This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic.
Not applicable. (All studies included in the review are available publicly.)
This systematic review summarizes and presents the current state of research quantifying the relationship between mental disorder and overdose for people who use opioids.
The protocol was published in Open Science Framework. We used the PECOS framework to frame the review question. Studies published between January 1, 2000, and January 4, 2021, from North America, Europe, the United Kingdom, Australia, and New Zealand were systematically identified and screened through searching electronic databases, citations, and by contacting experts. Risk of bias assessments were performed. Data were synthesized using the lumping technique.
Overall, 6512 records were screened and 38 were selected for inclusion. 37 of the 38 studies included in this review show a connection between at least one aspect of mental disorder and opioid overdose. The largest body of evidence exists for internalizing disorders generally and mood disorders specifically, followed by anxiety disorders, although there is also moderate evidence to support the relationship between thought disorders (e.g., schizophrenia, bipolar disorder) and opioid overdose. Moderate evidence also was found for the association between any disorder and overdose.
Nearly all reviewed studies found a connection between mental disorder and overdose, and the evidence suggests that having mental disorder is associated with experiencing fatal and non-fatal opioid overdose, but causal direction remains unclear.
Keywords: Opioids, Mental illness, Psychiatric disorder, Toxicity, Drug-related harmOpioid toxicity deaths, commonly referred to as opioid overdose has become a global crisis [1, 2] that has intensified during the COVID-19 pandemic [3]. Some evidence exists to suggest that opioid overdose is commonly experienced by individuals with co-occurring mental disorder [4–7] but literature that examines these two phenomena together is sparse. This review summarizes the evidence on mental disorder and the risk of opioid overdose to advance understanding of this relationship.
Although past research indicates that individuals with mental disorders are at increased risk for overdose [8–12], there are multiple considerations that have precluded definitive conclusions from being made in this area. First, disentangling the symptoms of substance use disorders, other non-substance use psychiatric disorders, and medication side effects can be challenging, which complicates the identification of the relationship between mental disorder and opioid overdose. For example, individuals with mental disorders are often prescribed opioids or other psychotropic medications [13] that may interact or cumulate with non-prescription or illicit opioids to increase overdose risk. In addition, the historical expansion of diagnostic scope for mental disorders may contribute to over-pathologizing drug use, criminality, and social deviance [14]. Thus, there is difficulty in distinguishing between the presence of a mental disorder and behaviors attributable to the use of drugs, leading to imprecision in the evidence base on mental disorder and opioid use [15].
It is also unclear whether there is an identifiable causal relationship between mental disorder and opioid overdose. While stigma, stress and social exclusion commonly experienced alongside opioid use may produce poor psychological outcomes, opioids may also be used as a form of self-medication for coping with emotional pain or mental disorder symptoms [16]. Otherwise stated, the direction of causality and potential moderators in the relationship are not well understood. Another possibility in discussions about causality, and one consistent with a framework introduced by Dasgupta and colleagues [17], would reflect a process of social causation (also known as indirect selection) [18, 19], where a third factor or set of factors produces suboptimal outcomes in both areas. In this case, where opioid use and mental disorder are both hypothesized to be occurring within processes of social causation, both outcomes may be responses to social and economic precarity [16]. Importantly, scholars have also proposed a countervailing theory of social selection or social drift [20] to explain the connection between mental disorder and socioeconomic marginalization which argues for the reverse: that experiencing mental disorder causes a downward shift in social class. Given these complex pathways, a fulsome understanding of these relationships is unlikely to produce a unidirectional explanation, these pathways likely co-exist, each contributing to understandings of the complexity of the relationship between opioid overdose and mental disorder.
Despite the clear need for conclusive evidence about the intersections between mental disorder and opioid overdose, this relationship has not been clarified or summarized in a systematic way. Specifically, gaps exist in the literature in: 1) establishing the volume and strength of literature supporting the connection is between mental disorder and overdose risk; and 2) understanding the empirical evidence that exists to support the theoretical relationships hypothesized between these phenomena.
The systematic review of the extant literature on mental disorder and opioid overdose presented here began as part of a larger investigation of the literature related to socioeconomic marginalization and opioid overdose, in which we found a preponderance of scientific evidence pointing to the potential role of mental disorder [21]. This systematic review employed an integrated knowledge translation process [22] whereby we partnered with decision makers in establishing the focus of the review, refining review questions and methodology, data retrieval and retrieval tool development, interpretation of review findings, identification of gaps, crafting of recommendations, and dissemination and application of review findings. Given our initial findings, the need for an independent review on mental disorder and overdose specifically was informed by decision-makers in the local community, municipal, provincial, and federal government who required further clarity regarding the evidence on mental disorder and overdose risk. This review responds to that need and systematically summarizes the literature on whether or not broadly defined mental disorder (including both symptoms of disorder and diagnosed mental disorder) are associated with opioid overdose. The review was designed to address the above omissions, summarize existing published research, provide a knowledge base for effective prevention and response strategies, and identify future directions for policy in this area.
