Association between depression and non-fatal overdoses among drug users: A systematic review and meta-analysis

Page 1

R

Au

FCGa

b

c

d

e

a

ARR2AA

KDDM

C

7

0h

Drug and Alcohol Dependence 134 (2014) 12– 21

Contents lists available at ScienceDirect

Drug and Alcohol Dependence

j ourna l ho me p age: www.elsev ier .com/ locate /drugalcdep

eview

ssociation between depression and non-fatal overdoses among drugsers: A systematic review and meta-analysis�

rancesco Bartoli a, Giuseppe Carràb,∗, Giulia Brambillaa, Daniele Carrettaa,ristina Crocamoc, Julia Neufeindd, Alex Baldacchinoe,erry Humphrisd, Massimo Clerici a

Department of Surgery and Interdisciplinary Medicine, University of Milano Bicocca, Milano 20126, ItalyMental Health Sciences Unit, Faculty of Brain Sciences, University College London, W1W 7EJ, UKDepartment of Mental Health, San Gerardo Hospital, Monza 20900, MB, ItalyMedical and Biological Sciences Building, University of St Andrews, North Haugh, St Andrews KY16 9TF, UKDivision of Neuroscience, Medical Research Institute, University of Dundee, Ninewells Hospital and Medical School, Dundee DD1 9SY, UK

r t i c l e i n f o

rticle history:eceived 22 July 2013eceived in revised form3 September 2013ccepted 13 October 2013vailable online 24 October 2013

eywords:epressionrug overdoseeta-analysis

a b s t r a c t

Background: Assessing factors associated with non-fatal overdose is important as these could be usefulto identify individuals with substance use disorders at high risk of adverse outcomes and consequences.Depression may play an important role in terms of overdose risk. We aimed to test if drug users sufferingfrom a depressive disorder might have significantly higher risk of non-fatal overdose as compared withdrug users without depression.Methods: We conducted a systematic review and meta-analysis. PubMed, Embase and Web of Knowledgewere searched. The pooled analyses were based on prevalence rates, risk difference (RD) and odds ratio(OR), reporting 95% confidence intervals (CIs). The combined estimates were obtained weighting eachstudy according to random effects model for meta-analysis.Results: Seven articles, involving 12,019 individuals, and run in the US, Canada, Sweden, Norway, andAustralia, were included. Pooled analyses comparing depressed with not depressed individuals high-

lighted a RD (95% CIs) for non-fatal overdose of 7.3% (4.8–9.7%) and an OR (95% CIs) of 1.45 (1.17–1.79).The subgroups analyses based on specific characteristics of included studies confirmed the associationbetween depression and overdose.Conclusions: Depressive disorders seem to be important factors associated to the risk of non-fatal over-dose. Longitudinal studies might appropriately clarify causal inference issues. Future research shouldaddress the role of depressive disorders as predictors of subsequent non-fatal overdoses.

© 2013 Elsevier Ireland Ltd. All rights reserved.

ontents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132. Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

2.1. Search strategy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132.2. Eligibility criteria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132.3. Data collection process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142.4. Quality assessment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142.5. Data analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

3. Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

3.1. Study selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3.2. Study characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

� Supplementary material can be found by accessing the online version of this paper. S∗ Corresponding author at: Mental Health Sciences Unit, Faculty of Brain Sciences, Unive

EJ, UK. Tel.: +44 20 7679 9428; fax: +44 20 7679 9426.E-mail addresses: [email protected], [email protected] (G. Carrà).

376-8716/$ – see front matter © 2013 Elsevier Ireland Ltd. All rights reserved.ttp://dx.doi.org/10.1016/j.drugalcdep.2013.10.007

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

ee Appendix A for more details.rsity College London, Charles Bell House, 67–73 Riding House Street, London W1W

Page 2

F. Bartoli et al. / Drug and Alcohol Dependence 134 (2014) 12– 21 13

3.3. Quality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153.4. Association between depression and previous non-fatal overdose . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

4. Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154.1. Summary of findings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154.2. Interpretation of findings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154.3. Strengths and limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174.4. Clinical perspectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194.5. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

Role of funding source . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20Contributors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20Conflict of interest . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20Appendix A. Supplementary data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

. . . . . .

1

sKo(2aai(0atdfegi(drHdat(othaicctrnoopTbEsodoe2

1. A target population composed by adults with at least one of the follow-ing characteristics. A) a problem drug use defined as intravenous drug useand/or long duration/regular use of illicit drugs, such as opiates, cocaine and/oramphetamines (Kraus et al., 2003); B) being in treatment in an addiction service

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. Introduction

Substance use disorders are related to serious consequences,uch as risk of overdose and early mortality (Cornelius et al., 2008;ertesz et al., 2012). Heroin, cocaine, and amphetamine use dis-rders show standardized mortality ratios of 9.1 (CI: 8.5–9.8), 6.4CI: 3.9–10.0), and 6.0 (CI: 4.2–8.3), respectively (Arendt et al.,011). The median estimated non-AIDS-related mortality ratemong persons who inject drugs has been estimated 1.08% pernnum (Degenhardt et al., 2006). Illicit drug overdose is a lead-ng cause of premature death and morbidity among opioid usersDarke and Hall, 2003). Overdose-related mortality accounts for.65 (0.55–0.75) deaths per 100 person-years, followed by traumand suicide related deaths, with values of 0.25 and 0.12, respec-ively (Degenhardt et al., 2011). People from Asia who injectrugs have the highest pooled overall crude mortality rate (5.25),ollowed by populations from North America (2.64) and West-rn Europe (2.31). An overall crude mortality is related to maleender, not being in treatment, and living in low- and middle-ncome countries (Mathers et al., 2013). A 2010 meta-analysisMerrall et al., 2010) showed an increased risk of drug-relatedeaths during the first 2 weeks after release from prison. The riskemained high up to at least the fourth week. Furthermore, asIV-serostatus may consistently affect outcomes of opiate depen-ents (Carrà et al., 2008a), HIV comorbidity seems to representlso an important factor associated with an increased risk of mor-ality for overdose (pooled risk ratio = 1.74; 95% CI: 1.45–2.09)Green et al., 2012). Despite being clearly more common than fatalverdose, non-fatal overdose has received considerably less atten-ion in the literature (Kerr et al., 2007), possibly because publicealth concerns have been focused on lethal consequences (Darkend Zador, 1996; Carrà et al., 2006). However, non-fatal overdoses an important cause of morbidity among illicit drug users andan result in a number of serious medical consequences, such asardiovascular and pulmonary complications, peripheral neuropa-hy, anoxia-induced cognitive impairment, rhabdomyolysis, andenal failure (Warner-Smith et al., 2001). Evidence shows also thaton-fatal overdoses consistently increase the risk of subsequentverdose mortality (Stoové et al., 2009). An improved assessmentf non-fatal overdose rates and risk factors may help clinicians forreventing the excess of overdose fatalities (Darke et al., 2003).he prevalence rates of lifetime history of non-fatal overdose rangeetween 13% and 69% among drug users (Bohnert et al., 2010).vidence suggests that severity of dependence, polysubstance use,erious withdrawal symptoms, history of suicide attempt, lengthf drug using career, number of network members who injectrugs, homelessness, all are important risk factors for non-fatalverdoses (Backmund et al., 2009; Bohnert et al., 2010; Brugal

t al., 2002; Coffin et al., 2007; Jenkins et al., 2011; Kerr et al.,007).

