Long-Term Impact of the Bloor Viaduct Suicide Barrier on Suicides in Toronto: A Time-Series Analysis: Effet à long terme de la barrière anti-suicide du viaduc Bloor sur les suicides à Toronto : une analyse chronologique

Abstract

Background

A suicide prevention barrier was installed at Toronto's Bloor Viaduct bridge in 2003. It was associated with short-term location substitution, possibly mediated by media effects that did not persist over 1 decade. The long-term impact of the barrier is unknown.

Methods

We examined rates of suicides by jumping from the Bloor Viaduct, other bridges and by other methods using coroner's records in Toronto (1998–2020). We used interrupted time-series Poisson regression analyses to model changes in quarterly bridge-related suicides after barrier installation. A secondary analysis explored the potential substitution effects of suicide by other methods.

Results

Of 5219 suicides from 1998 to 2020, 303 were by jumping from bridges. After controlling for covariates, installation of the Bloor Viaduct suicide barrier was associated with a 49% step decrease in bridge-related suicide in the next quarter in Toronto (incidence rate ratio [IRR] = 0.51, 95% CI, 0.30 to 0.86) with no rebound increase in bridge-related suicide during the subsequent 17 years after the original drop (IRR = 0.99, 95% CI, 0.96 to 1.03). There was also no associated change in suicides by other methods after the barrier (IRR = 1.04, 95% CI, 0.90 to 1.20).

Conclusions

Contrary to initial findings, these results indicate an enduring suicide prevention effect of the Bloor Viaduct suicide barrier. They support the long-term utility of structural interventions at high-frequency sites for suicide.

Introduction

Restricting access to lethal means is an evidence-based suicide prevention strategy and it is 1 of 4 key population-level prevention strategies encouraged by the World Health Organization's LIVE LIFE implementation guide.^1–7 Structural interventions have long been used to prevent suicide at high-frequency sites for suicide including iconic bridges, buildings, and natural peaks.^8–17 Pirkis et al. ¹⁸ conducted a meta-analysis demonstrating that such interventions tend to decrease suicides at the frequently used locations themselves, with some evidence of location substitution at neighbouring sites but still resulting in substantial net reductions in suicides in an area. This latter finding is of crucial importance in understanding whether barriers truly lower suicide rates across a city or region; the degree to which this occurs in a sustained manner following barrier installation remains an open scientific question. There is a dearth of studies on the long-term effects of barriers at frequently used locations for suicide.

Installation of the Bloor Viaduct suicide barrier initiated a natural experiment at a high-frequency site for suicide in Toronto, Canada that provides a relatively unique opportunity to address the long-term effect of such barriers. Prior to the “Luminous Veil” barrier, installed in 2003, the Bloor Viaduct had the second-highest yearly suicide counts of any bridge in North America after the Golden Gate Bridge in San Francisco.¹⁶ At the turn of the millennium, the 9 suicides occurring at the Bloor Viaduct each year accounted for half of suicides by jumping from bridges and 4% of all suicides in Toronto. ¹⁶ An initial study by our group examining the first 4 years after the barrier indicated that substantial location substitution appeared to be occurring with a significant rise in suicides at other bridges that diminished the apparent effect of the barrier on overall suicides. ¹⁶ However, a follow-up study examining a decade of data after the barrier's creation found that this increase in suicides at other bridges was a transient observation, as the substitution effect no longer existed in later years; the finding that substitution appeared to be present in the first few years, but not subsequent years may have resulted from media attention garnered by the barrier during and in the aftermath of its construction. ¹⁵

Now 2 decades after the barrier's completion, we seek an answer as to whether the barrier led to a longer-term reduction of suicides by jumping from any bridges in Toronto, and whether there was a substitution effect of suicide by other methods. Our a priori hypothesis is that, since its installation, the barrier reduced suicide by jumping from bridges in Toronto, with no substitution effect, and was thus an effective strategy for preventing suicide.

Methods

Data Sources

This study adopted a retrospective observational study design. The outcomes of interest were suicides by jumping from bridges and by other methods in the City of Toronto (1998–2020). Records of suicide deaths were available from the Office of the Chief Coroner of Ontario in the context of a larger ongoing study by our team. ¹⁹ These were used to calculate quarterly counts of suicide deaths by jumping from bridges and by other methods in Toronto between April 1998 and December 2020 (the most recent year for which complete data was available at the time of analysis). The monthly unemployment rate in Ontario ²⁰ and the consumer price index (CPI) in Toronto ²¹ between 1998 and 2020 were retrieved from Statistics Canada. The population of Toronto was obtained from the Canadian Census of Population 1996–2021,^22–27 which was conducted every 5 years. Linear growth within each 5-year period was assumed to interpolate the quarterly population in Toronto.

Statistical Analysis

Interrupted time-series analysis using Poisson regression was performed to investigate the association between completion of the Luminous Veil on the Bloor Viaduct and quarterly rates of suicide by jumping from bridges in Toronto between 1998 and 2020. The dependent variable was quarterly counts of suicides by jumping from bridges. Construction of the Luminous Veil spanned April 2002 to June 2003. ¹⁶ The categorical independent variable was thus defined as the period of time before the Luminous Veil was completed (June 2003 and before) compared with a period of time after it was completed (July 2003 and after).

A “time since intervention” variable (time × the construction of Luminous Veil) was input into the model to indicate the time passed after the installation of the barrier. The quarterly count of suicide cases by jumping from bridges in the previous quarter was input into the model to account for the temporal autocorrelation of the outcome variable. Other potential confounders were added to the model, including the quarterly count of suicides by other methods in the same quarter, quarterly unemployment rate in Ontario and CPI in Toronto and seasonal variation in suicide counts. The log-term of Toronto's population in the corresponding quarter was included in the model as the offset variable.

A Poisson regression was also used to test for a method substitution effect. The outcome was quarterly counts of suicide by methods other than jumping from bridges. The model included the intervention, the interaction term, the linear trend, the quarterly count of suicide cases by other methods in the previous quarter (lagged term), the quarterly unemployment rate, CPI, seasonal variation as well as the log-term of Toronto's population in the corresponding quarter. The overdispersion assumption for Poisson regressions and the autocorrelation assumption for time-series analysis were assessed.

One of the limitations of the time-series design is that, if only focused on a single location (in this case Toronto), it would not detect the impact of confounders on suicide rates across the entire region which may influence results. To address this problem, we replicated the same analysis exploring the association between the construction of the Luminous Veil on the Bloor Viaduct and quarterly rates of suicide by jumping from heights in the second and third largest metropolitan areas in the province of Ontario (Ottawa and Hamilton) between 1998 and 2020. The construction of the Luminous Veil should have had no impact on suicides in these cities and therefore they act as a control. Similar variables were used following the primary model, including the intervention, the interaction term, the linear trend, the quarterly count of suicide cases by jumping from height in the previous quarter (lagged term), the quarterly unemployment rate in Ontario, seasonal variation, and the log-term of Ottawa and Hamilton's population in the corresponding quarter. The populations of Ottawa and Hamilton were also obtained from the Canadian Census of Population 1996–2021,^22–27 and the quarterly population was calculated following the strategy used for the Toronto population.

Results are presented as incidence rate ratios (IRRs) and 95% confidence intervals (CIs), and the significance level was set at 2-sided p < 0.05. Analyses were conducted using R, version 4.2.1 (R Foundation for Statistical Computing, Vienna, Austria). This study was approved by the Sunnybrook Health Sciences Research Ethics Board (#1980).

Results

There were 5219 suicide deaths identified across the study period, among which 303 deaths were from jumping from a bridge. Only 2 suicides have occurred at the Bloor Viaduct during the period since the Luminous Veil was installed (July 2003 to December 2020) compared to 48 (April 1998 to June 2003) indicating that the barrier was highly effective in reducing suicide from that bridge. Interrupted time-series analysis results are shown in Table 1 and Figure 1. Time trends were not detected in bridge-related suicide during the preinstallation period (April 1998 to June 2003); that is, counts of suicide by jumping from bridges were stable during the years immediately prior to installation of the Luminous Veil.

Table 1.

The Association of the Construction of the Luminous Veil with Bridge-Related Suicide Rates and Method Substitution Effects in Toronto, 1998–2020.

	Crude model Incident rate ratio (95% CI)	Adjusted model Incident rate ratio (95% CI)	Substitution effect Incident rate ratio (95% CI)
Intervention-barrier on Bloor Viaduct	0.47 (0.30 to 0.75) b	0.51 (0.30 to 0.86) c	1.04 (0.90–1.20)
Time (quarter)	1.00 (0.97 to 1.04)	0.99 (0.88 to 1.12)	1.02 (0.99–1.05)
Intervention × time	1.00 (0.97 to 1.03)	0.99 (0.96 to 1.03)	1.00 (0.99–1.01)
Counts of bridge suicides in the last quarter		1.04 (0.97 to 1.11)	NA
Counts of suicide cases by other methods		1.01 (1.00 to 1.03)	NA
Counts of suicide cases by other methods in the last quarter		NA	1.00 (1.00–1.00)
Unemployment rate		0.98 (0.85 to 1.12)	0.97 (0.94–1.00)
Consumer price index		1.02 (0.85 to 1.22)	0.97 (0.93–1.01)
Seasonal effect
Quarter 1 (Jan to Mar)		Reference	Reference
Quarter 2 (Apr to Jun)		1.23 (0.84 to 1.81)	1.15 (1.06–1.25) b
Quarter 3 (Jul to Sep)		1.28 (0.89 to 1.86)	1.16 (1.06–1.27) a
Quarter 4 (Oct to Dec)		1.59 (1.14 to 2.24) b	0.97 (0.88–1.05)

Note. ^ap-value <0.001; ^bp-value <0.01; ^cp-value <0.05. NA = not applicable.