Using the Systematic Reviews and Meta-Analyses (PRISMA) checklist, the search sought studies that included measures of mental disorder and opioid-related fatal and non-fatal overdose published in English peer-reviewed journals or by governmental sources between January 1, 2000 and January 4, 2021. Only studies conducted in North America, Europe, the United Kingdom, Australia, and New Zealand were eligible for inclusion, as study stakeholders expressed a desire for a review of studies with the highest contextual and policy comparability. The specific terms related to mental disorder included in the search strategy are summarized in Table Table1, 1 , and a summary of the Medline search terms used in the review is provided in Appendix A. The PECOS framework (Population, Exposure, Comparison, Outcome, Study Design) used to frame our specific mental disorder and opioid overdose research question [23] is listed in Table Table1. 1 . Two research assistants first conducted title and abstract screening, and subsequently independently assessed full text articles to determine final inclusion eligibility. Differences in opinion about inclusion were resolved by a senior team member. More details about study selection are provided in the protocol, published on the Open Science Framework site [24].
Mental Health: mental health, mental illness, psychiatric disorder, health care access, social service access; despair, anguish, disability, vulnerability, stigma, social isolation, social exclusion, marginalization
Overdose (fatal and non-fatal): poisoning, drug-related poisoning, side-effects/adverse reactions, toxicity, death, morbidity, mortality, overdose
Opioids: People who use opioids (medical/non-medical), prescription and non-prescription, oral and injection
Population: People who use opioids in North America, Europe, the United Kingdom, Australia, and New Zealand. Articles were only included if they had opioids (any opioid, including opioid agonist therapy) identified as a cause of overdose. Poly drug-related overdose papers were included if they included opioid overdose in the cases
Exposure: Any measure of mental health as an independent variable in the article (e.g., despair, hopelessness, psychological distress, psychiatric disorder). Articles were included if they had any measure of mental health, broadly defined, as an independent variable in the article. Articles that include mental health variables as controls in their multivariable regression models, for example, were included as long as there were empirical results that showed the effects of the mental health variables on opioid overdose
Comparison: Quantitative studies with comparisons between groups with different levels of mental health
Outcomes: Opioid-related fatal and non-fatal overdose. Articles were included only if they had overdose as a unique/isolated outcome. Articles examining drug-related harm or mortality might include overdose but also include death or harm from other factors (e.g., motor vehicle accidents) and as such, were excluded. Articles comparing the risk of overdose between different types of opioids were excluded, as they do not address the review question of whether mental health issues increase the risk of overdose
Study design: Any study design including quantitative data. Articles that contained empirical data were included. Case-reports, letters, commentaries, reviews, and editorials were excluded
1 Terms related to these key concepts were entered into all computer databases, combined using appropriate Boolean operators. All terms were searched both as subject headings as well as keywords. See Appendix A for a summary of the Medline search terms included
Two research assistants used a standardized form to extract data from the included studies, including year, journal, and type of publication, author’s name, study location, sample, study period, study design, recruitment and sampling, sample size, response rates, age range, operationalization of mental disorder, type of opioid overdose outcome, and statistical analysis. Inconsistencies in the extracted data were noted by the research assistants and resolved through discussion with a senior team member.
This review approached the relationship between mental disorder (including both diagnosed disorder—medically diagnosed, assessed, or self-reported—and symptoms of disorder) and opioid overdose broadly and purposely allowed inclusion of different study types, as long as they provided evidence on the review question regarding whether mental disorder and associated symptoms are associated with increased occurrence of opioid overdose (including fatal and non-fatal overdose, overdose-related hospitalizations, and intentional vs. unintentional overdose). As a result, we employed the lumping synthesis technique where all evidence related to mental disorder and opioid overdose was included despite differences between study design and outcome measures [25]. This strategy allows for synthesis and identification of common findings in the relationships between mental disorder and opioid overdose that remain despite minor differences in study participants, context, and design, similar to the approach taken by others investigating determinants of overdose [26]. Substantial conceptual and methodological heterogeneity across the included studies precluded meta-analysis. Instead, findings are summarized by mental disorder variables. The data extraction process for this review included an assessment of bias, for which we used the study quality tools of the National Heart, Lung, and Blood Institute (presented in Table Table3) 3 ) [27]. Two independent reviewers assessed risk of bias.
