The Global Burden of Disease Methodology Has Been Good for Mental Disorders: But Not Good Enough

The global burden of disease (GBD) studies have revolutionized how we quantitate the burden of disease.¹ Bold in vision, virtuosic in methodology, encyclopedic in scope—the findings have been highly influential. GBD has allowed the health community to combine a measure of premature mortality (years of life lost [YLLs]) with a measure of disability (years lived with disability [YLDs]) in order to derive an overall measure of disease burden (disability adjusted life-years; DALYs). These innovative health metrics have highlighted the increasing contribution of chronic disability to the total burden of disease in the epidemiological transition seen in all countries and territories of the world. Total YLDs are influenced by the years of life lived with a disorder and a disability weight allocated to that disorder. Prevalent disorders with small disability weights may contribute more YLDs to the total compared to rare disorders with large disability weights. Within the common disorders with mild to moderate disability, those with an early onset, and a persistent or recurrent trajectory, will figure prominently in total YLDs. Mental disorders, among others, often have an early onset² and thus meet these criteria. If cardiovascular and respiratory disorders are the chronic disorders of the elderly, mental disorders are the chronic disorders of people of all ages—children, adolescents, middle-aged, and old.

The contribution of mental disorders to disease burden has been put into sharp focus by the GBD methodology.³ In the “league table of disorders,” mental disorders make a major contribution to the YLDs associated with noncommunicable disorders and was the Top 2 YLD burden of all diseases in the world in 2017.⁴ So, three cheers for the GBD—these methods have been good for mental disorders.

Vigo and colleagues⁵ have drawn attention to instances where the GBD may underestimate the true contribution of mental disorders to the total disease burden. In the target article,⁶ they revisit this topic, with data extracted from the 2017 estimates for Canada, the United States, and Mexico. They provide a convincing case that mental disorders should include the YLDs from intentional self-harm and the YLLs from suicide along with the officially sanctioned mental disorder-associated causes of death (alcohol use disorders, drug use disorders, eating disorders). Furthermore, they argue that mental disorders should harvest YLLs and YLD from selected neurological disorders, which are currently reported separately from mental disorders in GBD. With the redistribution of disorders into the “mental, neurological, substance use disorders and self-harm,” Vigo and colleagues estimated that the burden associated with this revised definition is the largest of all disorder groupings, especially in the United States.

Where the boundary should be drawn between neurological and psychiatric disorders is debatable (and influenced more by history and discipline loyalties than by empirical data). However, we agree that GBD’s current methodology systematically underestimates the burden of mental disorders. We wish to advocate for an additional change to the GBD methods that would contribute to a more realistic estimation of the disease burden of mental disorders. The current method links all YLLs to a single death based on a restricted set of causes of death (that does not include most mental disorders). We know that people with mental disorders have an increased risk of comorbidity with general medical conditions that contribute substantially to premature mortality.⁷ For example, on average, men and women with a mental disorder die 10 and 7 years earlier, respectively, than the age- and sex-matched general population. Furthermore, while they have an increased risk of premature mortality due to suicide, most of the premature morality associated with mental disorders is attributed to common general medical conditions (e.g., circulatory and respiratory disorders, diabetes). The mechanisms underpinning this premature mortality are complex and multifactorial (e.g., shared genetic and/or environmental risk factors, the influence of psychotropic medications, the influence of mental disorders on subsequent socioeconomic status).

Within the current GBD framework, if a person has a mental disorder (e.g., persistent depression from age 30 years) and subsequently dies from a myocardial infarct at age 60, YLDs will accumulate for depression (30 years multiplied by the disability weight linked to depression). However, the YLLs (29.32 years; corresponding to the difference between age at death, 60 years, and expected global norm for life expectancy at that age) will be allocated totally to cardiovascular disorders. While the prior mental disorder may have contributed substantially to the premature mortality, GBD does not attribute any of the resultant YLLs to mental disorders. The GBD has a scientific scotoma when it comes to the complex pathways that contribute to comorbidity and causes of death. Currently, there are no objective, comprehensive methods to redistribute YLLs to prior disorders. However, the GBD may be able to explore the links between mental disorders and premature mortality via their comparative risk assessment (CRA) process.⁸ These are well-established for many exposure–outcome pairs (e.g., smoking and lung cancer, obesity and diabetes, childhood bullying and adult depression and anxiety disorders) and currently include exposures such as alcohol use and drug use, and disease-associated biomarkers. While disorders are not currently included as exposures in the CRA process, it is feasible that the GBD could include prior mental disorders and later general medical conditions within this framework. Thus, GBD could start the long journey down the path of providing two types of analyses: (a) the current framework of direct disorder associated YLDs and YLLs and (b) an expanded framework that also includes indirect risk-attributable burden, where mental disorders are seen as risk factors for later general medical conditions.

Galileo made a concise recommendation for the scientific process—“measure what is measurable—make measurable what is not so.” The GBD has made a wide range of health metrics measurable. Now, two decades later, the field needs to make what is currently unmeasurable measurable. GBD may need to widen the portfolio of health metrics to accommodate this type of real-world complexity. As the maxim states bluntly—“If you can’t count it, it won’t count.”⁹ We argue that the GBD methods have been very important for mental health, but there is room for improvement. These methods have been good for mental health—but not good enough.