Comparison of mortality and causes of death between treatment-engaged individuals with severe alcohol or opioid use disorder: a prospective 19-year clinical cohort study

Abstract

Background

Comparisons of long-term mortality between alcohol-use-disorder (AUD) and opioid-use-disorder (OUD) groups within clinical cohorts are scarce. This study aimed to do this within one clinical cohort comprising groups suffering various severe substance use disorders.

Methods

Design: Observational cohort design. Prospective study of mortality 1998–2016, Cox regression. Sample/setting: 464 persons in specialist SUD treatment at start of observation period. Three subgroups defined by SUD characteristics at the beginning of the observation period, were studied: participants with 1) OUD, on opioid agonist treatment (OAT), (OUD/OAT-group), 2) AUD, no other SUDs (AUD-alone group), 3) AUD as part of a poly-substance-use disorder (AUD/PSUD group). Measures/covariates: Comparison of mortality risks and causes of death (somatic diseases/traumatic causes/substance-induced deaths/unknown cause) between OUD/OAT versus AUD-alone, and AUD-alone versus AUD/PSUD groups. Hazard ratio (HR) adjusted for baseline age (1 January 1998), sex, and socioeconomic/SUD characteristics.

Results

AUD-alone versus OUD/OAT: Unadjusted mortality risk was higher in the AUD-alone group [HR 1.99 (95% CI 1.33–2.99)]. However, in adjusted analysis type of SUD showed no statistically significant association with increased mortality [aHR 0.91 (0.54–1.54)], only male sex and increasing baseline age were independently associated with increased mortality risk. Cause-of-death distributions were very similar and somatic diseases caused over 50% of deaths in both groups. AUD-alone versus AUD/PSUD: Unadjusted mortality risk was higher in the AUD-alone group [(HR 2.57 (1.65–4.00)]. However, in adjusted analysis type of SUD showed no statistically significant association with increased mortality [aHR 1.40 (0.81–2.44)], only male sex, increasing baseline age, and “not being in work or education at study inclusion” were independently associated with increased mortality risk. Substance-induced deaths were most common in the AUD/PSUD group (43%).

Conclusion

In adjusted analyses there were no statistically significant differences in mortality risk between the three subgroups. Somatic diseases were the main cause of death. Besides efforts to improve social rehabilitation and to treat the substance use disorders as such, early and targeted lifestyle interventions, and measures for early detection, treatment and follow-up of somatic disorders are essential to reduce mortality in both AUD and OUD populations.

Clinical trial number

Not applicable

Background

It is well established that people with substance use disorders (SUDs) suffer increased overall mortality risk compared to the general population [1]. The increased risk comprises both substance-induced deaths (mostly fatal overdoses) [2], traumatic deaths (accidents, suicide and homicide) as well as deaths of a variety of acute and chronic somatic (physical) diseases [3, 4].

Alcohol use disorder (AUD) and opioid use disorder (OUD) are the two substance use disorders associated with the greatest overall global negative impact on public health and mortality [5]. Globally, in 2019 2.6 million deaths were attributable to alcohol use and alcohol use disorder. An estimated 400 million people were living with alcohol use disorder (including harmful use) and of them 209 million with alcohol dependence [5].

A 2013 meta-analysis showed a relative age-adjusted mortality risk (AUD versus non-AUD) for men of 3.38 (95% Confidence Interval (CI) 2.98–3.84) and for women 4.57 (95% CI 3.86–5.42). The relative risk among individuals below 40 years of age was 9 among men and 13 among women and lower at higher age and higher in clinical samples than in population studies [6]. People with AUD as part of a polysubstance use disorder (PSUD), tend to have higher mortality risk compared to people with AUD and no other SUDs [7–9]. Mortality among people with AUD is reduced when alcohol consumption is reduced, at the group level [10].

In 2022, about 60 million people used extramedical opioids [11]. In 2019, about 450,000 deaths were attributable to opioid use and opioid use disorder of which about 25% were fatal overdoses [11]. A 2011 meta-analysis on mortality among “dependent users of heroin and other opioids” showed a standardized mortality ratio of 14.66 (95% CI 12.82–16.50). Total crude mortality rate per 100 person-years was 2.09 (95% CI 1.93–2.26) while the crude mortality rate was higher and standardized mortality ratio lower among men than women [12]. In a 2020 meta-analysis on mortality among “people using extramedical opioids” the all-cause crude mortality rate per 100 person-years was 1.6 (95% CI 1.4–1.8) and all-cause standardized mortality ratio 10.0 (95% CI 7.6–13.2) [13]. The all-cause mortality rate and the rate of overdose deaths among persons with OUD as well as premature mortality due to a variety of comorbid disorders are significantly reduced when on versus off opioid agonist treatment (OAT) [14–18].

