Patterns, circumstances and risk factors associated with non-fatal substance overdose in a cohort of homeless population: an observational study
https://ror.org/03angcq70grid.6572.60000 0004 1936 7486College of Medical and Dental Sciences, University of Birmingham, Birmingham, UK
https://ror.org/01tgmhj36grid.8096.70000 0001 0675 4565Centre for Healthcare and Communities, Coventry University, Coventry, UK
https://ror.org/01nrxwf90grid.4305.20000 0004 1936 7988Centre for Homelessness and Inclusion Health, School of Health in Social Science, University of Edinburgh, Edinburgh, UK
https://ror.org/00cjeg736grid.450453.3Birmingham and Solihull Mental Health NHS Foundations Trust, Birmingham, UK
https://ror.org/0220mzb33grid.13097.3c0000 0001 2322 6764Florence Nightingale Faculty of Nursing, Midwifery and Palliative Care, King’s College London, London, UK
Abstract
Background
Non-fatal overdoses frequently precede fatal overdoses, thus identifying risk factors for non-fatal overdoses could help develop strategies to prevent substance related deaths.
Aim
This study aimed to identify patterns, circumstances and risk factors leading to non-fatal substance overdose in people experiencing homelessness.
Method
All recorded cases of non-fatal substance overdose from a population of people experiencing homelessness registered at a specialist homelessness primary care centre in England were identified using electronic medical records. Overdose details and patient characteristics were extracted. The heterogeneity between variables in people with and without a recorded non-fatal overdose were tested and multivariable logistic regressions were used to identify the risk factors of non-fatal overdoses.
Results
From the 1221 registered patients, 194(16%) were identified as having had a non-fatal overdose with 428 overdoses between them. Half were polypharmacy events with the main substances of overdose being: heroin, paracetamol, benzodiazepines, cocaine, antipsychotics, SSRIs and synthetic cannabinoids. Risk of non-fatal overdose was greater in females, white ethnicity, ages 36–45, and in those with a recorded use of tobacco, alcohol or illicit substance use. Chronic physical and mental health conditions increased the risk of non-fatal overdose including respiratory conditions, blood borne viruses, migraines, anxiety and depression.
Conclusion
With a high number of non-fatal overdoses within this population, identifying individuals at risk based on the factors identified in this research could enable primary care providers to apply prevention actions such as overdose awareness and naloxone provision to avoid drug harm and deaths. Future work should explore the role of chronic physical conditions and their treatment on non-fatal overdose risks.
Supplementary Information
The online version contains supplementary material available at 10.1007/s11096-024-01812-z.
Impact statements
- Non-fatal overdose events recorded in primary care can enable service providers to offer prevention actions such as overdose awareness and naloxone provision to avoid future fatal events.
- Understanding of risk factors for non-fatal overdose in homeless populations identified in this study such as female sex, white ethnicity, history of alcohol and substance use disorders can allow targeted approaches for prevention actions.
- Addressing wider determinants of health such as physical, mental health, addiction, social and housing issues is key to preventing substance overdose.
Introduction
Globally, substance overdose deaths remain a key public health concern. In England and Wales, a total of 4859 deaths related to substance poisoning were recorded in the year 2021, equivalent to 84.4 deaths per million people [1]. In particular, substance overdose was the most important cause of death among people experiencing homelessness (PEH), responsible for over a third of all deaths [2].
Rising homelessness and associated early deaths necessitate patterns and risk factors leading to overdose incidents in PEH to be better understood to strengthen prevention actions. In England alone, during 2020–21, 282,000 people were identified as homeless or threatened with homelessness, with numbers expected to increase further due to the current cost of living crisis [3]. PEH typically have a complex and unique set of medical and social needs, including increased early-onset age-related diseases [4], co-existing substance use and mental health disorders [5, 6] often, in part, a result of psychological and/or childhood trauma [7, 8].
Overdoses can be fatal or non-fatal (NFOD), with the latter a predicting factor of future fatal overdose [8]. PEH are at an increased risk of NFOD [9] and they are likely to have multiple overdoses, with one study reporting a mean of 3.2 NFODs over six months [10]. NFODs frequently require help from ambulances and/or present to emergency departments (ED) with significant cost implications to the health system and society [11–13]. Within England, substance overdose is known to account for 17.9% of all ED visits by PEH compared to only 1.9% of the general population with rates more than doubling in the last ten years [12].
