefits that could be realised by understanding (A) how investment in inclusion health will contribute to broader efforts to improve population health and reduce inequalities, (B) how resources from whole-of-population disease-specific programmes can be allocated towards inclusion health groups, particularly those that align with other government priorities and (C) how investment in specialist services tailored to the needs of inclusion health groups can be integrated with whole-of-population disease-specific programmes.

As such, we aim to demonstrate how a shift in policy framing from ‘inclusion health versus the rest of the population’ to ‘the impact of social exclusion on broader population health’ can make a utilitarian argument for inclusion health, suggesting that policy development and investment in inclusion health will also result in measurable benefit to the whole population. To illustrate this, we conducted this proof-of-concept analysis of premature mortality using various sources of published data. We discuss ways in which the results could be applied to policy and resource prioritisation, as well as the methodological strengths and limitations of this approach and considerations for future research.

Reframing existing evidence using population attributable fractions

Population attributable fractions (PAFs) are commonly used to quantify the excess burden of a health outcome attributable to an exposure.14 15 This approach can help to frame the contribution of inclusion health groups to broader population health outcomes. It also allows a simple way of comparing the relative contribution of social exclusion across different health outcomes. We used a rapid pragmatic approach to estimating PAFs by collating data from a range of published epidemiological studies of inclusion health groups. Similar approaches to estimating PAFs using multiple data sources have been used by well-known studies such as the Global Burden of Disease.14 16

To estimate PAFs for premature mortality attributable to extreme social exclusion, we used Levin’s PAF formula14 17 and standardised mortality ratios (SMRs) as the relative risk (RR) inputs. We then used these PAFs to calculate the absolute number of attributable premature deaths in England.18 Online supplemental appendix 1 provides further detail on our analytical methods. Online supplemental appendix 2 shows the data inputs and sources used for each PAF, including the SMR, population size (p) and total number of deaths in England. We used a summary SMR from a meta-analysis of mortality in high-income country studies to calculate an all-cause PAF for inclusion health groups (including people experiencing homelessness, prisoners and people who use illicit drugs but not including Gypsy, Roma and Traveller populations or migrants in vulnerable circumstances),5 followed by an illustrative example of cause-specific PAFs using data from a recent longitudinal study of mortality in people who use illicit opioids in England.19 We estimated population sizes using Office for National Statistics (ONS) routine statistical reports20 21 and the integrated statistical profile of severe and multiple disadvantage from the Hard Edges report which used administrative and survey data.12 We estimated uncertainty using a Monte Carlo method in which we generated simulations of the PAF by sampling from distributions of each parameter. To calculate the absolute number of attributable deaths, all available ONS data on all-cause and cause-specific deaths for people aged 15–64 years in England were totalled across 2013–2021.22 23 All data used in this study are publicly available.

Supplemental material

The contribution of inclusion health to premature deaths in England

The resultant PAFs and number of deaths attributed to inclusion health groups are presented in figure 1 and online supplemental appendix 2. Since the majority of individuals within inclusion health groups are under 65 years12 and the meta-analysis we used to estimate PAFs included studies contained a very small number of deaths over 65 years, all attributable deaths in this paper could be considered as premature mortality.

Figure 1

PAFs and number of deaths attributable to opioid use. Right graph shows PAFs and 95% CIs. Left graph shows the total number of deaths in England 2013–2021 for each cause of death (red) and number of deaths attributable to opioid use (blue), with a break in the x-axis between 250 and 650 thousand. Cause-specific mortality is grouped by International Classification of Diseases 10th Revision (ICD-10) chapter (upper case labels) or specific conditions as outlined Lewer et al’s3 study (lower case labels). PAF, population attributable fraction. COPD, chronic obstructive pulmonary disease.

For all-cause mortality in inclusion health groups, we found that 11.99% (95% CI 10.03% to 14.29%) of all premature deaths in England are attributable to the circumstances of people who experience homelessness, use drugs, and/or have been in prison. In other words, almost 1 in 8 premature deaths in England, equating to a total of 80 435 (95% CI 67 264 to 95 866) deaths between 2013 and 2021, could have been avoided if these individuals did not experience these characteristics of social exclusion. This estimate does not include excess premature deaths in Gypsy, Roma and Traveller populations, migrants in vulnerable circumstances, or other socially excluded groups for which we did not have population estimtes. This finding compares to 35.6% of premature mortality attributable to area-based inequalities in England.15 Despite inclusion health groups accounting for less than 2% of the adult population in England (and only 1% in this example limited to people experiencing homelessness, prisoners and individuals with substance use disorders), their relative contribution to premature mortality in the country is remarkably high.