Although the methods in these meta-analyses are different, they generally indicate that the individual mortality risk is higher among people with OUD than with AUD. However, findings differ by study setting and sample characteristics (e.g., clinical versus population studies, different treatment modalities, different societies with varying substance use patterns). Higher mortality risk among people with OUD compared to AUD in the meta-analyses may be due to OUD being a more severe disorder with a higher individual mortality risk than AUD. However, this difference could also possibly be related to selection bias in that AUD studies to a greater extent include people with less severe consequences of their substance use disorder than OUD studies, which are largely based on clinical cohorts. In line with this, higher mortality rates and standardized mortality ratios than those from the meta-analyses have been reported among people suffering from severe AUD [19]. Furthermore, based on summarized knowledge, we would expect a higher proportion of drug induced deaths in the OUD/OAT and the AUD/PSUD groups, respectively, compared to the AUD-alone group [5].

However, mortality and causes of death have rarely been studied over a long period of time within the same clinical cohort consisting of persons with severe OUD and AUD respectively [20]. All participants in this study met the criteria for AUD or OUD, respectively, and were in inpatient or outpatient treatment for AUD or OUD within the specialist health services at the start of the observation period. We consider this as a criterion for them falling into the groups with severe AUD or OUD, respectively. Therefore, the rationale for this study was to study long-term mortality and causes of death in subgroups of cohort participants with severe OUD and AUD in the same clinical cohort, and to examine whether our findings were consistent with the findings in the cited meta-analyses.

The cohort in this study was established by merging two already existing cohorts that were originally established in the same catchment area and during the same period (Fig. 1 and Methods, below). The subgroups included in the present study, were 1) participants with OUD who was or had previously been on opioid agonist treatment when recruited, (OUD/OAT group), 2) participants with AUD exclusively when recruited, (AUD-alone group) and 3) participants with AUD as part of a poly-substance-use disorder when recruited (AUD/PSUD group).

Fig. 1.

Merging of the two original cohorts. Dual diagnosis (DD)-Innlandet cohort (recruited 1997) and opioid agonist treatment (OAT)-Innlandet cohort (recruited 2007–2008, started OAT for the first time 1998–2007). The DD-Innlandet cohort consisted of persons in specialist substance-use-disorder treatment, with and without concurrent mental disorders, when recruited. The cohorts were merged into the joint DD/OAT Innlandet cohort in 2015. Previously published findings (DD-Innlandet cohort, ref 21 and 24, and OAT-Innlandet cohort, ref 22 and 23) and findings not previously published (this article) ¹ persons with alcohol use disorder and no other substance use disorders, N = 130. ² participants with alcohol use disorder as part of a poly-substance-use disorder when recruited to the study, N = 117. ³ persons with opioid use disorder who was or had previously been on opioid agonist treatment when recruited, N = 200. Coinciding with the original OAT-Innlandet cohort

The aims of this study were 1) to compare mortality risks and causes of death between the OUD/OAT and the AUD-alone groups and 2) to compare mortality risks and causes of death between the AUD-alone and the AUD/PSUD groups, respectively. The research questions were: 1a) Is mortality risk higher in the OUD/OAT group than in the AUD-alone group?, 1b) Is the distribution of causes of death different in the OUD/OAT and AUD-alone groups?, 2a) Is mortality risk higher in the AUD/PSUD group than in the AUD-alone group?, and 2b) Is the distribution of causes of death different in the AUD/PSUD and AUD-alone groups?

Methods

Design

The study has an observational cohort design. Mortality and causes of death were studied prospectively as of 1 January 1998 through 31 December 2016.

Sample and setting

The sample is described in Fig. 1. Two original cohorts – the Dual Diagnosis (DD)-Innlandet and the OAT-Innlandet cohorts – were merged into the joint DD/OAT- Innlandet cohort in 2015. The AUD-alone and AUD/PSUD subgroups in this study were parts of the DD-Innlandet cohort while the OUD/OAT subgroup coincides with the original OAT-Innlandet cohort. Details concerning recruitment to the DD-Innlandet and OAT-Innlandet cohorts are previously described [21, 22].

DD-Innlandet cohort

The DD-Innlandet cohort was recruited in 1997 among persons receiving inpatient or outpatient specialist SUD treatment in the county Innlandet in Norway at that time. The DD-Innlandet cohort comprised in total 291 persons who consented to participate in the study out of 690 eligible, corresponding to a participation rate of 42%. The 291 participants did not differ significantly from the rest of the eligible concerning key socio-economic, and substance use variables except for age; the participants were on average three years older than the non-participants. Compared with a national sample (n = 5000) drawn from the nationwide SUD treatment system the DD-Innlandet cohort was skewed towards having fewer young dependent drug users than the national sample [21]. One-hundred-and-thirty persons (45%) were defined as AUD-alone, i.e., having AUD but no other SUDs (DSM IV), and these constituted the AUD-alone group in the present study. One-hundred-and-seventeen (40%) had AUD as part of a poly-substance-use disorder and constituted the AUD/PSUD group. Psychoactive substances included in this group were opioids, alcohol, cannabis, benzodiazepines, amphetamine and cocaine. The remaining 44 (15%) suffered other SUDs than AUD and were not included in the subgroups in this study (AUD-alone and AUD/PSUD).

