Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2022 Jun 6.
Published in final edited form as: Addict Behav. 2021 Dec 20;127:107215. doi: 10.1016/j.addbeh.2021.107215

Identifying pathways to recent non-fatal overdose among people who use opioids non-medically: how do psychological pain and unmet mental health need contribute to overdose risk?

Catherine Tomko a,b,*, Kristin E Schneider b, Saba Rouhani a, Glenna J Urquhart a, Ju Nyeong Park a, Miles Morris a, Susan G Sherman a
PMCID: PMC9169246  NIHMSID: NIHMS1803162  PMID: 34953432

Abstract

Background:

Significant associations exist between psychological pain, unmet mental health need, and frequency and severity of substance use among people who use drugs (PWUD), but no studies have analyzed the relationship of these variables to non-fatal overdose.

Methods:

We conducted a cross-sectional survey of people who used opioids non-medically in Baltimore, Maryland (n=563) as part of a broader harm reduction-focused evaluation (PROMOTE). The outcome was self-reported recent (past 6 months) non-fatal overdose; exposures of interest were recent self-reported unmet mental health need, experiencing daily “long-lasting psychological or mental pain” (vs. <daily), and daily multi-opioid use (vs. none/one opioid used). Path analysis was used to model direct relationships between these variables, personal characteristics (race, gender, experiencing homelessness, drug injection) and overdose.

Results:

30% of the sample had experienced a recent non-fatal overdose, 46% reported unmet mental health need, 21% reported daily psychological pain, and 62% used multiple types of opioids daily. After adjusting for covariates, daily multi-opioid use (aOR=1.78, p=0.03) and unmet mental health need (aOR=2.05, p=0.01) were associated with direct, significant increased risk of recent overdose. Significant pathways associated with increased odds of unmet mental health need included woman gender (aOR=2.23, p=0.003) and daily psychological pain (aOR=4.14, p=0.002). In turn, unmet mental health need associated was with greater odds of daily multi-opioid use (aOR=1.57, p=0.05).

Discussion:

Unmet mental heath need and daily psychological pain are common experiences in this sample of PWUD. Unmet mental health need appears on several pathways to overdose and associated risk factors; improving access to mental healthcare for PWUD (particularly women) expressing need may be an important harm reduction measure.

Keywords: opioids, overdose, substance use, mental health, mental healthcare, path analysis

1. Introduction

Drug overdoses have increased dramatically in the past two decades in the United States (U.S.), and early data suggest they have soared an estimated 24% amid the COVID-19 pandemic (Holland et al., 2021; Slavova, Rock, Bush, Quesinberry, & Walsh, 2020; Wilson, Kariisa, Seth, Smith, & Davis, 2020). During the pandemic, there has been an unprecedented influx of mental health stressors, worsening social and economic conditions, and interrupted access to health services (Holland et al., 2021; McGinty, Presskreischer, Han, & Barry, 2020). These changes are increasingly being linked to a cascade of health consequences, including substance-use related harms (Jones, Guy Jr, & Board, 2021). Provisional data from the Centers for Disease Control and Prevention estimate more than 83,000 deaths were reported from July 2019-June 2020, underscoring the potential impacts of mental health on drug use-related harms in the past year (Centers for Disease Control and Prevention, 2021).

In addition to the influence of complex environmental factors, substance use may be understood through the self-medication hypothesis, which posits that individuals often use drugs and alcohol to alleviate feelings of psychological distress rather than as a self-destructive act or a “moral failure” (Khantzian, 1985, 2003). For people who use drugs (PWUD), substance use may alleviate mental distress (before it escalates to diagnosed mental illness), including distress associated with structural and interpersonal stigmas, discrimination, and chronic stressors such as housing instability. There is support for the self-medication hypothesis in individuals with diagnosed mental illness or dually-diagnosed patients meeting clinical criteria for substance use disorder and mental illness (Khantzian, 2003; Martins, Sampson, Cerda, & Galea, 2015; Priester et al., 2016; Robinson, Sareen, Cox, & Bolton, 2011). There is also support for the self-medication hypothesis in individuals who are undiagnosed or have sub-syndromal symptoms (i.e., exhibiting elevated levels of distress without meeting diagnostic criteria), though these studies have been limited (Robinson et al., 2011; Schwartz et al., 2006). Self-medication through the use of drugs or alcohol is often linked to traumatic life events or stressors, such as interpersonal violence or criminal victimization, and may drive a “feedback loop” between substance use and mental distress that can perpetuate both (Garland, Pettus-Davis, & Howard, 2013). Adverse childhood experiences have been found to predict initiation of opioid pill use among young adults, with self-medication guiding hypotheses about the reasons for this relationship (Guarino et al., 2021). Early evidence from studies of patients prescribed opioids for non-cancer pain suggests that psychological factors, such as depression and anxiety, may be more strongly associated with opioid misuse than chronic pain (Grattan, Sullivan, Saunders, Campbell, & Von Korff, 2012; Martel, Edwards, & Jamison, 2020).

Consistent with the self-medication hypothesis, unmet mental health need may contribute to greater substance use and be a possible pathway in the relationship between mental distress and overdose. The National Survey on Drug Use and Health (NSDUH) defines their measure of unmet mental health need as an individual “feeling a perceived need” for mental health care but not being able to receive it. The implication of this definition is that an individual has made some assessment of their own need and expressed some level of desire for care (Substance Abuse and Mental Health Services Administration, 2019). Unmet mental health need is common in the general population, owing to barriers such as cost, transportation, mental illness stigma, and long waits or difficulty scheduling and keeping appointments (Alang, 2015; Mojtabai, 2009; Mojtabai et al., 2011). These barriers are amplified among PWUD relative to other groups (Mojtabai, Chen, Kaufmann, & Crum, 2014; Priester et al., 2016; Verduin, Carter, Brady, Myrick, & Timmerman, 2005). In one study of people with opioid use disorder, individuals reporting the most serious mental illness also reported the highest levels of unmet mental health need (Novak, Feder, Ali, & Chen, 2019). Structural impediments to healthcare for PWUD can increase mental health inequities and contribute to a risk environment for drug use-related harms, including overdose. For example, using behavioral healthcare may require abstinence from drugs, PWUD may feel unwelcome or stigmatized from clinicians or other patients, or there may be high out-of-pocket costs or low capacity for new patients (Novak et al., 2019; Yang, Roman-Urrestarazu, McKee, & Brayne, 2019). It is important to consider the relationship between unmet mental health need and overdose in a community-based sample of PWUD because these groups have known disparities in mental health outcomes yet common marginalization from the mental healthcare system may prevent their inclusion in studies restricted to clinical samples (Priester et al., 2016).

