http://informahealthcare.

com/ada
ISSN: 0095-2990 (print), 1097-9891 (electronic)

Am J Drug Alcohol Abuse, 2014; 40(4): 269–273

! 2014 Informa Healthcare USA, Inc. DOI: 10.3109/00952990.2014.910520

Does the legalization of medical marijuana increase completed suicide?

Melanie Rylander, MD1,2,4, Carolyn Valdez, MS3, and Abraham M. Nussbaum, MD1,4
1
Departments of Behavioral Health, 2Departments of Internal Medicine, 3Departments of Patient Safety and Quality, Denver Health, Denver,
Colorado, and 4Department of Psychiatry, University of Colorado School of Medicine, Colorado, USA

Abstract Keywords
Introduction: Suicide is among the 10 most common causes of death in the United States. Altitude, cannabis, legalization of marijuana,
Researchers have identified a number of factors associated with completed suicide, including medical marijuana, suicide, unemployment
marijuana use, and increased land elevation. Colorado is an ideal state to test the strength of
these associations. The state has a completed suicide rate well above the national average and History
over the past 15 years has permitted first the medical and, as 2014, the recreational use of
marijuana. Objectives: To determine if there is a correlation between medical marijuana use, as Received 18 November 2013
assessed by the number of medical marijuana registrants and completed suicides per county in Revised 19 March 2014
Colorado. Methods: The number of medical marijuana registrants was used as a proxy for Accepted 23 March 2014
marijuana use. Analysis variables included total medical marijuana registrants, medical Published online 20 June 2014
marijuana dispensaries per county, total suicide deaths, mechanism of suicide death, gender,
total suicide hospitalizations, total unemployment, and county-level information such as mean
elevation and whether the county was urban or rural. Analysis was performed with mixed
model Poisson regression using generalized linear modeling techniques. Results: We found no
consistent association between the number of marijuana registrants and completed suicide
after controlling for multiple known risk factors for completed suicide. Conclusion: The
legalization of medical marijuana may not have an adverse impact on suicide rates. Given the
concern for the increased use of marijuana after its legalization, our negative findings provide
some reassurance. However, this conclusion needs to be examined in light of the limitations of
our study and may not be generalizable to those with existing severe mental illness. This
finding may have significant public health implications for the presumable increase in
marijuana use that may follow legalization.

