Prevalence of childhood maltreatment among people with opioid use disorder: A Systematic Review and Meta-analysis

Abstract

Background:

Experience of childhood maltreatment (CM) is a risk factor for opioid use disorder (OUD). CM is also associated with comorbid mental disorders and poor treatment outcomes among people with OUD. To our knowledge, this is the first systematic review and meta-analysis to estimate the prevalence of CM among people with OUD.

Methods:

We searched MEDLINE, EMBASE, and PsycINFO to identify observational studies that evaluated CM among people with OUD from January 1990 to June 2020. Prevalence of each CM type, sample characteristics, and methodological factors were extracted from each eligible study. Random-effects meta-analyses were used to pool prevalence estimates. Stratified meta-analyses were used to assess heterogeneity.

Results:

Of the 6,438 publications identified, 113 studies reported quantitative CM data among people with OUD and 62 studies (k=62; N=21,871) were included in primary analyses. Among people with OUD, the estimated prevalence of sexual abuse was 41% (95%CI 36–47%; k=38) among women and 16% (95%CI 12–20%; k=25) among men. Among all people with OUD, prevalence estimates were 38% (95%CI 33–44%; k=48) for physical abuse, 43% (95%CI 38–49%; k=31) for emotional abuse, 38% (95%CI 30–46%; k=17) for physical neglect, and 42% (95%CI 32–51%; k=17) for emotional neglect. Sex, history of injecting drug use, recruitment methods, and method of assessing CM were associated with substantial heterogeneity.

Conclusions:

People with OUD frequently report CM, supporting the need for trauma-informed interventions among this population. Future research should consider the impact of CM on OUD presentations and when assessment is appropriate, use of validated instruments.

1. Introduction

Globally, an estimated 40.5 million people live with opioid use disorder (OUD) (GBD 2017 causes of death collaborators, 2018). OUD is a chronic relapsing disorder that increases risk of blood borne virus transmission (Camacho et al., 1996), non-fatal overdose (Brady et al., 2017), and mortality (Degenhardt et al., 2019; Larney et al., 2020). The mortality rate of people with OUD is ten times that of comparable general population samples (Larney et al., 2020) and deaths attributable to OUD have increased exponentially over the past four decades (Hall et al., 2020). Risk factors for OUD include socioeconomic disadvantage (Needle et al., 1990), peer illicit substance use (Tucker et al., 2020), mental disorders (Martins et al., 2009), and the experience of childhood maltreatment (CM; Conroy et al., 2009).

CM is defined as the experience of sexual abuse, physical abuse, emotional abuse, or neglect of an individual under 18 years old (World Health Organization, 2006). Definitions for each type of CM are provided in Panel 1. Associations between CM and psychopathology related to substance use disorder are well-documented (Cicchetti and Handley, 2019). CM increases risk of behavioural disorders during childhood (Jones et al., 2004; Kendall-Tackett, 2004), youth mental health issues (Cicchetti and Valentino, 2006), and earlier onset of opioid and injection drug use (Kerr et al., 2009; Larance et al., 2018; Taplin et al., 2014). CM increases both risk of OUD (Conroy et al., 2009) and OUD severity. Among people with OUD, people who report experience of CM have higher rates of comorbid mental and substance use disorders (Darke, 2013; Maloney et al., 2007), engagement in risky injecting (McCurdy et al., 2010; Stein et al., 2017) and sexual behaviour (Engstrom et al., 2016), non-fatal overdose (Lake et al., 2015; Stein et al., 2017), and suicidal behaviour (Darke and Ross, 2002).

Panel 1: Definitions of each type of childhood maltreatment.

Sexual abuse

An attempted or completed sexual act or exploitation of a child (World Health Organization, 2006). Perpetrators of sexual abuse may be adults or other children in positions of power over the victim (World Health Organization, 2006). Examples include penetrative sexual abuse, sexual contact, and noncontact sexual abuse (e.g. child pornography) (Leeb et al., 2008).

Physical abuse

An intentional physical act by a caregiver that causes, or potentially causes, harm to a child’s health, survival, development, or dignity (World Health Organization, 2006). Examples include hitting, beating, kicking, and burning. Often inflicted with the object of punishment (World Health Organization, 2006) and recognised by physical injury (e.g. bruises) (Leeb et al., 2008).

Emotional abuse

An act or pattern of non-physical acts by a caregiver that have the potential to damage a child’s mental, physical, or social health and development (World Health Organization, 2006). Acts of emotional abuse may convey rejection and produce a hostile environment (World Health Organization, 2006). Examples include belittling, blaming, threating, ridiculing, terrorizing, and isolating (Leeb et al., 2008; World Health Organization, 2006). Also described as ‘psychological abuse’ (Leeb et al., 2008).

Physical neglect

Failure of a caregiver to provide for the child’s physical development and well-being, when they are in a position to do so (Leeb et al., 2008; World Health Organization, 2006). Examples include failure to provide adequate nutrition, appropriate clothing, hygiene, or shelter (Barnett et al., 1993; World Health Organization, 2006). The most commonly reported form of neglect (Stoltenborgh et al., 2015).

Emotional neglect

A caregiver’s failure to seek care for mental, emotional, or behavioural health problems and/or the omission of adequate affection (Barnett et al., 1993; Leeb et al., 2008). Examples include ignoring a younger child’s cry or an older child’s attempt to interact (Barnett et al., 1993; Leeb et al., 2008).

Panel notes:

Definitions derived from the World Health Organization (WHO) and United States Centers for Disease Control and Prevention (CDC)

Although studies of people with OUD report higher rates of CM than control groups (Conroy et al., 2009) and general population samples (Afifi et al., 2012; Evans et al., 2020), estimates vary widely (Conroy et al., 2009; Kumar et al., 2016).

To our knowledge, there has not yet been a systematic review and meta-analysis of the prevalence of any type of CM among people with OUD. Systematic reviews and meta-analyses of prevalence of CM exist for community samples (Stoltenborgh et al., 2015) and certain clinical populations, for example, people with psychosis (Bonoldi et al., 2013). These reviews have found that definitions of each type of CM are subject to cultural-geographical interpretation (Elliott and Urquiza, 2006; Stoltenborgh et al., 2013b, 2015), evolve over time (Miller-Perrin and Perrin, 2013), and vary by the methodology and instrument used to assess the experience (Baldwin et al., 2019).