Figure 1.

The relationship between installation of the Bloor Viaduct suicide barrier (red line) and bridge-related suicide rates in Toronto, 1998–2020.

After controlling for covariates, the construction of the Luminous Veil was associated with a 49% step decrease in bridge-related suicide in the next quarter in Toronto (IRR 0.51, 95% CI, 0.30 to 0.86). The postintervention time trend indicated that there was no statistically significant rebound in bridge-related suicide after the original drop (IRR 0.99; 95% CI, 0.96 to 1.03); that is, the observed reduction persisted over time. We did identify a seasonal effect, with the suicide rate from October to December being 60% higher than that from January to March (IRR 1.59; 95% CI, 1.14 to 2.24). The model assumptions of autocorrelation (p = 0.86) and overdispersion (p = 0.60) were not violated.

There was no evidence of method substitution as indicated by stable counts of suicide by methods other than jumping from bridges right after the construction of the Luminous Veil (IRR 1.04; 95% CI, 0.90 to 1.20). Furthermore, there was also no long-term change in suicide by other methods after the intervention (IRR 1.00; 95% CI, 0.99 to 1.01). The control group comparison also identified that the construction of the Luminous Veil did not have an impact on the quarterly rate of suicide by jumping from height in Ottawa and Hamilton although there was a trend in the direction of a reduction in suicides by jumping from height following the barrier in 1 of the 2 control cities (Ottawa: IRR 0.50; 95% CI, 0.26 to 1.01) (Table 2).

Table 2.

The Association of the Construction of the Luminous Veil With Jumping-Related Suicide Rates in Ottawa and Hamilton, 1998–2020.

	Ottawa Incident rate ratio (95% CI)	Hamilton Incident rate ratio (95% CI)
Intervention-barrier on Bloor Viaduct	0.50 (0.26 to 1.01)	1.17 (0.44 to 3.43)
Time (quarter)	1.03 (0.98 to 1.09)	0.95 (0.89 to 1.02)
Intervention × time	0.98 (0.93 to 1.04)	1.06 (0.99 to 1.14)
Counts of jumping suicides in the last quarter	0.94 (0.85 to 1.03)	0.98 (0.85 to 1.12)
Counts of suicide cases by other methods	0.99 (0.96 to 1.03)	0.95 (0.90 to 1.00)
Counts of suicide cases by other methods in the last quarter	0.99 (0.95 to 1.02)	0.99 (0.95.1.04)
Unemployment rate	1.01 (0.90 to 1.13)	1.05 (0.91 to 1.19)
Seasonal effect
Quarter 1 (Jan to Mar)	Reference	Reference
Quarter 2 (Apr to Jun)	1.17 (0.79 to 1.74)	1.13 (0.67 to 1.90)
Quarter 3 (Jul to Sep)	1.26 (0.86 to 1.87)	1.42 (0.87 to 2.34)
Quarter 4 (Oct to Dec)	0.89 (0.58 to 1.36)	0.98 (0.57 to 1.68)

Note. ^ap-value <0.001; ^bp-value <0.01; ^cp-value <0.05.

Discussion

In this retrospective observational study, we found that the installation of the Luminous Veil barrier at the Bloor Viaduct was associated with a reduction in suicides by jumping from bridges in Toronto that has persisted over nearly 2 decades. Half of all suicides by jumping from bridges in Toronto prior to the barrier occurred at the Bloor Viaduct.^15,16 In this context, our findings of a long-term ∼50% drop in suicides by this method do not indicate any method and location substitution. Our results here concur with the findings of our earlier follow-up study examining the first decade of postbarrier data showing that suicides from bridges in Toronto had decreased with no location substitution, ¹⁵ albeit now confirming that finding over nearly 2 decades. Specifically, our results are consistent with the notion that a high proportion of suicides that might otherwise have occurred at the Bloor Viaduct were likely truly prevented. This is supported by our further finding that there was no evidence of method substitution (i.e., no increase in deaths by other suicide methods in Toronto) both immediately and in the long term. Taken together, the nearly 9 yearly suicides potentially prevented over 17.5 years, would equate to an estimate of more than 150 lives saved.