Study design and sample characteristics of included studies
60 + : 43.9%; Sex: 93.3% male
Veterans Health Administration (VHA) National Patient Care Database (2004–2008)
Linked to National Death Index
Therapeutic community program participants
Completed survey for the Psychiatric and Addictive Dual Disorders in Italy (PADDI- TC) project (2010)
Prescription Pain Medication Dataset from the Utah Department of Health (2008–2009)
Office of the Medical Examiner
Linked to the Labour Commission database
N: 2,752,612; Age: Mean (SD): 17.2 (2.5);
Consortium to Study Opioid Risks and Trend (CONSORT) Study participants (1997–2005)
Linked from automated health care data, electronic medical records, and medical-record reviews
Wide-ranging Online Data for Epidemiological Research from the Centers for Disease Control and Prevention
Linked to Behavioral Risk Factor Surveillance System (BRFSS)
Personality data from adults who responded to the 44-item Big Five Inventory online
Linked to Centers for Disease Control Wonder Database
N: 14,898; Age: Mean (SD): 56.3 (16.0);
Oregon Medicaid claims data
Linked to vital statistics and prescription drug monitoring program (PDMP) data
≥ 56: 76.7%; Sex: 93.3% male
White: 73.3%; Black: 19.1%;
Hispanic: 5.2%; Not Hispanic: 94.8%
N: 1048; Age: Median (IQR):
[Control: 39 (31–45)]; Sex: 62.2% male
Office of the Chief Coroner of Ontario
Linked to Canadian Institute for Health Information Discharge Abstract Database (CIHI-DAD)
N: 1359; Age: Median (IQR): [Opioid Deaths: 44 (35–51)]
[Non-opioid deaths: 46 (37–54)];
N: 1650; Age: Median (IQR): [Case: 49 (40–55)]
[Control: 45 (38–52)];
Non-Hispanic White: 5101 (56.8%)
Non-Hispanic Black: 1383 (15.4%)
Hispanic: 463 (5.2%)
N: 61,170; Age: Mean (SE):
46.8 (0.19); Sex: 49.9% male
N:1284; Age: Mean (SD): [Case: 44.5 (11.2)]
[Control: 44.9 (11.6)]; Sex: 36.9% male
Oklahoma state Medicaid pharmacy and medical claims
Linked to medical examiner reports from the Oklahoma State Department of Health’s (OSDH) Fatal Unintentional Poisoning Surveillance System
Age: Median: 34; Sex: 86.2% male
Prison release data from the NC Department of Public Safety
Linked to NC death records from the NC Division of Public Health
Age: 14–39: 36.4%, 40–70 + : 63.6%; Sex: 50.1% male
N: 59,784; Age: Mean (median): [Male: 39 (37)]
[Female: 39 (36)]; Sex: 67.0% male
N:8987; Age: Median (IQR): [Case: 62 (10)]
[Control: 62 (16)]; Sex: 92.1% male
a Panel data are multi-dimensional data involving measurements over time
The number of studies retained in each step of the review process can be found in the PRISMA flow diagram in Fig. 1 . The primary search strategy found 7200 original articles that met the initial screening criteria. The review and screening process led to a final dataset of 38 articles on mental disorder and overdose.
PRISMA flow diagram
Eleven studies had fatal overdose as the outcome or drew their sample from a population who had experienced fatal overdose [28–38]. Thirteen included only non-fatal overdose as the sole overdose outcome in their studies [15, 39–50], and twelve articles included both fatal and non-fatal overdose [51–60, 63, 64]. Two studies included data from emergency department (ED) visits and hospitalizations for overdose but did not specify whether the overdoses were fatal or not [61, 62]. Most studies either did not report intention of overdose (n = 17) [15, 33, 35, 39–41, 46, 48, 51–53, 58, 59, 61–64] or examined both intentional and unintentional overdose (n = 17) [30–32, 36–38, 42–45, 47, 49, 54–57, 60] with four examining only unintentional overdose [28, 29, 34, 50]. Many studies (n = 20) included overdose that was attributable to prescription and non-prescription opioids [29, 32, 35–37, 41–47, 49, 50, 56–58, 60, 63, 64], while eleven investigated only prescription opioid overdoses [28, 33, 34, 38, 48, 51–54, 61, 62] and two investigated only non-prescription opioid overdoses [15, 39]. Five studies did not report what type of opioids were included in their data [30, 31, 40, 55, 59]. The included studies analyzed data from the United States (n = 28) [28–32, 34, 35, 42–46, 48–59, 61–64], Australia (n = 4) [15, 36, 39, 47], Canada (n = 4) [33, 37, 38, 60], Italy (n = 1) [40] and Spain (n = 1) [41]. Study designs included nine cross-sectional analyses [15, 29, 37, 39, 40, 42, 45, 49, 50], fourteen cohort studies [32, 35, 41, 43, 44, 46, 51–55, 58, 59, 63], eight case–control studies [33, 34, 38, 47, 56, 57, 60, 64] two studies that used nested case–control designs [48, 61] two ecological studies [30, 31] one study that used panel data [36] and one study that used a case-cohort design [28]. A summary of the study design, data sources, and sample characteristics for the 38 included studies can be found in Table Table2 2 .