OAT-Innlandet cohort

The OAT-Innlandet cohort was established in 2007–2008 and comprised 200 persons with opioid dependence (ICD-10, mainly heroin users) who started OAT with methadone or buprenorphine for the first time as of the start of the Norwegian national OAT programme in 1998 until 30 June 2007 in Innlandet county. One-hundred-and-eighty-seven of the 281 eligible consented to participate, and 13 persons who had died after OAT initiation but before the cohort was established were included. The cohort thus included 200 participants with a 72% participation rate.

Further, the OAT-Innlandet cohort included both participants with OUD exclusively and participants with other comorbid SUDs, including AUD. Of the 130 participants with interview information about alcohol history, 59 (45%) reported lifetime AUD at study recruitment in 2007–2008 with a median duration of five years. The sample was considered as representative for OAT patients in Norway at that time [22, 23]. Moreover, it is well established that OAT is a strong protective factor against premature death in OUD populations [14–18]. In the OUD/OAT group retention in OAT was very high. Of the total observation time 86.4% was on OAT, 9.0% off OAT and 4.6% with unknown OAT status. Of the time with known OAT status 90.5% was on OAT. Furthermore, in the DD-Innlandet cohort 89 out of 291 persons had an OUD lifetime diagnosis at study baseline [24]. Of these, 27 were also included in the OAT-Innlandet cohort. We did not know whether the remaining 62 received OAT during the observation period without being part of the OAT Innlandet cohort. Thus, the OUD/OAT group in this study was limited to persons who with certainty had been included in OAT.

The participants in the two original cohorts were recruited – in the same period and from the same catchment area – from specialist SUD treatment services within the national specialist health care system in Norway. As mentioned above we considered the fact that all participants were in specialist (second level) SUD treatment at the start of the observation period as a criterion for belonging to the groups with “severe AUD or OUD”, respectively. In 2015, the two cohorts were merged into the joint OAT/DD-Innlandet cohort. A follow-up study covering mortality and causes of death during the period from 1998 to 2016 in the joint cohort was performed in 2015–2017. Key cohort characteristics, causes of death and mortality rates in the original DD-Innlandet [21, 24] and OAT-Innlandet [22, 23] cohorts are previously published and summarized in Tables 1 and 3 and the Results section of this paper.

Table 1.

Sample characteristics, mortality estimates and causes of death, AUD-alone and OUD/OAT groups

	N AUD-alonea	N OUD/OATb	AUD-alone	OUD/OAT	P-value
Characteristics
Sex (male), n (%)	130	200	101 (77.7)	133 (66.5)	0.029*c
Age at study baseline,1 Jan 1998, years, mean (SD)	130	200	46.2 (9.9)	31.8 (7.5)	< 0.001*d
Died through 31 Dec 2016, n (%)	130	200	66 (50.8)	41 (20.5)	< 0.001*c
Age at death, mean (SD)	130	200	58.9 (10.6)	48.7 (6.8)	< 0.001*d
Age at onset AUD/OUD, respectively, mean (min–max)	130	126	28.9 (14–58)	18.1 (8–35)	0.001*e
Married or cohabitant at inclusion in study, n (%)	130	140	56 (43.1)	48 (34.3)	0.138c
Completed upper secondary school at inclusion in study, n (%)	128	132	76 (59.4)	42 (31.8)	< 0.001*c
Five years or more ordinary work experience at inclusion in study, n (%)	117	130	94 (80.3)	64 (49.2)	< 0.001*c
Having children, yes, n (%)	118	128	91 (77.1)	89 (69.5)	0.180c
Mortality
Crude mortality rate per 100 person years (95% Confidence Interval (CI))	130	200	3.5 (2.8–4.5)	1.6 (1.2–2.2)
Standardized mortality ratio (95% CI)	130	200	3.4 (2.6–4.2)	8.4 (6.1–11.2)
Causes of death
Somatic disease, n (%)			38 (58)	22 (54)
Traumatic causes, n (%)			7 (11)	6 (15)
Substance induced deaths, n (%)f			18 (27)	11 (27)
Unknown cause, n (%)			3 (5)	2 (5)
Total			66	41

