HIPRC: Best Practices

Adolescent Suicide

Suicide Interventions

Restriction of Access to Means of Suicide

Background

Following the premises of injury control, passive prevention strategies may be more effective than active strategies. Thus, one strategy is to limit access to lethal means of suicide.

1. Restricting access to guns

The most common means of suicide is firearms. Much of the attention to firearms has focused on homicide and assault, but more people in the US and in other countries of the world lose their lives to guns involved in suicide than in homicide.

Decreased access to guns can occur through safe storage of guns and through legislation restricting access to guns. There are no studies of the impact of safe storage programs on gun related suicides. There have been however a number of studies of the impact of gun laws on suicide rates. These are primarily ecological comparisons of different jurisdictions.

Review of gun restriction studies:

Author	Cummings, et al., 1997
Study design and target population	Ecological design US Population
Intervention	Effect of child access prevention ("CAP") laws.
Outcomes	Gun suicide rates.
Results
Study quality and conclusions	Well done time series study.

Author	Cantor & Slater, 1995
Study design and target population	Ecological study Comparison of firearm suicides before and after law in Queensland, Australia.
Intervention	Law requiring owners of long guns to be licensed, went into effect January 1, 1992.
Outcomes	Gun suicides in two years before (1990-1991) compared to 2 years after (1992-1993) in metropolitan, provincial city, and rural areas.
Results	Rates of adolescent gun suicide decreased by 57% in metropolitan areas, and 48% in provincial areas, but did not changes in rural areas. The overall number of suicides by men of all ages decreased in provincial areas but did not change in the other areas nor for women in any areas.
Study quality and conclusions	Weak study. Shows there may have been an effect on gun suicides but there was also evidence for a substitution effect.

Author	Loftin et al., 1991
Study design and target population	Ecological District of Columbia, 1968-1987 compared to neighboring areas in Maryland and Virginia.
Intervention	1976 D.C. law banning sale or ownership of handguns in D.C.
Outcomes	Suicide rates
Results	Suicide rates declined by 2.3% in D.C. There were no similar declines in MD or VA. Suicide by other means did not increase.
Study quality and conclusions	The data suggest that restrictions on access to gung in the District of Columbia was effective.

Author	Sloan et al., 1990
Study design and target population	Ecological study Comparison of King County Washington with the Vancouver Metropolitan area
Intervention	Restricted access to handguns in Vancouver
Outcomes	Suicide fatalities, 1985-1987
Results	There was no difference in overall suicide rates between the two areas (RR=0.97, 0.87-1.09). However, persons 15 to 24 years had a lower rate of suicide in Vancouver (RR=0.72, 0.54-0.98). This was virtually all due to an almost 10-fold higher rate of suicide by handguns in King County.
Study quality and conclusions	This study is a well done study but is limited by the fact that it is an ecological comparison of two different cities in two different countries. The study provides some evidence that restricting access to handguns can decrease the risk of youth suicide form handguns.

Author	Lester and Murrell, 1982
Study design and target population	Time series analysis US states, 1960 and 1970
Intervention	Strictness of state handgun control laws in 1968
Outcomes	Suicide rates by guns and by other causes 1960, 1970
Results	Inverse correlation between strictness of suicide laws and suicide rates by guns (r=0.46 in 1960 and -0.52 in 1970). States with stricter laws had higher rates of suicide by other means
Study quality and conclusions	No data by age. Correlation ecological analysis has serious limitations.

Summary on studies of restricting access to guns

The only studies which have been done to date are ecological analyses or correlation analyses. As a result, the conclusions which can be drawn on the impact of firearm restrictions are not clear. Case control studies examining the effect of gun availability are reviewed separately below.

2. Gun availability and risk of adolescent suicide

One design which has been used to examine the effect of gun availability is the case-control method in which homes where there has been a suicide are compared to homes where there has not been a suicide on gun ava ilability. In these studies, it is important to look at ALL suicides so that an substitution effect can be included. We review the studies here which include specific data on adolescents.

Review of gun availability and risk of adolescent suicide studies:

Author	Cummings et al., 1997
Study design and target population	Group Health Cooperative patients. Purchase of a handgun.
Intervention	Suicide fatality between 1980 and 1992.
Outcomes	GHC members matched on age, gender, and zip code of residence.
Results	RR of suicide for family purchase of gun: 1.9 (95% CI 1.4, 2.5). RR of suicide for personal purchase of gun: 2.0 (95% CI 1.4, 2.8). Time since family purchase of less than 1 year: 5.7 (2.4, 13.5).
Study quality and conclusions	No separate data on adolescents.

