TOBACCO USE AND SUICIDE RELATED OUTCOMES
By
Jonathan P. Troost

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
Epidemiology - Doctor of Philosophy
2014

ABSTRACT
TOBACCO USE AND SUICIDE RELATED OUTCOMES
By
Jonathan P. Troost
There is a growing body of evidence which suggests that tobacco cigarette smoking
(hereinafter, simply ‘smoking’) is an independent predictor of suicide as well as suicidal
behaviors such as suicidal ideation (thinking about committing suicide), suicidal planning, and
suicide attempts. Such evidence is by no means overwhelming and the possibility of this being a
causal association has yet to be determined. Still, many contributors in this area have proposed
that cigarette smoking could affect suicide and suicide related outcomes through biological
effects on serotonin metabolism. If nicotine and/or other chemicals found in tobacco are
indeed an etiologic link between cigarette smoking and suicide, then one would expect other
types of tobacco use to predict suicide and suicidal behaviors as well. Using US nationally
representative cross-sectional data from the National Survey on Drug Use and Health (NSDUH)
and longitudinal data from the National Epidemiological Study on Alcohol and Alcohol-Related
Conditions (NESARC) I will test for the independent effects of cigarette smoking and smokeless
tobacco use on subsequent suicide related outcomes with the hypothesis that cigarette
smoking will predict suicidal behaviors while smokeless tobacco use will not. This is under the
assumption that the possible causal effects of cigarette smoking on suicide risk have nothing to
do with either the combustion of tobacco, the efficiency or speed with which nicotine or other
agents would reach the brain, or other pharmacological differences between smoking tobacco
and using smokeless tobacco. I will also challenge the smoking-suicide hypothesis by controlling

for mental health co-morbidities using data from the Collaborative Psychiatric Epidemiology
Surveys (CPES), in the US.
Summary:
A consistent finding within this dissertation is that smokeless tobacco use does not
appear to alter the estimated risk of suicide related outcomes while cigarette smoking does.
Also, while a considerable proportion of the relationship between cigarette smoking and suicide
is explained by the presence of mental disorders, this work suggests smokers are still at
increased odds of suicide regardless of their mental health. An important implication of these
findings is that quitting cigarette smoking could potentially decrease one’s risk of suicide. This
dissertation found no evidence of those who recently quit smoking to be at a higher risk for
suicide related outcomes, but future studies should build on this work by tracking changes in
suicide risk in the proximal weeks and months following smoking cessation. Studies on nonUnited States or Western populations are also needed as the evidence in favor of a positive
association between cigarette smoking and suicide is currently western-centric, and if a true
biological association exists, one would expect to observe it in all countries where smokers are
found

ACKNOWLEDGEMENTS

I would like to thank my dissertation committee members: Drs. James C. Anthony,
Naomi Breslau, Qing Lu, and Carlos Rios-Bedoya.
In addition to the members of the dissertation guidance committee, I would also like to
acknowledge the following individuals in the Department of Epidemiology & Biostatistics for
their input on dissertation content, support and general advice during my graduate education:
Mat Reeves, Brian Fairman, Catalina Lopez and David Barondess as well as Nancy Bieber for her
constant assistance as the department’s Graduate Secretary. I would also like to acknowledge
the following funding from the National Institute on Drug Abuse which was integral to funding
my graduate education: NIDA T32DA021129; K05DA15799.

iv

TABLE OF CONTENTS

LIST OF TABLES…………………………………………………………………………………………………………………………vii
LIST OF FIGURES………………………………………………………………………………………………………………………. xi
CHAPTER 1: INTRODUCTION AND AIMS……………………………………………………………………………………1
CHAPTER 2: BACKGROUND……………………………………………………………..………………………………………..3
Biological mechanism:..………………………………………………………………………………………………….7
CHAPTER 3: A PROSPECTIVE COHORT STUDY ON THE EFFECTS OF TOBACCO USE ON SUICIDE
RELATED OUTCOMES………………………………………………………………………………………………………………14
Chapter 3 Aim:..……………………………………………………………………………………………………………14
Chapter 3 Methods:..……………………………………………………………………………………………………14
Data and participants……………………………………………………………………………………….14
Assessment of exposure: tobacco variables…………………………………………………......15
Assessment of the outcome: suicide-related outcome variables………………………..15
Data analysis…………………………………………………………………………………………………….16
Chapter 3 Results:………………………………………………………………………………………………………18
CHAPTER 4: A NATIONALLY REPRESENTATIVE CROSS-SECTIONAL STUDY ON THE
RELATIONSHIP BETWEEN TOBACCO USE AND PAST YEAR SUICIDE RELATED OUTCOMES……..22
Chapter 4 Aim:..……………………………………………………………………………………………………………22
Chapter 4 Methods:..…………………………………………………………………………………………………...22
Data and participants……………………………………………………………………………………….22
Assessment of exposure: tobacco variables………………………………………………………23
Assessment of the outcome: suicide-related outcome variables………………………..24
NESARC vs. NSDUH……………………………………………………………………………………………26
Chapter 4 Results:.……………………………………………………………………………………………………….27
CHAPTER 5: THE RELATIONSHIP BETWEEN CIGARETTE SMOKING AND SUICIDE RELATED
OUTCOMES: WHAT CAN WE ATTRIBUTE TO MENTAL HEALTH AND SOCIAL SUPPORT?...........35
Chapter 5 Aim:..……………………………………………………………………………………………………………35
Chapter 5 Methods:..……………………………………………………………………………………………………35
Data and participants…………………………………………………………………………………………….35
Assessment of exposure: tobacco variables…………………………………………………………….36
Assessment of the outcome: suicide-related outcome variables……………………………..36
Data analysis…………………………………………………………………………………………………………38
v

Chapter 5 Results:..………………………………………………………………………………………………………40
CHAPTER 6: DISCUSSION, LIMITATIONS, & FUTURE RESEARCH…………………………………….………..45
CHAPTER 7: SUMMARY……………………………………………………………………………………………………………57
APPENDICES…………………………………………………………………………………………………………………………...58
APPENDIX A: Tables and Figures......................................................................................59
APPENDIX B: Educational background/other research experiences………………………….116
APPENDIX C: List of non-dissertation related abstracts…………………………………………….119
BIBLIOGRAPHY………………………………………………………………………………………………………………………127

vi

LIST OF TABLES

Table 3.1 Distribution of newly incident cases of suicide attempt by covariates of interest. Data
from the US National Epidemiologic Survey on Alcohol and Alcohol Related Conditions
(NESARC), Wave I 2001-2002 (n=43,093), Wave II 2004-2005 (n=33,687)………………………………..62
Table 3.2 Distribution of newly incident cases of suicidal ideation by covariates of interest. Data
from the US National Epidemiologic Survey on Alcohol and Alcohol Related Conditions
(NESARC), Wave I 2001-2002 (n=43,093), Wave II 2004 2005(n=31,544).………………………………..64
Table 3.3 Distribution of newly incident cases of suicidal behavior (suicidal ideation and
attempts combined by covariates of interest. Data from the US National Epidemiologic Survey
on Alcohol and Alcohol Related Conditions (NESARC), Wave I 2001-2002 (n=43,093), Wave II
2004-2005(n=34,513).……………………………………………………………………………………………………………..66
Table 3.4 Unadjusted and adjusted models for the associations between covariates of interest
and newly incident suicide attempts. Data from the US National Epidemiologic Survey on
Alcohol and Alcohol Related Conditions (NESARC), Wave I 2001-2002 (n=43,093), Wave II 20042005 (n=34,653).……………………………………………………………………………………………………………………..68
Table 3.5 Unadjusted and adjusted models for the associations between covariates of interest
and newly incident suicide attempts. Data from the US National Epidemiologic Survey on
Alcohol and Alcohol Related Conditions (NESARC), Wave I 2001-2002 (n=43,093), Wave II 20042005 (n=34,653).……………………………………………………………………………………………………………………..70
Table 3.6 Unadjusted and adjusted models for the associations between covariates of interest
and newly incident suicidal ideation. Data from the US National Epidemiologic Survey on
Alcohol and Alcohol Related Conditions (NESARC), Wave I 2001-2002 (n=43,093), Wave II 20042005 (n=34,653).……………………………………………………………………………………………………………………..72
Table 3.7 Unadjusted and adjusted models for the associations between covariates of interest
and newly incident suicidal ideation. Data from the US National Epidemiologic Survey on
Alcohol and Alcohol Related Conditions (NESARC), Wave I 2001-2002 (n=43,093), Wave II 20042005 (n=34,653).……………………………………………………………………………………………………………………..74

vii

Table 3.8 Unadjusted and adjusted models for the associations between covariates of interest
and newly incident suicidal behavior. Data from the US National Epidemiologic Survey on
Alcohol and Alcohol Related Conditions (NESARC), Wave I 2001-2002 (n=43,093), Wave II 20042005 (n=34,653).……………………………………………………………………………………………………………………..76
Table 3.9 Unadjusted and adjusted models for the associations between covariates of interest
and newly incident suicidal behavior. Data from the US National Epidemiologic Survey on
Alcohol and Alcohol Related Conditions (NESARC), Wave I 2001-2002 (n=43,093), Wave II 20042005 (n=34,653).……………………………………………………………………………………………………………………..78
Table 3.10 Models for the associations between covariates of interest and occurrence of
suicidal behavior. Data from the US National Epidemiologic Survey on Alcohol and Alcohol
Related Conditions (NESARC), Wave I 2001-2002 (n=43,093), Wave II 2004-2005 (n=34,653).…80
Table 4.1 Suicidal behaviors among current, former and never tobacco users. Data from the US
National Survey on Drug Use and Health (NSDUH), 2008-2010 (n=114,048).……………………………81
Table 4.2 Characteristics of the sample by lifetime tobacco use………………………………………………82
Table 4.3 Tobacco use characteristics of the sample.…………………………………………………..………….84
Table 4.4 Crude and covariate adjustment logistic regressions predicting past year suicidal
related outcomes by recency of tobacco use. Data from the US National Survey on Drug Use
and Health (NSDUH), 2008-2010 (n=114,048).…………………………………………………………………………85
Table 4.5. Logistic regressions predicting past year suicidal related outcomes by recency of
tobacco use stratified by history of major depression†. Data from the US National Survey on
Drug Use and Health (NSDUH), 2008-2010 (n=114,048).………………………………………………………….87
Table 4.6. Logistic regressions predicting past year suicidal related outcomes by recency of
tobacco use stratified by history of major depression†. Data from the US National Survey on
Drug Use and Health (NSDUH), 2008-2010 (n=114,048).…………………………………………….……………89
Table 4.7. Logistic regressions predicting past year suicidal related outcomes by recency of
tobacco use stratified by past year alcohol use disorders†. Data from the US National Survey on
Drug Use and Health (NSDUH), 2008-2010 (n=114,048).…………………………………………………………91
Table 4.8 Mean number of days used tobacco in the past 30 days by past year suicide related
outcome status. Data from the US National Survey on Drug Use and Health (NSDUH), 20082010
(n=114,048).……………………………………………………………………………………………………………………………93

viii

Table 4.9 Significance tests for a possible dose response between past year suicidal related
outcomes and number of days used tobacco in the past 30 days. Data from the US National
Survey on Drug Use and Health (NSDUH), 2008-2010* …………………………………………………………..94
Table 4.10 Frequency of suicide related outcomes among tobacco users who are either
exclusive cigarette smokers or exclusive smokeless tobacco users. Data from the US National
Survey on Drug Use and Health (NSDUH), 2008-2010……………………………………………………………..95
Table 4.11 Frequency of suicide related outcomes among tobacco users who currently use only
one type of tobacco. Data from the US National Survey on Drug Use and Health (NSDUH), 20082010……………………………………………………………………………………………………………………………………….96
Table 5.1 Descriptive characteristics of current, formers and never smokers. Data from the
Collaborative Psychiatric Epidemiology Surveys (CPES), 2001-2003 (n=10,341).……………………..97
Table 5.2 Lifetime diagnosis of mental disorders by smoking status. Data from the Collaborative
Psychiatric Epidemiology Surveys (CPES), 2001-2003 (n=10,341).………………….……………………….98
Table 5.3 Social support assessment items and response proportions. Data from the
Collaborative Psychiatric Epidemiology Surveys (CPES), 2001-2003 (n=10,341).……… ……………100
Table 5.4 Social support assessment items and response proportions: exploratory factor
analysis sample. Data from the Collaborative Psychiatric Epidemiology Surveys (CPES), 20012003 (n=4,649).……………………………………………………………………………………………………………………101
Table 5.5 Social support assessment items and response proportions: confirmatory factor
analysis sample. Data from the Collaborative Psychiatric Epidemiology Surveys (CPES), 20012003 (n=5,692).……………………………………………………………………………………………………………………..102
Table 5.6 Estimated associations between cigarette smoking and suicide related outcomes:
unadjusted and adjusted for descriptive characteristics. Data from the Collaborative Psychiatric
Epidemiology Surveys (CPES), 2001-2003 (n=10,341).……………………………………………………………103
Table 5.7 Estimated associations between cigarette smoking and suicide related outcomes:
unadjusted and adjusted for mental disorders (bi-variable). Data from the Collaborative
Psychiatric Epidemiology Surveys (CPES), 2001-2003 (n=10,341).…………………………….……………104
Table 5.8 Estimated associations between cigarette smoking and suicide related outcomes:
unadjusted and adjusted for mental disorders(multi-variable). Data from the Collaborative
Psychiatric Epidemiology Surveys (CPES), 2001-2003 (n=10,341).………………………………………….106
Table 5.9 Model fit indices for factor structure of 12 item social support scale. Data from the
Collaborative Psychiatric Epidemiology Surveys (CPES), 2001-2003 (n=4,649).………………………108
ix

Table 5.10 Results from exploratory factor analysis of 12 social support items. Data from the
Collaborative Psychiatric Epidemiology Surveys (CPES), 2001-2003 (n=4,649).………………………109
Table 5.11 Results of a confirmatory factor analysis of a 3 factor solution for 12 social support
items. Data from the Collaborative Psychiatric Epidemiology Surveys (CPES), 2001-2003
(n=5,692)……………………………………………………………………………………………………………………………….111
Table 5.12 Distributions of social support factor scores. Data from the Collaborative Psychiatric
Epidemiology Surveys (CPES), 2001-2003 (n=10,341).……………………………………………………………112
Table 5.13 Estimated associations between cigarette smoking and suicide related outcomes:
unadjusted and adjusted for social support (multi-variable). Data from the Collaborative
Psychiatric Epidemiology Surveys (CPES), 2001-2003 (n=10,341).………………………………………….113
Table 5.14 Estimated associations between cigarette smoking and suicide related outcomes:
unadjusted and adjusted for social support, descriptive characteristics and mental disorders.
Data from the Collaborative Psychiatric Epidemiology Surveys (CPES), 2001-2003(n=10,341).114
Table 5.15 Exploratory analysis of subgroup variation in the effects of cigarette smoking on
suicide related outcomes by mental disorder status. Data from the Collaborative Psychiatric
Epidemiology Surveys (CPES), 2001-2003 (n=10,341).……………………………………………………………115

x

LIST OF FIGURES

Figure 3.1 Flow chart identifying newly incident cases of suicide attempts with complete followup. Data from the US National Epidemiologic Survey on Alcohol and Alcohol Related Conditions
(NESARC), Wave I 2001-2002 (n=43,093), Wave II 2004-2005 (n=34,653)……..…………………………59
Figure 3.2 Flow chart identifying newly incident cases of suicidal ideation with complete followup. Data from the US National Epidemiologic Survey on Alcohol and Alcohol Related Conditions
(NESARC), Wave I 2001-2002 (n=43,093), Wave II 2004-2005 (n=34,653).………….......................60
Figure 3.3 Flow chart identifying newly incident cases of suicidal behavior (suicidal ideation and
attempts combined) with complete follow-up. Data from the US National Epidemiologic
Surveyon Alcohol and Alcohol Related Conditions (NESARC), Wave I 2001-2002 (n=43,093),
Wave II 2004-2005 (n=34,653)…………………………………………………………………………………………………61

xi

CHAPTER 1: INTRODUCTION AND AIMS
Before a thorough review of the literature and detailed description of the methodology,
I will briefly describe this dissertation work. There is a growing body of evidence which suggests
that tobacco cigarette smoking (hereinafter, simply ‘smoking’) is an independent predictor of
suicide as well as suicidal behaviors such as suicidal ideation (thinking about committing
suicide), suicidal planning, and suicide attempts. Such evidence is by no means overwhelming
and whether or not this is a causal association has yet to be determined. Still, many
contributors in this research topic have proposed that cigarette smoking could affect suicide
and other suicide related outcomes through its biological effects on serotonin metabolism.
This dissertation will challenge the ‘smoking-suicide’ hypothesis in the following ways:
(1) If nicotine and/or other chemicals found in tobacco are indeed the etiologic
link between cigarette smoking and suicide, then one would expect other
types of tobacco use to predict suicide and suicidal behaviors as well. Using
US nationally representative cross-sectional data from the National Survey
on Drug Use and Health (NSDUH) and longitudinal data from the National
Epidemiological Study on Alcohol and Alcohol-Related Conditions (NESARC) I
will test for the independent effects of cigarette smoking and smokeless
tobacco use on subsequent suicide related outcomes with the hypothesis
that cigarette smoking will predict suicidal behaviors while smokeless
tobacco use will not. This is under the assumption that the possible causal
effects of cigarette smoking on suicide risk have nothing to do with either the
combustion of tobacco, the efficiency or speed of which nicotine or other
1

agents would reach the brain, or other pharmacological differences between
smoking tobacco and using smokeless tobacco.
(2) Smokers are more likely to have mental health comorbidities (such as major
depression and bipolar disorder) which are known, important causes of
suicide. Previous research assessing if the observed relationship between
cigarette smoking and suicide is in fact confounded by mental disorder status
has been met with mixed results. I will test for the effects of cigarette
smoking on suicide related outcomes while controlling for the diagnosis of
several mental disorders using data from the Collaborative Psychiatric
Epidemiology Surveys.

