Intended for healthcare professionals

  1. Ellicott C Matthay, postdoctoral scholar
  1. University of California, San Francisco, Department of Epidemiology and Biostatistics, Box 0560, 550 16th Street, 2nd floor, San Francisco, CA 94143, USA
  1. Correspondence to: ellicott.matthay{at}

Overall trends hide extreme heterogeneity across and within countries

National and international agencies seeking to reduce suicide need timely and accurate mortality data to set targets, benchmark progress, and identify those in need of intervention. Countries, however, vary in the quality of data sources and analytical methods used to produce cause-specific mortality estimates, if they produce them at all. The Global Burden of Disease Study addresses these concerns by generating comparable estimates of death, disability, and injury by using uniform metrics across countries and time. Investigators draw data from all available sources on the frequency of health conditions, make corrections for known biases, and incorporate estimates into a sophisticated modelling framework that draws information across space and time in areas where data are weak or nonexistent.1

In a linked article, Naghavi and colleagues (doi:10.1136/bmj.l94) analysed findings from the 2016 iteration of the Global Burden of Disease Study and identified remarkable changes in suicide rates worldwide.2 They describe patterns of suicide mortality and years of life lost globally, regionally, and for 195 countries and territories by age, sex, and Socio-demographic index from 1990 to 2016. Their results indicate substantial reductions in suicide globally—a 33% decline in the age standardised suicide rate between 1990 and 2016—but underscore that suicide remains a leading cause of years of life lost in many parts of the world. Further, their findings draw attention to the remarkable heterogeneity in suicide trends across countries and demographic subgroups that warrant further investigation.

Exceptionally high (Lesotho, Lithuania), rising (Zimbabwe, Jamaica), low (Lebanon, Syria), or falling (China, Denmark) suicide rates prompt the question of whether estimated patterns are real. The lack of complete and high-quality vital registration data in most countries leads to heavy reliance on modelling; results could reflect modelling assumptions rather than underlying data. Denominators used for rates could be subject even more uncertainty,3 as counting deaths is easier than counting those at risk of death. The Global Burden of Disease Study’s Data Input Sources Tool is useful in this regard, as it indicates those places where strengthening the accuracy and completeness of data collection systems is especially needed.

Results could also reflect data problems such as under-reporting, differential reporting, or misclassification of cause of death owing to the sensitive and illegal nature of suicide in many countries.4 The study methodology incorporates numerous strengths and innovations that likely enhance the validity of suicide mortality estimates, however, it is challenging to fully correct for these limitations and the reported uncertainty intervals do not fully account for them. Results should therefore be interpreted with some caution.

If accurate, study estimates can form the basis for follow-up studies to investigate the reasons for increases, declines, and inequalities. For example, between 1990 and 2016, age standardised suicide rates in China fell 64% from 23.4 to 8.4 per 100 000, but age standardised rates in Zimbabwe rose 96% from 14.2 to 27.8 per 100 000. Variation by age and sex and in the suicide sex ratio across development levels is similarly extensive and intriguing. Identifying the major drivers of these variations has the potential to uncover major insights that could inform prevention efforts in numerous settings. Simultaneously, the extreme heterogeneity in trends highlights that suicide is sociologically complex, making detailed, context-specific case studies, empirical analysis, and formal hypothesis testing necessary.

The Global Burden of Disease Study’s broad scope makes it an excellent tool for priority setting, but inevitably entails trade-offs. Because the framework must be relevant and feasible to execute across 264 causes of death, it does not incorporate factors important for suicide but not other outcomes. This study and other Global Burden of Disease research therefore lay the groundwork for future suicide studies to incorporate such factors and inform prevention efforts. 15678

For example, determining the means of suicide is crucial to evaluate and inform means related interventions, historically one of the most successful avenues for suicide prevention.9 Global estimates of “firearm” and “nonfirearm” means pave the way for future incorporation of more detailed means.1 Estimates of burden attributable to drug and alcohol use can similarly be enhanced by the addition of risk factors such as access to means, mental disorder prevalence and treatment, family history of self harm, and exposure to violence.510 Further, to inform national health plans that incorporate equity goals, stratification by demographic variables beyond age and sex, such as religion, race or ethnicity, occupation, and subnational geographic units is essential.11

Naghavi and colleagues’ findings will spur research that could inform future policy. Results could prove useful to governments, international agencies, donors, civic organizations, physicians, and the public to identify the places and groups at highest risk of self harm and to set priorities for interventions. This is particularly true for countries without complete vital registration systems where trends and high-risk groups are not well established. As new data and methods emerge, regular updating of suicide mortality estimates will be needed to inform research, policies, and recommendations with the best available evidence.



Log in

Log in through your institution


* For online subscription