Background
Mortality rates among opioid addicts are higher than for the general population, although there are differences between countries and regions [
1]. Standardized mortality ratios (SMRs) vary between 15 and 28 in different studies [
2‐
4]. In spite of the increased mortality, and hence the need for intervention, Cook et al. found that 67% of patients who attended an emergency department because of acute opioid overdose were not referred to any therapeutic programme for drug addiction [
5]. During the last few decades, studies have focused on abstinence versus continued drug use following participation in various treatment programmes, as an outcome measure [
6], even though there is still some doubt about whether such programmes have a long-term effect on mortality. In a 15-year follow-up study, Sorensen et al. found that people who had achieved stable abstinence from injectable narcotics use were at lower risk of premature death when compared with those who continued using drugs [
7]. However, even in the presumed abstinence group, there was a sevenfold increase in SMR. This raises the question whether referral to a detoxification programme has a satisfactory effect on long-term mortality.
Although the majority of deaths among opioid addicts are from accidental poisonings [
8], deaths from other unnatural causes, such as suicide, have also increased [
9,
10]. The co-morbidity between drug dependence and suicide has been described both as an association between drug dependence and mental illness [
11], and between known risk factors for both drug dependence and suicide, such as gender, living alone and unemployment [
12]. In an empirical review, Wilcox et al. estimated the SMR for suicide associated with opioid dependence, but did not estimate SMRs for other diseases [
10]. In another empirical review, Harris and Barraclough calculated SMRs for both natural and unnatural causes of death associated with opioid dependence [
13].
Review studies, which have proven useful in pointing out the increased mortality in this patient group, have used expected number of deaths based on WHO data, rather than patient data. The analyses in Harris and Barraclough's review were based on 10 papers with a follow-up period ranging from 1 to 28 years, and the SMRs for natural causes of death were based on only one study. Longer prospective follow-up studies of opioid addicts show how the overall mortality rate changes over time [
14], it being expected that cause-specific mortality would also differ between long-term and short-term follow-up studies. Although cause-specific mortality rates were measured in one four-year prospective cohort study, no SMRs were obtained [
15]. Prospective cohort studies in the field are relatively uncommon [
15], and even rarer is the opportunity to compare the cohort with a well-defined background population that includes a whole city. However, long-term prospective studies of excess mortality among opioid addicts, in particular, would be useful for obtaining more specific information about opioid addictions as a subgroup of substance use disorders.
During recent decades, the AIDS epidemic has had a major effect on this group of patients [
16]. Although mortality among opioid addicts increased from 1980 to 1988 [
17], this increase can only be partly explained by the emergence of AIDS [
18]. Cause-specific SMRs during recent decades would be helpful in studying mortality in this group of patients.
The aims of this study were to study the mortality rate and causes of death among opioid addicts who had been treated for self-poisoning or admitted to voluntary detoxification in 1980 and 1981 during a 20-year follow-up investigation. The study compared this cohort with the general population. The study's design allowed an investigation of whether participation in a detoxification programme can be protective.
Discussion
This prospective study followed all hospital-treated opioid addicts from the same large city, up to as long as 20 years. All patients were traced during this period, thus minimizing selection bias. Causes of death were obtained for all patients, enabling cause-specific mortality ratios to be determined so that they could be compared with those of the general population.
This study's main finding was the high mortality rate of 37.8%. One-third of females and almost half of males died. This high mortality rate was observed in a young patient group, for which the median age during 1980 and 1981 was 24 years. When corrected for the expected number versus the observed number of deaths, SMRs showed no statistically significant gender differences. Although in absolute numbers males had a higher mortality ratio than females, the expected number of deaths among males was higher as well. Therefore, the SMRs for males and females were almost identical, there being a 23-fold increase in mortality. In this study, the effect of opioid addiction seemed to overrule the effects of age and gender on mortality.
The SMR of 23.6, found for this cohort, is similar to SMRs obtained in other long-term European studies [
1], although there were differences in the inclusion criteria for these studies. SMRs were 15 times higher for male drug users in Rome compared with the general population [
2], 22 times higher for drug injectors in Glasgow [
3], and 28.5 times higher for heroin addicts in Catalonia, Spain [
4].
The lack of gender differences has also been observed in other studies, such as those conducted with homeless people who had a drug addiction [
23]. However, other studies have found higher SMRs for either males [
8,
15] or females [
2,
13]. This diversity of findings may be due to different inclusion criteria. Gender differences have been observed when notified addicts or drug users, who were recruited from drug treatment centres, were included whereas the present study included hospital-treated opioid addicts. Therefore, this subject sample may have been more prone to accidental poisoning, such incidents being less correlated to gender perhaps than to behavioural traits, such as the pattern of drug use. Irrespective of toxic compound, the SMR for hospital-treated opioid addicts was much higher than that for hospital-treated self-poisonings. In a 20-year follow-up study of all self-poisonings treated in Oslo hospitals in 1980, the SMR was 4.6 (95% CI, 4.1–5.1), compared with 23.6 in the present cohort [
24]. Opioid addicts are therefore at special risk. There were a higher number of deaths in the first five-year period of the study, and a decrease in SMR from 32.4 in the first five-year period to 13.4 in the last five-year period. It is suggested that those who take the highest risks will probably die early, leading to a decrease in mortality for the cohort as a whole. It has been shown that the number of active drug addicts declines mainly from death, rather than from long-term abstinence [
25]. The decrease in SMR may also be due partly to increased mortality in the general population over time, leading to a relative reduction in the ratio.