Sample characteristics, crude mortality rates, standardized mortality ratios and causes of death. Comparison of participants with alcohol use disorder (AUD) and no other substance use disorders when recruited to the study (AUD-alone group, N=130) versus participants with opioid use disorder (OUD) who were or had previously been on opioid agonist treatment (OAT) when recruited to the study (OUD/OAT group, N=200). Significance level 5%. The findings for the groups separately are previously published: AUD-alone group (Hjemsæter, BMC Psychiatry, 2019, ref 24), OUD/OAT group (Skeie, European Addiction Research, 2022, ref 23)

*Statistically significant

^aNumber in AUD-alone group with data regarding various covariates

^bNumber in OUD/OAT group with data regarding various covariates

^cPearson Chi-square test

^dIndependent-Samples T-test

^eMann-Witney U-test

^fMore had overdose and fewer SUD as such as registered main cause of death in the OAT group (not in table)

Table 3.

Sample characteristics and causes of death, AUD-alone group and AUD/PSUD group

	N AUD-alonea	N AUD/PSUDa	AUD-alone	AUD-PSUD	p-value
Sex (male), n (%)	130	117	101 (77.7)	85 (72.6)	0.359c
Age at baseline,1 Jan 1998, years, mean (SD)b	130	117	46.2 (9.9)	32.9 (8.9)	< 0.001*d
Died through 31 Dec 2016, n (%)	130	117	66 (50.8)	28 (23.9)	< 0.001*c
Age at death, mean (SD)	66	28	58,9 (10.6)	48.4 (10.8)	< 0.001*d
Age at onset AUD, mean (SD)	130	116	28.9 (11.4)	17.6 (7.0)	< 0.001*d
Married or cohabitant at inclusion in study, n (%)	130	115	56 (43.1)	25 (21.7)	< 0.001*c
Completed upper secondary school at inclusion in study, n (%)	128	115	76 (59,4)	45 (39.1)	0.002*c
Not in work or education at study inclusion (%)	130	114	80 (61.5)	88 (72.2)	0.008* c
HSCL-25 score at inclusion in study (SD)	117	100	1.96 (0.60)	2.14 (0.58)	0.027* d
Causes of death
Somatic disease, n (%)			38 (58)	10 (36)
Traumatic causes, n (%)			7 (11)	3 (11)
Substance induced deaths, n (%)			18 (27)	12 (43)
Unknown cause, n (%)			3 (5)	3 (11)
Total, n (%)			66 (101)	28 (101)

Sample characteristics and causes of death, AUD-alone group and AUD/PSUD groups. Comparison of participants with alcohol use disorder and no other substance use disorders when recruited to the study (AUD-alone group, N = 130) versus participants with alcohol use disorder as part of a poly-substance-use disorder (PSUD) when recruited to the study (AUD/PSUD, N = 117). Pearson Chi-square test for categorial variables, independent-samples T-test for continuous variables. Significance level 5%. * Statistically significant

* Statistically significant

^a Number in analysis, AUD-alone and AUD/PSUD groups, respectively

^b Standard deviation

^c Chi-square test

^d Independent-Samples T-test

Observation period

Mortality was studied prospectively as of 1 January 1998 through 31 December 2016. For the AUD-alone and AUD/PSUD groups the start of the individual observation period was set to 1 January 1998. In the OUD/OAT group the individual start was set to the 15^th of the month they were first included in the OAT programme due to lack of information of the exact date of OAT-medication initiation. The total observation time in the OUD/OAT group comprised time on (86%) and off (9%) OAT, and with unknown OAT status (5%) [23]. The end of the individual observation period for all was 31 December 2016 or by earlier death. No participants were excluded due to lack of mortality information.

Data sources and collection

Sex, date of birth and information about baseline socioeconomic characteristics and SUD diagnoses were known from the original two cohorts based on interviews (DD-Innlandet and OAT-Innlandet cohorts) and record information (OAT-Innlandet cohort). Crude mortality rates and standardized mortality ratios were drawn from the original DD-Innlandet and OAT-Innlandet cohorts [23, 24]. Data on time and cause of death were drawn from the Norwegian Cause of Death Registry [25]. Background population mortality data from the reference year 2008 (mid-year of the observation period) were drawn from Statistics Norway [26].

Measures and covariates

Comparisons of mortality risks between the AUD-alone and OUD/OAT groups and between the AUD alone and the AUD/PSUD groups were estimated as adjusted hazard ratios (aHR).

Causes of death were categorized in four groups: somatic (physical) diseases, traumatic causes (homicide, suicide, accidents), substance induced deaths (mostly fatal overdoses, a few cases where OUD or AUD, respectively, was registered as cause of death), and unknown cause of death.