Nevertheless, there are several gaps in the literature. First, while research has documented the positive associations between poor mental health (or mental illness), unmet mental health need, and the severity of substance use (Smith et al., 2017; Yang et al., 2019), the relationship between unmet mental health need and non-fatal overdose has not yet been explicitly enumerated Second, relationships between social determinants of mental health (e.g., gender), substance use, and non-fatal overdose are complex and the existing literature lacks research that describes specific and direct pathways between them in order to clearly identify targets for intervention on overdose prevention. Non-fatal overdose is associated with a host of physical morbidities following overdose, including cognitive impairments and neuropathy, and deleterious mental health impacts including symptoms of post-traumatic stress disorder (Schneider et al., 2020; Warner-Smith, Darke, & Day, 2002). Non-fatal overdose is also a strong predictor of future fatal overdose, making recognition of pre-cursor events, such as non-fatal overdose experiences, critical for fatal overdose prevention (Caudarella et al., 2016). Finally, the few studies of unmet mental health need and substance use have used data from large, national surveys of the general U.S. population, including the NSDUH, which does not sample individuals experiencing homelessness or housing instability, a population experiencing substantial substance use and mental health burden (Ayano, Belete, Duko, Tsegay, & Dachew, 2021; Ayano, Solomon, Tsegay, Yohannes, & Abraha, 2020; Padgett, Henwood, Abrams, & Drake, 2008). Data from a community-based sample of PWUD is a unique opportunity to understand the relationship between mental health need and overdose in a sample comprised of the individuals at highest-risk of overdose.

To address this gap, we examined associations between psychological pain, unmet mental health need, and non-fatal overdose among a sample of people who used opioids non-medically in Baltimore, and conducted a path analysis to further elucidate pathways between them. We hypothesized that unmet mental health need would have a significant direct path to overdose, and also lie on the pathway between psychological distress, substance use, and overdose.

2. Methods

2.1. Study setting

Baltimore, Maryland is one of the hardest-hit U.S. cities in the opioid crisis, with an age-adjusted overdose mortality rate of 57 per 100,000 residents in 2018; this was the highest such rate per county in a state that ranked third overall in overdose fatalities (Maryland Department of Health, 2018; Wilson et al., 2020).

2.2. Recruitment and eligibility

The Peer harm Reduction of Maryland Outreach Tiered Evaluation (PROMOTE) was a multi-method cross-sectional study of PWUD in Baltimore City and Anne Arundel County, Maryland. This analysis included data collected in Baltimore City. Participants were recruited from 15 street-based locations within two distinct waves of data collection (July – October 2018 and April – July 2019). Recruitment sites were developed using publicly available 2017 drug arrest data from Baltimore’s Central Booking facility; the data were used to create heat maps (ArcMap 10.4.1) to identify concentrated areas of high drug arrests in Baltimore. We extracted the frequencies of drug arrests according to weekday and time of day to establish sampling intervals. Combined, these data created a sampling frame consisting of geographic area, day of the week, and time period of the day representing Baltimore’s greatest drug activity. Eligibility criteria for PROMOTE required participants to be at least 18 years old and report non-medical use of any opioid in the previous month. After providing informed consent, eligible participants completed a 30-minute survey on tablets via Audio Computer-Assisted Self-Interview (ACASI) software. Participants were given a $25 VISA gift card in remuneration and all study activities and protocols were approved by the Johns Hopkins Bloomberg School of Public Health Institutional Review Board.

2.3. Outcome

We assessed recent non-fatal overdose with the following survey question: “In the past 6 months, how many times have you overdosed?”. Answers were dichotomized as “no” (zero overdoses) vs. “yes” (at least one overdose).

2.4. Covariates

We measured demographics and background characteristics including: age, race, education, gender (cis- and transgender participants were combined into man and woman categories), structural vulnerabilities (experiencing homelessness in the past 6 months, health insurance coverage, arrest in the previous year, weekly food insecurity), and recent (past 3-month) involvement in selling, making, or trading illegal drugs or selling sex. Frequency of psychological pain was assessed with the following question: “How often in the past 6 months have you had any long-lasting psychological or mental pain?” with responses categorized as “never”, “sometimes”, or “daily”. Recent (past 6-month) unmet mental health need was assessed using a binary (“yes”/ “no”) question adapted from the National Survey on Drug Use and Health:“During the past 6 months, was there a time when you needed mental health treatment or counseling but didn't get it?”

Finally, we gathered detailed data on substances used, route of administration, and frequency of use in the past 3 months. We then created binary indicators for daily use of cocaine, heroin, fentanyl, concomitant heroin and cocaine (speedball), and concomitant fentanyl and cocaine by route of administration (i.e., injection and non-injection separately). We also created binary indicators for daily use of fentanyl and non-medical prescription opioids (any route). We then created a variable to characterize the categories of opioids used daily regardless of route. This variable included heroin, fentanyl, and non-medical prescription opioids; participants reporting only daily speedball or fentanyl and cocaine use were categorized as using one opioid daily, for example.

2.5. Statistical analysis

We omitted two participants missing outcome data from our analyses, resulting in a final sample of n=563. We produced descriptive analyses and determined bivariate differences in key variables stratified by recent overdose via Pearson’s Chi-square tests for categorical variables and t-tests for continuous variables. As we were interested in the relationship between overdose, psychological distress, and unmet mental health need, we selected variables for inclusion in the multivariable model a priori based on their known association with these variables. For example, we selected daily multi-opioid use because it is associated with increased overdose risk and we selected race, gender, and recently experiencing homelessness because of established differential risk of mental distress/illness and overdose (Alang, 2015; Cook et al., 2014; Martins et al., 2015; Wang & Xie, 2019). We used general linear models with Poisson distribution and robust variances for univariate and multivariable models. Poisson regression models provide better estimates of relative risk than logistic regression models in the case of non-rare outcomes (i.e., the present outcome) (Dean, 1989). All analyses were conducted using StataSE 15.1 (College Station, TX).

