Drug and Alcohol Dependence 162 (2016) 176–181
Contents lists available at ScienceDirect
Drug and Alcohol Dependence journal homepage: www.elsevier.com/locate/drugalcdep
Full length article
Risk factors for unnatural death: Fatal accidental intoxication, undetermined intent and suicide: Register follow-up in a criminal justice population with substance use problems Martin O. Olsson ∗ , Louise Bradvik, Agneta Öjehagen, Anders Hakansson Lund University, Department of Clinical Sciences Lund, Psychiatry, Psykiatri Lund, Baravägen 1, S-221 85 Lund, Sweden, Sweden
a r t i c l e
i n f o
Article history: Received 27 October 2015 Received in revised form 2 March 2016 Accepted 7 March 2016 Available online 17 March 2016 Keywords: Suicide Overdose Undetermined intent Criminal justice Substance use disorders Prison
a b s t r a c t Background: Risk factors for suicide and fatal accidental intoxication are extensively studied, while risk factors for intoxications/injuries of undetermined intent are less well known. The latter have shown an overlap with suicides, but also with fatal accidental intoxications. The objective was to analyze potential differences and similarities in the patterns of risk factors for accidental intoxications, injuries/intoxications with undetermined intent, and suicides, respectively. Methods: A follow-up register study was conducted, using data from ASI interviews with clients in the criminal justice system in Sweden (n = 6744), followed in the National Causes of Death Register. A set of risk factors from the ASI interview were tested in bivariate analysis with the respective cause of death, yielding significant risk factors further analyzed in three Cox regression models. Results: In Cox regression analyses, death from fatal accidental intoxication was associated with male gender (HR 4.09), use of heroin (HR 2.86), and use of cannabis (HR 1.94), and death from intoxication/injury of undetermined intent was associated with use of heroin (HR 3.48), binge drinking of alcohol (HR 2.46) and previous psychiatric hospitalization (HR 2.41), while negatively associated with depression (HR 0.33). Death from suicide was associated with previous suicide attempts (HR 2.78) and use of sedatives (HR 2.17). Conclusions: In this population of criminal justice clients with reported substance use problems, fatal injuries/intoxications with undetermined intent – like fatal accidental intoxications – appear to be associated with substance use variables, and cannot readily be assumed to represent the same background factors as suicide. © 2016 Published by Elsevier Ireland Ltd.
1. Introduction Substance use disorders, either alone or in combination with other psychiatric disorders, constitute important risk factors for premature death, both from natural and unnatural causes. Natural death, in medico-legal practice, is caused by disease only. Unnatural deaths are results from intoxication or trauma (accidents, suicide, death of undetermined intent, and homicide/manslaughter). Self-inflicted death is categorized in ICD-10 as accidental (X40–49), of undetermined intent (Y10–34), or intentional (X60–84). Recent research (Bohnert et al., 2013) has pointed to a risk of misclassification of suicide deaths, where coroners are likely to use the categories accidental intoxication or undetermined
∗ Corresponding author. E-mail address: Martin
[email protected] (M.O. Olsson). http://dx.doi.org/10.1016/j.drugalcdep.2016.03.009 0376-8716/© 2016 Published by Elsevier Ireland Ltd.
intent, especially for individuals with a history of substance use disorder. The same author (Bohnert et al., 2012) points to a risk of overlap between the different categories: while some cases of intoxication registered as ‘unintentional’ or ‘undetermined’ could be misclassified suicide deaths, a psychiatric history could as well have a strong association with unintentional, accidental death. In two meta-analyses, Harris and Barraclough (1997, 1998) concluded that both alcohol and drug use disorders, constituted risks for premature death from unnatural causes, primarily suicidal or quasi-suicidal deaths. They found an increased risk for completed suicide, amounting to 5.5 times higher for alcohol-dependent males and 18 times higher for females. Similarly, the mortality risk for completed suicide for opioid users was 10 times higher than expected; for overdose cases, not classified as suicides, and other violent causes, the mortality risk was 13 times higher. Wilcox et al. 2004 augmented these findings, showing a robust relationship between completed suicide and alcohol use disorder, but they also
M.O. Olsson et al. / Drug and Alcohol Dependence 162 (2016) 176–181