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

Comorbidity for a mental disorder, involving more than half ofdrug-abusing individuals (Regier et al., 1990), may play an impor-tant role on the likelihood of nonfatal overdose. Drug users withco-occurring mental disorders report poorer prognosis, in terms ofremission and relapse likelihood (Hasin et al., 2002), emergencydepartment use (Curran et al., 2003), family/social relationships(Carrà and Clerici, 2006; Carrà and Johnson, 2009; Carrà et al., 2012),medical comorbidity (Rosenberg et al., 2001). Little is known on thepotential influence of depression on non-fatal overdose (Conneret al., 2008). Depression among drug users in residential programsis related to lower adherence and higher drop-out rates (Ravndaland Vaglum, 1994). Depressed methadone patients may be moresensitive to negative opioid effects and withdrawal (Elkader et al.,2009), and such a comorbid disorder seems associated also to anincreased risk of suicide attempt and ideation among drug and opi-oid users (Aharonovich et al., 2002; Darke and Ross, 2002; Phillipset al., 2004).

To the best of our knowledge this is the first meta-analysis thatsystematically analyzes the relationship between depression andnon-fatal overdose. We tested the hypothesis that drug users suf-fering from a depressive disorder would have shown rates, andexperienced risk, of non-fatal overdose significantly higher thannon-depressed drug users.

2. Methods

The present systematic review and meta-analysis was conducted according tothe Metaanalyses Of Observational Studies in Epidemiology (MOOSE) guidelines(Stroup et al., 2000).

2.1. Search strategy

We searched the Electronic databases PubMed, Embase and Web of Knowledgefor papers published up to September 2012. Additionally, we explored the refer-ence list of a relatively recent relevant systematic review (Bohnert et al., 2010) onthe association between overdose and suicide attempts in samples of substanceusers. We searched for papers in English, Spanish, French, German, Portuguese orItalian. Search phrases combined thesaurus terms related to overdose, depression,and mental disorders. In addition, synonyms of thesaurus terms were used for freesearch in titles and abstracts. Full search strategies are detailed in SupplementaryMaterial1.

2.2. Eligibility criteria

We included any observational study that provided:

1 Supplementary material can be found by accessing the online version of thispaper. See Appendix A for more details.

Page 3

14 F. Bartoli et al. / Drug and Alcohol Dependence 134 (2014) 12– 21

ram o

2

3

4

nfWi

-

--

2

iisao

erdrnwi

Fig. 1. Flow diag

(e.g., a specialized residential centre, a maintenance treatment clinic, a therapeu-tic community, or an inpatient detoxification unit); C) on medication-assistedtreatment for substance use disorders (e.g., methadone, buprenorphine, naltrex-one, or different pharmacological agents for non-opioid use disorders).

. Data on prevalence rates of previous non-fatal overdoses in a sample of drugsusers suffering from a depressive disorder, including major depressive disor-der, dysthymic disorder, or depressive disorder NOS, according to DSM-IV-TR(American Psychiatric Association, 2000), or previous Diagnostic and StatisticalManuals, as DSM-5 (American Psychiatric Association, 2013) was released imme-diately after we searched and screened the literature, or to ICD-10 InternationalClassifications of Diseases (World Health Organisation, 1992).

. Additional rates of non-fatal overdoses in a comparison group of drug userswithout depression.

. Data from clinical (inpatients or outpatients) and non-clinical (e.g., untreated,criminal justice, and institutionalized) populations.

We aimed at providing a summary result (crude cross-sectional association ofon-fatal overdoses) in a very specific subgroup of subjects, i.e., drug users suffering

rom a depressive disorder, thus we necessarily had restrictive eligibility criteria.e did not primarily aim at investigating heterogeneity. Thus, we excluded studies

f:

They used longitudinal/prospective designs where only incidence, and not preva-lence, of non-fatal overdoses was estimated.

There was data overlap with other studies included in the meta-analysis. Raw data could not be retrieved, since results of the study were adjusted for oneor more potential confounders and the authors did not provide the exact numberof cases.

.3. Data collection process

We performed a preliminary screening based on titles and abstracts, in order tonclude potentially relevant articles. After the first screening, studies were retrievedn full text to test the final eligibility, in order to explore whether specific inclu-ion/exclusion criteria were met. The eligibility assessment was performed by twouthors independently (FB and GC). Discordances on the inclusion or the exclusionf articles were analyzed, and disagreements were resolved by consensus.

We developed a sheet for the extraction of the following information fromach included study: year of publication; country; study design; inclusion crite-ia; setting; sample size; mean age; percentages of men and women; definition and

iagnostic strategies of depression; definition of non-fatal overdose; prevalence andaw numbers of cases of non-fatal overdoses in index and control subsamples. Ifumerical data were not reported, they were derived from percentages. If raw dataere not reported, we contacted the corresponding author in order to obtain this

nformation.

f search results.

2.4. Quality assessment

Errors in measuring exposure or outcome in observational studies may causeinformation bias (Copeland et al., 1977; Jepsen et al., 2004). Therefore, we conducteda quality evaluation, assessing if the included studies were affected by the risk of bothdepression and overdose misclassification. Measurement error or misclassificationmay result from poor validity of instruments used. We explored if proper methodsto assess depression, e.g., standardized and structured interviews such as the struc-tured clinical interview for DSM-IV-TR for Axis I disorders (SCID-I; First et al., 2002)or the Mini-International Neuropsychiatric Interview (MINI; Sheehan et al., 1998),were used. Furthermore, we evaluated if adequate definitions of non-fatal overdosewere provided, including also information on overdose related symptoms, such asloss of consciousness, seizures or respiratory depression, or the requirement of ahospitalization (Coffin et al., 2007; McGregor et al., 1998; Milloy et al., 2010; Polliniet al., 2006). Finally, we tested the potential effect of poor quality on results of ourmeta-analysis performing sensitivity analyses including only high quality paperswith low risk of overdose/depression misclassification.