We aimed to:

1. Estimate the prevalence of each major type of CM among people with OUD;

2. Examine sample characteristics and methodological factors associated with heterogeneity for each prevalence estimate;

3. Compare estimates of CM among people with OUD to global prevalence estimates of CM from a series of global reviews by Stoltenborgh et al (Stoltenborgh et al., 2015).

2. Methods

This review followed the PRISMA reporting guidelines (eAppendix 1; Moher, 2009) and was registered with PROSPERO (registration number: CRD42018094623).

2.1. Search strategy and study selection

Search terms were tailored for MEDLINE, Embase and PsycINFO databases using the OVID interface. Searches identified citations that reported any type of CM data among people with OUD published from January 1^st, 1990 to June 18^th, 2020. We aimed to estimate global prevalence of CM among people with OUD, therefore we did not restrict by language and used Google Translate to translate non-English studies. More details on the search strategy are provided in the supplementary material (eAppendix 2).

Study screening was conducted in Covidence, a web-based systematic review management tool (Veritas Health Innovation, 2016). After removing duplicates, we conducted title and abstract screening to identify observational studies of participants with OUD or proxy measures. Proxy measures for OUD included receiving treatment for opioid use (e.g. Opioid Agonist Therapy (OAT), detoxification, etc.), daily illicit opioid use, or frequent opioid use among people who inject drugs (PWID). Evaluation of CM was not considered at title and abstract screening stage to avoid selection bias. Relevant reviews were noted at title and abstract screening and citations were screened for eligibility. At the full-text screening stage, studies were excluded if the authors did not report any type of quantitative CM data among people with OUD (or proxy characteristics). Studies with less than 40 participants with OUD were also excluded. Full-text screening was conducted by two team members.

Authors were contacted if CM data and assessment of OUD (or proxy) were reported separately among the sample. Authors of citations that reported only continuous or aggregate types of CM data and those that did not report sexual abuse by sex were also contacted. eAppendix 3 includes details regarding the study selection process.

2.2. Data extraction

Data were extracted into a Microsoft Access database and double-checked by a second team member. Conflicts were resolved by discussion and referred to a third reviewer if necessary. eAppendix 4 includes details on the variables extracted and double-checking process.

2.2.1. CM Prevalence

The number and proportion of participants reporting each type of CM were extracted. Details on the instrument utilised and definition of maltreatment were collected. Details on the definition of CM included the instrument or scale utilised, cut-off used for continuous measurements, description of the event(s) or experience, and age cut-off for “childhood”.

2.2.2. Study characteristics

Sample characteristics, publication year, and recruitment method details were collected. Sample characteristics included age, proportion of men, proportion of people with a history of injecting drug use (IDU), description of opioid use patterns (i.e. details for ‘proxy’ OUD), childhood socio-economic status (SES) of the sample, history of parental SUD, and study location. Recruitment method details included setting and details on the participant inclusion and exclusion criteria. Recruitment setting of each study was stratified by four categories: sub-samples of general population studies, samples of people prescribed opioids for chronic non-cancer pain (CNCP), people with OUD recruited from OUD treatment settings (i.e. OAT, rehabilitation programs, supervised detoxification, and other in-patient services), and people recruited from non-treatment settings (i.e. community samples of PWID out of OAT, harm reduction services, and prisons).

2.2.3. Risk of bias and study quality

Study risk of bias was assessed by three domains: study inclusion or exclusion criteria, definition of OUD, and definition of CM, according to definitions described in Panel 1 (Leeb et al., 2008; World Health Organization, 2006). Risk of bias measures were adapted from published systematic reviews of people with OUD (Larney et al., 2020) and those with CM (Stoltenborgh et al., 2015). Risk of inclusion or exclusion criteria bias was a qualitative measure based on associations between CM and pre-defined risk of bias variables (e.g. studies that excluded participants with comorbid substance use or mental disorders were “high risk of underestimate”). These categorisations are not indicative of study quality, as the aims of included studies were not necessarily aligned with the aims of the current study. Each of the three domains was assessed by two independent reviewers and conflicts were resolved by a third party if necessary. Further details are provided in the supplementary material.

Statistical analyses were conducted in STATA 16.1 (StataCorp, 2017) using the metaprop_one command (Nyaga et al., 2014). Stratified random-effects meta-analyses were conducted by subgroup according to sex, history of IDU, recruitment setting, region, language of publication and language of survey administration. Random-effects meta-analyses were used as studies of retrospective CM data often report substantial heterogeneity (Baldwin et al., 2019). Stratified meta-analyses and meta-regressions were repeated by sex given that sex is a confounding variable in the experience of CM (Stoltenborgh et al., 2011). Sensitivity analyses were conducted by excluding studies at high risk of any form of bias, studies of adolescents, studies requiring translation, and studies before 2010.

Meta-regression analyses were conducted using the metareg function (Harbord and Higgins, 2009) to determine associations with continuous study-level variables, such as average sample age, percentage of men in the samples, and publication year.

3. Results

3.1. Included studies

Of the 113 studies (k=113) eligible for inclusion, 62 were included in one of the six primary analyses. Of the studies omitted from the primary analyses, 15 reported only continuous CM data, 8 reported only aggregate types of CM (i.e. ‘childhood adversity’), and 28 were secondary publications (i.e. participant overlap with primary studies). The PRISMA Flow diagram (Figure 1) provides further detail on the study inclusion and exclusion decisions. eAppendix 5 presents information on eligible studies excluded from the primary analyses.

Figure 1: PRISMA flowchart for included studies in the current review.

Notes: k=number of studies; OUD=Opioid Use Disorder; CM=Childhood Maltreatment; CSA= Childhood Sexual Abuse, CPA= Childhood Physical Abuse, CEA= Childhood Emotional Abuse, CPN=Childhood Physical Neglect, CEN=Childhood Emotional Neglect

Of the k=62 studies included in the primary analyses, there were samples from 18 different countries. Most studies recruited participants from North America (k=36; 58%) and evaluated CM among people in treatment for illicit OUD (e.g. heroin; k=45; 73%). Only one study evaluated history of CM among people with OUD who were prescribed opioids for CNCP. Over half of the included studies were published in 2010 or later (k=32; 52%) and 38 studies reported results among both men and women (61%). All estimates were self-report from observational studies, and most were published in English (94%). Table 1 presents detailed study-level descriptions.

