By continuing to browse this site you agree to us using cookies as described in About Cookies
Notice: Wiley Online Library will be unavailable on Saturday 7th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 08.00 EDT / 13.00 BST / 17:30 IST / 20.00 SGT and Sunday 8th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 06.00 EDT / 11.00 BST / 15:30 IST / 18.00 SGT for essential maintenance. Apologies for the inconvenience.
Restricting access to a suicide hotspot does not shift the problem to another location. An experiment of two river bridges in Brisbane, Australia
Australian Institute for Suicide Research and Prevention
National Centre of Excellence in Suicide Prevention
WHO Collaborating Centre for Research and Training in Suicide Prevention, Griffith University, Queensland
Correspondence to: Dr Chi-kin Law, Australian Institute for Suicide Research and Prevention, Mt Gravatt Campus, Griffith University,176 Messines Ridge Road, Mt Gravatt, Queensland 4122; e-mail: firstname.lastname@example.org
The authors have stated they have no conflict of interest
Background: Restricting access to lethal means is a well-established strategy for suicide prevention. However, the hypothesis of subsequent method substitution remains difficult to verify. In the case of jumping from high places (‘hotspots’), most studies have been unable to control for a potential shift in suicide locations. This investigation aims to evaluate the short- and long-term effect of safety barriers on Brisbane's Gateway Bridge and to examine whether there was substitution of suicide location.
Methods: Data on suicide by jumping – between 1990 and 2012, in Brisbane, Australia – were obtained from the Queensland Suicide Register. The effects of barrier installation at the Gateway Bridge were assessed through a natural experiment setting. Descriptive and Poisson regression analyses were used.
Results: Of the 277 suicides by jumping in Brisbane that were identified, almost half (n=126) occurred from the Gateway or Story Bridges. After the installation of barriers on the Gateway Bridge, in 1993, the number of suicides from this site dropped 53.0% in the period 1994–1997 (p=0.041) and a further reduction was found in subsequent years. Analyses confirmed that there was no evidence of displacement to a neighbouring suicide hotspot (Story Bridge) or other locations.
Conclusions: The safety barriers were effective in preventing suicide from the Gateway Bridge, and no evidence of substitution of location was found.
Restricting access to lethal means for suicide is a well-established community-level prevention strategy. Published studies have widely demonstrated the effect of this in reducing numbers of self-inflicted deaths occurring by jumping from high places1–4 or railway station platforms,5,6 use of firearms7–11 and poisoning using domestic gases12,13 or drugs.14–16 In a recent meta-analysis of 11 intervention studies conducted in eight ‘hot-spot’ sites for suicide by jumping around the world, installation of barriers and safety nets (structural interventions) were found to have a prominent effect on reducing the number of suicides by jumping in the areas studied.
Whether barrier installation to prohibit access to suicide at a hotspot will cause substitution to nearby sites has been a challenge to the effectiveness of this prevention strategy and may partly reduce the net gain of the intervention.17 Due to environmental constraints, the choice of a relevant control site, which should be a notorious hotspot of suicide within the same agglomeration, has been a major challenge for most intervention studies. As a comparable suicide hotspot is not necessarily available in every city/agglomeration; there has been a wide variation in the choice of controls for previous intervention studies on suicide hotspots. With very few exceptions,18 most intervention studies have used “all other jumping sites” as a control site to examine the effects of barrier installation at suicide hotspots.3,18–20 However, as highlighted by Cox et al,21 motives and epidemiological profiles of those who jump from locations that have acquired notoriety, and consequently also acquired some symbolic meaning, may not be comparable with those who jump from other locations (e.g. residential buildings), which may affect the validity of such analyses.
Additionally, studies have only assessed the immediate effect of barrier installation by comparing the number of suicides between two separate periods through before-and-after2,3,18–20,22,23 or reversal (after-before-after) analysis.1,24 There was a study exploring the long-term effectiveness of installing a safety fence at the Memorial Bridge in Augusta, Maine, in the US, which indicated there was no substitution effect at other jumping sites in a 12-year follow-up period.23 However, relatively little is known about the long-term persistence of the effect of setting up a bridge safety fence in a suicide hotspot and the influence of epidemiological changes in other suicide locations over a longer follow-up period. Another area of research that has to date received very little attention is the distance travelled between one's place of residence and the location of suicide. To the authors’ knowledge, the only study performed on this relevant aspect of suicide from hotspots to date was by Seiden and Spence,25 who found that 84.7% of the persons who jumped from Golden Gate Bridge in San Francisco between 1937 and 1979 in fact drove past another bridge (Bay Bridge) of comparable height and lethality on the way there, demonstrating the appeal of the chosen suicide location over any other.