2.5. Data analysis

We analyzed data using Stata version 10.0 SE. The pooled analyses were basedon prevalence rates, risk difference (RD) and odds ratio (OR) with related 95% con-fidence intervals. Results were summarized using conventional forest plots. Thepooled estimates were obtained by weighting each study according to randomeffects model for meta-analysis. This method weights studies more evenly and isconsidered more suitable for meta-analyses with substantial heterogeneity (Rileyet al., 2011). We assessed the presence and the level of heterogeneity using Q statisticand I2 index, respectively. In order to explore potential sources of heterogeneity, weperformed subgroup analyses based on characteristics of included studies: generalcharacteristics (publishing journal; year of publication; geographical area); char-acteristics of recruited samples (size; nature of substance use disorders; source ofrecruitment); temporal range (current or lifetime) of overdose and depression. Weconducted further subanalyses based on validity and quality of data, as described inthe previous subparagraph. We checked if authors used standardized interviews toassess depression and if adequate definitions of non-fatal overdose were provided. Afunnel plot was created for the visual inspection of potential risk of publication bias.The Egger’s linear regression test was used to evaluate the statistical significance ofpublication bias.

3. Results

3.1. Study selection

PubMed, Embase and Web of Knowledge provided 521, 421and 400 records, respectively. We searched also within 66 refer-ences from a recent relevant review (Bohnert et al., 2010). The

Page 4

hol D

ppieei

3

A

FeeTeo2leeHreaeMv2jit

3

ed(IueidSo

p2tors

3o

inc(((d

F. Bartoli et al. / Drug and Alco

reliminary screening by reading titles and abstracts identified 66otentially eligible papers. There were seven studies that fitted our

nclusion criteria (Bohnert et al., 2011; Darke et al., 2009; Fischert al., 2004; Hakansson et al., 2008; Havens et al., 2011; Maloneyt al., 2009; Rossow and Lauritzen, 1999). Detailed reasons for inel-gibility are described in flow diagram (Fig. 1).

.2. Study characteristics

Detailed characteristics of included papers are shown in Table 1.ll eligible articles were in English language.

Three papers were from North America (Bohnert et al., 2011;ischer et al., 2004; Havens et al., 2011), two from Australia (Darket al., 2009; Maloney et al., 2009), one from Sweden (Hakanssont al., 2008) and one from Norway (Rossow and Lauritzen, 1999).wo papers were published in 2011 (Bohnert et al., 2011; Havenst al., 2011), two in 2009 (Darke et al., 2009; Maloney et al., 2009),ne in 2008 (Hakansson et al., 2008), one in 2004 (Fischer et al.,004), and one in 1999 (Rossow and Lauritzen, 1999). Target popu-

ations were based on opioid and any drug users, in four (Darket al., 2009; Fischer et al., 2004; Hakansson et al., 2008; Maloneyt al., 2009) and three articles respectively (Bohnert et al., 2011;avens et al., 2011; Rossow and Lauritzen, 1999). Sample sizes

anged from 400 (Havens et al., 2011) to 5892 individuals (Bohnertt al., 2011). Most of the participants were males, with percent-ges ranging from 59% (Havens et al., 2011) to 88% (Hakanssont al., 2008). Four studies (Bohnert et al., 2011; Darke et al., 2009;aloney et al., 2009; Rossow and Lauritzen, 1999) included indi-

iduals from clinical settings, whereas three studies (Fischer et al.,004; Hakansson et al., 2008; Havens et al., 2011) recruited sub-

ects from non-clinical settings (drug users who were not enrolledn treatment, criminal justice clients, and rural drug users, respec-ively).

.3. Quality

Four papers (Darke et al., 2009; Fischer et al., 2004; Havenst al., 2011; Maloney et al., 2009) provided valid methods to assessepression, using standardized diagnostic interviews, such as MINISheehan et al., 1998) or the Composite International Diagnosticnterview (CIDI; World Health Organization, 1993). Three paperssed lower quality methods to assess depression, as two (Bohnertt al., 2011; Rossow and Lauritzen, 1999) did not use standard-zed interviews neither nor specific scales to make a diagnosis ofepression, and one (Hakansson et al., 2008) used the Addictioneverity Index (ASI) (McLellan et al., 1980) for the lifetime historyf depressive symptoms.

An appropriate definition of overdose was provided in fourapers (Bohnert et al., 2011; Fischer et al., 2004; Hakansson et al.,008; Maloney et al., 2009). In these studies, overdose defini-ion was not only made by asking the patients if they ever hadverdosed, but also, by collecting further relevant information onelated symptoms (e.g., loss of consciousness or respiratory depres-ion) or about the need for hospitalization.

.4. Association between depression and previous non-fatalverdose

The included studies involved 12,019 individuals, 6536 suffer-ng from depression, 5483 without depression. A lifetime history ofon-fatal overdose was found in 3792 subjects. The random effectombined prevalence (95% CIs) of previous overdose(s) was 38.5%

24.4–53.5%) among people suffering from depression and 30.8%16.7–47.1%) among people without depression. The combined RD95% CIs) for non-fatal overdose comparing depressed and non-epressed individuals was 7.3% (4.8–9.7%; p < 0.001). The pooled

ependence 134 (2014) 12– 21 15

OR (95% CIs) assessing the association between depression and ahistory of non-fatal overdose was 1.45 (1.17–1.79; p < 0.001). Het-erogeneity was high (p < 0.001; I2: 79.2% [49.2–88.3%]). Forest plotsare shown in Figs. 2 and 3.

Subgroup analyses based on specific characteristics of thepooled studies (including publishing journal, geographical area,period of publication, kind of population selected, samples size,setting, overdose and depression definitions), confirmed the asso-ciation between depression and non-fatal overdoses as statisticallysignificant (Table 2).

A sensitivity analysis based just on studies providing high qual-ity assessment of depression reported a statistically significantassociation between depression and a history of non-fatal over-dose, OR (95% CIs) = 1.31 (1.02–1.69; p = 0.03). Similarly, analysesincluding only papers with robust definition of overdose showedan OR of 1.48 (1.06–2.08; p = 0.02).

Funnel plot and Egger’s linear regression test for publication biasare shown in Fig. 4.

4. Discussion

4.1. Summary of findings

This systematic review on the potential burden of depressionon the risk of non-fatal overdose identified seven studies with datasuitable for meta-analysis. The association between depression andprevious non-fatal overdose(s) was statistically significant. Pro-portion of individuals with previous non-fatal overdose(s) wasabout 7% higher among depressed drug users as compared to theirnon-depressed counterpart. Furthermore, subjects suffering fromdepression were 1.5 times more likely to report a history of oneor more overdoses than their counterpart without this-current orlifetime-disorder. A high level of heterogeneity among studies wasfound. We explored potential sources of heterogeneity performingsubgroups analyses based on specific characteristics of includedstudies. The results did not appear to be affected by the natureof considered populations (drug users or opioid users). Further-more, subgroups analyses based on variables, such as period ofpublication, geographical area, sample size, source of recruitment,time considered for assessing depression and overdose (currentor lifetime), quality of reported data, consistently confirmed theassociation.