To model the complex relationship between mental health, multiple opioid use, and overdose, we conducted a path analysis to determine direct pathways. We specified the path model with the hypothesized relationships between variables. A path model is a type of structural equation model that allows simultaneous estimation of multiple, specific pathways on dependent variables. Path models allow researchers to specify assumed directionality between variables. The simultaneous estimation capability distinguishes path models from multivariable logistic or linear models. In addition to the independent variables in path models (e.g., gender, psychological pain), race and homelessness were also included as covariates in all paths. We used maximum likelihood estimator with logit link. Mplus 8.0 was used for path analyses.

3. Results

3.1. Participant characteristics and bivariate differences

Sample prevalence of recent overdose was 30%. On average, participants were 48 years old, 60% were men, and 72% were non-Hispanic Black (Table 1). Participants who reported a recent overdose reported significantly higher levels of recently experiencing homelessness (81% vs. 61%), past year arrests (29% vs. 18%), weekly food insecurity (74% vs. 55%), and recent involvement in the drug trade via selling, making or trading illegal drugs (70% vs. 54%) and recent sex work (43% vs. 23%) compared to those who did not.

Table 1.

Demographic, structural vulnerability, mental health, and substance use variables stratified by overdose a sample of people who use drugs in Baltimore, Maryland (n=563)

Overdose, past 6 months
Total (n=563)
N (col %) 		No (n=395)
N (col %) 		Yes (n=168)
N (col %) 		p
Age (mean, SD) 47.7 (10.6) 48.0 (10.8) 47.0 (10.2) 0.32
Race
 Non-Hispanic White 95 (17.3) 59 (15.4) 36 (21.7) 0.05
 Non-Hispanic Black 397 (72.2) 289 (75.3) 108 (65.1)
 Hispanic/mixed race/other race 58 (10.5) 36 (9.4) 22 (13.3)
Education
 Less than high school graduate or GED 228 (40.6) 161 (40.9) 67 (39.9) 0.89
 High school graduate/GED or greater 334 (59.4) 233 (59.1) 101 (60.1)
Gender
 Man 335 (60.1) 243 (62.5) 92 (54.8) 0.09
 Woman 222 (39.9) 146 (37.5) 76 (45.2)
Experienced homelessness, past 6 months 378 (67.1) 242 (61.3) 136 (81.0) <0.001
Has health insurance coverage 450 (79.9) 314 (79.5) 136 (81.0) 0.63
Arrested, past year 118 (21.0) 70 (17.7) 48 (28.6) 0.004
Food insecure, weekly 342 (60.7) 218 (55.2) 124 (73.8) <0.001
Sold, made, traded illegal drugs, past 3 months 326 (58.5) 210 (53.6) 116 (70.3) <0.001
Sold sex, past 3 months 162 (28.8) 90 (22.8) 72 (42.9) <0.001
Experiencing psychological pain
 Never 208 (37.3) 167 (42.7) 41 (24.7) <0.001
 Sometimes 233 (41.8) 162 (41.4) 71 (42.8)
 Daily 116 (20.8) 62 (15.9) 54 (32.5)
Unmet mental health need, past 3 months 259 (46.2) 156 (39.6) 103 (61.7) <0.001
Inject drugs daily 238 (42.7) 149 (37.9) 89 (53.9) <0.001
 Daily cocaine injection 84 (14.9) 44 (11.1) 40 (23.8) <0.001
 Daily speedball injection 111 (19.7) 59 (14.9) 52 (31.0) <0.001
 Daily heroin injection 167 (29.7) 101 (25.6) 66 (39.3) 0.001
 Daily fentanyl injection 109 (19.4) 64 (16.2) 45 (27.1) 0.003
Daily fentanyl snorting or smoking 112 (19.9) 74 (18.7) 38 (22.6) 0.29
Daily heroin snorting or smoking 166 (29.5) 115 (29.1) 51 (30.4) 0.77
Daily cocaine snorting 98 (17.5) 66 (16.8) 32 (19.2) 0.49
Daily fentanyl and cocaine use 81 (14.4) 42 (10.6) 39 (23.4) <0.001
Daily fentanyl use (any route) 334 (59.9) 225 (57.4) 109 (65.7) 0.07
Daily nonmedical prescription opioid use (any route) 203 (36.2) 132 (33.5) 71 (42.5) 0.04
No. of opioid types used daily
None 107 (19.0) 89 (22.5) 18 (10.7) <0.001
One 110 (19.5) 84 (21.3) 26 (15.5)
Multiple 346 (61.5) 222 (56.2) 124 (73.8)
Witnessed overdose, ever 308 (54.7) 214 (54.2) 94 (56.0) 0.70
Use drugs for physical pain 233 (41.5) 161 (40.9) 72 (42.9) 0.66
Currently has naloxone 297 (52.8) 205 (51.9) 92 (55.1) 0.49
Use drugs with at least one other person 445 (79.0) 310 (78.5) 135 (80.4) 0.62
Ever used fentanyl test strips 100 (17.8) 66 (16.7) 34 (20.2) 0.32
Open in a new tab

Daily psychological pain was significantly greater for those who recently overdosed (33% vs. 16%). Unmet mental health need was also significantly greater for those who recently overdosed compared to those who had not (62% vs. 40%). Daily drug use differed significantly between participants who recently overdosed and those who did not including: daily use of multiple opioids (74% vs. 56%); daily drug injection (any drug, 54% vs. 38%); daily injection of cocaine (24% vs. 11%), speedball (31% vs. 15%), heroin (39% vs. 26%), and fentanyl (27% vs. 16%); daily fentanyl and cocaine use (23% vs. 11%); and daily non-medical prescription opioid use (43% vs. 34%).

