A Wider Range of Conditions Contributing to Death in a Cohort of People Who Have Injected Drugs. Findings from an Edinburgh Cohort.

The morbidity and mortality attributed to injecting drug use is a substantial contributor to any study on causes of premature death. Understanding the extent of this may be limited by diculties in observing and recording outcomes over several decades. Historic studies have recorded information in a period when blood borne virus and drug deaths were a smaller proportion or, in the cases of Hepatitis C and HIV/AIDS, absent from National mortality gures.


Abstract Background
The morbidity and mortality attributed to injecting drug use is a substantial contributor to any study on causes of premature death. Understanding the extent of this may be limited by di culties in observing and recording outcomes over several decades. Historic studies have recorded information in a period when blood borne virus and drug deaths were a smaller proportion or, in the cases of Hepatitis C and HIV/AIDS, absent from National mortality gures.

Design and setting
A cohort of people who had, ever, injected drugs was established over a prolonged period of observation in one, community based, medical practice in Edinburgh (UK). Outcomes were measured in the clinical situation and by accessing death certi cates from national, UK, registers.

Findings
Causes of death in a cohort of 794 people who inject drugs (PWIDs) varied over time, some conditions relating to single pathological diagnoses and others were more complicated, multimorbid, and cumulative over time. HIV/AIDS was a striking cause of death until 1995 when antiviral chemotherapy was introduced. Drug related deaths (mainly overdose) remained a signi cant cause of death and death due to alcohol, respiratory, cardiovascular and cancer (mainly lung) increased over time. A wide range of other causes including suicide and violence and trauma were recorded.

Conclusions
Mortality resulting from present or historic drug use may be underestimated in current recoding systems, which largely record deaths from overdose or a single pathological event in an acute situation. The range of conditions causing or contributing to premature death is enormous re ecting multiple risks associated with drug use.

Background
Over the last 40 years cohort studies have reported on survival and outcomes for people who have injected drugs (1)(2)(3)(4). These studies record various morbid processes and events causing death, the most common over time being opiate poisoning, often by accidental overdose. O cial statistics in the UK usually relate to overdose and problems of de nition of what constitutes a drug related death are acknowledged as well as the inconsistency between countries. (5,6). Deaths from other causes are often obscured either because no opiates were present when the nal event occurred, the person died of apparently unrelated cause or past drug use was unknown by the recoding medical attendant. Attempts have been made to describe this, larger, group and to attribute causation to comorbid drug use (1,(7)(8)(9)(10)(11)(12).
Injecting drugs causes more harm than other modes of use and consequently many studies have concentrated on this group rather than the less well de ned people who use drugs without injecting. This study presents a wider view of drug related mortality, the data recording all causes many of which were generated directly by injecting drugs.

Methods
People who had ever injected an illegal drug were recoded in a large community NHS practice by searching the caseload for drug dependence and injecting drug use. From 1984 until 2015 various studies were conducted each requiring separate consent and ethical approval (13)(14)(15)(16). Prospectively participants were " agged" at the National Records O ce for Scotland and ethical permission to use these data for this study was again granted in 2019 (below).
Information retrieved from national records were anonymised and data presented is from death certi cates.

Results
There were 794 people who had at some time in their lives injected drugs in the cohort ( Table 1). The age range at rst injection was wide (11-38 years, mean 19.64, for those who had died compared to 12-39 years mean 20.02 for those who did not die) as was age at recruitment (16-54 years for those dead and 16-52 for those still alive). Years of follow up ranged from less than 1 to 34 (mean 12.50) and years from rst injection less than 1 to 53 (mean 20.86). Age of recruitment, years of follow up, and age at death were similar for men and women ( Table 2).  Table 3 are additional entries on the death certi cate and refer to conditions present at the time of death but not necessarily contributing to the death in the judgement of the certifying doctor.