4.2. Interpretation of findings

We found a significant association between depression and non-fatal overdose. This finding cannot be necessarily extended to casesof fatal overdose. However, previous evidence (Kinner et al., 2012;Warner-Smith et al., 2001) highlighted that causes of fatal and non-fatal overdoses are likely to be similar. Furthermore, due to thecross-sectional nature of data included in this meta-analysis, we areable to confirm the significance of the association between depres-sion and non-fatal overdose, but we cannot assume any causal linkof depression on the subsequent risk of overdose. The directionof the association remained uncertain, since most of the papersinvestigated lifetime overdoses and depressive disorders withoutexplanation of temporal relationship between the events. Althoughsome prospective studies (Britton et al., 2010; Darke et al., 2009)analyzing the role of depression as predictor of non-fatal overdosewere available in the literature, additional large size longitudinalstudies are required in order to explore casual inference.

Factors mediating the relationship between depression andoverdose remain unknown. Depression may increase the risk ofnon-fatal overdose, because people with depression may be moreprone to use higher doses and more substances, resulting in more

Page 5

16 F. Bartoli et al. / Drug and Alcohol Dependence 134 (2014) 12– 21

Table 1Studies exploring association between depression and previous non-fatal overdoses.

Study Location Population(years of datacollection)

N Mean age(±SD)*

Males(%)

Definition ofdepression

Definition ofnon-fataloverdose

Associationbetweendepression andnon-fatal overdoseOR (95% CIs)

Bohnert et al.(2011)

USA Individualsenteringaddictionstreatment frommulti-sitestudy(1993–1994)

5892 N/A (75%between 25and 44)

71 History ofdepressivesymptoms

Lifetimehistory ofoverdoserequiringhospitalization

2.00 (1.73–2.31)

Darke et al. (2009) Australia(GreaterSidney)

Entrants totreatment forheroindependencefrom 19agencies(2001–2002)

429 29.5 ± 8.0 65 DSM-IV MDDaccording toCIDI

Last 12 monthshistory ofoverdose

1.33 (0.81–2.20)

Fischer et al. (2004) Canada(Vancouver,Edmonton,Toronto,Montréal andQuébec)

Regular illicitopioid userswho were notenrolled intreatment(2002)

651 34.8 ± 9.4 67 MDD accordingto CIDI shortform

Last 6 monthsoverdosedefined as adrug-userelatedexperiencewhere thesubject lostconsciousnessand/or hadconvulsionsbecause of toostrong of adrug(s) ortaking toomuch drug(s)

1.73 (1.14–2.63)

Hakansson et al.(2008)

Sweden Subjectsreporting useof opioidsduring the 30days prior toincarcerationfrom multi-sitestudy of theSwedish Prisonand ProbationService(2001–2006)

1096 31.8 (N/A) 88 Lifetimehistory ofdepressivesymptomsaccording toASI

Lifetimeoverdoseaccording toASI, defined aslife-threateningphysicalcomplicationsfollowing drugintake

1.34 (1.06–1.71)

Havens et al.(2011)

USA(Appalachiancounty inKentucky)

Rural drugusers(2008–2010)

400 Median: 31IQR: 26–38

59 MDD accordingto MINI

Lifetimehistory ofoverdose

1.50 (0.92–2.44)

Maloney et al.(2009)

Australia (NewSouth Wales)

Subjects fromopioidmaintenancetreatmentclinics(2004–2008)

1500 36.4 (N/A) 60 DSM-IV MDDaccording to astructuredinterview

Lifetimehistory ofopioidoverdosedefined as anevent whereloss ofconsciousnessand respiratorydepressionoccurs thatdoes not resultin a fataloutcome

1.06 (0.86–1.31)

Rossow andLauritzen (1999)

Norway Drug addictsadmitted totreatment inthe larger drugtreatmentclinics inNorway(1992–1993)

2051 27 ± 5.9 (men)25 ± 6.1(women)

64 Depressionaccording totherapists’clinicalevaluation

Lifetimehistory of life-threateningoverdose

1.36 (1.14–1.62)

ASI, Addiction Severity Index; CESD, Center for Epidemiologic Studies Depression Scale; CIDI, Composite International Diagnostic Interview; DSM-IV, Diagnostic and StatisticalManual of Mental Disorders, Fourth Edition; IQR, interquartile range; MDD, major depressive disorder; MINI, Mini-International Neuropsychiatric Interview; OR, odds ratio(calculated from raw data or estimated from reported percentages).

* SD, standard deviation (if available).

Page 6

F. Bartoli et al. / Drug and Alcohol Dependence 134 (2014) 12– 21 17

presse

strtsspredbg

4

sftttGtaWtSdi

Fig. 2. Random effect pooled risk difference (de

evere dependence syndromes. Depression may be an importantrigger for high-risk injection practices, continued drug use, or drugelapse (Stein et al., 2003). Specifically, situations involving nega-ive mood states account frequently for the risk of relapse acrosseveral types of addictive substances (Stein et al., 2004). Depres-ive disorders seem to reduce the likelihood of remission and toredict relapse in substance use (Hasin et al., 2002). Therefore, theelationship between non-fatal overdose and depression might bexplained by the fact that depressed individuals are simply heavierrug users. Thus, depression may be not an independent factor,ut a mediator of a more severe substance use disorder linked to areater likelihood of overdose.

.3. Strengths and limitations

The well-known advantage of meta-analyses of observationaltudies is based on the fact that these allow the synthesis of resultsrom a large amount of studies, providing findings more robust thanhose from individual studies data. However, although observa-ional studies are an important source in epidemiological research,hey are prone to many methodological issues (Egger et al., 1998;rimes and Schulz, 2002), which could also affect our results. Since

hese limitations may influence confidence in results of meta-nalyses, we paid critical attention to the quality of included papers.e found that some studies had methodological issues on impor-

ant items, such as depression and overdose misclassification risk.ome papers provided inadequate assessments of depression andefinitions of overdose. In order to reduce the effect of these lim-

tations, we performed subanalyses based on best quality data on

d vs. non-depressed) for non-fatal overdose(s).

depression and history of overdose assessment. These subanalysesconfirmed the statistically significant association between depres-sion and previous non-fatal overdose(s). Indeed, drug misuse isoften difficult to be screened using questionnaires, because of inac-curate reporting (Fendrich et al., 1999). This may be even more truefor multiple drugs or drugs other than heroin non-fatal overdoses,as these show symptoms markedly different and more difficult tobe assessed than those related to opioids (Fairbairn et al., 2008;Kaye and Darke, 2004). Indeed, different types of drug use couldnot be controlled for, though under optimal conditions, it would beadequate to control for at least opiate vs. stimulant dependence.This further limitation needs to be acknowledged. These limita-tions may lead at least partially to an underestimation of the realprevalence rates of non-fatal overdoses in our meta-analysis. Fur-thermore, most of the included papers did not provide methodsappropriate for a clear differentiation between unintentional andintentional overdoses. This issue is particularly important sincedepression is a well-known risk factor for intentional overdose andsuicide attempts among illicit drug users (Neale, 2000; Brådviket al., 2007). Therefore, despite previous research suggested thatthe majority of overdoses among drug users are likely to be unin-tentional (Best et al., 2000), we cannot exclude that at least a partof the association between depression and overdose is related to aself-injurious intent.