3.2. Univariate and multivariable regression models

In unadjusted models, compared to no recent overdose, participants had significantly greater risk of overdose if they recently experienced homelessness (rate ratio[RR]=2.08, 95% confidence interval [CI]=1.48, 2.93); used multiple types of opioids daily compared to no daily opioid use (RR=2.13, 95% CI=1.37, 3.23); experienced psychological pain sometimes (RR=1.55, 95% CI=1.11, 2.16) or daily compared to never (RR= 2.36, 95% CI=1.68, 3.31); had unmet mental health need (RR=1.88, 95% CI=1.44, 2.44); and injected drugs daily (RR=1.58, 95% CI=1.22, 2.04). Non-Hispanic Black participants were less likely than non-Hispanic white participants to report recent overdose (RR=0.72, 95% CI=0.53, 0.97) (Table 2).

Table 2.

Multivariable Poisson regressions modeling odds of overdose in a sample of people who use drugs in Baltimore, Maryland (n=533)

Overdose, past 6 months
Unadjusted models Adjusted model
RR (95% CI) p aRR (95% CI) p
Race (ref = NH white)
 NH Black 0.72 (0.53, 0.97) 0.03 0.99 (0.74, 1.33) 0.95
 Hispanic/mixed race/other race 1.00 (0.66, 1.52) 0.99 1.09 (0.73, 1.63) 0.67
Gender (ref = man/male)
 Woman/female 1.25 (0.97, 1.60) 0.09 1.10 (0.86, 1.42) 0.46
Experienced homelessness (ref = no) 2.08 (1.48, 2.93) <0.001 1.87 (1.31, 2.68) 0.001
No. of opioid types used daily (ref = none)
 One 1.41 (0.82, 2.41) 0.22 1.34 (0.86, 2.07) 0.20
 Multiple 2.13 (1.37, 3.23) 0.001 1.55 (1.02, 2.36) 0.04
Experiencing psychological pain (ref = never)
 Sometimes 1.55 (1.11, 2.16) 0.01 1.19 (0.83, 1.71) 0.35
 Daily 2.36 (1.68, 3.31) <0.001 1.66 (1.13, 2.45) 0.01
Unmet mental health need (ref = no) 1.88 (1.44, 2.44) <0.001 1.38 (1.02, 1.87) 0.04
Inject drugs daily (ref = no) 1.58 (1.22, 2.04) 0.001 1.19 (0.90, 1.57) 0.23
Open in a new tab

In the adjusted model that controlled for multiple covariates, compared to no recent overdose, participants were more likely to have recently overdosed if they recently experienced homelessness (adjusted rate ratio [aRR]=1.87, 95% CI=1.31, 2.68); used multiple opioids daily compared to no daily opioid use (aRR=1.55, 95% CI=1.02, 2.36); experienced daily psychological pain (aRR=1.66, 95% CI=1.13, 2.45); and had unmet mental health need (aRR=1.38, 95% CI=1.02, 1.87) (Table 2).

3.3. Path analysis

After adjusting each path for covariates, we found evidence of several significant direct paths throughout the model. Woman gender was associated with greater adjusted odds of experiencing daily psychological pain (adjusted odds ratio [aOR]=1.68, 95% CI =1.18-2.40, p=0.05) compared to man gender. Woman gender (aOR=2.23, 95% CI=1.64-3.03, p=0.003) and experiencing daily psychological pain (aOR=4.14, 95% CI=2.75-6.23, p=0.002) were associated with greater odds of unmet mental health need compared to man gender and experiencing less frequent psychological pain, respectively. In turn, unmet mental health need was associated with greater odds of daily multi-opioid use (aOR=1.57, 95% CI=1.15-2.13, p=0.05), which in turn was associated with greater odds of overdose (aOR=1.78, 95% CI=1.28-2.47, p=0.03). Unmet mental health need was also directly associated with two-fold increase in odds of recent overdose (aOR=2.05, 95% CI=1.47-2.86, p=0.01).

4. Discussion

This study illustrates mechanisms through which psychological pain and unmet mental health need increase risks of overdose in a street-based sample of people in Baltimore who have used opioids non-medically. Nearly half the sample reported unmet mental health need, which has a highly significant direct relationship with overdose risk, and about two-thirds reported at least some psychological pain, which acted through unmet mental health need to amplify overdose risk. Woman gender was particularly salient to this dynamic as it was associated with both greater odds of psychological pain and unmet mental health need, both risk factors for recent non-fatal overdose. Our findings extend prior work—which has found associations between psychological trauma or unmet mental health need and severity of substance use—by considering non-fatal overdose, a critically important outcome as overdose mortality continues to increase in the U.S. (Narendorf, Cross, Santa Maria, Swank, & Bordnick, 2017). Taken together, these findings highlight an urgent need to scale up access to mental health services among people at risk of overdose.

Results show that greater psychological pain is associated with increased odds of unmet mental health need, creating a cycle of pain and unmet need. If the self-medication hypothesis holds, then devising acceptable (and feasible) modalities to interrupt this cycle (i.e., scaling up access to mental health care) will be important tools to prevent overdose. The goal may not be to eliminate all drug use but to create more opportunities for individuals to access more healthcare options and to engage in safer drug use. In this way, mental health service capacity for vulnerable populations is situated on a continuum of overdose risk and de-escalation of risk (Park et al., 2020). Harm reduction therapy is an emerging modality of mental health care that may be a promising avenue to expand access to care without a prerequisite that individuals are abstinent from drugs. Harm reduction therapy meets individuals where they are—literally and figuratively: therapists may engagein care with PWUD at drop-in centers, on the street, or other locations where PWUD spend time, and also present a range of options for care that are nonjudgemental and not driven by encouraging abstinence from drugs (Little & Franskoviak, 2010; Logan & Marlatt, 2010; Mancini & Linhorst, 2010).