All variables except for time of death and cause of death were variables from inclusion in the study. Data about time on and off OAT during the whole observation period were available for the OUD/OAT group but corresponding treatment-history data from the AUD-alone and AUD/PSUD groups were not available. Sex, age at study baseline 1 January 1998, age at onset OUD or AUD, respectively, being married/cohabitant, and completed upper secondary school at study inclusion were used as covariates in both the OUD/OAT versus AUD-alone and the AUD-alone versus AUD/PSUD Cox regression models. Further, five years or more of ordinary work experience and having children at study inclusion were used as covariates in the AUD-alone versus OUD/OAT model and not in work or education at study inclusion and anxiety/depression symptom score [27] were used in the AUD-alone vs. AUD/PSUD model.

The merging of the two cohorts posed some methodological problems. Twenty-seven participants belonged to both the original DD-Innlandet and the OAT-Innlandet cohorts and all of them were placed in the OUD/OAT subgroup in the joint cohort. However, several of these had a history of AUD as well at the start of the observation period. Further, some participants in the OAT-Innlandet cohort who were not participants in the DD-Innlandet cohort, would also have met the inclusion criteria for the AUD/PSUD subgroup in the DD-Innlandet cohort and consequently also in the joint cohort. The distinction between the OUD/OAT and the AUD/PSUD subgroups in the joint cohort was therefore somewhat imprecise. This would have been a problem if the AUD-alone subgroup was to be compared with both the OUD/OAT and the AUD/PSUD subgroups. Therefore, the group comparisons were carried out in two steps. The comparison of mortality (Cox regression models) and causes of death between the OUD/OAT and the AUD-alone groups was performed in the joint DD/OAT-Innlandet cohort as these groups were mutually exclusive in this cohort. The comparison of mortality (Cox regression models) and causes of death between the AUD-alone and the AUD/PSUD groups was done in the original DD-Innlandet cohort as these groups were mutually exclusive in that cohort.

Statistics

Bivariate comparisons of group characteristics were done by descriptive statistics, χ²-test for categorical variables and independent-samples-t-test (normally distributed) or Mann Witney U-test (non-normally distributed) for continuous variables (Tables 1 and 3). Comparison of mortality risks as HR with 95% CI were estimated in Cox regression models with time in study as time axis (Tables 2 and 4). Only participants with data on all covariates in each model were included in the analyses and this is specified in Tables 2 and 4. The significance level was set to 5%. Analyses were performed in SPSS v. 26.

Table 2.

Comparison of all-cause mortality risk between the AUD-alone group and the OUD/OAT group

			Model 11		Model 22
	Unadjusted HR	p-value	Adjusted HR	p-value	Adjusted HR	p-value
	(CI 95%)		CI 95%)		(CI 95%)
Type of SUD: AUD alone (OUD/OAT reference)	1.99 (1.33–2.99)	< 0.001*	1.12 (0.59–2.11)	0.733	0.91 (0.54–1.54)	0.773
Sex (male)	2.38 (1.42–4.00)	0.001*	2.24 (1.19–4.23)	0.013*	2.08 (1.24–3.50)	0.006*
Age at baseline, 1 Jan 1998, per year	1.06 (1.04–1.08)	< 0.001*	1.06 (1.02–1.09)	< 0.001*	1.05(1.03–1.08)	< 0.001*
Age at onset AUD or OUD, respectively, per year	1.03 (1.01–1.05)	0.003*	1.00 (0.97–1.03)	0.932
Married or cohabitant at inclusion in study, yes	0.90 (0.60–1.37)	0.638
Not completed upper secondary school at inclusion in study, yes	1.18 (0.78–1.80)	0.428
Five years or more ordinary work experience at inclusion in study, yes	1.66 (0.99–2.78)	0.056	0.87 (0.49–1.54)	0.623
Having children, yes	1.02 (0.61–1.69)	0.950

Comparison of all-cause mortality risk between participants with alcohol use disorder and no other substance use disorders when recruited to the study (AUD-alone group, N=130), and participants with opioid use disorder who was or had previously been on opioid agonist treatment when recruited to the study (OUD/OAT group, N=200). Cox regression models, unadjusted and adjusted hazard ratio (HR), confidence interval (CI) 95%

¹Model 1 including SUD type, sex, age at baseline, age at AUD/OUD onset, five years or more ordinary work experience at inclusion in study. Number of observations: 241

²Model 2 including SUD type, sex and age at baseline. Number of observations: 329

* Statistically significant

Table 4.

