Cause-Specific Mortality among Patients in Treatment for Opioid Use Disorder in Multiple Settings: A Prospective Comparative Cohort Study

Opioid use disorder (OUD) is considered a chronic, relapsing disease that often requires long-term treatment. Opioid maintenance treatment (OMT) has been shown to be effective in reducing all-cause and overdose mortality [1], crime [2, 3], and HIV and hepatitis C (HCV) transmission [4, 5]. Methadone, buprenorphine, or buprenorphine/naloxone combination are the most common medications used as part of OMT and are considered essential medicines by the World Health Organization [6]. However, despite the benefits of OMT, when compared with the general population, people with OUD face a 15 times increased risk of premature death [7]. Further, overdose mortality is over three times higher during periods outside of OMT treatment when compared to in-treatment periods [7].

It has often been observed that among people with OUD, the leading cause of death is overdose [7, 8]. In a large analysis of 19 cohorts, it was found that poisoning and substance-related deaths accounted for one-third of the deaths for people who use extra-medical opioids [8]. However, overdose is not the only risk for premature death, and several studies have found high rates of deaths related to HCV and AIDS [9] and suicide including those from other causes than poisoning [10].

The 2019 Global Burden of Disease Report describes that different age groups have distinct leading causes of death [11]. In the report, ischemic heart disease and stroke were reported as leading causes of death, particularly for those aged 50 and older. Among younger populations (under age 24), these conditions were not present in the top 20 causes of death, and the leading causes of deaths were self-harm and substance use disorders [11].

For many, OMT is a life-long and effective treatment, resulting in an ageing OMT population [12, 13]. Ageing OMT populations face high rates of medical problems [14], including physical conditions and other comorbidities [15, 16]. Additionally, some studies have found the risk of deaths related to drug use to increase with age, with studies from the United Kingdom reporting a quadrupled risk for those 45 and older compared to 25-35 year olds [17, 18]. The increased risk of death with older age may be attributed to several factors, including the high prevalence of comorbid conditions. Multi-morbidity, particularly comorbid psychiatric, circulatory, respiratory, or hepatic-pancreatic morbidity, is common among OMT patients. When compared to gender- and sex-matched general population counterparts, OMT patients have been found to be seven times more likely to have multi-morbidity [19].

Given the increased risk of premature death among people in OMT, investigations into causes of death are warranted to better tailor medical management. However, there is little evidence available on non-overdose, cause-specific mortality among OMT patients, particularly on how these causes vary across gender and age groups in different countries. Using large multinational cohorts of nationwide mortality registry data from patients who had received OMT in Czechia, Denmark, and Norway, this study aimed to (1) describe all non-overdose causes of death in the three national cohorts, (2) compare non-overdose mortality rates among different age and gender groups, and (3) compare age-standardized mortality rates (ASMRs) between different OMT and general populations.

This study was a prospective comparative cohort study that utilized a cohort design linking national registry databases through unique personal identification numbers in Czechia, Denmark, and Norway.

The study population comprised of people who received OMT at least once from Czechia, Denmark, and Norway during the study period. In total, 29,486 OMT patients were included in the study: 5,217 in Czechia from the period from January 1, 2000, to December 31, 2019; 12,880 in Denmark from the period from January 1, 2000, to December 31, 2018; and 11,389 in Norway from the period from January 1, 2010, to December 31, 2019. Given that patients came in and out of treatment during this period, the study population includes both new and existing patients.

Methadone is available in 12 specialized OMT centres in the country. The medication cost is fully covered, and the treatment is generally accompanied by follow-up psychosocial services. Medications that contain buprenorphine can be prescribed by physicians, which results in better accessibility of the medication; however, patients must cover the cost. OMT patients can be referred by both OMT clinics and physicians to specialized health services for additional physical or mental health treatment.

Danish OMT has a focus on high treatment coverage for people with OUD [20]. Denmark is considered to have a liberal OMT program in regards to concurrent use of alcohol and illicit drugs and has a high prescription rate of methadone. Access to additional public outpatient health care services, such as mental health services take place through referral from a physician at the treatment site, OMT clinic, or from general practice. Treatment is publicly funded and free for all patients [21].