For those who do seek substance use treatment, integration of mental health treatment is typically lacking or nonexistent (Grella & Hser, 1997; Mauro, Furr-Holden, Strain, Crum, & Mojtabai, 2016). In addition, patients with comorbid psychiatric conditions are less likely to complete drug treatment and therefore are at greater risk of overdose (Friesen & Kurdyak, 2020; Mason, Aplasca, Morales-Theodore, Zaharakis, & Linker, 2016; Ravndal & Amundsen, 2010). Another study found that, in a sample of patients with opioid use disorder and mental illness presenting for drug detoxification, mental illnesses were frequently treated with prescription medications such as gabapentin or benzodiazepines which heighten the risk of opioid overdose (Wilens, Zulauf, Ryland, Carrellas, & Catalina-Wellington, 2015). Healthcare professionals should consider both substance use and mental health conditions in designing holistic care; harm reduction training and a comprehensive understanding of how to help patients navigate multiple prescriptions, particularly during relapsing opioid use, is critical. Mental healthcare’s integration into substance use treatment models may be one option for PWUD to access mental healthcare, but there are present limitations such as those outlined above in this service delivery that should be addressed in a way that considers PWUD preferences and healthcare needs.

Self-medication can serve as a readily available coping mechanism when PWUD face social, organizational, and structural barriers to mental health care (Meyerson et al., 2021; Murphy, Yoder, Pathak, & Avery, 2021; Paquette, Syvertsen, & Pollini, 2018). Our sample of PWUD had all recently used opioids, so we attempted to explore severity of opioid use in our path model through the multi-opioid variable. We found some evidence to support the self-medication hypothesis, including significant paths between daily psychological pain, unmet mental health need and daily multi-opioid use; further, there was not a significant direct path between psychological pain and multi-opioid use. While we cannot determine if unmet mental health need mediates the pathway between psychological pain and multi-opioid use using the current cross-sectional data, these results show clear trends in association that should raise the potential for future investigation. Longitudinal data and a larger sample size could better determine if and how unmet mental health need mediates this pathway and supports the self-medication hypothesis. Building off the present results, future research may also consider the association between unmet mental health need and multiple non-fatal overdoses.

Gender was significantly associated with daily psychological pain and unmet mental health need, with women having higher odds than men of reporting both, yet gender was not directly associated with overdose. These results show a potential mediating role of psychological pain and unmet mental health need on women’s overdose risk, one that warrants further exploration of gender differences in mechanisms of overdose risk. It is well-established that women are more likely to be diagnosed with mental illness or experience mental distress compared to men, but there is variable evidence that mental healthcare-seeking differs significantly between men and women (Kessler et al., 2005; Riecher-Rossler, 2017). Our findings may depart from some of these findings of women’s more frequent mental healthcare-seeking likely because of the highly marginalized nature of our sample. For example, in work with women who use drugs and sell sex, women on the street experience frequent and acute stressors including physical and sexual violence and financial dependence on others (i.e., intimate partners) that may increase psychological distress and serve as a barrier to mental health care (Barreto, Shoveller, Braschel, Duff, & Shannon, 2019; Beattie, Smilenova, Krishnaratne, & Mazzuca, 2020). This result also demonstrates the benefit of a path analysis since the significance of gender was not evident when only considering the multivariable model.

There are several limitations to this analysis. First, data are cross-sectional so we cannot determine causal mediation, rather just establish relationships for further study. Second, our measures are self-reported and subject to recall and desirability biases. Questions about psychological pain and unmet mental health need are crude measures that will benefit from more detailed data collection in the future. For example, using validated measures of mental health conditions and well-being can add precision in understanding which particular symptoms or conditions may be associated with elevated overdose risk (Bohnert et al., 2012). Relatedly, we did not provide participants with a definition of overdose, instead leaving the definition up to interpretation by participants. Strict definitions of overdose have not been agreed upon and this may introduce recall bias into the analysis, though research has found that knowledge of opioid overdoses and their symptoms is high among people who use opioids (Chronister et al., 2018; Nielsen et al., 2018). A strength of this study, however, is in the community-based sample. Most prior studies of mental health or illness and overdose risk have used data from insurance or hospital claims to gather participants’ detailed medical histories, however, this omits medically-underserved or otherwise marginalized PWUD experiencing mental distress but who have not or cannot seek clinical care. We also clarify that people may use drugs for pleasure, though this departs from the strict definition of the self-medication hypothesis. We also caution against concluding that mental illness causes overdose. Our question of unmet mental health need specifically refers to participants who wanted mental health treatment; this group is motivated or self-reporting need but is experiencing barriers to getting it. Rather, these results demonstrate how structural barriers to care may increase overdose risk and the urgent need for increased facilitation of mental health services for marginalized populations. A final limitation is that our results may not be fully generalizable to other contexts, particularly among mostly White, rural contexts, where predictors of overdose have differed from urban locations. There is evidence to suggest that life stressors associated with opioid overdoses differ by urban and rural populations, where poverty and unemployment were associated with greater odds of overdose in urban areas compared to rural ones (Pear et al., 2019). Additionally, a relative paucity of available mental health care in rural compared to urban settings may mean that our results are underestimating the association between unmet mental health need and overdose risk (Laditka, Laditka, & Probst, 2009; Pullen & Oser, 2014).

When considering findings about non-fatal overdose, it is important to contextualize results the possibility that these data are subject to survival bias, i.e., individuals who were at highest risk of overdose may have already experienced a fatal overdose and cannot, therefore, be included in this data. Reviewing predictors of fatal overdose can contextualize our results further. For example, male sex and history of mental illness are two established predictors of fatal overdose (Brady, Giglio, Keyes, DiMaggio, & Li, 2017; Krawczyk et al., 2020; Saloner & Maclean, 2020). In the present results, woman gender may play a significant role in the path analysis because women are less likely than men to have died of an overdose already. Similarly, individuals with a history of mental illness being more likely to die of an overdose may have implications for the magnitude of association between unmet mental health need and non-fatal overdose. Recruiting a prospective cohort of PWUD and gathering robust mortality data in the event of individuals who fatally overdose can uncover the extent of this potential bias.

Amidst the COVID-19 pandemic, the world is experiencing the dual crises of unprecedented mental health strains—in part due to social isolation and economic anxieties—and reduced mental and physical healthcare capacities (Pfefferbaum & North, 2020). These crises are acutely felt by people who are actively using drugs and people who use medications to treat an opioid use disorder who have seen increased difficulties in timely receipt of their medication (Murphy et al., 2021). This study provides evidence of substantial psychological pain and unmet health need, even prior to the pandemic, and illustrates possible impacts on overdose. Present findings suggest that these mechanisms will be important to assess when measuring the long-term effects of the pandemic on the health of PWUD. Though this study was conducted before the pandemic, the present findings suggest that research into the long-term mental health effects of the pandemic on PWUD will be important in understanding and identifying mechanisms of overdose risk.