Table 1:

Description of studies included in primary analyses

Study Author, Year	Location	Total (N)	Men (n)	Women (n)	Setting	Description	OUD Definition	Inclusion / Exclusion Criteria	IDU Status	CSA-Women	CSA-Men	CPA	CEA	CPN	CEN
Afifi et al., 2012	USA (National)	700	350	268	General Population	Subset from representative population of USA adults. The National Epidemiologic Survey on Alcohol and Related Conditions (NESARC)-II.	DSM-4 OUD	Excl: Institutionalised	-	40%	14%	38%	20%	43%	18%
Alexander, 2018 (Alexander et al., 2019)	Philadelphia, Pennsylvania, USA	175	-	175	OAT	Women receiving methadone with a child or pregnant enrolled in a mindfulness intervention trial.	OAT	Incl: Women with children < 3 years or pregnant		-	-	-	30%	-	-
Bailey et al., 1994	Norco, California, USA	354	354	-	Prison	Men in the Civil Addict Program, a rehabilitation ‘alternative’ to prison for ‘narcotic addicts’.	Other treatment	Excl: LTFU 24 years after recruitment	-	-	5%	-	-	-	-
Bartholomew et al., 2005	Texas, USA (One city)	137	-	137	OAT	Women in a no-fee methadone program at a private clinic.	OAT	-	-	39%	-	-	-	-	-
Blatchley et al., 2000	Baltimore & Washington-DC, USA	248	-	248	OAT	Women in OAT with adolescent children. Children enrolled in a psycho-educational program.	OAT	Incl: Mothers with custody of children	-	32%	-	-	-	-	-
SAMHSA, 2009 (From Bohnert et al., 2011)	USA (National)	1275	892	393	OAT, Rehabilitation	Adults surveyed as part of the National Treatment Improvement Evaluation Study (NTIES), a survey of publicly funded treatment centres in the USA	Other treatment	Incl: Admitted to treatment for heroin or ‘street’ methadone use	75%-Ever	22%	5%	36%	-	-	-
Browne et al., 1998	Dublin, Ireland	52	28	24	OAT	Consecutive patients admitted for methadone detoxification (n=33) or OAT (n=19). Patients referred by GPs, addiction councillors, or selfreferral.	OAT	-	-	-	-	23%	-	-	-
Campbell et al., 2016	Australia (National)	381	203	178	Prescribed for CNCP	Subset of people with pharmaceutical opioid dependence prescribed opioids for CNCP. Recruited from community pharmacies in Australia.	ICD-10 Dependence	Excl: Only prescribed opioids for OAT or cancer pain	38%-Ever	46%	22%	43%	55%	17%	-
Cash and Wilke, 2003	USA (National)	336	-	336	Rehabilitation, Detoxification	Subset of mothers from the DATOS study. Recruited from drug treatment programs, hospitals, and criminal justice programs.	Other treatment	Incl: Women with children	-	27%	-	18%	-	-	-
Cohen et al., 2009	USA (Multisite)	214	-	214	OAT	Sexually active women enrolled in one of seven OAT programs participating in a national multisite clinical trial for HIV/STD interventions.	OAT	Incl: OAT > 30 days, Past 6-month HIV risk; Excl: Pregnant	-	40%	-	-	-	-	-
Darke and Torok, 2013 (Darke and Torok, 2014; Torok et al., 2016)	Sydney, NSW, Australia	300	201	99	Harm Reduction	Subset of people who frequent needle and syringe programs and use heroin.	Frequent Illicit Use	Incl: > Weekly injecting	100%-Current	-	-	77%	-	-	-
Davis et al., 2019a	Illinois, USA	2067	1373	694	Rehabilitation, Detoxification	Subset of adolescents with OUD in Chestnut Health Systems, a substance use treatment provider in Illinois.	DSM-4 OUD	Incl: Adolescents only	-	30%	3%	61%	42%	-	-
Derefinko et al., 2019	Tennessee, USA	87	65	12	OAT	Medical chart review of a rural clinic providing OAT and counselling.	OAT	-	-	-	-	-	45%	-	40%
Dissabandara et al., 2009	Kandy, Sri Lanka	278	278	-	Prison	Male inmates in prison for drug related offenses. 98% past-month heroin use, mean use: 13 years.	Frequent Illicit Use	-	16%-Ever	-	7%	22%	-	-	-
El-Bassel et al., 2001 (El-Bassel et al., 2000)	NYC, USA	280	-	28-	OAT	Women recruited from OAT clinics in Harlem through outreach and staff referrals (1995–96). 32% sex workers	OAT	Incl: ≥ 2 HIV-risk behaviours	-	31%	-	38%	-	-	-
Engstrom et al., 2012 (El-Bassel N. et al., 2019; Engstrom et al., 2016; Panchanadeswaran et al., 2008; Rojas, 2006)	NYC, USA	390	-	390	OAT	Women recruited from OAT clinics in a sexual relationship. Subset of random sample from OAT sites in NYC.	OAT	Incl: Sexually active with men	-	58%	-	38%	-	-	-
Evans et al., 2020 (Blanco C. et al., 2020; Hassan and Le Foll, 2019)	USA (National)	778	390	388	General Population	Subset of participants with lifetime illicit OUD from a 2012–13 representative USA household survey (NESARC-III)	DSM-5 OUD	Excl: Institutionalised	-	36%	12%	38%	30%	54%	19%
Fudalej et al., 2015	Warsaw, Poland	240	169	71	OAT	Sample of people with opioid dependence from an OAT Clinic.	ICD-10 Dependence	-	-	-	-	24%	-	-	-
Gardner S.M. et al., 2020	Columbus, Ohio, USA	40	16	24	Rehabilitation, Detoxification	Adolescents, young adults receiving outpatient treatment for OUD.	Any OUD	-	-	-	-	35%	57%	35%	43%
Garfield et al., 2017	Melbourne, Australia	121	87	34	OAT, Rehabilitation, Detoxification	People receiving OAT (n=90) and recently abstinent people (n=31) from residential services.	DSM-4 Dependence	Excl: Psychosis, bipolar, major depression	-	-	28%	45%	50%	36%	49%
Gilbert et al., 1997 (Gilbert et al., 2000)	NYC, USA	151	-	151	OAT	Women enrolled in OAT programs in Bronx/Harlem, recruited through staff referrals and printed announcements in 1994.	OAT	-	-	27%	-	39%	-	-	-
Golden, 2018	Alaska, USA	43	-	43	OAT	Sample of 43 Alaskan women seeking opioid-treatment without PTSD from a dissertation.	OAT	Excl: Comorbid PTSD	-	40%	-	30%	25%	-	-
Grella et al., 1995	Los Angeles, California, USA	209	-	209	Snowballing, Harm Reduction	Subset of women recruited to an OAT trial by outreach and snowballing. Targeted sex workers.	OAT	Excl: LTFU (13%)	100%-Ever	38%	-	40%	57%	-	-
Heffernan et al., 2000	NYC, USA	136	84	52	Hospital	Subset of opiate users (> daily use) from a sample of consecutive admissions to a psychiatric hospital.	Frequent Illicit Use	-	-	27%	-	46%	-	-	-
Hien et al., 2000	Long Island, NY, USA	96	47	49	OAT	People at a rehabilitation centre seeking methadone for OUD.	OAT	-	-	29%	4%	23%	-	-	-
Isralowitz et al., 2002; Isralowitz and Bar Hamburger, 2002	Negev, Israel	152	98	54	OAT, Detoxification	Israeli-born patients in an outpatient drug treatment program.	OAT	-	-	26%	5%	32%	-	-	-
Isralowitz, 2001	Negev, Israel	154	93	61	OAT, Detoxification	Russia/Ukraine-born patients in an outpatient drug treatment program.	OAT, other treatment	-	-	43%	12%	40%	-	-	-
Kaboski, 2013	Chicago, Illinois, USA	205	-	205	OAT	Mothers of infants and pre-schoolers at one of seven public OAT clinics.	OAT	-	-	35%	-	33%	-	-	-
Kang et al., 2002	NYC, USA	432	294	138	Snowballing, Harm Reduction	Participants recruited from a list of OAT treatment dropouts and by local outreach workers from 1997–1999.	OAT	Incl: Only people who left OAT	-	48%	31%	60%	58%	66%	-
Khosravani et al., 2019 (Ghorbani et al., 2019)	Tehran, Iran	350	350	-	Detoxification	Men with heroin dependence at a treatment centre in Tehran (N=450). Survey 2 weeks post detoxification.	DSM-4 Heroin Dependence	Excl: Health comorbidity, refusal, women	50%-Ever	-	-	-	74%	-	59%
Koyuncu et al., 2003	Istanbul, Turkey	100	89	11	Detoxification	People with heroin dependence at an inpatient hospital detoxification centre.	DSM-4 Heroin Dependence	Excl: Comorbid psychiatric diagnosis	-	-	-	59%	62%	-	-
Kumar et al., 2016	Little Rock, Arkansas, USA	113	63	50	OAT	Chart review of adults receiving buprenorphine at an outpatient clinic.	OAT	-	27%-Ever	16%	18%	19%	25%	23%	18%