2

CHAPTER 2: BACKGROUND
More deaths are caused by suicide each year than from liver, colon and stomach cancers
combined. An estimated 1 million people commit suicide in the world each year, accounting for
2% of all deaths.(WHO) Suicide rates have increased by 60% in the past 45 years, and it is the
second leading cause of death among those 10-24 years old.(Mathers & Loncar, 2006)
Additionally, non-fatal suicide attempts are a burden by themselves. In the United States
specifically, half a million individuals require medical care after a suicide attempt at an average
cost of $33,000 per year.(E.K. Moscicki, 1995)
Smoking has been shown to be associated with an increased risk of suicide in
epidemiological studies since the monumental British Doctors study first published in the
1970s.(Doll & Peto, 1976; Doll, Peto, Boreham, & Sutherland, 2004) Around the same time,
Friberg et al., found a two-fold increase of completed suicide among current smokers vs. never
smokers using data from a Swedish twin registry.(Friberg, Cederlof, Lorich, Lundman, & De
Faire, 1973) Similar results were found within the Nurses’ Health Study in the US. This study
was helpful because of its large sample and ability to control for possible confounding variables
such as chronic disease history. Cigarette smoking predicted completed suicide during followup, and a strong dose-response between number of cigarettes smoked at baseline and suicide
was also found.(Hemenway, Solnick, & Colditz, 1993)
Given the limitations of the observational data, not all were convinced of a causal
association at that time. Dr. George Davey-Smith provided an analysis on cigarette smoking and
completed suicide using data from the Multiple Risk Factor Intervention Trial (MRFIT) in the US.
Despite finding results in favor of an association that was consistent with the other studies on
3

this topic up until this point, he concluded that “it is improbable that cigarette smoking is a
direct contributory cause of suicide”. Davey-Smith found a consistent relationship between
cigarette smoking and suicide as well as cigarette smoking and homicide, and argued they were
both explained by an omitted confounding variable (or a set of confounding factors). DaveySmith suggests that while suggested biological mechanisms between smoking and suicide were
plausible, they were not likely: “A hypothesis that an exposure can cause a disease cannot be
said to have been very strongly challenged by the requirement that it should be biologically
plausible”.(G. D. Smith, Phillips, & Neaton, 1992)
Still, prospective studies on the topic continued to find a significant relationship
between smoking and completed suicide. Scandinavian population based studies, as well as a
follow-up study of Harvard Alumni in the US found dose-dependent relationships between
baseline smoking and suicide during follow-up.(Paffenbarger, Lee, & Leung, 1994; Rantakallio,
Laara, & Koiranen, 1995; Tverdal, Thelle, Stensvold, Leren, & Bjartveit, 1993) A very strong
dose-response between cigarette smoking and suicide was found by Miller et al., using data
from the Health Professionals Follow-up Study in the US.(M. Miller, Hemenway, & Rimm, 2000)
The same group of investigators found consistent results in a US study of active duty Army
soldiers.(M. Miller, Hemenway, Bell, Yore, & Amoroso, 2000)
Leistikow et al., studied this relationship using a nationally representative sample of
adults in the US and found a three-fold increase among heavy smokers (those who smoke >25
cigarettes/day) compared to never smokers.(Leistikow, Martin, & Samuels, 2000) Also, unlike
the Davey-Smith et al., contribution, this study found a specific relationship between cigarette

4

smoking and suicide, and smoking did not predict homicide or accidental injury as strongly as
suicide after statistical adjustment.
A notable limitation of the studies discussed this far is that they had limited ability to
control for background confounding variables that might influence which individuals both
become cigarette smokers and complete suicide. These studies had adjusted for basic
descriptive variables such as age, sex, race, marital status, and education as well as prior
chronic disease diagnosis (cardiovascular disease, cancer, diabetes). Important omitted
variables would include previous mental disturbances.
Another Finnish population study on this topic addressed this limitation by adjusting for
symptoms of depressed mood, stress and anxiety, and psychotropic medication prescription.
After statistical adjustment, the authors still detected a two-fold increase in suicide risk when
comparing current smokers to never smokers. (Tanskanen et al., 2000) In addition, while the
authors do not clearly explain their rationale of this, they made a stratification by type of
suicide (violent vs. non-violent) and found no difference in relative risk estimate across the two
groups. Another Finish cohort study on cigarette smoking and completed suicide found
relatively similar results after adjusting for alcohol use, social class, school performance, other
drug use, and hospital treated psychiatric disorders, although the smoking-suicide relationship
was only found in men.(Riala et al., 2007)
Iwasaki et al., conducted a prospective study on this topic among Japanese men. These
researchers categorized their smoking exposure variable as the number of ‘pack-years’ at
baseline (where a ‘pack-year’ would indicate a year of smoking 1 pack/day). This study found
5

little evidence in favor of a statistically significant association. Specifically, the only statistically
significant finding was that those with 60+ pack years at baseline were at a 2.1-fold increase
compared to those who had never smoked. This study also adjusted for list of important
confounding variables such as psychological stress, history of chronic disease and psychiatric
medication.(Iwasaki, Akechi, Uchitomi, & Tsugane, 2005) As indicated above, few studies on
this topic have been conducted outside of the US or Scandinavia. A prospective general
population study in China failed to detect even a crude (unadjusted) association between
cigarette smoking and suicide. (Rebholz et al., 2011)
It is of interest that these negative findings are from Japan and China, where tobacco
smoking is a highly prevalent behavior, especially among men. One might suspect that tobacco
smoking is not a strong source of social stigma where smoking prevalence is as large as it is in
Japan or China. Nonetheless, the idea that social stigma attached to smoking might translate
into depressed mood and suicide-related outcomes has not received substantial support. To
illustrate, studying smokers in the United States, where smokers often face stigma, Martini,
Wagner, and Anthony (2002) speculated that some of the low mood effects of tobacco smoking
might be attributable to feelings of social stigma (e.g., those associated with having failed to
quit when there is considerable social pressure to quit), and found odds of depression to be
greater among current smokers and to be lower among former smokers, with declining odds
across a gradient of elapsed time since last smoke. The research team turned to Chile, where
smoking prevalence among adolescents is above 50%, and there is little stigma attached to
smoking. Even so, in a low stigma environment, they found predicted associations consistent
with the tobacco-depression association (Caris, unpublished).
6

Other prior studies have considered the effect of cigarette smoking on suicidal related
behaviors (instead of completed suicide) which are known to predict completed suicide.(E.K.
Moscicki, 1995) Recent epidemiological studies have addressed this question in with mixed
results. Kessler et al., showed that the increased risk of suicide attempts among smokers can be
explained by background psychiatric disorders influencing the risk of starting to smoke.(Kessler
et al., 2007; Kessler, Borges, Sampson, Miller, & Nock, 2009) Conversely, two other recent
studies using US nationally representative data compare current smokers vs. former smokers
effectively holding constant factors that influence initiation of smoking.(Berlin, Covey,
Donohue, & Agostiv, 2011; Clarke et al., 2010; Clarke el al., 2008) These studies find that
current smokers are more likely to attempt suicide than those who have quit. A large
prospective study by Breslau et al., found an increased risk (RR=1.7) of suicide attempts among
currently daily smokers relative to never daily smokers. There was no increased risk among
former daily smokers.(Breslau, Schultz, Johnson, Peterson, & Davis, 2005)
When limiting the outcome to specifically completed suicide, a recent meta-analysis of
prospective studies suggests cigarette smoking to be a significant predictor of suicide. The
magnitude of this association was an estimated relative risk of 1.8 for current vs. never smokers
and an estimated 1.3 for former vs. never smokers. There was also a significant dose response
between number of cigarettes smoked per day and subsequent suicide.(Li et al., 2012)
Biological mechanism:
Many contributors have begun describing possible biological mechanisms that would
account for a relationship between smoking and suicide. While such a mechanism is currently
7

unclear, smoking is thought, by some, to affect suicide risk because of monoamine oxidase
inhibitors, such as nicotine, found in tobacco.(Berlin & Anthenelli, 2001; Fowler, Logan, Wang,
& Volkow, 2003)
Monoamine oxidase is an enzyme found in the outer mitochondrial membranes of most
cells in the human body which is responsible for, as its name implies, the oxidation (in this case,
breaking down) of amines. (Singer & Ramsay, 1995) There are two subtypes of monoamine
oxidase, MAO A and MAO B. While both types are known to oxidize dopamine, tryamine and
octopamine(Youdim & Riederer, 1993), MAO has been shown to preferentially oxidize
norephinephrine and serotonin(Johnston, 1968) while MAO B oxidizes phenethylamine.(Knoll &
Magyar, 1972)
Early human studies on this topic compared platelet levels of MAO B across smokers
and non-smokers. Oreland et al., was the first to find a significantly lower level of MAO B in
smoking women than in non-smoking women.(Oreland, Fowler, & Schalling, 1981) These
investigators also demonstrated that there was no difference in MAO levels between non and
former smokers and this was later replicated in an independent sample. (Norman, Chamberlain,
& French, 1987). A large study part of the Collaborative study on the Genetics of Alcoholism
later indicated a dose-dependent decrease of MAO B levels in current smokers where MAO B
levels decreased as number of cigarettes smoked per day increased. (Saccone et al., 1999)
Differences in monoamine oxidase B levels were also detected when measuring brain
levels instead of platelet levels using positron emission tomography techniques. Specifically,
brains of living smokers have an estimated 40% decrease in level of monoamine oxidase
8

(Fowler, Volkow, Wang, Pappas, Logan, MacGregor, et al., 1996). Former smokers, however, do
not differ in levels of monoamine oxidase B compared to never smokers, which is further
evidence suggesting that this difference is indeed a pharmacological effect and not due to
genetic factors.(Norman, et al., 1987) Smoking induced changes in serotonin levels would be
thought to affect suicide risk because lower serotonin levels are a strong predictor of suicide.
(Pandey, 1997)
Fewer studies have been done on MAO A between smokers and non-smokers because
of difficulties in measurement. However, results appear consistent with MAO B, and cigarette
smokers have decreased levels of MAO A compared to non-smokers. (Fowler, Volkow, Wang,
Pappas, Logan, Shea, et al., 1996; Geracioti et al., 1999)
As shown above, most research of this type has focused on cigarette smoking. Although
there is evidence that there are monoamine oxidase-inhibiting functions of raw tobacco plant
extracts, (Berlin & Anthenelli, 2001) it is unclear if the same clinical differences in MAO levels
are observed among smokeless tobacco users. Still, intravenous exposure to nicotine has been
shown to decrease serotonin and its metabolites in animals.(Bang & Commons, 2011; Kenny,
File, & Rattray, 2001) Similar results have emerged when intravenously exposing non-smoking
individuals to nicotine.(Newhouse et al., 1988) Smokers and smokeless tobacco users have also
been shown to have lower levels of serotonin in their cerebro-spinal fluid compared to
nonsmokers (Berggren, Eriksson, Fahlke, Blennow, & Balldin, 2007; Berggren, Fahlke, Eriksson,
& Balldin, 2003) although smokeless tobacco users and non-tobacco users did not differ with
respect to platelet MAO B activity.(Berggren, et al., 2007)

9

There are other common harmful chemicals found in both cigarettes and smokeless
tobacco. The following carcinogens currently have no suggested biological mechanisms in which
influence suicide and suicide related behaviors however, it is important to consider the
common constituents of the two products including: benzene, formaldehyde, ammonia,
acetone, arsenic, nitrosamines, polonium 210, and cadmium. Also, certain chemicals are found
in cigarette smoke but not in smokeless tobacco: hydrogen cyanide, carbon monoxide and
tar.(NCI, 1992) Of course, there are thousands of chemicals found in cigarette smoke not found
in smokeless tobacco. On one hand, this greatly complicates and seems to limit the idea of
testing for an effect of smokeless tobacco on suicide—it seems like the two types of tobacco
products are too different. However, it is important to remember that the currently
hypothesized model for smoking causing suicide is through nicotine and its effects on serotonin
regulation.
In addition to proposed biological pathways involving serotonin and MAO activity,
nicotine is also thought to alter suicide risk by its activation of the hypothalamic-pituitaryadrenal(HPA) axis in the brain (Rohleder, 2006). Animal studies have demonstrated that
nicotine alters HPA in a dose-dependent fashion (McKlveen, 2010), and clinical studies have
shown that dysregulation of the HPA axis is associated with an increased risk of suicide
attempts among depressed patients (Jokinen, 2007).
There is an important inconsistency in the literature on this topic with respect to mental
comorbidities. As described earlier, in some studies controlling for mental illnesses, such as

10

major depression, accounts for the observed association between cigarette smoking and
suicide, however, this is not always the case.
In addition to mental illness, I believe social support is another important possible
confounding variable in the relationship between cigarette smoking and suicide. Social
solidarity was first shown as an important predictor of suicide in Durkheim’s classic studies on
suicide published in 1897. Durkheim’s work showed lower rates of suicide in individuals with a
greater degree of social integration (for example, non-married individuals were at a greater risk
of suicide compared to married individuals and Protestants living in mostly Catholic areas were
more likely to commit suicide than Catholics).(Durkheim, 1951) Contemporary research has
expanded this concept of solidarity within in a community to a broader concept of social
support, which refers to a network of family, friends and other persons surrounding an
individual who provide emotional support. A commonly used definition of social support is “the
presence of others or the resources provided by them, prior to, during and following a stressful
event”(Sarason, 1985)Prior research has demonstrated that social support is a modifiable risk
factor of suicide.(Berkman, Glass, Brissette, & Seeman, 2000) Social support is also known to be
associated with cigarette smoking in two main ways: lower levels of social support increases the
risk of starting to smoke, and also decreases the risk of smoking cessation after smoking
initiation.(Lakon, Hipp, & Timberlake, 2010; Romano, Bloom, & Syme, 1991)
To the best of my knowledge, no prior study on the relationship between cigarette
smoking and suicide has adjusted for social support. This dissertation will aide in this area.

11

As stated in the introduction, this dissertation will aide in the assessment of a causal
relationship between cigarette smoking and suicide. Below is a chapter by chapter description
of the main aims:
Chapter 3: This chapter will challenge the biological relationship between tobacco use
and suicide by considering the effects of smokeless tobacco on suicidal behavior risk. This will
be a novel contribution as the effects of smokeless tobacco on suicide risk have never been
studied. The use of the nationally representative, prospective NESARC data will help in
identifying newly-incident cases of suicidal behavior as another limitation of many previous
studies has been the inability to control for previous suicidal behaviors.
Chapter 4: This chapter will be similar in scope to chapter 3 in that it will consider the
effects of cigarette smoking and smokeless tobacco use on suicide related behaviors. Unlike the
longitudinal NESARC data in chapter 3, this chapter will use multiple years of cross-sectional
data from the National Survey(s) on Drug Use and Health. Combining data from the 2008-2010
surveys will result in a large sample, capable of generating relatively stable estimates on a rare
outcome.
Chapter 5: This chapter will focus on the associated between cigarette smoking and
provide a stronger ability to control for psychiatric comorbidities as well as differences in social
support. This will be an important contribution as it will add a detailed assessment of mental
illness. Data for this chapter are from the Collaborative Psychiatric Epidemiology Surveys (CEPS)
which is a collection of three large, mental health focused cross-sectional surveys in the United
States collected between 2001 and 2003.
12

Finally, before describing the study details, I would like to draw the reader’s attention to
an important limitation in the interpretation of findings. These chapters each have a greater
ability to confirm or corroborate prior evidence than they do to refute it. As an example, if
chapters 3 and 4 find that smokeless tobacco use and cigarette smoking both predict suicide
related outcomes, then the interpretation is fairly simple: tobacco use might play a biological
role in suicide because we now see multiple types of tobacco predicting suicide related
outcomes. However, if smokeless tobacco does not predict suicide related outcomes, the
interpretation is not as easy. While I might like to argue this is evidence against a smoking and
suicide hypothesis, it is important to remember that cigarette smoking might play a biological
role in suicide due to factors found in cigarettes or cigarette smoke that are not found in
smokeless tobacco.
I admit this is an important limitation of this dissertation, but still, one could argue this is
present at some degree in most research. For example, almost all results that are in the same
direction with what has already been published are easier to interpret and discuss than those
that are different.