Author	Beautrais et al. 1996
Study design and target population	Christchuch NZ, suicide and suicide attempts. Access to gun in home.
Intervention	197 suicides and 302 suicide attempts between 1991 and 1995.
Outcomes	Age and gender matched controls from voter registration files.
Results	OR for access to gun was 1.4 (95% CI 0.96, 1.99). For males, adjusted OR 1.00 (95% CI 0.74, 1.34).
Study quality and conclusions	94% response rate among case families and 95% among non-fatal cases. GSW accounted for only 1.3% of non-fatal and 13.3% of fatalities. No separate data on adolescents. No evidence of an effect of access on suicide risk.

Author	Brent et al., 1993
Study design and target population	Pittsburgh adolescents. Firearms in the home.
Intervention	67 adolescent suicide victims.
Outcomes	67 demographically matched community controls.
Results	After adjusting for differences in rates of psychiatric disorders between suicide victims and controls, the association between suicide and any gun (OR=4.4, 95% CI 1.1 to 17.5). Presence of handguns (OR=9.4, 95% CI= 1.7 to 53.9) in the home. Long guns in the home were associated with suicide only in rural areas, whereas handguns were more closely associated with suicide in urban areas. Handguns (OR= 12.9, 95% CI=1.5 to 110.9) and loaded guns (OR= 32.3, 95% CI=2.5 to 413.4) in the home were particularly significant risk factors for suicide in those with no apparent psychiatric disorder.
Study quality and conclusions

Author	Kellerman, Rivara, et al.,1992
Study design and target population	Shelby county, TN and King county, WA suicide deaths. Home ownership of guns.
Intervention	Suicide occuring in the home.
Outcomes	Age, gender, race, and neighborhood matched controls.
Results	For all ages adjusted odds ratio: 4.8 (95% CI, 2.7 to 8.5). For adolescents, aOR:10.
Study quality and conclusions	Used neighborhood controls to control for other factors associated with owning a gun. Effect was strongest in the adolescent age group.

Author	Brent et al., 1991
Study design and target population	Pittsburgh adolescents. Gund in the home.
Intervention	47 adolescent suicide victims.
Outcomes	The two psychiatric inpatient control groups were 47 suicide attempters and 47 never-suicidal psychiatric controls, frequency-matched to the suicide victims on age, gender, and county of origin.
Results	Guns in the home of suicide victims as in the homes of attemptors (aOR, 2.1; 95% CI 1.2 to 3.7) or psychiatric controls (aOR, 2.2; 95% CI, 1.4 to 3.5). Handguns were not associated with suicide to any statistically significantly greater extent than long guns. No effect of gun storage methods.
Study quality and conclusions	Small study.

3. Restricting access to non-firearm means

One of the more common non-firearm methods of committing suicide is carbon monoxide. In 1957, the mean carbon monoxide content of domestic gas (derived from coal) in Great Britain was 12%. Self-asphyxiation with domestic gas accounted for over 40% of all British suicides. Following the discovery of natural gas in the North Sea, homes were converted to natural gas for domestic use; it had a CO content of only 2%. As a result, overall suicide rates dropped 26% and suicide by CO asphyxiation fell from 40% to less than 10%. However, the recent study by Gunnell below indicates some substitution may have occurred.

Review of restricting access to non-firearm means studies:

Author	Hawton, et al., 2001
Study design and target population	Time series study England and Wales, 1996-1999. People aged 12 and older.
Intervention	1998 law in UK limiting sale of salicylates or acetaminophen to 32 tablets in pharmacies and 16 in other retail outlets; warnings on packages.
Outcomes	Deaths: suicide, accidental, self-poisoning, undetermined cause. A&E visits for poisonings at 7 hospitals. Sales data
Results	In the year following the change, there was a 21% decrease (95% CI -34, -5) in acetaminophen deaths and a 48% (-70, -11) reduction in salicylate deaths. There was a 30% decrease in admissions for liver disease related to acetaminophen poisoning, a 59% decrease in the number awaiting liver transplant and a 66% decrease in liver transplants because of acteminophen poisoning. The total number of tablets of acetaminophen did not change, although there was a drop in the number of tablets of salicylates sold.
Study quality and conclusions	There appeared to be a significant association between the law and a reduction in poisoning related to these compounds. They do not provide any data on other types of poisonings in self-harm attempts, so we do not know if there is any substitution effect.

Author	Gunnell, et al., 2000
Study design and target population	Ecological time series study, England and Wales 1950-1975.
Intervention	Change in CO content of domestic gas.
Outcomes	Rates of suicide.
Results	For males 15-34 years, overall rate of suicide increased by 37% between 1950-52 and 1973-1975. There was a 34% decrease in suicides by gas but a 299% increase in suicide by drug overdose. For females, there was a 54% increase overall, and 89% decrease in suicide by gas, but a 305% increase in suicide by drug OD.
Study quality and conclusions	No separate data just for adolescents.