Usual limitations of systematic reviews of observational studies

need to be acknowledged also in our study. We actually includedonly published studies with sufficient data, excluding for exam-ple conference abstracts because these often cannot give reliableinformation on patients’ characteristics, inclusion criteria, and

Page 7

18 F. Bartoli et al. / Drug and Alcohol Dependence 134 (2014) 12– 21

Fig. 3. Random effect pooled odds ratio (depressed vs. non-depressed) for non-fatal overdose(s).

s. Egge

awesi

Fig. 4. Funnel plot for publication bia

ssessment of outcome, relevant results are often different from

hat is shown in the final publication, and they have not gen-

rally undergone the rigorous peer review process required forcientific journals papers. This increases the likelihood that biasnfluences the results (Crowther et al., 2010). However, search of

r’s test: −1,42 [−6,85; 4,02], p = 0,53.

comprehensive databases, such as PubMed, Embase, and Web of

Science-as well as the hand searching of references list of a rel-evant recent review-has largely granted an extensive coverage ofpublished literature on the topic (Lemeshow et al., 2005). Nonethe-less, as only seven papers were included, and these only focused on

Page 8

F. Bartoli et al. / Drug and Alcohol Dependence 134 (2014) 12– 21 19

Table 2Subanalyses of included studies.

N studies N subjects Odds ratio [random, 95% CI] p I2 (%)

General characteristicsPublishing journalDrug and alcohol dependence 3 7388 1.63 [1.20, 2.21] 0.002 76Other journals 4 4631 1.29 [1.07, 1.57] 0.009 47Period of publication2008–2011 5 9317 1.42 [1.05, 1.93] 0.02 85Before 2 2702 1.42 [1.18, 1.72] <0.001 11Geographical areaNorth America 3 6943 1.93 [1.70, 2.20] <0.001 0Northern Europe 2 3147 1.35 [1.17, 1.56] <0.001 0Australia 2 1929 1.10 [0.90, 1.33] 0.35 0

Characteristics of samplesSize<1000 3 1480 1.54 [1.18, 2.01] 0.002 0≥1000 4 10,539 1.41 [1.06, 1.88] 0.02 89NatureAny drug users 3 8343 1.62 [1.19, 2.20] 0.002 82Opioid users 4 3676 1.28 [1.05, 1.57] 0.02 42Source of recruitmentClinical setting 4 9872 1.41 [1.03, 1.95] 0.03 89Non-clinical setting 3 2147 1.44 [1.19, 1.75] <0.001 0

OverdoseTemporal rangeRecent (within past year) 2 1080 1.56 [1.13, 2.15] 0.007 0Lifetime 5 10,939 1.42 [1.10, 1.84] 0.007 86Quality of definitionHigh 4 9139 1.48 [1.06, 2.08] 0.02 89Low 3 2880 1.37 [1.17, 1.60] <0.001 0

DepressionTemporal rangeCurrent 3 2880 1.37 [1.17, 1.60] <0.001 0Lifetime 4 9139 1.48 [1.06, 2.08] 0.02 89

staStibpwa

4

p2oatddauf2cpAod

Quality of assessmentHigh 4 2980

Low 3 9039

pecific narrow regions, the results here may not be generalizableo all geographical locations. We also need to consider the risk thatn amount of negative or uncertain results remained unpublished.ome of included observational studies could be vulnerable to mul-iple analyses aimed to search for more significant results, resultingn an analysis of published studies, prone to a selective publicationias (Loder et al., 2010). Although the visual inspection of the funnellot and the Egger’s test showed the lack of risk of publication bias,e cannot exclude that this result was due to chance, as included

rticles were less than 10 (Sterne et al., 2008).

.4. Clinical perspectives

Clinicians should regularly assess symptoms of depression ineople suffering from a substance use disorder (Carrà and Clerici,006). The accuracy of assessment and screening of depressionften represents a critical issue, since tools commonly used inddiction research showed poor specificity and modest predic-ive power of screening (Delgadillo et al., 2011). Furthermore, aifferentiation between independent mood disorders from thoseue to continued substance misuse is often difficult (McIntoshnd Ritson, 2001). Thus, a primary depression remains frequentlyndetected and undertreated. This might be an important issueor establishing prognosis and optimal treatment (Mowla et al.,008). At the same time, treatment of depression represents ahallenge, because of lack of data supporting the use of a specific

harmacological intervention for depression among drug users.

relatively recent meta-analysis on pharmacological treatmentf depression among individuals on an opioid agonist for heroinependence, found low evidence supporting the clinical use of

1.31 [1.02, 1.69] 0.03 431.55 [1.17, 2.06] 0.002 86

antidepressants (Pani et al., 2010). Further data on the treatment ofmajor depressive and dysthymic disorders, showed no difference inthe reduction of depressive symptoms comparing antidepressantand placebo among individuals with comorbid opiate-use disor-ders (Pedrelli et al., 2011). Larger randomized trials are thereforeneeded, hopefully assessing also the role of more complex inter-ventions other the pharmacological ones, such as psychosocial orbehavioral interventions. Heterogeneous psychosocial treatmentsoffered in addition to pharmacological detoxification may be effec-tive in terms of completion of treatment, reduction of opiate use,and compliance, in terms of clinical absences during the treatment(Amato et al., 2008).