found a relationship between heavy drinking, opioid use disorder, intravenous drug use and mixed drug use and suicide. Thus, in subjects with a substance use disorder, self-inflicted death, such as fatal accidental intoxication, intoxication with undetermined intent and suicide, are over-represented, both for individuals with alcohol use disorder and drug use disorder. Risk factors for a fatal accidental intoxication among substance users have been examined in several studies. Studies of death records show that polysubstance use and toxic interactions, including interactions with alcohol and prescription drugs, are important risk factors for fatal accidental intoxications, especially for heroin users (Darke et al., 2000; Darke and Ross, 2001; Coffin et al., 2003; Cone et al., 2004; Seymour et al., 2000; Darke et al., 2010). In prison populations, in addition to substance-related causes for unnatural death, several psychiatric conditions are known to be over-represented (Fazel and Danesh, 2002), constituting risk factors for unnatural death. A cohort study of criminal justice clients in the Causes of Death Register in Sweden (Hakansson and Berglund, 2013) showed that 84% of the deceased individuals died from unnatural or substance-related causes. Common causes of unnatural death were accidental poisoning (27%), poisoning/injury with undetermined intent (12%) and suicide (10%). In criminal justice populations, the risks for unnatural death by fatal accidental intoxication (Merrall et al., 2010) or suicide (Pratt et al., 2006) are known to be highly increased after release from prison, which calls for further attention for the study of risk factors of these events in criminal justice clients. In population-based surveys of deaths by poisoning or suicides in general (Runeson et al., 2010; Hawton et al., 2013), suicide and death of undetermined intent are merged together in the analyses, thus possibly overlapping with overdose cases. Runeson et al. (2010) showed, in a follow-up of a Swedish long-term cohort, that the most common method of suicide (including ‘uncertain suicide’) at follow-up was poisoning, further indicating the possible overlap between substance-induced death and suicide. Björkenstam et al. 2014 compared, in a register study, suicides with deaths classified as undetermined intent in the general population in Sweden. They found that ‘undetermined intent’ was more common among individuals who had been hospitalized for substance abuse. Bohnert et al. (2010) investigated deaths registered by coroners and medical examiners as involving an undetermined intent in a cohort of Veterans Affairs patients, and found that substance use disorders had a stronger association with ‘indeterminate intent overdoses’ than ‘intentional overdoses’, as well as with unintentional overdoses than with the intentional ones. The same author concluded, in an extensive review of the literature, that ı´further research is needed to understand the overlap of, and differences between, suicide and accidental overdose among individuals who misuse substances, particularly individuals who primarily use substances other than heroin (Bohnert et al., 2010). The present study aimed to analyze potential differences and similarities in the patterns of risk factors for fatal accidental intoxications, injuries/intoxications with undetermined intent, and suicides, respectively. A follow-up register study was conducted, combining national register data on causes of death in Sweden with demographic, psychiatric and drug use data from prison and probation clients with substance use problems.
2. Methods 2.1. Baseline data This prospective follow-up study is based on a database of criminal justice clients with substance use problems, assessed with the Addiction Severity Index (ASI; McLellan et al., 1980, 1992, 2006).
177
Since 2001, this has been implemented by the Swedish Prison and Probation services in an increasing number of correctional facilities in Sweden (Tengvald et al., 2004). The ASI is a semi-structured interview instrument which examines substance-related problems in different aspects of life, widely used as a tool for assessment in clinical work and in addiction research (McLellan et al., 2006). Several previous analyses have been published based on the present database of ASI assessments, collected between 2001 and 2006, when data were blinded and delivered to the research group (Hakansson and Berglund, 2013; Hakansson et al., 2008, 2009, 2011). The version used, ASI-X (Öberg et al., 1999), is a modified version of ASI, based on the European standard version EuropASI (Kokkevi and Hartgers, 1995). The present database contains interviews with 7085 clients, assessed in prison (72%), on probation (17%), on remand, i.e., held in custody before trial (5%) or, for the remaining clients, in other correctional institutions. From the original sample of 7085 clients, 341 clients were excluded