Excess mortality among people in homelessness with substance use disorders: a Swedish cohort study

Background People in homelessness have an increased risk of substance use disorders (SUDs) and poor health outcomes. This cohort study aimed to investigate the association between homelessness and mortality in people with SUDs, adjusting for age, sex, narcotic use, intravenous drug use and inpatient care for SUDs. Methods Data from the Swedish National Addiction Care Quality Register in the Stockholm region were used to analyse mortality risk in people with SUDs (n=8397), including 637 in homelessness, 1135 in precarious housing and 6625 in stable housing, at baseline. HRs and CIs were calculated using Cox regression. Results Mortality was increased for people in homelessness (HR 2.30; 95% CI 1.70 to 3.12) and precarious housing (HR 1.23; 95% CI 0.86 to 1.75) compared with those in stable housing. The association between homelessness and mortality decreased (HR 1.27; 95% CI 0.91 to 1.78) after adjusting for narcotic use (HR 1.28; 95% CI 1.00 to 1.63), intravenous drug use (HR 1.98; 95% CI 1.52 to 2.58) and inpatient care for SUDs (HR 1.96; 95% CI 1.57 to 2.45). Standardised mortality ratios (SMRs) showed that mortality among people in homelessness with SUDs was 13.6 times higher than the general population (SMR=13.6; 95% CI 10.2 to 17.9), and 3.7 times higher in people in stable housing with SUDs (SMR=3.7; 95% CI 3.2 to 4.1). Conclusion Homelessness increased mortality, but the risk decreased after adjusting for narcotic use, intravenous drug use and inpatient care for SUDs. Interventions are needed to reduce excess mortality among people in homelessness with SUDs.


INTRODUCTION
2][3] An estimated 700 000 people are homeless in Europe, 4 of which approximately 30 000 are in Sweden. 5The intersection of physical and mental health issues, and substance use disorders (SUDs), disproportionately impacts people in homelessness. 1 6Homelessness has consistently been associated with poor health outcomes 1 6 7 and excess mortality. 2 8-15In comparison to the general population, previous studies have calculated standardised mortality ratios (SMRs) between 2.0 and 6.7 for people in homelessness. 1 2 9-17However, more knowledge is needed to understand if the problem is homelessness itself and to what extent other factors contribute to excess mortality. 2 8 14levated risks of mortality among people in homelessness have been partly explained by exposure to risk factors, such as smoking, alcohol, substance use 1 8 10 18 and psychiatric disorders, 8 12 15 19 which may coexist. 1 2 20 Psychiatric disorders are more prevalent among people in homelessness compared with the general population, with SUDs being the most common psychiatric disorder among this population. 1 8 19Inpatient admissions for psychiatric disorders and SUDs are higher among people in homelessness compared with those not in homelessness. 14Furthermore, the year following discharge from inpatient care is a high-risk period for homelessness. 21ntravenous drug use has been associated with an increased risk of mortality among people in homelessness. 3 22-24Homelessness potentiates relapse to drug use and intravenous drug-related risk behaviour, 22 in addition to the risk of HIV 23 24 and hepatitis C virus transmission in people using intravenous drugs. 23ystematic reviews on mortality among people in homelessness 3 25 reveal that the majority of studies have been conducted in North America, 3 11-13 18 26-29 WHAT IS ALREADY KNOWN ON THIS TOPIC ⇒ People in homelessness have an increased risk of substance use disorders (SUDs).SUDs are associated with an increased risk of mortality.⇒ People in homelessness have an increased risk of excess mortality but the association between excess mortality and SUDs requires further investigation.

WHAT THIS STUDY ADDS
⇒ Few previous studies have investigated the association between homelessness and SUDs in a large cohort study.⇒ Among the study population, the risk of mortality among people in homelessness at baseline was higher than among people in stable housing.⇒ The increased mortality risk among people in homelessness with SUDs was no longer significant after adjusting for narcotic use, intravenous drug use and inpatient care for SUDs.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
⇒ This study underscores the need for policies and targeted interventions to support people in homelessness with SUDs to decrease excess mortality in this vulnerable population.

Original research
or are based on older data collections 15 16 19 ; thus, there is a need to develop up-to-date insights from other regions.This study uses data from the Swedish National Addiction Care Quality Register (Swedish name: Bättre Beroendevård) which records housing status and provides a new and unique opportunity to perform research among people in homelessness or precarious housing who have SUDs (hereafter, quality register).