Lake et al., 2015 (Braitstein et al., 2003; Cheng et al., 2018)	Vancouver, Canada	2393	1489	903	Snowballing, Harm Reduction	People with a history of OAT or ≥ daily illicit opioid use from the VIDUS/ARYS/ACCESS cohorts.	OAT, Frequent Illicit Use	-	100%-Ever	46%	21%	36%	51%	32%	53%
Lovell, 2002	Marseille, France	111	111	-	Snowballing, Harm Reduction	People with past year IDU recruited from 3 areas in Marseille. > 82% on OAT and injecting duration 10 years.	Frequent Illicit Use	-	100%-Ever	-	-	18%	15%	-	-
McCurdy et al., 2010	Dar es Salaam, Tanzania	169	-	169	Snowballing, Harm Reduction	Women injected recently and had sex in the month prior. Of 559 screened, 249 remained at 5-month FU.	Frequent Illicit Use	Incl: Current IDU Excl: LTFU at 5 months (44%)	100%-Past 2 days	25%	-	-	-	-	-
Medrano et al., 1999	San Antonio, Texas, USA	80	-	80	Snowballing, Harm Reduction	Subset of participants that primarily use opioids from women recruited to an AIDS prevention program.	Frequent Illicit Use	Excl: Past month drug treatment	100%-Ever	60%	-	57%	47%	64%	79%
Mirhashem et al., 2017	Northern New England, USA	84	45	39	Snowballing, Harm Reduction	Participants with lifetime SUD, identified opioids as primary substance. 29% had received OAT.	DSM-4 OUD	Excl: Active psychosis	-	-	27%	51%	49%	58%	49%
Moselhy et al., 2010 (Elhammady et al., 2014; Moselhy, 2009)	Birmingham, United Kingdom	120	90	30	Rehabilitation, Detoxification	Outpatient attendees at a drug treatment service without mental or SUD comorbidities.	Other treatment	Excl: Other illicit substance use, Alcohol SUD, psychosis, trauma	-	-	-	27%	-	-	-
Naqavi et al., 2011	Bardsir, Iran	212	179	33	Rehabilitation, Detoxification	People who visited one of four drug treatment centres	DSM-4 Dependence	-	-		-	-	-	34%	52%
Nelson et al., 2006	Australia (National)	73	29	44	General Population	Young adults from Australian Twin Register. Participants with OUD born 1964–71. Substance use resembled general population.	DSM-4 OUD	Excl: LTFU	-	57%	-	-	-	-	-
Nyamathi et al., 2010	Los Angeles, California, USA	256	152	104	OAT	People in OAT with moderate-heavy alcohol use at five sites promoting HAV/HBV vaccinations.	OAT	Incl: Moderateheavy alcohol use	-	-	-	25%	-	-	-
Orellana et al., 2014 (Orellana, 2010)	NYC, USA	356	356	-	OAT	Males in heterosexual relationships in OAT. Parent study was a sample of males enrolled in at seven sites.	OAT	Incl: OAT > 3 months; Past year sexual partner	-	-	34%	-	-	-	-
Palis et al., 2016	Vancouver, Canada	172	96	96	Snowballing, Harm Reduction	Participants used opioids chronically from GeMa trial (≥5 years use, chronic use 6 months prior to survey, ≥1 OAT attempt). Recruited through snowballing and community.	OAT	Incl: ≥5 years use, ≥1 OAT attempt	100%-Ever	67%	29%	48%	70%	40%	43%
Peles et al., 2016 (Peles et al., 2012)	Tel Aviv, Israel	125	49	76	OAT	People at OAT clinic and the CSA prevalence determined. Of initial group, 79 patients after LTFU. Next wave, invited all women in OAT to participate: 46 agreed.	DSM-4 OUD	Incl: Daily heroin use for ≥ 1 year Excl: LTFU (19%)	-	89%	24%	-	-	-	-
Pinto et al., 2011	USA (Multisite)	137	-	137	Rehabilitation, Detoxification	Women with > sub-threshold PTSD from six community-based treatment sites in different US regions from Women and Trauma Study (WTS).	DSM-4 Dependence	Incl: Past 6-month substance use; DSM-4 PTSD	-	72%	-	59%
Plotzker et al., 2007	Philadelphia, Pennsylvania, USA	113	-	113	Snowballing, Harm Reduction	Women recruited from two needle and syringe programs.	Frequent Illicit Use	-	100%- Past 3 month	56%	-	68%	-	-	-
Rodriguez et al., 2017	Tulsa, Oklahoma, USA	48	-	-	OAT	Receiving buprenorphine at an addiction centre medicine practice. Patients approached during regular clinic visit with their physician.	OAT	-	-	-	-	-	52%	-	48%
Rovis D. et al., 2019	Rijeka, Croatia, Koper, Slovenia	167	129	38	Rehabilitation, Detoxification	Participants recruited from two in-treatment Rehab. Programs in Croatia and Slovenia. OAT (86%)	DSM-5 OUD	-	-	-	6%	13%	12%	8%	7%
Sansone et al., 2009	Dayton, Ohio, USA	113	61	52	Rehabilitation, Detoxification	Participants presented for admission to sub-acute detoxification 24-hr care unit that uses buprenorphine.	Other Treatment	Excl: Withdrawal or psychosis	-	-	-	40%	60%	23%	-
Santos Goni et al., 2010	Castilla & León, Spain	50	-	50	Rehabilitation, Detoxification	Subset of women with opiate dependence from a sample of women at 27 residential centres.	ICD-10 Dependence	Excl: Severe depression, psychosis	-	33%*	-	47%	59%	-	-
Sartor et al., 2014	USA (Multisite)	3513	2178	1335	General Population	Sub-set of clinical and general community samples of people with OUD. Recruited by family pairs from five east coast university medical schools for case-control studies.	DSM-4 Dependence	Excl: Major psychotic illness	-	32%	11%	13%	-	-	-
Schiff et al., 2010 (Schiff et al., 2006, 2002)	Israel (Multisite)	144	-	144	OAT	Women from methadone clinics in Israel were approached for their consent to participate.	OAT	Excl: Participant, social worker refused (25%)	-	72%	-	-	-	-	-
Shand et al., 2011 (Conroy et al., 2009; Larance et al., 2018; Larney et al, 2016; Maloney et al., 2010, 2009, 2007; Shand et al., 2010)	Sydney, NSW, Australia	1513	914	599	OAT	Participants from CATS Study were receiving OAT for heroin dependence at 34 of 35 OAT clinics in Sydney.	OAT	-	-	71%	36%	62%	53%	37%	61%
Shannon, 2007	Lexington, Kentucky, USA	68	-	68	Rehabilitation, Detoxification	Pregnant women entered unit to receive methadone-supervised detoxification or OAT.	OAT, other treatment	Incl: Pregnant	-	39%	-	30%	50%	-	-
Somer et al., 2010 (Somer, 2003)	Northern Israel	149	-	111	OAT, Rehabilitation, Detoxification	Receiving treatment in main heroin recovery program of northern Israel. Detoxification (n=49), outpatient treatment (n=48), OAT (n=52).	OAT	Excl: < 3 weeks in program -	-	-	30%	28%	-	42%
Stein et al., 2017	Fall River, Massachusetts, USA	457	326	131	Rehabilitation, Detoxification	Sample of patients seeking and admitted to inpatient opioid detoxification at 24-hour medically supervised detoxification facility.	Other treatment	-	-	42%	11%	36%	48%	18%	42%
Teegen and Zumbeck, 2000	Northern Germany	122	96	26	Rehabilitation, Detoxification	Participants were recruited from 10 addiction therapy facilities, all of whom use opioids. 53% in OAT.	Other treatment	-	-	-	4%	16%	-	-	-
Vogel et al., 2011	Basel, Switzerland	193	128	65	OAT	People from OAT, diacetylmorphine outpatient treatment centres.	OAT	-	-	40%	20%	31%	34%	40%	41%
Walker et al., 2014	Dallas, Texas	72	25	47	Rehabilitation, Detoxification	Adolescents with “cheese” heroin SUD from residential drug treatment or juvenile detention programs.	Other treatment	Excl: Psychosis, withdrawal, suicidal/homicidal ideation	13%-Ever	17%	-	17%	11%	-	-
Wang et al., 2010	Shanghai, China	341	123	218	Rehabilitation, Detoxification	Participants were currently injecting and recruited through convenience sampling from three government operated drug rehabilitation facilities.	DSM-4 Heroin Dependence	Excl: Cardiovascular, neurological issues	100%-Past Month	29%	20%	56%	36%	-	-
Weiss et al., 2019	Tel Aviv, Israel	51	33	18	OAT	Patients at an OAT clinic. All clients approached (81% participation).	OAT	Incl: OAT > 3 months	63% -Ever	-	-	53%	-	-	-
Wickersham et al., 2016	Kuala Lumpur, Malaysia	60	-	60	Snowballing, Harm Reduction	Sub-set of women in OAT recruited from community sites and shelters.	OAT	Incl: Past year other illicit substance use	-	27%	-	47%	-	-	-