because they did not report a substance use problem, or were assessed with the adolescent instrument version ADAD (Friedman and Utada, 1989) instead of the ASI, or were excluded because of failure to complete the interview or because answers were judged to be distorted by misinterpretation or inability to understand the questions. Thus, the final data were based on 6744 individuals. Consistent with the intention of the Swedish Prison and Probation Service, in the efforts to enhance focus on drug problems, the database assessed is an overrepresentation of illicit drugs compared to alcohol, as well as an overrepresentation of women in comparison to the whole Swedish criminal justice system. There is also an overrepresentation of clients sentenced for acquisitive crimes (such as stealing and shoplifting) and drug crimes. In this sample (N = 6744) the primary drugs, i.e., the dominating problem, were the following: amphetamine (32%), polydrug use (more than two illicit substances on one day, 17%), cannabis (13%), alcohol (12%) heroin (9%) alcohol and drugs combined (7%), sedatives (3%), cocaine (3%) and less than 1% for opioid analgesics and methadone, inhalants and hallucinogenic drugs. Two percent reported that their primary drug was ‘other drugs’. The attrition rate was low; from previous investigations of the cohort by Hakansson et al. (2008, 2009, 2010, 2011), using post hoc surveys to the interviewers, it has been estimated that around 6% of the clients approached for assessment refused the interview. The time span elapsed from intake to the criminal justice facility (prison or probation office) where the client was interviewed, to the interview, was 60 days (median 27 days; 98% were interviewed within the first year). Baseline data were the following factors, identified in the ASI instrument, thus constituting the independent variables. 2.1.1 Demographic factors were: age, gender, homelessness 30 days before incarceration, country of birth (born in Nordic countries [Sweden, Denmark, Finland, Iceland and Norway], versus other countries). 2.1.2 Psychiatric factors were previous (either 30 days prior to incarceration or previously in life) psychiatric hospitalization, suicide attempt, depression, anxiety, cognitive symptoms, hallucinations, or difficulties controlling violent behavior. 2.1.3 Substance use factors (substances used during the last 30 days before incarceration) were: binge drinking of alcohol, use of heroin, other opioids, sedatives/tranquilizers, cocaine, amphetamine, cannabis, or injection drug use, as well as any lifetime history of drug overdose. In the ASI instrument, overdose is defined as ‘life-threatening physical complications following drug intake’ (Öberg et al., 1999). 2.1.4 Criminality factors were: main type of crime in the current verdict: violent crime, property crime, or drug crime
178
M.O. Olsson et al. / Drug and Alcohol Dependence 162 (2016) 176–181
Table 1 Clinical characteristics. Baseline data from ASI interview. Client characteristics
Fatal accidental intoxication X40–X49 n = 66 (n/%)
Injury/intoxication of undetermined intent Y10–Y34 n = 32 (n/%)
Suicide X60–X84 n = 26 (n/%)
Alive or deceased from another cause n = 6620 (n/%)
Age (mean) Male gender Homeless Born in Nordic countries Psychiatric hospitalization Suicide attempt Depression Anxiety Cognitive symptoms Hallucinations Violent behavior Alcohol (binge drinking) Heroin Other opioids Sedatives Cocaine Amphetamine Cannabis Injection use Overdose Violent crime Property crime Drug crime
31.3 64 (97.0)* 13 (19.7) 56 (84.8) 12 (18.2) 18 (27.3) 28 (42.4) 33 (50.0) 36 (64.5) 8 (12.1) 22 (33.3) 28 (42.4) 33 (50.0)*** 15 (22.7)* 34 (51.5)*** 8 (12.1) 38 (57.6) 52 (78.8)*** 42 (63.6)*** 32 (48.5)*** 10 (15.2) 19 (28.8) 15 (22.7)
31.2 31 (96.9) 6 (18.8) 29 (90.6) 9 (28.1)* 7 (21.9) 11 (34.4)* 22 (68.8) 20 (62.5) 3 (9.4) 14 (43.8) 21 (65.6)** 16 (50.0)*** 7 (21.9) 13 (40.6) 4 (12.5) 14 (43.8) 24 (75.0)* 18 (56.3) 15 (46.9)** 3 (9.4) 8 (25.0) 6 (18.8)
31.0 22 (84.6) 3 (11.5) 23 (88.5) 5 (19.2) 12 (46.2)** 17 (65.4) 15 (57.7) 13 (50.0) 3 (11.5) 12 (46.2) 13 (50.0) 3 (11.5) 0 (0.0) 13 (50.0)* 2 (7.7) 10 (38.5) 11 (42.3) 9 (34.6) 8 (30.8) 4 (15.4) 6 (23.1) 4 (15.4)
33.0 5843 (88.3) 1169 (17.7) 5451 (82.3) 997 (15.1) 1402 (21.2) 3431 ((51.9) 3568 (53.9) 3488 (52.7) 864 (13.1) 2749 (41.5) 2775 (41.9) 1170 (17.7) 868 (13.1) 1912 (28.9) 3460 (52.3) 3480 (52.6) 3480 (52.6) 2790 (42.1) 1551 (23.4) 814 (12.3) 1430 (21.6) 1545 (23.3)
Binary comparisons (chi-square test for categorical variables and Student’s t-tests for continuous variables) between each category of self-inflicted death and the group alive, or deceased from other causes (n = 6620). *p < 0.05, **p < 0.01, ***p < 0.001 compared to group comprising alive subjects and subjects deceased from other causes.