NFODs are a strong predictor of subsequent overdose (fatal and non-fatal), and fatal overdose is the most common cause of death in PEH [14]. The importance of this public health epidemic of substance use disorder, overdose, and the resulting increased pressures on emergency services is emphasised by national initiatives that are encouraging innovative models of research and care [15, 16]. However, there is a dearth of descriptive or analytical studies of overdose in PEH, with the extent, nature and risk factors (e.g. sex, physical health conditions) for NFODs poorly understood.
Aim
This study aimed to identify patterns, circumstances and risk factors leading to non-fatal substance overdose in people experiencing homelessness.
Ethics approval
The study was reviewed and approved by institutional review board of Birmingham and Solihull Mental Health NHS Foundation Trust, approval reference number: 2023/SE0369.
Method
Participants included all patients registered at a specialist homeless primary healthcare centre in the West Midlands region of England. The centre provides access to various healthcare professionals—GPs, nurses, and podiatrists—alongside psychotherapy and hosts a street outreach service. A substance dependence treatment centre is located in the city in a separate setting.
Data were collected using the electronic patient record system (EMIS Web). This system contains internal consultation and prescription records and external documents from other services (e.g. secondary care, and ambulance services). The reporting function was used to identify all non-fatal overdoses that had ever been recorded based on specified coded terms (Electronic Supplementary Material 1). The coded terms are set by the Systemised Nomenclature of Medicine Clinical Terms (SNOMED CT) and used throughout the National Health System (NHS) in the UK. Relevant codes are added to patient records by staff at the practice based on consultations and external documents received including ambulance and hospital discharge summaries. General characteristics were also extracted based on these coded items for all participants (age, gender, race, accommodation status) alongside terms to identify mental and physical health problems on record (Electronic Supplementary Material 2) and substance use (Electronic Supplementary Material 3).
Where a substance overdose was identified, medical records were searched for information regarding the overdose. This included documents from ambulance reports and hospitals (discharge letters from A&E and hospital wards, liaison psychiatry letters where relevant), free text searching to identify relevant consultation notes and free text searching within the ‘problems’ tab.
The following information was recorded for each overdose where available: number of overdoses for that individual, the substances overdosed on (including if alcohol was involved—Electronic Supplementary Material 4), ambulance attendance, administration of naloxone, ED attendance, hospital admissions and any suicidal intent. Free text information was also documented if relevant.
Data analysis and statistics
The exported text data were changed to binary form using a custom MATLAB (R2023b) program. Statistical analysis was conducted using Stata version 16 to identify factors associated with NFOD. Differences in the characteristics of those with a NFOD and those without were determined using chi-square tests and Wilcoxon rank sum tests for normally and non-normally distributed data, respectively. Odds ratios and 95% confidence intervals were calculated using multivariable logistic regression. All data were adjusted for sex, age group, smoking status, BMI status and alcohol intake status.
Results
Population characteristics
