Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Morbidity and mortality in homeless individuals, prisoners, sex workers, and individuals with substance use disorders in high-income countries: a systematic review and meta-analysis

Morbidity and mortality in homeless individuals, prisoners, sex workers, and individuals with... Articles Morbidity and mortality in homeless individuals, prisoners, sex workers, and individuals with substance use disorders in high-income countries: a systematic review and meta-analysis Robert W Aldridge, Alistair Story, Stephen W Hwang, Merete Nordentoft, Serena A Luchenski, Greg Hartwell, Emily J Tweed, Dan Lewer, Srinivasa Vittal Katikireddi, Andrew C Hayward Summary Lancet 2018; 391: 241–50 Background Inclusion health focuses on people in extremely poor health due to poverty, marginalisation, and multimorbidity. We aimed to review morbidity and mortality data on four overlapping populations who experience Published Online November 11, 2017 considerable social exclusion: homeless populations, individuals with substance use disorders, sex workers, and http://dx.doi.org/10.1016/ imprisoned individuals. S0140-6736(17)31869-X See Comment page 186 Methods For this systematic review and meta-analysis, we searched MEDLINE, Embase, and the Cochrane Library See Review page 266 for studies published between Jan 1, 2005, and Oct 1, 2015. We included only systematic reviews, meta-analyses, Centre for Public Health Data interventional studies, and observational studies that had morbidity and mortality outcomes, were published in Science, Institute of Health English, from high-income countries, and were done in populations with a history of homelessness, imprisonment, Informatics (R W Aldridge PhD, A Story PhD, S A Luchenski FFPH, sex work, or substance use disorder (excluding cannabis and alcohol use). Studies with only perinatal outcomes and D Lewer MSc, studies of individuals with a specific health condition or those recruited from intensive care or high dependency Prof A C Hayward MD), The Farr hospital units were excluded. We screened studies using systematic review software and extracted data from Institute of Health Informatics published reports. Primary outcomes were measures of morbidity (prevalence or incidence) and mortality Research (R W Aldridge, A Story, S A Luchenski, D Lewer, (standardised mortality ratios [SMRs] and mortality rates). Summary estimates were calculated using a random Prof A C Hayward), and the effects model. Institute of Epidemiology and Health Care (Prof A C Hayward), Findings Our search identified 7946 articles, of which 337 studies were included for analysis. All-cause standardised University College London, London, UK; University College mortality ratios were significantly increased in 91 (99%) of 92 extracted datapoints and were 11·86 (95% CI London NHS Foundation Trust, 10·42–13·30; I²=94·1%) in female individuals and 7·88 (7·03–8·74; I²=99·1%) in men. Summary SMR estimates for London, UK (A Story); Centre for the International Classification of Diseases disease categories with two or more included datapoints were highest for Urban Health Solutions, Li Ka deaths due to injury, poisoning, and other external causes, in both men (7·89; 95% CI 6·40–9·37; I²=98·1%) and Shing Knowledge Institute, St Michael’s Hospital, women (18·72; 13·73–23·71; I²=91·5%). Disease prevalence was consistently raised across the following categories: Toronto, ON, Canada infections (eg, highest reported was 90% for hepatitis C, 67 [65%] of 103 individuals for hepatitis B, and 133 [51%] of (Prof S W Hwang MD); Mental 263 individuals for latent tuberculosis infection), mental health (eg, highest reported was 9 [4%] of 227 individuals for Health Centre Copenhagen and schizophrenia), cardiovascular conditions (eg, highest reported was 32 [13%] of 247 individuals for coronary heart Institute of Clinical Medicine, Faculty of Health and Medical disease), and respiratory conditions (eg, highest reported was 9 [26%] of 35 individuals for asthma). Sciences, University of Copenhagen, Denmark Interpretation Our study shows that homeless populations, individuals with substance use disorders, sex workers, (Prof M Nordentoft DMSc); and imprisoned individuals experience extreme health inequities across a wide range of health conditions, with the Department of Social and Environmental Health relative effect of exclusion being greater in female individuals than male individuals. The high heterogeneity between Research, London School of studies should be explored further using improved data collection in population subgroups. The extreme health Hygiene & Tropical Medicine, inequity identified demands intensive cross-sectoral policy and service action to prevent exclusion and improve health London, UK (G Hartwell MFPH); outcomes in individuals who are already marginalised. and Medical Research Council/ Scottish Government Chief Scientist Office Social and Funding Wellcome Trust, National Institute for Health Research, NHS England, NHS Research Scotland Scottish Public Health Sciences Unit, Senior Clinical Fellowship, Medical Research Council, Chief Scientist Office, and the Central and North West London University of Glasgow, NHS Trust. Glasgow, UK (E J Tweed MFPH, S Vittal Katikireddi PhD) Correspondence to: Copyright © The Author(s). Published by Elsevier Ltd. This is an Open Access article under the CC BY 4.0 license. Dr Robert W Aldridge, The Farr Institute of Health Informatics Introduction observed social gradients in health do not capture the full Research, University College Inclusion health is a research, service, and policy agenda extent of health inequities for individuals who experience London, London NW1 2DA, UK [email protected] that aims to prevent and redress health and social considerable social exclusion. inequities among people in extremely poor health due to Previous research has described the high prevalence of 1 2 poverty, marginalisation, and multimorbidity. The substance use disorders in homeless populations, prison­ 3 4 association between socio economic status and health ers, and sex workers, and the increased prevalence 5 6 outcomes is well established. However, these commonly of homelessness in prisoners and sex workers. These www.thelancet.com Vol 391 January 20, 2018 241 Articles Research in context Evidence before this study mortality outcomes across a range of inclusion health A comprehensive body of research exists on the health effect of populations. We found that the extent of the health inequity inequity, much of which focuses on disparities in morbidity seen in our inclusion health populations greatly exceeded and mortality, and is based on common measures of that previously observed between populations with high and socioeconomic status, such as neighbourhood deprivation and low socioeconomic status and was consistent across inclusion occupational class. A consistent association has been found health populations. Mortality rates are extremely high across between ill health and increasing levels of social deprivation, the International Classification of Diseases, tenth revision which has underpinned a broad range of social policies and disease categories in inclusion health populations, and our public health initiatives. Such analyses cannot adequately review is the first to show that relative risks are consistently assess the extent of health inequity faced by individuals who higher in female than male individuals. experience considerable social exclusion. In preparation for this Implications of all the available evidence Review, we searched the Cochrane Library, MEDLINE, and The extreme burden of disease experienced by inclusion health Embase databases for articles published between Jan 1, 2005, populations demands a cross-sectoral response to prevent and Sept 30, 2013. We searched for systematic reviews, meta- considerable social exclusion and an improvement in services analyses, cohort studies, and cross-sectional studies containing that work with these populations. Our analyses focused on morbidity and mortality outcomes for the four inclusion health relative measures of mortality and therefore future work should populations of interest (substance use disorders, homeless examine absolute measures in greater detail. Inclusion health populations, prisoners, and sex workers). We only included populations are often invisible within routine health data. full-text articles published in English. Full search terms are This limitation can be addressed by modifying the instruments listed in the appendix. The studies identified described the used to collect such data or through data linkage studies. highly overlapping nature of inclusion health populations, Services that provide for inclusion health populations should the increased risk factors for disease, and poor mortality aim to deliver health and social services for overlapping outcomes compared with the general population. Previous marginalised groups to tackle the poor health outcomes found systematic reviews have analysed health outcomes of in this study. These services should also have a greater focus on individual inclusion health populations, but none have prevention and management of more common conditions in examined the populations together. addition to those traditionally considered high risk for inclusion Added value of this study health groups. Our systematic review and meta-analysis provides the first comprehensive examination to date of morbidity and marginalised populations have common inter secting Methods characteristics and adverse life experiences that lead to Search strategy and selection criteria considerable social exclusion, making them powerful For this systematic review and meta­ analysis, we determinants of marginalisation in high­ income settings. searched the Cochrane Library, MEDLINE, and Embase When considered separately, marginalised popu lations for articles published between Jan 1, 2005, and Oct 1, 2015. 8–10 See Online for appendix have been shown to have high all­ cause mortality. Full search terms are provided in the appendix. We However, despite the considerable overlap in risk factors searched for articles about the populations of interest and the substantially increased mortality observed in (homeless individuals, prisoners, sex workers, and these populations, no previous review has examined the individuals with substance use disorders, excluding outcomes of these groups together. cannabis and alcohol use) from systematic reviews, No universally agreed theoretical framework exists to meta­ analyses, inter ventional studies, and observational describe inclusion health. In this Article, we build on studies that had morbidity and mortality outcomes. We existing social exclusion theory and consider the so­ called included studies identified from references of included linked and cumulative factors and processes that confound articles. We only included full­ text articles published in individual and group capacity for hope, opportunity, English that were done in high­ income countries 11 13 reciprocity, and participation. Our analysis is also (classified according to the World Bank classification ). informed by an intersectionality perspective, which We excluded studies with only perinatal outcomes and focuses on how social characteristics combine to have an did not include data on perinatal outcomes from studies 2,12 effect on health. that otherwise met our inclusion criteria. We excluded Our systematic review therefore aims to examine articles that limited the study population to individuals mortality and morbidity in homeless populations, with a specific health condition and studies that recruited prisoners, sex workers, and individuals with substance participants exclusively from intensive care or high use disorders, who experience considerable exclusion. dependency hospital units. 