There was an increased mortality for all causes of death among opioid addicts when compared with the general population, including both natural and unnatural causes of death. This was consistent with what is known about substance abusers in general [
13]. The causes of death were mainly drug-related, as has been observed in other studies [
3,
15]. Five suicides occurred in the cohort. The increased risk of suicide among opioid addicts accords with other studies [
9,
10]. There has been controversy about the possible existence of a substantial number of hidden suicides among accidental poisonings. So far, the results support the hypothesis that most deliberate poisonings are accidental [
26,
27]. In the present study, the causes of death were obtained from Statistics Norway, and our group reported on the validity of these data in 1985 [
28]. Since then, autopsy rates have been declining in Norway, and the reliability of death certificates has been questioned [
29]. Therefore, the suicide numbers may well be too low. Mortality in the study population was compared with mortality rates from the whole country. Although the expected lifespan is lower in Oslo than the average for the country as a whole, it is less than one year below the average. Although SMR values found in the current study may be somewhat high, this will not change the study's major findings.
Only five deaths were registered as specific natural causes of death, three being cancer and two being cardiovascular diseases. The SMRs increased significantly, possibly due to confounding factors such as lower socio-economic status or tobacco use. Nevertheless, opioid addicts represent a group that is at high risk for excess mortality, not only from drug-related deaths, but also because of their increased risk of death from cancer and cardiovascular diseases. The increased risk of death, even from natural causes, is consistent with the increased mortality associated with other mental disorders [
13].
Only one death could be classified as being caused by AIDS, although three deaths were stated as "disorders involving the immune mechanism". In other studies, AIDS has accounted for a majority of the deaths among opioid addicts [
4]. A study of HIV-positive opioid addicts in Oslo found that drug overdose was a major cause of death, thus overriding the effect of AIDS on mortality [
30].
The major cause of death was drug dependence, as registered in the other diseases category. In order to compare this cohort with the general population, we used the same categories as Statistics Norway. Although the SMR for this single cause of death was not calculated, this value would have provided minimal additional information. If drug dependence is a chronic disease, a symptom of which is opioid addiction, one would expect that only those suffering from the disease would die from it; that is, there would be a low number of deaths in the background population. Even when all other diseases were considered in the general population, death from drug dependence outnumbers the total number of deaths from all other diseases. As the classification of drug-related deaths is problematic in the ICD system, the term "drug-related deaths", used in mortality statistics, is currently being developed by the European Monitoring Centre for Drugs and Drug Addiction.
The category, drug dependence, is not equivalent literally to any category in ICD-10. Chapter X4 can be used for accidental poisonings, whereas F10 to F19 cover psychiatric and behavioural disorders caused by drug dependence. The incongruence of these classification systems makes it difficult to know if the categories cover the same spectrum of patients.
In this study, mortality did not change from referral to voluntarily detoxification. Our hypothesis was that those treated solely for self-poisonings would have a poorer prognosis, since such people were not sent to drug addiction units on a regular basis. Sorensen et al. found a significant decrease in mortality for those who achieved abstinence [
7], but in the present study, it was not known whether patients completed or discontinued the detoxification programme. Those who joined the programme voluntarily may have also been in a poorer physical condition, since this was one of the criteria for admittance. Some of the patients admitted for detoxification were transferred to other units specialising in the treatment of addicts, whereas others were admitted for a few days of detoxification in the medical department because of their extremely poor medical and physical state. Generally, these patients were not motivated to partake in treatment required to achieve abstinence, and were among those with the most serious drug addiction problems. In addition, we did not have information about the completion rate for those transferred for further treatment.
Unfortunately, in retrospect, we were unable to trace the status of each patient at discharge, but could only do so for the group as a whole. However, the high level of mortality in both subgroups supports the hypothesis that referral to a detoxification programme alone is not sufficiently effective to prevent the excess mortality for this group. Patients probably need a longer and more closely supervised follow-up, both to improve their physical health and to achieve abstinence.
Since the relatively small numbers of deaths in each category did not make it meaningful to obtain cause-specific SMRs for each time period, it was not possible to investigate how cause-specific mortality ratios change over time. Subgroups might have been used in the statistical analyses to evaluate the effect of repeated treatments on mortality. However, the resulting groups would be too small to reveal any statistically significant differences. It might be worthwhile investigating in larger studies whether there could still be a difference between patients seeking repeated treatments and other patients.
The fact that mortality in this group was quite high when compared with the general population is a great challenge to our society. It is worth emphasizing that not all drug addicts die early. More research is needed to discover what makes those patients who survive different from those who die. This information would be helpful both for choosing therapy and for identifying those at special risk.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
MAB helped obtain information on each patient's status and their cause of death, participated in the design of the study, and drafted the manuscript. ASB helped obtain information on each patient's status and their cause of death, and helped draft the manuscript. AO helped obtain information on each patient's status and their cause of death, and helped draft the manuscript. TH performed the statistical analyses. DJ obtained the cohort of hospitalized opioid addicts admitted for self-poisoning and participated in the design of the study. MR obtained the cohort of patients admitted for voluntarily detoxification. OE conceived of the study, participated in its design and coordination, and helped draft the manuscript. All authors read and approved the final version of the manuscript.