Now posts ↓

Thursday 1 May 2008

Postscript 2: Search and suicide

Now have a research author response: see below.

The responses on the British Medical Journal (BMJ) website are mounting up to the 'search and suicide' research they published and which I dissembled.

I mentioned before how a Japanese researcher responded about how they are using the web to directly intervene - Outreach in the Real World - here's a bit more detail. (NB: The BMJ may force you to sign up to view these responses in full).
Not only are online approaches expanding, but outreach-based interventions in the real world are expanding as well. The "Guidelines for Response to Advance Suicide Warning Cases on the Internet," released in October 2005, includes a statement that Internet community organizations can disclose the content of private communications to a third party in emergency situations. As a result, the National Public Safety Commission in Japan reported that among 121 cases of suicide warning on the Internet in 2007, the police identified 105 individuals based on provider information. Of these, 72 cases were intervened by police and other aid measures, while the remaining 33 cases were determined as pranks. Sixteen cases could not be identified because of access from unspecified terminals, such as Internet cafes.
I have absolutely zero problem with this - it's about saving life in the final analysis, not privacy. I can't see why the NHS cannot negotiate similar protocols in the UK. The model is there to learn from. Is that happening? I very much doubt it.

Other BMJ responses are mainly from Hong Kong researchers, but all give a very different take on where research should happen.

'Suicide and internet use levels-- the evidence is lacking', for example, says "the evidence for an association between internet use levels per se and suicide ideation are scant .. it is practically impossible to differentiate visitors drawn by morbid curiosity, boredom, transient dysphoric mood or even academic interest from those with actual suicidal intent, caution should be exercised in interpreting their results."

Another, 'Positive and negative influences of the Internet on suicide', says "the Internet may have beneficial effects on suicide and to better understand the phenomenon both positive and negative influences of the Internet on suicide should be considered." These researchers say that "suicide prevention and intervention can be provided through the Internet for those who have suicidal ideation and visit those sites".

Especially if you use the web to put your positive services in the middle, as the Japanese appear to have learnt.

An American researcher says, "we suggest further studies to be conducted by Biddle and her colleagues comparing the actual traffic to both pro-suicide and suicide prevention websites and focusing future studies on the actual online behaviors of suicidal individuals."

Too right. "Actual online behaviors" being a crucial element to any research around search.

It's very noticeable that there aren't any responses from the UK - how does Biddle et al's research get funded I wonder and why are these other countries seemingly looking at far better and more productive avenues vs. suicide and the web than the sort of web-bashing, headline-grabbing rubbish we're churning out?

NB: I have received no response to email from either the researchers ('Biddle et al') or the charity Sane, which was quoted by the BBC. Help is obviously not wanted as they know it all. Neither, it must be said, have I had much response from search industry bodies and media I contacted. As I have said numerous times before, the industry is lousy at PR. They will wait until regulatory bodies and ill-informed politicians start proposing draconian and irrational legislation before they start dealing with their social impact.

May 5th addition: research author response

Lead author David Gunnell, Professor of Epidemiology Department of Social Medicine, University of Bristol, has responded to my critique amongst others on the BMJ's website. I will note that I wrote to Gunnell and he didn't reply. This is relevant in the context of this response.
We are pleased to see the interest and broad range of views provoked by our paper. There appears to be consensus amongst correspondents about the potential of the Internet both to provide help and to cause harm. Several correspondents have helpfully highlighted additional approaches to altering the results of Internet searching in favour of minimising risk to individuals experiencing serious suicidal thoughts and planning suicide.