5. Conclusions

We presented findings enumerating direct pathways between mental pain, opioid use, unmet mental health need, and overdose in a sample of people who used opioid non-medically in Baltimore. This is, to our knowledge, one of the first studies to explicitly explore the relationship between unmet mental health need and overdose. Further, the large community-based sample of PWUD allows for an understanding of this relationship among individuals at the highest risk of overdose who frequently are underserved by health services. We found evidence that unmet mental health need appears on several direct pathways to overdose, suggesting a need to prioritize mental health service provision as a harm reduction tool for PWUD.

Figure 1. Path analysis of significant direct pathways to experiencing recent overdose in a sample of who use drugs in Baltimore, Maryland (n=533).

Figure 1.

Open in a new tab

Note: All paths also controlled for race and homelessness. Paths not statistically significant are displayed with dashed lines.

*p≤0.05, **p≤0.01

Table 3.

Indirect effects on probability of overdose in a sample of people who use drugs in Baltimore, Maryland (n=541)

Indirect path Estimate 95% CI P
Daily psychological pain → Unmet mental health need → Overdose 0.16 0.09, 0.22 <0.001
Gender → Unmet mental health need → Overdose 0.12 0.05, 0.19 0.007
Gender → Daily psychological pain → Unmet mental health need → Overdose 0.05 0.01, 0.08 0.04
Open in a new tab