Norwegian OMT can be considered “low-threshold,” with very few patients being discharged from treatment, regardless of continued illegal drug use while in treatment [22]. Patients have the right to mental health services [23], which are often integrated into OMT clinics. Patients can also be referred by the OMT clinics to additional, more specialized mental health services if necessary. OMT is funded by the government and free for all patients in Norway.

In all three countries, nationwide health registers were linked using personal identification numbers that are assigned at birth, or upon moving to the country. In each country, physicians are legally required to report data to these registries. The protocol for the overall comparative registry linkage study can be found elsewhere [24]. All three cohorts have been analysed separately.

For Czechia, data on OMT patients were obtained from the National Register of Addiction Treatment. This registry contains OMT patient information, including demographic data, treatment dates, and type of OMT medication used. For mortality data, data were obtained from the Information System on Deaths. This registry includes the cause of death diagnosis (ICD-10) and date of death.

For Denmark, information on patients who receive OMT, date of treatment, and type of medication was obtained through the Danish Registry of Drug Abusers Undergoing Treatment. Mortality data for Denmark were obtained using the Cause of Death Registry, which includes cause of death diagnosis (ICD-10) and date of death [25].

For Norway, the Norwegian Prescription Database (NorPD) provided data on OMT medications dispensed and dispensation date [26]. Since Norway does not have a specific OMT registry, NorPD provided a proxy indication by identifying OMT patients based on prescriptions. This database identifies approximately 90% of OMT patients in Norway [27]. Opioids used for identification of OMT patients included methadone oral solution (Anatomical Therapeutic Chemical [ATC] code N07BC02) and high-dose buprenorphine tablets (≥2 mg sublingual tablets, N07BC01 [buprenorphine] or N07BC51 [buprenorphine-naloxone]), all almost solely prescribed for the treatment of OUD. Mortality data for Norway were obtained using the Cause of Death Registry, which includes the cause of death diagnosis (ICD-10) and date of death.

Causes of death were categorized into the following categories: neoplasms (C00-D49), cardiovascular (I00-I99), respiratory (J00-J99), intentional self-harm (X60-X84) (excluding overdoses death with intentional poisoning), accidents (V01-X59), infectious (A00-B99), and digestive (K00-K95). All other causes of death were categorized as “other.”

In order to present a complete picture, overdoses were also included. Overdose coding can include cases of intentional, accidental, and undetermined intent. This study used the European Monitoring Centre for Drugs and Drug Addiction (EMCDDA) selection B+ criteria of drug-induced deaths to define overdoses. These criteria are based upon ICD-10 codes and are operated by selecting deaths where the underlying causes are either (1) mental and behavioural disorders caused by illicit drug use (diagnoses F11-F12, F14-F16, F19) or (2) accidental poisoning (X41, X42), intentional poisoning (X61, X62, X64), or poisoning of undetermined intent (Y11, Y12) by: opium (T40.0), heroin (T40.1), other opioids (T40.2), methadone (T40.3), other synthetic opioids (T40.4), cocaine (T40.5), other and unspecified narcotics (T40.6), cannabis (T40.7), lysergide (T40.8), other and unspecified psychodysleptics (T40.9), or psychostimulants (T43.6). Using the EMCDDA criteria, overdoses could include intentional poisoning (X61, X62, X64), if accompanied by the substance code (T codes).

Descriptive statistics were performed for the three study populations. Crude mortality rates (CMRs) were calculated as deaths per 1,000 person years (PYs) for cause-specific mortality and were displayed by gender and age groups (under 40 years old, 40–50 years old, and over 50 years old). Age groups were chosen due to the different age distributions from the three cohorts. PYs were calculated from the start of OMT until the person died, or until the end of the study period. Age was determined during the follow-up period, meaning that each person contributed to different age groups as they aged during the study period. In Czechia and Denmark, the start date of treatment was available. For Norway, the start date was the first date an OMT medication prescription was filled.