Overall, 1221 patients were registered at the practice during data collection (Table 1). Males comprised the majority of the population (84.9%) and the population had a mean age of 39.7 ± 12.2 years and a mean time registered with the practice of 3.9 ± 4.3 years. Asylum seekers accounted for 14.5% of the patients. Based on their current address, 86 (7.0%) of patients were recorded has having no fixed abode, 152 (12.4%) in a hostel, 167 (13.7%) in a hotel, 88 (7.2%) in supported accommodation and the remaining 645 (52.8%) had a residential address recorded that represented some form of temporary accommodation (permanent living situations do not remain at the specialist practice). Of the total population, 319 (26.1%) were recorded as rough sleeping at some point.Characteristic Total participants (n = 1221) At least one NFOD on record p-value Yes (n = 194) No (n = 1027) n (% total) n (% those with overdose) n (% those without overdose) Age groups (years) 16–25 153 (12.5%) 4 (2.1%) 149 (14.5%) < 0.001 26–35 334 (27.3%) 37 (19.1%) 297 (28.9%) 36–45 373 (30.5%) 94 (48.4%) 279 (27.2%) 45 + 361 (29.6%) 59 (30.4%) 302 (29.4%) Sex (male) 1037 (84.9%) 154 (79.4%) 883 (86.0%) 0.018 Body mass index (kg/m2) Normal (18.5–25.0 kg/m2) 516 (42.3%) 102 (52.6%) 414 (40.3%) < 0.001 Underweight (< 18.5 kg/m2) 59 (4.8%) 13 (6.7%) 46 (4.5%) Overweight/obese (> 25 kg/m2) 364 (29.8%) 65 (33.5%) 299 (29.1%) Not recorded 281 (23.0%) 14 (7.2%) 267 (26.0%) Ethnicity White 285 (23.3%) 75 (38.7%) 210 (20.4%) < 0.001 Black/African/Caribbean 139 (11.4%) 11 (5.7%) 128 (12.5%) Asian 127 (10.4%) 10 (5.1%) 117 (11.4%) Mixed 141 (11.5%) 45 (23.2%) 96 (9.3%) Others 44 (3.6%) 0 44 (4.3%) Unknown/not recorded 485 (39.7%) 53 (27.3%) 432 (42.1%) Asylum seekers 177 (14.5%) 20 (10.3%) 157 (15.3%) 0.071 Accommodation (from address) No fixed abode 86 (7.0%) 20 (10.3%) 66 (6.3%) Hostel 152 (12.4%) 26 (13.4%) 126 (12.3%) 0.019 Hotel 167 (13.7%) 33 (17.0%) 134 (13.0%) Supported accommodation 88 (7.2%) 11 (5.7%) 77 (7.5%) Unknown/temporary housing 570 (46.7%) 70 (36.1%) 500 (48.7%) Rough sleeping (ever recorded) 319 (26.1%) 55 (28.4%) 264 (25.7%) 0.442 Smoking Never smoker 209 (17.1%) 6 (3.1%) 203 (19.8%) < 0.001 Ex-smoker 52 (4.3%) 6 (3.1%) 46 (4.5%) Current smoker 680 (55.7%) 171 (88.1%) 509 (49.6%) No information recorded 280 (22.9%) 11 (5.7%) 269 (26.2%) Cigarettes consumption No. of cigarettes per day (n = 396) 11.3 (8.5%) 12.7 (8.7%) 10.9 (8.4%) 0.032 Grams of tobacco per week (n = 460) 17.5 (22.3%) 32.4 (22.8%) 15.2 (21.4%) < 0.001 Alcohol consumption No (no record of it) 681 (55.8%) 59 (30.4%) 622 (60.6%) < 0.001 Yes 540 (44.2%) 135 (69.6%) 405 (39.4%) Alcohol dependent 265 (21.7%) 90 (46.4%) 175 (17.03%) < 0.001 Not alcohol dependent 275 (50.9%) 45 (33.3%) 230 (56.8%) Alcohol consumption (units/week) Up to 14 196 (48.9%) 38 (40.4%) 158 (51.5%) 0.061 More than 14 205 (51.1%) 56 (59.6%) 149 (48.5%) Alcohol-related information Refused to discuss drinking 19 (1.6%) 4 (2.1%) 15 (1.5%) 0.535 Alcohol overdose 63 (5.2%) 32 (16.5%) 31 (3.0%) < 0.001 Alcohol/drug referral accepted 59 (4.8%) 20 (10.3%) 39 (3.8%) < 0.001 Alcohol /drug referral declined 17 (1.4%) 6 (3.1%) 11 (1.1%) 0.028 Education about alcohol provided 294 (24.1%) 80 (41.2%) 214 (20.8%) < 0.001 Blood pressure measurement (too few with low BP to include) Normal (90–140 SBP) 810 (67.7%) 153 (80.5%) 657 (65.2%) < 0.001 High (> 140 SBP) 131 (10.9%) 24 (12.6%) 107 (10.6%) No measurement 255 (21.3%) 13 (6.8%) 242 (24.0%) Illicit substance use Substance misuse present 442 (36.2%) 146 (75.3%) 296 (28.8%) < 0.001 Poly-substance misuse 272 (22.3%) 102 (52.6%) 170 (16.5%) < 0.001 Cannabis 114 (9.3%) 37 (19.1%) 77 (7.5%) < 0.001 Cocaine 194 (15.9%) 72 (37.1%) 122 (11.9%) < 0.001 Heroin 257 (21.0%) 91 (46.9%) 166 (16.2%) < 