Brisbane – the state capital of Queensland – is one of the few cities with two bridges that were identified as suicide hotspots during the 1990s (i.e. the Gateway Bridge and the Story Bridge).26 In response to an emerging number of jumping suicides at the Gateway Bridge following its opening in 1986, safety fencing was installed in 1993.27 This provided an opportunity to examine the effectiveness of barriers in reducing the risk of suicide while monitoring for a potential shifting of people attempting suicide at another suicide hotspot located close by. This unique natural experiment offers a methodological approach that has not been previously applied in the international literature.
The aims of our study were threefold: (1) to measure the immediate effectiveness of installing a safety fence to prevent suicides by jumping from the Gateway Bridge; (2) to examine whether there was a subsequent increase in suicides by jumping from the Story Bridge or other sites in Brisbane; and (3) to evaluate whether the effect of barriers at the Gateway Bridge on lowering the incidence of suicides by jumping was sustained over a longer period of time.
Data and methods
Figure 1 shows the locations of the two bridges and their surroundings. The Gateway Bridge (officially renamed to the ‘Sir Leo Hielscher Bridges’ in May 2010) is a pair of 65-metre-high bridges over the Brisbane River, linking an industrial area on the north and a suburban area on the south of Brisbane.27 The original bridge became a suicide hotspot when an audience of 200,000 witnessed a jumping suicide from the bridge during the opening ceremony in January 1986.26 Previous studies also indicated that most suicide attempters at the Gateway Bridge tended to drive and parked their cars at the apex of the bridge before jumping, since its location is not within convenient walking distance for most Brisbane residents.27 After numerous suicides involving people jumping off the bridge in the early 1990s, a pair of fencing barriers about 3.3 metres high was retrofitted along the sidewalk of the bridge in 1993.27,28 After the new duplication bridge was built in May 2010, the barrier was replaced with a similar one with a height of 3.6 metres on the original bridge at the end of November 2010.28
The Story Bridge, opened in 1935, is a 74-metre-high cantilever bridge that crosses the Brisbane River from Fortitude Valley on the north to Kangaroo Point on the south. It lies 12 km upstream from the Gateway Bridge and has three lanes of vehicular traffic each way and a pedestrian walkway on each side.26,27 Previous studies indicated that those who committed suicide at the Story Bridge tended to walk to the bridge, as it is in the centre of Brisbane city and within walking distance for most residents.27 Although it has been a notorious hotspot for suicide for decades,26 to date no physical barriers have been installed on the Story Bridge and it could therefore be used as the control-comparison site in the present analysis.
The present analysis covers the 23 years from 1990 to 2012. Suicide data were retrieved from the Queensland Suicide Register (QSR), an independent and comprehensive mortality database that includes information on all identified suicides by Queensland residents from 1990 onwards.29
Inclusion criteria were the location of suicide (Greater Brisbane Region or Statistical Area Level 4: 301–305, as specified by the 2011 Australian Statistical Geography Standard)30 and cause of death being either jumping from high place (ICD-10: X80) or drowning (ICD-10: X71).31 Also, since Queensland is the second-largest state in Australia, with an area of 1.85 million km2, it is unlikely that the installation of barrier on a bridge in Brisbane would influence the epidemiology of jumping suicides across the whole state.
Suicide rates at each jumping location were calculated using the population of Brisbane, which was retrieved from the Australian Bureau of Statistics. To examine the effect of barrier installation for preventing suicide by jumping at the Gateway Bridge, descriptive analyses and Poisson regression analyses were applied. The latter examined the statistical significance of any change in numbers of suicides by jumping stratified from each jumping location (i.e. the Gateway Bridge, the Story Bridge, other bridges and other jumping sites in the Brisbane area) before (1990–1993) and after (1994–2012) the installation of the barriers on the Gateway Bridge.
