Do Fences Work?
Research Papers
We believe the advisors to Cornell University about the fences have overstated the results of the studies to Cornell University, to the Mayor and to Common Council. In fact, the most recent study on the effectiveness of suicide barriers (the Toronto study) indicates that people found alternative locations to end their life and the overall suicide rate was unaffected.
There are a range of other options that have not been considered because the Cornell advisors have overstated the effectiveness of suicide barriers and means restriction. Among those options that have not been considered: signs and telephone hotline, suicide patrols, improved lighting/visibility at particular sites, training for Cornell University staff to identify individuals in distress, mentoring system for students to identify peers in distress, and doing nothing.
The key question about the fences is: Do they work?
A national survey found that "34% of respondents believed that every single jumper would have found another way to complete suicide and an additional 40% believed that most would have completed suicide using other means" (Miller et al., 2007).
Contrary to these numbers, the Cornell University advisors who assisted in the decision to install bridge barriers insist that: "A small number of studies have formally evaluated the impact of installing barriers at suicide sites. All studies show barriers are effective in reducing suicides from that site, without displacement to neighboring sites, and sometimes accompanied by a reduction in suicides by jumping in the surrounding region." (Beautrais et al., 2010).
This statement is not true. All studies DO NOT agree about the effectiveness of barriers. One thing that is not addressed in most of the studies is how the overall suicide rate in a community is affected by suicide barriers placed at one site. To us this is the only relevant question. If people use other sites or other means for suicide then fences may actually cause more suicides than they prevent. We need evidence not speculation to make rational decisions about the need for barriers on the bridges.
One issue that is not answered by the studies is what is the message that fences send to would be suicide attempters. We believe that fences send a cruel and callous message that says go away, move on, don't kill yourself here, we don't want to see it. Sadly, the people who may have been identified as suicidal and saved on the bridges are now forced to suffer alone. There is absolutely no evidence that fences reduce the overall suicide rate in any community. Spreading misinformation does not help anyone, especially the people struggling with mental illness who need help not fences.
We have tried to provide open source sites to the research papers. Some of them are not working right now. We're working on that as well as adding other research papers.
There are a range of other options that have not been considered because the Cornell advisors have overstated the effectiveness of suicide barriers and means restriction. Among those options that have not been considered: signs and telephone hotline, suicide patrols, improved lighting/visibility at particular sites, training for Cornell University staff to identify individuals in distress, mentoring system for students to identify peers in distress, and doing nothing.
The key question about the fences is: Do they work?
A national survey found that "34% of respondents believed that every single jumper would have found another way to complete suicide and an additional 40% believed that most would have completed suicide using other means" (Miller et al., 2007).
Contrary to these numbers, the Cornell University advisors who assisted in the decision to install bridge barriers insist that: "A small number of studies have formally evaluated the impact of installing barriers at suicide sites. All studies show barriers are effective in reducing suicides from that site, without displacement to neighboring sites, and sometimes accompanied by a reduction in suicides by jumping in the surrounding region." (Beautrais et al., 2010).
This statement is not true. All studies DO NOT agree about the effectiveness of barriers. One thing that is not addressed in most of the studies is how the overall suicide rate in a community is affected by suicide barriers placed at one site. To us this is the only relevant question. If people use other sites or other means for suicide then fences may actually cause more suicides than they prevent. We need evidence not speculation to make rational decisions about the need for barriers on the bridges.
One issue that is not answered by the studies is what is the message that fences send to would be suicide attempters. We believe that fences send a cruel and callous message that says go away, move on, don't kill yourself here, we don't want to see it. Sadly, the people who may have been identified as suicidal and saved on the bridges are now forced to suffer alone. There is absolutely no evidence that fences reduce the overall suicide rate in any community. Spreading misinformation does not help anyone, especially the people struggling with mental illness who need help not fences.
We have tried to provide open source sites to the research papers. Some of them are not working right now. We're working on that as well as adding other research papers.
Effect of a barrier at Bloor Street Viaduct on suicide rates in Toronto: natural experiment (Sinyor & Levitt, 2010).