0.001 Methadone (misuse of) 22 (1.8%) 10 (5.1%) 12 (1.2%) < 0.001 Opioid (not otherwise specified) 187 (15.3%) 73 (37.6%) 114 (11.1%) < 0.001 Synthetic cannabinoids (e.g. mamba, spice) 77 (6.3%) 28 (14.4%) 49 (4.8%) < 0.001 Benzodiazepine 32 (2.6%) 21 (10.8%) 11 (1.1%) < 0.001 Ecstasy (MDMA) 6 (0.5%) 4 (2.1%) 2 (0.2%) 0.001 Amphetamines 17 (1.4%) 10 (5.1%) 7 (0.7%) < 0.001 Physical health co-morbidities At least one present 462 (37.8%) 129 (66.5%) 333 (32.4%) – Respiratory 145 (11.9%) 51 (26.3%) 94 (9.1%) < 0.001 Asthma 131 (10.7%) 48 (24.7%) 83 (8.1%) < 0.001 COPD 29 (2.4%) 10 (5.1%) 19 (1.8%) 0.006 BBV 167 (13.7%) 64 (33.0%) 103 (10.0%) < 0.001 Hepatitis C 157 (12.9%) 62 (32.0%) 95 (9.2%) < 0.001 HIV 14 (1.1%) 3 (1.5%) 11 (1.1%) 0.568 Cardiovascular 16 (1.3%) 3 (1.5%) 13 (1.3%) 0.753 GI bleed or ulcer 74 (6.1%) 23 (11.9%) 51 (5.0%) < 0.001 Diabetes type 2 44 (3.6%) 4 (2.1%) 40 (3.9%) 0.209 Epilepsy 35 (2.9%) 15 (7.7%) 20 (1.9%) < 0.001 Hypertension 71 (5.8%) 11 (5.7%) 60 (5.8%) 0.925 Leg/foot ulcer 51 (4.2%) 15 (7.7%) 36 (3.5%) 0.007 STI 15 (1.2%) 5 (2.6%) 10 (1.0%) 0.063 Migraine 44 (3.6%) 14 (7.2%) 30 (2.9%) 0.003 Mental health co-morbidities At least one present 619 (50.7%) 164 (84.5%) 455 (44.3%) – ADHD 21 (1.7%) 8 (4.1%) 13 (1.3%) 0.005 Depression 410 (33.6%) 112 (57.7%) 298 (29.0%) < 0.001 Anxiety 276 (22.6%) 90 (46.4%) 186 (18.1%) < 0.001 PTSD 62 (5.1%) 16 (8.2%) 46 (4.5%) 0.028 Bipolar 21 (1.7%) 7 (3.6%) 14 (1.4%) 0.027 Personality disorder 82 (6.7%) 48 (24.7%) 34 (3.3%) < 0.001 Psychosis (inc. Schizophrenia) 85 (7.0%) 23 (11.9%) 62 (6.0%) 0.003 Substance misuse 442 (36.2%) 146 (75.3%) 296 (28.8%) < 0.001 Drug-related mental behaviour 55 (4.5%) 16 (8.2%) 39 (3.8%) 0.006 Alcohol related 20 (1.6%) 9 (4.6%) 11 (1.1%) < 0.001 Recorded self-harm or suicide attempt 185 (15.1%) 81 (41.7%) 104 (10.1%) < 0.001
A total of 462 (37.8%) had at least one chronic condition reported. Of the conditions included, blood-borne viruses (BBV) were recorded in 167 (13.7%) of the population, and respiratory conditions were recorded in 145 (11.9%). There was a high prevalence of mental health conditions within the population, with 619 (50.7%) having at least one mental health condition documented. Depression was the most common (n = 410, 33.6%), followed by anxiety (n = 276, 22.6%). An attempt at suicide or self-harm at any time had been recorded in 185 (15.2%).
Substance use
Current smokers made up 55.7% (n = 680) of the total patient group, and 44.2% (n = 540) were recorded as consuming alcohol. This included 265 (21.7%) noted to be problem drinkers.
Illegal substance misuse was identified in 442 patients (36.2%). Heroin was the most common substance of use, with 257 (58.1%) of the substance misuse population reported as users (21.0% of the total population). As shown in Fig. 1, other common substances included cocaine (194, 43.9% substance misuse population; 15.9% total population), cannabis (25.8% of substance misuse population, 9.3% total population) and synthetic cannabinoids such as mamba and spice (n = 77, 17.4% substance misuse, 6.3% total population).
Non-fatal overdoses
In total, 194 patients (15.9%) were recorded as ever having had a NFOD, with 428 overdoses between them. The median number of overdoses was 1, with 100 patients (51.5% of patients with NFOD) having just one recorded episode. Two overdoses were recorded in 45 people (23.2% of patients with NFOD), three in 22 people (11.3% of patients with NFOD), four in 12 people (6.2% of patients with NFOD) and 15 people (7.6% of patients with NFOD) had 5 or more overdoses recorded. The maximum number of episodes recorded for a single patient was 21. In the five years prior to the data collection, 94 patients (7.7% of all patients) had a recorded overdose.