Introduction citizen-initiated amendment legalized the possession,

distribution, and recreational use of one ounce or less of
Colorado has received national attention for its approach to
marijuana. Simultaneously, members of the Colorado psychi-
marijuana. In 2000, a citizen-initiated amendment to the state
atric community have voiced concerns regarding the adverse
constitution allowed for the medical use of marijuana. After
psychiatric effects of marijuana use, especially with regards to
the federal government announced in 2009 (1) that it would
suicide (3).
not routinely prosecute users and distributors, the number of
In large cohort studies, the association of marijuana use
registrants to the state’s medical marijuana registry increased
and suicide has been inconsistent (4–9). Degenhardt et al.
dramatically. By 2011, more than 2% of the state’s population
examined a series of cohort and cross sectional studies and
had registered to use marijuana for a medical purpose (2).
found a modest association for heavy or problematic cannabis
Concerns have been raised that the criteria to obtain a
use and depressive symptoms. They also found a modest
medical marijuana card is extremely broad, and that many of
association between early onset regular cannabis use and later
those on the medical marijuana registry are recreational
onset depression. No association was found for baseline
users who have found a way to obtain the drug legally (2,3).
depression with later onset of cannabis use, arguing against
Thus, the population of medical marijuana users may be
the self-medication hypothesis (4). Lynskey et al. examined
very similar to general recreational users. In 2012, another
same sex twin pairs discordant for cannabis dependence to
determine relationships between cannabis use and major
depressive disorder, suicidal behavior, and suicidal ideations.
Cannabis-dependent individuals had an odds of suicide
attempt or ideation that was 2.5–2.9 times higher than their
Address correspondence to Melanie Rylander, MD, Departments of
non-cannabis-dependent twin. Those who initiated use prior
Behavioral Health and Internal Medicine, Denver Health, 777 Bannock
Street, MC 0490, Denver, CO 80204-4507, USA. Tel: +1 303 602 6938. to age 17 had 3.5 times the rate of subsequent suicide
Fax: +1 303 602 6930. E-mail: melanie.rylander@dhha.org attempts (5). Pedersen et al. also observed increased rates of
270 M. Rylander et al.
suicide attempts in cannabis users even after controlling for
confounders (6). Rasic et al. found no association between
cannabis use and suicide amongst high school student
though did observe an association between heavy use and
depression but not suicidal ideations or attempts (7). In a
30-year longitudinal study, van Ours et al. found that intensive
cannabis use, defined as several times per week, was
associated with later onset of suicidal ideations amongst
males but not females after controlling for confounders (8).
However, not all studies support these findings. In a
longitudinal study of over 50,000 men, Price et al. found no
association between cannabis use and suicides after control-
ling for psychological and behavior problems (9). The
positive associations are particularly concerning for a state
like Colorado, where the prevalence of both completed
suicides and marijuana use exceed the national averages.
Between 2000 and 2010, the age-adjusted national rate
of completed suicide was 11.2 per 100,000; in Colorado,
the rate was 16.3 per 100,000 (10,11). Thus, the relationship
between marijuana use and suicide remains unclear.
Differences in study populations, designs, and controlled
confounders likely explain much of the discrepancy in
findings.
To date, no studies have examined the relationships
between medical marijuana use and completed suicides in
Colorado. The legalization of medical marijuana may increase
the frequency and intensity of marijuana use (12). States that
have legalized medical marijuana have been found to have
higher rates of marijuana use. Given the concerns raised about
potential adverse psychiatric effects of marijuana use, this
may increase psychiatric morbidity and mortality. However, a
recent study by Anderson et al. found that legalization of
medical marijuana across all states was not associated with
increased suicide rates (13).
Colorado is an ideal state to explore the association
between suicide and marijuana use, not only because of
the prevalence of medical marijuana use and completed
suicide, but because the state regulates the use of medical
marijuana. The Colorado Department of Public Health and
Environment (CDPHE) has maintained a registry of medical
marijuana users since 2000, which provides anonymous
demographic information about registrants. In addition, the
CDPHE also maintains a database of completed suicides.
Examining correlations between these data sets after
controlling for other known risk factors for suicide may
give insight into if and how medical marijuana use impacts
the prevalence of completed suicide. Furthermore, because
medical marijuana users may be similar to the general
population and recreational users (14), the impact of
medical marijuana use on suicide rates may have implica-
tions for the general population in a state with legalized
recreational marijuana. In addition to established risk factors
such as unemployment and living in a rural community, we
also controlled for elevation as this has been associated with
suicide even after controlling for traditional risk factors such
as age, male gender, unemployment, and access to firearms
(15,16). In a state with mountainous regions such as
Colorado, altitude may be a significant risk factor for
suicide that needs to be incorporated into statistical models
(15,16).
Am J Drug Alcohol Abuse, 2014; 40(4): 269–273
Methods
Data collection and sources
The CDPHE provided county-level data for the number of
suicides per year, number of suicide-related hospitalizations,
mechanism of suicide death, gender of suicide completers and
age, and number of individuals on the medical marijuana
registry for all 64 counties in Colorado.
Medical marijuana registrants were used as a proxy for
marijuana use. Although the medical use of marijuana was
permitted in November 2000, we restricted our analysis
to CDPHE data gathered from 2004–2010. Prior to 2004,
the collection and recording of data from the medical
marijuana registry was irregular, leaving several gaps in
the data which rendered its analysis unfeasible. During 2011,
the number of registrants began to decline without clear
explanation. The CDPHE speculates (personal communica-
tion) that the decline was due to patients delaying renewal
because they anticipated decreased registration fees and
concerns that the confidentiality of the registry was
compromised and might affect employment or the ability to
purchase firearms or ammunition. Since we could not assess
these possibilities, we excluded all data after 2010 because the
irregularities in the data after 2010 made it unreliable for
analysis.
In addition, we used the number of medical marijuana
dispensaries per county as a proxy for use in that county.
Addresses for marijuana dispensaries were matched to
individual counties by zip code. Unemployment data were
obtained from Colorado Department of Labor and
Employment (CDLE), and the mean elevation of individual
counties was also obtained from the National Geographic
Survey (NGS). Designation of counties as urban versus rural
was obtained from CDPHE.
Data analysis
Analysis variables included total medical marijuana regis-
trants, medical marijuana dispensaries per county, total
suicide deaths, mechanism of suicide death (firearm, hanging,
poison, other), gender, total suicide hospitalizations, total
unemployment, and county-level information such as mean
elevation and whether the county was urban or rural. Total
medical marijuana registrants per county, per year and
medical marijuana dispensaries per county were the main
explanatory variables and total suicide deaths was the main
dependent variable, total unemployment was a secondary
explanatory variable. Total registrants, total suicide deaths,
total suicide hospitalizations, total unemployment, and mean
county elevation were all count data, whereas county type
(rural/urban) was categorical.
Data were first examined for any existing associations.
This was done by stratifying count data by year and analyzing
the correlation between two variables at a time. Count data
had to be log-transformed in order to make them normal and
meet the criteria for correlation analyses to be performed.
Analysis was performed on count and categorical data for
all Colorado counties and included years. All analyses were
conducted using SAS Statistical Software Version 9.3
(SAS Institute, Cary, NC, USA). Analysis was performed
DOI: 10.3109/00952990.2014.910520 Correlations between medical marijuana and suicide 271
Figure 1. Medical Marijuana Registrants in 140000 1000
Colorado from 2004–2010 (primary axis) and
Completed Suicides in Colorado from 900
2004–2010 (secondary axis). Source: 120000
CDPHE. 800