AIM
This cohort study aimed to investigate the association between homelessness and mortality in people with SUDs, adjusting for age, sex, narcotic use, intravenous drug use and inpatient care for SUDs.

METHODS
Data were retrieved from the quality register.The purpose of the quality register is to contribute to improved quality and more equal addiction care nationwide by measuring quality indicators. 30Inclusion criteria for the quality register are that individuals must have a SUD diagnosis, as defined by the International Classification of Diseases, 10th Revision (ICD-10 codes, F10-F19 31 ), and be receiving treatment at a specialised addiction care clinic.
For this study, we used data from the Stockholm healthcare region where the quality register is integrated into electronic healthcare record systems; thus, patient data are directly transferred to the quality register with a high coverage rate. 30he quality register data encompasses a range of information, including sociodemographic characteristics, diagnoses, pharmaceutical treatment, psychosocial treatment, self-reported substance use and outcomes.All data are collected by healthcare staff during healthcare visits to specialised addiction centres.Through structured interviews, the healthcare staff use predefined terms to ask the patient about various questions, including the patient's intravenous drug use and substance use.
The quality register is updated about whether a patient is alive or deceased from national statistics (Statistics Sweden) daily.The mortality data in the quality register are sourced from the Swedish National Death Register.Linkage between different data sources (ie, the quality register and the Swedish National Death Register) was performed on the individual level using the personal identification number (PIN).PINs, which are unique for all Swedish citizens and permanent residents, are used in all Swedish national health and sociodemographic registers.

Study population
The cohort consisted of 8397 patients with SUDs that fulfilled the inclusion criteria.Table 1 presents the baseline characteristics of the people in homelessness (n=637), in precarious housing (n=1135) or in stable housing (n=6625).Inclusion criteria for the study were patients: (1) aged 18 years and over with a SUD diagnosis (ICD-10 codes, F10-F19) 31 ; and (2) registered in the quality register with their housing status between 2018 and 2020.The inclusion date refers to the first recorded housing status.The start of follow-up was at the inclusion date (ie, 2018-2020) and follow-up continued until death, or 31 December 2022.Baseline information including psychiatric diagnoses and self-report variables was collected at baseline and 365 days prior (ie, information that is registered during these 365 days).Other than mortality data, no data pertaining to events beyond baseline are included in the analyses.

Variables
Table 2 presents an overview of the key variables included in the analyses.

Statistical analysis
Statistical comparisons between groups were made using χ 2 tests for categorical variables and the analysis of variance for the continuous variable, that is, age (see tables 1 and 3).In table 1, the categorical variables are presented as counts and percentages, and the continuous variable is presented with mean and SD.Cox regression models were used to estimate the HRs with 95% CIs.All models were adjusted for age and sex, and tested to ensure that the assumptions for Cox regression were fulfilled.Estimated HRs were additionally interpreted in terms of the probabilistic index. 32SMRs with 95% CIs were calculated as the ratio of observed to expected number of deaths. 33he expected number of deaths was calculated using general population mortality rates from the Swedish National Board of Health and Welfare, and stratified by 5-year age groups and sex.

Baseline characteristics of the study population
At baseline, there were more men than women within each housing group.The proportion of women living in stable housing was 35.6% (n=2361), in homelessness it was 25.6% (n=163) and in precarious housing it was 25.6% (n=291).Among the study population, mean age for people in precarious housing at baseline was younger (M=33.9years; SD=12.1)than those in homelessness (M=40.7 years; SD=12.6), or in stable housing (M=47.1 years; SD=14.0).All differences in sex and age were statistically significant.Table 1 presents the descriptive statistics associated with the substance use and psychiatric variables for the people in homelessness, precarious housing and stable housing at baseline.The results in table 1 indicate that a significantly higher proportion of those experiencing homelessness had used narcotics (n=467; 73.3%) compared with those in precarious (n=681; 60.0%) or stable housing (n=1913; 28.9%).Similarly, a significantly higher proportion of those experiencing homelessness had used intravenous drugs (n=246; 38.6%) compared with those in precarious Sex was defined as male or female.

Age Continuous
Age was calculated at baseline.

Housing status Homelessness=1 Precarious housing=2 Stable housing=3
Patients were asked the following question at the addiction clinic: "how was your living situation in the last 30 days?"Housing status was based on three predefined groups: homelessness, precarious housing and stable housing.