242 www.thelancet.com Vol 391 January 20, 2018 Articles We recognise that social exclusion has a major effect on stratifying the analyses by country and by inclusion health in other social groups, including Gypsies and health population group. Travellers, migrants, ethnic minorities, indigenous com­ munities, and sexual and gender minorities. Although Role of the funding source these groups experience social exclusion in many high­ The funders of the study had no role in study design, income settings, they were considered beyond the scope data collection, data analysis, data interpretation, writing of this systematic review. of the report, or the decision to submit the paper for publication. All authors had full access to all the data in Data analysis the study and had final responsibility for the decision to RWA screened titles, abstracts, and full texts using submit for publication. Covidence systematic review software. All authors con­ tributed to data extraction, and data were double­ checked Results by a second researcher (RWA, EJT, GH, or SVK). We identified 7946 articles, of which 1274 were duplicates Extracted items included study design, year or years of (figure 1). Of the 711 full ­ text articles retrieved, 418 met study, country, number of participants, primary outcomes, the inclusion criteria. We excluded a further 81 articles and summary descriptions of the study popu lation. We because of overlapping data. A total of 337 studies were tried to contact authors if we were unable to locate papers included in this Article, which included 2835 datapoints or required additional information about the data or study. (ie, effect estimates for a unique population) af ter the We attempted to identify and exclude duplicate data removal of 384 duplicates. from research studies presented in separate publications. The studies were from 38 countries (appendix). The For cases in which we identified multiple studies with USA contributed 698 datapoints, Australia contributed duplicated or overlapping data (by population, time, place, 460, Sweden contributed 309, Canada contributed 257, and and outcome) we selected the study with the largest or the UK contributed 234. Populations with substance use most representative sample size, and when these were disorders were the most studied subgroup, accounting for also similar, we present the most recent study. We followed 1193 (42·1%) of 2835 datapoints, followed by prisoners the PRISMA reporting guidelines in the presentation of (769 [27·1%]), homeless populations (754 [26·6%]), and sex our manuscript. A review protocol was not published workers (119 [4·2%]). before this review was done. Infectious diseases and mental and behavioural disorders Outcomes included were measures of morbidity and were the two most studied ICD­ 10 categories with infectious mortality for conditions defined in the International diseases accounting for 898 (31·6%) of 2835 datapoints, Classification of Diseases, tenth revision (ICD­ 10). Out­ and mental and behavioural disorders accounting for comes were reported using a variety of measures. To ensure maximum comparability across studies for mor­ tality outcomes, we extracted, in order of pref erence, the 7946 potentially eligible studies identified by database search first of the following measures: standardised mortality ratio (SMR), hazard ratio, mortality rate ratio, or crude 1274 duplicates excluded mortality rate. For consistency with most studies included in this Article, we have not multiplied SMRs by 100. In our results, a value of 1 equates to no difference between the 6672 identified for screening expected and observed mortality rate. For morbidity outcomes, we extracted, in order of preference, the first of 5961 excluded on title and abstract the following measures: prevalence, incidence, prevalence risk ratio, incidence rate ratio, prevalence odds ratio, or incidence odds ratio. When available, we used data in 711 full-text articles assessed for eligibility which the comparison group was a socially deprived population or measures were adjusted for area­ based or 293 excluded after full-text screening income­ based deprivation. A link to all extracted data is included in the appendix. For the quantitative findings analysed in this study, we 418 met the inclusion criteria focused the synthesis on SMRs. SMRs for all­ cause mortality and by ICD­ 10 disease category were summar­ 81 excluded because of overlapping ised in forest plots. We anticipated high levels of data with other studies heterogeneity, and therefore did summary estimates with random effects models using Stata version 13. We used 337 included in quantitative synthesis the I² statistic to indicate the proportion of total variation in study estimates due to heterogeneity. Figure 1: Study selection We explored potential sources of hetero geneity by www.thelancet.com Vol 391 January 20, 2018 243 Articles External causes Circulatory Infectious and parasitic diseases Injury and poisoning Neoplasms Digestive Genitourinary Skin Respiratory Mental and behavioural disorders Nervous system Endocrine Musculoskeletal 1 Symptoms, signs, and abnormal clinical and laboratory findings 2 Diseases of the blood SMR 3 Diseases of the eye and adnexa 0 510152025 4 Diseases of the ear and mastoid process Figure 2: Treemap summarising the amount of available data grouped according to the ICD-10 disease categories and summary estimates of SMRs Box sizes indicate the total number of datapoints included in this Article. SMRs used are summary estimates for the ICD-10 disease categories for both sexes combined. Grey boxes (SMR of 0) indicate that none of the studies included in this Article reported SMR for both sexes combined. ICD-10=International Classification of Diseases, tenth revision. SMR=standardised mortality ratio. 715 (25·2%) datapoints (figure 2, appendix p 4). Injury and ICD­ 10 categories (appendix pp 6–7). In some ICD­ 10 poisoning only accounted for 98 (3·4%) of all extracted categories, the summary SMRs for both sexes combined datapoints. did not fall between the male and female estimates Our all­ cause meta­ analyses focused on SMRs because the meta­ analyses used data from different 8–10,15–40 and included 29 studies, which contributed studies (rather than the estimate for both sexes combined 92 datapoints (table, figure 3, appendix). 91 (99%) of the being drawn from the male and female populations). 92 all­ cause SMRs were increased and overall we estimated We identified 201 papers reporting outcomes for that summary all­ cause SMRs were higher in female infectious and parasitic diseases. Summary estimates of individuals (11·86 [95% CI 10·42–13·30]; I²=94·1%; SMRs for infectious diseases were increased in male figure 3) than male individuals (7·88 [7·03–8·74]; individuals (2·83 [95% CI 1·61–4·05]; I²=65·4%; I²=99·1%; figure 3). We provide summary estimates of appendix p 6) and female individuals (5·58 [1·46–9·70]; SMRs; however, the I² statistic indicated that data were I²=60·0%; appendix p 6) and both sexes combined heterogeneous in many of our analyses and therefore (11·43 [6·91–15·94; I²=97·0%; appendix p 6). Disease these summary measures must be interpreted with prevalence was high but heterogeneous and ranged from 41 42 43 appropriate caution. Heterogeneity was not substantially 0% to 54% for HIV infection, from less than 0·1% to 42 44 reduced when analyses were stratified by population 90% for hepatitis C, from 2% (two of 119) to 65% 45 46 subgroup (appendix). Insufficient data were available to (67 of 103) for hepatitis B, and from 1% (one of 82) to do subgroup analyses by country. 51% (133 of 263) for latent tuberculosis infection. Summary SMRs were higher in female individuals Summary estimates of SMRs for injury, poisoning, and than in male individuals for mortality in each of the other external causes were the highest across all of the 244 www.thelancet.com Vol 391 January 20, 2018 Articles Study years Country Participants (n) Population description Homeless people Nielsen et al 1999–2009 Denmark 32 711 Women aged 16 years or older with at least one contact with a homeless shelter Roy et al 1995–2001 Canada 829 Individuals aged 14–25 years with unstable housing Vila-Rodriguez et al 2008–11 Canada 293 Prospective community sample of adults living in single-room occupancy hotel Prisoners Graham et al 1996–2007 UK 76 627 Male individuals imprisoned for the first time betw een 1996 and 2007 Kariminia et al 1988–2002 Australia 85 203 All adults who had been in full-time custody Individuals with substance use disorders Arendt et al 1996–2006 Denmark 20 581 People receiving treatment in specialist institutions for substance use disorder, who reported cocaine as their primary substance Bargagli et al 1996–2002 Netherlands 2575 Male opiate users aged 15–69 years entering treatment Barrio et al 2004–06 Spain 714 Regular cocaine users recruited from drug scenes and non-treatment settings Bjornaas et al 1980–2000 Norway 185 Individuals with opioid addiction admitted to hospital because of self-poisoning Darke et al 2001–09 Australia 615 Opioid users Degenhardt et al 1985–2005 Australia 43 789 People who are opioid-dependent treated with opioid substitution therapy Evans et al 2005–07 USA 644 Injecting drug users younger than 30 years Gibson et al 1980–2006 Australia 2489 Opioid users Hser et al 2000–02 USA 4447 Women who were admitted to drug treatment programmes Lee et al 2006–08 Taiwan 10 842 Heroin users attending opioid substitution therapy Mathers et al 1980–99 Denmark 101 People who injected opioids and other drugs Merrall et al 1996–2006 UK 69 456 People in contact with drug treatment services 29,30 Nyhlen et al 1970–2006 Sweden 561 Substance abusers admitted for inpatient detoxification Pavarin et al 1988–2012 Italy 471 Individuals who had visited a public treatment centre for problems due to cocaine use Rehm et al 1994–2000 Switzerland 6281 Participants in heroin-assisted treatment Rosca et al 1999–2008 Israel 9818 Patients who had ever been treated or were currently in treatment in methadone maintenance treatment clinics Singleton et al 1997–2002 Czech 3039 Drug users admitted to hospital for drug-related problems Republic Spittal et al 1996–2002 Canada 520 Injecting drug users recruited through self-referral and street outreach Stoove et al 1990–2006 Australia 220 Injecting drug users recruited from the community van Santen et al 1985–2012 Netherlands 1254 Individuals recruited from local methadone outposts, a sexually transmitted diseases clinic, and by word of mouth Zabransky et al 1996–2008 Czech 151 Injecting drug users aged 15–18 years Republic Degenhardt et al 1996–2004 Canada 717 People who injected cocaine daily Degenhardt et al 1985–2006 Australia 42 676 Opioid users Table: Studies included in the standardised all-cause mortality ratio meta-analyses ICD­ 10 categories, in male individuals (7·89 [95% CI were from populations with substance use disorders 6·40–9·37]; I²=98·1%; appendix p 7), female individuals only. Only two studies included data on male individuals, (18·72 [13·73–23·71]; I²=91·5%; appendix p 7), and both one study on female individuals, and two studies on both sexes combined (23·53 [15·34–31·71]; I²=99·6%; appendix sexes combined. Prevalence of major depression in p 7). However, these categories only accounted for 98 (3%) inclusion health populations ranged from 3% (one of of 2835 extracted datapoints. Summary SMR estimates 38 individuals) in the month before assessment to were also increased for external causes of morbidity and a 53% (25 of 47 individuals) lifetime prevalence. mortality in male individuals (6·52 [95% CI 5·54–7·51; Prevalence of schizophrenia ranged from 0·9% (212 of I²=97·4%; appendix p 7), female individuals (13·15 23 530 individuals; we estimated the numerator on the [9·87–16·43]; I²=93·7%; appendix p 7), and both sexes basis of data in the original article) to 4% (nine of 51 49 combined (8·50 [6·89–10·10]; I²=97·5%; appendix p 7). No 227 individuals), and from 0% (none of 53 individuals) data from studies that included sex workers were used in to 45% (221 of 495 individuals; numerator estimated) any of the SMR estimates for injuries or external causes. for bipolar disorder. SMRs for mental and behavioural disorders for male Summary estimates of SMRs for neoplasms were individuals and female individuals were exclusively from increased in male individuals (1·61 [95% CI 1·30–1·92]; prison populations and data for both sexes combined I²=88·7%; appendix p 6), female individuals (1·91 www.thelancet.com Vol 391 January 20, 2018 245 Articles A Male individuals B Female individuals SMR (95% CI) SMR (95% CI) Homeless Homeless 10 10 Nielsen et al (2011) 5·60 (5·40–5·80) Nielsen et al (2011) 6·70 (6·20–7·10) Prisoners Prisoners 17 17 Kariminia et al (2007) 3·70 (3·60–3·80) Kariminia et al (2007) 7·80 (7·10–8·50) 8 8 Graham et al (2015) 2·30 (2·20–2·40) Graham et al (2015) 5·70 (5·10–6·20) SUD SUD 32 32 Rehm et al (2005) 8·40 (6·00–11·60) Rehm et al (2005) 17·20 (10·00–29·60) 18 18 Bargagli et al (2006) Bargagli et al (2006) 7·20 (6·10–8·40) 12·20 (8·70–17·20) 18 18 Bargagli et al (2006) 21·10 (19·80–22·50) Bargagli et al (2006) 53·70 (47·40–60·90) 18 18 Bargagli et al (2006) 7·90 (7·30–8·60) Bargagli et al (2006) 10·40 (8·90–12·10) 18 18 Bargagli et al (2006) 10·70 (8·80–13·10) Bargagli et al (2006) 11·40 (6·90–18·90) 18 18 Bargagli et al (2006) 6·30 (5·70–7·00) Bargagli et al (2006) 16·70 (13·20–21·00) 18 18 Bargagli et al (2006) Bargagli et al (2006) 12·20 (8·50–17·60) 15·80 (7·10–35·10) 18 18 Bargagli et al (2006) 13·60 (12·20–15·10) Bargagli et al (2006) 37·70 (30·20–47·10) 18 18 Bargagli et al (2006) 9·90 (8·50–11·60) Bargagli et al (2006) 10·20 (7·10–14·80) 35 35 Spittal et al (2006) 20·70 (17·20–24·20) Spittal et al (2006) 47·30 (36·10–58·50) 20 20 Bjornaas et al (2008) 23·40 (17·60–31·10) Bjornaas et al (2008) 24·20 (16·10–36·40) 40 40 Degenhardt et al (2009) Degenhardt et al (2009) 5·90 (5·70–6·10) 8·70 (8·10–9·20) 34 34 Singleton et al (2009) 5·87 (4·13–8·09) Singleton et al (2009) 7·84 (3·92–14·02) 9 9 Arendt et al (2011) 5·20 (3·00–9·00) Arendt et al (2011) 16·30 (6·80–39·20) 9 9 Arendt et al (2011) 6·00 (4·20–8·70) Arendt et al (2011) 5·80 (2·40–13·90) 