Saturday, 19 March 2016

What can 'big data' tell us about suicide-related behaviours?

The findings reported by Yu-Wen Lin and colleagues [1] (open-access available here) examining the "characteristics and suicide methods of patients with suicide-related behaviors" and "influential factors for repeated suicide-related behaviors and death by suicide" might not make for 'great dinner-party conversation' but are nevertheless important.

Drawing on data from one of the world's premier 'big data' research sources - the Taiwanese National Health Insurance Research Database (NHIRD) - mentioned quite a few times on this blog (see here for example), researchers scanned the database for instances where people were "hospitalized with a code indicating suicide or self-inflicted injury (E950-E959) according to the International Classification of Disease, Ninth Revision, Clinical Modification" between 1997 and 2010. They then set about looking at various factors associated with this patient population (N=57,874).

With the cold, objective spectacles of science on, various important findings are discussed. Approximately 4% of those hospitalised for suicide-related behaviours had unfortunately died. Most were male and most "had died in their first suicide attempt." The most frequently cited method of suicide was self-poisoning.

Analysis by gender revealed some important trends. Males were more likely to die by suicide but females were at "significantly increased odds of repeated suicide-related behaviors when compared with men." Further: "Those who were female, had been hospitalized for suicide-related behaviors at a younger age, had a low income, had a psychiatric disorder (i.e., personality disorder, major depressive disorder, bipolar disorder, schizophrenia, alcohol-related disorder, or adjustment disorder), had a catastrophic illness, or had been hospitalized for suicide-related behaviors that involved two methods of self-inflicted injury had a higher risk of hospitalization for repeated suicide-related behaviors." And factors for males: "had been hospitalized for suicide-related behaviors at an older age, had low income, had schizophrenia, showed repeated suicide-related behaviors, had a catastrophic illness, or had adopted a single lethal method had an increased risk of death by suicide." I might also add that the experience of being hospitalised for suicide-related behaviours at a young age also increased the risk for subsequent "repeated suicide-related behaviors."

There are some important trends emerging from this and other data. Perhaps one of the most important variables was the suggestion that a psychiatric disorder seemed to be part and parcel of that risk profile, whether for suicide-related behaviours or death by suicide. I know this is not new news but does well to reiterate how effective treatment/management of psychiatric labels should remain an absolute priority. I'd also encourage greater screening and assessment [2] among populations who might be particularly vulnerable to mental health issues too (see here for example - particularly timely in light of recent media reports).

The finding that self-poisoning was the primary method of suicide-related behaviour and completed suicides also ties in with other independent data. Accepting that one cannot control every single agent that might be used for self-poisoning, I do believe there are additional measures that can be taken. So, as per the data on the use of paracetamol as a method of suicide (which is by the way, a really horrible way to die), there are ways and means to potentially reduce risk [3] particularly when overdose is an 'impulsive' decision. One might even argue that removing such medicines from the OTC bracket might also help despite the inconvenience to other users. I suppose it depends on how much value you place on life vs. inconvenience or the use of additional resources. I'd also direct readers to some important discussions on another self-poisoning method quite frequently discussed in the research literature: pesticides (see here).  As per the findings reported by Gunnell and colleagues [4], there are solutions here too: "(a) the use of pesticides most toxic to humans was restricted, (b) pesticides could be safely stored in rural communities, and (c) the accessibility and quality of care for poisoning could be improved."

As I've discussed previously on this blog, suicide (whether attempted or completed) is a very personal thing (see here) with no magical 'one-size-fits-all' formula of what to look for in terms of risk profile or indeed, how to help. That there may however be some rough-and-ready factors generally associated with cases is a start in terms of focusing on those who might need most help, including that when it comes to the engineering of social policy [5].

And just in case you need it, there is always someone to talk to (see here).


[1] Lin YW. et al. Influential Factors for and Outcomes of Hospitalized Patients with Suicide-Related Behaviors: A National Record Study in Taiwan from 1997-2010. PLoS One. 2016 Feb 22;11(2):e0149559.

[2] Murray D. Is it time to abandon suicide risk assessment? British Journal of Psychiatry Open. 2016; 2: e1-e2.

[3] Simkin S. et al. What can be done to reduce mortality from paracetamol overdoses? A patient interview study. QJM. 2012 Jan;105(1):41-51.

[4] Gunnell D. et al. The global distribution of fatal pesticide self-poisoning: systematic review. BMC Public Health. 2007 Dec 21;7:357.

[5] Antonakakis N. & Collins A. The impact of fiscal austerity on suicide: On the empirics of a modern Greek tragedy. Social Science & Medicine. 2014; 112: 39-50.

---------- Lin YW, Huang HC, Lin MF, Shyu ML, Tsai PL, & Chang HJ (2016). Influential Factors for and Outcomes of Hospitalized Patients with Suicide-Related Behaviors: A National Record Study in Taiwan from 1997-2010. PloS one, 11 (2) PMID: 26900930