We only focused on suicides by jumping (instead of suicides by any method) in Brisbane in the years following the installation of barriers on the Gateway Bridge because jumping from heights is a relatively rare method (accounting for, on average, 3.5% of all suicide deaths in Queensland 1990–2010; QSR, unpublished data) and changes in the annual number of these deaths were not likely to provide sufficient statistical power to affect the trends of suicides by all methods in Queensland. A preliminary analysis based on yearly data was compiled and we found it will be less likely to detect any major changes in suicide epidemiology over the period (see the supplementary file). Therefore, incidence of suicide cases was grouped into four-year periods (1990–1993; 1994–1997; 1998–2001; 2002–2005; 2006–2009) and the three years 2010–2012. The Poisson regression equation is written as follows:
The average relative change of suicide risk at the Gateway Bridge between the pre-installation and post-installation periods was estimated by taking the exponential form (e) of the slope coefficient (β) from the Poisson regression model with the following equation:
The installation of barriers would be considered effective for the immediate prevention of suicide if there were a significant reduction in the number of deaths from the Gateway Bridge during the first four-year period after the barrier installation (1994–1997) when compared to the pre-installation period (1990–1993). A long-term persistence of the barrier's effectiveness would be confirmed by a significantly lower incidence of suicides by jumping from the Gateway Bridge in all the years following the installation (1994–2012).
Concurrently, to assess whether the barrier would lead to a shift of people dying by suicide from other jumping sites in Brisbane, we separately fitted the described model to the Story Bridge, other bridges in Brisbane and other jumping sites in Brisbane. If there was no displacement effect, there should be no significant increase in other jumping sites in Brisbane after the barrier installation, when compared with the pre-installation period.
In all analyses, a p-value smaller than 0.05 was considered to be statistically significant.
A total of 277 suicides by jumping from a high place were identified for the period 1990 to 2012. The method of jumping from a height accounted for 5.2% of all suicides by residents of Brisbane (n=5,232). Of those, 146 (45.5%) occurred from bridges in Brisbane, most commonly from the Gateway Bridge (n=38) and the Story Bridge (n=88). Apart from jumping suicides, there were seven suicide cases by other methods (five by hanging, one by car exhaust poisoning and one by crashing of motor vehicle) that occurred at the bridges in Brisbane (three at the Story Bridge and the remaining four at other bridges in Brisbane) and were excluded from the study.
Table 1 shows the pattern of suicide by jumping at different sites in Brisbane from 1990 to 2012. There has been an apparent and significant reduction in the number of jumping suicides at the Gateway Bridge since the barrier installation in 1993. The overall incidence of suicide reduced by 87.5% (p<0.001) following the fencing and the change did not appear to cause displacement to other locations of suicide by jumping across Brisbane during the same period at the Story Bridge (percentage change=−16.7%, p=0.520), other bridges (percentage change=−60.0%, p=0.060), or other jumping sites (percentage change=+55.4%, p=0.131). Overall, a 32.4% reduction of suicides by jumping was detected in Brisbane (p=0.009). However, this did not pose a significant impact on all suicides for residents in Brisbane (percentage change=+1.2%, p=0.758).
Table 1. Suicides by jumping from a high place and percentage change in suicide with respect to the level during the pre-installation period (1990–1993) from different sites in Brisbane, 1990–2012.
Further analysis of the changes of suicide rate over time shows that the incidence of suicide at the Gateway Bridge was only reduced by 53.0% (p=0.041) during the first 4-year period following the fencing (1994–1997). A further reduction in risk of suicide at the Gateway Bridge was found in all subsequent periods after 1998, with only five suicide cases reported at the Gateway Bridge from 1998 to 2009 and none since 2010. This also resulted in a significantly lower rate of all suicides by jumping in Brisbane in the period 1998–2001 (relative change=−48.2%, p=0.003) and the period 2006–2009 (relative change=−54.6%, p=0.000) when compared to the pre-installation level. In contrast, changes in all jumping suicides were found as not significant in 2002–2005 (relative change=−16.2%, p=0.346) due to the elevated incidence at other jumping sites (relative change=+111.7%, p=0.021). Similarly, such changes were also found to be not significant in 2010–2012 (relative change=−16.2%, p=0.364) due to an elevated incidence at the Story Bridge (relative change=+37.7%, p=0.340) and at other jumping sites (relative change=95.0%, p=0.048).
The present study adds valuable knowledge to the discussion on restricting access to lethal means for suicide with a unique methodological design that enabled comparisons of incidence of suicide by jumping from two adjacent bridges, identified as concurrent suicide hotspots in one city, following the installation of safety barriers on one of them.
In line with the meta-analysis conducted by Pirkis et al.,17 this study provides empirical support that the barriers constructed at the Gateway Bridge were effective in preventing suicides by jumping with no immediate signs of displacement to another neighbouring suicide hotspot (e.g. the Story Bridge) or other locations. This indicated that suicidal individuals generally did not seek alternative sites during the follow-up period of 19 years (from 1994 to 2012). Furthermore, the installation of higher barriers after renovation of the bridge in 2010 may have completely dissuaded people from considering suicide at that location.
This study also demonstrated the importance of examining the long-term effects of barrier installation at the hotspot. As referred to our findings, the suicide rate at the Gateway Bridge only reduced by 53.0% during the first four-year period after the barrier installation, which was considerably lower than the overall reduction of 87.5%. This indicated that the immediate effect of barrier installation is not sufficient to reflect its true impact at a suicide hotspot and a longer follow-up period is needed for reporting in future. This phenomenon may suggest some suicidal individuals would have developed a determined plan to commit suicide at that particular location and the presence of barrier may not be able to prevent all of them from jumping at that location. There has been no specific investigation on this hypothesis, which would deserve further attention in future research.
Although jumping from a high place is not a very common method of ending one's life, a comparatively higher percentage of suicide deaths by this method for Brisbane residents occurred over the study period (5.3%) when compared with the overall figure in Queensland (3.5%). An increasing trend of suicide by jumping at other jumping sites has also been noted since early 2000s. This observation is likely a reflection of the influence of suicide hotspots on the patterns and profile of suicide mortality in Brisbane, as results showed that almost half of these suicides occurred from these two bridges alone (the Gateway and Story Bridges). Suicide prevention strategies require regular evaluation and adjustment. Although barrier installation at the Gateway Bridge has effectively reduced the suicide incidence at that site, and from current data there is no evidence of a subsequent substitution effect, constant revision of suicide prevention strategy may be required given the changing epidemiology of suicide.
The findings of this study should be interpreted in light of its limitations. First, as suicide by jumping only constituted a small proportion of cases in Brisbane, it is difficult (if not impossible) to estimate how the physical barriers at the Gateway Bridge have affected the total number of suicides in the area, i.e., whether being prevented from jumping from the chosen bridge would lead not only to a substitution of location, but also of suicide method. Some studies have pointed out that suicidal individuals do not easily displace to an alternative lethal means of suicide;15,32–35 method substitution remains one of the major concerns that should not be ignored in future research. Second, reliable data for non-fatal suicidal acts incidence were not available, which could underestimate the actual effect of barriers on preventing suicide. Third, apart from the lack of a barrier, our data did not contain any information about the structural weak points of the original Gateway Bridge in 1986 for people to commit suicide. Fourth, it would have been ideal to have an analysis that starts from the establishment of the bridge in 1986. The present analysis was based on the Queensland Suicide Register that started to collect suicide records in Queensland in 1990. Therefore, data prior to 1990 are not available unless other sources are used. According to the work by Cantor and Hill,26 10 jumping suicides occurred at the Gateway Bridge and two suicides at the Story Bridge in the first 17 months (i.e. from January 1986 to May 1987) after the opening of the Gateway. However, this remains insufficient to fully cover the gap before 1990, while no information on jumping suicide in Brisbane is available for the remaining period from June 1987 to December 1989. Finally, we have not considered whether the installation of suicide barriers would bring a net gain to society from an economic perspective.5,6 The installation of new barriers at the Gateway Bridge in 2010 incurred a direct cost of $2.2 million29 and further investigations are required, not only to evaluate their cost-effectiveness in terms of the number of lives saved, but also to be used as a benchmark in assisting policymakers when endorsing similar interventions at sites that continue to attract heightened numbers of people at risk of suicide.
We acknowledge financial support from the Australian Commonwealth Government, Department of Health. The Queensland Suicide Register is funded by Queensland Mental Health Commission. The authors thank Dr Delaney Skerrett, who assisted with the preparation and proofreading of the manuscript.
Additional supprting information may be found in the online version of this article:
Table 1: Suicides by jumping from a high place and percentage change in suicide with respect to the level during the pre-installation period (1990–1993) from different sites in Brisbane, 1990–2012.