2.2. The swedish national causes of death register Mortality data were collected from the national Causes of Death Register, held by the Swedish National Board of Health and Welfare, where deaths were registered until censorship on December 31, 2008. According to Swedish legislation, all unnatural deaths, as well as deaths in ‘known substance abusers’, are to be examined post mortem by forensic medical examiners, after a formal application to the police authority. Although the present register data do not report systematically whether a case has been assessed in a forensic medical examination, it can be expected, given the legislation, that a large proportion of the deaths in this population have been subject to this examination. For the cases of self-inflicted death, the diagnoses used for causes of death are either ‘accidental intoxication’ (X40–49), ‘suicide or injury of undetermined intent’ (Y10–34), or ‘suicide’ (X60–84), according to the assessment of the certifying physician. The categories were studied separately, comparing each category with the group of subjects who survived or died from other causes, not related to any of these three diagnostic categories. At follow-up, in total, 166 individuals were deceased, and amongst them, 66 died from accidental poisoning, 32 from intoxication or injury of undetermined intent (30 from intoxication, two from injury), and 26 from suicide. Each category was compared for itself, with those who survived, or had died by another cause (n = 6620). After excluding two cases of injuries from the category ‘injury/intoxication of undetermined intent’, a sensitivity analysis was performed, yielding a subgroup of only subjects deceased from ‘intoxication of undetermined intent’.
2.3. Statistical analyses Binary associations between the different categories of selfinflicted death and baseline data were calculated with chi square analysis for categorical data and Student’s t-test for continuous variables. The time of follow-up varied between the subjects (from
prison release until death or censorship), and therefore, a Cox regression survival analysis was chosen for multivariate analysis. Outcome variables were suicide, injury/intoxication of undetermined intent, and fatal accidental intoxication, respectively, each compared to subjects alive or deceased from other causes. Among the baseline factors listed above, those with a significant association in the binary analysis were analyzed in three separate Cox regression models, yielding Hazard Ratios (HR) as a measure of increased risk for each of the causes of death assessed here. In each of the three Cox regression models, we compared the outcome variables with subjects alive or deceased from other causes. The time variable was defined as time under risk from release from prison (for clients interviewed in prison) or from the interview (for clients interviewed on probation or remand [held in custody before trial]) until death or until data were censored, on December 31, 2008. Calculations were made using 95% confidence intervals. Statistical analyses were carried out in SPSS software, version 20. The study was approved by the Lund Medical Ethics Committee, Sweden.
3. Results Clients were followed for an average of 1402 days (3.8 years). Those deceased from fatal accidental intoxication (n = 66) were followed for an average of 861 days (2.4 years), from injury/intoxication of undetermined intent (n = 32) for 1022 days (2.8 years) and from suicide (n = 26) for 923 days (2.5 years). Baseline characteristics and binary associations with the cause of death are reported in Table 1. In binary analyses, significant associations with fatal accidental intoxication were seen for male gender, use of heroin, other opioids, sedatives/tranquilizers, use of cannabis, injection drug use and previous overdoses. Significant associations with injury/intoxication of undetermined intent were seen for previous psychiatric hospitalization, binge drinking of alcohol, and use of heroin, cannabis, and previous overdose. Significant association with death from
M.O. Olsson et al. / Drug and Alcohol Dependence 162 (2016) 176–181 Table 2 Significant risk factors for different causes of self-inflicted death. Cox regression. Risk factor
Hazard ratio
Fatal accidental intoxication X40–X49 (n = 66) 4.09 Male gender Use of heroin 2.86 Use of cannabis 1.94
95% CI 1.00a –16.75 1.59–5.14 1.03–3.67
p-Value p = 0.05 p < 0.001 p < 0.05
Injury/intoxication with undetermined intent Y10–Y34 (n = 32) 3.48 1.56–7.73 Use of heroin 2.41 1.08–5.35 Psychiatric hospitalization 2.46 1.17–5.18 Alcohol, binge drinking Depression 0.33 0.16–0.71
p < 0.01 p < 0.05 p < 0.05 p < 0.01
Suicide X60–X84 (n = 26) Previous suicide attempt Use of sedatives/tranquilizers
p < 0.05 p = 0.05
a
2.78 2.17
1.28–6.05 1.00a –4.70
Rounded off to two decimals. Confidence interval includes 1.