13

CHAPTER 3: A PROSPECTIVE COHORT STUDY ON THE EFFECTS OF TOBACCO USE ON SUICIDE
RELATED OUTCOMES
Chapter 3 Aim:
This chapter will challenge the biological relationship between tobacco use and suicide
by considering the effects of smokeless tobacco on suicidal behavior risk. This will be a novel
contribution as the effects of smokeless tobacco on suicide risk have never been studied. The
use of the nationally representative, prospective NESARC data will help in identifying newlyincident cases of suicidal behavior as another limitation of many previous studies has been the
inability to control for previous suicidal behaviors.
Chapter 3 Methods:
Data and participants
The National Epidemiologic Survey on Alcohol and Alcohol Related Conditions (NESARC)
is a nationally representative longitudinal study focused on alcohol use the in the United States.
From August 2001 to September 2003, 43,093 individuals were gathered as the Wave 1 sample.
The U.S. Bureau of Census executed the field work and data collection for the National Institute
on Alcohol Abuse and Alcoholism. Interviews were conducted within a respondents home using
a computer assisted personal interview to improve validity.(Grant & Kaplan, 2005; Grant,
Kaplan, Shepard, & Moore, 2003)
As mentioned earlier, the target population was nationally representative. Specifically, it
was representative of adult (18 years or older) U.S. citizen and non-citizens living in the country
(including the District of Colombia, Alaska and Hawaii). All non-institutionalized persons were
eligible. This included those living in households, boarding houses, rooming houses, non14

transient hotels/motels, shelters, facilities for housing workers, college quarters and group
homes. The sampling approach used a multi-staged, complex survey approach where first a
census defined, county-based primary sampling unit (PSU) was selected, followed by a selection
of an eligible housing unit within each PSU and finally a selection of an individual within each
household. Young adults (those 18-24 years old) were oversampled. The household response
rate of 89% multiplied by the person response rate of 93% gave an overall response rate of
81%.(Grant & Kaplan, 2005; Grant, et al., 2003)
Assessment of exposure: tobacco variables
Tobacco use status was determined at the baseline wave 1 interview. Participants were
asked if, in their lifetime, they had smoked 100 or more cigarettes, if they had used snuff 20 or
more times and if they had used chewing tobacco 20 or more times. A person who had met any
of these thresholds was considered a lifetime tobacco user. For each tobacco product, if a
participant was a lifetime user, they were also asked more detailed questions about their use
such as their age of first use and if they are a never, former or current user. Current users were
those who had used within the past year, and former users were those who had used at some
point in their life but not in the past year.
Assessment of the outcome: suicide-related outcome variables
Suicide related outcomes were assessed as part of a structured interview for Major
Depression. All participants were asked if they had “ever had a 2 week period where they felt
sad, blue depressed or down most of the time” and if they “ever had a 2 week period when
they didn’t care about things they usually cared about”. Any participant who said yes to either
of these questions screened positive for Major Depression and was administered the entire
15

depression module. Within this module, all participants were asked if “during the time when
your mood was at its lowest or when you enjoyed or cared the least about things, did you
attempt suicide?” They were also asked if they had thought about committing suicide. Those
who had attempted suicide were treated as suicide attempters and those who had thought
about committing suicide were treated as having suicidal ideation.
The same module was asked again in the wave 2 interview. When participants were
asked the depression screening items, they were asked specifically if they had either of these
experiences since their last interview. Those who had were asked about suicide attempts and
suicidal ideation in the 2-3 year period since their last interview. By combining information from
the two waves, it was possible to exclude those who had previous suicide attempts or ideation
from the sample (because they are not at risk for becoming incident cases) and follow-up on
who becomes a new case of suicide attempt or ideation. (The specific numbers and proportions
will be discussed in the results section.)
Data analysis
The analysis tested the extent to which baseline cigarette smoking and smokeless
tobacco use predicted first suicide attempt or ideation during the 2-3 year follow-up period.
This was tested by a weighted logistic regression adjusting for the complex survey design
features in Stata v.12.0. There were 3 main outcomes of interest. (1) First suicide attempt.
(Among those with no previous suicide attempt as of wave 1, what proportion attempt suicide
by wave 2?) (2) First suicidal ideation. (Among those with no previous suicidal ideation as of
wave 1, what proportion developed suicidal ideation by wave 2?) (3) First suicidal behavior.

16

(Among those with no history of suicide attempt or ideation as of wave 1, what proportion had
either a suicide attempt or ideation by wave 2).
Univariable models for cigarette smoking and smokeless tobacco use were fit using
‘never user’ as a reference. Then the estimates were adjusted for sex, race/ethnicity
(categorized as Non-Hispanic White, Non-Hispanic Black, Non-Hispanic American
Indian/Alaskan native, Non-Hispanic Asian/Native Hawaiian/Pacific Islander or Hispanic), age
(18-24, 25-34, 35-44, 45-64, 65+) and education (less than high school, high school/GED, some
college, college or more). Also, cigarette smoking and smokeless tobacco use were included in
the same model (so the estimates for cigarette smoking are adjusted for smokeless tobacco use
and vice versa).
The other main confounding variables of interest that might affect suicide were major
depression and alcohol use disorders (abuse and or dependence). Age of onset data for both
depression and alcohol use disorders were collected as part of the interview. Using this
information along with age of onset of smokeless tobacco and cigarette smoking, it was
possible to adjust for depression and alcohol use disorders that occurred before cigarette
smoking and smokeless tobacco use. After the adjustment for socio-economic variables there
were two more adjusted models: one for cigarette smoking adjusting for prior depression and
alcohol use disorders, and one for smokeless tobacco use adjusting for prior depression and
alcohol use disorders.
Also, in an attempt to improve sample size, I also asked how baseline tobacco use
predicted the re-occurrence in addition to the incidence of suicide related behaviors. For this

17

step, I re-ran the main analyses in this chapter without excluding those who had baseline
suicide related behaviors.
Finally, I tested for the effect of cumulative tobacco use duration defined as the total
years of cigarette and smokeless tobacco use. This was calculated by taking the difference of
the participant’s age at assessment and their age of onset. The hypothesis being that among
users those with longer durations of tobacco use will have a higher rate of suicide related
outcomes compared to those with shorter durations if tobacco use is indeed acting as a cause.
The reader should note that these are not truly duration variables, but instead measure elapsed
time since onset since we are uncertain if they smoked every year between the year of onset
and the survey year.
Chapter 3 Results:
Figures 3.1, 3.2 and 3.3 each provide a flowchart showing how of the three analysis
samples was derived. For example, in Figure 3.1, 43,093 participants completed the Wave 1
interview. 1,157 had a suicide attempt prior to Wave 1. Since they were not at risk for an
incident suicide attempt during the follow-up, they were set aside. Among the 41,936 with no
history of suicide attempt, 33,687 completed the Wave 2 interview and 8,249 were lost to
follow-up. This gave a suicide attempt analysis sample of 33,687 individuals, 143 of which were
newly incident suicide attempters.
In Figure 3.2, 43,093 participants completed the Wave 1 interview. 3,655 had a history
of suicidal ideation prior to Wave 1. Since they were not at risk for incident suicidal ideation
during the follow-up, they were set aside. Among the 39,438 with no history of suicide attempt,
3,551 completed the Wave 2 interview and 7,886 were lost to follow-up. This gave a suicidal
18

ideation analysis sample of 33,687 individuals, 688 of which had newly-incident suicidal
ideation
Finally, in Figure 3.2, 43,093 participants completed the Wave 1 interview. 3,696 had a
history of suicidal behavior prior to Wave 1. Since they were not at risk for incident suicidal
behavior during the follow-up, they were set aside. Among the 31,520 with no history of
suicidal behavior, 31,520 completed the Wave 2 interview and 7,877 were lost to follow-up.
This gave a suicidal behavior analysis sample of 31,520 individuals, 702 of which had newly
incident suicidal behavior.
Tables 3.1, 3.2 and 3.3 provide a description of each of the analysis samples described in
Figure 3.1, 3.2 and 3.3 by covariates of central interest. Briefly, the 2-3 year incidence estimates
for suicide attempt, suicidal ideation and any suicidal behavior were 0.4%, 2.1% and 2.2%.
Current smokers appeared more likely to make a suicide attempt (0.7% vs. 0.3%), have suicidal
ideation (3.6%vs. 1.6%) or any suicidal behavior (3.7% vs. 1.7%) compared to never smokers. A
similar trend was observed for smokeless tobacco use: suicide attempt (1.0% vs. 0.4%), suicidal
ideation (3.0% vs. 2.1%) and suicidal behavior (3.4% vs. 2.1%). There were marked differences
in suicidal behavior between current chewers and never chewers (8.2% vs. 2.1%), but no
difference was seen by snuff use.
Table 3.4 provides the unadjusted and adjusted analyses for suicide attempt. Current
smokers were at an increased risk relative to never smokers (OR=2.2) while former and never
smokers did not differ. This did not appear to change after statistical adjustment. There was no
statistically significant association between smokeless tobacco use and suicide attempt,
however, this is likely limited by low lower. The point estimate for current vs. never users of
19

smokeless tobacco was similar to that for cigarette smokers (OR=2.7 unadjusted; OR= 2.8
adjusted). Table 3.5 adjusts the estimates in Table 3.4 by prior depression and alcohol use. This
adjustment did not affect the estimate for cigarette smoking, and resulted in a shift away from
the null for smokeless tobacco use (OR=3.4).
The results of the suicidal ideation analysis are found in Table 3.6. Prior to statistical
adjustment, current cigarette smoking predicted suicidal ideation (OR=2.3). There was no
increased risk among former smokers. Also, there was no association between smokeless
tobacco use and suicidal ideation. Statistical adjustment did not affect the smokeless tobacco
estimates. There was, however, a shift away from the null among former smokers (OR=1.6).
Adjusting for depression and alcohol use disorders in Table 3.7 provided little change to the
estimates. Tables 3.8 and 3.9 provide the results for the suicidal behavior analysis, but given the
considerable weight given to suicidal ideation in this analysis, the results do not differ from
Tables 3.6 and 3.7.
I also rephrased the research question to assess occurrence of suicide related outcomes
(first occurrence or reoccurrence). This added 1,116 current smokers, 579 former smokers and
1,386 never smokers who were previously excluded due to prior suicide related behaviors. In
terms of smokeless tobacco use, this step added 74 current smokeless tobacco users, 127
former smokeless tobacco users and 2,928 never smokeless tobacco users.
Table 3.10 compares shows results for models predicting suicide related outcomes as a
function of both cigarette smoking and smokeless tobacco use. This was done for a sample that
excludes those who had suicide related behaviors at baseline, and for a sample that did not
make this exclusion. In both samples, current cigarette smoking at baseline was independently
20

associated with each suicide related outcome. Current smokeless tobacco at baseline use was
associated with suicide attempts in the first occurrence analysis, but not in the first
occurrence/reoccurrence analysis.
Finally, I estimated the effect of elapsed time since cigarette onset and elapsed time
since smokeless tobacco use on suicide related outcomes. These models failed to detect a
crude or adjusted association between years since onset and suicide related outcomes
(ORs=1.0 p>0.05).

21

CHAPTER 4: A NATIONALLY REPRESENTATIVE CROSS-SECTIONAL STUDY ON THE
RELATIONSHIP BETWEEN TOBACCO USE AND PAST YEAR SUICIDE RELATED OUTCOMES
Chapter 4 Aim:
This chapter will be similar in scope to chapter 3 in that it will consider the effects of
cigarette smoking and smokeless tobacco use on suicide related behaviors. Unlike the
longitudinal NESARC data in chapter 3, this chapter will use multiple years of cross-sectional
data from the National Survey(s) on Drug Use and Health. Combining data from the 2008-2010
surveys will result in a large sample, capable of generating relatively stable estimates on a rare
outcome.
Chapter 4 Methods:
Data and participants
Cross-sectional samples were collected as part of the National Survey(s) on Drug Use
and Health (NSDUH). The NSDUH’s sampling approach involved a multi-stage probability
sampling of non-institutionalized community dwelling US residents aged 12 and older.
Individuals living in households as well as those in college dorms, other group home facilities,
those living on military bases and those without a permanent place of residence were all
included in the sampling frame. Further details of the sampling approach are described
elsewhere.(Barondess, Meyer, Boinapally, Fairman, & Anthony, 2010; SAMHSA, 2010b) The
study protocol was reviewed and approved by the cognizant institutional review board for
protection of human subjects in research.(SAMHSA, 2010a)
The NSDUH is conducted annually and is designed in a way that allows for combining
data across multiple years.(Barondess, et al., 2010; Meier, Troost, & Anthony, 2012) Each
22

survey has used a computer-assisted interview where respondents key in answers using a
laptop computer to improve validity. After 2008, the NSDUH began collecting data on suicide
related outcomes. To optimize the sample size, I combined data from 2008-2010. The total
sample sizes from 2008-2010 were 55,739, 55,772 and 57,837 respectively.
Assessment of exposure: tobacco variables
As described above, I was interested in the extent that tobacco use, that is, use of both
cigarettes and smokeless tobacco products, predict suicide related outcomes. Previously
validation work has shown strong validity in the NSDUH’s self-report tobacco module.(Harrison,
Martin, Eney, & Harrington, 2007) Participants were asked ‘Have you ever smoked part or all of
a cigarette?’ Those whose had were asked another set of questions regarding their smoking
experience such as ‘How long has it been since you last smoked part or all of a cigarette?’ with
the following possible responses: ‘Within the past 30 days’, ‘More than 30 days ago, but within
the past 12 months’, ‘More than 12 months ago but within the past 3 years’ or ‘More than 3
years ago’. I classified those who had smoked in the past year as ‘current’ smokers. Those who
had last used more than 12 months ago but within the past 3 years were classified as ‘recent
quitters’. Those who had last smoked more than 3 years ago were labeled as ‘former’ smokers.
Finally, those who had never smoked were simply ‘never’ smokers.
Analogous questions were asked regarding smokeless tobacco use. After a brief
description of the types of smokeless tobacco, participants were asked ‘Have you ever used
snuff, even once?’ and ‘Have you ever used chew, even once?’ For both snuff and chew,
participants were also asked the recency of use question that were asked to cigarette smokers.

23

For the main analysis, I combined information about snuff and chew use into simply smokeless
tobacco use, however, I considered each substance separately in a secondary analysis.
Assessment of the outcome: suicide-related outcome variables
Given the cross-sectional nature of the data, I was unable to measure completed suicide
among participants and instead I focused on the following suicide-related outcomes: suicidal
ideation, suicidal plans, and suicide attempts. In addition to these experiences being signs of
psychological distress by themselves, they are also strong predictors of future completed
suicide,(E. K. Moscicki, 1995) and I considered them worth studying.
All adult participants, those aged 18+, were asked ‘At any time in the past 12 months,
up to and including today, did you seriously think about trying to kill yourself?’ Those who
indicated that they had were labeled as having past year suicidal ideation and asked the
following question on suicide plans: ‘during the past 12 months, did you make any plans to kill
yourself?’ Those with a positive response were classified as having a past year suicide plan and
were asked a final question on suicide attempts ‘during the past 12 months, did you try to kill
yourself?’ To be clear, because the question was only asked to adults, this analysis is limited to
persons 18+ despite there being participants aged 12-17 in the sample.
Data analysis
Past year suicide ideation, plans and attempts were treated as separate outcomes in
three separate series of logistic regressions. The main predictors of interest were cigarette
smoking and smokeless tobacco use, both coded as a categorical variable describing the
recency of use in four categories using never users as a reference. Analyses were conducted in

24

Stata v11 with ‘svy’ commands to take into the account the complex survey design and
weighting in the NSDUH.
After crude associations between both cigarette and smokeless tobacco use and each of
the three suicide-related outcomes were estimated, adjustments were made for sex, race (nonHispanic White vs. other), age (18-25, 26-35 and 35+), income (<$20k, $20k-$49k, $50k-$74k
and ≥$75k) and population density (Metropolitan Statistical Area (MSA)>1 million, MSA<1
million, non-MSA).
The other confounding variables of interest included history of major depression and
alcohol abuse. I decided to condition on these variables by stratifying instead of adding them to
our multi-variable models. I stratified based on past year DSM-IV alcohol abuse and/or
dependence as well as past year DSM-IV cannabis abuse/dependence. I stratified on depression
in two ways: first, I stratified on whether or not the participant indicated that ‘a doctor or other
medical professional’ had ever told them they had depression. I also stratified based on a
structured module to assess DSM-IV Major Depression.
A secondary analysis repeated the above steps but separated smokeless tobacco use
into two categories: snuff use and chewing tobacco use.
I also tested for a dose-response between tobacco use and suicide-related outcomes
among current users. Here, I regressed the log-odds of past year suicide-related outcomes on
the number of days used each type of tobacco in the 30 days prior to the assessment.
The final analysis attempted to tease apart the specific effects of cigarette smoking and
smokeless tobacco use on suicide related outcomes by setting aside users of multiple tobacco
products. While this analysis was limited by sample size, it was done as follows: (1) test for
25

differences in the frequency of each suicide related outcome among current, former and never
smokers who had all never used smokeless tobacco. Similarly, (2) test for differences in the
frequency of each suicide related outcome among current, former and never smokeless
tobacco users who had never smoked cigarettes.
NESARC vs. NSDUH
Given the apparent similarities between the NESARC (chapter 3) and NSDUH (chapter 4)
surveys, it seems appropriate to provide a brief comparison of methods before presenting the
NSDUH results.
Both surveys use the non-institutionalized, civilian population of the US as a target
population, and their sampling frames include group quarters (e.g., college dorms, group
homes) in addition to typically conceptualized households. The NESARC sampled adults (age 18
and up) while the NSDUH sampled adolescents as well (age 12 and up) although only those 18
and up are used from the NSDUH here. Both surveys used complex sampling designs with
sociodemographic stratifications. Weights were assigned to make the sample representative of
the US population, although the specific sampling methods differed. In the NESARC, the PSUs
approximated the US Census Current population Survey county-based PSUs, and the second
stage consisted of dwelling unit selection. Blacks and Hispanic dwelling units were oversampled.
Finally, one person was selected from each dwelling unit.(Grant, et al., 2003) The NSDUH’s PSUs
were 500,000 area segments (groups of adjacent census blocks). The next stage involved the
selection of a dwelling unit from within a selected segments boundaries. Adolescents and
young adults were oversampled within each dwelling unit—one third of the sample was

26

between 12 and 25 years old. Target samples were 3600 for the 8 largest states and 900 in the
other states and in D.C.
Grucza et al., have outlined important differences in the two surveys in regards to
privacy/anonymity considerations, response rate and diagnostic instrumentation. Namely, that
the NSDUH is expected to generate more accurate data because of its use of a computerized
self-administration method. Also, data collection was anonymous for the NSDUH, so a
participant’s name was not linked with their answers. Also, the NSDUH’s interviews were
conducted by RTI, a private firm, while the NESARC was conducted by the census bureau.
(Grucza, Przybeck, & Cloninger, 2007)
I believe these studies complement each other for the sake of this dissertation research.
The NESARC data has the advantage of being longitudinal, while the NSDUH has the advantage
of better confidentiality and, perhaps, more better validity with respect to the responses from
the respondents.
Chapter 4 Results:
The total adult sample size in the NSDUH population from 2008-2010 was 114,048.
Table 4.1 shows the distribution of these individuals by tobacco use status and past year
suicidal ideation, plans and attempts. 37,186 were never smokers, 43,070 were current (past
year), 6,553 had quit smoking between 1-3 years prior to the interview, and 27,239 had quit >3
years prior to the interview. For smokeless tobacco use (combining both snuff and chewing
tobacco use) there were 89,789 never smokeless tobacco users, 8,444 current users, 3,227
individuals who had quit using smokeless tobacco 1-3 years prior to the interview and 12,558

27

who had quit >3 years prior to the interview. Table 4.1 also shows frequencies specifically for
snuff and chew users.
Other information in Table 4.1 displays the estimated prevalence of past year suicide
related outcomes by tobacco use. For the entire population, 3.8% had experienced suicidal
ideation, 1.1% had made a suicidal plan and 0.5% had attempted suicide. These proportions
appeared to differ by tobacco use status. For example, the prevalence of past year suicide
attempt was 1.0% among current smokers and 0.3% among never smokers and between 0.3%
and 0.4% for former smokers. This trend was less clear among smokeless tobacco users where
former users who quit 1-3 years prior to the interview had the highest prevalence of suicide
attempt at 1.0%, compared to 0.8% for current users, 0.4% for never users and 0.5% for former
users who quit >3 years prior to the interview.
Table 4.2 provides a description of the data by other covariates of interest. While males
were more likely to be ever users of any type of tobacco compared to females, this was more
pronounced for smokeless tobacco than for cigarettes. For examples 74.8% of males and 64.2%
of females had smoked a cigarette at some point in their lifetimes while 33.4% of males and
only 5.3% had used either type of smokeless tobacco. Those with depression, assessed by
either self-report or a structured interview, were more likely to be cigarette smokers, but did
not appear to differ in terms of smokeless tobacco use. Those with alcohol use disorder, a
known strong predictor of suicide, were more likely to be cigarette smokers, and almost twice
more likely to be smokeless tobacco users than those without an alcohol use disorder.
Table 4.3 describes the sample by crossing cigarette smoking status and smokeless
tobacco use status. For example, among those who had never used smokeless tobacco, 35,134
28

individuals had never smoked, 29,243 were current smokers, 4,681 were recent quitters, and
20,731 were former smokers; among those who had never smoked a cigarette, 35,134 were
never smokers, 831 were current smokers, 248 were recent quitters and 973 were former
users. This table demonstrates there is a reasonable amount of independence between
cigarette smoking and smokeless tobacco use in the US. Indeed, the two are related, but not
necessarily all smokeless tobacco users smoke cigarettes and vice versa.
Table 4.4 provides crude and adjusted odds ratios for the estimated effects of tobacco
use status on suicide related outcomes. Cigarette smoking predicted suicidal ideation before
and after adjustment for background variables. Current users had a 4.3 fold increase in odds of
suicidal ideation and recent quitters had a 1.5 increase in odds compared to never users. Similar
results were found for suicidal plans and suicide attempts: current smokers were between 2-3
times and recent quitters were ~1.5 times more likely to have experienced the outcomes
compared to never smokers. Former smokers, who had quit for at least 3 years, did not appear
to differ from never smokers in terms of planning and attempts.
The unadjusted estimates in Table 4.4 for smokeless tobacco show a similar trend to
those seen for cigarette smoking, albeit at a smaller effect size: current users of either
smokeless tobacco had a 1.6-1.9 fold increase in odds, depending on the outcome, compared to
never users. Recent quitters were also at increased odds for suicidal ideation and suicide
attempts, and former users had increased odds for ideation. However, after statistical
adjustment, most of the ORs were shifted toward the null. There was a significant OR of 1.2 for
former vs. never smokeless tobacco use, and all others were non-significant (with p>0.10).

29

The next two panels of Table 4.4 provide estimates for chewing tobacco and snuff use
separately. Note, the ‘smokeless tobacco estimates’ panel adjusts for cigarette smoking, the
‘chewing tobacco estimates’ panel adjusts for cigarette smoking and snuff use and the ‘snuff
estimates’ panel adjusts for cigarette smoking and chewing tobacco use. Similar results were
found for chewing tobacco and snuff separately as when they were combined into one
exposure: neither type of smokeless tobacco use predicted suicide related outcomes as
consistently as cigarette smoking. After statistical adjustment, current chewers were more
likely to have suicidal ideation compared to never chewers (OR=1.6) and former snuff users
were more likely to have experienced suicidal ideation and suicidal planning compared to never
snuff users (ORs=1.4 and 1.5 respectively).
Tables 4.5-4.8 provide stratifications in the unadjusted estimates in Table 4.4 by
depression and alcohol use disorders. Table 4.5 stratifies based on the structured interview
depression module in the NSDUH. Cigarette smoking predicted all suicide related outcomes in
both those with and without lifetime depression, although the estimated effect appeared
stronger in the non-depressed (current vs. never OR=1.8 and 2.6 for those with and without
depression respectively). Unadjusted smokeless tobacco estimates generally remained
unchanged before and after stratification. However, chew use was a stronger predictor of
suicidal planning and attempts among those without depression than among those with
depression. Table 4.6 stratifies the estimates by depression status, as in Table 4.5, but here
depression was assessed by a self-report item (“has a doctor, nurse or other health professional
ever told you that you have major depression?”). The results are not appreciably different from
those found in Table 4.5.
30

Table 4.7 stratified the unadjusted estimates in Table 4.4 by whether or not the
participant had a current alcohol use disorder (DSM-IV Alcohol Abuse or Alcohol Dependence).
Cigarette smoking results were similar to those in the depression stratifications. Current
smoking predicted suicidal ideation, and this estimated effect was stronger in those without an
alcohol use disorder compared to those with one; current smoking also predicted suicidal
planning among those without an alcohol use disorder, and suicide attempts in both those with
and without alcohol use disorders. The estimates for smokeless tobacco use were noticeably
attenuated after stratification after stratification by alcohol use disorder status: the only
evidence of smokeless tobacco predicting suicide related outcomes was a 1.3 fold increase in
odds of suicidal ideation in former users compared to never users without an alcohol use
disorder, and a 2.1 increase in odds of suicide attempt in recent quitters compared to never
users without an alcohol use disorder. When separating the estimated effects of chewing
tobacco and snuff, chewing tobacco had a slightly more consistent relationship with suicide
related outcomes than snuff.
Other results consider a possible dose-response relationship between tobacco use and
suicidal outcomes. Specifically, if either the reported number of days smoked or days used
smokeless tobacco in the 30 days prior to the assessment predicted past year-suicide related
outcomes. Table 4.8 shows the weighted mean number of days used each tobacco product and
its standard error by each suicide related outcome. A series of regressions tested for the effects
of the number of days an individual used cigarette, chew or smokeless in the 30 days prior to
the interview on the log-odds or suicidal ideation, plans and attempts. These regressions were
also stratified by depression and alcohol use disorders. The p-values of these regressions are
31

shown in Table 4.9. The only consistent dose-response trend was an inverse association
between snuff use and suicidal ideation.
Finally, Table 4.10 examines the effects of tobacco use on suicide related outcomes
among those who are either exclusive cigarette smokers, or exclusive smokeless tobacco users.
While the results of the smokeless tobacco analysis is limited by the small number of smokeless
tobacco users who had never smoked and had suicide related outcomes, the p-values in the
footnote indicate that cigarette smoking consistently predicted suicide related outcomes, while
smokeless tobacco use did not. To be clear: among those who had never used smokeless
tobacco, cigarette smoking was a significant predictor of suicidal ideation, plans and attempts;
among those who had never smoked a cigarette, smokeless tobacco was not a significant
predictor of said suicide related outcomes.
Realizing the relatively small sample of smokeless tobacco users who had never smoked
a single cigarette, I modified this analysis by excluding current users (instead of ever users) or
each respective tobacco type. These results are shown in Table 4.11 where one can see that
cigarette smoking predicted all suicide related outcomes among those who were not currently
using smokeless tobacco, and there was no evidence that smokeless tobacco use altered the
occurrence of suicide related outcomes among those who were not current smokers.
(Significance tests from a logistic regression are displayed in the footnote.)
As a final exploratory step, I tested for effect modification between cigarette smoking
and smokeless tobacco on each suicide related outcome. This involved crossing recency of
cigarette smoking by recency of smokeless tobacco use. For the estimated effects of tobacco on
suicidal ideation, current cigarette smokers (p=<0.01) and recent quitters (p<0.01) were at an
32

increased odds of suicidal ideation, and there was a significant subgroup variation effect of
current smokeless tobacco use was smaller in those who had quit smoking 1-3 years prior. For
suicidal planning, current cigarette smokers (p=<0.01) and recent quitters (p<0.01) were at an
increased odds of suicidal ideation, and there was a significant subgroup variation term
showing former cigarette smokers who had quit for at least 3 years and quit using smokeless
tobacco 1-3 years prior were at a lower odds for suicidal planning than other former cigarette
smokers (although this estimate was very imprecise and difficult to interpret). For the effect on
suicide attempts, current cigarette smokers and recent quitters were at an increased odds for
suicide attempt, and there was no subgroup variation by smokeless tobacco use. Also, there
was no evidence in these models that smokeless tobacco use is associated with an increased
odds in suicide attempt. and none of the smokeless tobacco or interaction terms were
significant. (These data are not shown in a table).
Given the small number of significant interactions terms (2/27) it is possible they are
due to chance, and both are non-significant after a Bonferroni correction for multiple
comparisons. With that in mind, I still explored the significant interaction term between former
smoking and current smokeless tobacco use on suicidal ideation using the following estimated
coefficients from the model: change in log odds of +0.51 for former smoking, +0.50 for current
smokeless tobacco use and -0.95 for the interaction term. A current smokeless tobacco
user/recent cigarette quitting does not differ from a never tobacco user in terms of occurrence
of suicidal ideation (OR=1.06). I compared to this to someone who had recently quit using
smokeless tobacco but was still smoking cigarettes using the following coefficients: +0.91 for
current smoking, -0.70 for former smokeless tobacco use (1-3 years prior) and +0.90 for their
33

interaction. This type of tobacco user would still be at an increased risk compared to a never
tobacco user (OR=2.96). This is all under the large assumption of these interaction terms being
of value, for which there is very little evidence in this analysis. However, the results suggest that
someone who is both a current smoker and a current smokeless tobacco user would see a great
reduction in their risk of suicidal ideation if they were to quit smoking, but they would see no
change in risk if they were to quit using smokeless tobacco.

34

CHAPTER 5: THE RELATIONSHIP BETWEEN CIGARETTE SMOKING AND SUICIDE RELATED
OUTCOMES: WHAT CAN WE ATTRIBUTE TO MENTAL HEALTH AND SOCIAL SUPPORT?
Chapter 5 Aim:
This chapter will focus on the associated between cigarette smoking and provide a
stronger ability to control for psychiatric comorbidities as well as differences in social support.
This will be an important contribution as it will add a detailed assessment of mental illness.
Data for this chapter are from the Collaborative Psychiatric Epidemiology Surveys (CEPS) which
is a collection of three large, mental health focused cross-sectional surveys in the United States
collected between 2001 and 2003.
Chapter 5 Methods:
Data and participants
The data for this chapter were drawn from the Collaborative Psychiatric Epidemiology
Surveys (CPES). This was a combination of the National Comorbidity Survey Replication (NCS-R),
and the the National Latino and Asian American Study (NLAAS). Each of these cross-section
samples were gathered from 2001 to 2003 in an effort to better understand the prevalence,
correlates, risk factors and suspected causes of mental illness in the US. The target population
of non-institutionalized adults (those over 18) was interviewed either face to face or over the
phone. (S. G. Heeringa et al., 2004; Pennell et al., 2004) Interviewers were trained to administer
the Mental Health Composite International Diagnostic Interview (WMH-CIDI, which is a tool
used to diagnose mental illness according to DSM-IV criteria.(Kessler & Ustun, 2004)
Multi-stage probability sampling was done at several levels: PSU of US metropolitan
statistical areas, single counties and groupings of countries. This step was followed by the
35

sampling of census blocks within metropolitan statistical areas. Afterwards, households were
selected from census blocks, and then finally a person was selected within each household. The
assessment uses a two-part design to limit the burden on the respondent. At baseline, all
individuals are screened for having any mental illness. All those who screened positive for asked
to take the second part of the interview several weeks later. A 25% enrichment sample of those
who were not screened positive for mental illness were also asked to take the part II interview.
Weighting factors adjusted for this imbalance in the analysis.
Assessment of exposure: tobacco variables
While these surveys were mental health focused, there was a brief assessment of
cigarette smoking status. Participants were asked if they “were a current smoker, an ex-smoker
or if they had never smoked.” They were also given “only smoked a few times” as a response
option. These individuals were treated as former smokers for the sake of this analysis.
Assessment of the outcome: suicide-related outcome variables
Questions regarding suicidal behavior were asked the part II interview. Participants
were asked if they had ever ‘seriously thought about killing themself’. Those who responded
that they had were designated as having suicidal ideation and were asked the following
question about suicidal plans: ‘have you made a plan to kill yourself?’ If participants had made a
suicidal plan, they were asked whether they had ever attempted suicide. Protocols for the NCSR and for the NLAAS were the same except participants in the NLAAS were able to indicate that
they had made a suicide attempt unconditional on if they had suicidal ideation or had made a
suicide plan.

36

The following lifetime diagnoses were treated as confounders in this analysis: bipolar
disorder (I and II), major depression, dysthymia, alcohol use disorder (alcohol dependence with
or without abuse), substance use disorder (substance dependence with or without abuse),
conduct disorder, post-traumatic stress disorder, generalized anxiety disorder and panic
disorder. In addition, to diagnosed mental illness, I was also interested in controlling for
differences in social support, as higher levels of social support have been shown to possibly
protect against smoking onset and possibly increase the likelihood of smoking cessation (Lakon,
et al., 2010; Romano, et al., 1991)
To measure this, I used 12 questions on social support asked to all participants:
(1) How often do you talk on the phone or get together with relatives who do not live
with you?
(2) How much can you open up to relatives who do not live with you if you need to talk
about your worries?
(3) How much can you open up to relatives who do not live with you if you need to talk
about your worries?
(4) How often do your relatives make too many demands on you?
(5) How often do your relatives argue with you
(6) How often do you talk on the phone or get together with friends?
(7) How much can you rely on your friends for help if you have a serious problem?
(8) How much can you open up to your friends if you need to talk about your worries?
(9) How often do your friends make too many demands on you?
(10)

How often do your friends argue with you?
37

(11)

When you have a problem or worry, how often do you let your partner worry

about it?
(12)

When you have a problem or worry, how often do you let someone (else) know

about it?
Respondents answered each of these items with Likert scale, ordinal responses. I
combined this information into factor scores using an exploratory/confirmatory factor
analysis method (details explained in the analysis section).
Data analysis
History of suicidal ideation, plans and attempts were treated as separate outcomes in
separate series of logistic regressions. The main exposure variable of interest was smoking
status, current, former or never.
For each outcome, the first steps involved unadjusted estimates for the effect of
smoking on suicide related outcomes. The next steps involved adjustment for the descriptive
characteristics: age, sex, race, income and education. Estimates were then adjusted by history
of mental illness.
Social support variables were created according to a factor analysis. I considered this an
appropriate analysis because of the latent nature of ‘social support’. Each of these 12 items
asks the participant about having a network of friends and relative that they are comfortable
with talking about their worries and how well they get along with the individuals in this
network. Another option would be to use a principal components analysis, however, such an
analysis would neglect the possible inter-factor correlations.