The majority of the 38 included studies operationalized mental disorder through diagnosed mental disorder, with only six studies reporting on symptoms of disorder [29, 38–40, 42, 45]. Overall, 37/38 of these studies found a significant association with the mental disorder and opioid overdose variables used, with the direction of the association suggesting that people experiencing mental disorder and associated symptoms were more likely to overdose than those who were not. Only one study presented results in the opposite direction [32]. A summary of measures and findings for the 38 included studies can be found in Table Table3 3 .
Summary of measures and findings for included studies
Mental disorders in individuals with chronic pain: AHR 1.87, 95%CI 1.48–2.38
Mental disorders in individuals with acute pain: AHR 1.77, 95% CI 1.19–2.65
Mental disorders in individuals with substance use disorder: AHR 1.73, 95% CI 1.10–2.72
No significant associations were found between mental disorders and overdose in individuals with cancer
Hopelessness: Wald’s statistic 6.12, p < 0.01
Antisocial behaviour: Wald’s statistic 8.21, p < 0.01
Prior mental illness: Wald’s statistic 4.15, p < 0.05
No significant associations were found between depression, reported self-harm and overdose
Mental disorder in opioid overdose decedents vs. poor mental disorder estimates for state-level population: 50.0% vs. 15.0%, p < 0.001
‘‘Poor mental disorder’’ in the population defined as stress, depression and problems with emotions for more than 7 days in the past 30 days
Adjustment disorders, Personality disorders, other mental disorders
Anxiety disorder: ARR 1.26, 95% CI 1.07–1.48
Mood disorders: ARR 1.29, 95% CI 1.12–1.50
No significant associations were found between adjustment disorders, personality disorders, other mental disorder disorders and overdose
Mental disorder in opioid overdose history vs. no opioid overdose history, 72.2% vs. 50.0% p-value = 0.013
History of suicide attempt, 50.0% vs. 17.2% p-value = 0.001
No significant associations were found between anxiety symptoms and overdose
Anti-Social Personality Disorder
Bipolar Disorder: OR 3.37, 95% CI 1.44–7.88
Anti-Social Personality Disorder: OR 2.20, 95% CI 1.15–4.21
BPD and ASPD: OR 4.70, 95% CI 2.54–8.69
Severe depression: AOR 2.46, 95% CI 1.24–4.89
PTSD: AOR 2.77, 95% CI 1.37–5.60
Psychosis: AOR 2.39, 95% CI 1.10–5.15
No significant associations were found between severe anxiety and overdose
Anxiety: AOR 1.24, 95% CI 1.12–1.36, p < 0.001
No significant associations were found between mood, PTSD and overdose
Anxiety: AHR 1.65, 95% CI 1.33–2.06, p-value < 0.0001
Mood disorders: AHR 2.77, 95% CI 2.26–3.34, p-value < 0.0001
Opioid -related near-fatal events involving mechanical ventilation
Psychoses were significantly associated with opioid-related hospitalizations (for those who have an opioid-related ED visit): AOR 5.40, 95% CI 4.85–6.00, p < 0.001and with frequency of ED visits: AOR 1.44, 95% CI 1.25–1.65, p < 0.001
Depression is significantly associated with opioid-related hospitalizations (for those with an opioid-related ED visit): AOR 2.71, 95% CI 2.46–2.97, p < 0.001
No significant associations were found between depression and frequent ED visits for overdose
Any mental disorder condition
Anxiety: Among adults with non-fatal opioid overdose from 2009 to 2016, those with a second overdose were more likely to have anxiety compared to those without a second overdose (47.0% vs. 39.5%, p = 0.013)
Bipolar disorder: Among adults with non-fatal opioid overdose from 2009 to 2016, those with a second overdose were more likely to have bipolar disorder compared to those without a second overdose (22.3% vs. 14.8%, p < 0.001)