Author	Birckmayer & Hemenway et al., 1999
Study design and target population	US, 1970-1980, Ecological design.
Intervention	Effect of minimum age drinking laws changed to age 21 in the US.
Outcomes	Suicide rates of 18-20 year olds.
Results	States with an 18-year old minimum drinking age had an 8% higher suicide rate among 18-20 year olds than states with an age 21 law. There was no effect of laws on rates of suicide among 15-17 year olds.
Study quality and conclusions	Ecological design. Mechanism of effect not clear

Author	Buckley et al., 1995
Study design and target population	Pre-test, post-test design Newcastle, UK Adolescents and adults
Intervention	Change from sale of carbamazepine in bottles to sale in blister packs
Outcomes	Amount of carbamazepine taken in suicide attempts
Results	65% reduction in number of tablets taken (mean of 7 compared to mean of 20). 53% reduction in total amount ingested (1400 mg compared to 3000 mg). No change in clinical outcomes
Study quality and conclusions	Study was simple before after study which did not control for other factors. Suggests that safety packaging may also affect intentional overdoses in teens and adults.

Author	Lester, 1990
Study design and target population	Time series analysis US 1950-1070
Intervention	Detoxification of domestic gas in the US
Outcomes	Suicide rates, overall and type specific
Results	Between 1950 and 1970, suicides by domestic gas decreased by 97% (0.73/100,000 to 0.02/100,000). The overall suicide rate remained essentially unchanged at 11/100,000. Suicide by MV exhaust increased by 70%.
Study quality and conclusions	No data for teens or young adults separately. Although rates of suicide by domestic gas decreased, overall suicide rates remained unchanged. However, even in 1950, domestic gas accounted for only 6% of suicides.

Author	Hingson et al., 1985
Study design and target population	Pre-test, post-test study Massachusetts 15-19 year olds compared to NY state (outside of NY city) 15-19 year olds
Intervention	Change in legal drinking age from 18 to 20
Outcomes	Suicide deaths
Results	5% increase in suicide deaths in Mass 18-19 year olds compared to 1% increase in NY (OR=1.04, 0.65-1.69).
Study quality and conclusions	No effect of the law on changes in suicide rates in the target age group

Author	Kreitman, 1976
Study design and target population	Time series analysis Great Britain
Intervention	Introduction of natural gas for domestic use in Great Britain, which essentially eliminated risk from CO.
Outcomes	Suicide rates
Results	Suicide rate declined by 26% overall and the proportion of suicides due to gas declined from 40% to less than 10%. However, this effect was most prominently seen in older adults. For teenage and young adult males and females, suicide by other means did increase by 9-33%, dampening the effect of the decrease in suicides by CO
Study quality and conclusions	Well-done study, but limited by the fact that it was an ecological analysis. Supporting evidence that elimination of lethal means of suicide can result in decrease in suicide rates for older adults, but that there is evidence of a substitution effect to a large degree among teens and young adults.

Author	Oliver and Hertzel, 1973
Study design and target population	Time series analysis Australia, 1955-1970
Intervention	Restriction on sale of barbiturates in 1967
Outcomes	Suicide rates
Results	Overall suicide rates peaked in 1967 then subsequently dropped. Likewise, there was a drop of drug suicides of 29% between 1967 peak and 1970 and a drop in the number of sedative prescriptions
Study quality and conclusions	There are no control comparisons so that it is difficult to determine if the drop was related solely to the restrictions on access to Phenobarbital.

Summary of studies evaluating impact of restricting access to non-firearm means

The recent data from the UK indicates that while there may have been some immediate effect on suicide rates with the change from coal gas to natural gas, over a 25 year period there appeared to be a substitution effect with a marked increase in suicide by drug overdoes.

The data on the effect of restricting access to commonly used drugs in suicide is suggestive that there was an effect on suicide rates. The recent study on changes in the number of salicylate and acetaminophen tablets allowed to be sold shows an associated change in suicide attempts related to these two compounds but does not look at any substitution effect. However, these were not strong designs and futher evaluation is needed.

There was no indication that changing the legal age for purchase of alcohol affected suicide rates in the Hingson study of one state, but strong evidence for an effect in the study by Birckmayer and Hemenway of the entire US.

Recommendations on programs to restrict access to poisons

Based on the above, these interventions appear promising. However, the data are best for CO, which in most countries is no longer a common means of suicide. Restricting access to other means of suicide from drugs or poisons should be accompanied by an evaluation to determine its effect.