38

For the first step, I first conducted an exploratory factor analysis using a geomin rotation
in Mplus version 5.0. I divided the sample into random exploratory and confirmatory halves.
The exploratory factor analysis helps in this context because I was uncertain about the number
of factors. I hypothesized that there would be two factors: one that primarily loaded the items
asking about relatives, and another onto the items about friends. To do this, I used the NLAAS
as the exploratory sample and the NCS-R as the confirmatory sample. Given the relatively small
number of items (12) I decided to test for 1-4 number factor solutions. Anything much larger
would limit the number of items per factor. Ordinal responses to each item were coded so that
a higher score represented a higher level of social support. The specific responses for each item
can be found in table 4.3.
Once the number of factors had been determined, all significant loadings (at p<0.05)
were used in a confirmatory factor analysis using the NCS-R sample. Factor loadings that were
significant at this level were used for estimating social support scores. This was done on the
entire sample. Social support scores were then used as a covariate when estimating the
association between smoking and suicide related outcomes.
As a final exploratory step, I considered the finding in chapter 5 were cigarette smoking
was a stronger predictor of suicide in those without major depression than in those with it. For
each mental disorder and each suicide related outcome, I ran a logistic regression with terms
for smoking, mental disorder and the interaction between the two. These models asked
whether the effects of cigarette smoking on the occurrence of suicidal outcomes might vary
with respect to whether the person had a mental disorder. Using evidence from chapter 5, I

39

approached this work with the hypothesis there would be a diminished effect of smoking on
suicide in those with a mental disorder (or at least those with depression).
Chapter 5 Results:
In total, there were 13,931 individuals in the combined NCS-R and NLAAS sample. This
analysis was limited to those who had taken the second part of the interview, as this was where
most of the relevant questions were asked. The total sample size for this study was 10,341.
Table 5.1 provides a brief description of the nationally representative sample. Prevalence
estimates of current and former smoking were comparable to other studies (24.3% current and
27.3% former).
Table 5.2 shows the proportions of suicide related outcomes and mental disorder status
by smoking status. Those with suicidal ideation, planning and attempts were more likely to be
current smokers than those without (38.8% vs. 22.4%, 45.2% vs. 23.5% and 50.1% vs. 23.4%
respectively). Similar results were seen for mental disorders. For example, 61% of those with
Bipolar II were current smokers while only 24.2% of non-Bipolar II cases were current smokers.
Table 5.3 provides a list of each social support item along with the response
proportions. The items were coded so that a high number represents a high level of social
support and a low number represents a low amount. Items were fairly well-distributed
although the distributions are left skewed (clustering towards higher levels of support). Tables
5.4 and 5.5 contain the same information in Table 5.3 with Table 5.4 showing responses for the
exploratory sample and table 5.5 containing responses for the confirmatory. There were little
to no differences between samples with respect to marginal response patterns.

40

Table 5.6 contains estimates from a series of weighted logistic regressions predicting
suicide related outcomes as a function of smoking status and socio-economic descriptive
characteristics: sex, age, race, income and education. There were significant unadjusted effects
of smoking on suicide related. ORs for the estimated effect of current relative to never smokers
were 2.4 for ideation, 2.7 for plans and 3.8 for attempts. Former smoking also predicted suicide
attempts (OR=1.4). Adjusting for sex, age, race, income and education had little effect on the
results.
Table 5.7 shifts to adjustment for mental disorder status. Here, before moving to a
multivariable model with more than one mental disorder, I first looked at how adding each
mental health diagnosis altered the OR between cigarette smoking and suicide related
outcomes For example, the estimates in the first three rows labeled “Bipolar I” are the effect of
smoking on suicide related outcomes adjusting for Bipolar I diagnosis. Current cigarette
smoking remained a significant predictor of each suicide related outcome after independently
adjusting for each mental disorder; no single diagnosis accounted for the observed association
between smoking and suicidal outcomes. Adjustment for alcohol use disorder resulted in the
greatest attenuation of the ORs (1.8, 1.9 and 1.8 for the estimated association with current
smoking on ideation, plans and attempts respectively).
Multivariable models with each diagnosis are shown in Table 5.8. After adjusting for all
mental disorders, current cigarette smoking continued to predict each suicide related outcome
(OR=1.4 for ideation and plans, OR=2.0 for attempts). Consistent with previous findings, bipolar
disorder had the strongest effect on each suicide related outcome.

41

The model fit indices for the factor analysis of social support are found in Table 5.9. Fit
indices are best with the 4-factor solution, although the CFI and TLI suggest good fit is also
found with the 3 factor solution. Also, the RMSEA of the 3-factor solution (0.081) was almost in
the acceptable range (<0.08). The Kaiser rule would also suggest the 3 factor model due to the
eigenvalues being <1 in the 4 factor solution.(Lance & Vandenburg, 2008) I also considered
factor interpretability in determining the number of factors. EFA factor loadings are found in
table 5.10. The single factor solution had mostly positive loadings with the strongest loadings
on “How much can you open up to relatives who do not live with you if you need to talk about
your worries?”, “How much can you open up to relatives who do not live with you if you need
to talk about your worries?”, “How much can you rely on your friends for help if you have a
serious problem?” and “How much can you open up to your friends if you need to talk about
your worries?”. This seemed reasonable as each of these items reflected having someone, a
friend or a relative, from which the individual can seek help from. There were also some
moderately strong negative loadings on items asking about demands and arguing with friends
and relatives. (Note: since all items are coded so that a higher value=higher social support, this
would mean individuals with a high score on this factor tend to have someone to talk to about
their problems but also that they argue frequently with those same individuals).
In the second factor solution factor 1 appeared to be similar to the single factor in the
one factor solution. The second factor had positive loadings for arguing and demand variables.
In this solution, a high amount of factor 1 seems to represent a high amount of support from
family and friends, while a high amount of factor 2 seems to represent a low amount of stress
(demands and arguments) from friends and family.
42

The 3-factor solution appears to separate factor 1 into 2 categories: factor 1, those who
have a high amount of support from their friends and family, and factor 2: those who have a
high amount of support from their friends. Here, factor 3 represents those who might not have
a high degree of support from friends or family, but they tend not to argue or feel demands
from their friends and family. The 4-factor solution is difficult to interpret because of a mono
item factor. Because of this in combination with the fit statistics, I chose to use the 3-factor
solution.
Table 5.11 shows the results from the CFA using all items with significant loadings in the
EFA. Factor structure in the EFA was supported by the CFA. (Although I considered dropping the
factor 1->item 12 loading. While it is still statistically significant, it is very low and loads better
onto another factor. Given the small amount of factor correlation, I decided to retain it.)
The distributions of factor scores are shown in table 5.12. There is a fair amount of
variation in each of the three scores by smoking status where never smokers have higher levels
of support than current smokers.
The effects of social support on suicide related outcomes are displayed in table 5.13.
Factors 1 and 3 were protective of suicidal ideation, plans and attempts. Interestingly, factor 2
had no effect on these outcomes. Adjusting for these scores had little effect on the association
between cigarette smoking and outcomes.
Table 5.14 summarizes most of the model results in this chapter and adds a final set of
models that controls for descriptive characteristics, mental disorder status and social support
(Model V and Model VI. Model VI is the same as Model V, but instead of adding a term for each
mental disorder, instead I added a term for total number of disorders). Although there was a
43

noticeable attenuation in the effect size current smoking continued to predict suicidal ideation
(OR=1.4), suicidal plans (OR=1.3) and suicide attempts (OR=1.8) To put this in perspective, the
independent effects of cigarette smoking on suicide attempts comparable to the independent
effects of substance use disorder, alcohol use disorder, conduct disorder, PTSD, generalized
anxiety disorder or dysthymia on suicide.
Finally, Table 5.15 explores the idea that smoking might be a stronger predictor of
suicide related outcomes in those without a mental disorder than in those with a mental
disorder (subgroup variation). This table shows the ORs for a product term between current
smoking and mental disorder status. For example, an OR>1 would suggest that the effects of
cigarette smoking on suicide are stronger in those with a given mental disorder, and an OR<1
would indicate that said effects are smaller in those with a mental disorder. For suicidal
ideation, significant ORs<1 were found for major depression, dysthymia and post-traumatic
stress disorder, for suicidal plans, significant ORs<1 were found for major depression and posttraumatic stress disorder, and there were no significant interactions for suicide attempts.

44

CHAPTER 6: DISCUSSION, LIMITATIONS, & FUTURE RESEARCH
The main results can be summarized succinctly: (1) In the NESARC longitudinal analyses,
tobacco cigarette smoking at baseline predicted newly-incident suicidal ideation (SI) and
attempts (SA) while baseline smokeless tobacco was an independent predictor of suicide
attempts (but not SI). (2) In addition to a general confirmation of NESARC longitudinal results
via cross-sectional analyses with other independently gathered data, the cross-sectional NSDUH
analyses disclosed that cigarette smokers who have never used smokeless tobacco were still at
increased odds for suicide related outcomes, but smokeless tobacco users who had never
smoked cigarettes were not. (3) In the cross-sectional CPES analysis with detailed control of
mental health status and social support, cigarette smoking had an independent association with
suicide related outcomes. In sum, the results partially support the smoking hypothesis under
study; some results seem to contradict what has been hypothesized.
I will start by a description of the results that support tobacco cigarette smoking as a
possible cause of suicide related outcomes. First, in all three analyses (the NESARC, the NSDUH
and the CPES), being a current cigarette smoker predicted suicidal ideation, plans and attempts
after covariate adjustment. These odds ratio (OR) estimates for suicide attempt ranged from
1.8 in the CPES analysis to 2.8 in the NSDUH analysis. The ORs for the estimated effect of
current smoking on suicidal ideation and suicidal plans remained statistically significant (with
p<0.05) after covariate adjustment in all analyses, but were not as strong as the ORs for current
smoking and suicide attempts. In the CPES analysis, these were fairly small (OR=1.4 for ideation
and 1.3 for attempts), however, they were still statistically significant (with p<0.05). These
estimates are consistent with published estimates from other studies. Also, current cigarette
45

smoking predicted suicide related outcomes after stratification by mental disorder status with
due consideration of social support levels. This work supports the idea that there is an
association between being a current smoker and having an increased risk of suicide related
outcomes, and that this association is independent of diagnosable mental disorders and levels
social support. Although I can still speculate that sub threshold mental illnesses could still act as
a confounder (e.g., controlling for depression as a continuous covariate instead of a categorical
covariate might drive the smoking-suicide association toward the null.)
In some of these analyses, the adjusted OR estimates for the effect of cigarette smoking
on suicide attempts were modest (e.g., 1.8). Some readers might not be impressed by the
magnitude of this association, and this judgment is somewhat justifiable. It is a relatively small
OR. I would note that this estimate comes from a model adjusting for many important
confounding variables. Also, it may be important to compare the 1.8 value to the other ORs
calculated in the modeling process. Although not shown in a table, , other OR estimates were
found when the log odds of suicide attempts was regressed on bipolar I status (OR=3.4, bipolar
II status (OR=2.6), major depression (OR=2.5), dysthymia (OR=2.0), alcohol use disorder
(OR=1.5), conduct disorder (OR=1.5), substance use disorders (OR=1.8), post-traumatic stress
disorder (OR=1.9), and generalized anxiety disorder (OR=2.2). When compared to these
estimates, I judge the 1.8 estimate for current smoking to be a bit more impressive.
Other evidence presented here somewhat contradicts the idea that being an active
cigarette smoker might cause suicide or suicide related outcomes. Although not true in every
modeling step, former smoking also tended to predict suicide related outcomes, although at a
lesser extent than current smoking. The strongest evidence on former smoking is from the
46

NESARC analysis where, after covariate adjustment, current smoking predicted suicidal
behavior with an OR of 2.0 while former smoking also did with an OR of 1.6. Similar results
were found in the NSDUH analysis, with ORs of ~2.7 and ~1.5 respectively.
As described in the introductory background material, one of the prominent prior
suggestions about the biological mechanism for this association is through differences in MAO
levels due to smoking. However, once an individual stops smoking, their MAO levels return to
normal.(Norman, et al., 1987) Therefore, if the transient MAO level is the central mechanism, it
does not make sense within this model that someone who has not smoked a cigarette for years
would remain at increased risk of suicide. A more complex mechanism must be at play, perhaps
one with cigarette smoking serving as a marker for an underlying suicide ideation/attempt
diathesis, or with an influence on transient monoamine pathways, followed by more durable
pathological process not well indicated by transient MAO levels.
In this context, it may be noteworthy that the CPES analysis showed a null association
between former smoking and suicide related outcomes—achieved after a more complete
control of mental health status and social support. Also, after these adjustments, the estimates
for cigarette smoking were greatly reduced (about half the size of the unadjusted estimates).
One interpretation is that some of the observed association between current smoking and
suicide and all of the association between former smoking and suicide might be attributable to
mental disorders and perhaps levels of social support (as measured via self-report items).
Perhaps this would explain why former smokers had increased odds of occurrence of suicide
related outcomes in the other two analyses, which did not adjust for these covariates. Still, it is
important to consider that the appropriate comparison might be “current smokers vs. former
47

smokers” instead of “current smokers vs. never smokers”. For example, the considerable
overlap in confidence intervals of the odds ratios for current and former smokers in chapter 3
shows that current smokers might not always differ from former smokers in terms of suicide
related outcomes. In short, being a former smoker should not independently predict suicide
related outcomes if cigarette smoking is an actual cause of suicide by its transient effects on
serotonin metabolism and MAO pathways. An excess risk of suicide among former smokers is
more likely due to characteristics that influence whether someone becomes a smoker, unless
there is some type of lasting change induced by smoking, over and above what is manifest in
transient MAO pathways.
Also speaking against SI/SA outcomes attributed to smoking, the evidence of these
analyses disclosed no clear dose-response relationship, whether when looking at elapsed time
since smoking onset or frequency of use among current users. Some experts make a judgment
that a dose-response relationship should be established before a causal inference is drawn.
These experts might use the absence of dose-response gradients as evidence against a causal
relationship.
The general failure to find an association between smokeless tobacco use and suicide
related outcomes deserves some discussion. If tobacco induced changes in MAO levels or some
other tobacco or nicotine alteration of biological mechanism were indeed responsible for a
smoking-suicide relationship, I would expect to find the same relationship among smokeless
tobacco users. This evidence is limited by our current biological understanding of the
relationship. For example, while smokeless tobacco delivers nicotine at a similar dose compared
to smoking cigarettes, nicotine reaches the brain faster when smoked than when absorbed
48

through the mouth and nose (~10 seconds vs. ~4 minutes) and this variation across tobacco
products might alter the effect of tobacco exposure on outcomes.(Benowitz, 2011) Still, there is
a possibility that cigarette smoking pharmacologically affects suicide risk through a current
unknown mechanism. For example, thousands of chemicals are found in cigarette smoke that
are not found in smokeless tobacco. It is completely possible that any of these, or perhaps a
combination of these other chemicals, affect the suicide outcomes. If so, there is an explanation
for the observed smoking-suicide relationship and the lack of smokeless tobacco-suicide
relationship.
The concept of combination of chemicals is worth stressing. Perhaps nicotine does play
a small, direct role on suicide risk, but this increased risk is not large enough to observe in these
studies. Instead the excess risk found in smokers could perhaps be due to an accumulation of
many different chemicals—not all of which are also found in smokeless tobacco.
Still, I consider the null findings about the association between smokeless tobacco use
and suicide outcomes to be noteworthy and novel. This dissertation research constitutes the
first studies ever to explore this relationship.
There is an important limitation of the dissertation research in relation to the outcome
variables under study. Beyond my own deep interest in studying “completed suicide” as an
outcome instead of SI and SA behaviors, I can perceive a problem with how these suicide
related variables generally have been measured. In keeping with contemporary practice, these
large, cross-sectional surveys ask individuals if they had ever seriously thought about
committing suicide. However, only those who had said yes were asked about suicidal planning,
and only those who indicated they had made a suicide plan had were asked about a suicide
49

attempt. The researchers who designed these assessments assumed these suicidal behaviors
followed an ordered hierarchy.(D. C. Miller, 1991) this might be a fair assumption in some types
of social sciences research with different outcomes—for example, in the United States, anyone
who has attained a college degree generally can be assumed to have attained their high school
diploma; anyone who has finished graduate school can be assumed to have also finished high
school and has a four year college degree. Previous research has shown this ordered hierarchy
is not always the case in suicide outcomes; some attempts are impulsive (e.g., a suicide
attempt without a clear period of ideation).(A. R. Smith et al., 2008) In most cases, this
trajectory might be accurate and those with suicide attempts have suicidal planning and
suicidal ideation (and so forth). However, an estimated 10-20% of suicide attempters do not
follow this assumed hierarchy .(Perez, 2005) As mentioned earlier, these suicide attempts are
thought to stem more from impulsive characteristics of the person, perhaps in response to an
acute stressor. In consequence, this dissertation research misses these types of suicides (i.e.,
lacks sensitivity to their potential presence).
It is not entirely clear how the omission of these suicide outcomes affected the results.
One must consider how they would expect tobacco exposure to influence suicide. For example,
if tobacco use were to increase impulsive behavior (which could lead to suicide) one might
expect a shift away from the null if these non-hierarchicical suicide attempts had been
detected. However, one could also argue that since tobacco users are more likely to be
impulsive (i.e., impulsive individuals more likely to start and continue using tobacco) that
impulsivity could act as a confounder in the observed relationship.(Acton, 2003; Krishnan-Sarin
et al., 2007; Waldeck & Miller, 1997)
50

This aspect of measurement is especially a limitation in the NESARC analyses in chapter
3. NESARC participants were only asked about suicide related outcomes as part of the
depression module, so those who had not reported a two week period of feeling sad, blue or
depressed (or a two week period of anhedonia) were not asked about suicide outcomes at all.
As a check on this source of error, I estimated how many suicide related outcomes
might have been omitted in the NESARC analyses using data from the CPES project. In CPES
data, an estimated 20% of those with suicidal ideation and 15% of those with suicide attempts
did not report a two week period of feeling sad, blue or depressed, indicating that a notable
fraction of possible outcomes where omitted in the NESARC analyses.
Another important set of possible confounding variables in research on cigarette
smoking and suicide outcomes can be found when considering suicide feasibility. When arguing
that the consistently observed association between smoking and suicide is a confounded
relationship, most investigators suggest it is confounded by factors that influence the desire to
commit suicide (e.g., Major Depression, low income, aggression, impulsivity). However, suicide
requires both a desire and a means. Factors such as access to firearms are known to increase
risk of suicide.(Lewiecki & Miller, 2013) Other factors might include access to harmful
chemicals, such as pesticides or pharmaceuticals, which would be lethal when ingested, or
living in an area with tall buildings and bridges, which a suicide attempter could jump. Future
work should consider these details as well, and clarify whether they are important in relation to
ideation and attempts. One might argue that these factors would only affect the relationship
between smoking and completed suicide (i.e., “Why would owning a gun affect one’s risk of
suicidal ideation, plans and attempts instead of only completed suicide?”). But I would argue
51

that access to a firearm or other effective means of suicide might also increase risk of other
suicidal behaviors as well; an individual with access to suicide means might view suicide as
more of a concrete possibility than someone who does not.
There was another limitation with respect to the measure of social support in chapter 5
in which not all items apply to widows, divorcees or those never married, as question 11 refers
to their spouse. Similarly, there are questions about relatives not living with the participant
which might be interpreted and answered differently by adopted children not living with their
biological family. However, this is an example of why it is advantageous to measure constructs
such as social support using multi-item scales.
Before discussion of possible implications that might arise from this research, I would
ask the reader to suspend judgment about the limitations on this research and imagine if
cigarette smoking were indeed a cause of suicide and that smokeless tobacco was not. Some
advocates might consider this situation to favor tobacco harm reduction approaches. There is
currently a debate over whether cigarette smokers should be encouraged to switch to using
smokeless tobacco due to the much larger chronic disease risk found in association with
smoking cigarettes versus using smokeless tobacco.(Benowitz, 2011) This is clearly a
controversial opinion, and, to the best of my knowledge, is not currently applied by public
health officials, but the results of this research might aid in that discussion. Perhaps a better
example would be in favor of other FDA licensed nicotine replacement therapy, or even the socalled 'e-cigarettes' with a 100% nicotine formulation, not yet regulated in the United States. If
cigarette smoking were to be a cause of suicide, but exposure to nicotine by itself is not, then

52

nicotine replacement therapies, such as nicotine patches and gum, or nicotine user who
exclusively consumes the e-cigarette , would not also increase an ex-smoker’s risk of suicide.
Also, if cigarette smoking were to be an established cause of suicide, it would be
important for practicing psychiatrists to work with their psychiatric patients and their patient’s
primary care physician on the role of smoking cessation plans, depending upon the nature and
degree of the possibly suicide-associated psychiatric disturbance. However, quitting smoking is
a stressful experience (e.g., withdrawal), and in some individuals it might increase suicide risk in
the short term. We would need a greater understanding of the effects of smoking cessation
attempts on short term mental health to best tailor possible interventions. Still, there is no
evidence in this dissertation that suggests quitting smoking will increase one’s risk of suicide
relative to current smoking. I consider this an important finding because one’s intuition might
tell them that quitting smoking might leave former smokers at an increased risk of suicide. also,
a recent study by Capron et al., reported that there is no short term increase in anxiety,
depression, or suicidal ideation among those who successfully quit smoking. Future studies
should continue to consider the short term effects on smoking cessation, or smoking cessation
attempts, on suicide (e.g., “What happens to an individual’s suicidal thoughts within the first
week without a cigarette?”).
Another area for future studies would be to add suicide as an outcome variable of
interest in ongoing smoking cessation trials. It would be unethical to randomly assign nonsmokers to start smoking, but one could ethically assign current smokers to smoking cessation
interventions. Unfortunately, these studies would likely be limited to treatment seeking
samples. Still, other ongoing, longitudinal studies collecting data on cigarette smoking could
53

provide useful analyses by using the within subjects case-crossover approach. In one of these
hypothetical studies, investigators would have repeated measures on smoking status (perhaps
annually) and assessments of suicide attempt. For each individual who attempts suicide
investigators would ask if the patient was smoking or not in a defined time window leading up
to the suicide attempt (ideally an interval as fine grained as the month prior to the suicide
attempt, but the year prior to the suicide attempt would work as well). They would then
compare the likelihood of smoking during this time period, the hazard interval, to the likelihood
of smoking at another period in their life. The major strength of this design is that it holds
constant important confounding variables about the person because the individual is used as
their own control. A limitation here would be that it would not control for time-sensitive
confounders that might influence whether someone attempted to quit smoking, successfully
quit smoking and also attempted suicide.
This raises another important limitation on research of this topic. The ability and desire
to quit smoking are very likely influenced by one’s suicidal tendencies. Someone who has
thoughts and plans to commit suicide probably is not putting healthy changes like quitting
smoking high on their priorities. This problem is actually two-fold: (1) as I stated, suicidal
behaviors might decrease one’s desire to quit smoking and (2) suicidal behaviors also likely
decrease the likelihood of smoking cessation among those who do have a desire to quit
smoking. This circumstance complicates studies on this topic. As a starting point on this issue, I
would suggest more qualitative studies on the relationship between smoking and suicide which
focuses on the individual’s motivations to quit or continue smoking. Do individuals at a high risk
for suicide continue smoking because they feel that the effects of suicide help their mood? Do
54

they simply not care about quitting smoking because they are unhappy with their lives and are
considering suicide? Do they lack the motivation to quit for other reasons? Are they interested
in quitting, but their failed cessation attempts give them a feeling of helplessness which
increases their suicidal feelings? These motivations are not currently well understood, and a
more qualitative investigation on this topic would be helpful.
On a similar note, in the general population of the United States, over half of smokers
are interested in quitting smoking and have made a quit attempt in the past year.(Lewiecki &
Miller, 2013). From this evidence, one could argue that only 50% of current smokers are
smokers because they want to be (i.e., they don’t want to quit), while the other 50% are
current smokers even though they have no desire to continue to smoke (i.e., they are trying to
quit). If cigarette smoking were in fact a cause of suicide with biological and pharmacological or
toxicological mechanisms, then it should predict suicide in each of these smoking subgroups.
One suspects that these mechanisms of a smoking-suicide association are not altered by the
person’s thoughts and feelings about smoking.
Another direction for future studies involves studying this relationship outside of the US.
As shown in the literature review, most of the evidence on the relationship between cigarette
smoking and suicide has been completed in the US or other Western countries. One study in
China and one in Japan found little to no evidence of a tobacco-suicide relationship. Outside of
this dissertation work, I have provided an outline for a future study on smoking and suicide
from a global perspective in the using data from the World Mental Health Surveys Initiative.
Such a study would be helpful because the WMH uses a standardized assessment across 21

55

different countries in the world. Notwithstanding the value of multi-country research on this
topic, studies within individual countries on smoking and suicide would also be helpful.

56

CHAPTER 7: SUMMARY
The relationship between smoking and suicide remains unclear and deserves further
study, even though we in the public health community already have identified many reasons to
prevent the use of tobacco products and to encourage users to quit. While the tobacco-suicide
relationship may not be the most important public health problem today in terms of disease
burden, I consider this a noteworthy topic because we generally view most cases of completed
suicide as feasibly preventable, and the same can be said about cigarette smoking onsets.
The main contribution of this dissertation is its evidence on smokeless tobacco. It is
noteworthy that smokeless tobacco use does not appear to increase suicide outcome risk while
cigarette smoking appears to account for an increased risk. In addition, while a considerable
proportion of the relationship between cigarette smoking and suicide may well be explained by
mental illness status, this work suggests smokers are at increased odds of suicide outcomes
such as ideation and attempt, even with common mental illnesses have been taken into
account.

57

APPENDICES

58

APPENDIX A: Tables and Figures

Figure 3.1 Flow chart identifying newly incident cases of suicide attempts with complete
follow-up. Data from the US National Epidemiologic Survey on Alcohol and Alcohol Related
Conditions (NESARC), Wave I 2001-2002 (n=43,093), Wave II 2004-2005 (n=34,653).
NESAC Wave I (2001-2002)
n=43,093 total participants

No history of suicide attempt at Wave 1

History of suicide attempt at Wave 1

n=41,936 (97.3%)

n=1,157 (2.7%)

s

Wave 2 follow-up
n=33,687 (80.3%)

Wave 2 follow-up

No Wave 2
follow-up

No Wave 2
follow-up

n=966 (83.5%)
n=191 (16.5%)

n=8,249
(19.7%)

n=33,687
143 newly incident cases of suicide attempt
33,535 individuals without a suicide attempt

9 with missing data on incident suicide attempt
Analysis sample=33,678

59

Figure 3.2 Flow chart identifying newly incident cases of suicidal ideation with complete
follow-up. Data from the US National Epidemiologic Survey on Alcohol and Alcohol Related
Conditions (NESARC), Wave I 2001-2002 (n=43,093), Wave II 2004-2005 (n=34,653).

NESAC Wave I (2001-2002)
n=43,093 total participants

History of suicidal ideation at
Wave 1

No history of suicidal ideation at Wave 1
n=39,438 (91.5%)

n=3,655 (8.5%)

Wave 2 followup

No Wave 2
follow-up

Wave 2 followup

No Wave 2
follow-up

n=31,552

n=7,886
(20.0%)

n=3,101

n=554(14.5%)

n=31,552
688 newly incident cases of suicidal ideation
30,856 individuals without suicidal ideation
8 with missing data on suicidal ideation
Analysis sample=31,544

60

Figure 3.3 Flow chart identifying newly incident cases of suicidal behavior (suicidal ideation
and attempts combined) with complete follow-up. Data from the US National Epidemiologic
Survey on Alcohol and Alcohol Related Conditions (NESARC), Wave I 2001-2002 (n=43,093),
Wave II 2004-2005 (n=34,653).

NESAC Wave I (2001-2002)

n=43,093 participants

No history of suicidal behavior at
Wave 1

History of suicidal behavior at
Wave 1

n=39,397 (91.4%)

n=3,696 (8.6%)

Wave 2
follow-up

No Wave 2
follow-up

Wave 2
follow-up

No Wave 2
follow-up

n=31,520

n=7,877
(20.0%)

n=3,133

n=563
(15.2%)

n=31,520
702 cases of newly incident suicidal behavior
30,811 individuals without suicidal behavior
7 with missing data on suicidal behavior
Analysis sample=31,520

61

Table 3.1 Distribution of newly incident cases of suicide attempt by covariates of interest.
Data from the US National Epidemiologic Survey on Alcohol and Alcohol Related Conditions
(NESARC), Wave I 2001-2002 (n=43,093), Wave II 2004-2005 (n=33,687).
All
Newly incident suicide attempt during
persons
follow-up
n
n
%*
All persons
33,678
143
0.4
Baseline cigarette smoking
Current
7,469
59
0.7
Former
6,495
21
0.3
Never
19,457
62
0.3
Missing
257
†
Baseline smokeless tobacco use
Current
854
6
0.9
Former
1,176
3
0.3
Never
31,406
133
0.4
Missing
242
Baseline chewing tobacco use
Current
396
2
0.9
Former
1,054
3
0.3
Never
31,980
137
0.4
Missing
248
Baseline snuff use
Current
588
4
0.8
Former
722
3
0.7
Never
32,122
135
0.4
Missing
246
§
Lifetime Major Depression at baseline
Yes
5,772
51
0.9
No
27,906
92
0.3
Lifetime Alcohol Use Disorder at baseline§
Yes
9,458
49
0.5
No
24,220
94
0.4
Sex
Male
14,275
44
0.3
Female
19,403
99
0.5
(Cont’d on next page)

62

Table 3.1, (cont’d)
All
persons
n

Newly incident suicide attempt during
follow-up
n
%*

Race/ethnicity
Non-Hispanic White
19,590
69
0.3
Non-Hispanic Black
6,420
18
0.3
Non-Hispanic American Indian/Alaska
551
7
1.1
Native
Non-Hispanic Asian/Native
941
6
0.5
Hawaiian/Pacific Islander
Hispanic
6,176
43
0.8
Age (in years)
18-24
3,802
32
0.8
25-34
6,055
31
0.6
35-44
7,294
45
0.5
45-64
10,589
30
0.2
65+
5,938
5
0.1
Education
Less than high school
5,544
39
0.5
High school/GED
9,688
40
0.4
Some college
10,121
46
0.5
College or more
8,325
18
0.2
†
§
*Weighted percentage; Chewing tobacco or snuff use; Depression and alcohol use disorders
were identified using a structured clinical interview. Alcohol Use Disorder included either DSMIV alcohol dependence with or without abuse.

63

Table 3.2 Distribution of newly incident cases of suicidal ideation by covariates of interest.
Data from the US National Epidemiologic Survey on Alcohol and Alcohol Related Conditions
(NESARC), Wave I 2001-2002 (n=43,093), Wave II 2004-2005 (n=31,544).
All Newly incident suicide attempt during
persons follow-up
n
n
%*
All persons
31,544
688
2.1
Baseline cigarette smoking
Current
6,754
252
3.6
Former
6,083
106
2.0
Never
18,450
324
1.6
Missing
257
----†
Baseline smokeless tobacco use
Current
802
22
3.0
Former
1,089
31
2.7
Never
29,411
629
2.1
Missing
242
----Baseline chewing tobacco use
Current
372
15
4.8
Former
985
25
2.1
Never
29,939
641
2.1
Missing
779
----Baseline snuff use
Current
554
15
2.8
Former
663
17
2.8
Never
30,081
650
2.1
Missing
246
----§
Lifetime Major Depression at baseline
Yes
3,959
181
4.6
No
27,585
507
1.8
Lifetime Alcohol Use Disorder at
baseline§
Yes
8,445
245
2.8
No
23,099
443
1.9
Sex
Male
13,526
268
1.9
Female
18,018
420
2.3
(Cont’d on next page)

64

Table 3.2, (cont’d)
All
persons
n

Newly incident suicide attempt during
follow-up
n
%*

Race/ethnicity
Non-Hispanic White
18,099
392
2.1
Non-Hispanic Black
6,137
118
2.0
Non-Hispanic American Indian/Alaska
484
19
3.8
Native
Non-Hispanic Asian/Native
918
21
2.2
Hawaiian/Pacific Islander
Hispanic
5,906
138
2.2
Age (in years)
18-24
3,511
122
3.2
25-34
5,656
141
2.4
35-44
6,835
180
2.6
45-64
9,784
190
1.9
65+
5,758
55
1.0
Education
Less than high school
5,263
140
2.6
High school/GED
9,154
202
2.3
Some college
9,377
222
2.4
College or more
7,750
124
1.4
†
§
*Weighted percentage; Chewing tobacco or snuff use; Depression and alcohol use disorders
were identified using a structured clinical interview. Alcohol Use Disorder included either DSMIV alcohol dependence with or without abuse .

65

Table 3.3 Distribution of newly incident cases of suicidal behavior (suicidal ideation and
attempts combined by covariates of interest. Data from the US National Epidemiologic Survey
on Alcohol and Alcohol Related Conditions (NESARC), Wave I 2001-2002 (n=43,093), Wave II
2004-2005 (n=34,513).
All
Newly incident suicide attempt during
persons
follow-up
n
n
%*
All persons
31,513
702
2.2
Baseline cigarette smoking
Current
6,740
254
3.7
Former
6,077
110
2.0
Never
18,446
332
1.7
Missing
257
----Baseline smokeless tobacco use†
Current
801
24
3.3
Former
1,085
30
2.6
Never
29,385
642
2.1
Missing
242
----Baseline chewing tobacco use
Current
371
16
5.4
Former
982
24
2.0
Never
29,912
655
2.1
Missing
248
----Baseline snuff use
Current
554
16
2.8
Former
661
17
3.0
Never
30,052
663
2.1
Missing
246
----Lifetime Major Depression at baseline§
Yes
3,931
520
4.7
No
27,582
182
1.8
§
Lifetime Alcohol Use Disorder at baseline
Yes
8,434
247
2.9
No
23,079
455
1.9
Sex
Male
13,520
270
1.9
Female
17,993
432
2.4
(Cont’d on next page)

66

Table 3.3,( cont’d)
All
persons

Newly incident suicide attempt during
follow-up
n
n
397
2.1
119
2.1
20
4.3

Race/ethnicity
Non-Hispanic White
18,087
Non-Hispanic Black
6,132
Non-Hispanic American Indian/Alaska
482
Native
Non-Hispanic asian/Native
914
22
2.2
Hawaiian/Pacific Islander
Hispanic
5,898
144
2.3
Age (in years)
18-24
3,509
124
3.2
25-34
5,650
145
2.5
35-44
6,830
182
2.6
45-64
9,770
195
1.9
65+
5,754
56
1.0
Education
Less than high school
5,259
146
2.6
High school/GED
9,141
202
2.3
Some college
9,361
227
2.5
College or more
7,752
127
1.4
*Weighted percentage; † Chewing tobacco or snuff use; § Depression and alcohol use disorders
were identified using a structured clinical interview. Alcohol Use Disorder included either DSMIV alcohol dependence with or without abuse.