Comparison of all-cause mortality risk between the AUD-alone and AUD/PSUD groups

			Model 11		Model 22
Variable	Unadjusted HR	p-value	Adjusted HR	p-value	Adjusted HR	p-value
	(CI 95%)		(CI 95%)		(CI 95%)
SUD type: AUD alone (reference AUD/PSUD)	2.57 (1.65–4.00)	< 0.001*	1.33 (0.74–2.40)	0.345	1.40 (0.81–2.44)	0.229
Sex (male)	1.80 (1.05–3.08)	0.033*	1.95 (1.04–3.64)	0.038*	1.86 (1.08–3.20)	0.024*
Age at baseline 1 Jan 1998, per year	1.06 (1.04–1.08)	< 0.001*	1.06 (1.03–1.09)	< 0.001*	1.06 (1.03–1.08)	< 0.001*
Age at onset AUD, per year	1.04 (1.02–1.05)	< 0.001*	1.00 (0.97–1.03)	0.842
Not married/cohabitant at inclusion in study	0.86 (0.56–1.31)	0.487
Completed upper secondary school at inclusion in study	1.19 (0.79–1.79)	0.398
Not in work or education at study inclusion	1.61 (1.00–2.60)	0.051	1.95 (1.16–3.28)	0.012*	1.93 (1.18–3.15)	0.009*
HSCL-25 score at baseline	0.65 (0.44–0.96)	0.029*	0.73 (0.49–1.08)	0.111

Comparison of all-cause mortality risk between participants with alcohol use disorder and no other substance use disorders when recruited to the study (AUD-alone group, N=130) and participants with alcohol use disorder as part of a poly-substance-use disorder when recruited to the study, (AUD/PSUD, N=117) groups. Cox regression models, unadjusted and adjusted hazard ratio (HR), confidence interval (CI) 95%.

¹Model 1 including SUD type, sex, age at baseline, age at onset AUD, not in work or education at study inclusion, HSCL-25 score at baseline. Number of observations: 212

² Model 2 including SUD type, sex, age at baseline, not in work or education at study inclusion. Number of observations: 242

* Statistically significant

Results

Comparison of the AUD-alone and OUD/OAT groups

Group characteristics

Table 1 compares previously reported group characteristics [21, 22]. Compared with the OUD/OAT group, the proportion of men (78% versus 67%) as well as baseline mean age (46 years versus 32 years), mean age at onset of primary SUD (AUD and OUD respectively) (29 years versus 18 years), the proportion that was dead at the end of the observation period (50% versus 21%) and mean age at death (59 years versus 49 years) were significantly higher in the AUD-alone group. The proportions that had completed upper secondary school (59% versus 32%) and had at least five years of work experience at study inclusion (80% versus 49%), were also significantly higher in the AUD-alone group. Crude mortality rate was higher and standardized mortality ratio lower in the AUD-alone group.

Mortality risk (research question 1a)

The AUD-alone group had a higher risk of mortality compared with the OUD/OAT group (unadjusted HR 1.99 (95% CI 1.33–2.99) (Table 2). However, when adjusted for sex, baseline age and key socioeconomic and SUD characteristics, type of SUD was not statistically significant (adjusted HR 0.91 (95% CI 0.54–1.54, OUD/OAT reference). In the adjusted model, only male sex and increasing baseline age were independently associated with increased mortality risk.

Causes of death (research question 1b)

The distribution of causes of death (somatic disease, traumatic causes, substance induced causes and unknown cause of death) was very similar in the AUD-alone and OUD/OAT groups and somatic diseases counted for more than half of the deaths in both groups (Table 1).

Comparison of the AUD-alone group versus the AUD/PSUD group

Group characteristics

Compared with the AUD/PSUD group, the AUD-alone group was on average older at baseline (46 versus 33 years) and at onset of AUD (29 versus 18 years). Participants in the AUD-alone group were also more often married or cohabiting, had on average higher levels of education, were more often in work or education, and scored lower on anxiety/depression symptoms at study entry (Table 3).

Mortality risk (research question 2a)

The AUD-alone group had a higher risk of mortality compared with the AUD/PSUD group [unadjusted HR 2.57 (95% CI 1.65–4.00)]. However, when adjusted for baseline age, sex and socioeconomic and SUD characteristics, type of SUD was not statistically significant [aHR 1.40 (95% CI 0.81–2.44)] (Table 4). Baseline age accounted for most of the downward HR adjustment for the AUD-alone versus AUD/PSUD comparison (not in table). Only male sex, increasing baseline age, and not being in work or education at study inclusion were significantly associated with increased mortality risk in the adjusted model.

Causes of death (research question 2b)

The proportion of substance-induced deaths was higher (43% versus 27%) and deaths due to somatic diseases lower (36% versus 56%) in the AUD/PSUD group compared to the AUD-alone group.

Discussion

Main findings

The crude mortality risk was twice as high in the AUD-alone group as in the OUD/OAT group, mainly due to age differences. In adjusted analyses, only baseline age and male sex, and not type of SUD, were independently associated with increased mortality. When causes of death were divided into four categories (somatic diseases, traumatic causes, substance induced deaths and unknown cause), the cause-of-death patterns in the two groups were very much alike and somatic diseases caused more than half of the deaths in both groups.