9 9 Arendt et al (2011) 8·70 (8·00–9·40) Arendt et al (2011) 12·20 (10·30–14·40) 9 9 Arendt et al (2011) Arendt et al (2011) 7·30 (6·10–8·80) 8·70 (6·60–11·30) 21 21 Darke et al (2011) 4·56 (3·09–6·47) Darke et al (2011) 18·57 (9·89–31·52) 21 24 Darke et al (2011) 2·95 (1·75–4·66) Gibson et al (2011) 6·40 (5·40–7·50) 24 29 Gibson et al (2011) 4·00 (3·50–4·50) Nyhlen et al (2011) 4·20 (2·99–5·41) 29 25 Nyhlen et al (2011) 5·60 (4·80–6·50) Hser et al (2012) 8·40 (7·20–9·60) 38 23 Zabransky et al (2011) Evans et al (2012) 14·38 (7·19–28·75) 19·10 (5·20–48·80) 23 26 Evans et al (2012) 6·70 (3·10–12·70) Lee et al (2013) 11·80 (6·10–20·60) 26 31 Lee et al (2013) 5·20 (4·40–6·10) Pavarin et al (2013) 25·41 (11·42–56·56) 31 31 Pavarin et al (2013) 8·93 (6·21–12·85) Pavarin et al (2013) 25·46 (8·21–78·93) 31 31 Pavarin et al (2013) 12·43 (6·88–22·44) Pavarin et al (2013) 25·36 (8·18–78·63) 31 2 Pavarin et al (2013) Overall (I =94·1%, p<0·0001) 7·62 (4·80–12·09) 11·86 (10·42–13·30) Overall (I =99·1%, p<0·0001) 7·88 (7·03–8·74) 12 2 3 5 10 08 30 50 0 C Male and female individuals Favours general population SMR (95% CI) Homeless Roy et al (2010) 11·60 (7·60–17·00) Roy et al (2010) 3·00 (1·00–6·90) Vila-Rodriguez et al (2013) 4·83 (2·91–8·01) SUD Mathers et al (2013) 15·75 (11·40–21·20) Rehm et al (2005) 9·70 (7·30–12·80) Mathers et al (2013) 13·01 (12·11–13·91) Mathers et al (2013) 7·77 (6·70–8·95) Mathers et al (2013) 8·15 (7·28–9·09) Mathers et al (2013) 16·40 (9·10–27·10) Mathers et al (2013) 4·38 (3·99–4·78) Degenhardt et al (2011) 4·74 (4·19–5·29) Bjornaas et al (2008) 23·60 (18·70–29·90) Mathers et al (2008) 29·13 (19·27–44·04) Stoove et al (2008) 6·08 (4·14–8·93) Degenhardt et al (2009) 6·40 (6·20–6·60) Singleton et al (2009) 6·22 (4·59–8·25) Arendt et al (2011) 6·40 (3·90–10·00) Arendt et al (2011) 6·00 (4·20–8·30) Arendt et al (2011) 9·10 (8·50–9·80) Arendt et al (2011) 7·70 (6·60–8·90) Gibson et al (2011) 4·60 (4·20–5·00) Mathers et al (2013) 10·30 (8·90–12·00) Mathers et al (2013) 9·00 (8·00–10·00) Mathers et al (2013) 27·60 (24·90–30·70) Mathers et al (2013) 14·40 (9·31–19·49) Nyhlen et al (2011) 5·94 (5·50–6·80) Rosca et al (2012) 12·20 (11·40–13·00) Evans et al (2012) 8·30 (4·40–14·30) Merrall et al (2012) 6·40 (6·00–6·90) Merrall et al (2012) 4·80 (4·60–5·00) Degenhardt et al (2014) 6·50 (6·30–6·70) Barrio et al (2013) 4·70 (2·40–9·00) van Santen et al (2014) 13·90 (12·60–15·30) Overall (I =97·7% p<0·0001) 8·56 (7·78–9·35) 12 2 3 5 10 08 30 50 0 Favours general population 246 www.thelancet.com Vol 391 January 20, 2018 Articles [1·33–2·49]; I²=62·8%; appendix p 6), and both sexes Our study comprehensively describes for the first time, combined (2·20 [1·61–2·79]; I²=90·6%; appendix p 6). to our knowledge, the relative mortality and morbidity Only 44 studies reported cardiovascular outcomes, burden in selected inclusion health populations. We have accounting for 149 (5%) of 2835 datapoints extracted for reviewed the existing literature in this area using a this Article. Summary SMRs for diseases of the comprehensive search strategy to identify the balance of circulatory system were increased in male individuals evidence available to inform policy making around (2·44 [95% CI 1·48–3·41]; I²=94·5%; appendix p 6), inclusion health. Data were extracted and reviewed by a female individuals (3·13 [1·75–4·52]; I²=51·5%; appendix second author to reduce the likelihood of errors. Our p 6), and both sexes combined (2·91 [2·04–3·77]; approach enabled the identification of relative gaps in both I²=85·8%; appendix p 6). The prevalence of coronary categories of disease and inclusion health categories. Our artery disease was 13% (32 of 247 individuals). analysis was informed by an intersectionality perspective, Standardised mortality ratios for respiratory diseases which focuses on how social characteristics in combination 7,57 were only reported for populations with substance use affect health. We have therefore specifically investigated disorders and prison populations, ranging from 1·8 how the health consequences of exclusion might vary as a (95% CI 1·5–2·1) in male Scottish prisoners to 7·9 result of other socially influenced characteristics, with (5·1–11·8) in populations with substance use disorders differences between sexes being particularly noteworthy. in Australia. The prevalence of asthma ranged from However, several limitations should be considered. 5·0% (10 525 of 210 501 individuals; numerator esti­ Caution must be taken when interpreting the summary 54 55 mated) to 26% (nine of 35 individuals). Summary estimates because of the heterogeneity of studies. The SMRs for gastrointestinal conditions included only data absence of internationally agreed definitions of inclusion from prison populations and populations of individuals health groups is likely to explain some of this variation. with substance use disorders, and were higher in Similarly, comparison groups varied, with some studies female individuals (7·89 [95% CI 5·81–9·97]; I²=66·1%; using the general population and others using groups appendix p 6) than male individuals (3·37 [2·58–4·15]; living in socially deprived areas. Studies also varied I²=93·1%; appendix p 6). according to the extent of adjustment for social deprivation and other risk factors. We used a random­ Discussion effects method and noted the recommendations that The excess mortality associated with considerable social meta­ analyses should be pursued whenever possible, exclusion is extreme. We found all­ cause mortality acknowledging heterogeneity. We limited our search to SMRs of 7·9 in male individuals and 11·9 in female articles published from 2005 onwards and therefore we individuals. By comparison, mortality rates for individuals have not examined longer­ term trends. Furthermore, for aged 15–64 years in the most deprived areas of England pragmatic reasons, we were unable to investigate other and Wales are 2·8 times higher than those in the least health inclusion groups and believe that further work is deprived areas for male individuals and 2·1 times higher needed to describe their health experiences. for female individuals. The relative excesses were We found that the SMRs were consistently higher for greatest for injury, poisoning, and external causes, but female than male individuals. Because general popu­ extend across almost all health conditions and across the lation mortality rates are lower in female individuals inclusion health populations that we studied. than male individuals for most conditions, this result The available body of evidence is largest for infectious does not necessarily indicate that outcomes were worse diseases, with a substantial amount of existing research in female inclusion health groups than in male groups. on morbidity associated with mental and behavioural These results might reflect an increased vulnerability of disorders. By contrast, evidence on non­ communicable women in inclusion health populations or different risk diseases and injury, poisoning, and external causes is distributions among female individuals and male scarce despite these causes having the highest SMRs individuals in inclusion health groups. SMRs are a across ICD­ 10 categories in our study. SMRs across relative measure, and the lower (but still greater than 1) disease categories were consistently higher in female SMRs for more common diseases such as cardiovascular than male individuals. Of the four inclusion health disease and cancer than for other conditions might populations considered, sex workers were the least well underplay the number of excess cases of mortality that investigated, which should be addressed as a matter of occurred as a result of these conditions. Conversely, high priority in future research. SMRs might not indicate a large number of excess deaths if the condition is rare. Further work should report absolute as well as relative measures of mortality. Figure 3: Forest plots of SMRs for all-cause mortality Data are presented for male individuals (A), female individuals (B), and overall (C). These extreme inequities demand an intensive cross­ Weights were assigned by random effects analysis. Several studies contribute sectoral policy and service response to prevent exclusion multiple rows of data because different populations with substance use disorders and improve health outcomes. An accompanying Review, 9,31 18 were studied, because different countries were included, or because different published in The Lancet outlines interventions that time periods were studied. SMR=standardised mortality ratio. SUD=substance use disorder. respond to these increases in morbidity and mortality. www.thelancet.com Vol 391 January 20, 2018 247 Articles Determining the burden of disease remains challeng­ overlapping marginalised groups. These services should ing in inclusion health populations because membership address not only diseases with extreme disparities, but of such populations is not recorded in most vital also prevention and management of more common registration and health information systems. Deaths and conditions with a lower relative risk but high excess health service use in excluded populations are therefore mortality, such as cardiovascular disease. The ability of largely invisible and neglected aspects of routine health and social policy to address the needs of the most statistics. By contrast, the availability of area­ based marginalised populations should be a key indicator of measures of social deprivation across high­ income quality. Such initiatives need to be sup ported by countries has allowed the impact of less extreme social information systems that can provide data for continuing inequalities to be measured at the major population advocacy, guide service development, and monitor the level. The outcomes of these measure ments have sup­ health of marginalised populations over time. ported extensive cross­ sectoral policy initiatives to Our study highlights an extreme health inequity that address these inequities. Better routine data is also persists in high­ income countries. An inclusion health needed to drive the policy response to the inclusion policy response must build on the evidence regarding health agenda. who is at risk and the events that trigger exclusion to Two broad potential approaches are available to address highlight the social and economic benefits of sustained this problem. First, health services could routinely record action to prevent social exclusion. membership of health inclusion groups. This would Contributors RWA, ACH, and AS proposed the hypothesis and idea for the systematic require agreed definitions of each group. Individuals review with all authors contributing to its development and the analysis responsible for recording data would need guidance plan. RWA did the literature search and reviewed studies for inclusion. to help them ascertain membership and avoid re­ All authors extracted and checked the data. RWA and DL did all inforcement of stigma. The feasibility of this approach meta­ analyses and RWA wrote the first draft of the manuscript. All authors reviewed and interpreted the results and edited the manuscript. outside of specialist services remains unclear. Alternatively, and more feasibly in the short term, data Declaration of interests AS is the clinical lead and manager of the Find & Treat service (University linkage methods could be used to match data from College Hospitals NHS Trust). ACH is a trustee of Pathway, a charity for services that work with inclusion health groups, with homeless people. All other authors declare no competing interests. vital registration data, electronic health records, and Acknowledgments existing disease surveillance systems. Data linkage has RWA was supported by an academic clinical lectureship from the been the primary method used to estimate SMRs in the National Institute for Health Research (NIHR) and and a Wellcome Trust Clinical Research Career Development Fellowship (206602/Z/17/Z). studies reported in this Article. These linked datasets AS is funded by University College London Hospitals Foundation Trust. would facilitate systematic estimates of mortality and ACH’s salary is provided by Central and North West London National morbidity over time and help to measure the effect of Health Service (NHS) Community Trust. EJT and SVK are funded by the interventions. Medical Research Council (MC_UU_12017/13 & MC_UU_12017/15) and the Scottish Government Chief Scientist Office (SPHSU13 & SPHSU15). To inform the content of this Article and the accomp­ 1 SVK is also funded by a National Research Service (NHS) Research anying Review we held an engagement workshop with Scotland Scottish Senior Clinical Fellowship (SCAF/15/02). The views 16 people with experience of homelessness and social expressed are those of the authors and not necessarily those of exclusion. We asked this group about their views on The Wellcome Trust, NIHR, NHS, NHS Research Scotland, Medical Research Council, or the Scottish Government Chief collecting operational data with ethical and appropriate Scientist Office. research governance approvals, but without specific References individual level consent. Although this sample was only 1 Luchenski S, Maguire N, Aldridge RW, et al. What works in inclusion small (and we acknowledge that people who face exclusion health: overview of effective interventions for marginalised and excluded populations. Lancet 2017; published online Nov 11. and are willing to attend a workshop might differ from http://dx.doi.org/10.1016/S0140­ 6736(17)31959­ 1. those who do not), acceptability of collection of this sort of 2 Bramley G, Fitzpatrick S, Edwards J, et al. Hard edges: data was extremely high. 13 (100%) of 13 participants were mapping severe and multiple disadvantage in England. 