No significant associations were found between depression, panic disorder, schizophrenia, any mental disorder and opioid overdose
Those with persistent overdoses were more likely to have a diagnosis of PTSD (AOR 3.84, 95% Cl l.41–10.46, p = 0.01) than those who have never overdosed
No significant associations were found between suicidal ideation and opioid overdose
Schizophrenia and other
Depression: AOR 1.29, 95% CI 1.09–1.53
Anxiety: AOR 1.65, 95% CI 1.37–1.97
Bipolar disorder: AOR 1.51, 95% CI 1.28–1.79
PTSD: AOR 0.73, 95% CI 0.58–0.92
Those who had fatal opioid overdose were more likely to have mental disorders (depression, 56.8% vs. 30.2%; anxiety, 63.4% vs. 30.9%; bipolar disorder, 35.6% vs. 14.0%; schizophrenia and other
psychotic disorders, 18.8% vs. 11.5%; ADHD, 12.0% vs. 4.6%; personality disorders, 9.8% vs. 4.1%)
Serious mental illness
Opioid-overdose hospitalizations vs. Non-Opioid-related hospitalizations:
Depression (56.4% vs 36.7%) p < 0.0001*
Serious mental illness (24.2% vs 9.7%) p < 0.0001*
PTSD (30.4% vs 19.0%) p < 0.0001*
Anxiety (22.3% vs 15.1%) p < 0.05*
The p values given here are for all opioid-related hospitalizations (including abuse, dependence and overdose) vs. non-opioid-related hospitalizations
Mood disorder: AOR 1.80, 95% CI 1.00–3.24
No significant associations were found between schizophrenia and opioid overdose deaths
Screening for Bipolar disorder
There was no significant difference between those with no overdose or suicide attempt and those with history of opioid overdose only (without suicide attempt) for anxiety disorder, depressive episode, and screening for bipolar disorder
Case subjects were more likely to have had anxiety disorders (63.2% vs. 54.9%, SD = 0.13), affective disorders (11.2% vs. 5.6%, SD = 0.19), received psychiatric care (34.7% vs. 27.5%, SD = 0.16), and have a history of self-harm (8.5% vs. 2.5%, SD = 0.26) in the year preceding opioid-related suicide
No significant associations were found between psychotic disorder and opioid-related suicide
Depressive Disorder Bipolar disorder
Depression in Commercial insured population (CIP): AOR (95% CI) 3.12 (2.84–3.42)
Bipolar disorder in CIP: AOR (95% CI) 2.18 (1.83–2.60). Bipolar disorder in Veterans Health Administration (VHA): AOR (95% CI) 1.68 (1.17–2.43). Schizophrenia in CIP: AOR (95% CI) 2.06 (1.17–3.69). Anxiety disorder in CIP; AOR (95%CI) 1.64 (1.50–1.80). ADHD in VHA: AOR (95% CI) 0.33 (0.11–0.99)
No significant association between PTSD/OCD and overdose in Commercial insured or VHA populations
Major depressive disorder
Bipolar disorder: AOR 1.78, 95% CI 1.20–2.65
Schizophrenia: AOR 1.15, 95% CI 1.04–1.27
Those who received in-prison mental disorder treatment were more likely to experience opioid overdose death after release (AHR 1.9, 95% CI 1.7–2.2) and 1 year after release (AHR 2.2, 95% CI 1.7–2.9)
No significant associations were found between in-prison mental disorder treatment and opioid overdose death 2 weeks after release
Mental health problems in deaths where codeine toxicity was a contributory factor: Intentional vs. accidental OD: OR (95% CI) 2.1 (1.6–2.7)
Mental health problems defined as any history of mental health problems recorded in the National Coronial Information System
OUD diagnosis and overdose in the year before or after delivery vs. OUD diagnosis but no overdose in the year before or after delivery vs. no diagnosis of OUD in the year prior
Anxiety: 82.1% vs. 60.2% vs. 18.29%, p < 0.001
Depression: 84.8% vs. 61.2% vs. 18.90%, p < 0.001
Major depressive disorder
Prevalence of diagnosed mental disorders in the past year and in the past 5 years was significantly different between overdoses cases and controls
Past year prevalence:
Past 5-year prevalence:
