Installation of a Bridge Barrier as a Suicide Prevention Strategy in Montréal, Québec, Canada

Abstract

Objectives. We investigated whether the installation of a suicide prevention barrier on Jacques-Cartier Bridge led to displacement of suicides to other jumping sites on Montréal Island and Montérégie, Québec, the 2 regions it connects.

Methods. Suicides on Montréal Island and Montérégie were extracted from chief coroners’ records. We used Poisson regression to assess changes in annual suicide rates by jumping from Jacques-Cartier Bridge and from other bridges and other sites and by other methods before (1990–June 2004) and after (2005–2009) installation of the barrier.

Results. Suicide rates by jumping from Jacques-Cartier Bridge decreased after installation of the barrier (incidence rate ratio [IRR] = 0.24; 95% confidence interval [CI] = 0.13, 0.43), which persisted when all bridges (IRR = 0.39; 95% CI = 0.27, 0.55) and all jumping sites (IRR = 0.66; 95% CI = 0.54, 0.80) in the regions were considered.

Conclusions. Little or no displacement to other jumping sites may occur after installation of a barrier at an iconic site such as Jacques-Cartier Bridge. A barrier’s design is important to its effectiveness and should be considered for new bridges with the potential to become symbolic suicide sites.

Physical availability and sociocultural acceptability are important considerations in the choice of method of suicide.¹ Restricting access to commonly used methods of suicide is widely recognized as a suicide prevention strategy. Several studies have indicated that detoxification of domestic gas; mandatory use of catalytic converters in motor vehicles; restrictions on pesticides, barbiturates, and analgesics; use of lower toxicity antidepressants; firearm control legislation; and construction of barriers at jumping sites have been effective in reducing suicides by those methods.¹ However, evidence for the success of some of these strategies remains equivocal (e.g., use of catalytic converters in Australia,² reduction of paracetamol pack size in the United Kingdom^3,4). Furthermore, restricting one method can result in substitution with another, although substitution may depend on the popularity of the method and the availability of alternative methods that are acceptable to the individual.^5–7 A change in overall suicide rates may be obscured if method substitution occurs or if the restricted suicide method accounts for a relatively small proportion of all suicides.^1,8

For suicide by jumping, displacement to other jumping sites is probably more likely than a change in method. Two studies found no shift to other jumping sites after installation of barriers, but they did not examine the effect on overall suicide rates.^9,10 By contrast, other studies have shown evidence of displacement, with overall suicide rates remaining unchanged.^11–13 In 1 of these latter studies, jumping suicides from other bridges and buildings in Toronto increased after the construction of a suicide barrier at Bloor Street Viaduct.¹³ The researchers suggested that the viaduct was not a uniquely attractive location for suicide and was therefore interchangeable with other sites. By contrast, some jumping sites are reportedly the only site a suicidal individual would consider (e.g., Golden Gate Bridge, Eiffel Tower, Empire State Building); their status as suicide magnets is enhanced by the ease of access, perceived lethality of the jump, notoriety as a popular suicide site, romantic view of death they encourage, media attention, and unique features such as being over water.^10,13–16

All of these factors existed for Jacques-Cartier Bridge, which spans the St. Lawrence River between Montréal Island and Montérégie, Québec, with an average of 10 suicides annually before the construction of a barrier in 2004 (P-A Perron, unpublished data, 2002). To create the barrier, the existing 1.1-meter steel palisade fencing was extended a further 1.4 meters and curved inwardly at the top, making it high and difficult to climb (see image available as a supplement to the online version of this article at http://www.ajph.org). An initiative to have a barrier installed was unsuccessful in 1996 and almost halted in 2002 because of the argument that a barrier would not decrease suicides but merely displace them to other sites. To test the validity of this argument, we assessed whether displacement to other jumping sites on Montréal Island and Montérégie occurred.

METHODS

For this natural experiment, we extracted suicide deaths among Québec residents from the data banks of the Québec chief coroner’s office for January 1, 1990–December 31, 2009. These data banks have information on all suicides that occur in the province.

Data and Variables

The date, site, region, and method (e.g., jumping from height, hanging, firearm) of suicide and the victims’ age, gender, and region of residence were recorded. Because the antisuicide barrier on Jacques-Cartier Bridge was under construction from July to December 2004, we excluded suicides occurring during this period (n = 593). When the date of death was missing, we imputed the date of the discovery of the body (n = 92). We classified the 13 years from 1990 to 2003 and the first 6 months of 2004 as being before the barrier and the 5 years from January 1, 2005, to December 31, 2009, as being after the barrier.

Jacques-Cartier Bridge is 1 of several bridges spanning the St. Lawrence River between Montréal Island and the Montérégie region on the south shore (Figure 1). We considered jumping sites within the 2 regions, including bridges connecting them, for potential evidence of displacement. The bridge used for a suicide was usually recorded, but in some cases the body was retrieved from the St. Lawrence River and, although bodily injury indicated a fall from height, the exact bridge used was unknown. We divided suicides into 6 categories: (1) jumps from Jacques-Cartier Bridge, (2) jumps from unknown bridge connecting Montréal Island and Montérégie, (3) jumps from other known bridges connecting Montréal Island and Montérégie, (4) jumps from other bridges on Montréal Island and Montérégie not spanning the St. Lawrence River, (5) jumps from other sites on Montréal Island and Montérégie, and (6) all suicides among Montréal Island and Montérégie residents (for greater detail, see table available as a supplement to the online version of this article at http://www.ajph.org).

FIGURE 1—

Bridges crossing the St. Lawrence River between Montréal Island and the Montérégie region, Québec.

Note. The inset map shows the boundaries of Montréal Island and Montérégie.

Source. CanMap RouteLogistics. Markham, Ontario: DMTI Spatial; 2009.

We derived population counts for Québec and Montréal Island and Montérégie from census data for 1991, 1996, 2001, and 2006 and assumed linear population growth for the periods 1991–1996, 1996–2001, and 2001–2006. We estimated population growth by extrapolating backward for 1990 to 1991 and forward for 2006 to 2009.¹⁷

Analyses

The study had a before-and-after design. The unit of analysis was suicide rates, and the intervention was the installation of the barrier. Because there was no evidence of overdispersion, we used Poisson regression analysis to compare annual suicide rates before and after the installation of the barrier on Jacques-Cartier Bridge. Incidence rate ratios (IRRs) were calculated and represent the ratio of the incidence rate before and after the installation of the barrier. To assess possible displacement, we ran regression models for suicides from Jacques-Cartier Bridge only and then combined them in a stepwise fashion with suicides from categories 2 through 6. If displacement occurred, the IRR for all jumping sites (categories 1–5) and all jumping sites excluding Jacques-Cartier Bridge (categories 2–5) would be equivalent.

We also performed a Poisson regression with linear spline comparing suicides from Jacques-Cartier Bridge with suicides from all jumping sites and all suicides on Montréal Island and Montérégie excluding Jacques-Cartier Bridge among Montréal Island and Montérégie residents. We used 2 cutpoints for the spline; the 1st cutpoint was the year with the most suicides by jumps (1996) and the 2nd cutpoint was the year in which the barrier was installed (2004). The model used was as follows:

We used 3 different models for each time period. The IRR represents the yearly change in rates, with an IRR of 1 indicating no temporal variation in suicide rates within the specified time frame. Displacement would be suggested if suicide rates from Jacques-Cartier Bridge significantly decreased after the installation of the barrier and the rate from other jumping sites increased within the same time frame. We calculated rates using the annual population of Montréal Island and Montérégie estimated by the Institut de Statistique du Québec. We also undertook a sensitivity analysis for all those residing in Québec province.

We analyzed demographic data using the 2-tailed, independent-samples t test for continuous variables and the 2-sided χ² test for categorical variables. We considered P < .05 to be statistically significant.

RESULTS

There were 25 781 suicides between 1990 and 2009 in Québec. Exclusion of those occurring between July 1, 2004, and December 31, 2004, resulted in a study sample of 25 188 suicides. Suicide rates for Montréal Island, Montérégie, and the rest of Québec increased from 1990 to 1999 but decreased thereafter (data available as a supplement to the online version of this article at http://www.ajph.org). Of the 8652 suicides on Montréal Island and Montérégie from 1990 through 2009, 652 were jumps, 162 (25%) of which were from Jacques-Cartier Bridge. Jacques-Cartier Bridge was the only single structure that attracted individuals from regions beyond neighboring municipalities (13% traveled from elsewhere in Québec). Before the installation of the barrier, an average of 10 individuals jumped from the bridge annually. Three people per year managed to scale the barrier between 2005 and 2008, with no suicides from the bridge recorded for 2009.

Results of the Poisson regression demonstrated a sharp decrease in suicide rates for Jacques-Cartier Bridge after the installation of the barrier (IRR = 0.24; 95% confidence interval [CI] = 0.13, 0.43; Table 1). In addition, the decrease in the IRR persisted when suicides from other bridges connecting Montréal Island and Montérégie (known and unknown) and from other bridges and jumping sites on Montréal Island and Montérégie were added incrementally. However, some overlap in CIs occurred when we compared suicide from all jumping sites (IRR = 0.66; 95% CI = 0.54, 0.80) with suicide from all jumping sites excluding Jacques-Cartier Bridge (IRR = 0.81; 95% CI = 0.66, 1.00; Table 1), thereby not completely ruling out the potential for some displacement.

TABLE 1—

Poisson Regression Analysis of Annual Suicide Rates by Jumping and Other Means for Residents of Montréal Island and Montérégie, Québec, Before and After the Installation of a Barrier on Jacques-Cartier Bridge

	Suicide Rate/100 000 Individuals		No. of Annual Suicides,a Corrected Mean
Variable	Before Installation	After Installation	Before Installation	After Installation	IRR (95% CI)	P
Jacques-Cartier Bridge (1)	0.324	0.079	10.0	2.6	0.24 (0.13, 0.43)	< .001
(1) + unknown bridges connecting Montréal Island and Montérégie (2)	0.378	0.085	11.7	2.8	0.23 (0.12, 0.39)	< .001
(1) + (2) + other known bridges connecting Montréal Island and Montérégie (3)	0.415	0.128	12.8	4.2	0.31 (0.19, 0.48)	< .001
(1) + (2) + (3) + other bridges on Montréal Island and Montérégie (4)	0.577	0.225	17.9	7.4	0.39 (0.27, 0.55)	< .001
(1) + (2) + (3) + (4) + other jumping sites on Montréal Island and Montérégie (5)	1.168	0.768	36.1	25.0	0.66 (0.54, 0.80)	< .001
(2) + (3) + (4) + (5)	0.844	0.687	26.1	22.5	0.82 (0.66, 1.01)	.055
All suicides (6)	15.563	11.915	467.3	388.6	0.77 (0.73, 0.81)	< .001

Note. CI = confidence interval; IRR = incidence rate ratio. Time period before barrier installation was January 1, 1990–June 30, 2004; time period after barrier installation was January 1, 2005–December 31, 2009. Rates were calculated using estimated populations for Montréal Island and Montérégie. The rates were not standardized for age.

^aCorrected per capita to suicides in 1990 population.

To further test the hypothesis that suicide on Jacques-Cartier Bridge was independent of temporal variation of suicide rates, we performed a linear spline analysis to evaluate the pattern of annual change in suicide rates (Table 2). Suicide from Jacques-Cartier Bridge, other jumping sites, and all suicides followed the same pattern in periods 1 and 2 but diverged strongly in period 3. Indeed, suicide by jumping from other sites was constant in period 3 with a CI overlapping all suicides but decreased for Jacques-Cartier Bridge with a CI not overlapping other jumping sites or all suicides.

TABLE 2—

Poisson Regression Analysis of Suicide Rates for Jacques-Cartier Bridge, Other Jumping Sites, and All Suicides Among Residents of Montréal Island and the Montérégie Region, Québec

Variable	IRR (95% CI)	P
Jacques-Cartier Bridge (category 1)
Period 1 (1990–1996)	1.06 (0.96, 1.16)	.28
Period 2 (1997–2004)	0.93 (0.87, 0.99)	.027
Period 3 (2005–2009)	0.66 (0.48, 0.90)	.009
Other jumping sites (categories 2–5)
Period 1 (1990–1996)	1.05 (0.99, 1.11)	.125
Period 2 (1997–2004)	0.94 (0.91, 0.98)	.002
Period 3 (2005–2009)	1.04 (0.93, 1.15)	.493
All suicides (category 6)
Period 1 (1990–1996)	1.02 (1.01, 1.04)	.001
Period 2 (1997–2004)	0.98 (0.98, 0.99)	< .001
Period 3 (2005–2009)	0.93 (0.91, 0.95)	< .001

Note. CI = confidence interval; IRR = incidence rate ratio. The year 1996 represents the year with the most suicide by jumps and 2005 the first year with barriers installed on Jacques-Cartier Bridge.

Individuals who jumped from Jacques-Cartier Bridge were younger than those who used other sites (Table 3). The average age of suicide victims was higher in the period after installation of the barrier for all sites except Jacques-Cartier Bridge, for which we found a decrease. Most suicides by jumping were males. The proportion of males who committed suicide by jumping changed little after the installation of the barrier, but the proportion decreased for all suicides in Québec. We found no differences in the results when we performed the analysis for all residents from the province of Québec (data not shown).