suicide were seen for previous suicide attempt and use of sedatives/tranquilizers. In the following Cox regression analyses, death from a fatal accidental intoxication was associated with use of heroin and use of cannabis, and with marginal significance, by male gender. Death from injury/intoxication of undetermined intent was positively associated with use of heroin, binge drinking of alcohol, and previous psychiatric hospitalization, and negatively associated with depression. After performing the sensitivity analysis, death from intoxication of undetermined intent was positively associated with use of heroin and binge drinking of alcohol, and negatively associated with depression. Death from suicide was positively associated with previous suicide attempts, and, with marginal significance, with the use of sedatives/tranquilizers (Table 2). 4. Discussion Our main finding is that in this sample of criminal justice clients, risk factors differ for the different diagnostic categories of self-inflicted death investigated. In particular, risk factors for selfinflicted death with undetermined intent differed from those of accidental deaths, and, importantly, differed from those of suicide deaths. These events with undetermined intent likely cannot be collapsed into the suicide category, given the discrepancy described here between risk factors for these two categories, respectively. Fatal accidental intoxications, often referred to as “overdoses”, in this sample, were positively associated with use of heroin, use of cannabis, and marginally significant, male gender. The finding that use of heroin is positively associated with fatal accidental intoxication is consistent with the literature on heroin overdose (Darke and Ross, 2000, 2002; Darke et al., 2010), describing the high prevalence of both fatal and non-fatal overdoses among heroin users. It is, however, intuitively surprising, that a history of heroin overdose is not associated with a later fatal accidental intoxication in this population of criminal justice clients with substance use problems. It is known, that the risk of overdose increases in polydrug use, especially when combinations of benzodiazepines or alcohol are used together with opioids. Such fatal overdoses are well known to occur accidentally, typically after release from prison and in opioiddependent subjects outside maintenance treatment (Darke et al., 2000; McGregor et al., 1998). The definition of “overdose” in the ASI manual is, however, not restricted to opioid overdosing, but describes unintentional, excess intake of a drug, a circumstance that could lead to reporting alleged “overdoses” of drugs without a fatal outcome (Hakansson and Berglund, 2013). The finding that use of cannabis is positively associated with a fatal accidental intoxication is interesting. While smoking of cannabis in itself cannot cause an overdose, it cannot be excluded that cannabis use in the present sample represents a proxy variable,
179
associated with a higher risk of a fatal accidental intoxication, for example through a more pronounced polysubstance use pattern. Death from injury/intoxication of undetermined intent was positively associated with use of heroin, binge drinking of alcohol and by previous psychiatric hospitalization, whereas it was negatively associated with depression. Thus, this category of causes of death was associated with two substance variables, and including an association with heroin use at least as strong as that of fatal accidental intoxication, and including a clear association with binge drinking of alcohol, a factor that can cause a fatal accidental intoxication in combination with other drugs. Given the substance-related risk factors for this category of causes of death, the present findings suggest that self-inflicted death with undetermined intent at least cannot readily be assumed to be comparable with suicide in populations with a problematic substance use. This is consistent with the findings of Björkenstam et al., (2014) who found, in a population-based register study, that individuals hospitalized for substance abuse were more likely to be classified as undetermined intent suicides. We could also conclude that there is a difference within a sample of subjects with substance use problems. Death from suicide was positively associated with previous suicide attempts, and, with marginal significance, by use of sedatives/tranquilizers. The association between previous suicide attempts and completed suicide is well established in the literature (Runeson et al., 2010; Niméus et al., 2002; Bradvik and Berglund, 2002; Skogman et al., 2004; Tidemalm et al., 2008), thus indicating its predictive value both in the general population and in this population of criminal justice clients. It is possible that use of sedatives/tranquilizers is an indicator, or a proxy variable, for psychiatric co-morbidity in this sample of criminal justice clients. Use of sedatives had an association with future psychiatric hospitalization in a previous study of this sample of criminal justice clients (Olsson et al., 2015). An association between psychiatric co-morbidity and use of sedatives/tranquilizers has been shown among individuals with substance use problems in the general population (Schuckit et al., 2002; Schepis and Hakes, 2013). Interestingly, a history of depression, a well-established risk factor for suicide (Holmstrand et al., 2015; Mattisson et al., 2015; Tidemalm et al., 2008), was not associated with suicide, but had a negative association with injury/intoxication of undetermined intent. It cannot be excluded that the reporting of a history of depression according to the ASI interview – based on selfreported data – in fact corresponds to a description of a negative mood resulting from active drug use. The ASI interview is not a diagnostic instrument, but can give information on self-reported symptoms. At the same time, the substance use could possibly mask a depression. The negative association with depressive symptoms at baseline is somewhat difficult to interpret. It cannot be excluded, that a history of depression may be protective in the present population, possibly through an additional mediating factor. More research is needed, in order to elucidate the generalizability and implications of this finding. This further supports the assumption that events of undetermined intent are different, compared to suicides, in this population. Traditionally, in suicide research, death of undetermined intent is often merged with suicides (Runeson et al., 2010). Specifically, in studies examining death of undetermined intent in the general population, authors have pointed to a possible undercounting of suicides, taking into account the psychiatric history of the victims (Carr et al., 2004; Öhberg and Lönnqvist, 1998). In contrast, our results indicate that the risk factors for injury/intoxication of undetermined intent had more in common with fatal accidental intoxications than with suicide, a possible illustration of the fact that an individual with a substance use problem is more susceptible to experiencing intoxications unintentionally, because of the
180
M.O. Olsson et al. / Drug and Alcohol Dependence 162 (2016) 176–181
pattern of drug use, and because of the pharmacological effects of a combination of drugs—illegal, or legally prescribed (Coffin et al., 2003; Cone et al., 2004). Our study has therefore, in contrast, pointed to a possible undercounting of fatal accidental intoxications in the category injury/intoxication of unknown intent, in this population of criminal justice clients with substance use problems.