Some contributors (Grohol, Canning, Fu, Knight) express surprise at our focus on methods of suicide.
Not me, wasn't expressed by me.
Furthermore Fu points out that some suicidal web-users may have already decided on their method of suicide and be searching for technical information about its implementation. We agree. However, as we explained in the opening paragraph of our paper, we focused on methods of suicide (rather than Internet resources for suicide in general) because research evidence suggests that one of the strongest media influences on suicidal behaviour concerns its affect on the choice of suicide method used.[2,3] Choice of method, in turn, influences the likelihood that a suicidal act will result in death,[4] and so may have an impact on suicide rates. We assumed that some Internet influences on suicide are likely to be similar to those of other media such as TV and newspaper reporting. In keeping with this Fu highlights the possible role of the Internet in relation to the spread of carbon monoxide poisoning as a method of suicide in Hong Kong. Thus we chose the search terms likely to be used by individuals with serious suicidal thoughts who are planning their suicide attempt. We felt it important that organisations offering support to suicidal individuals should endeavour to ensure their sites occur high up on the list of 'hits' retrieved by searches making inquiries about suicide methods.
But they didn't focus on methods. The search terms they chose, which they didn't explain the genesis of in detail, barely included specific methods - Fu's point. If, as there is strong evidence of, people take clues from media reporting - 'copycats' - then surely it would help to look at what methods are shown/described in reporting and take cues from there? This they didn't do. And it's difficult to make recommendations - especially such dramatic ones as they did - when the initial sources "terms likely to be used by individuals with serious suicidal thoughts who are planning their suicide attempt" are so apparently weak.
We agree with Grohol and other correspondents that additional aspects of the Internet in relation to preventing and provoking suicide are worthy of academic study. Indeed this is something we wish to pursue. However, we felt in an area about which much is conjectured, but little empirical data exist, a focussed inquiry into one relevant aspect was appropriate.
In which case they picked the wrong area. Surely going to source - what is known about how suicidal people connect to information and then act on that information, taking cues from well-developed experiences like the Japanese - would be a more productive avenue. For example, what is the role of UK social networks? Where is research lacking on the actual connections prior to suicides? The assumption here was 'search engines' - and everything flowed from those assumptions. Again and again, the main issue here is researchers who aren't webbies, don't understand how the web works, make assumptions, don't talk to the right experts, then leap off with claims which are - plainly - false and suggest solutions they are not qualified to suggest and lead people down more false trails.
In a single brief research paper it would not have been possible to cover the range of aspects suggested by Grohol. We disagree with Grohol's suggestion that we painted a pessimistic picture.
Talk to the PR, what headlines have they generated? How helpful are those headlines? It is highly irresponsible for such researchers to step back here and say 'we're not responsible'.
We described a number of beneficial aspects of the Internet - highlighting advice and information sites. Indeed we report that 13% of our hits were for sites offering support or information about suicide and 12% discouraged suicide. We also highlight that in England, whilst use of the Internet has increased in recent years, rates of suicide have declined. Thus harmful impacts of the Internet on individual acts of suicide are either being offset by beneficial effects or the impact of other suicide prevention activities.
Yes, they mentioned this. But it wasn't their headline. As I showed, searches for 'suicide' appear to be going down. But this wasn't their headline or reported anywhere at all in the subsequent headlines generated from the cues they gave reporters (i.e. the press release). Look at it this way, if the media is a causal factor the researchers just failed to get the media to report the existence of suicide help and to only focus on the existence of harmful websites.
Some of the assertions made by Canning are incorrect. The sites we reviewed were restricted to the first 10 results for each search. The searches 'methods of suicide' and 'suicide methods', which, like most users, we carried out without the quotation marks, do, in fact, yield different sites.
Ok, point taken. Search engines are improving. Though results are barely different. What else makes up "some .. are incorrect".
The interview data we used to inform our search strategy referred to interviews that were conducted with people who had survived nearly fatal suicide attempts. Two of the twelve individuals we have interviewed to date used the Internet to research their methods and indicated the terms they used. We can think of few better methods of identifying search terms than from such individuals.
This detail wasn't in the paper. Two individuals? And remembering terms they used? On this headlines around the world was built? If you want 'better methods', look to the Japanese. It is actually possible to backtrack from the machine they used and find exactly what they searched on. Security services and police do this sort of thing all the time. Further, Gunnell's response doesn't take on board my point that existing industry methods will tell you what terms are most searched on. They couldn't think of methods because they didn't think to ask experts in that field.
Canning makes some good points regarding the relative use of different search engines. We agree, weighting our findings according to site use may have been more informative.
No, not doing it completely undermines the credibility of the numbers reported. Especially the headline numbers cited at the top of the paper.
We did report that searches of Google, the most frequently used search engine, retrieved the highest number of pro-suicide hits and so weighting our findings by site would be unlikely to substantially alter out findings. Canning also provides a series of useful suggestions regarding strategies for reducing the accessibility of pro-suicide sites and working with the Search Engines themselves.
Why do I have 'useful suggestions'? Because I work in the web area. These researchers don't. They couldn't 'think of few better methods' because they didn't 'think' to collaborate with people such as myself or - actually - true experts who do this sort of thing for a living.
The influence of the Internet upon an individual's risk of suicide and upon population suicide rates is currently uncertain. Our study has shed some light on one aspect of the Internet. Further research is needed to ensure there is a more evidence-based debate on the Public Health impact of the Internet on suicide.
But the paper still stands and the publicity is out there. Is this response letter going to do anything about that? Are the researchers going to take their public responsibility seriously?

This is not an academic question. Decision makers are, unfortunately, far more likely to listen to people like Gunnell than people like me. Gunnell does not indicate here that they have learnt from the reaction that further research must happen by drawing on collaborations with industry and experts in the web field.

This is defensive. This is about covering their academic asses. How does this approach help us actually deal with doing the right sort of research which will lead to the right sort of solutions and actually help those they claim to want to help?

1. Biddle L, Donovan J, Hawton K, Kapur N, Gunnell D. Suicide and the Internet: BMJ 2008; 336:800-802
2. Hawton K, Williams K. Influences of the media on suicide. BMJ 2002; 325:1374-1375
3. Schmidtke A, Hafner H. The Werther effect after television films: new evidence for an old hypothesis. Psychol Med 1988;18:665-676.
4. Miller M, Azrael D, Hemenway D. The epidemiology of case fatality rates for suicide in the Northeast. Ann Emerg Med 2004;43:723-730

NB: this response was published in full the next day by the BMJ.


One thing which I did notice on returning to the BMJ's website to look at responses is that they are carrying a blog by a breast cancer victim which appears to be doing it's job - explaining day-to-day reality to practitioners.
I really appreciate the way you are looking at things.Your blog is an eye opener. As a practising physician, it has changed my outlook towards my patients and their illnesses.
Or sortof a blog, posts form part of a category called 'From the other side', which isn't flagged at their top level navigation (and NB again, probably behind a firewall).

It's by Anna Donald.
I should introduce myself before launching into a blog which I hope is not too depressing: living in the shadow of death. This is my starting point, as I was diagnosed with metastatic breast cancer (I lit up “like a Christmas tree” on the scans) in February 2007. It was not a complete surprise - I’d had primary breast cancer - not terribly high risk - in 2003. But disappointing that so much treatment hadn’t cured it.

I hope to write about things that may interest doctors, other health workers and policy makers about what it’s like to have life threatening disease; to be on the other side of the doctor-patient divide, and to experience 21st century health care for a chronic disease (Sydney’s hospitals are pretty similar to Britain’s), from a quality-of-care perspective.
One commentator puts the value of Anna having a voice in a nutshell:
There is a tremendous amount that we can learn from you as you write to us about your experience and insights. We all have challenges which can bring us to a brink (physically, emotionally, spiritually), and it is encouraging and healing to share thoughts and experiences. This can help us to gain empathy and become better health advocates, personally (for others and for ourselves) and professionally

No comments:

Post a Comment