In this study on mortality in the total Norwegian OAT population, both somatic and drug-induced causes of death were frequent during OAT. In the 2-year observation period, 1.4% of the patients died. In general, CMRs increased with age, and this pattern was more pronounced for somatic causes than other causes of death. The CMR was also higher in patients taking methadone compared with buprenorphine. In the multiple regression model, we found that increasing somatic comorbidity, as measured by the Charlson comorbidity index, reduced the odds of dying of a drug-induced cause of death compared with other causes of death.
In line with previous Norwegian studies, we found that somatic causes of death among OAT patients predominated [8, 17]. Non-communicable diseases such as cancer and COPD take time to develop and are associated with both age and the lifestyle factors prevalent among OAT patients. High rates of pulmonary diseases and increased cancer risk are consistent with previous findings in ageing OAT patients [27,28,29,30,31]. COPD and emphysema are independent risk factors for lung cancers, together with smoking, and predict reduced survival [32, 33]. Liver cirrhosis and liver cancer due to hepatitis C also contribute substantially to morbidity and mortality among opioid users [34]; however, despite a high prevalence of hepatitis C among the deceased in our study, only 14 patients died of liver disease. Some of the deaths of somatic origin were probably more directly associated with injecting drug use. Acute bacterial skin and soft tissue infections are common among injecting drug users [35], and bacteraemia often causes severe focal infections and sepsis. Persisting infections and inflammation caused by continued injecting and skin popping (subcutaneous injecting) are also associated with AA amyloidosis [36, 37]. AA amyloidosis was not encountered among heroin users in Norway until 2005 [37] but is now an emerging issue among the ageing OAT patients. Injection-related health risks other than blood-borne viral infections in OAT patients who continue to use drugs might be an under-researched topic.
Although the overdose risk is reduced during OAT, nevertheless 42% of the patients in our study had a drug-induced cause of death. None of the patients died of an overdose in the first month after initiating OAT. The increased risk of fatal overdose during initiation of methadone may vary according to treatment setting [4, 38]. According to Norwegian OAT guidelines, both buprenorphine and methadone should be initiated under monitoring and observation, and inpatient detoxification at the initiation of OAT is common [19]. Methadone was judged to be the main intoxicant in 31 of the 84 drug-induced deaths and, in all except two cases, the patient was taking methadone as the OAT medication. The interpretation of this finding is not straightforward. It is difficult to determine the precise role of OAT medication in fatal overdoses [39]. The instituted dose of methadone may become dangerous because of increasing vulnerability as OAT patients age and comorbidity levels rise. The overdose risk among OAT patients is associated with several factors such as somatic and psychiatric comorbidities, co-prescribing, previous non-fatal overdoses and polydrug use [40,41,42], which may make it difficult to ascertain the exact cause of death. In addition, the post-mortem examiner is not always informed about the OAT status. Thus, the number of methadone deaths might represent an overestimation, and may in fact have been caused by single or multiple somatic causes in combination with regular prescribed methadone doses.
An all-cause mortality rate of 1.4/100 PY during OAT was the same as found in an earlier Norwegian study [17], but higher than the rate of 0.93/100 PY found in a systematic review and meta-analysis [4]. In line with previous studies, CMRs increased with age, and were higher in men and for patients taking methadone compared with buprenorphine [2,3,4, 8]. Suggested explanations for increased CMRs among patients taking methadone are methadone-induced prolongation of the QTc interval, increasing the risk of ventricular cardiac arrhythmia (torsades de pointes) and “sudden death”, ingestion of alcohol and BZD, physical comorbidities and harder-to-support patients [15, 16]. In the Norwegian setting, the difference in CMRs might be explained by a “veteran effect”. Until 2001, methadone was the only OAT medication. Patients taking methadone in our study had been treated in OAT for significantly longer than those taking buprenorphine, and most likely had a longer drug career. In addition, patients with a severe or terminal disease such as cancer taking buprenorphine are often converted to methadone or other opioids.
In the regression analysis, we found an association between increased somatic morbidity and reduced odds of a drug-induced cause of death. The Charlson comorbidity index was moderately correlated with age, which could be one explanation why age did not remain significant in the multiple model. Multimorbidity (i.e., having two or more chronic diseases) is associated with increased risk of mortality, functional decline, polypharmacy, increased number of hospital admissions and poorer quality of life [43]. Multimorbidity usually increases with age [43], but patients in OAT have high rates of chronic diseases across all age groups [44, 45]. Several of the patients in our study had multiple severe and potentially fatal medical conditions, and thus several competing disease end-points.
Somewhat surprisingly, given the superior safety profile of buprenorphine, we did not find that taking methadone increased the odds of drug-induced cause of death compared with buprenorphine. The lack of association between the other covariates and drug-induced cause of death could be because the two groups were quite similar, which makes differences less likely to detect. Risk factors not included in the model (e.g., prescription medication, drug use) could be another explanation.
Our findings have several implications. Multimorbidity in OAT patients calls for a broad range of patient-oriented and organizational measures, such as improved treatment and follow-up of chronic diseases and multidisciplinary teamwork and co-ordination of care [43, 44]. The high prevalence of COPD and pulmonary cancer suggests that a stronger focus on tailored tobacco harm-reduction approaches and smoking cessation is important for this patient group, and as early in their lives as possible, to reduce cumulative risk. OAT patients should be offered spirometry and lung image tests [32, 33]. Overdose prevention is a multifaceted challenge [14]. Further measures may include improved follow-up after non-fatal overdose, reviewing older patients’ methadone dosage in the context of somatic comorbidities (e.g., reduced liver and kidney function) and offering regular electrocardiograms to patients aged > 45 years. Distribution of intranasal naloxone to at-risk populations is also relevant [15, 46, 47].
Strengths and limitations
The strengths of our study include the use of register data that were combined with information from hospital medical records. This gave in-depth information about the fatalities that were not accessible using register data alone. The national OAT programme is organized within the public specialist health care service in Norway, and has a monopoly of this treatment modality; thus, we were able to study mortality in a complete, national OAT population. The high rate of forensic or medical autopsy also strengthens the validity of the findings. A valid cause of death was not established in only two patients (1%).
Our study has several limitations. Almost half (47%) of the questionnaires were completed by clinicians other than physicians, who do not always have access to somatic medical records. Thus, we cannot rule out the possibility of information bias. Regarding somatic comorbidity, we have no data on smoking status, but the smoking prevalence among Norwegian OAT patients is high and similar to the 69–94% prevalence reported in earlier studies [27, 45, 48, 49]. In addition, the number of non-fatal overdoses is probably under-estimated, because most overdoses in Norway are attended by the ambulance service only. A higher number of participants would have allowed for more variables in the regression analysis. We did not have information on the changes in variables that can vary over time, such as prescription of BZD, psychotropic medications and changes in OAT medication before death. The broad categories of prescribed medication (at least one prescription of benzodiazepine and psychotropic medication in the year before death) limited their use as covariates in the regression analyses.