ASMRs per 1,000 were calculated to compare OMT patients in different countries. The direct standardization method using data from each country was used to standardize for age. Rates were standardized to the European standard population 2013 [28]. For the general population, data were retrieved from a published report [29]. The report is from 2019 and included fully calculated data for those under age 65. We therefore used these data to compare the general population and OMT populations from the three countries. In addition, we calculated ratios between ASMR for the different causes of mortality from different countries and the ASMRs for the general population. For Denmark, for any causes with n < 6, data were imputed due to the rules from Statistics Denmark prohibiting the export of any data cells lower than six individual observations.

In all, this study included 29,486 people who received OMT in the three countries. Characteristics of the cohorts are reported in Table 1. The mean age at study start was highest in Norway, at 39.8 for males, and the lowest in Czechia at 29.3 for females. The cohorts from Norway and Czechia comprised of one-third females, whereas the rate in the Danish cohort was slightly lower; 24.5% (n = 3,158). In all three countries, women on average were younger at the start of treatment.

Among the three cohorts for the first OMT medication prescribed, Denmark had the highest rate of methadone prescribed (n = 9,951, 77%), Czechia had the highest rates of buprenorphine (n = 2,020, 39%) and Norway had the highest rates of buprenorphine/naloxone (n = 3,554, 31%). Czechia had similar prescribing patterns to Norway. The age at death was similar among the Danish and Norwegian cohorts, with a mean age of approximately 48 for both males and females. The mean age of death was youngest in Czechia, with a mean age of 36 for women and 38 for men.

Among the three cohorts, a total of 5,322 deaths were recorded in the study period (18% of the cohort). Tables 2-4 and Figure 1 show the frequency of causes of death and CMR for each cohort. Figure 1 shows a comparison between the countries.

While there was variation in the top causes of death between the three cohorts, overdoses were in the top three leading causes for each cohort (Czechia n = 47, 11%; Denmark n = 951, 28%; Norway n = 617, 41%). Accidents were among the top three causes for Czechia (n = 78, 19%) and Denmark (n = 380, 11%). After accidents, cardiovascular causes were the leading cause of death in Czechia (n = 74, 18%). Neoplasms were a leading cause in Denmark (n = 331, 10%) and Norway (n = 175, 12%). For all three cohorts for suicide, the main contributing diagnosis was intentional self-harm by hanging, strangulation, and suffocation (ICD-10: X70). In all three cohorts, infectious causes were among the least reported causes of death. “Other” causes of death comprised between 14 and 20% for the three cohorts.

In total, physical causes of death (neoplasms, cardiovascular, respiratory, digestive, and infectious) made up slightly more than one-third of the deaths for each of the cohorts (Czech Republic: n = 165, 39%; Denmark: n = 1,310, 39%; Norway: n = 520, 34%). While physical causes made up similar proportions among the cohorts, there was variation between the cohorts in terms of the specific types of physical deaths. Cardiovascular causes of death were particularly high in Czechia compared to Norway and Denmark (Tables 2-4). Also of note, deaths due to neoplasms were the third leading cause of death in Denmark and the second leading cause in Norway; however, in Czechia neoplasms were among the least likely causes of death.

Causes of death for the different genders are displayed in Figure 2a, b. The CMR per 1,000 PY was highest in Denmark for both men and women at 26.8 and 24.6, respectively (Table 3), and lowest in Czechia at 9.1 for men and 5.1 for women (Table 2). In all three cohorts, the CMR for overdose was higher in men than women. Suicide rates were similar among men and women in Denmark and Norway. However, the CMR for suicide was over three times higher for men in Czechia (CMR 1.0) than for women (CMR 0.3). Cardiovascular deaths were the leading cause of death for women in Czechia, with a CMR higher than overdose and suicide combined. After accidents, the CMR for cardiovascular causes of death was the highest for men in Czechia (CMR 1.6). Conversely, cardiovascular causes were not in the top three leading causes of death for men or women in Denmark and Norway. Of other notable gender differences, the CMR for accidents for Norwegian women was approximately half compared to Norwegian men.