2015. happy for homeless hostel records to be collected, eight http://lankellychase.org.uk/multiple­ disadvantage/publications/hard­ edges/ (accessed Dec 2, 2016). (73%) of 11 agreed to the collection of criminal records, 3 Fazel S, Bains P, Doll H. Substance abuse and dependence in eight (62%) of 13 to health records, and 11 (85%) of 13 to prisoners: a systematic review. Addiction 2006; 101: 181–91. these records being linked together. 4 Gilchrist G, Singleton N, Donmall M, Jones A. Prevalence and factors associated with sex trading in the year prior to entering treatment for A vertical approach to tackling inclusion health (ie, one drug misuse in England. Drug Alcohol Depend 2015; 152: 116–22. that focuses on specific diseases or specific risk groups) 5 Greenberg GA, Rosenheck RA. Homelessness in the state and federal can overlook multimorbidity and the social issues faced prison population. Crim Behav Ment Health 2008; 18: 88–103. by excluded populations. This approach can result in 6 Croxford S, Platt L, Hope VD, Cullen KJ, Parry JV, Ncube F. Sex work amongst people who inject drugs in England, Wales and Northern inefficiencies and missed opportunities for pre vention, Ireland: findings from a national survey of health harms and early diagnosis, and management, and missed behaviours. Int J Drug Policy 2015; 26: 429–33. opportunities for mitigation of social risk factors. The 7 Fitzpatrick S, Bramley G, Johnsen S. Pathways into multiple exclusion homelessness in seven UK cities. Urban Stud 2013; emerging field of inclusion health should advocate for 50: 148–68. and deliver joined up health and social services for 248 www.thelancet.com Vol 391 January 20, 2018 Articles 8 Graham L, Fischbacher CM, Stockton D, Fraser A, Fleming M, 30 Nyhlen A, Fridell M, Backstrom M, Hesse M, Krantz P. Greig K. Understanding extreme mortality among prisoners: Substance abuse and psychiatric co­ morbidity as predictors of a national cohort study in Scotland using data linkage. premature mortality in Swedish drug abusers a prospective Eur J Public Health 2015; 25: 879–85. longitudinal study 1970–2006. BMC Psychiatry 2011; 11: 122. 9 Arendt M, Munk­ Jørgensen P, Sher L, Jensen SO. Mortality among 31 Pavarin RM. Mortality risk for cocaine abusers in relation to individuals with cannabis, cocaine, amphetamine, MDMA, heroin use: a follow­ up study. Subst Use Misuse 2013; 48: 702–10. and opioid use disorders: a nationwide follow­ up study of Danish 32 Rehm J, Frick U, Hartwig C, Gutzwiller F, Gschwend P, substance users in treatment. Drug Alcohol Depend 2011; Uchtenhagen A. Mortality in heroin­ assisted treatment in 114: 134–39. Switzerland 1994–2000. Drug Alcohol Depend 2005; 79: 137–43. 10 Nielsen SF, Hjorthøj CR, Erlangsen A, Nordentoft M. 33 Rosca P, Haklai Z, Goldberger N, Zohar P, Margolis A, Psychiatric disorders and mortality among people in homeless Ponizovsky AM. Mortality and causes of death among users of shelters in Denmark: a nationwide register­ based cohort study. methadone maintenance treatment in Israel, 1999–2008. Lancet 2011; 377: 2205–14. Drug Alcohol Depend 2012; 125: 160–63. 11 Peace R. Social exclusion: a concept in need of definition? 34 Singleton J, Degenhardt L, Hall W, Zabransky T. Mortality among Soc Policy J NZ 2001; 16: 17–36. amphetamine users: a systematic review of cohort studies. 12 Fitzpatrick S, Bramley G, Johnse S. Multiple exclusion Drug Alcohol Depend 2009; 105: 1–8. homelessness in the UK: an overview of key findings. 2012. 35 Spittal PM, Hogg RS, Li K, et al. Drastic elevations in mortality http://mcnevaluation.co.uk/download/public/observatory/ among female injection drug users in a Canadian setting. resource­ type/research­ evaluation/2012%20Multiple%20 AIDS Care 2006; 18: 101–08. Exclusion%20Homelessness%20­ %20breifing%20paper%201.pdf 36 Stoové MA, Dietze PM, Aitken CK, Jolley D. Mortality among (accessed Dec 2, 2016). injecting drug users in Melbourne: a 16­ year follow­ up of the 13 The World Bank. World Bank country and lending groups. Victorian Injecting Cohort Study (VICS). Drug Alcohol Depend https://datahelpdesk.worldbank.org/knowledgebase/ 2008; 96: 281–85. articles/906519­ world­ bank­ country­ and­ lending­ groups (accessed 37 van Santen DK, van der Helm JJ, Grady BPX, et al. Nov 2, 2016). Temporal trends in mortality among people who use drugs 14 Higgins JPT, Thompson SG. Quantifying heterogeneity in a compared with the general Dutch population differ by hepatitis C meta­ analysis. Stat Med 2002; 21: 1539–58. virus and HIV infection status. AIDS 2014; 28: 2589–99. 15 Roy É, Haley N, Boudreau J­ F, Leclerc P, Boivin J­ F. The challenge 38 Zabransky T, Csemy L, Grohmannova K, Janikova B, Brenza J. of understanding mortality changes among street youth. Mortality of cohort of very young injecting drug users in Prague, J Urban Health 2010; 87: 95–101. 1996–2010. Cent Eur J Public Health 2011; 19: 152–57. 16 Vila­ Rodriguez F, Panenka WJ, Lang DJ, et al. The hotel study: 39 Degenhardt L, Singleton J, Calabria B, et al. Mortality among multimorbidity in a community sample living in marginal cocaine users: a systematic review of cohort studies. housing. Am J Psychiatry 2013; 170: 1413–22. Drug Alcohol Depend 2011; 113: 88–95. 17 Kariminia A, Butler TG, Corben SP, et al. Extreme cause­ specific 40 Degenhardt L, Randall D, Hall W, Law M, Butler T, Burns L. mortality in a cohort of adult prisoners—1988 to 2002: Mortality among clients of a state­ wide opioid pharmacotherapy a data­ linkage study. Int J Epidemiol 2007; 36: 310–16. program over 20 years: risk factors and lives saved. 18 Bargagli AM, Hickman M, Davoli M, et al. Drug­ related mortality Drug Alcohol Depend 2009; 105: 9–15. and its impact on adult mortality in eight European countries. 41 Gilles M, Swingler E, Craven C, Larson A. Prison health and Eur J Public Health 2006; 16: 198–202. public health responses at a regional prison in Western Australia. 19 Barrio G, Molist G, de la Fuente L, et al. Mortality in a cohort of Aust N Z J Public Health 2008; 32: 549–53. young primary cocaine users: controlling the effect of the riskiest 42 Des Jarlais DC, Perlis T, Arasteh K, et al. Reductions in hepatitis C drug­ use behaviors. Addict Behav 2013; 38: 1601–04. virus and HIV infections among injecting drug users in New York 20 Bjornaas MA, Bekken AS, Ojlert A, et al. A 20­ year prospective City, 1990–2001. AIDS 2005; 19 (suppl 3): S20–25. study of mortality and causes of death among hospitalized opioid 43 del Amo J, González C, Losana J, et al. Influence of age and addicts in Oslo. BMC Psychiatry 2008; 8: 8. geographical origin in the prevalence of high risk human 21 Darke S, Mills KL, Ross J, Teesson M. Rates and correlates of papillomavirus in migrant female sex workers in Spain. mortality amongst heroin users: findings from the Australian Sex Transm Infect 2005; 81: 79–84. Treatment Outcome Study (ATOS), 2001–2009. 44 Havens JR, Walker R, Leukefeld CG. Prevalence of opioid Drug Alcohol Depend 2011; 115: 190–95. analgesic injection among rural nonmedical opioid analgesic 22 Degenhardt L, Larney S, Randall D, Burns L, Hall W. Causes of users. Drug Alcohol Depend 2007; 87: 98–102. death in a cohort treated for opioid dependence between 1985 and 45 Bart G, Piccolo P, Zhang L, Jacobson I, Schaefer RA, Kreek MJ. 2005. Addiction 2014; 109: 90–99. Markers for hepatitis A, B and C in methadone maintained 23 Evans JL, Tsui JI, Hahn JA, Davidson PJ, Lum PJ, Page K. patients: an unexpectedly high co­ infection with silent hepatitis B. Mortality among young injection drug users in San Francisco: Addiction 2008; 103: 681–86. a 10­ year follow­ up of the UFO study. Am J Epidemiol 2012; 46 Carbonara S, Babudieri S, Longo B, et al. Correlates of 175: 302–08. Mycobacterium tuberculosis infection in a prison population. 24 Gibson A, Randall D, Degenhardt L. The increasing mortality Eur Respir J 2005; 25: 1070–76. burden of liver disease among opioid­ dependent people: cohort 47 Tabuchi T, Takatorige T, Hirayama Y, et al. Tuberculosis infection study. Addiction 2011; 106: 2186–92. among homeless persons and caregivers in a 25 Hser YI, Kagihara J, Huang D, Evans E, Messina N. Mortality high­ tuberculosis­ prevalence area in Japan: a cross­ sectional study. among substance­ using mothers in California: a 10­ year BMC Infect Dis 2011; 11: 22. prospective study. Addiction 2012; 107: 215–22. 48 Prinsloo B, Parr C, Fenton J. Mental illness among the homeless: 26 Lee CTC, Chen VCH, Tan HKL, et al. Suicide and other­ cause prevalence study in a Dublin homeless hostel. Ir J Psychol Med mortality among heroin users in Taiwan: a prospective study. 2012; 29: 22–26. Addict Behav 2013; 38: 2619–23. 49 Lubman DI, Allen NB, Rogers N, Cementon E, Bonomo Y. 27 Mathers BM, Degenhardt L, Bucello C, Lemon J, Wiessing L, The impact of co­ occurring mood and anxiety disorders among Hickman M. Mortality among people who inject drugs: a systematic substance­ abusing youth. J Affect Disord 2007; 103: 105–12. review and meta­ analysis. Bull World Health Organ 2013; 91: 102–23. 50 Baillargeon J, Penn JV, Thomas CR, Temple JR, Baillargeon G, 28 Merrall ELC, Bird SM, Hutchinson SJ. Mortality of those who Murray OJ. Psychiatric disorders and suicide in the nation’s largest attended drug services in Scotland 1996–2006: record­ linkage study. state prison system. J Am Acad Psychiatry Law 2009; 37: 188–93. Int J Drug Policy 2012; 23: 24–32. 51 Vergara­ Moragues E, González­ Saiz F, Lozano OM, et al. 29 Nyhlen A, Fridell M, Hesse M, Krantz P. Causes of premature Psychiatric comorbidity in cocaine users treated in therapeutic mortality in Swedish drug abusers: a prospective longitudinal study community: substance­ induced versus independent disorders. 1970–2006. J Forensic Leg Med 2011; 18: 66–72. Psychiatry Res 2012; 200: 734–41. www.thelancet.com Vol 391 January 20, 2018 249 Articles 52 Darke S, Williamson A, Ross J, Teesson M. Attempted suicide among 59 Marmot M. Fair society, healthy lives: the Marmot Review: strategic heroin users: 12­ month outcomes from the Australian Treatment review of health inequalities in England post­ 2010. 2010. Outcome Study (ATOS). Drug Alcohol Depend 2005; 78: 177–86. https://www.gov.uk/dfid ­ research­ outputs/fair­ society­ healthy­ lives­ the­ marmot­ review­ strategic­ review­ of­ health­ inequalities­ in­ 53 Brown RT, Kiely DK, Bharel M, Mitchell SL. Geriatric syndromes in england­ post­ 2010 (accessed Oct 22, 2017). older homeless adults. J Gen Intern Med 2012; 27: 16–22. 60 Katikireddi SV, Valles SA. Coupled ethical­ epistemic analysis of 54 Harzke AJ, Baillargeon JG, Pruitt SL, Pulvino JS, Paar DP, Kelley MF. public health research and practice: categorizing variables to Prevalence of chronic medical conditions among inmates in the Texas improve population health and equity. Am J Public Health 2015; prison system. J Urban Health 2010; 87: 486–503. 105: e36–42. 55 O’Brien KK, Schuttke A, Alhakeem A, et al. Health, perceived quality 61 Aldridge RW, Shaji K, Hayward AC, Abubakar I. Accuracy of of life and health services use among homeless illicit drug users. probabilistic linkage using the enhanced matching system for Drug Alcohol Depend 2015; 154: 139–45. public health and epidemiological studies. PloS One 2015; 56 Romeri E, Baker A, Griffiths C. Mortality by deprivation and cause of 10: e0136179. death in England and Wales, 1999–2003. Health Stat Q 2006; 32: 19–34. 62 Barnett K, Mercer SW, Norbury M, Watt G, Wyke S, Guthrie B. 57 Kapilashrami A, Hill S, Meer N. What can health inequalities Epidemiology of multimorbidity and implications for health care, researchers learn from an intersectionality perspective? research, and medical education: a cross­ sectional study. Lancet Understanding social dynamics with an inter­ categorical approach? 2012; 380: 37–43. Soc Theory Health 2015; 13: 288–307. 58 Ioannidis JPA, Patsopoulos NA, Rothstein HR. Reasons or excuses for avoiding meta­ analysis in forest plots. BMJ 2008; 336: 1413–15. 250 www.thelancet.com Vol 391 January 20, 2018 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png The Lancet Unpaywall

Morbidity and mortality in homeless individuals, prisoners, sex workers, and individuals with substance use disorders in high-income countries: a systematic review and meta-analysis

Download PDF
10 pages

Loading next page...