TABLE 3—

Demographic Characteristics of People Who Committed Suicide on Montréal Island and Montérégie, Québec, Before and After the Installation of a Barrier on Jacques-Cartier Bridge

	Age, Years			Male
Variable	Before Barrier, Mean (SD)	After Barrier, Mean (SD)	P	Before Barrier, No. (%)	After Barrier, No. (%)	P
Jacques-Cartier Bridge (category 1)	34.9 (12.2)	31.8 (12.2)	.388	121 (82.9)	12 (92.3)	.378
Other bridges connecting Montréal Island and Montérégie (category 3)	38.9 (11.2)	44.1 (16.1)	.456	14 (82.4)	6 (85.7)	.841
Bridges on Montréal Island and Montérégie (category 4)	46.4 (18.0)	47.0 (16.6)	.889	72 (80.0)	20 (87.0)	.444
Other jumping sites (category 5)	43.5 (17.8)	47.4 (19.0)	.029	388 (73.8)	97 (77.0)	.487
All suicides (category 6)	41.3 (15.9)	45.7 (15.9)	< .001	15 435 (79.0)	4 326 (77.8)	.041

Note. Time period before barrier installation was January 1, 1990–June 30, 2004; time period after barrier installation was January 1, 2005–December 31, 2009.

DISCUSSION

In line with other studies showing that barriers prevent suicides at the sites at which they are placed,^9–13,18 we found that the construction of the barrier on Jacques-Cartier Bridge was successful in significantly reducing the number of suicides from the bridge. In addition, we found little evidence of displacement to other bridges on Montréal Island and Montérégie. This result resembles those of studies concerning other iconic suicide sites such as the Memorial Bridge between Maine and New Hampshire,⁹ Duke Ellington Bridge in Washington, DC,¹⁹ and the Minster Terrace in Bern, Switzerland,¹⁰ but is unlike those of a study regarding a less symbolic site, the Bloor Street Viaduct in Toronto, Ontario, which had several comparable bridges nearby.¹³ Jacques-Cartier Bridge, with a bicycle path and pedestrian walkway, is more accessible than other bridges connecting Montréal Island and Montérégie. In addition to its accessibility, highly publicized suicides from the bridge by well-known individuals may have contributed to its suicide magnet status. We also observed that individuals were willing to travel long distances to commit suicide on Jacques-Cartier Bridge but this was not observed for the Bloor Street Viaduct.¹³ Thus, the degree of importance that a bridge holds for someone contemplating suicide may be reflected in the distance he or she is willing to travel to its location.¹³

Unfortunately, the barrier was not effective in preventing all suicides. After installation of the barrier, Jacques-Cartier Bridge still represented 10% of all jumping suicides in both regions (or 11% if one adds suicides from unknown bridges) connecting Montréal Island and Montérégie; there was only one further suicide from an unknown bridge site after the installation of the barrier, and we assumed that it was from Jacques-Cartier Bridge. A similar finding was reported after the installation of barriers on the Clifton suspension bridge,⁸ which, together with our results, suggests the need to pay careful attention to the design of barriers and access to the bridge.

In a natural experiment, several variables beyond the researchers’ control may bias results. Variables other than the barrier may possibly have influenced suicide rates at the same time at which the barrier was installed. For example, we observed an overall decrease in rates of suicide by all methods and by jumping in the province of Québec. However, the decreasing trend is unlikely to explain all of our results. Our findings suggest that the reduction in suicides from Jacques-Cartier Bridge, with little evidence of displacement to other sites, had an independent distinct impact on rates of jumping suicides in general. Indeed, suicides from Jacques-Cartier Bridge represented 28% of all jumping suicides on Montréal Island and Montérégie before the installation of the barrier. If one assumes all suicides from unknown bridges connecting Montréal Island and Montérégie were from Jacques-Cartier Bridge, this proportion increases to 32%. With a single structure representing such a large proportion of suicides, it is not surprising that reducing access to it would have a significant impact given little or no displacement. Given the ecological nature of this study, it is not possible to infer findings at the individual level.

Québec suicide records have been shown to be valid and reliable, with little underreporting.²⁰ The decision to ascribe a death to suicide is based on the balance of evidence indicating suicidal intent. Although some unintentional deaths or homicides may be misclassified as suicides after a fall from height, it is unlikely to present a serious bias. Using the date when the body was discovered as the date of death for some cases will have added some inaccuracy but is unlikely to affect annual trends.

In conclusion, our results, together with those of previous studies, suggest that little or no evidence of displacement may be found when a barrier is installed at an iconic site, whereas this may not be true for sites that are not suicide magnets. The specific design of the barrier is important for its efficacy and should be considered in the construction of new bridges with the potential to become symbolic suicide sites, such as the new Champlain Bridge that will be built between Montréal Island and Montérégie. For other sites, an area-wide strategy of removing access to jumping sites may be needed to reduce suicide. However, preventing access to commonly used methods of suicide does not address an individual’s psychological distress, and other interventions, such as depression screening and treatment, should also be undertaken.^1,21