67

Table 3.4 Unadjusted and adjusted models for the associations between covariates of interest
and newly incident suicide attempts. Data from the US National Epidemiologic Survey on
Alcohol and Alcohol Related Conditions (NESARC), Wave I 2001-2002 (n=43,093), Wave II
2004-2005 (n=34,653).
Unadjusted
Adjusted I
OR[95% CI]
p-value OR[95% CI] p-value
Baseline cigarette smoking
Current
2.2[1.4,3.4]
0.770 2.0[1.3,3.1]
0.002
Former
0.9[0.5,1.8]
<0.001 1.6[0.8,3.2]
0.165
Never
1.0(ref)
--1.0(ref)
--Baseline smokeless tobacco use†
Current
2.4[0.9,6.3]
0.071 2.7[0.9,8.6]
0.083
Former
0.7[0.2,2.3]
0.504 0.8[0.2, 3.1]
0.783
Never
1.0(ref)
--1.0(ref)
--Major depression prior to cigarette onset
Yes
4.2[1.3,13.7]
0.020
No
1.0(ref)
--1.0(ref)
--Major depression prior to SLT onset
Yes
N/A
No
1.0(ref)
--1.0(ref)
--Alcohol use disorder prior to cigarette onset
Yes
2.1[0.5,8.5]
0.302
No
1.0(ref)
--1.0(ref)
--Alcohol use disorder prior to SLT onset
Yes
0.2[0.1,1.2]
0.071
No
1.0(ref)
--1.0(ref)
--Sex
Male
0.5[0.3,0.9]
0.010 0.4[0.2,0.7]
0.444
Female
1.0(ref)
--1.0(ref)
--Race/ethnicity
Non-Hispanic White
1.0(ref)
--1.0(ref)
--Non-Hispanic Black
0.7[0.4,1.4]
0.373 0.7[0.3,1.3]
0.239
Non-Hispanic American Indian/Alaska Native
3.4[1.3,8.9]
0.011 2.7[1.1,6.7]
0.037
Non-Hispanic Asian/Native Hawaiian/Pacific
1.5[0.5,4.8]
0.472
1.8[.6,5.6]
0.324
Islander
Hispanic
2.3[1.4,3.9]
0.002 1.9[1.0,3.4]
0.034
Age (in years)
18-24
1.0(ref)
--1.0(ref)
--25-34
0.7[0.4,1.4]
0.312 0.8[0.4,1.5]
0.510
35-44
0.6[0.4,1.1]
0.099 0.7[0.4,1.2]
0.212
45-64
0.3[0.1,0.5]
<0.001 0.3[0.2,0.6]
0.001
65+
0.1[0.1,0.3]
<0.001 0.1[0.1,0.3] <0.001
(Cont’d on next page)

68

Table 3.4, (cont’d)

Education
Less than high school
High school/GED
Some college
College or more

69

Unadjusted
OR[95% CI]
p-value

Adjusted I
OR[95% CI] p-value

1.0(ref)
0.8[0.5,1.4]
0.9[0.5,1.5]
0.3[0.2,0.7]

1.0(ref)
0.8[0.5,1.5]
0.8[0.4,1.5]
0.4[0.2,0.8]

--0.436
0.647
0.005

--0.506
0.528
0.017

Table 3.5 Unadjusted and adjusted models for the associations between covariates of
interest and newly incident suicide attempts. Data from the US National Epidemiologic
Survey on Alcohol and Alcohol Related Conditions (NESARC), Wave I 2001-2002 (n=43,093),
Wave II 2004-2005 (n=34,653).
Adjusted II Cigarette
Adjusted II SLT
OR[95% CI]
p-value OR[95% CI] p-value
Baseline cigarette smoking
Current
2.0[1.3,3.1]
0.003
Former
1.6[0.8,3.2]
0.177
Never
1.0(ref)
--1.0(ref)
--†
Baseline smokeless tobacco use
Current
3.2[1.0,10.
0.043
0]
Former
1.1[0.2,4.5]
0.945
Never
1.0(ref)
--1.0(ref)
--Major depression prior to cigarette onset
Yes
2.1[0.6,7.4]
0.235
No
1.0(ref)
--1.0(ref)
--Major depression prior to SLT onset
Yes
N/A
N/A
No
1.0(ref)
--1.0(ref)
--Alcohol use disorder prior to cigarette
onset
Yes
1.7[0.4,7.1]
0.463 0.1[0.1,1.0]
0.055
No
1.0(ref)
--1.0(ref)
--Alcohol use disorder prior to SLT onset
Yes
No
1.0(ref)
--1.0(ref)
--Sex
Male
0.5[0.3,0.8]
0.002
Female
1.0(ref)
--1.0(ref)
--Race/ethnicity
Non-Hispanic White
1.0(ref)
--1.0(ref)
--Non-Hispanic Black
0.6[0.3,1.2]
0.176 0.6[0.3,1.2]
0.128
Non-Hispanic American Indian/Alaska Native
2.8[1.1,7.2]
0.036 2.8[1.1,6.8]
0.027
Non-Hispanic Asian/Native Hawaiian/Pacific
1.7[0.5,5.3]
0.358 1.5[0.5,4.8]
0.460
Islander
Hispanic
1.8[1.0,3.2]
0.046 1.6[0.9,3.0]
0.129

(Cont’d on next page)

70

Table 3.5, (cont’d)

Age (in years)
18-24
25-34
35-44
45-64
65+
Education
Less than high school
High school/GED
Some college
College or more

Adjusted II Cigarette
OR[95% CI] p-value

Adjusted II SLT
OR[95% CI] p-value

1.0(ref)
0.8[0.4,1.5]
0.7[0.4,1.2]
0.3[0.2,0.6]
0.1[0.1,0.3]

1.0(ref)
0.6[0.3,1.2]
0.7[0.4,1.3]
0.3[0.2,0.7]
0.1[0.1,0.3]

--0.128
0.289
0.002
<0.001

--1.0(ref)
0.517 0.8[0.4,1.4]
0.516 0.8[0.4,1.4]0
.391
0.013 0.3[0.2,0.7]

--0.442
0.391

1.0(ref)
0.8[0.5,1.5]
0.8[0.4,1.5]
0.4[0.2,0.8]

71

--0.543
0.224
0.001
<0.001

0.005

Table 3.6 Unadjusted and adjusted models for the associations between covariates of
interest and newly incident suicidal ideation. Data from the US National Epidemiologic
Survey on Alcohol and Alcohol Related Conditions (NESARC), Wave I 2001-2002 (n=43,093),
Wave II 2004-2005 (n=34,653).
Unadjusted
Adjusted I
OR[95% CI] p-value OR[95% CI] p-value
Baseline cigarette smoking
Current
2.3[1.9,2.8] <0.001 2.1[1.7,2.6] <0.001
Former
1.2[0.9,1.6]
0.154 1.6[1.2,2.1]
0.002
Never
1.0(ref)
--1.0(ref)
--Baseline smokeless tobacco use†
Current
1.4[0.8,2.1]
0.150 1.3[0.8,2.1]
0.374
Former
1.3[0.8,2.1]
0.306 1.2[0.7,2.0]
0.513
Never
1.0(ref)
--1.0(ref)
--Major depression prior to cigarette onset
Yes
1.5[0.6,3.8]
0.397
No
1.0(ref)
--1.0(ref)
--Major depression prior to SLT onset
Yes
5.5[1.0,29.3
0.045
]
No
1.0(ref)
--1.0(ref)
--Alcohol use disorder prior to cigarette
onset
Yes
1.8[0.8,4.3]
0.183
No
1.0(ref)
--1.0(ref)
--Alcohol use disorder prior to SLT onset
Yes
3.2[1.3,8.2]
0.011
No
1.0(ref)
--1.0(ref)
--Sex
Male
0.8[0.7,1.0]
0.026 0.7[0.6,0.9]
0.001
Female
1.0(ref)
--1.0(ref)
--Race/ethnicity
Non-Hispanic White
1.0(ref)
--1.0(ref)
--Non-Hispanic Black
1.0[0.7,1.3]
0.901 0.9[0.7,1.2]
0.559
Non-Hispanic American Indian/Alaska
1.9[1.1,3.2]
0.021 1.6[0.9,2.7]
0.087
Native
Non-Hispanic Asian/Native Hawaiian/Pacific 1.1[0.6,1.9]
0.850 1.2[0.6,2.2]
0.581
Islander
Hispanic
1.1[0.8,1.4]
0.611 0.9[0.7,1.3]
0.695
(Cont’d on next page)

72

Table 3.6, (cont’d)

Age (in years)
18-24
25-34
35-44
45-64
65+
Education
Less than high school
High school/GED
Some college
College or more

Unadjusted
OR[95% CI] p-value

Adjusted I
OR[95% CI] p-value

1.0(ref)
0.7[0.5,1.0]
0.8[0.6,1.0]
0.6[0.4,0.8]
0.3[0.2,0.4[

--0.047
0.102
<0.001
<0.001

1.0(ref)
0.8[0.6,1.1]
0.8[0.6,1.1]
0.6[0.4,0.8]
0.3[0.2,0.4]

--0.142
0.164
0.001
<0.001

1.0(ref)
0.9[0.7,1.2]
0.9[0.7,1.2]
0.5[0.4,0.7]

--0.387
0.622
<0.001

1.0(ref)
0.8[0.6,1.1]
0.8[0.6,1.1]
0.6[0.4,0.8]

--0.164
0.220
0.002

73

Table 3.7 Unadjusted and adjusted models for the associations between covariates of
interest and newly incident suicidal ideation. Data from the US National Epidemiologic
Survey on Alcohol and Alcohol Related Conditions (NESARC), Wave I 2001-2002 (n=43,093),
Wave II 2004-2005 (n=34,653).
Adjusted II Cigarette
Adjusted II SLT
OR[95% CI] p-value OR[95% CI] p-value
Baseline cigarette smoking
Current
2.1[1.7,2.6] <0.001
Former
1.5[1.2,2.1]
0.003
Never
1.0(ref)
--1.0(ref)
--†
Baseline smokeless tobacco use
Current
1.1[0.7,1.9]
0.623
Former
1.1[0.7,1.9]
0.647
Never
1.0(ref)
--1.0(ref)
--Major depression prior to cigarette onset
Yes
0.9[0.3,2.3]
0.792
No
1.0(ref)
--1.0(ref)
--Major depression prior to SLT onset
Yes
2.5[0.3,22.8
0.423
]
No
1.0(ref)
--1.0(ref)
--Alcohol use disorder prior to cigarette
onset
Yes
1.4[0.6,3.3]
0.491 3.5[1.3,9.7]
0.013
No
1.0(ref)
--1.0(ref)
--Alcohol use disorder prior to SLT onset
Yes
No
1.0(ref)
--1.0(ref)
--Sex
Male
0.7[0.6,0.9]
0.001 0.8[0.6,0.9]
0.006
Female
1.0(ref)
--1.0(ref)
--Race/ethnicity
Non-Hispanic White
1.0(ref)
--1.0(ref)
--Non-Hispanic Black
0.9[0.7,1.2] 0.4910 0.8[0.6,1.1]
0.217
Non-Hispanic American Indian/Alaska
1.6[0.9,2.7]
0.085 1.6[1.0,2.8]
0.058
Native
Non-Hispanic Asian/Native
1.2[0.6,2.2]
0.613 1.0[0.6,1.9]
0.909
Hawaiian/Pacific Islander
Hispanic
0.9[0.7,1.3]
0.668 0.8[0.6,1.1]
0.169
(Cont’d on next page)

74

Table 3.7, (cont’d)

Age (in years)
18-24
25-34
35-44
45-64
65+
Education
Less than high school
High school/GED
Some college
College or more

Adjusted II Cigarette
OR[95% CI] p-value

Adjusted II SLT
OR[95% CI] p-value

1.0(ref)
0.8[0.6,1.1]
0.8[0.6,1.1]
0.6[0.4,0.8]
0.3[0.2,0.4]

--0.126
0.145
0.001
<0.001

1.0(ref)
0.8[0.6,1.1]
0.8[0.6,1.1]
0.6[0.4,0.8]
0.3[0.2,0.4]

--0.217
0.216
0.001
<0.001

1.0(ref)
0.8[0.6,1.1]
0.8[0.6,1.1]
0.6[0.4,0.8]

--0.178
0.220
0.001

1.0(ref)
0.8[0.6,1.0]
0.8[0.6,1.0]
0.5[0.3,0.7]

--0.095
0.083
<0.001

75

Table 3.8 Unadjusted and adjusted models for the associations between covariates of
interest and newly incident suicidal behavior. Data from the US National Epidemiologic
Survey on Alcohol and Alcohol Related Conditions (NESARC), Wave I 2001-2002 (n=43,093),
Wave II 2004-2005 (n=34,653).
Unadjusted
Adjusted I
OR[95% CI] p-value
OR[95% CI] p-value
Baseline cigarette smoking
Current
2.2[1.9,2.7] <0.001 2.0[1.7,2.5] <0.001
Former
1.2[0.9,1.6]
0.160 1.6[1.2,2.1]
0.002
Never
1.0(ref)
--1.0(ref)
--Baseline smokeless tobacco use†
Current
1.6[1.0,2.5]
0.074 1.4[0.7,2.3]
0.194
Former
1.2[0.7,2.0]
0.452 1.1[0.7,1.9]
0.625
Never
1.0(ref)
--1.0(ref)
--Major depression prior to cigarette onset
Yes
1.5[0.6,3.9]
0.373
No
1.0(ref)
--1.0(ref)
--Major depression prior to SLT onset
Yes
5.4[1.0,28.7]
0.048
No
1.0(ref)
--1.0(ref)
--Alcohol use disorder prior to cigarette
onset
Yes
1.8[0.8,4.3]
0.177
No
1.0(ref)
--1.0(ref)
--Alcohol use disorder prior to SLT onset
Yes
3.3[1.3,8.0]
0.010
No
1.0(ref)
--1.0(ref)
--Sex
Male
0.8[0.7,0.9]
0.011 0.7[0.6,0.9]
0.001
Female
1.0(ref)
--1.0(ref)
--Race/ethnicity
Non-Hispanic White
1.0(ref)
--1.0(ref)
--Non-Hispanic Black
1.0[0.7,1.3]
0.868 0.9[0.7,1.2]
0.535
Non-Hispanic American Indian/Alaska
2.1[1.2,3.5]
0.007 1.8[1.0,3.0]
0.035
Native
Non-Hispanic Asian/Native
1.1[0.6,1.9]
0.847 1.2[0.6,2.2]
0.585
Hawaiian/Pacific Islander
Hispanic
1.1[0.9,1.4]
0.421 1.0[0.7,1.3]
0.895
(Cont’d on next page)

76

Table 3.8, (cont’d)

Age (in years)
18-24
25-34
35-44
45-64
65+
Education
Less than high school
High school/GED
Some college
College or more

Unadjusted
OR[95% CI] p-value

Adjusted I
OR[95% CI] p-value

1.0(ref)
0.7[0.5,1.0]
0.8[0.6,1.1]
0.6[0.4,0.8]
0.3[0.2,0.4]

--0.077
0.115
<0.001
<0.001

1.0(ref)
0.8[0.6,1.1]
0.8[0.6,1.1]
0.6[0.4,0.8]
0.3[0.2,0.4]

--0.214
0.190
0.002
<0.001

1.0(ref)
0.9[0.7,1.1]
0.9[0.7,1.2]
0.5[0.4,0.7]

--0.286
0.634
<0.001

1.0(ref)
0.8[0.6,1.1]
0.8[0.6,1.1]
0.6[0.4,0.8]