Suicides by jumping off the Bloor Street Viaduct in Toronto, Canada, were studied from 1993-2007. In 2003 a 5 meter high suicide barrier was installed. Yearly rates of suicide by jumping in Toronto remained unchanged between the periods before and after the construction of the barrier (56.4 v 56.6, P=0.95). A mean of 9.3 suicides occurred annually at Bloor Street Viaduct before the barrier and none after the barrier was erected (P<0.01). Nevertheless, the annual rate of suicides by jumping from other bridges increased significantly (8.7 v 14.2, P=0.01) and suicides by jumping from buildings increased non-significantly (38.5 v 42.7, P=0.32).
http://www.bmj.com/content/341/bmj.c2884.long
http://www.bmj.com/content/341/bmj.c2884.long
Garrett Glascow's Review of Suicide Barrier Studies
After reviewing the scientific research Professor Garrett Glascow concludes that there is no evidence that erecting a Suicide Barriers would save lives on a bridge in California. His arguments are general arguments about the ineffectiveness of suicide barriers. He reviews the research on suicide barriers and the weaknesses on those studies.
http://www.polsci.ucsb.edu/faculty/glasgow/barrier_report.pdf
http://www.polsci.ucsb.edu/faculty/glasgow/barrier_report.pdf
(Beautrais, 2001)
Removing bridge barriers stimulates suicides: an unfortunate natural experiment (Beautrais et al., 2009).
This study was performed by Annette L. Beautrais, one of Cornell University's three advisors regarding the installation of suicide barriers on Ithaca's bridges. Suicides by jumping off the Grafton bridge in Auckland, New Zealand were studied from 1991-2006. Suicide barriers were erected in 1937, removed during 1996 and reinstalled at the start of 2003. From 1991-1995 (when the old barrier was in place) there were 5 suicides. From 1997-2002 (when there was no barrier) there were 19 suicides. From 2003-2006 (when a barrier was in place) there were no suicides reported from this particular bridge. From these numbers the authors conclude that "safety barriers are effective in preventing suicide: their removal increases suicides; their reinstatement prevents suicides."
NOTE: This study was conducted on a bridge next to a psychiatric hospital. Most of the suicides were escaped psychiatric patients from that hospital. This study contains no control group at all, no data on suicides in Auckland by other means (means substitution) or data whether other locations (location substitution) for suicide was used.
There is no direct link to the the actual research paper. It used to be an open link but since Beatrais has collaborated with Cornell and been criticized by us the links have been removed. Now the links lead to a summary of the research paper. The flawed conclusion is stated in this summary but the serious limitations that make the study practically useless are not mentioned in this summary. This is a good example of how Cornell has distorted the findings of the research by highlighting only the "conclusion" but not informing people of what the study actually looks at or finds. Please read the entire research paper.
To read this flawed study:
Google "Removing bridge barriers stimulates suicides"
Then click on the pdf result at www.gannett.cornell.edu or www.otaga.ac.nz (NOTE: these links will NOT lead you to the full article. You must google as directed then select the result listed here.)
NOTE: This study was conducted on a bridge next to a psychiatric hospital. Most of the suicides were escaped psychiatric patients from that hospital. This study contains no control group at all, no data on suicides in Auckland by other means (means substitution) or data whether other locations (location substitution) for suicide was used.
There is no direct link to the the actual research paper. It used to be an open link but since Beatrais has collaborated with Cornell and been criticized by us the links have been removed. Now the links lead to a summary of the research paper. The flawed conclusion is stated in this summary but the serious limitations that make the study practically useless are not mentioned in this summary. This is a good example of how Cornell has distorted the findings of the research by highlighting only the "conclusion" but not informing people of what the study actually looks at or finds. Please read the entire research paper.
To read this flawed study:
Google "Removing bridge barriers stimulates suicides"
Then click on the pdf result at www.gannett.cornell.edu or www.otaga.ac.nz (NOTE: these links will NOT lead you to the full article. You must google as directed then select the result listed here.)
Preventing suicide by jumping: the effect of a bridge safety fence (Pelletier, 2007).