4.5. Conclusions

Despite some limitations, this systematic review and meta-analysis show that depression may be an important factorassociated with the risk of non-fatal overdose. Estimate of thepooled association and related subgroup analyses confirmed ourhypothesis. Accurate assessment and screening of depression, dis-tinguishing independent disorders from those due to substancemisuse, is needed for preventing the excess of overdose (Darkeet al., 2003). However, also substance-induced depression has clin-ical significance and should not be dismissed particularly as a riskperiod for non-fatal overdose (Aharonovich et al., 2002). Specialattention may be needed to prevent non-fatal overdose in such

patients, including careful review of the history of suicidal behav-ior as well as previous treatment of depression. DSM-IV subtypesof depression based on the timing of the occurrence of depres-sion in relation to substance dependence may also be important in

Page 9

2 hol D

eattosro

R

C

FFtFdlaFaamra

C

p

A

i2

R

A

A

A

A

A

B

B

B

B

B

0 F. Bartoli et al. / Drug and Alco

valuating non-fatal overdose risk among drug users suffering from depressive disorder. Effective pharmacological and psychosocialreatments for depression in drug users may require careful adapta-ion and implementation because of clinical complexity but also forrganizational barriers (Carrà et al., 2008b). Future meta-analyseshould be based on longitudinal studies addressed to clarify theole of depression as potential predictor of subsequent non-fatalverdoses.

ole of funding source

Nothing declared.

ontributors

Giuseppe Carrà and Massimo Clerici designed the study;rancesco Bartoli and Alex Baldacchino wrote the protocol;rancesco Bartoli, Daniele Carretta and Giulia Brambilla performedhe literature systematic search and first screening of articles;rancesco Bartoli and Giuseppe Carrà checked the papers for theefinitive eligibility in the meta-analysis; Daniele Carretta and Giu-

ia Brambilla performed the data extraction from included papersnd developed the tables with all relative relevant information;rancesco Bartoli and Cristina Crocamo performed the statisticalnalysis; Cristina Crocamo created the figures; Francesco Bartolind Giuseppe Carrà wrote the first and the second draft of theanuscript; Alex Baldacchino, Gerry Humphris, Julia Neufeind

evised and edited first and second draft of the manuscript. Alluthors contributed to and have approved the final manuscript.

onflict of interest

The authors have no conflict of interest in relationship to thisaper.

ppendix A. Supplementary data

Supplementary data associated with this article can be found,n the online version, at http://dx.doi.org/10.1016/j.drugalcdep.013.10.007.

eferences

haronovich, E., Liu, X., Nunes, E., Hasin, D.S., 2002. Suicide attempts in substanceabusers: effects of major depression in relation to substance use disorders. Am.J. Psychiatry 159, 1600–1602.

mato, L., Minozzi, S., Davoli, M., Vecchi, S., Ferri, M.M., Mayet, S., 2008. Psychosocialand pharmacological treatments versus pharmacological treatments for opioiddetoxification. Cochrane Database Syst. Rev. 4, CD005031.

merican Psychiatric Association, 2000. Diagnostic and Statistical Manual of MentalDisorders, 4th ed. American Psychiatric Association, Washington, DC (text rev.).

merican Psychiatric Association, 2013. Diagnostic and Statistical Manual of MentalDisorders, 5th ed. American Psychiatric Publishing, Arlington, VA.

rendt, M., Munk-Jørgensen, P., Sher, L., Jensen, S.O., 2011. Mortality among indi-viduals with cannabis, cocaine, amphetamine, MDMA, and opioid use disorders:a nationwide follow-up study of Danish substance users in treatment. DrugAlcohol Depend. 114, 134–139.

ackmund, M., Schuetz, C., Meyer, K., Edlin, B.R., Reimer, J., 2009. The risk of emer-gency room treatment due to overdose in injection drug users. J. Addict. Dis. 28,68–73.

est, D., Gossop, M., Man, L., Finch, E., Greenwood, J., Strang, J., 2000. Accidental anddeliberate overdose among opiate addicts in methadone maintenance treat-ment: are deliberate overdoses systematically different? Drug Alcohol Rev. 19,213–216.

ohnert, A.S., Roeder, K., Ilgen, M.A., 2010. Unintentional overdose and suicideamong substance users: a review of overlap and risk factors. Drug AlcoholDepend. 110, 183–192.

ohnert, A.S., Roeder, K.M., Ilgen, M.A., 2011. Suicide attempts and overdoses among

adults entering addictions treatment: comparing correlates in a U.S, NationalStudy. Drug Alcohol Depend. 119, 106–112.

rådvik, L., Frank, A., Hulenvik, P., Medvedeo, A., Berglund, M., 2007. Heroin addictsreporting previous heroin overdoses also report suicide attempts. Suicide LifeThreat. Behav. 37, 475–481.

ependence 134 (2014) 12– 21

Britton, P.C., Wines Jr., J.D., Conner, K.R., 2010. Non-fatal overdose in the 12 monthsfollowing treatment for substance use disorders. Drug Alcohol Depend. 107,51–55.

Brugal, M.T., Barrio, G., De La Fuente, L., Regidor, E., Royuela, L., Suelves, J.M., 2002.Factors associated with non-fatal heroin overdose: assessing the effect of fre-quency and route of heroin administration. Addiction 97, 319–327.

Carrà, G., Clerici, M., 2006. Dual diagnosis – policy and practice in Italy. Am. J. Addict.15, 125–130.

Carrà, G., Lusignani, G.S., Sciarini, P., Barale, F., Marinoni, A., Clerici, M., 2008b. Andhow shall we deal with adaptation and implementation of NICE schizophreniaguidelines in Italy? Epidemiol. Psichiatr. Soc. 17, 258–262.

Carrà, G., Johnson, S., 2009. Variations in rates of comorbid substance use in psychosisbetween mental health settings and geographical areas in the UK. Soc. PsychiatryPsychiatr. Epidemiol. 44, 429–447.

Carrà, G., Johnson, S., Bebbington, P., Angermeyer, M.C., Heider, D., Brugha, T.,Azorin, J.M., Toumi, M., 2012. The lifetime and past-year prevalence of dualdiagnosis in people with schizophrenia across Europe: findings from the Euro-pean Schizophrenia Cohort (EuroSC). Eur. Arch. Psychiatry Clin. Neurosci. 262,607–616.

Carrà, G., Montomoli, C., Monti, M.C., Clerici, M., 2008a. Does HIV serostatus affectoutcomes of dually diagnosed opiate dependents in residential treatment? Epi-demiol. Psichiatr. Soc. 17, 77–81.

Carrà, G., Scioli, R., Monti, M.C., Marinoni, A., 2006. Severity profiles of substanceabusing patients in Italian community addiction facilities: influence of psychi-atric concurrent disorders. Eur. Addict. Res. 12, 96–101.

Coffin, P.O., Tracy, M., Bucciarelli, A., Ompad, D., Vlahov, D., Galea, S., 2007. Identi-fying injection drug users at risk of nonfatal overdose. Acad. Emerg. Med. 14,616–623.

Conner, K.R., Pinquart, M., Holbrook, A.P., 2008. Meta-analysis of depression andsubstance use and impairment among cocaine users. Drug Alcohol Depend. 98,13–23.

Copeland, K.T., Checkoway, H., McMichael, A.J., Holbrook, R.H., 1977. Bias due to mis-classification in the estimation of relative risk. Am. J. Epidemiol. 105, 488–495.