Further, comparing the AUD-alone and AUD/PSUD groups, type of SUD was not associated with mortality risk in adjusted analyses. Only male sex, baseline age, and not being in employment or education at study baseline were independently associated with increased mortality risk in adjusted analyses.

Long-term mortality risk and cause-of-death patterns in the AUD-alone and OUD/OAT groups

Both the mortality risks (adjusted HR) and the cause-of-death patterns in the AUD-only and OUD/OAT groups were very similar. The similarity in adjusted mortality risks was a striking finding that contrast with results from previous systematic reviews where people with OUD typically are shown to have higher mortality risk than people with AUD [12–18], but the similarity in mortality risks in our study is more in line with findings in a recent Australian study [20]. Several factors may have contributed to this.

First, the alcohol group in this study consisted of persons with severe AUD receiving inpatient or outpatient specialist AUD treatment when recruited to the study and may deviate from the average of AUD samples in the meta-analyses.

Second, the treatment for people with OUD and AUD differs. Opioid agonist treatment is the most widely used form of treatment for opioid dependence. This is a long-term, mainly outpatient, treatment with well-established significant mortality-reducing effect, mainly during treatment, and it is most often combined with psycho-social rehabilitation programs. Thus, OAT combines harm-reducing and rehabilitation approaches [28–30]. In the case of AUD, long-term medicinal treatment options to reduce craving and alcohol intake is less established and far less used, and relatively short-term psycho-social treatment is the main intervention [10]. Although we lack detailed data on the course of treatment and to what extent the patients received pharmaceutical and psychosocial treatment during the observation period, especially concerning the AUD groups, we assume that they had access to “standard treatment for OUD and AUD available in Norway 1998–2016”. We might expect OUD to be potentially more harmful and with higher mortality risk, while untreated, but OAT may possibly be more long-term risk reducing than standard AUD treatment for severe AUD.

Third, OAT reduces overdose mortality [14, 15] and the retention in OAT was very high in the OUD/OAT group. In our study, somatic diseases accounted for most of the deaths in both groups. This emphasizes the impact of organ pathologies associated with severe alcohol and opioid use, both diseases directly caused by substance-use-behaviours (e.g. extramedical injecting and chronical hepatitis C) and diseases partly attributable to pathological organ stress due to extensive long-lasting substance intake (e.g. alcohol-induced liver cirrhosis). Unhealthy lifestyle factors such as smoking, inadequate nutrition and insufficient physical activity also contribute to increased somatic morbidity and mortality.

Finally, people with severe SUDs are often socially deprived and marginalized and suffer poverty, housing problems and unhealthy lifestyle habits as mentioned above. Low socio-economic status (SES) is generally associated with multimorbidity and increased mortality [11]. In this study, SES was higher in the alcohol-alone group than in the OUD/OAT group at inclusion in the study. However, higher SES at a younger age may be offset by severe negative physical and mental health effects and social consequences related to extensive alcohol use later in life and by both groups sharing lifestyle risk factors. Also, many people with severe AUD may experience a rapid and dramatic drop in SES, despite education, professional experience and social networks, when the alcohol disorder becomes sufficiently severe. This may help to explain why higher baseline SES is not associated with reduced mortality risk in this study.

The similar adjusted mortality risk for the AUD-alone and OUD/OAT groups suggests that severe AUD and OUD may have comparable long-term adverse health effects.

Alcohol use disorder and polysubstance use

Concurrent problem use and dependence of multiple psychoactive substances – polysubstance use and dependence – generally worsens the negative health consequences compared to single substance use. Polysubstance use, including alcohol, increases mortality risk among drug users [7–9]. In contrast, in adjusted regression analysis, we did not find a statistically significant higher mortality risk in the AUD/PSUD group compared to the AUD-alone group. The AUD/PSUD group was on average 13 years younger at baseline and on average 11 years younger at AUD debut, scored lower on being married or cohabitant, having completed upper secondary school and being in work or education at inclusion in the study. A possible explanation for our finding may be that the AUD-alone group consisted of persons with a more severe AUD, though occuring later in life, and with no or little extramedical drug use while people in the AUD/PSUD group had less severe AUD but more social problems and use of illegal substances at a younger age [24]. Thus, this may reflect and underscore the major adverse health effects and the increased mortality risk – especially due to somatic diseases – related to severe AUD [3]. The AUD-alone group in our study consisted of a sample of people with severe AUD, and in this group mortality risk may be on the same level as among people with AUD as part of a polysubstance use disorder.

Strengths and limitations

Strengths

Although the AUD and OUD groups originally originated from two different cohorts, it was possible to merge these into a joint cohort. The main strengths of this study are the prospective study of mortality, the long follow-up time, no loss to follow-up, and a rich data set from the study baseline and from the Norwegian Cause of Death Registry. These made it possible to compare mortality between the AUD-alone and OUD/OAT and the AUD-alone and AUD/PSUD groups, respectively.