No information beyond the coded term of substance overdose was available for 121 (28.2%) of the 428 overdoses. In total, there were 264 (61.7%) incidents that had a recorded substance of overdose, and 208 (48.6%) had information regarding the management.
Substance of overdose
Of the 264 NFODs that had a substance recorded, single substance overdoses were seen in exactly half (50.0%) of these, with poly-substance overdoses accounting for the other half. Alcohol use was implicated in a further 51 (19.3%) of these overdoses.
As seen in Fig. 2, for single substance overdoses (n = 132), the most commonly identified substances were: heroin (n = 30, 22.7%), paracetamol (n = 30, 22.7%), synthetic cannabinoids (n = 9, 6.8%), SSRIs (n = 9 6.8%), anti-psychotics (n = 7, 5.3%), benzodiazepines (n = 7, 5.3%) and co-codamol (n = 5, 3.8%). For polysubstance overdoses, paracetamol (n = 37, 28.0%) was the most common substance followed by heroin (n = 31, 23.5%), cocaine (n = 32, 24.2%), benzodiazepines (n = 31, 23.5%), anti-psychotics (n = 21, 15.9%), SSRIs (n = 19, 14.4%), mirtazapine (n = 18, 13.6%), synthetic cannabinoids (n = 16, 12.1%), pregabalin (n = 16, 12.1%), sleeping tablets – unspecified, but could include benzodiazepines (n = 16, 12.1%), co-codamol (n = 13, 9.8%) and methadone (n = 13, 9.8%).
Of the 194 who had a recorded overdose, 146 (75.3%) had coded substance abuse on record. Heroin was the most common substance of use, with 91 (46.9% NFOD population) reported to be users. Other opioids not specified (n = 74, 37.6%), cocaine (n = 54, 37.1%), cannabis (n = 37, 19.1%), synthetic cannabinoids (n = 28, 14.4%) and benzodiazepines (n = 22, 11.3%) were also popular substances of abuse.
Use of emergency services in NFOD
Of the 208 overdoses with management information, 110 (52.9%) had an ambulance call out (Table 2). Of these, 21 (10.1%) involved the patient self-discharging from the ambulance prior to reaching the hospital. Naloxone was administered in 32 (15.4%) cases, primarily given by the ambulance service. Overall, 148 (71.2%) NFODs led to a presentation to the ED, with self-discharge before assessment in 21 cases (10.1%).Management n % total overdoses % overdoses with recorded information No management information 220 51.4 – Ambulance called 110 25.7 52.9 ED attended 148 34.6 71.2 Ambulance + ED 78 18.2 37.5 Ambulance or ED attended 181 42.3 87.0 Naloxone administered 32 7.5 15.4 Hospital admission 57 13.3 27.4 Self-discharged from ambulance 21 4.9 10.1 Self-discharged from hospital/ED 21 4.9 10.1
In total, 57 overdose events (27.4%) led to hospital admissions. For 25 of these 57, a length of stay was recorded. The majority (n = 14) stayed for one day, 7 for two days, two individuals for 3 days, one person for 4 days, one for 11 days and one for 25 days. The short single-day stays were reported for observation, whereas longer stays were due to co-existing issues, including aspiration pneumonia (25 days), mental health reasons (11 days) and leg ulcers (3 days).
Circumstances leading to NFOD
In some of the medical notes, there was descriptive text regarding the overdose. In general, two subsets of overdoses appeared. Firstly, was the group that intended to self-harm or attempt suicide. In 50 cases (12% of NFODs with available data), direct references to mental health were made, with 25 (9% of NFODs with available data) noting suicidal intent. These were often triggered by life events such as relationship breakups or a family member dying. These overdoses often involved taking multiple different street substances and/or prescribed medications.
A second subset (n = 91, 44% of NFOD with available data) involved regular substance users who overdosed on street drugs. This group appeared more likely to self-discharge from emergency care. Of 91 overdoses identified as caused by street drugs (predominantly heroin and cocaine), self-discharge from ambulances or ED occurred in 31 (34.1%) cases accounting for 73.8% of the self-discharge events. While it is not possible to be certain, many overdose events were likely unintentional due to factors such as reduced tolerance from being in prison noted.