Cornelius, J.R., Reynolds, M., Martz, B.M., Clark, D.B., Kirisci, L., Tarter, R., 2008. Pre-mature mortality among males with substance use disorders. Addict. Behav. 33,156–160.

Crowther, M., Lim, W., Crowther, M.A., 2010. Systematic review and meta-analysismethodology. Blood 116, 3140–3146.

Curran, G.M., Sullivan, G., Williams, K., Han, X., Collins, K., Keys, J., Kotrla, K.J., 2003.Emergency department use of persons with comorbid psychiatric and substanceabuse disorders. Ann. Emerg. Med. 41, 659–667.

Darke, S., Zador, D., 1996. Fatal heroin ‘overdose’: a review. Addiction 91, 1765–1772.Darke, S., Hall, W., 2003. Heroin overdose: research and evidence-based interven-

tion. J. Urban Health 80, 189–200.Darke, S., Mattick, R.P., Degenhardt, L., 2003. The ratio of non-fatal to fatal heroin

overdose. Addiction 98, 1169–1172.Darke, S., Mills, K., Teesson, M., Ross, J., Williamson, A., Havard, A., 2009. Patterns

of major depression and drug-related problems amongst heroin users across 36months. Psychiatry Res. 166, 7–14.

Darke, S., Ross, J., 2002. Suicide among heroin users: rates, risk factors and methods.Addiction 97, 1383–1394.

Degenhardt, L., Bucello, C., Mathers, B., Briegleb, C., Ali, H., Hickman, M., McLaren, J.,2011. Mortality among regular or dependent users of heroin and other opioids:a systematic review and meta-analysis of cohort studies. Addiction 106, 32–51.

Degenhardt, L., Hall, W., Warner-Smith, M., 2006. Using cohort studies to esti-mate mortality among injecting drug users that is not attributable to AIDS. Sex.Transm. Infect. 82, 56–63.

Delgadillo, J., Payne, S., Gilbody, S., Godfrey, C., Gore, S., Jessop, D., Dale, V., 2011.How reliable is depression screening in alcohol and drug users? A validation ofbrief and ultra-brief questionnaires. J. Affect. Disord. 134, 266–271.

Egger, M., Schneider, M., Davey Smith, G., 1998. Spurious precision? Metaanalysisof observational studies. BMJ 316, 140–144.

Elkader, A.K., Brands, B., Dunn, E., Selby, P., Sproule, B.A., 2009. Major depressivedisorder and patient satisfaction in relation to methadone pharmacokineticsand pharmacodynamics in stabilized methadone maintenance patients. J. Clin.Psychopharmacol. 29, 77–81.

Fairbairn, N., Wood, E., Stoltz, J.A., Li, K., Montaner, J., Kerr, T., 2008. Crystal metham-phetamine use associated with non-fatal overdose among a cohort of injectiondrug users in Vancouver. Public Health 122, 70–78.

Fendrich, M., Johnson, T.P., Sudman, S., Wislar, J.S., Spiehler, V., 1999. Validity ofdrug use reporting in a high-risk community sample: a comparison of cocaineand heroin survey reports with hair tests. Am. J. Epidemiol. 149, 955–962.

First, M.B., Spitzer, R.L., Gibbon, M., Williams, J.B.W., 2002. Structured ClinicalInterview for DSM-IV-TR Axis I Disorders, Research Version, Patient Edition(SCID-I/P). Biometrics Research. New York State Psychiatric Institute, New York.

Fischer, B., Brissette, S., Brochu, S., Bruneau, J., el-Guebaly, N., Noël, L., Rehm, J.,Tyndall, M., Wild, C., Mun, P., Haydon, E., Baliunas, D., 2004. Determinants ofoverdose incidents among illicit opioid users in 5 Canadian cities. CMAJ 171,235–239.

Green, T.C., McGowan, S.K., Yokell, M.A., Pouget, E.R., Rich, J.D., 2012. HIV infectionand risk of overdose: a systematic review and meta-analysis. AIDS 26, 403–417.

Grimes, D.A., Schulz, K.F., 2002. Bias and causal associations in observationalresearch. Lancet 359, 248–252.

Hakansson, A., Schlyter, F., Berglund, M., 2008. Factors associated with history ofnon-fatal overdose among opioid users in the Swedish criminal justice system.Drug Alcohol Depend. 94, 48–55.

Page 10

hol D

H

H

J

J

K

K

K

K

K

L

L

M

M

M

M

M

M

M

M

N

F. Bartoli et al. / Drug and Alco

asin, D., Liu, X., Nunes, E., McCloud, S., Samet, S., Endicott, J., 2002. Effects of majordepression on remission and relapse of substance dependence. Arch. Gen. Psy-chiatry 59, 375–380.

avens, J.R., Oser, C.B., Knudsen, H.K., Lofwall, M., Stoops, W.W., Walsh, S.L., Leuke-feld, C.G., Kral, A.H., 2011. Individual and network factors associated withnon-fatal overdose among rural Appalachian drug users. Drug Alcohol Depend.115, 107–112.

enkins, L.M., Banta-Green, C.J., Maynard, C., Kingston, S., Hanrahan, M., Merrill,J.O., Coffin, P.O., 2011. Risk factors for nonfatal overdose at Seattle-area syringeexchanges. J. Urban Health 88, 118–128.

epsen, P., Johnsen, S.P., Gillman, M.W., Sørensen, H.T., 2004. Interpretation of obser-vational studies. Heart 90, 956–960.

aye, S., Darke, S., 2004. Non-fatal cocaine overdose among injectingand non-injecting cocaine users in Sydney, Australia. Addiction 99,1315–1322.

err, T., Fairbairn, N., Tyndall, M., Marsh, D., Li, K., Montaner, J., Wood, E., 2007. Pre-dictors of non-fatal overdose among a cohort of polysubstance-using injectiondrug users. Drug Alcohol Depend. 87, 39–45.

ertesz, S.G., Khodneva, Y., Richman, J., Tucker, J.A., Safford, M.M., Jones, B.,Schumacher, J., Pletcher, M.J., 2012. Trajectories of drug use and mortal-ity outcomes among adults followed over 18 years. J. Gen. Intern. Med. 27,808–816.

inner, S.A., Milloy, M.J., Wood, E., Qi, J., Zhang, R., Kerr, T., 2012. Incidence and riskfactors for non-fatal overdose among a cohort of recently incarcerated illicitdrug users. Addict. Behav. 37, 691–696.

raus, L., Augustin, R., Frischer, M., Kümmler, P., Uhl, A., Wiessing, L., 2003. Estimat-ing prevalence of problem drug use at national level in countries of the EuropeanUnion and Norway. Addiction 98, 471–485.

emeshow, A.R., Blum, R.E., Berlin, J.A., Stoto, M.A., Colditz, G.A., 2005. Searching oneor two databases was insufficient for meta-analysis of observational studies. J.Clin. Epidemiol. 58, 867–873.