100000 700

600

Suicide Deaths
80000

Registrants
500
60000
400

40000 300
Total Registrants
Total Suicide Deaths 200
20000
100

0 0
2004 2005 2006 2007 2008 2009 2010

Table 1. The range, means, and standard deviations for completed with mixed model Poisson regression using generalized linear
suicides, medical marijuana registrants, and unemployment. modeling techniques.
Year Variable Range Mean SD
Results
2004 Total suicide deaths 0–121 12.1 25.9
Registrants 0–83 8.3 14.5 Figure 1 shows the number of individuals on the medical
Total unemployment 20–20 082 2260.4 4678.2
marijuana registry from 2004–2010 as well as the total
2005 Total suicide deaths 0–106 12.1 25.5
Registrants 0–95 11.7 18.7
number of suicides in Colorado over the same time.
Total unemployment 17–17 615 2103.8 4299.8 Registrants steadily increased up until 2009 when there was
2006 Total suicide deaths 0–92 11.0 21.7 an exponential increase in registrants, following the federal
Registrants 0–116 16.6 27.9 government’s announcement that they would not pursue
Total unemployment 19–14 829 1804.2 3668.0 prosecution. During the same period, completed suicides
2007 Total suicide deaths 0–100 12.5 25.5 ranged from 792–940 per 100,000. Table 1 shows the
Registrants 0–253 32.8 59.6 range, means, and standard deviations for completed
Total unemployment 18–13 317 1612.5 3305.6
suicides, medical marijuana registrants, and unemployment.
2008 Total suicide deaths 0–106 12.3 25.0
Registrants 0–618 76.7 151.3 Table 2 displays the unadjusted Pearson Correlation
Total unemployment 21–17 340 2079.5 4285.8 Coefficients by year for each of the key log-transformed
2009 Total suicide deaths 0–131 14.5 28.9 variables. All correlations were strong and statistically
Registrants 3–6053 651.0 1329.8 significant with the correlation between unemployment
Total unemployment 23–29 437 3526.6 7163.1 and medical marijuana use being the highest. However,
2010 Total suicide deaths 0–115 13.5 26.6 the correlation between suicide deaths per year per county
Registrants 13–17 181 1842.8 3696.0
Total unemployment 27–32 085 3866.4 7810.9 and medical marijuana registrants per year per county dis-
appeared when unemployment rates per year per county
SD, Standard deviation. were factored in (Table 3). When adjusting for medical
marijuana registrants, the correlation between unemployment
rates and suicides remained significant (Table 3). Separating
Table 2. Pearson Correlation Coefficients by year for two-way suicides by age, gender, or method did not alter our results.
correlations between each of the key variables.
After adjusting for unemployment, mean county elevation,
and urban versus rural county status, medical marijuana
2004 2005 2006 2007 2008 2009 2010
registrants per year was not a significant predictor of
log10(Suicide deaths)  0.41 0.53 0.60 0.61 0.68 0.70 0.70 suicide rates (p ¼ 0.13) (Table 3). Urban counties were
log10(MMJ registrants)
log10(Suicide deaths)  0.73 0.73 0.74 0.75 0.77 0.79 0.71 associated with a higher suicide rate (p50.001) after
log10(Unemployment) adjusting for unemployment, mean county elevation, and
log10(MMJ registrants)  0.80 0.83 0.84 0.86 0.91 0.94 0.96 medical marijuana registrants. Mean elevation by county was
log10(Unemployment) associated with lower suicide rates but this association
All correlations were significant at the  ¼ 0.05 level. MMJ, Medical disappeared after adjusting for urban versus rural status,
marijuana. unemployment, and medical marijuana registrants (Table 4).
There was no relationship between maximum or mean
272 M. Rylander et al.
Table 3. Pearson Partial Correlation Coefficients by year for two-way correlations between each of the key variables.
2004
log10(Suicide deaths)  log10 0.32 0.05
(medical marijuana registrants) (p ¼ 0.31) (p ¼ 0.80)
Partial: log10(Unemployment)
log10(Suicide deaths)  log10 0.65
(Unemployment) (p50.001) (p ¼ 0.01)
Partial: log10(marijuana registrants)
Table 4. Results for mixed model Poisson regression for completed
suicides.
Type III tests of fixed effects
Effect Num DF Den DF F value
Year 6 430 6.7
MMJ registrants 1 430 2.35
Total unemployment 1 430 7.64
Maximum county elevation 1 52.91 1.33
Urban county 1 45.18 40.3
MMJ, Medical marijuana.
county elevation and number of medical marijuana registrants
(Table 5).
The medical marijuana registry is only a by-proxy
measure of medical marijuana use, because a person could
register for permission to use medical marijuana without
consuming marijuana and some dispensaries may inconsist-
ently verify registry cards. A subsequent analysis exploring
the relationship between the number of medical marijuana
dispensaries per county and per county suicides did not
identify a significant relationship between the number of
marijuana dispensaries per county and suicides.
Discussion
Our results did not show a significant correlation between the
number of medical marijuana registrants and suicides.
Although prior studies have shown positive correlations
with marijuana use and suicide attempts and suicidal ideations
even after adjusting for anxiety, depression, and stressful life
events, the role of marijuana use in completed suicides
remains unclear (6,8). The difficulty in conducting research
on the role of medical marijuana use in suicide completions is
a reflection of the small number of suicides, limitations in
obtaining accurate information on marijuana use in deceased
subjects, and controlling for confounding factors such as
comorbid substance use, depression and anxiety. While
previous studies have found that marijuana use was a
significant independent risk factor for suicidal behaviors in
high school students after controlling for anxiety and
depression (17), suicide completers represents a different
population than those engaged in suicidal behaviors, and it is
unclear whether this data can be extrapolated. Additionally,
medical marijuana users may be different from general users
though data suggests that medical marijuana users are not
different from the general population (14).
The lack of correlation between unemployment and
suicides after adjusting for county elevation and urban
versus rural status was surprising and not readily explainable.
Similarly, the lack of correlation between county elevation
Am J Drug Alcohol Abuse, 2014; 40(4): 269–273
2005 2006 2007 2008 2009 2010
0.04 0.08 0.01 0.08 0.11
(p ¼ 0.82) (p ¼ 0.68) (p ¼ 0.98) (p ¼ 0.63) (p ¼ 0.53)
0.51 0.53 0.55 0.50 0.52 0.20
(p ¼ 0.003) (p ¼ 0.002) (p ¼ 0.004) (p ¼ 0.001) (p ¼ 0.25)
and suicides was also surprising given prior studies showing
positive correlations after adjusting for age, gender, income,
and gun ownership (15,16). These studies focused on state-by-
state data whereby our data was restricted to the state of
Pr4F Colorado. In the absence of attempts to replicate this finding,
50.0001
it is unclear if the restriction to one state alters this association
0.1256 or if there is a unique confounder in Colorado.
0.006 Our study did show significant positive correlations with
0.2534 medical marijuana registrants and unemployment. Analysis of
50.0001
medical marijuana use on gaining and maintaining employ-
ment is beyond the scope of this study. However, this
relationship warrants further investigation particularly in light
of the recent legalization of marijuana.
There are several limitations to our study. The first is that
the medical marijuana registry is a proxy measure of marijuana
use. At this time, it is not possible to obtain accurate data
about the number of marijuana users in Colorado or to
quantify that use. Individuals on the medical marijuana registry
may not represent marijuana users in the general populations.
However, there has been speculation that because of the
broad eligibility criteria and minimal oversight of the methods
used to deem medical marijuana necessary, that registrants
often are similar to, if not the same as, recreational users (2,3).
We attempted to mitigate this limitation by also using the
number of medical marijuana dispensaries per county and
obtained the same results. The number of dispensaries is also
an imperfect by-proxy measure, as it may be more related to
access than use. However, given the difficulty of an accurate
measure of the true number of medical marijuana users, we had
to use imperfect markers. Our results were the same with both
by proxy measures of use. We were also unable to control for
psychiatric comorbidities including substance use, as this
information is not available in public records. Additionally, the
number of recorded suicides suffers limitations inherent to
retrospective analysis of records. It is likely not an accurate
record of the total suicides committed secondary to misclassi-
fication of cause of death. However, it is the most accurate
record available.
This study attempted to draw correlations between num-
bers of people on the medical marijuana registry as a proxy
measure of marijuana use and completed suicides. No such
correlation was observed after controlling for confounders. At
face value, this may seem somewhat reassuring given the
presumed increase in marijuana use with its recent legaliza-
tion. However, several prior studies have shown positive
associations between suicidal ideations, attempts and mari-
juana use. Given limitations in data sources, a prospective
design may be needed to better quantify this risk in light of
the recent legalization of the recreational use of marijuana in
Colorado.
DOI: 10.3109/00952990.2014.910520 Correlations between medical marijuana and suicide 273
Table 5. Pearson Correlation Coefficients between suicide deaths and mean county elevation.

Correlation between suicide deaths and mean county elevation

Pearson correlation Variables 2004 2005 2006 2007 2008 2009 2010
Unadjusted log10(Suicide deaths)  0.36 0.15 0.45 0.50 0.37 0.35 0.26
Mean county elevation (p ¼ 0.04) (p ¼ 0.45) (p ¼ 0.01) (p ¼ 0.004) (p ¼ 0.04) (p ¼ 0.04) (p ¼ 0.13)
Adjusted (partial) log10(Suicide deaths)  0.14 0.03 0.08 0.23 0.08 0.09 0.09
Mean county elevation (p ¼ 0.50) (p ¼ 0.90) (p ¼ 0.68) (p ¼ 0.23) (p ¼ 0.68) (p ¼ 0.63) (p ¼ 0.62)
Partial: log10(MMJ registrants)
log10(Unemployment)
Urban county (0/1)

Correlation between marijuana registrants and county elevation (maximum and mean)
Pearson correlation Variables 2004 2005 2006 2007 2008 2009 2010

Unadjusted log10(MMJ registrants)  0.00 0.00 0.05 0.14 0.18 0.20 0.24
Maximum county elevation (p ¼ 0.99) (p ¼ 0.99) (p ¼ 0.76) (p ¼ 0.32) (p ¼ 0.16) (p ¼ 0.11) (p ¼ 0.06)
Unadjusted log10(MMJ registrants)  0.11 0.16 0.03 0.01 0.06 0.04 0.06
Mean county elevation (p ¼ 0.48) (p ¼ 0.31) (p ¼ 0.84) (p ¼ 0.97) (p ¼ 0.62) (p ¼ 0.77) (p ¼ 0.62)

MMJ, Medical marijuana.

Acknowledgements 7. Rasic D, Weerasinghe S, Asbridge M, Langille DB.

The authors thank the Colorado Department of Public Health
and Environment (CDPHE) for providing additional data.
Declaration of interest
The authors report no conflicts of interest. The authors alone
are responsible for the content and writing of this paper.
References
1. Ogden DW. Memorandum for selected United States Attorneys:
Investigations and Prosecutions in States Authorizing the Medical
Use of Marijuana. Available from http://www.justice.gov/opa/pr/
2009/October/09-ag-1119.html [last accessed 19 Oct 2009].
2. Nussbaum AM, Boyer JA, Kondrad EC. ‘‘But my doctor recom-
mended pot’’: medical marijuana and the patient-physician
relationship. J Gen Int Med 2011;26:1364–1367.
3. Nussbaum AM, Thurstone C, Binswanger IA. Medical marijuana
use and suicide attempt in a patient with major depressive disorder.
Am J Psychiatry 2011;168:778–781.
4. Degenhardt L, Hall W, Lynskey M. Exploring the association
between cannabis use and depression. Addiction 2003;98:
1493–1504.
5. Lynskey MT, Glowinski AL, Todorov AA, Bucholz KK,
Madden PA, Nelson EC, Statham DJ, et al. Major depressive
disorder, suicidal ideation, and suicide attempt in twins discordant
for cannabis dependence and early-onset cannabis use. Arch Gen
Psychiatry 2004;61:1026–1032.
6. Pedersen W. Does cannabis use lead to depression and suicidal
behaviours? A population-based longitudinal study. Acta Psychiatr
Scand 2008;118:395–403.
Longitudinal associations of cannabis and illicit drug use with
depression, suicidal ideation and suicidal attempts among Nova
Scotia high school students. Drug Alcohol Depend 2013;129:
49–53.
8. van Ours JC, Williams J, Fergusson D, Horwood LJ. Cannabis use
and suicidal ideation. J Health Econ 2013;32:524–537.
9. Price C, Hemmingsson T, Lewis G, Zammit S, Allebeck P.
Cannabis and suicide: longitudinal study. Br J Psych 2009;195:
492–497.
10. Colorado Department of Public Health and Environment.
Completed suicides by method 2000–2010. Health Statistics
Section.
11. Center for Disease Control. 2000–2010, United States suicide
injury deaths and rates per 100,000. All races, both sexes, all ages.
QISWARS Injury Mortality Report.
12. Friese B, Grube JW. Legalization of medical marijuana and
marijuana use among youths. Drugs 2013;20:33–39.
13. Anderson DM, Rees D, Sabia J. Medial marijuana laws and suicides
by gender and age. Am J Public Health 2014. Jan 16 Epub ahead of
print.
14. Reinarman C, Nunberg H, Lanthier F, Heddleston T. Who
are medical marijuana patients? Population characteristics from
nine California assessment clinics. J Pyschoact Drugs 2011;43:
128–135.
15. Brenner B, Cheng D, Clark S, Camargo CA. Positive association
between altitude and suicide in 2584 U.S. counties. High Alt Med
Biol 2011;12:31–35.
16. Kim N, Mickelson JB, Brenner BE, Haws CA, Yurgelun-Todd DA,
Renshaw PF. Altitude, gun ownership, rural areas, and suicide.
Am J Psychiatry 2011;168:49–54.
17. Chabrol H, Chauchard E, Girabet J. Cannabis use and suicidal
behaviours in high-school students. Addict Behav 2008;33:
152–155.
Copyright of American Journal of Drug & Alcohol Abuse is the property of Taylor & Francis
Ltd and its content may not be copied or emailed to multiple sites or posted to a listserv
without the copyright holder's express written permission. However, users may print,
download, or email articles for individual use.