Table Notes: Secondary studies cited in parentheses; Abbreviations: OAT=Opioid Agonist Therapy, IDU=Injecting Drug Use, OUD=Opioid Use Disorder; CM=Childhood Maltreatment, CSA= Childhood Sexual Abuse, CPA= Childhood Physical Abuse, CEA= Childhood Emotional Abuse, CPN=Childhood Physical Neglect, CEN=Childhood Emotional Neglect, LTFU: Lost to follow-up, PTSD: posttraumatic stress disorder, Incl= Key inclusion criteria for study, Excl= Key exclusion criteria for study

3.2. Prevalence of CM

Studies of people with OUD most commonly report frequency of childhood sexual abuse (k=53; n=20,522). Among samples of people with OUD, 41% of women reported a history of childhood sexual abuse (95%CI: 36–47%; k=38; n=8,478), compared to 16% of men (95%CI: 12–20%; k=25; n=9,940). A history of childhood physical abuse was reported by 38% of all people with OUD (95%CI: 33–44%; k=48; n=18,324) and 43% reported a history of emotional abuse (95%CI: 38–49%; k=31; n=11,8030). Prevalence of physical neglect was reported in 17 studies (n=7,504; 38%, 95%CI: 30–46%) and prevalence of emotional neglect was reported in 17 studies (n=6,964; 42%, 95%CI: 32–51%) among people with OUD. Every pooled estimate of CM reported substantial heterogeneity (Table 2). eAppendix 7 presents forest plots by study for each analysis.

