From the Editor

Do suicide barriers really work, or do they cost money that could have been better spent elsewhere?

This debate raged in Toronto about a decade and a half ago when some argued that a suicide barrier must be added to the Bloor Viaduct. That bridge, which connects downtown Toronto with its east end, was considered a “suicide magnet” – a place that proved popular for suicide attempts. In North America, only the Golden Gate Bridge in San Francisco had more suicides associated with it. Media coverage of the debate included an article in The New York Times; it ran with the dramatic headline: “A Veil of Deterrence for a Bridge With a Dark Side.”

And in 2003, the barrier went up.

So did the suicide rate go down?

This week, we look at a new paper just published by Sunnybrook’s Mark Sinyor et al. Drawing on years of data, they consider the utility of the Bloor Viaduct suicide barrier. They find that it has saved lives. The paper obviously has implications for other bridges and cities.

The Bloor Viaduct: big debate, big outcome?

The authors also looked at media reporting on suicide.


Suicide Barriers and Outcomes

“Did the suicide barrier work after all? Revisiting the Bloor Viaduct natural experiment and its impact on suicide rates in Toronto”

Mark Sinyor, Ayal Schaffer, Donald A Redelmeier, Alex Kiss, Yasunori Nishikawa, Amy H Cheung, Anthony J Levitt, Jane Pirkis

BMJ Open, May 2017


Means restriction is arguably the population-based suicide prevention strategy with the most robust evidence base. Restricting access to common methods of suicide is thought to disrupt the suicide process because suicidal crises may be short-lived and people often report a preference for specific means. Critical reviews of the literature suggest only a small risk of people seeking out other ways to die from suicide once a specific method is no longer available.

Special attention has been given to suicide by jumping from height with suicide prevention barriers constructed at the Empire State Building, the Eiffel Tower and bridges around the world including the Bloor Street Viaduct in Toronto, Canada, the second most frequented bridge worldwide for suicide death after the Golden Gate Bridge in San Francisco. Our group studied the impact of the barrier 4 years after it was completed and demonstrated that, although there had been no further suicide deaths at the Bloor Street Viaduct, the number of suicide deaths by jumping from a height in Toronto was unchanged due to a statistically significant rise in suicide deaths at other bridges. This finding is in contrast to those from a meta-analysis conducted by Pirkis et al examining six studies which showed that, despite a 44% increase in suicide deaths at other jumping sites, barriers still resulted in a 28% net reduction of suicide deaths by jumping in the study cities.

Suicide barriers are prominent, visible and often controversial interventions that may garner substantial attention from the media. This is notable given the well described ‘Werther Effect’ in which media reporting on suicide is thought to have a causal relationship with increased rates of death in an area through social contagion. Media reporting specifically on suicide by jumping has been positively associated with suicide rates and therefore may be particularly likely to result in contagion effects. Given evidence that media reporting on deaths at particular jumping sites encourages ‘copycat’ behaviour, it is plausible that such media reports occurring proximal to the construction of a suicide barrier may have a similar impact.

The construction of a suicide barrier therefore may function as a population-based natural experiment that ‘tests’ two potentially opposing impacts on suicide death; that is, it may test the value of a means restriction strategy designed to reduce suicide deaths as well as the impact of the consequent media discussion that may inadvertently increase them.

sri_mark-sinyorMark Sinyor

So opens a new paper by Sinyor et al. looking at suicide in Toronto, before and after the construction of the suicide barrier on the Bloor Viaduct.

Here’s what they did:

  • Drawing on data from the Office of The Chief Corner of Ontario, the authors looked at the number of suicides for the 11 years before the suicide barrier’s construction (January 1993 to December 2003) and the 11 years after (January 2004 to December 2014).
  • Suicide deaths were further divided: “all suicides in Toronto, suicides in Toronto by jumping… and suicides in Toronto by means other than jumping.” Suicides by jumping were further divided whether it was from a bridge or a building; the location of the bridge was also obtained. To calculate a rate, they used census data from Statistics Canada.
  • To consider media reports, they used a media tracking service, and looked at 11 local and national publications in the Toronto area. A search was done for articles on “suicide” with coders then reviewing the articles for terms like “Viaduct” and “Jump.” The articles were then rated in different ways, including whether they expressed a negative view of the barrier or if they expressed a message of hope: suicide is preventable.
  • Statistical analyses were done, including a Poisson regression analysis.

Here’s what they found:

  • “Per-capita rates at that location have declined from 9.0 deaths per year before the barrier to 0.1 deaths per year after the barrier (p=0.002).” In other words, the barrier worked at the Bloor Viaduct.
  • But what about other bridges? “Suicide deaths from bridges in Toronto have also declined by a similar absolute number (18.8 deaths per year before the barrier vs 10.0 deaths per year after the barrier, p<0.0001)… There has been no statistically significant rise in deaths by jumping from other bridges in the city overall, walking distance bridges, the nearest comparison bridge or from buildings. There was a numeric but non-significant reduction in overall suicide deaths by jumping (57.0 deaths per year before the barrier vs 51.3 deaths per year after the barrier, p=0.07). Suicide deaths from the nearest comparison bridge rose in the years during which the barrier was constructed and in the 2  years afterwards, but suicide deaths at that location have since declined to prebarrier levels.”
  • “Per capita rates of suicide overall and by means other than jumping have also declined significantly over the study period (p<0.0001; p=0.001, respectively).” (!)
  • In terms of media reporting, there was a mixed picture but “messages of hope were associated with a decrease in suicide deaths by jumping from buildings only (IRR 0.9869).”


They note:

What this study adds is the long-term data showing that suicide deaths from all bridges in Toronto have declined by a similar number to those prevented at the Bloor Street Viaduct with no increases in deaths at nearby bridges. These results differ substantially from the earlier, short-term findings.

A few thoughts:

  1. This is a good study.
  1. This is a study that has implications far beyond Toronto. The results are also consistent with other studies on bridges and suicide barriers. For example, after city officials constructed a barrier for the Clifton suspension bridge in Bristol, UK, death by suicide at the bridge dropped by 90% with no rise in (male) suicides by jumping in other sites in the city. Note that Sinyor et al. had a bigger dataset covering far more years. The British Journal of Psychiatry paper on the Clifton suspension bridge can be found here:
  1. The implications for suicide prevention are clear. Indeed, this paper may well save lives. (!!)
  1. So, should Reversing Falls Bridge in Saint John, New Brunswick, have a suicide barrier? Based on the findings, Dr. Sinyor argues yes in a CBC interview. You can find it here:
  2. As for the Golden Gate Bridge, construction began this spring on a suicide net that will span the whole length of the bridge.The Golden Gate Bridge in San Francisco, CA at sunset.Golden Gate Bridge: big bridge and, soon, big safety net
  3. The study is not without limits, of course. As the authors note, the study is an uncontrolled natural experiment, meaning that there could be factors unaccounted for that influence the findings. “For example, although we were able to control for population growth per capita, we could not control for other population-based factors such as knowledge of the Bloor Street Viaduct as a suicide hotspot or societal changes that might have impacted on chosen suicide methods.”
  4. On a larger note: the quality of suicide research has markedly improved in recent years. Nice.


Reading of the Week. Every week I pick articles and papers from the world of Psychiatry.