Risk factors for non-fatal overdose
Several factors were associated with individuals with a recorded NFOD compared to those without (Table 3). Male sex was associated with a reduced risk of NFOD (OR: 0.52, 95% CI 0.33–0.81), white ethnicity a greater risk compared to Black/African/Caribbean (OR: 0.42, 95% CI 0.21–0.85) or Asian (OR: 0.42, 95% CI 0.20–0.89) and age 36–45 associated with higher risks (OR: 3.86, 95% CI 1.33–11.22).Characteristic n (% total participants) Odds ratio (OR) 95% CI Age (years) 16–25 153 (12.5%) Ref 26–35 334 (27.3%) 2.02 0.68–6.05 36–45 373 (30.5%) 3.86 1.33–11.22 45+ 361 (29.6%) 1.97 0.66–5.82 Sex Female 184 (15.1%) Ref Male 1037 (84.9%) 0.52 0.33–0.81 Body mass index (kg/m2) Normal (18.5–25.0 kg/m2) 516 (42.3%) Ref Underweight (< 18.5 kg/m2) 59 (4.8%) 1.01 0.50–2.05 Overweight/obese (> 25 kg/m2) 364 (29.8%) 1.14 0.78–1.67 Not recorded 281 (23.0%) 0.76 0.37–1.55 Ethnicity White 285 (23.3%) Ref Black/African/Caribbean 139 (11.4%) 0.42 0.21–0.85 Asian 127 (10.4%) 0.42 0.20–0.89 Mixed 141 (11.5%) 1.36 0.85–2.17 Unknown/not recorded 485 (39.7%) 0.68 0.44–1.06 Number of years registered at practice less than one year 434 (35.5%) Ref 1–2 years 207 (16.9%) 0.94 0.54–1.65 2–4 years 183 (15.0%) 0.7 0.38–1.27 4–6 years 134 (11.0%) 0.72 0.40–1.30 > 6 years 263 (21.5%) 0.71 0.42–1.20 Accommodation from address Supported accommodation 88 (7.2%) Ref No fixed abode 86 (7.0%) 1.12 0.60–2.09 Hostel 152 (12.4%) 0.72 0.42–1.24 Hotel 167 (13.7%) 1.10 0.65–1.85 Unknown/temporary housing 570 (46.7%) 0.80 0.38–1.68 Rough sleeping (ever recorded) 319 (26.1%) 1.12 0.76–1.64 Smoking Never smoker 209 (17.1%) Ref Ex-smoker 52 (4.3%) 5.45 1.41–21.00 Current smoker 680 (55.7%) 10.84 3.84–30.58 No information recorded 280 (22.9%) 2.79 0.79–9.82 Alcohol consumption Never 681 (55.8%) Ref Ever drinker 540 (44.2%) 2.19 1.51–3.17 Alcohol dependence Alcohol dependent 265 (49.1%) Ref Not alcohol dependent 275 (50.9%) 2.28 1.59–3.29 Alcohol consumption (units/week) Up to 14 196 (48.9%) Ref More than 14 205 (51.1%) 1.18 0.71–1.99 Alcohol related complications Delirium tremens/withdrawal 19 (1.56%) 4.1 1.39–12.12 Blood pressure measurement Normal (90–140 SBP) 810 (67.7%) Ref High (> 140 SBP) 131 (10.9%) 0.93 0.54–1.57 No measurement 255 (21.3%) 1.12 0.49–2.58 Illicit substance use Any illicit substance misuse 442 (36.2%) 5.86 3.58–9.58 Poly-substance misuse 272 (22.3%) 2.73 1.88–3.95 Cannabis 114 (9.3%) 1.77 1.11–2.83 Cocaine 194 (15.9%) 2.35 1.60–3.47 Heroin 257 (21.0%) 2.14 1.48–3.12 Opioid (not specified) 187 (15.3%) 2.19 1.47–3.26 Synthetic cannabinoids (mamba, spice etc.) 77 (6.3%) 1.74 1.01–2.99 Benzodiazepine 32 (2.6%) 5.45 2.41–12.31 Ecstasy (MDMA) 6 (0.5%) 6.72 1.15–39.20 Amphetamines 17 (1.4%) 4.81 1.62–14.24 Physical co-morbidities Respiratory 145 (11.9%) 1.84 1.20–2.81 Asthma 131 (10.7%) 1.96 1.26–3.04 COPD 29 (2.4%) 1.63 0.69–3.85 BBV 167 (13.7%) 2.19 1.46–3.28 Hepatitis C 157 (12.9%) 2.25 1.49–3.39 HIV 14 (1.1%) 0.71 0.18–2.73 Cardiovascular 16 (1.3%) 1.19 0.30–4.64 GI bleed or ulcer 74 (6.1%) 1.51 0.85–2.68 Type 2 diabetes 44 (3.6%) 0.38 0.12–1.13 Hypertension 71 (5.8%) 0.75 0.36–1.56 Leg/foot ulcer 51 (4.2%) 1.15 0.59–2.24 STI 15 (1.2%) 1.08 0.33–3.50 Migraine 44 (3.6%) 2.21 1.06–4.62 Mental health co-morbidities ADHD 21 (1.7%) 3.12 1.14–8.52 Depression 410 (33.6%) 1.79 1.26–2.55 Anxiety 276 (22.6%) 2.27 1.59–3.26 PTSD 62 (5.1%) 1.75 0.90–3.42 Bipolar 21 (1.7%) 1.14 0.43–3.03 Personality disorder 82 (6.7%) 4.92 2.96–8.17 Psychosis (inc Schizophrenia) disorder 85 (7.0%) 1.38 0.80–2.38 Drug-related mental and behaviour 55 (4.5%) 1.38 0.73–2.63 Alcohol related 20 (1.6%) 3.69 1.28–10.61 Recorded self-harm or suicide attempt 185 (15.1%) 3.53 2.38–5.22