4.1. Strengths and limitations An important strength of the study lies in the sample size, in addition to the high rates of available causes of death in national registers used here. The study uses prospective data, and the sample size makes it possible to calculate statistical associations with acceptable power despite the study of relatively rare events. Also, the ASI is a well-documented instrument, regarded as a standard instrument in addiction research. An important limitation of the present paper is the low number of cases deceased from each separate cause of self-inflicted death, due to the low incidence in the population. This is particularly an issue for the categories of suicides and events of undetermined intent, and mainly in relation to the number of potential risk factors possible to include in the model (Vittinghoff and McCulloch, 2007). For these reasons, we attempted to limit the number of variables, introducing only those factors into the Cox regression models, which appeared to be significant in the initial binary models. In all three Cox regression analyses, for each type of self-inflicted death, respectively, clients deceased from the two remaining types of self-inflicted death were excluded, in order to allow for a conservative estimate of correlates of each type of self-inflicted death. While this implies a certain risk of bias due to competing risks related to different causes of death, the demonstration of potential differences in risk factor patterns between the various categories of self-inflicted death is rather conservative than the opposite. The study is restricted to criminal justice clients, which means that the possibility to generalize the findings to the general population is limited. However, the risk of death by suicide or fatal accidental intoxication is increased among criminal justice clients (Pratt et al., 2006; Merrall et al., 2010), which calls for attention to this specific group. Additionally, the majority of the sample is male, which also restricts the generalizability. Baseline data are based on self-reported symptoms, an obvious limitation, as well as the variation in the time span from incarceration to the ASI interview, which may increase the risk of recall bias. The time span from the ASI interview to the actual fatality is also a limitation. We have studied correlates and not proximal risk factors for death, which means that we have no information on the situation immediately prior to the self-inflicted death. In addition, we have not had access to data about the cases of death beyond the diagnoses causing death according to the death certificate. For example, we did not have access to toxicological analyses or detailed reporting of the circumstances preceding the fatalities. Instead, however, causes of death data were based on death certificates issued by the physician responsible for the reporting of causes of death for medico-legal purposes, according to a structured diagnostic classification. Thus, as the present study is a naturalistic observational study rather than an experimental study, the study cannot control for subjective differences in diagnostic procedures between physicians issuing death certificates, but instead highlights the potential correlates of fatalities deemed to be selfinflicted but with an undetermined intent. The present findings, despite potential limitations, may add to the present knowledge of risk factors for common causes of death among clients in the criminal justice system.
5. Conclusion Our findings indicate, in this sample of criminal justice clients with a substance use problem, that injury/intoxication of undetermined intent may be a phenomenon distinct from suicides, more closely linked to substance use characteristics than to suicides. Further research is needed in order to investigate predictors for this category of self-inflicted death among substance users in general, in non-forensic settings. The question of unnatural deaths after prison sentences is complex and must be addressed by both health care providers and the criminal justice system. Hopefully, the classification of unnatural deaths in forensic practice, as well as epidemiological reporting of causes of death, could benefit from our findings, since we think they are relevant, although limited. Epidemiological reporting on causes of death may tend to be inaccurate, if deaths of undetermined intent in substance users are reported as suicides. Conflict of interests The authors have no conflicts of interests to report related to the present study. Role of funding source While the present project partially was funded by the Research and Development Department at Region Skane, this institution did not have any influence on the study design or on the writing of the paper. Contributors Olsson and Hakansson planned and conducted literature search for the project. Hakansson was responsible for obtaining data and the statistical analysis. Olsson was the main responsible for the writing of the manuscript. Hakansson, Bradvik and Öjehagen contributed to, and approved the final manuscript. References Björkenstam, C., Johansson, L.