Findings from the Global Burden of Disease Study 2010 (GBD 2010) have reinforced the understanding of the significant impact that mental, neurological, and substance use disorders have on population health (Murray and others 2012; Whiteford and others 2013). One key finding was the health transition from communicable to noncommunicable diseases across all regions. This transition was particularly evident in low- and middle-income countries (LMICs) (Murray and others 2012), where the proportion of burden attributable to noncommunicable disease increased from 36 percent in 1990 to 49 percent in 2010, compared with an increase from 80 percent to 83 percent in high-income countries (HICs) (IHME 2013).

GBD 2010 estimates that the majority of disease burden caused by mental, neurological, and substance use disorders is from nonfatal health loss; only 15 percent of the total burden is from mortality in years of life lost (YLLs) (IHME 2013). This finding may erroneously lead to the interpretation that premature death in people with mental, neurological, and substance use disorders is inconsequential. A recent review has shown higher mortality risks than the general population for a range of mental disorders, with a standardized mortality ratio (SMR) as high as 14.7 for opioid use disorders (Chesney, Goodwin, and Fazel 2014). Excess mortality in people with epilepsy is reported to be two-to three-fold higher than that of the general population, with an increased risk up to six-fold higher in LMICs (Diop and others 2005). A significant proportion of these deaths is preventable (Diop and others 2005; Jette and Trevathan 2014).

There are multiple causes for lower life expectancy in people with mental disorders (Chang and others 2011; Crump and others 2013; Lawrence, Hancock, and Kisely 2013). Self-harm is an important cause of death, but the majority of premature deaths are caused by chronic physical disease, particularly ischemic heart disease (IHD), stroke, type II diabetes, respiratory diseases, and cancer (Crump and others 2013; Lawrence, Hancock, and Kisely 2013). Dementia is an independent risk factor for premature death; and patients with physical impairment, inactivity, and medical comorbidities are at increased risk (Park and others 2014).

In many HICs, the life expectancy gap between those with mental disorders and the general population is widening. The general population enjoys a longer life, while the lifespan for those with mental, neurological, and substance use disorders remains significantly lower and unchanged (Lawrence, Hancock, and Kisely 2013). Information on the extent and causes of premature mortality in people with mental, neurological, and substance use disorders in LMICs is sparse, but these groups are understood to experience reduced life expectancy, although causes of death may vary across regions.

This chapter explores the cause-specific and excess mortality of individual mental, neurological, and substance use disorders estimated by GBD 2010 and discusses the results. We present the additional burden that can be attributed to these disorders, using GBD results for comparative risk assessments (CRAs) assessing mental, neurological, and substance use disorders as risk factors for other health outcomes. We focus on the following mental, neurological, and substance use disorders:

1. Mental disorders, including schizophrenia, major depressive disorder, anxiety disorders, bipolar disorder, autistic disorder, and disruptive behavioral disorders (attention-deficit hyperactivity disorder [ADHD] and conduct disorder [CD])

2. Substance use disorders, including alcohol use disorders (alcohol dependence and fetal alcohol syndrome) and opioid, cocaine, cannabis, and amphetamine dependence

3. Neurological disorders, including dementia, epilepsy, and migraine.

For the purposes of GBD 2010, countries were grouped into 21 regions and 7 super-regions based on geographic proximity and levels of child and adult mortality (IHME 2014; Murray and others 2012). Regions were further grouped into developed and developing categories using the GBD 2010 method. Details of countries in each region and super-region can be found on the Institute for Health Metrics and Evaluation (IHME) website (IHME 2014).

The mortality associated with a disease can be quantified using two different, yet complementary, methods employed as part of the GBD analyses. First, cause-specific mortality draws on vital registration systems and verbal autopsy studies that identify deaths attributed to a single underlying cause using the International Classification of Diseases (ICD) death coding system. Second, GBD creates natural history models of disease, drawing on a range of epidemiological inputs, which ultimately provide epidemiological estimates for parameters including excess mortality—that is, the all-cause mortality rate in a population with the disorder above the all-cause mortality rate observed in a population without the disorder. By definition, the estimates of excess deaths include cause-specific deaths.

Although arbitrary, the ICD conventions are a necessary attempt to deal with the multi-causal nature of mortality and avoid the double-counting of deaths. Despite the system’s clear strengths, cause-specific mortality estimated via the ICD obscures the contribution of other underlying causes of death—for example, suicide as a direct result of major depressive disorder—and likely underestimates the true number of deaths attributable to a particular disorder. However, the estimation of excess mortality using natural history models often includes deaths from causal and noncausal origins and likely overestimates the true number of deaths attributable to a particular disorder. The challenge is to parse out causal contributions to mortality, beyond those already identified as cause-specific, from the effects of confounders.

The quantification of the burden attributable to risk factors requires approaches such as CRA, which is now an integral part of the GBD studies. The fundamental approach is to calculate the proportion of deaths or disease burden caused by specific risk factors—for example, lung cancer caused by tobacco smoking—while holding all other independent factors constant. A counterfactual approach is used to compare the burden associated to an outcome with the amount expected in a hypothetical situation of ideal risk factor exposure, for example, zero prevalence. This provides a consistent method for estimating the changes in population health when decreasing or increasing the level of exposure to risk factors (Lim and others 2012).