The ASMR differed between the cohorts and for the genders (Tables 5-7). Notably, cardiovascular causes in Czechia were higher than in the other cohorts, particularly for women in OMT (ASMR 3.59). Additionally, in Denmark, men in OMT died of an overdose at a rate (9.92) nearly twice that of women in OMT (5.01). However, in all three cohorts, the OMT population (for both genders) had an ASMR that exceeded the general population.

Causes of death among the age groups for each country are presented in Figure 3a–c. Regarding the CMR overall, each of the cohorts presented differently across the age groups. However, despite these differences, when sudden, unexpected causes of death (accidents, suicide, overdose) and physical causes (neoplasms, cardiovascular, respiratory, and digestive) are grouped, we see similar trends between the groups regarding age. For all three groups, we see the highest rates of sudden, unexpected deaths in the youngest group that decreases in the older age groups. Similarly, physical causes were lowest in the youngest group and increased in the older age groups for all three cohorts.

In Czechia and Denmark, the total CMR was highest in the under 40 age group (CMR 8.8 and 32.1, respectively) and lowest in the 40–50 age group (CMR 4.2) in Czechia and in the over 50 group (18.3) in Denmark (online suppl. Table 1a; for all online suppl. material, see https://doi.org/10.1159/000530822). Inversely, Norway’s under 40 age group had the lowest total CMR (12.9) (online suppl. Table 1b, c) with a higher CMR in the older age groups

In Denmark and Norway, overdoses and suicide dominated within the youngest age group, whereas neoplasms and respiratory deaths increased with age (Fig. 3b, c). However, for the Danish and Norwegian cohorts, while overdoses were highest in the youngest group, overdoses remained a leading cause of death for the 40–50 age group in Denmark (Fig. 3b) and across all age groups in Norway (Fig. 3c). In Czechia, cardiovascular deaths contributed to over one-half of the deaths in the over 50 age group, with no overdose deaths reported in the oldest age group (Fig. 3a). Digestive-related deaths, mainly alcohol-related liver disease, increased with age in Norway, with over twice as many deaths occurring in the over 50-age group compared to the 40-50 year olds. For the over 50 age group, Denmark and Norway reported neoplasms to be the leading cause of death. Overdose in Denmark’s under 40 age group represented the highest CMR across all ages and causes for all three cohorts (CMR 11.8).

Among the nearly 30,000 patients who ever received OMT in the three countries, Czech Republic, Denmark, and Norway, we found large variations in causes of death across the countries, and between gender and age groups. However, all three cohorts saw the highest rates of sudden, unexpected deaths in the younger age group that decreased in the older groups, while physical causes of death increased among the older age groups.

In the general population, circulatory system, cancer (malignant neoplasms), and respiratory disease are the three leading causes of death in Europe [30]. This is similar to the general population in this study’s three countries, with neoplasms and cardiovascular causes being among the top three main causes for each country. However, the OMT population differed somewhat, with the addition of overdose, accidents, and digestive causes among the top three main causes (in addition to cardiovascular and neoplasms).

International trends indicate that patients in OMT are living longer [31]. Several settings have seen increases in the age of patients in OMT [13, 32‒34], including an increasing age for treatment initiation [32]. However, despite these improvements in longevity, this ageing cohort can be considered a neglected group within an already marginalized population [12, 35] with unique challenges and needs. Older OMT patients have complex medical issues and accumulated adverse health effects from long-term drug use, with high rates of psychiatric comorbidities [36], physical complaints [15], and death [17, 18]. Across all age groups, we found physical causes of death (neoplasms, cardiovascular, respiratory, digestive, and infectious) to make up one-third of the total causes, which is similar to other studies reporting high rates of physical causes of death among OMT patients [37‒39]. In addition, among the oldest age group, we found physical causes such as neoplasms to surpass overdose as the leading cause of death in Denmark. This increase in physical causes of death (and decrease in sudden, unexpected deaths) among the older age groups demonstrates a possible indication of success of OMT. While the younger groups experience sudden, unexpected causes of death, this decrease in the older groups can be partly explained by stability that the older groups experience.