Suicides by jumping off the Memorial bridge in Augusta, Maine, were studied from 1960-2005. In 1983 a 11 feet high suicide barrier was installed. From 1960-1983 a total of 14 suicides from the memorial bridge were identified (6% of all suicides in Augusta) of which 10 of the individuals had a documented history of psychiatric illness (71%). Between 1983-2005 there were no suicides from this particular bridge. Jumping from other high places was unaffected (N = 9 from 1960-1963 and N = 9 from 1983-2005).
The author of the study admits that data for 1968 was not available. And, that the data for the years 1960 to 1974 do not meet current data quality standards and may underestimate the number of suicides from the Memorial Bridge during that period. The suicide rate for Maine was unchanged after the the suicide barrier was erected on the Memorial Bridge. In Augusta the suicide rate slightly decreased from 26/100,000 per year before the barrier to 23.8/100,000 per year after the barrier. The author acknowledges that the decline in the suicide rate in Augusta after the installation of the safety fence was not statistically significant,
Conclusion: Installation of the a fence prevented suicides from the particular bridge only, but had no statistically significant effect on death by jumping from other sites or the general suicide rate in Augusta.
NOTE: This study was conducted on a bridge next to a psychiatric hospital. Most of the suicides were escaped psychiatric patients from that hospital.
http://www.glendon.org/content/_common/attachments/barriers/preventing_suicide_by_jumping._the_effect_of_a_bridge_safety_fence.pdf
(open access through Glendon Association) (NOTE: we've had trouble with this link)
The author of the study admits that data for 1968 was not available. And, that the data for the years 1960 to 1974 do not meet current data quality standards and may underestimate the number of suicides from the Memorial Bridge during that period. The suicide rate for Maine was unchanged after the the suicide barrier was erected on the Memorial Bridge. In Augusta the suicide rate slightly decreased from 26/100,000 per year before the barrier to 23.8/100,000 per year after the barrier. The author acknowledges that the decline in the suicide rate in Augusta after the installation of the safety fence was not statistically significant,
Conclusion: Installation of the a fence prevented suicides from the particular bridge only, but had no statistically significant effect on death by jumping from other sites or the general suicide rate in Augusta.
NOTE: This study was conducted on a bridge next to a psychiatric hospital. Most of the suicides were escaped psychiatric patients from that hospital.
http://www.glendon.org/content/_common/attachments/barriers/preventing_suicide_by_jumping._the_effect_of_a_bridge_safety_fence.pdf
(open access through Glendon Association) (NOTE: we've had trouble with this link)
Effect of barriers on the Clifton suspension bridge, England, on local patterns of suicide: implications for prevention (Bennewith et al., 2007).
Suicides by jumping off the Clifton suspension bridge in Bristol, England were studied from 1994-2003. In 1998 a 2 meter high suicide barrier was installed covering only the span of the bridge. From 1994-1998 a total of 41 suicides from the Clifton bridge were identified (8.2/year). Between 1998-2003 there were 20 suicides from this particular bridge.
Before installation of the barrier in 1998 - 30 of 31 (97%) cases occurred from the span of the bridge. After installation of the barrier 8 of 17 jumped from the buttresses of the bridge (although not explicitly stated 9 of 17 appeared to have climbed the fence). There was a non-significant increase in the number of deaths from jumping from sites other than the suspension bridge (6.2/y to 8.4/y) and a non-significant decrease in mean number of deaths from jumping from all sites in the surrounding area (14.3/y to 12.4/y).
Conclusion: Installation of the a partial fence failed to prevent suicides from the Clifton suspension bridge and had no significant effect on death by jumping from other sites.
In direct contradiction to the results of this study, the Cornell report 2010 states: "Bennewith and colleagues examined the effect of installation of barriers on the Clifton suspension bridge, Bristol, England in 1998 on local suicides by jumping. Bridge deaths halved from 8.2 per year (1994 - 1998) to 4.0 per year (1999 -2003; P<0.008). (Note: Only the main arches were fenced; suicides migrated to the unfenced edges of the bridge). Although 90% of the suicides from the bridge were by males, there was no evidence of an increase in male suicide by jumping from other sites in the Bristol area after erection of barriers. The authors claim this study provides evidence for the effectiveness of barriers on bridges in preventing site-specific suicides and suicides by jumping overall in the surrounding area." (Beautrais et al., 2010)
http://www.glendon.org/content/_common/attachments/barriers/effects_of_barriers_on_the_clifton_suspension_bridge.pdf(open access through Glendon Association) (NOTE: we've had trouble with this link)
Before installation of the barrier in 1998 - 30 of 31 (97%) cases occurred from the span of the bridge. After installation of the barrier 8 of 17 jumped from the buttresses of the bridge (although not explicitly stated 9 of 17 appeared to have climbed the fence). There was a non-significant increase in the number of deaths from jumping from sites other than the suspension bridge (6.2/y to 8.4/y) and a non-significant decrease in mean number of deaths from jumping from all sites in the surrounding area (14.3/y to 12.4/y).
Conclusion: Installation of the a partial fence failed to prevent suicides from the Clifton suspension bridge and had no significant effect on death by jumping from other sites.
In direct contradiction to the results of this study, the Cornell report 2010 states: "Bennewith and colleagues examined the effect of installation of barriers on the Clifton suspension bridge, Bristol, England in 1998 on local suicides by jumping. Bridge deaths halved from 8.2 per year (1994 - 1998) to 4.0 per year (1999 -2003; P<0.008). (Note: Only the main arches were fenced; suicides migrated to the unfenced edges of the bridge). Although 90% of the suicides from the bridge were by males, there was no evidence of an increase in male suicide by jumping from other sites in the Bristol area after erection of barriers. The authors claim this study provides evidence for the effectiveness of barriers on bridges in preventing site-specific suicides and suicides by jumping overall in the surrounding area." (Beautrais et al., 2010)
http://www.glendon.org/content/_common/attachments/barriers/effects_of_barriers_on_the_clifton_suspension_bridge.pdf(open access through Glendon Association) (NOTE: we've had trouble with this link)
Securing a Suicide Hot Spot: Effects of a Safety Net at the Bern Muenster Terrace (Reisch & Michel, 2005).
In this study by Thomas Reisch and Konrad Michel, which was published in Suicide and Life Threaten...ing
Behavior in 2005, showed no significant reduction in jumping suicides
in Bern, Switzerland after a barrier went up on the city's iconic
suicide bridge. There were 45 jumping suicides per year in Bern in the 4
years before a barrier went up and 44 jumping suicides in Bern in the 4
years after the barrier went up. Had the barrier worked, one might have
expected roughly 35.
In direct contradiction to the results of this study 2010 Cornell report states... " "Reisch and Michel (2005) reported that the city of Bern had a high percentage of suicides by jumping (28.6%). The highest number of deaths (mean 2.5 per year) occurred at the Muenster Terrace. In 1998, after a series of suicides, a safety net was built to prevent people leaping from the terrace and to avoid traumatization of people living in the street below. After the installation of the net no suicides occurred from the terrace. The number of people jumping from all high places in Bern was significantly lower compared to the years before, indicating to the authors that no immediate shift to other nearby jumping sites took place. Furthermore, they found a moderate correlation between the number of media reports and the number of persons resident outside Bern committing suicide by jumping from high places in the city." (Beautrais et al., 2010).
http://www.glendon.org/content/_common/attachments/barriers/securing_a_hot_spot.pdf (open access through Glendon Association) (NOTE: we've had trouble with this link)
If you can't get to the pdf of this study you can put the entire title of the study above in a Google search and scroll down for the link.
In direct contradiction to the results of this study 2010 Cornell report states... " "Reisch and Michel (2005) reported that the city of Bern had a high percentage of suicides by jumping (28.6%). The highest number of deaths (mean 2.5 per year) occurred at the Muenster Terrace. In 1998, after a series of suicides, a safety net was built to prevent people leaping from the terrace and to avoid traumatization of people living in the street below. After the installation of the net no suicides occurred from the terrace. The number of people jumping from all high places in Bern was significantly lower compared to the years before, indicating to the authors that no immediate shift to other nearby jumping sites took place. Furthermore, they found a moderate correlation between the number of media reports and the number of persons resident outside Bern committing suicide by jumping from high places in the city." (Beautrais et al., 2010).
http://www.glendon.org/content/_common/attachments/barriers/securing_a_hot_spot.pdf (open access through Glendon Association) (NOTE: we've had trouble with this link)
If you can't get to the pdf of this study you can put the entire title of the study above in a Google search and scroll down for the link.