Table 2:

Pooled prevalence estimates for each type of childhood maltreatment in people with opioid use disorder stratified by sample characteristics

	Sexual Abuse (Women)			Sexual Abuse (Men)			Physical Abuse			Emotional Abuse			Physical Neglect			Emotional Neglect
Strata	Study (k) (Total N)	Estimate (95%CI) p-value	I2	Study (k) (Total N)	Estimate (95%CI) p-value	I2	Study (k) (Total N)	Estimate (95%CI) p-value	I2	Study (k) (Total N)	Estimate (95%CI) p-value	I2	Study (k) (Total N)	Estimate (95%CI) p-value	I2	Study (k) (Total N)	Estimate (95%CI) p-value	I2
Community Sample Estimates (95%CI)1	k=193	18 (16–20)	-	k=104	8 (6–9)	-	k=157	23 (20–26)	-	k=42	36 (28–45)		k=13	16 (12–22)	-	k=18	18 (13–25)	-
Sex					-
Total	-	-	-	-	-	-	48 (18324)	38 (33–44)	99	33 (11030)	43 (38–49)	97	17 (7504)	38 (30–45)	96	17 (6964)	42 (32–51)	99
		p<0.01*						p=0.40			p=0.04*			p=0.38			p=0.29
Men	-	-	-	25 (9940)	16 (12–20)	97	23 (9438)	36 (27–46)	99	15 (5847)	39 (31–48)	98	12 (4121)	34 (25–44)	98	11 (3960)	38 (24–51)	99
Women	38 (8478)	41 (36–47)	96	-	-	-	35 (7418)	41 (35–47)	97	19 (3735)	50 (44–57)	94	13 (2793)	40 (32–49)	95	11 (2458)	45 (32–57)	98
Lifetime injecting history of sample		p=0.01*			p=0.01*			p<0.01*			p=0.68			p<0.01*			p<0.01*
<100%	4 (658)	25 (12–39)	92	4 (1429)	12 (6–18)	92	7 (2426)	30 (23–38)	93	3 (915)	41 (14–68)	99	2 (513)	18 (15–21)	-	2 (463)	44 (40–49)	-
100%	8 (2222)	49 (36–61)	97	4 (2443)	27 (17–36)	95	9 (4884)	51 (40–63)	98	7 (4423)	47 (39–56)	96	4 (3792)	40 (32–47)	95	5 (3775)	58 (50–67)	95
Recruitment Setting		p=0.56			p=0.56			p=0.12			p<0.01*			p<0.01*			p<0.01*
General population subsample	3 (699)	41 (33–49)	-	2 (740)	13 (11–15)	-	2 (1397)	38 (36–41)	-	2 (1128)	26 (23–29)	-	2 (1397)	49 (46–52)	-	2 (1397)	19 (17–21)	-
CNCP patients with OUD	1 (178)	46 (39–54)	-	1 (203)	22 (16–28)	-	1 (381)	43 (38–48)	-	1 (380)	55 (50–60)	-	1 (400)	17 (13–21)	-	-	-	-
OUD treatment settings2	26 (5971)	39 (32–47)	97	17 (7120)	14 (10–18)	98	34 (12780)	35 (28–42)	99	21 (6308)	42 (36–49)	96	9 (2783)	30 (20–39)	96	11 (3092)	41 (28–54)	98
Non-Treatment Settings3	8 (1630)	46 (37–54)	91	5 (1877)	22 (12–32)	95	11 (3766)	47 (36–58)	98	7 (3214)	50 (40–60)	95	5 (2924)	52 (34–70)	98	4 (2475)	56 (44–68)	92

Table Notes:

^*=Significant (<0.05) results subgroup comparison from stratified meta-analysis;

¹Estimates from series of meta-analyses by Stoltenborgh et al.¹;

Abbreviations:

²Treatment settings include Opioid Agonist Treatment (OAT), rehabilitation, and other inpatient services;

³Non-Treatment Settings include harm reduction services, community samples of people who inject drugs, and prisons; k=number of studies, CNCP=Chronic Non-Cancer Pain

3.3. Sample characteristic associations

In stratified sub-analyses (Table 2), women reported significantly higher rates of childhood sexual and emotional abuse than men with OUD. Samples of people with a history of IDU and OUD had substantially higher rates of all types of CM, excluding emotional abuse, than samples with lower rates of history of IDU and OUD. Prevalence rates of sexual abuse among men, emotional abuse, physical neglect, and emotional neglect varied substantially in stratified analyses of recruitment setting. Meta-regressions of average sample age and CM prevalence did not produce significant results, however, in stratified meta-analyses of age, studies of adolescents reported significantly lower rates of sexual abuse among women and men and emotional abuse among all people with OUD. Further information is presented in eAppendix 7 of the supplementary material. There was insufficient information to examine childhood SES and parental SUD as potential sources of heterogeneity.

3.4. Quality and risk of bias of included studies

Table 3 displays results from stratified analyses of each prevalence estimate according to risk of bias. Study inclusion or exclusion criteria was associated with substantial variance for prevalence estimates of sexual abuse among men and emotional abuse among women. The method of assessment of OUD (i.e. evaluation assessed via structured interviews versus proxy measures) did not affect prevalence estimates of any type of CM. Studies that used similar definitions to the “gold standard” WHO or CDC definitions reported significantly higher prevalence rates of sexual abuse among men, physical abuse, and physical neglect compared to less rigorous measures. eAppendix 6 provides further detail on risk of bias measures and classifications by study.