-A., Nordström, P., Thiblin, I., Fugelstad, A., Hallqvist, J., Ljung, R., 2014. Suicide or undetermined intent? A register-based study of signs of misclassification. Popul. Health Metr. 12, 11. Bohnert, A.S.B., Roeder, K., Ilgen, M.A., 2010. Unintentional overdose and suicide among substance users: a review of overlap and risk factors. Drug Alcohol Depend. 110, 183–192. Bohnert, A.S.B., Ilgen, M.A., Ignacio, R.V., McCarthy, J.F., Valenstein, M., Blow, F.C., 2012. Risk of death from accidental overdose associated with psychiatric and substance use disorders. Am. J. Psychol. 169, 64–70. Bohnert, A.S.B., McCarthy, J.F., Ignacio, R.V., Ilgen, M.A., Eisenberg, A., Blow, F.C., 2013. Misclassification of suicide deaths: examining the psychiatric history of overdose decedents. Inj. Prev 19, 326–330. Bradvik, L., Berglund, M., 2002. Aspects of the suicidal career in severe depression. A comparison between attempts in suicides and controls. Arch. Suicide Res. 6, 339–349. Carr, J.R., Hoge, C.W., Gardner, J., Potter, R., 2004. Suicide surveillance in the US military—reporting and classification biases in rate calculations. Suicide Life Threat. Behav. 34, 233–241. Coffin, P.O., Galea, S., Ahern, J., Leon, A.C., Vlahov, D., Tardiff, K., 2003. Opiates, cocaine and alcohol combinations in accidental overdose deaths in New York City. Addiction 98, 739–747. Cone, E.J., Fant, R.V., Rohay, J.M., Caplan, Y.H., Ballina, M., Reder, R.F., Haddox, J.D., 2004. 2004 Oxycodone involvement in drug abuse deaths: II. Evidence for toxic multiple drug–drug interactions. J. Anal. Toxicol. 28, 616–624. Darke, S., Ross, J., 2001. The relationship between suicide and heroin overdose among maintenance patients in Sydney Australia. Addiction 96, 1443–1453. Darke, S., Ross, J., 2002. Suicide among heroin users: rates, risk factors and methods. Addiction 97, 1383–1394. Darke, S., Ross, J., Zador, J., Sunjic, S., 2000. Heroin-related deaths in new South Wales Australia, 1992–1996. Drug Alcohol Depend. 60, 141–150. Darke, S., Mills, K.L., Ross, J., Teeson, M., 2010. Rates and correlates of mortality amongst heroin users: findings from the Australian Treatment Outcome Study (ATOS), 2001–2009. Drug Alcohol Depend. 115, 190–195.
M.O. Olsson et al. / Drug and Alcohol Dependence 162 (2016) 176–181 Fazel, S., Danesh, J., 2002. Serious mental disorder in 23,000 prisoners: a systematic review of 62 surveys. Lancet 359, 545–550. Friedman, A., Utada, A., 1989. A method for diagnosing and planning the treatment of adolescent drug abusers (the Adolescent Drug Abuse Diagnosis (ADAD) instrument). J. Drug Educ. 19, 285–312. Hakansson, A., Berglund, M., 2013. All-cause mortality in criminal justice clients with substance use problems—a prospective follow-up study. Drug Alcohol Depend. 133, 499–504. Hakansson, A., Schlyter, F., Berglund, M., 2008. Factors associated with history of non-fatal overdose among opioid users in the Swedish criminal justice system. Drug Alcohol Depend. 94, 48–55. Hakansson, A., Schlyter, F., Berglund, M., 2009. Characteristics of primary amphetamine users in Sweden: a criminal justice population examined with the Addiction Severity Index (ASI). Eur. Addict. Res. 15, 10–18. Hakansson, A., Brådvik, L., Schlyter, F., Berglund, M., 2010. Factors associated with the history of attempted suicide. Crisis 31, 12–21. Hakansson, A., Schlyter, F., Berglund, M., 2011. Associations between polysubstance use and psychiatric problems in a criminal justice population in Sweden. Drug Alcohol Depend. 118, 5–11. Harris, E.C., Barraclough, B., 1997. Suicide as an outcome for mental disorders. A meta-analysis. Br. J. Psychiatry 170, 205–228. Harris, E.C., Barraclough, B., 1998. Excess mortality of mental disorder. Br. J. Psychiatry 173, 11–53. Hawton, K., Bergen, H., Simkin, S., Dodd, S., Pocock, P., Bernal, W., et al., 2013. Long term effect of reduced pack sizes of paracetamol in poisoning deaths and liver transplant activity in England and Wales: interrupted time series analyses. BMJ 346, f403. Holmstrand, C., Bogren, M., Mattisson, C., Brådvik, L., 2015. Long-term suicide risk in people with no, one or more mental disorders, the Lundby study 1947–1997. Acta Psychiatr. Scand., 1–11 [Epub ahead of print]. Kokkevi, A., Hartgers, C., 1995. EuropASI: European adaptation of a multidimensional assessment instrument for drug and alcohol dependence. Eur. Addict. Res. 1, 208–210. Mattisson, C., Bogren, M., Brådvik, L., Horstmann, V., 2015. Mortality of subjects with mood disorders in the Lundby Community cohort: a follow-up over 50 years. J. Affect. Disord. 178, 98–106. McGregor, C., Darke, S., Ali, R., Christie, P., 1998. Experience of non-fatal overdose among heroin users in Adelaide, Australia: circumstances and risk perceptions. Addiction 93, 701–711. McLellan, A.T., Luborsky, L., Woody, G.E., OBrien, C.P., 1980. An improved diagnostic evaluation instrument for substance abuse patients. The Addiction Severity Index. J. Nerv. Ment. Dis. 168, 26–33. McLellan, A.T., Kushner, H., Metzger, D., Peters, R., Smith, I., Grissom, G., Pettinati, H., Argeriou, M., 1992. The fifth edition of the addiction severity index. J. Subst. Abuse Treat. 9, 199–213.