References

  1. Alang SM (2015). Sociodemographic disparities associated with perceived causes of unmet need for mental health care. Psychiatric rehabilitation journal, 38(4), 293. [DOI] [PubMed] [Google Scholar]
  2. Ayano G, Belete A, Duko B, Tsegay L, & Dachew BA (2021). Systematic review and meta-analysis of the prevalence of depressive symptoms, dysthymia and major depressive disorders among homeless people. BMJ Open, 11(2), e040061. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Ayano G, Solomon M, Tsegay L, Yohannes K, & Abraha M (2020). A systematic review and meta-analysis of the prevalence of post-traumatic stress disorder among homeless people. Psychiatric quarterly, 91, 949–963. [DOI] [PubMed] [Google Scholar]
  4. Barreto D, Shoveller J, Braschel M, Duff P, & Shannon K (2019). The Effect of Violence and Intersecting Structural Inequities on High Rates of Food Insecurity among Marginalized Sex Workers in a Canadian Setting. J Urban Health, 96(4), 605–615. doi: 10.1007/s11524-018-0281-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Beattie TS, Smilenova B, Krishnaratne S, & Mazzuca A (2020). Mental health problems among female sex workers in low- and middle-income countries: A systematic review and meta-analysis. PLoS Med, 17(9), e1003297. doi: 10.1371/journal.pmed.1003297 [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Bohnert AS, Ilgen MA, Ignacio RV, McCarthy JF, Valenstein M, & Blow FC (2012). Risk of death from accidental overdose associated with psychiatric and substance use disorders. Am J Psychiatry, 169(1), 64–70. doi: 10.1176/appi.ajp.2011.10101476 [DOI] [PubMed] [Google Scholar]
  7. Brady JE, Giglio R, Keyes KM, DiMaggio C, & Li G (2017). Risk markers for fatal and non-fatal prescription drug overdose: a meta-analysis. Inj Epidemiol, 4(1), 24. doi: 10.1186/s40621-017-0118-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Caudarella A, Dong H, Milloy M, Kerr T, Wood E, & Hayashi K (2016). Non-fatal overdose as a risk factor for subsequent fatal overdose among people who inject drugs. Drug and alcohol dependence, 162, 51–55. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Centers for Disease Control and Prevention. (2021, March 17, 2021). Understanding the Epidemic. Retrieved from https://www.cdc.gov/drugoverdose/epidemic/index.html
  10. Chronister KJ, Lintzeris N, Jackson A, Ivan M, Dietze PM, Lenton S, … van Beek I (2018). Findings and lessons learnt from implementing Australia's first health service based take-home naloxone program. Drug and Alcohol Review, 37(4), 464–471. [DOI] [PubMed] [Google Scholar]
  11. Cook BL, Zuvekas SH, Carson N, Wayne GF, Vesper A, & McGuire TG (2014). Assessing racial/ethnic disparities in treatment across episodes of mental health care. Health Services Research, 49(1), 206–229. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Dean C L. JF (1989). Tests for detecting overdispersion in Poisson regression models. Journal of the American Statistical Association, 84(406), 467–472. [Google Scholar]
  13. Friesen EL, & Kurdyak P (2020). The impact of psychiatric comorbidity on treatment discontinuation among individuals receiving medications for opioid use disorder. Drug and alcohol dependence, 216, 108244. [DOI] [PubMed] [Google Scholar]
  14. Garland EL, Pettus-Davis C, & Howard MO (2013). Self-medication among traumatized youth: Structural equation modeling of pathways between trauma history, substance misuse, and psychological distress. Journal of behavioral medicine, 36(2), 175–185. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Grattan A, Sullivan MD, Saunders KW, Campbell CI, & Von Korff MR (2012). Depression and prescription opioid misuse among chronic opioid therapy recipients with no history of substance abuse. The Annals of Family Medicine, 10(4), 304–311. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Grella CE, & Hser Y-I (1997). A county survey of mental health services in drug treatment programs. Psychiatric Services (Washington, DC), 48(7), 950–952. [DOI] [PubMed] [Google Scholar]
  17. Guarino H, Mateu-Gelabert P, Quinn K, Sirikantraporn S, Ruggles KV, Syckes C, … Friedman SR (2021). Adverse Childhood Experiences Predict Early Initiation of Opioid Use Behaviors. Frontiers in Sociology, 6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Holland KM, Jones C, Vivolo-Kantor AM, Idaikkadar N, Zwald M, Hoots B, … Chen MS (2021). Trends in US emergency department visits for mental health, overdose, and violence outcomes before and during the COVID-19 pandemic. JAMA psychiatry. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Hooper D, Coughlan J, & Mullen M (2007). Structural equation modeling: Guidelines for determining model fit (Vol. 6). [Google Scholar]
  20. Jones CM, Guy GP Jr, & Board A (2021). Comparing actual and forecasted numbers of unique patients dispensed select medications for opioid use disorder, opioid overdose reversal, and mental health, during the COVID-19 pandemic, United States, January 2019 to May 2020. Drug and alcohol dependence, 219, 108486. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Kessler RC, Demler O, Frank RG, Olfson M, Pincus HA, Walters EE, … Zaslavsky AM (2005). Prevalence and treatment of mental disorders, 1990 to 2003. N Engl J Med, 352(24), 2515–2523. doi: 10.1056/NEJMsa043266 [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Khantzian EJ (1985). The self-medication hypothesis of addictive disorders: focus on heroin and cocaine dependence. Am J Psychiatry, 142(11), 1259–1264. doi: 10.1176/ajp.142.11.1259 [DOI] [PubMed] [Google Scholar]
  23. Khantzian EJ (2003). The self-medication hypothesis revisited: The dually diagnosed patient. Primary Psychiatry, 10(9), 47–54. [Google Scholar]
  24. Krawczyk N, Eisenberg M, Schneider KE, Richards TM, Lyons BC, Jackson K, … Saloner B (2020). Predictors of overdose death among high-risk emergency department patients with substance-related encounters: a data linkage cohort study. Annals of Emergency Medicine, 75(1), 1–12. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Laditka JN, Laditka SB, & Probst JC (2009). Health care access in rural areas: evidence that hospitalization for ambulatory care-sensitive conditions in the United States may increase with the level of rurality. Health & Place, 15(3), 761–770. [DOI] [PubMed] [Google Scholar]
  26. Little J, & Franskoviak P (2010). So glad you came! Harm reduction therapy in community settings. Journal of Clinical Psychology, 66(2), 175–188. [DOI] [PubMed] [Google Scholar]
  27. Logan DE, & Marlatt GA (2010). Harm reduction therapy: A practice-friendly review of research. Journal of Clinical Psychology, 66(2), 201–214. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Mancini MA, & Linhorst DM (2010). Harm reduction in community mental health settings. Journal of social work in disability & rehabilitation, 9(2-3), 130–147. [DOI] [PubMed] [Google Scholar]
  29. Martel MO, Edwards RR, & Jamison RN (2020). The relative contribution of pain and psychological factors to opioid misuse: A 6-month observational study. American Psychologist, 75(6), 772. [DOI] [PubMed] [Google Scholar]
  30. Martins SS, Sampson L, Cerda M, & Galea S (2015). Worldwide Prevalence and Trends in Unintentional Drug Overdose: A Systematic Review of the Literature. Am J Public Health, 105(11), 2373. doi: 10.2105/AJPH.2015.302843a [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Maryland Department of Health. (2018). Unintentional Drug- and Alcohol-Related Intoxication Deaths in Maryland: Annual Report 2018. Retrieved from https://bha.health.maryland.gov/Documents/Annual_2018_Drug_Intox_Report.pdf
  32. Mason MJ, Aplasca A, Morales-Theodore R, Zaharakis N, & Linker J (2016). Psychiatric comorbidity and complications. Child and Adolescent Psychiatric Clinics, 25(3), 521–532. [DOI] [PubMed] [Google Scholar]