Table 3:

Pooled prevalence estimates for each type of childhood maltreatment in people with opioid use disorder stratified by risk of bias measures

	Sexual Abuse (Women)			Sexual Abuse (Men)			Physical Abuse			Emotional Abuse			Physical Neglect			Emotional Neglect
Strata	Study (k) (Total N)	Estimate (95%CI) p-value	I2	Study (k) (Total N)	Estimate (95%CI) p-value	I2	Study (k) (Total N)	Estimate (95%CI) p-value	I2	Study (k) (Total N)	Estimate (95%CI) p-value	I2	Study (k) (Total N)	Estimate (95%CI) p-value	I2	Study (k) (Total N)	Estimate (95%CI) p-value	I2
Inclusion/Exclusion criteria1		p=0.17			p<0.01*			p=0.09			p=0.01*			p=0.35			p=0.10
Low risk of inclusion/exclusion criteria bias	26 (5637)	39 (32–46)	97	19 (6399)	15 (11–19)	96	31 (11756)	35 (28–41)	98	20 (5554)	40 (32–48)	97	12 (4543)	34 (25–43)	98	13 (4452)	37 (25–49)	99
High risk of underestimate	4 (921)	40 (16–64)	97	3 (1805)	9 (3–15)	-	8 (2597)	40 (26–55)	96	6 (2437)	43 (29–56)	95	1 (121)	36 (27–45)	-	1 (121)	49 (40–58)	-
High risk of overestimate	8 (1920)	50 (41–59)	94	3 (1736)	26 (19–34)	-	9 (3971)	49 (38–60)	98	5 (3039)	57 (50–64)	88	4 (2840)	50 (30–71)	99	3 (2391)	58 (43–73)	-
OUD evaluation		p=0.09			p=0.66			p=0.42			p=0.99			p=1.00			p=0.37
Structured evaluation (DSM, ICD)	10 (3634)	50 (37–63)	98	10 (5354)	17 (10–24)	98	15 (9485)	42 (29–55)	99	12 (6219)	40 (32–49)	98	8 (3667)	38 (26–50)	98	7 (3405)	36 (18–55)	99
Proxy measure (i.e. OAT, ~daily use, etc.)	28 (4844)	38 (33–43)	94	15 (4586)	15 (10–20)	96	33 (8839)	36 (32–41)	95	19 (4811)	43 (36–50)	95	9 (3837)	38 (27–49)	98	10 (3559)	45 (37–53)	94
CM Definition		p=0.18			p<0.01*			p=0.04*			p=0.60			p<0.01*			-
Similar to WHO or CDC Definitions2	16 (3774)	45 (36–59)	97	16 (6038)	21 (14–27)	98	25 (11298)	44 (38–49)	98	22 (8237)	43 (37–50)	97	15 (6991)	43 (35–50)	97	17 (6694)	42 (32–51)	99
Specific (e.g. ‘severe’, <15 year, etc.)	13 (3782)	35 (29–41)	93	4 (3183)	7 (3–10)	93	8 (4498)	30 (19–41)	98	8 (2729)	43 (30–55)	96	1 (113)	23 (16–32)	-	-	-	-
Non-specific (e.g. were you abused?)	9 (922)	43 (30–56)	95	5 (719)	10 (4–15)	85	15 (2044)	34 (27–40)	91	1 (64)	50 (37–63)	-	1 (400)	17 (13–21)	-	-	-	-

Table Notes:

¹Further information on each study provided in eAppendix 6; :

²See Panel 1 for additional details; Abbreviations: k=number of studies, CI= Confidence Intervals, OUD= Opioid Use Disorder, DSM=Diagnostic and Statistical Manual of Mental Disorders, OAT= Opioid Agonist Therapy, ICD=International Classification of Diseases, WHO=World Health Organization, CDC=Centers for Disease Control and Prevention

Publication year was associated with lower prevalence of physical and emotional neglect in meta-regressions. In stratified meta-analyses, studies published after 2010 reported higher rates of sexual abuse among men and significantly lower rates of emotional neglect among all people with OUD. Studies published in English reported higher rates of sexual abuse among men, physical neglect, and emotional neglect, however, only two non-English studies were included in these analyses. More detail is provided in eAppendix 7.

3.5. Sensitivity analyses

We performed sensitivity analyses by pooling CM estimates of studies with low risk of inclusion or exclusion criteria bias, studies excluding adolescents, studies with “gold-standard” definitions of CM, studies published in English, and studies published after 2010. Results from sensitivity analyses were similar to those presented in Table 1, however, I² values remained significant. We present the results of sensitivity analyses in eAppendix 8.

3.6. Comparison to community samples

Figure 2 presents prevalence estimates for each type of CM among people with OUD alongside global, self-report estimates of each CM type from a series of reviews by Stoltenborgh et al (Stoltenborgh et al., 2015, 2013a, 2013b, 2012, 2011). Similar to the current review, studies included in the pooled estimates from Stoltenborgh et al were predominantly from North America (Stoltenborgh et al., 2015). Although it is difficult to formally compare the pooled estimates of CM prevalence from the different reviews, all pooled self-report estimates of CM prevalence are higher among people with OUD compared to community samples. Excluding emotional abuse, the 95% CIs of the two random-effects meta-analyses do not overlap.

Figure 2:

Prevalence estimates of childhood maltreatment (CM) among people with opioid use disorder (OUD) versus community samples

4. Discussion

4.1. Major findings and implications

The experience of CM among people with OUD was common, ranging between 16–43% across different types of CM. With the exception of emotional abuse, people with OUD report significantly higher rates of all types of CM when compared to community samples (Stoltenborgh et al., 2015). Our findings provide further evidence for CM as a risk factor for OUD (Conroy et al., 2009; Darke, 2013; Rossow and Lauritzen, 2001). Children who are abused or neglected often experience adversity and stress, which can increase risk of OUD later in life (Cicchetti and Handley, 2019). Additionally, CM and OUD share common environmental risk factors, including low SES (Young et al., 2007; Zielinski and Bradshaw, 2006) and parental SUD (Chatterjee et al., 2018), contributing to the elevated prevalence rates of CM among people with OUD.

There is clearly an integral relationship between a history of CM and OUD in adolescence and adulthood. Findings from the current review provide further evidence that prevention efforts of OUD may consider the long-term impact of CM. Policies that aim to reduce the overall prevalence of CM and opioid use following experience of CM, particularly during adolescence (Davis et al., 2019a), could help to reduce the burden of OUD in the general population (Afifi et al., 2012). Findings from the current review also have implications for interventions for people with OUD. Evidence suggests that experience of CM may affect an individual’s cognitions, beliefs, and feelings about their opioid use (Cicchetti and Handley, 2019). Additionally, there is strong evidence to support the relationship between CM and externalising disorders (Darke, 2011), which can play an important role in the frequency of opioid use (Cicchetti and Handley, 2019). The associations between CM and harms related to OUD, including overdose (Lake et al., 2015), alongside the extensive prevalence of CM among people with OUD reported in the current review, provides further evidence for interventions that aim to integrate CM experiences into OUD treatment.