oder, E., Groves, T., Macauley, D., 2010. Registration of observational studies. BMJ340, c950.

aloney, E., Degenhardt, L., Darke, S., Nelson, E.C., 2009. Are non-fatal opioidoverdoses misclassified suicide attempts? Comparing the associated correlates.Addict. Behav. 34, 723–729.

athers, B.M., Degenhardt, L., Bucello, C., Lemon, J., Wiessing, L., Hickman, M.,2013. Mortality among people who inject drugs: a systematic review and meta-analysis. Bull. World Health Organ. 91, 102–123.

cGregor, C., Darke, S., Ali, R., Christie, P., 1998. Experience of non-fatal overdoseamong heroin users in Adelaide, Australia: circumstances and risk perceptions.Addiction 93, 701–711.

cIntosh, C., Ritson, B., 2001. Treating depression complicated by substance misuse.Adv. Psychiatr. Treat. 7, 357–364.

cLellan, A.T., Luborsky, L., Woody, G.E., O’Brien, C.P., 1980. An improved diagnos-tic evaluation instrument for substance abuse patients. The Addiction SeverityIndex. J. Nerv. Ment. Dis. 168, 26–33.

errall, E.L., Kariminia, A., Binswanger, I.A., Hobbs, M.S., Farrell, M., Marsden, J.,Hutchinson, S.J., Bird, S.M., 2010. Meta-analysis of drug-related deaths soon afterrelease from prison. Addiction 105, 1545–1554.

illoy, M.J., Fairbairn, N., Hayashi, K., Suwannawong, P., Kaplan, K., Wood, E., Kerr, T.,2010. Overdose experiences among injection drug users in Bangkok, Thailand.

Harm Reduct. J. 7, 9.

owla, A., Kianpor, M., Bahtoee, M., Sabayan, B., Chohedri, A.H., 2008. Comparisonof clinical characteristics of opium-induced and independent major depressivedisorder. Am. J. Drug Alcohol Abuse 34, 415–421.

eale, J., 2000. Suicidal intent in non-fatal illicit drug overdose. Addiction 95, 85–93.

ependence 134 (2014) 12– 21 21

Pani, P.P., Vacca, R., Trogu, E., Amato, L., Davoli, M., 2010. Pharmacological treatmentfor depression during opioid agonist treatment for opioid dependence. CochraneDatabase Syst. Rev. 9, CD008373.

Pedrelli, P., Iovieno, N., Vitali, M., Tedeschini, E., Bentley, K.H., Papakostas, G.I., 2011.Treatment of major depressive disorder and dysthymic disorder with antide-pressants in patients with comorbid opiate use disorders enrolled in methadonemaintenance therapy: a meta-analysis. J. Clin. Psychopharmacol. 31, 582–586.

Phillips, J., Carpenter, K.M., Nunes, E.V., 2004. Suicide risk in depressed methadone-maintained patients: associations with clinical and demographic characteristics.Am. J. Addict. 13, 327–332.

Pollini, R.A., McCall, L., Mehta, S.H., Vlahov, D., Strathdee, S.A., 2006. Nonfatal over-dose and subsequent drug treatment among injection drug users. Drug AlcoholDepend. 83, 104–110.

Ravndal, E., Vaglum, P., 1994. Self-reported depression as a predictor of dropout ina hierarchical therapeutic community. J. Subst. Abuse Treat. 11, 471–479.

Regier, D.A., Farmer, M.E., Rae, D.S., Locke, B.Z., Keith, S.J., Judd, L.L., Goodwin,F.K., 1990. Comorbidity of mental disorders with alcohol and other drugabuse. Results from the Epidemiologic Catchment Area (ECA) study. JAMA 264,2511–2518.

Riley, R.D., Higgins, J.P., Deeks, J.J., 2011. Interpretation of random effects metaanal-yses. BMJ 342, d549.

Rosenberg, S.D., Goodman, L.A., Osher, F.C., Swartz, M.S., Essock, S.M., Butterfield,M.I., Constantine, N.T., Wolford, G.L., Salyers, M.P., 2001. Prevalence of HIV, hep-atitis B, and hepatitis C in people with severe mental illness. Am. J. Public Health91, 31–37.

Rossow, I., Lauritzen, G., 1999. Balancing on the edge of death: suicide attempts andlife-threatening overdoses among drug addicts. Addiction 94, 209–219.

Sheehan, D.V., Lecrubier, Y., Sheehan, K.H., Amorim, P., Janavs, J., Weiller, E., Her-gueta, T., Baker, R., Dunbar, G.C., 1998. The Mini-International NeuropsychiatricInterview (M.I.N.I.): the development and validation of a structured diag-nostic psychiatric interview for DSM-IV and ICD-10. J. Clin. Psychiatry 59,22–33.

Stein, M.D., Herman, D.S., Solomon, D.A., Anthony, J.L., Anderson, B.J., Ramsey, S.E.,Miller, I.W., 2004. Adherence to treatment of depression in active injection drugusers: the Minerva study. J. Subst. Abuse Treat. 26, 87–93.

Stein, M.D., Solomon, D.A., Herman, D.S., Anderson, B.J., Miller, I., 2003. Depres-sion severity and HIV injection risk behaviors. Am. J. Psychiatry 160,1659–1662.

Sterne, J.A.C., Egger, M., Moher, D., 2008. Chapter 10: Addressing reporting biases.In: Higgins, J.P.T., Green, S. (Eds.), Cochrane Handbook for Systematic Reviewsof Interventions: Cochrane Book Series. John Wiley & Sons, Ltd, Chichester, UK,Available from www.cochrane-handbook.org (accessed on 13.07.13).

Stoové, M.A., Dietze, P.M., Jolley, D., 2009. Overdose deaths following previousnon-fatal heroin overdose: record linkage of ambulance attendance and deathregistry data. Drug Alcohol Rev. 28, 347–352.

Stroup, D.F., Berlin, J.A., Morton, S.C., Olkin, I., Williamson, G.D., Rennie, D.,Moher, D., Becker, B.J., Sipe, T.A., Thacker, S.B., 2000. Meta-analysis ofobservational studies in epidemiology: a proposal for reporting, Meta-analysis Of Observational Studies in Epidemiology (MOOSE) group. JAMA 283,2008–2012.

Warner-Smith, M., Darke, S., Lynskey, M., Hall, W., 2001. Heroin overdose: causesand consequences. Addiction 96, 1113–1125.

World Health Organization, 1992. ICD-10 Classifications of Mental and BehaviouralDisorder: Clinical Descriptions and Diagnostic Guidelines. World Health Orga-nization, Geneva.

World Health Organization, 1993. Composite International Diagnostic Interview(version 1.1). World Health Organization, Geneva.