Duke Ellington Bridge, Washington D.C.
O'Carroll & Silverman (1994)
In 1986 a suicide barrier was installed on the Duke Ellington Bridge which had an average of four deaths per year from that bridge. On the neighboring Taft Bridge there were on average two deaths per year. After construction of the suicide barrier on the Duke Ellington Bridge there was one death from that Bridge. Over the same period, the suicide rate on the neighboring Taft Bridge, which had no suicide barriers, remained the same .O'Carroll suggests a
number of criteria that must be included in future studies to draw
any reliable conclusion about the efficacy of fences.
The study does not look at the overall suicide rate in the Washington, D.C. area to examine if the rate is decreased by the number of suicides that were stopped on the Duke Ellington Bridge. The study does not look at suicides from any other bridges or high places in the Washington, D.C. to determine if the suicide rates at those locations went up.
Important Note: The authors of this study admit that the data collected is insufficient to conclude that suicide barriers save lives.
Unfortunately, only the first page of this research paper is open source:
http://onlinelibrary.wiley.com/doi/10.1111/j.1943-278X.1994.tb00666.x/full
The study does not look at the overall suicide rate in the Washington, D.C. area to examine if the rate is decreased by the number of suicides that were stopped on the Duke Ellington Bridge. The study does not look at suicides from any other bridges or high places in the Washington, D.C. to determine if the suicide rates at those locations went up.
Important Note: The authors of this study admit that the data collected is insufficient to conclude that suicide barriers save lives.
Unfortunately, only the first page of this research paper is open source:
http://onlinelibrary.wiley.com/doi/10.1111/j.1943-278X.1994.tb00666.x/full
The Golden Gate Bridge Study
Where are They Now? A Follow-up Study of Suicide Attempters from the Golden Gate Bridge (Seiden, 1978)
This misleading and false study is one of the most cited studies to support the claim that means restriction is effective in suicide prevention. In this study the author follows up with 515 subjects who were involved in “any incident in which a subject commits an overt act toward an attempt to commit suicide” at the Golden Gate Bridge from 1937-1971 (GGB group) and compares that group with 184 persons hospitalized for attempting suicide by other means at the San Francisco General Hospital emergency room from 1956-1957 (SFGH group).
The unjustified high impact of this clearly outdated publication stems from the false conclusion that “after 26-plus years the vast majority of GGB suicide attempters (about 94%) are still alive or have died from natural causes“. This claim is false and misleading for two reasons: (1) the study followed up with only 12 of the 515 subjects after a 26 year period; (2) of the deaths from the GGB group, approximately 40% of the people who died during the study after being stopped from jumping from the Golden Gate bridge had committed suicide by 1971, shockingly 12 % were suicides by jumping.
1. The claim of a “median follow up period of 26-plus years“ in the study is false and misleading. The study was conducted in 1971. Only 12 of 515 cases (2.3%) were recorded from 1937-1946 and could be followed up for 26 years in 1971. 310 of 515 cases (60%) occurred from 1962-1971 and could only be followed up for a maximum of 9 years. 132 of 515 cases (26%) occurred from 1967-1971 with a maximum follow up period of only 4 years. Hence, for the median case occurring in 1963 only a maximum of 8 years follow-up was possible in 1971.
2. During the 1937-1971 study period 64 deaths were recorded in the GGB group (12.5% of the total group); of these 49.2% died of natural cause, 11.1% died in accidents and 39.7% died by suicide - 12.5% were suicides by jumping. (7 returned to jump from the GGB, and one jumped from the Bay bridge.)
During the 1937-1971 study period 47 deaths were recorded in the SFGH group (25.5% of the total group); of these 57.4% died of natural causes, 12.8% died in accidents 27.7% died by suicide.
In comparison with the US population at large in 1960 (92.9% died of natural causes, 5.5% died in accidents and 1.1% died by suicide).