Intravenous drug use Ever=1 Never=0
Ever: (i) injected drugs previously but not during the last 12 months; (ii) injected drugs during the last 12 months but not during the last 30 days; or (iii) injected drugs during the last 30 days.Never: no, never injected drugs.
Inpatient care for SUDs Yes=1 No=0 Yes: ≥1 day of specialised residential treatment for SUDs (ICD-10 F10-F19) in the 12 months prior to baseline.
Selected psychoactive drug use Yes=1 No=0 Yes: Used psychoactive drugs in the 12 months prior to baseline including other narcotic stimulants, other opioids, buprenorphine-naloxone, hallucinogens, heroin, methadone, mono-buprenorphine, stimulants with hallucinogenic effects.

Mortality
Date of death.
No: Both alcohol-related and drug-related cause of death, or non-alcohol-related cause of death.

Other causes Yes=1 No=0
Yes: Other ICD-10 codes than those included for alcohol related, drug related or external causes of death.

Unknown causes Yes=1 No=0
Yes: ICD-10 code unknown at the time of data collection.

SMRs, number of deaths and causes of death
Table 3 presents the SMRs, in addition to the number of deaths, causes of death, mean age at death and mean years of follow-up.Among the study population, the SMR for people in homelessness at baseline was 13.6 (CI 10.2 to 17.9), the SMR for people in precarious housing was 8.7 (CI 6.2 to 12.0) and the SMR for people in stable housing was 3.7 (CI 3.2 to 4.1).A higher proportion of people in homelessness at baseline died (n=51; 8.0%) than those in precarious housing (n=38; 3.3%), or in stable housing (n=297; 4.5%).Furthermore, a higher proportion of people in homelessness at baseline (n=13; 25.5%) died from unknown causes than those in precarious housing (n=6; 15.8%), or in stable housing (n=42; 14.1%).A lower proportion of people in homelessness at baseline (n=1; 2.0%) died from alcohol-related causes than those in precarious housing (n=4; 10.5%), or in stable housing (n=69; 23.2%).

Cox regression results
The crude and adjusted models, with p values, are presented in figure 1.
The crude model showed that among the study population, people in homelessness at baseline had an increased risk of mortality (HR 2.30; 95% CI 1.70 to 3.12) compared with those in stable housing (figure 1).According to the probabilistic index, the probability that the period was longer for a person in stable housing compared with a person in homelessness was 69.7% (95% CI 63.0% to 75.7%).In the study population, people in precarious housing at baseline did not have a significant increase in risk of mortality (HR 1.23; 95% CI 0.86 to 1.75).
In the model presented in figure 1 when adjusting for narcotic use, intravenous drug use, and inpatient care for SUDs, the HR for mortality among people in homelessness was not significant (HR 1.27; 95% CI 0.91 to 1.78).Narcotic use (HR 1.28; 95% CI 1.00 to 1.63), intravenous drug use (HR 1.98; 95% CI 1.52 to 2.58) and inpatient care for SUDs (HR 1.96; 95% CI 1.57 to 2.45) were associated with significantly increased mortality among people with SUDs.In the study population, people in precarious housing at baseline did not have a significant increase in risk of mortality (HR 1.02; 95% CI 0.72 to 1.46).

Sensitivity analyses
In a sensitivity analysis using a shorter 1-year follow-up of the main crude model, the HRs for mortality among people in homelessness (HR 3.02; 95% CI 1.94 to 4.72) and precarious housing (HR 1.72; 95% CI 1.03 to 2.85) were significant.In the adjusted model with a 1-year follow-up, the HR for mortality among people in homelessness (HR 1.70; 95% CI 1.04 to 2.80) became significant.The HR for precarious housing (HR 1.44; 95% CI 0.86 to 2.40) was still not significant.
In another sensitivity analysis adjusting for selective psychoactive drug use, the HRs for mortality among people in

Original research
homelessness remained similar to main results (online supplemental table 1).