  33. Mauro PM, Furr-Holden CD, Strain EC, Crum RM, & Mojtabai R (2016). Classifying substance use disorder treatment facilities with co-located mental health services: a latent class analysis approach. Drug and alcohol dependence, 163, 108–115. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. McGinty EE, Presskreischer R, Han H, & Barry CL (2020). Psychological distress and loneliness reported by US adults in 2018 and April 2020. Jama, 324(1), 93–94. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Meyerson BE, Russell DM, Kichler M, Atkin T, Fox G, & Coles HB (2021). I don't even want to go to the doctor when I get sick now: Healthcare experiences and discrimination reported by people who use drugs, Arizona 2019. International Journal of Drug Policy, 103112. [DOI] [PubMed] [Google Scholar]
  36. Mojtabai R (2009). Unmet need for treatment of major depression in the United States. Psychiatric Services, 60(3), 297–305. [DOI] [PubMed] [Google Scholar]
  37. Mojtabai R, Chen L-Y, Kaufmann CN, & Crum RM (2014). Comparing barriers to mental health treatment and substance use disorder treatment among individuals with comorbid major depression and substance use disorders. Journal of substance abuse treatment, 46(2), 268–273. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Mojtabai R, Olfson M, Sampson NA, Jin R, Druss B, Wang PS, … Kessler RC (2011). Barriers to mental health treatment: Results from the National Comorbidity Survey Replication (NCS-R). Psychological medicine, 41(8), 1751. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Murphy SM, Yoder J, Pathak J, & Avery J (2021). Healthcare utilization patterns among persons who use drugs during the COVID-19 pandemic. Journal of substance abuse treatment, 121, 108177. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Narendorf SC, Cross MB, Santa Maria D, Swank PR, & Bordnick PS (2017). Relations between mental health diagnoses, mental health treatment, and substance use in homeless youth. Drug and alcohol dependence, 175, 1–8. [DOI] [PubMed] [Google Scholar]
  41. Nielsen S, Peacock A, Lintzeris N, Bruno R, Larance B, & Degenhardt L (2018). Knowledge of opioid overdose and attitudes to supply of take-home naloxone among people with chronic noncancer pain prescribed opioids. Pain Medicine, 19(3), 533–540. [DOI] [PubMed] [Google Scholar]
  42. Novak P, Feder KA, Ali MM, & Chen J (2019). Behavioral health treatment utilization among individuals with co-occurring opioid use disorder and mental illness: evidence from a national survey. Journal of substance abuse treatment, 98, 47–52. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Padgett DK, Hawkins RL, Abrams C, & Davis A (2006). In their own words: Trauma and substance abuse in the lives of formerly homeless women with serious mental illness. American journal of orthopsychiatry, 76(4), 461–467. [DOI] [PubMed] [Google Scholar]
  44. Padgett DK, Henwood B, Abrams C, & Drake RE (2008). Social relationships among persons who have experienced serious mental illness, substance abuse, and homelessness: Implications for recovery. Am J Orthopsychiatry, 78(3), 333–339. doi: 10.1037/a0014155 [DOI] [PubMed] [Google Scholar]
  45. Paquette CE, Syvertsen JL, & Pollini RA (2018). Stigma at every turn: health services experiences among people who inject drugs. International Journal of Drug Policy, 57, 104–110. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Park JN, Rouhani S, Beletsky L, Vincent L, Saloner B, & Sherman SG (2020). Situating the continuum of overdose risk in the social determinants of health: a new conceptual framework. The Milbank Quarterly, 98(3), 700–746. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Pear VA, Ponicki WR, Gaidus A, Keyes KM, Martins SS, Fink DS, … Cerda M (2019). Urban-rural variation in the socioeconomic determinants of opioid overdose. Drug Alcohol Depend, 195, 66–73. doi: 10.1016/j.drugalcdep.2018.11.024 [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Pfefferbaum B, & North CS (2020). Mental health and the Covid-19 pandemic. New England Journal of Medicine, 383(6), 510–512. [DOI] [PubMed] [Google Scholar]
  49. Priester MA, Browne T, Iachini A, Clone S, DeHart D, & Seay KD (2016). Treatment access barriers and disparities among individuals with co-occurring mental health and substance use disorders: an integrative literature review. Journal of substance abuse treatment, 61, 47–59. [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Pullen E, & Oser C (2014). Barriers to substance abuse treatment in rural and urban communities: counselor perspectives. Substance use & misuse, 49(7), 891–901. [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Ravndal E, & Amundsen EJ (2010). Mortality among drug users after discharge from inpatient treatment: an 8-year prospective study. Drug and alcohol dependence, 108(1-2), 65–69. [DOI] [PubMed] [Google Scholar]
  52. Riecher-Rossler A (2017). Sex and gender differences in mental disorders. Lancet Psychiatry, 4(1), 8–9. doi: 10.1016/S2215-0366(16)30348-0 [DOI] [PubMed] [Google Scholar]
  53. Robinson J, Sareen J, Cox BJ, & Bolton JM (2011). Role of self-medication in the development of comorbid anxiety and substance use disorders: a longitudinal investigation. Archives of General Psychiatry, 68(8), 800–807. [DOI] [PubMed] [Google Scholar]
  54. Saloner B, & Maclean JC (2020). Specialty Substance Use Disorder Treatment Admissions Steadily Increased In The Four Years After Medicaid Expansion: Study looks at whether Medicaid expansion led to more low-income adults with substance use disorders receiving treatment. Health Affairs, 39(3), 453–461. [DOI] [PMC free article] [PubMed] [Google Scholar]
  55. Schneider KE, Tomko C, Nestadt DF, Silberzahn BE, White RH, & Sherman SG (2020). Conceptualizing overdose trauma: The relationships between experiencing and witnessing overdoses with PTSD symptoms among street-recruited female sex workers in Baltimore, Maryland. International Journal of Drug Policy, 102859. [DOI] [PMC free article] [PubMed] [Google Scholar]
  56. Schwartz AC, Bradley R, Penza KM, Sexton M, Jay D, Haggard PJ, … Ressler KJ (2006). Pain medication use among patients with posttraumatic stress disorder. Psychosomatics, 47(2), 136–142. [DOI] [PMC free article] [PubMed] [Google Scholar]
  57. Slavova S, Rock P, Bush HM, Quesinberry D, & Walsh SL (2020). Signal of increased opioid overdose during COVID-19 from emergency medical services data. Drug and alcohol dependence, 214, 108176. [DOI] [PMC free article] [PubMed] [Google Scholar]
  58. Smith LL, Yan F, Charles M, Mohiuddin K, Tyus D, Adekeye O, & Holden KB (2017). Exploring the link between substance use and mental health status: what can we learn from the self-medication theory? Journal of health care for the poor and underserved, 28(2), 113–131. [DOI] [PubMed] [Google Scholar]
  59. Substance Abuse and Mental Health Services Administration. (2019). National Survey on Drug Use and Health 2009-2019. Retrieved from: https://datafiles.samhsa.gov/
  60. Verduin ML, Carter RE, Brady KT, Myrick H, & Timmerman MA (2005). Health service use among persons with comorbid bipolar and substance use disorders. Psychiatric Services, 56(4), 475–480. [DOI] [PubMed] [Google Scholar]
  61. Wang N, & Xie X (2019). Associations of health insurance coverage, mental health problems, and drug use with mental health service use in US adults: an analysis of 2013 National Survey on Drug Use and Health. Aging & mental health, 23(4), 439–446. [DOI] [PubMed] [Google Scholar]
  62. Warner-Smith M, Darke S, & Day C (2002). Morbidity associated with non-fatal heroin overdose. Addiction, 97(8), 963–967. doi: 10.1046/j.1360-0443.2002.00132.x [DOI] [PubMed] [Google Scholar]
  63. Wilens T, Zulauf C, Ryland D, Carrellas N, & Catalina-Wellington I (2015). Prescription medication misuse among opioid dependent patients seeking inpatient detoxification. The American journal on addictions, 24(2), 173–177. [DOI] [PubMed] [Google Scholar]
  64. Wilson N, Kariisa M, Seth P, Smith H, & Davis NL (2020). Drug and opioid-involved overdose deaths—United States, 2017–2018. Morbidity and Mortality Weekly Report, 69(11), 290–297. [DOI] [PMC free article] [PubMed] [Google Scholar]
  65. Yang JC, Roman-Urrestarazu A, McKee M, & Brayne C (2019). Demographic, socioeconomic, and health correlates of unmet need for mental health treatment in the United States, 2002–16: evidence from the national surveys on drug use and health. International journal for equity in health, 18(1), 1–11. [DOI] [PMC free article] [PubMed] [Google Scholar]

RESOURCES