 
/lp/unpaywall/morbidity-and-mortality-in-homeless-individuals-prisoners-sex-workers-cgR0SFloAI

References (62)

Publisher
Unpaywall
ISSN
0140-6736
DOI
10.1016/s0140-6736(17)31869-x
Publisher site
See Article on Publisher Site

Abstract

Articles Morbidity and mortality in homeless individuals, prisoners, sex workers, and individuals with substance use disorders in high-income countries: a systematic review and meta-analysis Robert W Aldridge, Alistair Story, Stephen W Hwang, Merete Nordentoft, Serena A Luchenski, Greg Hartwell, Emily J Tweed, Dan Lewer, Srinivasa Vittal Katikireddi, Andrew C Hayward Summary Lancet 2018; 391: 241–50 Background Inclusion health focuses on people in extremely poor health due to poverty, marginalisation, and multimorbidity. We aimed to review morbidity and mortality data on four overlapping populations who experience Published Online November 11, 2017 considerable social exclusion: homeless populations, individuals with substance use disorders, sex workers, and http://dx.doi.org/10.1016/ imprisoned individuals. S0140-6736(17)31869-X See Comment page 186 Methods For this systematic review and meta-analysis, we searched MEDLINE, Embase, and the Cochrane Library See Review page 266 for studies published between Jan 1, 2005, and Oct 1, 2015. We included only systematic reviews, meta-analyses, Centre for Public Health Data interventional studies, and observational studies that had morbidity and mortality outcomes, were published in Science, Institute of Health English, from high-income countries, and were done in populations with a history of homelessness, imprisonment, Informatics (R W Aldridge PhD, A Story PhD, S A Luchenski FFPH, sex work, or substance use disorder (excluding cannabis and alcohol use). Studies with only perinatal outcomes and D Lewer MSc, studies of individuals with a specific health condition or those recruited from intensive care or high dependency Prof A C Hayward MD), The Farr hospital units were excluded. We screened studies using systematic review software and extracted data from Institute of Health Informatics published reports. Primary outcomes were measures of morbidity (prevalence or incidence) and mortality Research (R W Aldridge, A Story, S A Luchenski, D Lewer, (standardised mortality ratios [SMRs] and mortality rates). Summary estimates were calculated using a random Prof A C Hayward), and the effects model. Institute of Epidemiology and Health Care (Prof A C Hayward), Findings Our search identified 7946 articles, of which 337 studies were included for analysis. All-cause standardised University College London, London, UK; University College mortality ratios were significantly increased in 91 (99%) of 92 extracted datapoints and were 11·86 (95% CI London NHS Foundation Trust, 10·42–13·30; I²=94·1%) in female individuals and 7·88 (7·03–8·74; I²=99·1%) in men. Summary SMR estimates for London, UK (A Story); Centre for the International Classification of Diseases disease categories with two or more included datapoints were highest for Urban Health Solutions, Li Ka deaths due to injury, poisoning, and other external causes, in both men (7·89; 95% CI 6·40–9·37; I²=98·1%) and Shing Knowledge Institute, St Michael’s Hospital, women (18·72; 13·73–23·71; I²=91·5%). Disease prevalence was consistently raised across the following categories: Toronto, ON, Canada infections (eg, highest reported was 90% for hepatitis C, 67 [65%] of 103 individuals for hepatitis B, and 133 [51%] of (Prof S W Hwang MD); Mental 263 individuals for latent tuberculosis infection), mental health (eg, highest reported was 9 [4%] of 227 individuals for Health Centre Copenhagen and schizophrenia), cardiovascular conditions (eg, highest reported was 32 [13%] of 247 individuals for coronary heart Institute of Clinical Medicine, Faculty of Health and Medical disease), and respiratory conditions (eg, highest reported was 9 [26%] of 35 individuals for asthma). Sciences, University of Copenhagen, Denmark Interpretation Our study shows that homeless populations, individuals with substance use disorders, sex workers, (Prof M Nordentoft DMSc); and imprisoned individuals experience extreme health inequities across a wide range of health conditions, with the Department of Social and Environmental Health relative effect of exclusion being greater in female individuals than male individuals. The high heterogeneity between Research, London School of studies should be explored further using improved data collection in population subgroups. The extreme health Hygiene & Tropical Medicine, inequity identified demands intensive cross-sectoral policy and service action to prevent exclusion and improve health London, UK (G Hartwell MFPH); outcomes in individuals who are already marginalised. and Medical Research Council/ Scottish Government Chief Scientist Office Social and Funding Wellcome Trust, National Institute for Health Research, NHS England, NHS Research Scotland Scottish Public Health Sciences Unit, Senior Clinical Fellowship, Medical Research Council, Chief Scientist Office, and the Central and North West London University of Glasgow, NHS Trust. Glasgow, UK (E J Tweed MFPH, S Vittal Katikireddi PhD) Correspondence to: Copyright © The Author(s). Published by Elsevier Ltd. This is an Open Access article under the CC BY 4.0 license. Dr Robert W Aldridge, The Farr Institute of Health Informatics Introduction observed social gradients in health do not capture the full Research, University College Inclusion health is a research, service, and policy agenda extent of health inequities for individuals who experience London, London NW1 2DA, UK [email protected] that aims to prevent and redress health and social considerable social exclusion. inequities among people in extremely poor health due to Previous research has described the high prevalence of 1 2 poverty, marginalisation, and multimorbidity. The substance use disorders in homeless populations, prison­ 3 4 association between socio economic status and health ers, and sex workers, and the increased prevalence 5 6 outcomes is well established. However, these commonly of homelessness in prisoners and sex workers. These www.thelancet.com Vol 391 January 20, 2018 241 Articles Research in context Evidence before this study mortality outcomes across a range of inclusion health A comprehensive body of research exists on the health effect of populations. We found that the extent of the health inequity inequity, much of which focuses on disparities in morbidity seen in our inclusion health populations greatly exceeded and mortality, and is based on common measures of that previously observed between populations with high and socioeconomic status, such as neighbourhood deprivation and low socioeconomic status and was consistent across inclusion occupational class. A consistent association has been found health populations. Mortality rates are extremely high across between ill health and increasing levels of social deprivation, the International Classification of Diseases, tenth revision which has underpinned a broad range of social policies and disease categories in inclusion health populations, and our public health initiatives. Such analyses cannot adequately review is the first to show that relative risks are consistently assess the extent of health inequity faced by individuals who higher in female than male individuals. experience considerable social exclusion. In preparation for this Implications of all the available evidence Review, we searched the Cochrane Library, MEDLINE, and The extreme burden of disease experienced by inclusion health Embase databases for articles published between Jan 1, 2005, populations demands a cross-sectoral response to prevent and Sept 30, 2013. We searched for systematic reviews, meta- considerable social exclusion and an improvement in services analyses, cohort studies, and cross-sectional studies containing that work with these populations. Our analyses focused on morbidity and mortality outcomes for the four inclusion health relative measures of mortality and therefore future work should populations of interest (substance use disorders, homeless examine absolute measures in greater detail. Inclusion health populations, prisoners, and sex workers). We only included populations are often invisible within routine health data. full-text articles published in English. Full search terms are This limitation can be addressed by modifying the instruments listed in the appendix. The studies identified described the used to collect such data or through data linkage studies. highly overlapping nature of inclusion health populations, Services that provide for inclusion health populations should the increased risk factors for disease, and poor mortality aim to deliver health and social services for overlapping outcomes compared with the general population. Previous marginalised groups to tackle the poor health outcomes found systematic reviews have analysed health outcomes of in this study. These services should also have a greater focus on individual inclusion health populations, but none have prevention and management of more common conditions in examined the populations together. addition to those traditionally considered high risk for inclusion Added value of this study health groups. Our systematic review and meta-analysis provides the first comprehensive examination to date of morbidity and marginalised populations have common inter secting Methods characteristics and adverse life experiences that lead to Search strategy and selection criteria considerable social exclusion, making them powerful For this systematic review and meta­ analysis, we determinants of marginalisation in high­ income settings. searched the Cochrane Library, MEDLINE, and Embase When considered separately, marginalised popu lations for articles published between Jan 1, 2005, and Oct 1, 2015. 8–10 See Online for appendix have been shown to have high all­ cause mortality. Full search terms are provided in the appendix. We However, despite the considerable overlap in risk factors searched for articles about the populations of interest and the substantially increased mortality observed in (homeless individuals, prisoners, sex workers, and these populations, no previous review has examined the individuals with substance use disorders, excluding outcomes of these groups together. cannabis and alcohol use) from systematic reviews, No universally agreed theoretical framework exists to meta­ analyses, inter ventional studies, and observational describe inclusion health. In this Article, we build on studies that had morbidity and mortality outcomes. We existing social exclusion theory and consider the so­ called included studies identified from references of included linked and cumulative factors and processes that confound articles. We only included full­ text articles published in individual and group capacity for hope, opportunity, English that were done in high­ income countries 11 13 reciprocity, and participation. Our analysis is also (classified according to the World Bank classification ). informed by an intersectionality perspective, which We excluded studies with only perinatal outcomes and focuses on how social characteristics combine to have an did not include data on perinatal outcomes from studies 2,12 effect on health. that otherwise met our inclusion criteria. We excluded Our systematic review therefore aims to examine articles that limited the study population to individuals mortality and morbidity in homeless populations, with a specific health condition and studies that recruited prisoners, sex workers, and individuals with substance participants exclusively from intensive care or high use disorders, who experience considerable exclusion. dependency hospital units. 242 www.thelancet.com Vol 391 January 20, 2018 Articles We recognise that social exclusion has a major effect on stratifying the analyses by country and by inclusion health in other social groups, including Gypsies and health population group. Travellers, migrants, ethnic minorities, indigenous com­ munities, and sexual and gender minorities. Although Role of the funding source these groups experience social exclusion in many high­ The funders of the study had no role in study design, income settings, they were considered beyond the scope data collection, data analysis, data interpretation, writing of this systematic review. of the report, or the decision to submit the paper for publication. All authors had full access to all the data in Data analysis the study and had final responsibility for the decision to RWA screened titles, abstracts, and full texts using submit for publication. Covidence systematic review software. All authors con­ tributed to data extraction, and data were double­ checked Results by a second researcher (RWA, EJT, GH, or SVK). We identified 7946 articles, of which 1274 were duplicates Extracted items included study design, year or years of (figure 1). Of the 711 full ­ text articles retrieved, 418 met study, country, number of participants, primary outcomes, the inclusion criteria. We excluded a further 81 articles and summary descriptions of the study popu lation. We because of overlapping data. A total of 337 studies were tried to contact authors if we were unable to locate papers included in this Article, which included 2835 datapoints or required additional information about the data or study. (ie, effect estimates for a unique population) af ter the We attempted to identify and exclude duplicate data removal of 384 duplicates. from research studies presented in separate publications. The studies were from 38 countries (appendix). The For cases in which we identified multiple studies with USA contributed 698 datapoints, Australia contributed duplicated or overlapping data (by population, time, place, 460, Sweden contributed 309, Canada contributed 257, and and outcome) we selected the study with the largest or the UK contributed 234. Populations with substance use most representative sample size, and when these were disorders were the most studied subgroup, accounting for also similar, we present the most recent study. We followed 1193 (42·1%) of 2835 datapoints, followed by prisoners the PRISMA reporting guidelines in the presentation of (769 [27·1%]), homeless populations (754 [26·6%]), and sex our manuscript. A review protocol was not published workers (119 [4·2%]). before this review was done. Infectious diseases and mental and behavioural disorders Outcomes included were measures of morbidity and were the two most studied ICD­ 10 categories with infectious mortality for conditions defined in the International diseases accounting for 898 (31·6%) of 2835 datapoints, Classification of Diseases, tenth revision (ICD­ 10). Out­ and mental and behavioural disorders accounting for comes were reported using a variety of measures. To ensure maximum comparability across studies for mor­ tality outcomes, we extracted, in order of pref erence, the 7946 potentially eligible studies identified by database search first of the following measures: standardised mortality ratio (SMR), hazard ratio, mortality rate ratio, or crude 1274 duplicates excluded mortality rate. For consistency with most studies included in this Article, we have not multiplied SMRs by 100. In our results, a value of 1 equates to no difference between the 6672 identified for screening expected and observed mortality rate. For morbidity outcomes, we extracted, in order of preference, the first of 5961 excluded on title and abstract the following measures: prevalence, incidence, prevalence risk ratio, incidence rate ratio, prevalence odds ratio, or incidence odds ratio. When available, we used data in 711 full-text articles assessed for eligibility which the comparison group was a socially deprived population or measures were adjusted for area­ based or 293 excluded after full-text screening income­ based deprivation. A link to all extracted data is included in the appendix. For the quantitative findings analysed in this study, we 418 met the inclusion criteria focused the synthesis on SMRs. SMRs for all­ cause mortality and by ICD­ 10 disease category were summar­ 81 excluded because of overlapping ised in forest plots. We anticipated high levels of data with other studies heterogeneity, and therefore did summary estimates with random effects models using Stata version 13. We used 337 included in quantitative synthesis the I² statistic to indicate the proportion of total variation in study estimates due to heterogeneity. Figure 1: Study selection We explored potential sources of hetero geneity by www.thelancet.com Vol 391 January 20, 2018 243 Articles External causes Circulatory Infectious and parasitic diseases Injury and poisoning Neoplasms Digestive Genitourinary Skin Respiratory Mental and behavioural disorders Nervous system Endocrine Musculoskeletal 1 Symptoms, signs, and abnormal clinical and laboratory findings 2 Diseases of the blood SMR 3 Diseases of the eye and adnexa 0 510152025 4 Diseases of the ear and mastoid process Figure 2: Treemap summarising the amount of available data grouped according to the ICD-10 disease categories and summary estimates of SMRs Box sizes indicate the total number of datapoints included in this Article. SMRs used are summary estimates for the ICD-10 disease categories for both sexes combined. Grey boxes (SMR of 0) indicate that none of the studies included in this Article reported SMR for both sexes combined. ICD-10=International Classification of Diseases, tenth revision. SMR=standardised mortality ratio. 715 (25·2%) datapoints (figure 2, appendix p 4). Injury and ICD­ 10 categories (appendix pp 6–7). In some ICD­ 10 poisoning only accounted for 98 (3·4%) of all extracted categories, the summary SMRs for both sexes combined datapoints. did not fall between the male and female estimates Our all­ cause meta­ analyses focused on SMRs because the meta­ analyses used data from different 8–10,15–40 and included 29 studies, which contributed studies (rather than the estimate for both sexes combined 92 datapoints (table, figure 3, appendix). 91 (99%) of the being drawn from the male and female populations). 92 all­ cause SMRs were increased and overall we estimated We identified 201 papers reporting outcomes for that summary all­ cause SMRs were higher in female infectious and parasitic diseases. Summary estimates of individuals (11·86 [95% CI 10·42–13·30]; I²=94·1%; SMRs for infectious diseases were increased in male figure 3) than male individuals (7·88 [7·03–8·74]; individuals (2·83 [95% CI 1·61–4·05]; I²=65·4%; I²=99·1%; figure 3). We provide summary estimates of appendix p 6) and female individuals (5·58 [1·46–9·70]; SMRs; however, the I² statistic indicated that data were I²=60·0%; appendix p 6) and both sexes combined heterogeneous in many of our analyses and therefore (11·43 [6·91–15·94; I²=97·0%; appendix p 6). Disease these summary measures must be interpreted with prevalence was high but heterogeneous and ranged from 41 42 43 appropriate caution. Heterogeneity was not substantially 0% to 54% for HIV infection, from less than 0·1% to 42 44 reduced when analyses were stratified by population 90% for hepatitis C, from 2% (two of 119) to 65% 45 46 subgroup (appendix). Insufficient data were available to (67 of 103) for hepatitis B, and from 1% (one of 82) to do subgroup analyses by country. 51% (133 of 263) for latent tuberculosis infection. Summary SMRs were higher in female individuals Summary estimates of SMRs for injury, poisoning, and than in male individuals for mortality in each of the other external causes were the highest across all of the 244 www.thelancet.com Vol 391 January 20, 2018 Articles Study years Country Participants (n) Population description Homeless people Nielsen et al 1999–2009 Denmark 32 711 Women aged 16 years or older with at least one contact with a homeless shelter Roy et al 1995–2001 Canada 829 Individuals aged 14–25 years with unstable housing Vila-Rodriguez et al 2008–11 Canada 293 Prospective community sample of adults living in single-room occupancy hotel Prisoners Graham et al 1996–2007 UK 76 627 Male individuals imprisoned for the first time betw een 1996 and 2007 Kariminia et al 1988–2002 Australia 85 203 All adults who had been in full-time custody Individuals with substance use disorders Arendt et al 1996–2006 Denmark 20 581 People receiving treatment in specialist institutions for substance use disorder, who reported cocaine as their primary substance Bargagli et al 1996–2002 Netherlands 2575 Male opiate users aged 15–69 years entering treatment Barrio et al 2004–06 Spain 714 Regular cocaine users recruited from drug scenes and non-treatment settings Bjornaas et al 1980–2000 Norway 185 Individuals with opioid addiction admitted to hospital because of self-poisoning Darke et al 2001–09 Australia 615 Opioid users Degenhardt et al 1985–2005 Australia 43 789 People who are opioid-dependent treated with opioid substitution therapy Evans et al 2005–07 USA 644 Injecting drug users younger than 30 years Gibson et al 1980–2006 Australia 2489 Opioid users Hser et al 2000–02 USA 4447 Women who were admitted to drug treatment programmes Lee et al 2006–08 Taiwan 10 842 Heroin users attending opioid substitution therapy Mathers et al 1980–99 Denmark 101 People who injected opioids and other drugs Merrall et al 1996–2006 UK 69 456 People in contact with drug treatment services 29,30 Nyhlen et al 1970–2006 Sweden 561 Substance abusers admitted for inpatient detoxification Pavarin et al 1988–2012 Italy 471 Individuals who had visited a public treatment centre for problems due to cocaine use Rehm et al 1994–2000 Switzerland 6281 Participants in heroin-assisted treatment Rosca et al 1999–2008 Israel 9818 Patients who had ever been treated or were currently in treatment in methadone maintenance treatment clinics Singleton et al 1997–2002 Czech 3039 Drug users admitted to hospital for drug-related problems Republic Spittal et al 1996–2002 Canada 520 Injecting drug users recruited through self-referral and street outreach Stoove et al 1990–2006 Australia 220 Injecting drug users recruited from the community van Santen et al 1985–2012 Netherlands 1254 Individuals recruited from local methadone outposts, a sexually transmitted diseases clinic, and by word of mouth Zabransky et al 1996–2008 Czech 151 Injecting drug users aged 15–18 years Republic Degenhardt et al 1996–2004 Canada 717 People who injected cocaine daily Degenhardt et al 1985–2006 Australia 42 676 Opioid users Table: Studies included in the standardised all-cause mortality ratio meta-analyses ICD­ 10 categories, in male individuals (7·89 [95% CI were from populations with substance use disorders 6·40–9·37]; I²=98·1%; appendix p 7), female individuals only. Only two studies included data on male individuals, (18·72 [13·73–23·71]; I²=91·5%; appendix p 7), and both one study on female individuals, and two studies on both sexes combined (23·53 [15·34–31·71]; I²=99·6%; appendix sexes combined. Prevalence of major depression in p 7). However, these categories only accounted for 98 (3%) inclusion health populations ranged from 3% (one of of 2835 extracted datapoints. Summary SMR estimates 38 individuals) in the month before assessment to were also increased for external causes of morbidity and a 53% (25 of 47 individuals) lifetime prevalence. mortality in male individuals (6·52 [95% CI 5·54–7·51; Prevalence of schizophrenia ranged from 0·9% (212 of I²=97·4%; appendix p 7), female individuals (13·15 23 530 individuals; we estimated the numerator on the [9·87–16·43]; I²=93·7%; appendix p 7), and both sexes basis of data in the original article) to 4% (nine of 51 49 combined (8·50 [6·89–10·10]; I²=97·5%; appendix p 7). No 227 individuals), and from 0% (none of 53 individuals) data from studies that included sex workers were used in to 45% (221 of 495 individuals; numerator estimated) any of the SMR estimates for injuries or external causes. for bipolar disorder. SMRs for mental and behavioural disorders for male Summary estimates of SMRs for neoplasms were individuals and female individuals were exclusively from increased in male individuals (1·61 [95% CI 1·30–1·92]; prison populations and data for both sexes combined I²=88·7%; appendix p 6), female individuals (1·91 www.thelancet.com Vol 391 January 20, 2018 245 Articles A Male individuals B Female individuals SMR (95% CI) SMR (95% CI) Homeless Homeless 10 10 Nielsen et al (2011) 5·60 (5·40–5·80) Nielsen et al (2011) 6·70 (6·20–7·10) Prisoners Prisoners 17 17 Kariminia et al (2007) 3·70 (3·60–3·80) Kariminia et al (2007) 7·80 (7·10–8·50) 8 8 Graham et al (2015) 2·30 (2·20–2·40) Graham et al (2015) 5·70 (5·10–6·20) SUD SUD 32 32 Rehm et al (2005) 8·40 (6·00–11·60) Rehm et al (2005) 17·20 (10·00–29·60) 18 18 Bargagli et al (2006) Bargagli et al (2006) 7·20 (6·10–8·40) 12·20 (8·70–17·20) 18 18 Bargagli et al (2006) 21·10 (19·80–22·50) Bargagli et al (2006) 53·70 (47·40–60·90) 18 18 Bargagli et al (2006) 7·90 (7·30–8·60) Bargagli et al (2006) 10·40 (8·90–12·10) 18 18 Bargagli et al (2006) 10·70 (8·80–13·10) Bargagli et al (2006) 11·40 (6·90–18·90) 18 18 Bargagli et al (2006) 6·30 (5·70–7·00) Bargagli et al (2006) 16·70 (13·20–21·00) 18 18 Bargagli et al (2006) Bargagli et al (2006) 12·20 (8·50–17·60) 15·80 (7·10–35·10) 18 18 Bargagli et al (2006) 13·60 (12·20–15·10) Bargagli et al (2006) 37·70 (30·20–47·10) 18 18 Bargagli et al (2006) 9·90 (8·50–11·60) Bargagli et al (2006) 10·20 (7·10–14·80) 35 35 Spittal et al (2006) 20·70 (17·20–24·20) Spittal et al (2006) 47·30 (36·10–58·50) 20 20 Bjornaas et al (2008) 23·40 (17·60–31·10) Bjornaas et al (2008) 24·20 (16·10–36·40) 40 40 Degenhardt et al (2009) Degenhardt et al (2009) 5·90 (5·70–6·10) 8·70 (8·10–9·20) 34 34 Singleton et al (2009) 5·87 (4·13–8·09) Singleton et al (2009) 7·84 (3·92–14·02) 9 9 Arendt et al (2011) 5·20 (3·00–9·00) Arendt et al (2011) 16·30 (6·80–39·20) 9 9 Arendt et al (2011) 6·00 (4·20–8·70) Arendt et al (2011) 5·80 (2·40–13·90) 9 9 Arendt et al (2011) 8·70 (8·00–9·40) Arendt et al (2011) 12·20 (10·30–14·40) 9 9 Arendt et al (2011) Arendt et al (2011) 7·30 (6·10–8·80) 8·70 (6·60–11·30) 21 21 Darke et al (2011) 4·56 (3·09–6·47) Darke et al (2011) 18·57 (9·89–31·52) 21 24 Darke et al (2011) 2·95 (1·75–4·66) Gibson et al (2011) 6·40 (5·40–7·50) 24 29 Gibson et al (2011) 4·00 (3·50–4·50) Nyhlen et al (2011) 4·20 (2·99–5·41) 29 25 Nyhlen et al (2011) 5·60 (4·80–6·50) Hser et al (2012) 8·40 (7·20–9·60) 38 23 Zabransky et al (2011) Evans et al (2012) 14·38 (7·19–28·75) 19·10 (5·20–48·80) 23 26 Evans et al (2012) 6·70 (3·10–12·70) Lee et al (2013) 11·80 (6·10–20·60) 26 31 Lee et al (2013) 5·20 (4·40–6·10) Pavarin et al (2013) 25·41 (11·42–56·56) 31 31 Pavarin et al (2013) 8·93 (6·21–12·85) Pavarin et al (2013) 25·46 (8·21–78·93) 31 31 Pavarin et al (2013) 12·43 (6·88–22·44) Pavarin et al (2013) 25·36 (8·18–78·63) 31 2 Pavarin et al (2013) Overall (I =94·1%, p<0·0001) 7·62 (4·80–12·09) 11·86 (10·42–13·30) Overall (I =99·1%, p<0·0001) 7·88 (7·03–8·74) 12 2 3 5 10 08 30 50 0 C Male and female individuals Favours general population SMR (95% CI) Homeless Roy et al (2010) 11·60 (7·60–17·00) Roy et al (2010) 3·00 (1·00–6·90) Vila-Rodriguez et al (2013) 4·83 (2·91–8·01) SUD Mathers et al (2013) 15·75 (11·40–21·20) Rehm et al (2005) 9·70 (7·30–12·80) Mathers et al (2013) 13·01 (12·11–13·91) Mathers et al (2013) 7·77 (6·70–8·95) Mathers et al (2013) 8·15 (7·28–9·09) Mathers et al (2013) 16·40 (9·10–27·10) Mathers et al (2013) 4·38 (3·99–4·78) Degenhardt et al (2011) 4·74 (4·19–5·29) Bjornaas et al (2008) 23·60 (18·70–29·90) Mathers et al (2008) 29·13 (19·27–44·04) Stoove et al (2008) 6·08 (4·14–8·93) Degenhardt et al (2009) 6·40 (6·20–6·60) Singleton et al (2009) 6·22 (4·59–8·25) Arendt et al (2011) 6·40 (3·90–10·00) Arendt et al (2011) 6·00 (4·20–8·30) Arendt et al (2011) 9·10 (8·50–9·80) Arendt et al (2011) 7·70 (6·60–8·90) Gibson et al (2011) 4·60 (4·20–5·00) Mathers et al (2013) 10·30 (8·90–12·00) Mathers et al (2013) 9·00 (8·00–10·00) Mathers et al (2013) 27·60 (24·90–30·70) Mathers et al (2013) 14·40 (9·31–19·49) Nyhlen et al (2011) 5·94 (5·50–6·80) Rosca et al (2012) 12·20 (11·40–13·00) Evans et al (2012) 8·30 (4·40–14·30) Merrall et al (2012) 6·40 (6·00–6·90) Merrall et al (2012) 4·80 (4·60–5·00) Degenhardt et al (2014) 6·50 (6·30–6·70) Barrio et al (2013) 4·70 (2·40–9·00) van Santen et al (2014) 13·90 (12·60–15·30) Overall (I =97·7% p<0·0001) 8·56 (7·78–9·35) 12 2 3 5 10 08 30 50 0 Favours general population 246 www.thelancet.com Vol 391 January 20, 2018 Articles [1·33–2·49]; I²=62·8%; appendix p 6), and both sexes Our study comprehensively describes for the first time, combined (2·20 [1·61–2·79]; I²=90·6%; appendix p 6). to our knowledge, the relative mortality and morbidity Only 44 studies reported cardiovascular outcomes, burden in selected inclusion health populations. We have accounting for 149 (5%) of 2835 datapoints extracted for reviewed the existing literature in this area using a this Article. Summary SMRs for diseases of the comprehensive search strategy to identify the balance of circulatory system were increased in male individuals evidence available to inform policy making around (2·44 [95% CI 1·48–3·41]; I²=94·5%; appendix p 6), inclusion health. Data were extracted and reviewed by a female individuals (3·13 [1·75–4·52]; I²=51·5%; appendix second author to reduce the likelihood of errors. Our p 6), and both sexes combined (2·91 [2·04–3·77]; approach enabled the identification of relative gaps in both I²=85·8%; appendix p 6). The prevalence of coronary categories of disease and inclusion health categories. Our artery disease was 13% (32 of 247 individuals). analysis was informed by an intersectionality perspective, Standardised mortality ratios for respiratory diseases which focuses on how social characteristics in combination 7,57 were only reported for populations with substance use affect health. We have therefore specifically investigated disorders and prison populations, ranging from 1·8 how the health consequences of exclusion might vary as a (95% CI 1·5–2·1) in male Scottish prisoners to 7·9 result of other socially influenced characteristics, with (5·1–11·8) in populations with substance use disorders differences between sexes being particularly noteworthy. in Australia. The prevalence of asthma ranged from However, several limitations should be considered. 5·0% (10 525 of 210 501 individuals; numerator esti­ Caution must be taken when interpreting the summary 54 55 mated) to 26% (nine of 35 individuals). Summary estimates because of the heterogeneity of studies. The SMRs for gastrointestinal conditions included only data absence of internationally agreed definitions of inclusion from prison populations and populations of individuals health groups is likely to explain some of this variation. with substance use disorders, and were higher in Similarly, comparison groups varied, with some studies female individuals (7·89 [95% CI 5·81–9·97]; I²=66·1%; using the general population and others using groups appendix p 6) than male individuals (3·37 [2·58–4·15]; living in socially deprived areas. Studies also varied I²=93·1%; appendix p 6). according to the extent of adjustment for social deprivation and other risk factors. We used a random­ Discussion effects method and noted the recommendations that The excess mortality associated with considerable social meta­ analyses should be pursued whenever possible, exclusion is extreme. We found all­ cause mortality acknowledging heterogeneity. We limited our search to SMRs of 7·9 in male individuals and 11·9 in female articles published from 2005 onwards and therefore we individuals. By comparison, mortality rates for individuals have not examined longer­ term trends. Furthermore, for aged 15–64 years in the most deprived areas of England pragmatic reasons, we were unable to investigate other and Wales are 2·8 times higher than those in the least health inclusion groups and believe that further work is deprived areas for male individuals and 2·1 times higher needed to describe their health experiences. for female individuals. The relative excesses were We found that the SMRs were consistently higher for greatest for injury, poisoning, and external causes, but female than male individuals. Because general popu­ extend across almost all health conditions and across the lation mortality rates are lower in female individuals inclusion health populations that we studied. than male individuals for most conditions, this result The available body of evidence is largest for infectious does not necessarily indicate that outcomes were worse diseases, with a substantial amount of existing research in female inclusion health groups than in male groups. on morbidity associated with mental and behavioural These results might reflect an increased vulnerability of disorders. By contrast, evidence on non­ communicable women in inclusion health populations or different risk diseases and injury, poisoning, and external causes is distributions among female individuals and male scarce despite these causes having the highest SMRs individuals in inclusion health groups. SMRs are a across ICD­ 10 categories in our study. SMRs across relative measure, and the lower (but still greater than 1) disease categories were consistently higher in female SMRs for more common diseases such as cardiovascular than male individuals. Of the four inclusion health disease and cancer than for other conditions might populations considered, sex workers were the least well underplay the number of excess cases of mortality that investigated, which should be addressed as a matter of occurred as a result of these conditions. Conversely, high priority in future research. SMRs might not indicate a large number of excess deaths if the condition is rare. Further work should report absolute as well as relative measures of mortality. Figure 3: Forest plots of SMRs for all-cause mortality Data are presented for male individuals (A), female individuals (B), and overall (C). These extreme inequities demand an intensive cross­ Weights were assigned by random effects analysis. Several studies contribute sectoral policy and service response to prevent exclusion multiple rows of data because different populations with substance use disorders and improve health outcomes. An accompanying Review, 9,31 18 were studied, because different countries were included, or because different published in The Lancet outlines interventions that time periods were studied. SMR=standardised mortality ratio. SUD=substance use disorder. respond to these increases in morbidity and mortality. www.thelancet.com Vol 391 January 20, 2018 247 Articles Determining the burden of disease remains challeng­ overlapping marginalised groups. These services should ing in inclusion health populations because membership address not only diseases with extreme disparities, but of such populations is not recorded in most vital also prevention and management of more common registration and health information systems. Deaths and conditions with a lower relative risk but high excess health service use in excluded populations are therefore mortality, such as cardiovascular disease. The ability of largely invisible and neglected aspects of routine health and social policy to address the needs of the most statistics. By contrast, the availability of area­ based marginalised populations should be a key indicator of measures of social deprivation across high­ income quality. Such initiatives need to be sup ported by countries has allowed the impact of less extreme social information systems that can provide data for continuing inequalities to be measured at the major population advocacy, guide service development, and monitor the level. The outcomes of these measure ments have sup­ health of marginalised populations over time. ported extensive cross­ sectoral policy initiatives to Our study highlights an extreme health inequity that address these inequities. Better routine data is also persists in high­ income countries. An inclusion health needed to drive the policy response to the inclusion policy response must build on the evidence regarding health agenda. who is at risk and the events that trigger exclusion to Two broad potential approaches are available to address highlight the social and economic benefits of sustained this problem. First, health services could routinely record action to prevent social exclusion. membership of health inclusion groups. This would Contributors RWA, ACH, and AS proposed the hypothesis and idea for the systematic require agreed definitions of each group. Individuals review with all authors contributing to its development and the analysis responsible for recording data would need guidance plan. RWA did the literature search and reviewed studies for inclusion. to help them ascertain membership and avoid re­ All authors extracted and checked the data. RWA and DL did all inforcement of stigma. The feasibility of this approach meta­ analyses and RWA wrote the first draft of the manuscript. All authors reviewed and interpreted the results and edited the manuscript. outside of specialist services remains unclear. Alternatively, and more feasibly in the short term, data Declaration of interests AS is the clinical lead and manager of the Find & Treat service (University linkage methods could be used to match data from College Hospitals NHS Trust). ACH is a trustee of Pathway, a charity for services that work with inclusion health groups, with homeless people. All other authors declare no competing interests. vital registration data, electronic health records, and Acknowledgments existing disease surveillance systems. Data linkage has RWA was supported by an academic clinical lectureship from the been the primary method used to estimate SMRs in the National Institute for Health Research (NIHR) and and a Wellcome Trust Clinical Research Career Development Fellowship (206602/Z/17/Z). studies reported in this Article. These linked datasets AS is funded by University College London Hospitals Foundation Trust. would facilitate systematic estimates of mortality and ACH’s salary is provided by Central and North West London National morbidity over time and help to measure the effect of Health Service (NHS) Community Trust. EJT and SVK are funded by the interventions. Medical Research Council (MC_UU_12017/13 & MC_UU_12017/15) and the Scottish Government Chief Scientist Office (SPHSU13 & SPHSU15). To inform the content of this Article and the accomp­ 1 SVK is also funded by a National Research Service (NHS) Research anying Review we held an engagement workshop with Scotland Scottish Senior Clinical Fellowship (SCAF/15/02). The views 16 people with experience of homelessness and social expressed are those of the authors and not necessarily those of exclusion. We asked this group about their views on The Wellcome Trust, NIHR, NHS, NHS Research Scotland, Medical Research Council, or the Scottish Government Chief collecting operational data with ethical and appropriate Scientist Office. research governance approvals, but without specific References individual level consent. Although this sample was only 1 Luchenski S, Maguire N, Aldridge RW, et al. What works in inclusion small (and we acknowledge that people who face exclusion health: overview of effective interventions for marginalised and excluded populations. Lancet 2017; published online Nov 11. and are willing to attend a workshop might differ from http://dx.doi.org/10.1016/S0140­ 6736(17)31959­ 1. those who do not), acceptability of collection of this sort of 2 Bramley G, Fitzpatrick S, Edwards J, et al. Hard edges: data was extremely high. 13 (100%) of 13 participants were mapping severe and multiple disadvantage in England. 