Summary of key findings
In the present study, homelessness at baseline was associated with an increased risk of mortality which supports previous studies. 2 8-15The relative mortality risk for people in homelessness decreased after adjusting for narcotic use, intravenous drug use and inpatient care for SUDs.This aligns with earlier research showing that alcohol and drug use contribute to excess mortality in this population. 15A higher proportion of people in homelessness had psychiatric comorbidities, in addition to narcotic use, intravenous drug use and inpatient care for SUDs, than people in precarious housing or stable housing.However, a lower proportion of people in homelessness used alcohol compared with people in precarious housing or stable housing.SMRs showed that mortality among people in homelessness with SUDs was 13.6 times higher than the general population, and 3.7 times higher in people in stable housing with SUDs.The estimated SMR for people in homelessness with SUDs was higher than previous estimates among people in homelessness, which range between 2.0 and 6.7 1 2 9-17 ; however, there are considerable differences between the study populations.
Similar to earlier studies, narcotic use, 15 18 intravenous drug use 3 22-24 and inpatient care, sometimes referred to as a treatment for SUDs, 15 were associated with excess mortality.

Implications for policy and practice
In terms of policy and practice implications, the results indicate a need for addiction care interventions to support people in homelessness with SUDs.The results support international research 35 and guidance 36 on health and social responses to homelessness and substance use, calling for integrated strategies connecting services (eg, addiction care, housing support, psychosocial services, among others) that are targeted to the individual's needs.
Using the shorter follow-up time of 1 year in the sensitivity analyses showed an increased mortality risk among the people in homelessness.A shorter follow-up provides potentially more up-to-date information on homelessness which this study demonstrated is a risk factor for mortality.

Strengths and limitations
A strength of this study is its use of the quality register, which provides unique individual-level data on homelessness from addiction care.It is challenging to access cohort data on homelessness since it is not included in national health data registers.The quality register provided access to a larger, populationbased cohort which addresses some of the limitations of earlier research based on smaller samples of people participating in specific healthcare programmes for homelessness, 25 or samples of community-based shelter records which do not account for different types of homelessness (ie, rough sleeping, temporary lodgings or institutions). 2 8 9 15 The present study's analysis focused on the psychiatric diagnoses in relation to addiction care and housing variables at baseline and did not account for changes over the follow-up period that may have impacted mortality.It is likely that many of those in the study population, particularly those experiencing homelessness, had somatic health comorbidities that may have contributed to increased mortality. 1 3 6 7 20Consequently, the results likely include physical health conditions that were beyond the scope of this study.
8][39] Severe narcotic users are more likely to under-report the frequency of substance use rather than deny it. 40A meta-analysis indicated that false omission rates are generally low, especially if there are no consequences and if people are asked about longer reference periods. 41Moreover, self-classifications of housing status may not be consistent if patients have different understandings of what constitutes homelessness.

Future directions
There were insufficient data to stratify our analyses according to sex but in future studies it is advisable to investigate differences between the sexes in terms of their experiences of housing status when living with SUDs, and how these differences may impact mortality.Furthermore, the large age difference between groups at baseline limited the interpretation of differences in mean age at death in a meaningful way.
The duration of homelessness could not be analysed because this information is not available in the quality register.There are limited national and regional statistics on the length of time a person spends in homelessness or other types of housing status.A report from the Swedish National Board of Health and Welfare estimated that more than two-thirds of people in homelessness in Sweden have been in homelessness for a year or longer, and approximately 3000 have been in homelessness for at least 10 years. 5However, significant variation in homelessness duration has been reported in different countries and regions. 4 5

CONCLUSION
Homelessness at baseline was associated with excess risk of mortality in the study population with SUDs.The excess risk of mortality decreased after adjusting for narcotic use, intravenous drug use and inpatient care for SUDs.The findings underscore the need for targeted interventions to support individuals experiencing both homelessness and SUDs, with the objective of reducing the elevated risk of mortality within this vulnerable population.
Contributors All authors contributed to the design of the study.HW and JH performed the statistical analysis.All authors contributed to the interpretation of the results and revision of the manuscript.SNG drafted the manuscript, and all authors contributed to and approved the final version.JW is the guarantor.

Figure 1
Figure 1 Cox regression analyses of mortality among people with substance use disorders (SUDs) in homelessness, precarious housing and stable housing (reference group).

Table 2
List of variables

*
31ternal causes include suicide, unknown suicide, accident or assault.ICD-10 coding according to the WHO's International Statistical Classification of Diseases and Related Health Problems (10th edition).31ICD-10,International Classification of Diseases, 10th Revision; PTSD, post-traumatic stress disorder.

Table 3
Standardised mortality ratios (SMRs), number of deaths and causes of death Q3: quartile 1 and quartile 3. †See table2for International Classification of Diseases, 10th Revision codes of the causes of death.‡External causes include suicide, unknown suicide, accident or assault.