Being an ex or current smoker increased the odds of a NFOD by 5.45 (95% CI 1.41–21.00) and 10.84 (95% CI 3.84–30.58) times, respectively, as did being ever reported as an alcohol drinker (OR: 2.19, 95% CI 1.51–3.17). However, being non-alcohol dependent increased the likelihood of NFOD compared to a reported alcohol dependence (OR: 2.28, 95% CI 1.59–3.29).
Having any illicit substance use substantially increased the odds of having a NFOD (OR: 5.86, 95% CI 3.58–9.58). Records of cannabis (OR: 1.77, 95% CI 1.11–2.83), cocaine (OR: 2.35 95% CI 1.60–3.47), heroin (OR: 2.14, 95% CI 1.48–3.12), opioids not specified (OR: 2.19 95% CI 1.47–3.26) and synthetic cannabinoids (OR: 1.74, 95% CI 1.01–2.99) were positively associated with NFOs.
In general, physical or mental health conditions increased the risk of an NFO, including the presence of a respiratory condition (OR: 1.84, 95% CI 1.20–2.81), particularly asthma (OR: 1.96, 95% CI 1.26–3.04), or a blood-borne virus (OR: 2.19, 95% CI 1.46–3.28) or known migraines (OR: 2.21, 95% CI 1.06–4.62). Further risks included having diagnosed anxiety (OR: 2.27, 95% CI 1.59–3.26), depression (OR: 1.79, 95% CI 1.26–2.55), ADHD (OR: 3.12, 95% CI 1.14–8.52) or a personality disorder (OR: 4.92, 95% CI 2.96–8.17).
Discussion
Summary of key findings
This study demonstrates a high proportion (15.9%) of PEH registered at a specialist primary care centre had a recorded NFOD. Risk factors identified for these overdoses included an increased risk in women, those of white ethnicity and those aged 36–45, the use of tobacco and alcohol, having a known substance misuse disorder and having at least one chronic health condition, specifically respiratory, blood borne viruses, migraines, anxiety and depression. The variety of substances used to overdose combined with the qualitative information suggests two types of overdoses: accidental overdose in street substance users and self-harm attempts. These overdoses frequently required help from emergency services, from which a significant number of patients self-discharged.
Interpretation
The two types of overdose identified, accidental overdose and overdose with suicidal or self-harm intent, are not often considered concurrently. However, both are large contributors to fatality amongst PEH. In England and Wales in 2021, 35% of deaths in PEH were caused by substance overdoses and 13.4% by suicide [2]. These two types of overdose are more likely to be at ends of a spectrum with considerable overlap rather than being distinct classes, with suicidal thoughts shown to be present in almost 50% of illicit NFOD [17] and suicidal ideation a risk factor for accidental overdose [18].
The mixture of illicit, over the counter, and prescription drugs reported in the NFOD most likely reflect the combination of these two types of overdoses, with paracetamol most commonly used for self-harm attempts in PEH [19], and heroin, benzodiazepines and cocaine frequent in street drug overdoses [10]. The importance of considering the type of overdose lies in the preventative approach from a health care perspective. Although sharing similarities that may benefit from comparable holistic and mental health care, there may be some variation in the approach to dealing with patients in each group. For example, ensuring naloxone for those likely to overdose on heroin vs limiting the prescribing of medicines potentially associated with overdose. Alternative approaches are in place in other UK cities with opportunities for heroin assisted treatment and supervised drug consumption in Glasgow, for example.
