‘Deaths of despair’: A term that needs to be retired

The term ‘deaths of despair’ gained a great deal of currency as shorthand for three causes of death—drug overdoses, suicides and alcohol-related liver disease [1]. In 2015, Case and Deaton [1] noted that life expectancy in the United State declined between 1999 and 2013 and that this was attributable to large increases in deaths from these three causes. These collectively came to be described as ‘deaths attributable to despair’ [1, 2]. As they later noted, what began as a descriptive tag resonated with the media and captured the interest of many researchers [2]. We believe that the term is well intentioned but mistaken in theory, of doubtful validity, and misleading in its implications for policies likely to reduce premature death.

The term ‘deaths of despair’ collates deaths caused by substance poisoning, suicide and alcohol-related disease to form a distinct epidemiological phenomenon driven by cumulative economic disadvantage [1, 2]. Socio-economic factors such as high unemployment and a loss of traditional social structures are argued to be responsible for a high level of societal despair. This despair is proposed to be the common factor driving substance poisoning, suicide and alcohol-related disease. Since its postulation in the United States, the terminology has been used across a range of other jurisdictions including Eastern Europe, the United Kingdom and Canada [3-5].

There are undoubtedly commonalities between events such as overdose, suicide and harmful alcohol use, which include social deprivation and trauma, and the term does serve to highlight such deprivation. A simple interpretation of the term that suggests a common entity and a central causal role for despair, however, does not capture the complexity of these phenomena. Let us first consider substance use and overdose. There is little evidence that despair is the primary driver of the initiation of substance use, its continuation, dependent use or overdose. Certainly, self-medication plays a role in problematic substance use, but the initiation of drug use involves a complex range of factors. Foremost in the United States was a massive increase in the prescribing and promotion of opioids in the 1990s [6]. Individual risks play important roles, and include impulsivity, attitudes and one's social network [7]. Once use has commenced, dependence becomes its own driver of continued use, involving a range of psychological, behavioural, social and physiological signs and symptoms [8]. For drugs such as the opioids or hypnosedatives, tolerance and withdrawal are powerful motivators for continued use [8]. There are a range of well-established risk factors for overdose including a history of overdose, dependence, higher drug purity/doses, the concomitant consumption of other drugs, reinstatement after a period of abstinence and injection as a route of administration [9, 10]. Most overdoses involve established, dependent drug users, most are men, and almost all are accidental [6, 9-12]. It is notable that in recent years we have seen large increases in deaths because of synthetic opioids such as fentanyl and among older people [10, 12].

In contrast, risk factors for suicide include a previous attempt or self-harm, a history of child abuse, major depression, poor health, mood disorders, post-traumatic stress disorder and psychotic illnesses [8]. Economic factors, such as unemployment, are certainly relevant, but are only one aspect of a complex clinical picture. Rather than aspects of a single entity, overdose and suicide are two distinct phenomena, associated with different risk factors. Even among dependent opioid users, a population with elevated suicide risk, overdose and suicide are predicted by separate factors [9].

It is not clear why alcohol-related liver disease would be expected to co-vary with overdose and suicide. Overdose and suicide may be characterised as acute risks. Alcohol-related liver disease is a chronic condition that progresses over years of heavy drinking. Such drinking patterns are driven by dependence. Other risk factors include problematic parental alcohol use, genetic influences, cultural attitudes, availability and peer alcohol use [8]. There is no evidence for the assumption that such drinking patterns are driven primarily by despair. Moreover, there appears to be a temporal incongruence in grouping the long-term sequelae of long-term drinking with overdose and suicide.

Epidemiological studies of population trends are inconsistent with grouping these causes into a distinct epidemiological phenomenon [3, 13-17]. These studies have reported divergent trends between these causes across different countries, and that trends vary by sex, age, birth cohort and spatial location [14, 15, 17]. In the case of heroin overdose, the heroin market appears a stronger predictor than an epidemiological entity driven by despair [16]. Indeed, heroin droughts result in fewer users and overdoses, whereas gluts have the reverse effect, independent of economic circumstances [18].

We believe that talk of ‘deaths of despair’ has given rise to misconceptions about the causal mechanisms that underlie these phenomena and, hence, the policies needed to reduce them. Assuming that ‘deaths of despair’ are driven by a common factor suggests that reducing despair will reduce these fatality rates. It is not clear how one can reduce societal despair apart from broad economic improvements. Although we can all agree on the benefits of a healthier and fairer economy, responses to these three major causes of premature death need to be specific to each.

There is no one way to reduce overdose mortality. One involves increasing enrolments in drug treatment programmes, both medication assisted and residential rehabilitation [6]. The risk of overdose and all-cause mortality is substantially lower among those enrolled in treatment [19]. Harm reduction approaches such as naloxone distribution to first responders and the family and friends of at-risk persons can also prevent deaths. Other suggested interventions include safer prescribing practices and non-opioid medications for chronic pain [6]. The most effective way to reduce alcohol-related morbidity and mortality may be a combination of increased alcohol taxes, reduced alcohol availability and increased treatment access. In the case of suicide, prevention requires screening for suicidality in medical settings and the treatment of known risk factors, such as depressive illnesses and psychosis, in at-risk persons.

In our view, it is time to retire the term ‘deaths of despair’. Although it did draw attention to real sociological issues, it has outlived its usefulness and does not describe a distinct epidemiological phenomenon. The terminology invites a collation of different phenomena and is a potential barrier to the adoption of more specific evidence-based interventions. These disparate phenomena need to be viewed separately and responded to accordingly.

Shane Darke: Conceptualization; writing—original draft. Michael Farrell: Conceptualization; writing—original draft. Wayne Hall: Conceptualization; writing—original draft. Julia Lappin: Conceptualization; writing—original draft.

None.