Outputs canalso be used to frame the degree of impact on population health outcomes in relative or absolute terms. For cause-specific premature mortality in the general population that is attributable to individuals with history of illicit opioid use (figure 1), the highest proportion of attributable deaths was for viral hepatitis (43.03%, 95% CI 30.35% to 56.61%). In absolute number of deaths, this equated to 508 (95% CI 358 to 669) attributable premature deaths between 2013 and 2021. Contrastingly, the highest number of absolute premature deaths attributed to illicit opioid use was for cancers (3844, 95% CI 3438 to 4285), but in relative terms the proportion attributable was only 1.57% (95% CI 1.41% to 1.75%) of all deaths due to cancer in the general population.

Implications for inclusion health policy

We anticipate four main policy applications of the outputs generated using this approach. First, they can be used to inform high-level policies within and outside of government, for example, policies to prevent social exclusion and its manifestations such as homelessness, illicit drug use and imprisonment. The outputs can also help to achieve alignment between priorities for inclusion health and broader cross-government priorities for population health, for example, England’s HIV Action Plan 2022–2025,24 the NHS Hepatitis C Elimination Programme25 and the NHS Long Term Plan.26

Second, outputs can be used to inform resource prioritisation and allocation. Situating inclusion health outcomes within the context of achieving broader population health improvements can inform future spending decisions for inclusion health programmes. They can also support national and local public health teams to prioritise internal budgets, particularly when doing so could enable both disease-specific and equity-based health improvement targets to be simultaneously achieved (eg, reducing health inequalities while addressing national and global hepatitis elimination targets).27 This reduces the tension between high-risk approaches and population approaches to public health interventions, thereby achieving both efficiency and equity.

Third, given the range of health conditions in which there is notable increased excess premature mortality attributable to extreme social exclusion, such evidence can inform service design to provide integrated prevention and management across conditions, tailored to the needs of inclusion health populations. These services may include specialist primary care, multidisciplinary teams and outreach. The large PAFs across different disease areas also demonstrates the need for close integration of inclusion health services with mainstream services and for mainstream services to better meet the needs of inclusion health groups.

Finally, this approach of reframing the inclusion health narrative can support policy development by informing policy analysis, cost analysis and implementation research. Sustainable, tailored and co-produced interventions are much needed in inclusion health.1 Yet, the burden of ill health in inclusion health groups is so vast that pilot interventions are often spread across numerous target health outcomes without adequate integration, resourcing or buy-in to evaluate and scale them up.28 In this regard, PAFs can support programme evaluation, policy analysis, cost analysis and implementation research to focus on specific areas where the greatest magnitude of population health benefit may be identified.

Methodological considerations and implications for future research

In all policy applications of PAF outputs for inclusion health, researchers must be clear in communicating the limitations of their methods and the uncertainties contained within the results. This includes disclosing the nature and source of the underlying data used, as well as the specific inclusion health groups and age ranges that the estimates concern. Application of these rapid PAF outputs to policy also makes causal assumptions about the relationship between various dimensions of social exclusion and resultant health outcomes. This implies a counterfactual that individuals within inclusion health groups never experienced social exclusion. This is a conceptual challenge given that, eradicating homelessness, for example, may be possible although difficult, but it is not possible to eradicate the historical social exclusion that has caused poor health outcomes today. Therefore, while the counterfactual is not a realistic scenario, the PAFs in this paper are a measure of the importance of public health investment in reducing social exclusion.

Differences in availability of peer-reviewed studies and quantitative estimates for certain inclusion health groups and health outcomes can also create blind spots. This is particularly relevant for inclusion health groups such as Gypsy, Roma and Traveller communities, asylum seekers and sex workers for whom little data and research exist. In this paper, we only produced cause-specific PAFs derived from a study about people who use illicit opioids, which was more detailed than for many inclusion health groups. The use of PAFs for policy development must, therefore, be balanced with qualitative research and grey literature evidence about the health and well-being of inclusion health groups for which high-quality research is scarce.