Alongside factors previously identified to be associated with NFOD including being female [9], a polysubstance user [9], being white [20] and known substance misuse, the most novel risk factors identified related to the association with chronic health conditions. This is already recognised from a mental health perspective, with mental health disorders shown to be the second highest risk factor for overdose behind substance use in a meta-analysis [21]. Chronic physical health conditions have not been previously considered a risk factor, but we do know there is a high level of multi-morbidity within this population, with a mean of 2.2 mental health problems and 5.4 physical health problems per person reported in a group of PEH with a recent NFOD [10].
Respiratory disease (particularly asthma), BBV (particularly Hep C) and migraines were associated with an increased risk of NFOD. While the timing of diagnosis and overdose events remains unknown, thus making the direction of these relationships unclear, chronic health conditions come with a large burden to PEH physically, mentally and socially [22].
Most overdose incidents reported here (87%) had contact with emergency services (ambulance or the emergency department), although this is likely an overestimation if considering all episodes of overdose in the community as many overdoses are likely to unrecorded where no contact with health services were made. Just over half (53%) arrived by ambulance. Self-discharge from emergency services was high, with 14% of those that attended ED self-discharging and 19% of those that had an ambulance call out refusing help. Similar reports of self-discharge from the ED have been documented [11]. Self-discharge following a poisoning episode could be associated with an increased risk of death [23]. Waiting environment, the length of time of the wait and waiting around other people in the emergency department could contribute to the self-discharge before assessment [24].
Strengths and weaknesses
This study is novel considering existing literature focus mainly on fatal overdose. However there are some limitations. While a large number of patient data and non-fatal overdose events were analysed, data originated from one specialist homelessness general practice and this can have limited generalisability. The recorded NFOD here only captures those that had medical attention, but it is likely a number of NFOD happen without medical attention in the community and in the streets. The EMIS system used to identify NFOD only allowed searching of coded instances of NFOD, meaning some that were not coded were missed from the evaluation. Furthermore, the free text data in discharge letters was analysed by only one researcher (a final year graduate medical student with a PhD degree), thus the reliability of this was not assessed. There were also missing data on the circumstances surrounding the overdoses. Finally, the population included at the practice also includes asylum seekers (14%) and it is worth noting that this population will vary in terms of their health care needs and less is known about substance use patterns in this population.
Further research
Future research should identify the direction and temporal association between physical and mental health conditions and overdose as well as consider the relationship between treatment burden and overdose. A greater understanding surrounding the psychopathology of mental health conditions in PEH such as anxiety and depression could help target primary prevention strategies while extending into more acute stressors could also improve our understanding around timings of overdose. Overall prevention strategies need to consider the type of any previous overdose, if present, and target those with risk factors for NFOD, inclusive of chronic mental and physical health conditions. From a medication perspective, this should focus on promoting user understanding around risks related to overdose from both prescribed and illicit substances. Language, health literacy and cultural barriers can often hinder effective communications between healthcare professionals and PEH leaving people at risk of medication misadventures and overdose due to lack of adequate knowledge and understanding. Future intervention should focus addressing these factors on an individual basis, but more substantially from a population perspective, such as the use of social media to target younger PEH, particularly in relation to novel psychoactive substance and their risks [25]. Given that NFOD are associated with future overdose mortality [8], research and practice based on multisector collaboration is key to document and apply prevention approaches to NFOD wherever identified including in the community, streets, temporary accommodations. When considering prevention approaches, it is important to consider wider determinants of health including barriers of PEH access to primary care such as experiences of stigma and discrimination in public health and healthcare setting, exclusion due to lack of proof of address, and perceived fragmentation of services [26–30].
Conclusion
Female sex, substance use disorder (legal and illegal substance) and chronic mental and physical comorbidities increased the odds of a NFOD in PEH. Identifying people with NFOD and awareness of risk factors can enable primary care providers to apply prevention actions such as provision of overdose awareness and naloxone to avoid future fatal overdose events. Future work should explore the role of chronic physical conditions and their treatment on NFOD risk.
Supplementary Information
Below is the link to the electronic supplementary material.
Acknowledgements
We would like to acknowledge the staff at the specialist homeless health centre for facilitating access to the data.
Funding
There is no funding associated with this study.
Conflicts of interest
Vibhu Paudyal is an Associate editor of the International Journal of Clinical Pharmacy. He had no role in handling the manuscript, specifically the processes of editorial review, peer review and decision making.