These results are in contrast to our group's initial findings which did identify some evidence for location substitution when examining yearly suicides in the first 4 years after the barrier was installed. ¹⁶ In retrospect, that earlier analysis may have been confounded by short-term data fluctuations (e.g., a transient media effect resulted in a small number of increased suicides at neighbouring bridges) and limited statistical power as a result of a small sample size and relatively few data points. At the time, we had justified conducting that preliminary analysis given our hypothesis that the maximum impact of a barrier should be apparent immediately after installation, ¹⁶ however, results of our current study underscore the need to examine more longitudinal data in studies of this kind.

Our study is an ecological, natural experiment in which we cannot say with absolute confidence that the Luminous Veil prevented suicides in Toronto. However, the strength, specificity, temporality, and plausibility of our findings strongly align with the Bradford Hill criteria for causation. ²⁸ They are also in line with the other long-term international studies of the impact of a barrier on a bridge. Berman et al. ⁹ recently examined suicide trends in the 30 years following the installation of a suicide prevention barrier at the Duke Ellington Bridge in Washington DC. The results of that study are essentially identical to ours—the barrier prevented suicide at the Duke Ellington bridge with no evidence of location or method substitution. ⁹ A limitation of that study is the relatively small number of suicides occurring at the Duke Ellington bridge prior to the barrier (2.8 per year) ⁹ which was less than one-third of the prebarrier yearly rate at the Bloor Viaduct. Another study investigated the long-term impact of installing a safety barrier on the West Gate Bridge in Victoria, Australia. ²⁹ Similar to what was observed historically at the Bloor Street Viaduct, suicide by jumping was a relatively frequent occurrence at the West Gate Bridge, accounting for approximately 40% of such incidents in the region. The results indicated that after the installation of the barrier, the rate of suicide by jumping at the West Gate Bridge dropped to zero. Additionally, accounting for population growth and pre-existing time trends, suicide by jumping at other bridges overall appeared reduced compared to what would have been expected with no intervention, with no indication of excess suicides at nonbridge locations. ²⁹ The fact that the current study, at a bridge with a higher base rate of suicides and in a vastly different geographic location, yielded similar results confirms the potential capacity for such barriers to represent a broadly effective suicide prevention intervention. Taken together, the results of these studies of barriers in different countries appear to be strong evidence that suicide barriers are effective long-term suicide prevention strategies.

These findings also offer potentially important insights into the psychological pathway of transition from suicidal thoughts to high-lethality suicide attempts. The integrated motivational–volitional model of suicidal behaviour presents access to the means of suicide and mental imagery of being dead or dying as key volitional moderators of the transition from thoughts to behaviour. ³⁰ With respect to suicides from iconic bridges, this can be understood in the context of evidence that frequently used locations for suicide differ from other locations. In a study of 6 suicide survivors from the Golden Gate Bridge from the 1970s, 4 of the 6 reported that they would not have attempted suicide by jumping from any other bridge. ³¹ Another study by Seiden and Spence ³² examining suicides at the Golden Gate Bridge and the San Francisco-Oakland Bay Bridge (a less iconic bridge of similar height and constructed around the same time as the Golden Gate Bridge) found dozens of cases of people driving over the Bay Bridge to die by suicide at the Golden Gate Bridge but no cases of the reverse phenomenon. These findings in San Francisco along with our findings and those in Washington DC underscore that high-frequency sites for suicide should, at least in some respect, be considered distinct suicide methods that likely have their own sway and associated mental imagery for individuals. Our study is impactful by identifying an actionable way in which suicides can be prevented in cities and regions with iconic suicide locations.

Our study has a few important limitations. As already mentioned, it is an uncontrolled natural experiment and, as such, factors other than the Bloor Viaduct suicide barrier and/or covariates in the analysis may have been responsible for observed changes in suicide rates. The demographics of the population of Toronto, the culture in the city, and access to mental health resources have evolved over time and could have impacted findings. We also relied on coroner investigations and cannot rule out that a small number of suicide deaths in Toronto were not identified or misclassified. Lastly, while not meeting the prespecified threshold for significance, there did appear to be a trend in the direction of a reduction in suicides by jumping from height following the barrier in 1 of the 2 control cities. Therefore the possibility that other confounders may have influenced results remains indicating the need for a more longitudinal study of the impact of suicide prevention barriers worldwide.

In this long-term study of the Bloor Viaduct suicide barrier natural experiment, we found that the barrier prevented suicides at the Bloor Viaduct, reduced suicides from bridges in Toronto by half, and was not associated with any evidence of location or method substitution. Given the limited number of high-frequency sites for suicide worldwide with sufficient suicide counts for robust analysis, these findings are relatively unique. Ultimately, it will be important to contextualize these findings with future outcomes of the suicide barrier which recently was completed at the Golden Gate Bridge. In the meantime, our results affirm the utility of means restriction strategies and governments and policy stakeholders should take note and strongly consider structural barriers at frequently used locations for suicide.