Conclusions
There are many issues constraining a clear view of the outcomes of drug use. Causes of death varied over time, most strikingly with the introduction of HIV infection into the cohort and the dramatic effect of antiviral chemotherapy in the middle of the 1990 decade (Fig. 1).
Other trends re ect the long incubation period of hepatitis C which was introduced into the cohort in the early 1980's when injecting rst emerged in this population of, then, young people and the increase in drug related, liver and cardiovascular and respiratory causes of death.
The range of other conditions recorded on death certi cates in this group is enormous and some are not directly related to drug injecting ( Table 3).
Scrutiny of this extensive list illustrates the collateral damage often evident in caseloads attending drug treatment centres. Many, if not most, of these co existing conditions at the time of death are seen to be related to a lifestyle associated with poverty, inequalities, and drug use.
Drug use over a lifetime is, for most, an intermittent problem rather than following a linear pattern. Drug use and associated risks are often intermittent following a pattern of relapse and remission (17). There is an acute risk of, one off, hazards such as overdose, infection with a blood borne virus or a vascular event but there is also a cumulative risk over time resulting in premature fatal or morbid events.
Drug use is sometimes a poorly de ned entity with many manifestations some more serious than others and some, like injecting behaviours, with the potential for immediate risk of death. Research often concentrates on short term problems and outcomes of treatment interventions and the impact of treatments such as agonist prescriptions, detoxi cation episodes and residential rehabilitation may not be immediate and therefore are more complicated to measure.
Recorded numbers of drug related deaths reported in the UK and Europe are largely about overdoses and take little account of the larger spectrum of mortality which should be attributed to drug related causes. Current de nitions of a drug related death, in the UK, depend upon the presence of a drug, controlled under the Misuse of Drugs Act 1971, to be present at the time of death. The impact of blood borne viruses and bacterial infections as well as the longer term outcomes from conditions connected to current or past drug use remain absent from totals of death from drug causes.
This study suggests that there are several deaths attributed to drug injecting for every one that is recorded as an overdose. The delayed impact of infection with a blood borne virus or long term drug use has been missing from many studies but needs to be considered when assessing the burden of disease in this population.
Drug policy is often ideologically based and highly political in its construction. Evidence in uences approaches to treatment and prevention only in response to crises and when damage cannot be ignored (18). Primary care and community based facilities are the ideal opportunity to engage and retain contact with people who have long term drug related issues and provides the possibility of a range of interventions including, but well beyond, the prescribing of Medically Assisted Opiate Agonist treatment.
The outcome of drug taking and the continuation or abstinence from illegal drugs is covered elsewhere in this cohort series but at the time of death and in the surviving group many continue to use a mixed economy of drugs, some prescribed and some accessed elsewhere. As in other studies the comorbid use of cigarettes and alcohol is common.

Strengths of this study
This study overcomes the issues of under classi cation of drug related deaths to an extent as it is a population based sample of people mainly recruited quite early in their lives and drug using careers and followed up in multiple ways. It therefore allows a clearer and more complete picture of the true population costs of drug use to be seen -and they are substantial over a life course -probably as substantial as the costs of other things we hear more about, like obesity and lack of exercise or poor diet.
But like all those things the damage is intrinsically linked to and exacerbated by deprivation. This study is the rst to systematically access death certi cates in a cohort of people who have injected drugs. Drug users are sometimes di cult to de ne but the inclusion of a single group identi ed by the characteristic of injecting drugs gives a clear picture of a group of people with, arguably, the most serious risk factor associated with drug use. The study has the advantage of following a group over time and identifying a changing pattern of disease caused by drug policy, emergence of new diseases and the effect of drugs over a long time period. Whereas many previous studies quantify mortality to mainly overdoses of drugs this study adds a clearer, more complete, picture of the multisystem damage. The unique nature of the point in time when HIV entered the drug using culture distinguishes this cohort from others over the past 4 decades of drug use in the UK.

Weaknesses of this study
A cohort from one locality clearly has limited generalizability and must be interpreted in context of local and national variation in treatment policy. Not all patients were recruited at the same time and therefore their risks of death from time dependent variables such as HIV/AIDS and drug policy evolution differs.
The variation in length of time spent injecting drugs was large some individuals injecting for a short period and others consistently over many years. All were included as risks such as blood borne virus transmission can occur during a single injecting event.

Declarations
Ethical approval and consent to participate.
Ethical approval was obtained from the West of Scotland ethics committee. REC reference: 20/WS/0008. Protocol number: AC19108. IRAS project ID: 272287 Consent to the studies conducted in this long running cohort study was granted for each project and is included in the papers published (and cited in this paper). Consent was not considered to be required again for this study when it was granted (above) as no named patient data was included.

Availability of data and materials
All data generated or analysed during this study are included in this article.  Figure 1