The numbers clearly indicate a highly significant enhancement of the prospect of a violent death for both the GGB and SFGH groups. Contrary to the central claim that “failed suicide attempters will not try again“, Seiden's study indicates quite the opposite. In the GGB group 31% of the deaths by suicide occurred within 6 months after the Golden Gate Bridge attempt.
Finally, Seiden tried to calculate the suicide rate of GGB suicide attempters (his result is 182.69/100000 per year), which is 16.6 times higher than those of the general US population in 1960 (approx. 11/100000 per year). Seiden again uses the false "median study period of 26 years, 7 months“, which is only accomplished for less than 2.6% of the GGB cases (as described above), and even fails to calculate the correct result [rate = (suicides in GGB*100000) / (N of GGB*median study period) = (35*100000) / (515*26.8) = 253.6 / 100000 per year, NOT 182.69]. Using the maximum follow up period of 9 years for more than 60% of the subjects a suicide rate of approximately 755/100000 per year is obtained, which is about 70 times higher than those for the general population in 1960.
Consequently, the widely cited conclusion of Seiden’s study is not supported by the data and no additional follow up study has ever been performed. We believe anyone who relies on this study is either knowingly misleading people or is unable to understand the basic foundations of good science.
http://www.seattlefriends.org/files/seiden_study.pdf (open Access)
The unjustified high impact of this clearly outdated publication stems from the false conclusion that “after 26-plus years the vast majority of GGB suicide attempters (about 94%) are still alive or have died from natural causes“. This claim is false and misleading for two reasons: (1) the study followed up with only 12 of the 515 subjects after a 26 year period; (2) of the deaths from the GGB group, approximately 40% of the people who died during the study after being stopped from jumping from the Golden Gate bridge had committed suicide by 1971, shockingly 12 % were suicides by jumping.
1. The claim of a “median follow up period of 26-plus years“ in the study is false and misleading. The study was conducted in 1971. Only 12 of 515 cases (2.3%) were recorded from 1937-1946 and could be followed up for 26 years in 1971. 310 of 515 cases (60%) occurred from 1962-1971 and could only be followed up for a maximum of 9 years. 132 of 515 cases (26%) occurred from 1967-1971 with a maximum follow up period of only 4 years. Hence, for the median case occurring in 1963 only a maximum of 8 years follow-up was possible in 1971.
2. During the 1937-1971 study period 64 deaths were recorded in the GGB group (12.5% of the total group); of these 49.2% died of natural cause, 11.1% died in accidents and 39.7% died by suicide - 12.5% were suicides by jumping. (7 returned to jump from the GGB, and one jumped from the Bay bridge.)
During the 1937-1971 study period 47 deaths were recorded in the SFGH group (25.5% of the total group); of these 57.4% died of natural causes, 12.8% died in accidents 27.7% died by suicide.
In comparison with the US population at large in 1960 (92.9% died of natural causes, 5.5% died in accidents and 1.1% died by suicide).
The numbers clearly indicate a highly significant enhancement of the prospect of a violent death for both the GGB and SFGH groups. Contrary to the central claim that “failed suicide attempters will not try again“, Seiden's study indicates quite the opposite. In the GGB group 31% of the deaths by suicide occurred within 6 months after the Golden Gate Bridge attempt.
Finally, Seiden tried to calculate the suicide rate of GGB suicide attempters (his result is 182.69/100000 per year), which is 16.6 times higher than those of the general US population in 1960 (approx. 11/100000 per year). Seiden again uses the false "median study period of 26 years, 7 months“, which is only accomplished for less than 2.6% of the GGB cases (as described above), and even fails to calculate the correct result [rate = (suicides in GGB*100000) / (N of GGB*median study period) = (35*100000) / (515*26.8) = 253.6 / 100000 per year, NOT 182.69]. Using the maximum follow up period of 9 years for more than 60% of the subjects a suicide rate of approximately 755/100000 per year is obtained, which is about 70 times higher than those for the general population in 1960.
Consequently, the widely cited conclusion of Seiden’s study is not supported by the data and no additional follow up study has ever been performed. We believe anyone who relies on this study is either knowingly misleading people or is unable to understand the basic foundations of good science.
http://www.seattlefriends.org/files/seiden_study.pdf (open Access)