Our findings underscore recommendations in clinical guidelines (Center for Substance Abuse Treatment, 2000) and research (Davis et al., 2019a, 2019b) for greater screening for CM among people in treatment for OUD. The high prevalence of CM within this population emphasises the importance of providing trauma-informed care for people with OUD, including the education of treatment providers around the sensitivities and complexities regarding disclosure of potentially traumatic experiences (Center for Substance Abuse Treatment, 2000; Elwyn and Smith, 2013). Trauma-informed care is critical for people with OUD because negative experiences with disclosure of victimisation is a significant treatment barrier, especially among women with OUD (Davis et al., 2019b; Green, 2006). Appropriate screening for experience of CM may also help to identify signs and symptoms of comorbid mental disorders, including depression (Center for Substance Abuse Treatment, 2000), post-traumatic stress disorder (Darke, 2013), and other anxiety-related disorders (Lawson et al., 2013). Since people with OUD and CM are more likely to present to treatment with comorbid mental disorders (Maloney et al., 2007); the findings of this review may have implications on the availability of treatment for comorbid mental disorders and OUD.

OAT is the most effective treatment for OUD (Degenhardt et al., 2019; Mattick et al., 2009; Sordo et al., 2017), including individuals with comorbid mental disorders (Trafton et al., 2006). However, people receiving OAT for OUD with a history of CM or comorbid mental disorders have higher levels of non-fatal overdose (Caudarella et al., 2016; Kang et al., 2002) and OAT treatment dropout (Hassan et al., 2017; Schiff et al., 2006), respectively. Adjunct psychosocial services have demonstrated mixed effectiveness in increasing retention in OAT and reducing psychological distress for people with OUD (Amato et al., 2011). There is some evidence that adjunct mental health and behavioural services may reduce symptoms of mental disorders among people with OUD (Fingleton et al., 2015; Hassan et al., 2017), including those who have experienced CM (Alexander, 2018). Examples of such services include psychopharmacological management of mental disorders (Hassan et al., 2017), cognitive behavioural therapy (Langdon et al., 2019; McHugh et al., 2017), and contingency management (Ainscough et al., 2017; Carroll and Weiss, 2017). Clinicians providing mental health treatment to people with OUD (Adams J.M. and Giroir B.P., 2019) may explore how experiences of CM can be integrated into comprehensive treatment plans for OUD and mental disorders (Davis et al., 2019b). Associations between inclusion or exclusion of participants with comorbid mental or substance use disorders and CM provides further evidence for the relationship between CM, polydrug use and mental health comorbidities. However, further research on the effective management of OUD among people with complex medical histories that include trauma and comorbid mental disorders is necessary.

Subgroup analyses of CM prevalence among different groups of people with OUD demonstrated that some groups of people with OUD have been exposed to higher rates of CM than others. Women with OUD reported significantly higher rates of childhood sexual abuse compared to men with OUD. Multiple studies included in our review report associations between CM and experience of revictimization and intimate partner violence among women with OUD (Engstrom et al., 2012, 2008; Green, 2006). People with a history of IDU and OUD, a clinical sample that commonly reports mental health comorbidities (Darke, 2011), reported elevated rates of multiple forms of CM. Finally, low rates of sexual abuse reported by adolescents are likely because these samples have not yet experienced the entire childhood period (Stoltenborgh et al., 2011) and fear of disclosure (Paine and Hansen, 2002) rather than true prevalence rates. Findings from the current review support further research on trauma-informed interventions (Alexander, 2018; Davis et al., 2019b) tailored to the needs and experiences of subgroups of people with OUD.

4.2. Limitations

Limitations of studies that assess CM prevalence among people with OUD are similar to limitations described in the wider literature on CM prevalence. For example, studies specific to people with OUD assess physical and sexual abuse most frequently (Stoltenborgh et al., 2015), despite evidence that all types of CM may have similar adverse health and social effects (Darke, 2011; Vachon et al., 2015). Several studies used questions that provided little detail on the experience of CM, which is critical to defining each CM type (Baldwin et al., 2019). The current review identified higher rates of CM in studies that used validated scales or definitions similar to the “gold standard” definitions for three of six pooled estimates, which implies that questions that require participant interpretation may be underestimating the true prevalence of CM.

Additionally, as all studies reported retrospective CM data, issues with memory and recall bias need to be considered when interpreting CM data (Baldwin et al., 2019). There is some concern that retrospective measures may identify a different group of people than prospective measurements of CM, particularly for individuals with substance use disorders (Elwyn and Smith, 2013). Despite these issues, use of retrospective measures do not necessarily indicate poor validity (Baldwin et al., 2019).

One notable limitation of existing evidence concerns sparse evidence on people with OUD whose primary opioid of concern is a pharmaceutical opioid, primarily because studies reporting on CM in such populations did not assess OUD as required for this review (Bouvier et al., 2019; Merrick et al., 2020; Williams J.R. et al., 2020). Given the widespread concerns over OUD among people prescribed opioids, more evidence on this population is needed.

Finally, there was substantial heterogeneity for each CM estimate. Heterogeneity is a common limitation in meta-analyses of retrospective CM data (Baldwin et al., 2019; Stoltenborgh et al., 2015), and inherent of many epidemiological reviews. Despite the heterogeneity, patterns identified in community samples (Stoltenborgh et al., 2015) were also identified in the current study, demonstrating consistency with the broader literature related to CM. For example, emotional abuse was the most prevalent type of CM reported by people with OUD and in community samples (Stoltenborgh et al., 2015; Vachon et al., 2015). Due to limited data, we were unable to control for potential sources of heterogeneity including childhood SES and parental SUD. However, we accounted for inclusion or exclusion of participants with comorbid mental or substance use disorders, language and year of publication, measurement of CM type, and geographic location in sensitivity analyses. Sensitivity analyses did not produce significantly different results to the pooled estimates of all studies of people with OUD that assessed CM.

4.3. Conclusions

This is the first systematic review and meta-analysis of CM prevalence among people with OUD. People with OUD commonly experience CM, and the prevalence of CM history is much higher than community samples. As such, comorbid mental health and SUD treatment providers should ensure that trauma-informed care for people with OUD is accessible, particularly for women and people with a history of IDU. Research on interventions that mitigate the long-term health and social consequences of CM among people with OUD is critical, and when appropriate, researchers should utilise validated instruments to assess history of CM.

Highlights.

• Childhood maltreatment (CM) is common among people with opioid use disorder (OUD)

• Women with OUD experience childhood sexual abuse (41%) more often than men (16%)

• People with OUD often report childhood physical abuse (43%) and neglect (40%)

• Most studies evaluate childhood sexual or physical abuse among OUD treatment samples

• “Gold standard” definitions of CM produce the highest prevalence rates of CM