Limitations

Generalizability

First, the aim of the study was to compare mortality between subgroups with severe OUD and AUD respectively. Thus, the findings do not shed light on mortality among people with alcohol and opioid use disorders in general. Second, the OUD/OAT group was limited to opioid dependent persons who was or had been included in OAT. It is well established that mortality is lower among opioid dependent people on OAT compared to their peers after leaving OAT and those never on OAT [14, 17, 18]. Further, the retention in treatment is very high in this cohort as at least 87% of the total observation time was on OAT [23]. As OAT is a strong protective factor against premature and especially overdose deaths [14–18], the high OAT retention clearly contributed to reduced mortality in the OAT group. This will reduce the transferability of our findings to OAT samples with less retention in treatment, and especially to groups never on OAT. However, as retention in OAT [31] and OAT coverage [32] in Norway is high, this cohort is representative of the opioid dependent population in Norway and probably in other countries with high OAT coverage and retention in OAT.

Covariates

All covariates in the Cox regression models refer to the status at the inclusion in the original DD-Innlandet and OAT-Innlandet cohorts. The lack of data on treatment course, changes in substance use and severity of the substance use disorders and in socioeconomic status over time, as well as data on somatic and mental health during the observation period, may be a limitation to the comparison of mortality across groups, as these factors may influence mortality. On the other hand, it is likely that all three groups (OUD/OAT, AUD-alone and OUD/PSUD) had access to “standard available treatment” for their substance use disorders during the observation period. Seen in this perspective, the study reflects mortality through the “natural course” of severe OUD and AUD in this cohort.

The establishing of the original DD-Innlandet and OAT-Innlandet cohorts

As the DD/OAT cohort was formed by merging the two original cohorts, we cannot rule out that differences in the establishing of these, as participation rates, may have influenced the findings. However, as the recruitment took place in the same geographical area among groups that entered specialist SUD treatment in the same period, we consider the merging of the two cohorts as methodologically acceptable.

Sample size and observation period

Long-term comparison of mortality between groups of patients with severe, though different, SUDs within the same clinical cohort is rare. Although the mortality data are somewhat old ending in 2016, the long observation period makes the findings interesting. Thus, the results should be of general relevance despite the limited sample size.

Conclusion

In this long-term clinical cohort study mortality was high in all three subgroups. In adjusted analyses there were no statistically significant differences in mortality risk between the OUD/OAT and AUD-alone groups and between the AUD-alone and AUD/PSUD groups. Increasing age and male sex was independently associated with increased mortality risk in both the OUD/OAT versus AUD-alone and the AUD-alone versus AUD/PSUD adjusted analyses, and “not being in work or education at study inclusion” in the AUD-alone versus AUD/PSUD comparison. Somatic diseases counted for more than half of the deaths in the OUD/OAT (54%) and AUD-alone group (58%), while substance-induced deaths were most common (43%) in the AUD/PSUD group. Besides efforts to improve social rehabilitation and to treat the substance use disorders as such, early and targeted lifestyle interventions, and measures for early detection, treatment and follow-up of somatic disorders are essential to reduce the high mortality in both the AUD and OUD populations.

Abbreviations

aHR

adjusted hazard ratio

AUD

alcohol use disorder

CI

confidence interval

DD

dual diagnosis

DSM IV

Diagnostic and Statistical Manual of Mental Disorders, 4th Edition (American Psychiatric Association)

HR

hazard ratio

ICD 10

International Statistical Classification of Diseases and Related Health Problems 10th Revision (World Health Organization)

OAT

opioid agonist treatment

OUD

opioid use disorder

PSUD

poly substance use disorder

SES

socio-economic status

SUD

substance use disorder

Authors contributions

Study conception and design: I.S., T.C., A.J.H., B.M., A.L., M.T., H.W., A.B.B. Acquisition of data: I.S., B.M., A.J.H. Analysis and interpretation of data: I.S., M.T. Drafting of the manuscript: I.S. Critical revision: All authors.

Funding

Innlandet Hospital Trust (150306 and 150924).

Data availability

The dataset is not publicly available to protect the privacy of participants. Queries about the data should be directed to the corresponding author.

Declarations

Ethics approval and consent to participate

The study was approved by the Regional Committee for Medical and Health Research Ethics, Health Region South-East (ID 2014/1936 C). All participants alive gave their written consent before taking part in the original study and could decline participation in the follow-up study. Exemption from professional secrecy duty of confidentiality for those dead was given by the Norwegian Directorate of Health (original studies, see Fig. 1) and the Ethics Committee (above, follow-up study).

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.