Additionally, our approach assumes a stable RR throughout the life course. However, this is unlikely to be the case. For example, while some SMRs for inclusion health groups are measured at a time of high risk, inclusion health groups are dynamic and changes in individuals’ circumstances may temporarily or more permanently reduce the excess mortality risks (eg, acquiring stable housing, changing immigration status). This affects both the RRs and the populations estimates. Additionally, SMRs are age-related and typically reduce with age as reference mortality rates in the general population increase, even while absolute excess mortality risks for inclusion health groups remain stable or even increase. Calculating the proportion of deaths attributable to extreme social exclusion based on the SMR measured among, for example, young people experiencing homelessness, will therefore overestimate the PAF. It may be valuable to estimate age-specific PAFs to address this issue.

Using RRs and p’s produced at one point or period in time to estimate PAFs assumes that the relative burden of the health outcome in question and the inclusion health groups have remained constant over time. With increasing longitudinal research and monitoring of inclusion health outcomes, these static inputs have the potential to become dynamic. Even when RRs are constrained by availability of published research, p’s and absolute number of deaths may be interchanged to correspond with time-period-specific policy questions.

We propose the present approach in the absence of more robust analysis to estimate PAFs with harmonised p’s and RRs from the same population-based datasets. As the phenotyping of social exclusion indicators and linkage of routine datasets improve over time,29 30 researchers should incorporate the calculation of PAFs (the relative metric and the absolute numbers attributable) as an additional step in their analyses. This will also allow the calculation of age-specific RRs rather than borrowing age-adjusted SMRs from existing studies. Longitudinal cohorts, repeated cross-sections and case–control designs with sufficient population coverage would allow researchers to investigate how PAFs change over time, within narrower age bands, and across regions and local areas. Examining PAFs for local policy action is of particular importance since geography and mobility is pertinent to the location of inclusion health groups. For example, while this paper estimated that inclusion health groups account for around 12% of all premature deaths in England, at a local level, people experiencing homelessness have been found to account for over 80% of premature deaths in Carfax, the Oxford ward with the highest SMR.31

Comparative approaches using PAFs should also complement other approaches to situating inclusion health within the broader context of population health. This may include estimation of years of life lost and years of healthy life lost due to disability, measurement of healthcare activity such as unplanned hospitalistations attributable to inclusion health groups, forecasting of future trends in morbidity and mortality, and comparative research on the impact of various interventions aimed at reducing deaths and disability.

Nevertheless, even with creative methods like the one proposed in this paper, the dearth of existing research on some inclusion health groups and health outcomes will continue to obscure the visibility of these groups until there is a sizeable improvement to data collection for social exclusion characteristics and greater investment in robust epidemiological research in this field.

Conclusion

Quantifying the impact of social exclusion on broader population health outcomes is a useful tool for policy development in inclusion health. While the methodological assumptions made by this approach means that outputs should be interpreted with caution, it allows for the reframing of existing research. It moves from juxtaposing inclusion health outcomes with the rest of the population to quantifying how much the burden of poor health in inclusion health groups contributes to wider population health outcomes and wider social inequalities in health. This, in turn, enables a utilitarian argument to be made for investment in inclusion health. Comparing previous estimates of premature mortality attributable to area-based inequalities in England (35%) to the 12% attributable to extreme social exclusion provides an indication of the central importance of inclusion health within policies to reduce health inequalities. This reframing of existing evidence supports greater understanding of how investing in and acting on inclusion health might simultaneously achieve other government priorities and population health targets.

Data availability statement

Data are available in a public, open access repository.

Ethics statements

Patient consent for publication

References

Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

Footnotes

  • Twitter @claire_x_zhang

  • Contributors Conceptualisation: CXZ, AH, DL, RA and IC-M. Methodology: CXZ, AH, DL, RA and IC-M. Production of visualisations: CXZ, DL and CC. Analysis: CXZ, DL and RA. Writing-original draft preparation: CXZ. Writing review and editing: All authors. Project administration: CXZ. Guarantor: CZ. All authors have read and agreed to the published version of the manuscript.

  • Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.