Although cancer is a leading cause of death among the general population in Europe [30], OMT patients are nearly twice as likely to die from cancer as the general population [40]; however, this risk is decreased for those in treatment when compared to those out of treatment [41]. This increased mortality may be due to a higher incidence of cancer in this group, or delayed diagnosis and/or treatment [40]. Public health efforts aimed at cancer prevention and early detection may not be equally accessible in this traditionally marginalized and underserved group. Further, people with OUD have increased exposures to known cancer risks, such as tobacco [42], alcohol use [43], and hepatitis C [44, 45]. In this study, neoplasm-related causes of death were the top causes in Denmark and Norway and were the highest in the over 50 group. This supports findings from others that have found crude cancer-specific mortality to increase over time, likely relating to the ageing OMT population [40].

In the general population, mortality related to cancer traditionally has higher rates in men than women [46]. Others have found high rates of neoplasm causes of death among OUD patients [47], with women being significantly more likely to die of cardiovascular and cancer causes [9]. Similarly, in this study, we found women to have high rates of neoplasms in Norway and Denmark and cardiovascular causes of death in Czechia.

In this study, cardiovascular causes of death were the leading cause of death for women in Czechia and accounted for half of the deaths in the over 50 age group. The higher rates in Czechia may relate to the fact that cardiovascular diseases at present form the leading cause of death in the general population despite a decline over the last 30 years [48]. Cardiovascular risk among illicit opioid using populations is a known concern [49‒51]. A recent registry study from Scotland found a high co-presence of respiratory and cardiovascular disease from autopsies for opioid-specific drug-related deaths (DRD) and the authors concluded that circulatory disease was likely related to the quadrupling of methadone-specific DRD in the age group 45 and older [52]. Another recent study found older OMT patients to have high cardiovascular risks compared to a gender- and age-matched sample from general populations [53].

While many of our findings indicate important areas for monitoring for ageing OMT patients, our study also revealed two areas of concern for younger age groups. Overdose and suicide rates were high in the youngest age group, particularly in Norway. While Czechia had lower rates of overdose reported, this could be partly explained by coding differences and the higher reported rates of accidents and suicide. Further, in the USA, the misclassification of suicide and unintentional poisoning underlines the difficulty in accurately detecting intentional versus unintentional poisoning deaths [54]. Others have found a high prevalence of suicide attempts among OMT patients [55, 56], and heroin users have been reported to be 14 times more likely to die from suicide than the general population [56]. Suicide and overdose are intertwined, with overdose being a common method of suicide [57]. Both conditions are considered preventable causes of death, and OMT offers opportunities for frequent, regular contact with health personnel including relevant outreach and intervention. Further, despite the high rates of suicide among OMT patients, others have found significant reductions in suicidal behaviour when receiving methadone to when they were not [58].

A strength of this study was the large sample size across three nationwide cohorts. In addition, the long study period without people being lost to follow-up allowed the ability to observe low-prevalence diseases and examine them within groups. The use of personal identifiers in all three countries enabled the ability to link different national registries. By observing causes of death other than only overdose, this study was able to explore additional causes of death. Further, the study design allowed for the examination of different causes of death between gender, age groups, and three countries, allowing for a broad overview.

This study had limitations to consider. First, to note were the potential variations in habitual coding practices for the causes of death among the three countries. Overdoses form a heterogeneous category regarding ICD diagnostic groups – they can be classified as either suicides, accidents, or events of undetermined intent, and moreover presence or influence of substances in causes of deaths other than poisoning may be coded differently in different countries. As it was shown, forensic toxicology rates and coding practices have a crucial role in the diagnosis of drug-related deaths and form “structural breaks” in monitoring the trends of drug-related death overtime [59]. In this study, we examined OMT populations that included people going in and out of treatment, and during different study periods. We have limited information to assess how this influenced the findings. In addition, complications from a range of conditions may contribute to deaths, and while the underlying cause of death may be determined to be one specifically serious condition, others may have destabilized the patient’s health before the underlying cause of death finally caused the patient to die.