181
McLellan, A.T., Cacciola, J.C., Alterman, A.I., Rikoon, S.H., Carise, D., 2006. The addiction severity index at 25: origins: contributions and transitions. Am. J. Addict. 15, 113–124. Merrall, E.L.C., Kariminia, A., Biswanger, I.A., Hobbs, M.S., Farrell, M., Marsden, J., Hutchinson, S.J., Bird, S.M., 2010. Meta-analysis of drug-related deaths soon after release from prison. Addiction 105, 1545–1554. Niméus, A., Alsén, M., Träskman-Bendz, L., 2002. High suicidal intent scores indicate future suicide. Arch. Suicide Res. 6, 211–219. Olsson, M.O., Öjehagen, A., Brådvik, L., Håkansson, A., 2015. Predictors of psychiatric hospitalization in ex-prisoners with substance use problems–a data-linkage study. J. Drug Issues, 1–12, http://dx.doi.org/10.1177/ 00222042615575374. Öberg, D., Zingmark, D., 1999. Sallmen, B. ASI-X. Version 1.1. Available at: http:// eib.emcdda.europa.eu/attachements.cfm/att 4121 EN asi%20en.pdf (accessed 09.03.15.). Öhberg, A., Lönnqvist, J., 1998. Suicides hidden among undetermined deaths. Acta Psychiatry Scand. 98, 214–218. Pratt, D., Piper, M., Appleby, L., Webb, R., Shaw, J., 2006. Suicide in recently released prisoners: a population-based cohort study. Lancet 368, 119–123. Runeson, B., Tidemalm, D., Dahlin, M., Lichtenstein, P., Långström, N., 2010. Method of attempted suicide as predictor of subsequent successful suicide: national long term cohort study. BMJ 340, c3222. Schepis, T.S., Hakes, J.K., 2013. Dose-related effects for the precipitation of psychopathology by opioid or tranquilizer/sedative non-medical prescription use: results from the National Epidemiological Survey on Alcohol and Related Conditions. J. Addict. Med. 7, 39–44. Schuckit, M.A., Smith, T.L., Kramer, J., Danko, G., Volpe, F.R., 2002. The prevalence and clinical course of sedative-hypnotic abuse and dependence in a large cohort. Am. J. Drug Alcohol Abuse 28, 73–90. Seymour, A., Oliver, J.S., Black, M., 2000. Drug-related deaths among recently released prisoners in the Strathclyde Region of Scotland. J. Forensic Sci. 45, 649–654. Skogman, K., Alsén, M., Öjehagen, A., 2004. Sex differences in risk factors for suicide after attempted suicide—a follow-up study of 1052 suicide attempters. Soc. Psychiatry Psychiatry Epidemiol. 39, 113–120. Tengvald, K., Andrén, A., Bergman, H., Engström, C., Nyström, S., Sallmén, B., Oberg, D., 2004. Implementing the Addiction Severity Index (ASI) in Swedish human services: experiences problems and prospects. J. Subst. Use 9, 163–171. Tidemalm, D., Långström, N., Lichtenstein, P., Runeson, B., 2008. Risk of suicide after suicide attempt according to coexisting psychiatric disorder: swedish cohort study with long term follow-up. BMJ 337, a2205. Vittinghoff, E., McCulloch, C.E., 2007. Relaxing the rule of ten events per variable in logistic and Cox regression. Am. J. Epidemiol. 165, 710–718. Wilcox, H.C., Conner, K.R., Caine, E.D., 2004. Association of alcohol and drug use disorders and completed suicide: an empirical review of cohort studies. Drug Alcohol Depend. 78 (Suppl), 511–519.