2015. happy for homeless hostel records to be collected, eight http://lankellychase.org.uk/multiple­ disadvantage/publications/hard­ edges/ (accessed Dec 2, 2016). (73%) of 11 agreed to the collection of criminal records, 3 Fazel S, Bains P, Doll H. Substance abuse and dependence in eight (62%) of 13 to health records, and 11 (85%) of 13 to prisoners: a systematic review. Addiction 2006; 101: 181–91. these records being linked together. 4 Gilchrist G, Singleton N, Donmall M, Jones A. Prevalence and factors associated with sex trading in the year prior to entering treatment for A vertical approach to tackling inclusion health (ie, one drug misuse in England. Drug Alcohol Depend 2015; 152: 116–22. that focuses on specific diseases or specific risk groups) 5 Greenberg GA, Rosenheck RA. Homelessness in the state and federal can overlook multimorbidity and the social issues faced prison population. Crim Behav Ment Health 2008; 18: 88–103. by excluded populations. This approach can result in 6 Croxford S, Platt L, Hope VD, Cullen KJ, Parry JV, Ncube F. Sex work amongst people who inject drugs in England, Wales and Northern inefficiencies and missed opportunities for pre vention, Ireland: findings from a national survey of health harms and early diagnosis, and management, and missed behaviours. Int J Drug Policy 2015; 26: 429–33. opportunities for mitigation of social risk factors. The 7 Fitzpatrick S, Bramley G, Johnsen S. Pathways into multiple exclusion homelessness in seven UK cities. Urban Stud 2013; emerging field of inclusion health should advocate for 50: 148–68. and deliver joined up health and social services for 248 www.thelancet.com Vol 391 January 20, 2018 Articles 8 Graham L, Fischbacher CM, Stockton D, Fraser A, Fleming M, 30 Nyhlen A, Fridell M, Backstrom M, Hesse M, Krantz P. Greig K. Understanding extreme mortality among prisoners: Substance abuse and psychiatric co­ morbidity as predictors of a national cohort study in Scotland using data linkage. premature mortality in Swedish drug abusers a prospective Eur J Public Health 2015; 25: 879–85. longitudinal study 1970–2006. BMC Psychiatry 2011; 11: 122. 9 Arendt M, Munk­ Jørgensen P, Sher L, Jensen SO. Mortality among 31 Pavarin RM. Mortality risk for cocaine abusers in relation to individuals with cannabis, cocaine, amphetamine, MDMA, heroin use: a follow­ up study. Subst Use Misuse 2013; 48: 702–10. and opioid use disorders: a nationwide follow­ up study of Danish 32 Rehm J, Frick U, Hartwig C, Gutzwiller F, Gschwend P, substance users in treatment. Drug Alcohol Depend 2011; Uchtenhagen A. Mortality in heroin­ assisted treatment in 114: 134–39. Switzerland 1994–2000. Drug Alcohol Depend 2005; 79: 137–43. 10 Nielsen SF, Hjorthøj CR, Erlangsen A, Nordentoft M. 33 Rosca P, Haklai Z, Goldberger N, Zohar P, Margolis A, Psychiatric disorders and mortality among people in homeless Ponizovsky AM. Mortality and causes of death among users of shelters in Denmark: a nationwide register­ based cohort study. methadone maintenance treatment in Israel, 1999–2008. Lancet 2011; 377: 2205–14. Drug Alcohol Depend 2012; 125: 160–63. 11 Peace R. Social exclusion: a concept in need of definition? 34 Singleton J, Degenhardt L, Hall W, Zabransky T. Mortality among Soc Policy J NZ 2001; 16: 17–36. amphetamine users: a systematic review of cohort studies. 12 Fitzpatrick S, Bramley G, Johnse S. Multiple exclusion Drug Alcohol Depend 2009; 105: 1–8. homelessness in the UK: an overview of key findings. 2012. 35 Spittal PM, Hogg RS, Li K, et al. Drastic elevations in mortality http://mcnevaluation.co.uk/download/public/observatory/ among female injection drug users in a Canadian setting. resource­ type/research­ evaluation/2012%20Multiple%20 AIDS Care 2006; 18: 101–08. Exclusion%20Homelessness%20­ %20breifing%20paper%201.pdf 36 Stoové MA, Dietze PM, Aitken CK, Jolley D. Mortality among (accessed Dec 2, 2016). injecting drug users in Melbourne: a 16­ year follow­ up of the 13 The World Bank. World Bank country and lending groups. Victorian Injecting Cohort Study (VICS). Drug Alcohol Depend https://datahelpdesk.worldbank.org/knowledgebase/ 2008; 96: 281–85. articles/906519­ world­ bank­ country­ and­ lending­ groups (accessed 37 van Santen DK, van der Helm JJ, Grady BPX, et al. Nov 2, 2016). Temporal trends in mortality among people who use drugs 14 Higgins JPT, Thompson SG. Quantifying heterogeneity in a compared with the general Dutch population differ by hepatitis C meta­ analysis. Stat Med 2002; 21: 1539–58. virus and HIV infection status. AIDS 2014; 28: 2589–99. 15 Roy É, Haley N, Boudreau J­ F, Leclerc P, Boivin J­ F. The challenge 38 Zabransky T, Csemy L, Grohmannova K, Janikova B, Brenza J. of understanding mortality changes among street youth. Mortality of cohort of very young injecting drug users in Prague, J Urban Health 2010; 87: 95–101. 1996–2010. Cent Eur J Public Health 2011; 19: 152–57. 16 Vila­ Rodriguez F, Panenka WJ, Lang DJ, et al. The hotel study: 39 Degenhardt L, Singleton J, Calabria B, et al. Mortality among multimorbidity in a community sample living in marginal cocaine users: a systematic review of cohort studies. housing. Am J Psychiatry 2013; 170: 1413–22. Drug Alcohol Depend 2011; 113: 88–95. 17 Kariminia A, Butler TG, Corben SP, et al. Extreme cause­ specific 40 Degenhardt L, Randall D, Hall W, Law M, Butler T, Burns L. mortality in a cohort of adult prisoners—1988 to 2002: Mortality among clients of a state­ wide opioid pharmacotherapy a data­ linkage study. Int J Epidemiol 2007; 36: 310–16. program over 20 years: risk factors and lives saved. 18 Bargagli AM, Hickman M, Davoli M, et al. Drug­ related mortality Drug Alcohol Depend 2009; 105: 9–15. and its impact on adult mortality in eight European countries. 41 Gilles M, Swingler E, Craven C, Larson A. Prison health and Eur J Public Health 2006; 16: 198–202. public health responses at a regional prison in Western Australia. 19 Barrio G, Molist G, de la Fuente L, et al. Mortality in a cohort of Aust N Z J Public Health 2008; 32: 549–53. young primary cocaine users: controlling the effect of the riskiest 42 Des Jarlais DC, Perlis T, Arasteh K, et al. Reductions in hepatitis C drug­ use behaviors. Addict Behav 2013; 38: 1601–04. virus and HIV infections among injecting drug users in New York 20 Bjornaas MA, Bekken AS, Ojlert A, et al. A 20­ year prospective City, 1990–2001. AIDS 2005; 19 (suppl 3): S20–25. study of mortality and causes of death among hospitalized opioid 43 del Amo J, González C, Losana J, et al. Influence of age and addicts in Oslo. BMC Psychiatry 2008; 8: 8. geographical origin in the prevalence of high risk human 21 Darke S, Mills KL, Ross J, Teesson M. Rates and correlates of papillomavirus in migrant female sex workers in Spain. mortality amongst heroin users: findings from the Australian Sex Transm Infect 2005; 81: 79–84. Treatment Outcome Study (ATOS), 2001–2009. 44 Havens JR, Walker R, Leukefeld CG. Prevalence of opioid Drug Alcohol Depend 2011; 115: 190–95. analgesic injection among rural nonmedical opioid analgesic 22 Degenhardt L, Larney S, Randall D, Burns L, Hall W. Causes of users. Drug Alcohol Depend 2007; 87: 98–102. death in a cohort treated for opioid dependence between 1985 and 45 Bart G, Piccolo P, Zhang L, Jacobson I, Schaefer RA, Kreek MJ. 2005. Addiction 2014; 109: 90–99. Markers for hepatitis A, B and C in methadone maintained 23 Evans JL, Tsui JI, Hahn JA, Davidson PJ, Lum PJ, Page K. patients: an unexpectedly high co­ infection with silent hepatitis B. Mortality among young injection drug users in San Francisco: Addiction 2008; 103: 681–86. a 10­ year follow­ up of the UFO study. Am J Epidemiol 2012; 46 Carbonara S, Babudieri S, Longo B, et al. Correlates of 175: 302–08. Mycobacterium tuberculosis infection in a prison population. 24 Gibson A, Randall D, Degenhardt L. The increasing mortality Eur Respir J 2005; 25: 1070–76. burden of liver disease among opioid­ dependent people: cohort 47 Tabuchi T, Takatorige T, Hirayama Y, et al. Tuberculosis infection study. Addiction 2011; 106: 2186–92. among homeless persons and caregivers in a 25 Hser YI, Kagihara J, Huang D, Evans E, Messina N. Mortality high­ tuberculosis­ prevalence area in Japan: a cross­ sectional study. among substance­ using mothers in California: a 10­ year BMC Infect Dis 2011; 11: 22. prospective study. Addiction 2012; 107: 215–22. 48 Prinsloo B, Parr C, Fenton J. Mental illness among the homeless: 26 Lee CTC, Chen VCH, Tan HKL, et al. Suicide and other­ cause prevalence study in a Dublin homeless hostel. Ir J Psychol Med mortality among heroin users in Taiwan: a prospective study. 2012; 29: 22–26. Addict Behav 2013; 38: 2619–23. 49 Lubman DI, Allen NB, Rogers N, Cementon E, Bonomo Y. 27 Mathers BM, Degenhardt L, Bucello C, Lemon J, Wiessing L, The impact of co­ occurring mood and anxiety disorders among Hickman M. Mortality among people who inject drugs: a systematic substance­ abusing youth. J Affect Disord 2007; 103: 105–12. review and meta­ analysis. Bull World Health Organ 2013; 91: 102–23. 50 Baillargeon J, Penn JV, Thomas CR, Temple JR, Baillargeon G, 28 Merrall ELC, Bird SM, Hutchinson SJ. Mortality of those who Murray OJ. Psychiatric disorders and suicide in the nation’s largest attended drug services in Scotland 1996–2006: record­ linkage study. state prison system. J Am Acad Psychiatry Law 2009; 37: 188–93. Int J Drug Policy 2012; 23: 24–32. 51 Vergara­ Moragues E, González­ Saiz F, Lozano OM, et al. 29 Nyhlen A, Fridell M, Hesse M, Krantz P. Causes of premature Psychiatric comorbidity in cocaine users treated in therapeutic mortality in Swedish drug abusers: a prospective longitudinal study community: substance­ induced versus independent disorders. 1970–2006. J Forensic Leg Med 2011; 18: 66–72. Psychiatry Res 2012; 200: 734–41. www.thelancet.com Vol 391 January 20, 2018 249 Articles 52 Darke S, Williamson A, Ross J, Teesson M. Attempted suicide among 59 Marmot M. Fair society, healthy lives: the Marmot Review: strategic heroin users: 12­ month outcomes from the Australian Treatment review of health inequalities in England post­ 2010. 2010. Outcome Study (ATOS). Drug Alcohol Depend 2005; 78: 177–86. https://www.gov.uk/dfid ­ research­ outputs/fair­ society­ healthy­ lives­ the­ marmot­ review­ strategic­ review­ of­ health­ inequalities­ in­ 53 Brown RT, Kiely DK, Bharel M, Mitchell SL. Geriatric syndromes in england­ post­ 2010 (accessed Oct 22, 2017). older homeless adults. J Gen Intern Med 2012; 27: 16–22. 60 Katikireddi SV, Valles SA. Coupled ethical­ epistemic analysis of 54 Harzke AJ, Baillargeon JG, Pruitt SL, Pulvino JS, Paar DP, Kelley MF. public health research and practice: categorizing variables to Prevalence of chronic medical conditions among inmates in the Texas improve population health and equity. Am J Public Health 2015; prison system. J Urban Health 2010; 87: 486–503. 105: e36–42. 55 O’Brien KK, Schuttke A, Alhakeem A, et al. Health, perceived quality 61 Aldridge RW, Shaji K, Hayward AC, Abubakar I. Accuracy of of life and health services use among homeless illicit drug users. probabilistic linkage using the enhanced matching system for Drug Alcohol Depend 2015; 154: 139–45. public health and epidemiological studies. PloS One 2015; 56 Romeri E, Baker A, Griffiths C. Mortality by deprivation and cause of 10: e0136179. death in England and Wales, 1999–2003. Health Stat Q 2006; 32: 19–34. 62 Barnett K, Mercer SW, Norbury M, Watt G, Wyke S, Guthrie B. 57 Kapilashrami A, Hill S, Meer N. What can health inequalities Epidemiology of multimorbidity and implications for health care, researchers learn from an intersectionality perspective? research, and medical education: a cross­ sectional study. Lancet Understanding social dynamics with an inter­ categorical approach? 2012; 380: 37–43. Soc Theory Health 2015; 13: 288–307. 58 Ioannidis JPA, Patsopoulos NA, Rothstein HR. Reasons or excuses for avoiding meta­ analysis in forest plots. BMJ 2008; 336: 1413–15. 250 www.thelancet.com Vol 391 January 20, 2018

Journal

The Lancet – Unpaywall

Published: Jan 1, 2018

There are no references for this article.