Additionally, while this study spanned several years in each of the cohorts, Norway had a shorter observation period than Czechia and Denmark. Patients from Norway could have started OMT prior to 2010, which may have increased the age at study start. However, a Norwegian study from prior to this study period showed a similar age for treatment start [60]. Further, while all three countries have national registers for OMT patients, there is always a possibility that some proportion of patients receive care without being picked up by the registers (e.g., because they receive care in prison settings, specialized dual diagnosis teams outside of standard care, or with their general practitioner). This loss of patients may also differ across countries. In addition, due to data security issues, it was not possible to merge the data together, which limited the ability to conduct pooled analyses. This study did not examine retention, termination, and re-entry, so it is unknown how such fluctuations could impact mortality rates across the different settings. Finally, the findings do not generalize to people with OUD who do not receive care.

Overdoses remain a leading cause of death for people in OMT. However, physical causes of death surpassed overdoses in some groups, including cardiovascular causes for women in Czechia and neoplasms for people over 50 years in Denmark. For the under 40 age group, suicide was the second leading cause of death in Norway. Our findings have implications for clinical practice and indicate areas for targeting prevention measures among specific groups of OMT patients. This includes health promotion for cardiovascular causes of death (smoking, diet, exercise; screening/diagnostics), suicide prevention efforts, especially within younger age groups, and overdose prevention measures. Life expectancy is changing for OMT patients, and our findings support more focus on preventative care, screening, and treatment for physical conditions. Future studies should examine specific causes of cancer within this group to further target specific cancer prevention and screening efforts for OMT patients.

This study protocol and use of data was reviewed and approved by the Ethics Committee of the General University Hospital in Prague, Czech Republic, approval number 36/19GrantAZVVES20201.LFUK; The Danish national data authority, number 2013540288 (updated March 13, 2020); and the Norwegian Regional Ethics Committee, number 2019/656/REC South-East C. Written informed consent from participants was not required in accordance with local/national guidelines.

Morten Hesse and Birgitte Thylstrup received a grant from Reckitt Benckiser in 2011, when the company distributed buprenorphine. Roman Gabrhelík is the shareholder of Adiquit Ltd., which is currently developing apps for addictions recovery. Nevertheless, no funding was related to this study and the activities had no role in the study design or the data collection, analysis, and interpretation of the data, writing the manuscript, or the decision to submit the paper for publication. The remaining authors have no conflicts of interest to declare.

The study was supported by the Ministry of Health of Czechia, Grant No. NU20-09-00066; Charles University institutional support programme Progress No. Q06/LF1; and by the Research Council of Norway (Grant No. 320360).

Concept and design: Roman Gabrhelík, Svetlana Skurtveit, Desiree Eide, and Morten Hesse. Acquisition, analysis, or interpretation of data: Svetlana Skurtveit, Blanka Nechanská, Gabriela Rolová, Roman Gabrhelík, Desiree Eide, Morten Hesse, and Abdu Kedir Seid. Statistical analysis: Blanka Nechanská, Gabriela Rolová, Svetlana Skurtveit, and Abdu Kedir Seid had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Drafting of the manuscript: Desiree Eide, Svetlana Skurtveit, and Roman Gabrhelík. Critical revisions of the manuscript for important intellectual content: Desiree Eide, Roman Gabrhelík, Svetlana Skurtveit, Thomas Clausen, Viktor Mravčík, Gabriela Rolová, Birgitte Thylstrup, Christian Tjagvad, and Ingvild Odsbu. Obtained funding: Roman Gabrhelík, Viktor Mravčík, Svetlana Skurtveit, and Morten Hesse. Supervision: Roman Gabrhelík, Viktor Mravčík, Svetlana Skurtveit, and Thomas Clausen. All authors have contributed and have approved the final version of the manuscript.

The project uses third-party data derived from state government registries and databases that are ultimately governed by their Ethics Committees and data custodians. Thus, any requests to share these data will be subject to formal approval from each data source used in this study. Further enquiries can be directed to the corresponding author.