--0.119
0.226
0.001

77

Table 3.9 Unadjusted and adjusted models for the associations between covariates of
interest and newly incident suicidal behavior. Data from the US National Epidemiologic
Survey on Alcohol and Alcohol Related Conditions (NESARC), Wave I 2001-2002
(n=43,093), Wave II 2004-2005 (n=34,653).
Adjusted II Cigarette
Adjusted II SLT
OR[95% CI] p-value OR[95% CI] p-value
Baseline cigarette smoking
Current
2.0[1.7,2.5] <0.001
Former
1.6[1.2,2.1]
0.003
Never
1.0(ref)
--†
Baseline smokeless tobacco use
Current
1.3[0.7,2.1]
0.380
Former
1.1[0.7,1.8]
0.767
Never
1.0(ref)
--1.0(ref)
--Major depression prior to cigarette onset
Yes
0.9[0.3,2.3]
0.813
No
1.0(ref)
--1.0(ref)
--Major depression prior to SLT onset
Yes
1.4[0.6,3.3]
0.467 2.4[0.2,23.5]
0.455
No
1.0(ref)
--1.0(ref)
--Alcohol use disorder prior to cigarette
onset
Yes
No
1.0(ref)
--1.0(ref)
--Alcohol use disorder prior to SLT onset
Yes
3.6[1.3,9.6]
0.013
No
1.0(ref)
--1.0(ref)
--Sex
Male
0.7[0.6,0.9]
0.001 0.7[0.6,0.9]
0.003
Female
1.0(ref)
--1.0(ref)
--Race/ethnicity
Non-Hispanic White
1.0(ref)
--1.0(ref)
--Non-Hispanic Black
0.9[0.7,1.2]
0.459 0.8[0.6,1.1]
0.212
Non-Hispanic American Indian/Alaska
1.8[1.0,3.0]
0.034 1.8[1.1,3.0]
0.022
Native
Non-Hispanic Asian/Native
1.2[0.6,2.2]
0.623 0.6[0.5,0.8]
0.908
Hawaiian/Pacific Islander
Hispanic
1.0[0.7,1.3]
0.847 0.8[0.6,1.1]
0.268
(Cont’d on next page)

78

Table 3.9, (cont’d)

Age (in years)
18-24
25-34
35-44
45-64
65+
Education
Less than high school
High school/GED
Some college
College or more

Adjusted II Cigarette
OR[95% CI] p-value

Adjusted II SLT
OR[95% CI] p-value

1.0(ref)
0.8[0.6,1.1]
0.8[0.6,1.1]
0.6[0.4,0.8]
0.3[0.2,0.4]

--0.188
0.160
0.001
<0.001

1.0(ref)
0.8[0.6,1.1]
0.9[0.6,1.1]
0.6[0.5,0.8]
0.3[0.2,0.4]

--0.233
0.249
0.002
<0.001

1.0(ref)
0.8[0.6,1.1]
0.8[0.6,1.1]
0.6[0.1,0.8]

--0.130
0.226
0.001

1.0(ref)
0.8[0.6,1.0]
0.8[0.6,1.0]
0.5[0.3,0.7]

--0.068
0.087
<0.001

79

Table 3.10 Models for the associations between covariates of interest and occurrence of
suicidal behavior. Data from the US National Epidemiologic Survey on Alcohol and Alcohol
Related Conditions (NESARC), Wave I 2001-2002 (n=43,093), Wave II 2004-2005
(n=34,653).
Suicide attempt
Suicidal ideation
Either suicidal
behavior
First occurrence
OR[95% CI]
pOR[95% CI]
p- OR[95% CI] p-value
value
value
Baseline cigarette
smoking
Current
2.3[2.0,2.7] <0.001
1.8[1.1,2.9] 0.023 2.2[1.8,2.7] <0.001
Former
1.2[1.0,1.5] 0.106
0.8[0.4,1.8] 0.620 1.2[0.9,1.6]
0.178
Never
1.0(ref)
--1.0(ref)
--1.0(ref)
--Baseline smokeless
tobacco use†
Current
1.1[0.7,1.7] 0.550
2.9[1.1,7.5] 0.034 1.4[0.8,2.2]
0.209
Former
1.1[0.7,1.3] 0.449
0.8[0.2,2.8] 0.700 1.0[0.8,2.2]
0.990
Never
1.0(ref)
--1.0(ref)
--1.0(ref)
--First occurrence or
reoccurrence
Baseline cigarette
smoking
Current
2.9[2.1,4.2] <0.001
2.5[2.2,2.9] <0.00 2.5[2.2,2.9] <0.001
1
Former
0.8[0.5,1.4] 0.456
1.1[0.9,1.3] 0.283 1.1[0.9,1.3]
0.313
Never
1.0(ref)
--1.0(ref)
--1.0(ref)
--Baseline smokeless
tobacco use†
Current
1.4[0.6,3.2] 0.487
1.1[0.7,1.6] 0.796 1.1[0.7,1.7]
0.608
Former
0.7[0.3,1.8] 0.430
0.8[0.5,1.2] 0.309 0.8[0.6,1.2]
0.391
Never
1.0(ref)
--1.0(ref)
--1.0(ref)
--*Estimates for cigarette smoking and smokeless tobacco are evaluated from the same
model

80

Table 4.1 Suicidal behaviors among current, former and never tobacco users. Data from the
US National Survey on Drug Use and Health (NSDUH), 2008-2010 (n=114,048).
All persons
Ideation
Plan
Attempt
n

%†

n

n

%†
n
%†
1,858
1.1 1,005
0.5

All persons
114,048
6,197
3.8
Cigarette Smokers
Never
37,186
1,445
2.5
392
0.1
193
0.3
Current
43,070
3,360
6.3 1,094
1.9
665
1.0
Former (quit 1-3 years)
6,553
335
4.5
92
1.2
46
0.4
Former (quit >3 years)
27,239
1,057
2.8
280
0.7
101
0.3
§
Smokeless Tobacco Users
Never
89,789
4,713
0.4 1,416
1.0
762
0.4
Current
8,444
560
0.5
173
1.6
109
0.8
Former (quit 1-3 years)
3,227
194
0.5
59
1.3
38
1.0
Former (quit >3 years)
12,588
730
0.5
210
1.2
96
0.5
Chewing Tobacco Users
Never
97,888
5,213
3.6 1,563
1.0
839
0.5
Current
3,934
295
7.0
92
2.1
64
1.2
Former (quit 1-3 years)
2,390
152
4.8
50
1.2
28
0.6
Former (quit >3 years)
9,836
537
4.2
153
1.0
74
0.5
Snuff Users
Never
94,358
4,956
3.7 1,494
1.0
799
0.4
Current
7,268
481
5.3
145
1.7
89
0.9
Former (quit 1-3 years)
2,674
173
5.5
49
1.4
31
1.1
Former (quit >3 years)
9,748
587
4.9
170
1.3
86
0.6
*75,964 persons at least of 18 years of age were surveyed. 250 were missing for recency of
cigarette use; 210 were missing for recency of smokeless tobacco use; 107 were missing for
recency of chewing tobacco use; 156 were missing for recency of snuff use; 274 were missing
for suicidal ideation; 2 were missing for suicidal plan; 4 were missing for suicide attempt.
†
Row percentages
§
Chewing tobacco use and/or snuff use

81

Table 4.2 Characteristics of the sample by lifetime tobacco use.
Total
Cigarette
Smokeless
sample
n
n
%*
n
%*
All persons
114,048 76,862 69.3 24,259 19.1
Sex
Male
53,097 38,073 74.8 19,634 33.4
Female
60,951 38,789 64.2 4,625
5.3
Race
Non-Hispanic
72,819 53,372 75.5 19,791 23.5
white
Non-Hispanic
13,974 7,500 58.3
976
9.2
black
Non-Hispanic
1,677 1,332 73.3
681 37.1
American
Indian/Alaskan
native
Non-Hispanic
532
333 53.1
94 15.4
Pacific Islander
Non-Hispanic
4,322 1,884 38.8
281
5.6
Asian
Non-Hispanic
3,061 2,292 77.2
747 24.4
multiracial
Hispanic
17,663 10,149 57.4 1,689
9.1
Age
18-25
56,546 35,976 63.2 12,032 79.3
26-34
17,009 12,114 71.2 4,048 76.8
≥35
40,493 28,772 70.2 8,179 82.2
Family income
<$20k
28,593 18,525 65.3 5,179 15.5
$20k-$49k
39,365 26,734 68.5 8,059 17.7
$50k-$74k
18,725 12,846 71.1 4,309 20.5
≥$75k
27,365 18,757 71.4 6,712 21.9
Pop. Density
MSA>1million
47,732 30,830 67.6 7,386 15.2
MSA<1million
57,211 39,394 70.9 13,935 22.5
Non-MSA
9,105 6,638 73.3 2,938 29.0
Depression:
interview
Yes
16,292 12,551 79.6 3,531 20.6
No
97,276 64,029 67.8 20,652 18.9
(Cont’d on next page)

82

Chewing

Snuff

n
16,160

%*
n
13.9 19,690

%*
14.2

13,924
2,236

26.0 16,102
2.6 3,588

25.3
3.9

13,259

17.3 16,245

17.7

611

6.3

687

6.0

494

28.4

567

27.8

53

7.7

74

11.7

173

3.8

203

3.4

519

18.5

599

17.9

1,051

6.2

1,315

6.6

7,259
2,635
6,266

12.3 10,251
15.1 3,371
14.0 6,068

17.8
19.3
12.3

3,263
5,405
2,945
4,547

10.7
12.9
15.1
16.2

4,231
6,473
3,494
5,492

11.5
12.9
15.3
16.5

4,609
9,414
2,137

10.7 5,953
16.6 11,314
22.4 2,423

11.1
16.9
22.0

2,309
13,800

14.5 2,898
13.9 16,735

15.9
14.0

Table 4.2, (cont’d)
Total
Cigarette
Smokeless
Chewing
Snuff
sample
n
n
%*
n
%*
n
%*
n
%*
Depression: selfreport
Yes
14,182 2,848 80.1 3,086 19.0 1,959 13.3 2,473 14.0
No
40,934 12,672 70.6 8,688 19.3 6,007 14.7 6,882 13.6
Alcohol use
disorder
Yes
13,332 11,790 88.4 5,403 37.4 3,651 27.2 4,631 30.8
No
100,716 65,072 67.8 18,856 17.6 12,509 12.8 15,059 12.9
*Row percentages

83

Table 4.3 Tobacco use characteristics of the sample
Cigarette Smokers
Total sample (n)

Never used
Never used chew
Never used snuff
smokeless (n)
(n)
(n)
Never
37,186
35,134
35,965
35,635
Current
43,070
29,243
22,567
31,564
Former (quit 1-3 years)
6,553
4,681
5,308
4,975
Former (quit >3 years)
27,239
20,731
34,048
22,184
Smokeless Tobacco Users
Total sample (n)
Never smoked
Never used chew
Never used snuff
cigarette (n)
(n)
(n)
Never
89,789
35,134
89,789
89,789
Current
8,444
831
2,542
810
Former (quit 1-3 years)
3,227
248
1,282
535
Former (quit >3 years)
12,588
973
4,275
3,224
*Among the 50,898 cigarette smokers, there were 36,311 who have never used smokeless tobacco (cigarette only
users).
†
Among the 15,958 smokeless tobacco users, there were 1,371 who had never smoked a cigarette (smokeless only
users).

84

Table 4.4 Crude and covariate adjustment logistic regressions predicting past year suicidal related outcomes by recency of
tobacco use. Data from the US National Survey on Drug Use and Health (NSDUH), 2008-2010 (n=114,048).
Cigarette smoking estimates
Past year suicidal ideation

Current
Former (quit
1-3 years)
Former (quit
>3 years)
Never

¶

Past year suicide plan**
†

p

uOR*

p

<0.001

2.3[2.1,2.6]

<0.001

2.8[2.2,3.6]

<0.001

1.5[1.2,1.9]

0.002

1.8[1.3,2.6]

1.1[1.0,1.3]

0.069

1.2[1.1,1.4]

0.002

1.1[0.8,1.5]

1.0

-

1.0 (Ref)

-

1.0 (Ref)

uOR*

p

2.6[2.4,3.0]
1.7[1.3,2.2]

aOR

Past year suicide attempt
†

p

uOR*

p

<0.001

2.5[1.9,3.2]

<0.001

3.2[2.3,4.4]

0.001

1.7[1.2,2.4]

0.005

1.5[0.9,2.3]

0.537

1.3[0.9,1.7]

0.147

-

1.0 (Ref)

-

aOR

††
†

p

<0.001

2.8[2.0,4.0]

0.001

0.105

1.3[0.8,2.1]

0.248

0.8[0.6,1.2]

0.308

1.1[0.7,1.7]

0.567

1.0 (Ref)

-

1.0 (Ref)

-

aOR

§

Smokeless tobacco estimates
Past year suicidal ideation
Past year suicide plan**
†
uOR*
p
aOR
p
uOR*
p
aOR†
¶

Current
Former (quit
1-3 years)
Former (quit
>3 years)
Never

p

Past year suicide attempt††
uOR*
p
aOR†

p

1.6[1.4,1.8]

<0.001

1.2[1.0,1.4]

0.071

1.6[1.2,2.1]

0.001

1.2[0.8,1.6]

0.357

1.9[1.4,2.6]

<0.001

1.3[0.9,1.9]

0.153

1.5[1.2,1.9]

<0.001

0.9[0.7,1.2]

0.470

1.4[0.9,2.2]

0.151

0.8[0.5,1.3]

0.367

2.3[1.4,3.7]

0.001

1.2[0.7,2.0]

0.423

1.3[1.2,1.5]

<0.001

1.2[1.1,1.5]

<0.001

1.2[1.0,1.5]

0.109

1.2[0.9,1.6]

0.162

1.2[0.8,1.6]

0.365

1.4[0.9,1.9]

0.101

1.0 (Ref)

-

1.0 (Ref)

-

1.0 (Ref)

-

1.0 (Ref)

-

1.0 (Ref)

-

1.0 (Ref)

-

Past year suicidal ideation
uOR*
p
aOR†
2.0[1.6,2.5]

Current
1.3[1.0,1.7]
Former (quit
1-3 years)
1.2[1.0,1.3]
Former (quit
>3 years)
1.0
Never
(Cont’d on next page)

¶

Chewing tobacco estimates
Past year suicide plan**
p
uOR*
p
aOR†

p

Past year suicide attempt††
uOR*
p
aOR†

p

<0.001

1.6[1.2,2.1]

0.003

2.1[1.4,3.1]

0.001

1.4[0.8,2.2]

0.200

2.6[1.8,4.0]

<0.001

1.6[0.9,2.8]

0.089

0.041

1.0[0.7,1.4]

0.908

1.1[0.7,1.8]

0.578

0.8[0.5,1.2]

0.260

1.3[0.8,2.2]

0.313

0.7[0.4,1.5]

0.354

0.039

1.2[1.0,1.4]

0.137

1.0[0.8,1.3]

0.831

0.9[0.6,1.3]

0.574

1.2[0.8,1.7]

0.447

1.3[0.7,2.4]

0.475

-

1.0 (Ref)

-

1.0 (Ref)

-

1.0 (Ref)

-

1.0 (Ref)

-

1.0 (Ref)

-

85

Table 4.4, (cont’d)

Past year suicidal ideation
uOR*
p
aOR†
1.5[1.3,1.8]

<0.001

1.1[0.9,1.4]

¶

p
0.399

Snuff estimates
Past year suicide plan**
uOR*
p
aOR†
1.7[1.2,2.2]

0.001

1.3[0.9,1.9]

p
0.150

Past year suicide attempt††
uOR*
p
aOR†
2.0[1.5,2.9]

<0.001

1.5[1.0,2.3]

p
0.065

Current
1.6[1.2,2.1]
0.003 1.1[0.8,1.6]
0.547 1.4[0.8,2.4]
0.198 1.1[0.6,2.1]
0.779
2.5[1.4,4.4]
0.003
1.8[0.9,3.4]
0.092
Former (quit
1-3 years)
1.4[1.2,1.6]
<0.001 1.4[1.2,1.7]
<0.001
1.3[1.0,
0.030 1.5[1.1,2.2]
0.018
1.4[0.9,2.2]
0.091
1.6[0.9,3.0]
0.106
Former (quit
1.7]
>3 years)
1.0 (Ref)
1.0 (Ref)
1.0 (Ref)
1.0 (Ref)
1.0 (Ref)
1.0 (Ref)
Never
†
* Unadjusted odds ratio with 95% confidence interval; Adjusted odds ratio with 95% confidence interval. Adjustments for sex,
race (non-Hispanic White vs. other), age(18-25, 26-35 and 35+), income(<$20k, $20k-$49k, $50k-$74k and ≥$75k) and population
density(MSA>1 million, MSA<1 million, non-MSA). Cigarette and smokeless tobacco estimates are from the same model; §
Smokeless tobacco users had either used chewing tobacco, snuff or both; ¶ Asked to all adults, “During the past 12 months, did
you seriously think about trying to kill yourself?”; ** Asked to all adults with ideation, “During the past 12 months, did you make
any plans to kill yourself?”; ††Asked to all adults with planning, “During the past 12 months, did you try to kill yourself?”;

86

Table 4.5. Logistic regressions predicting past year suicidal related outcomes by recency of tobacco use stratified by history of
major depression†. Data from the US National Survey on Drug Use and Health (NSDUH), 2008-2010 (n=114,048).
Cigarette smoking estimates
Past year suicidal ideation
With depression

Current
Former (quit
1-3 years)
Former (quit
>3 years)
Never

¶

Past year suicide plan**

Without depression

With depression

Past year suicide attempt

Without depression

With depression

††

Without depression

OR*

p

OR*

p

OR*

p

OR*

p

OR*

p

OR*

p

1.7[1.4,2.1]

<0.001

2.1[1.8,2.5]

<0.001

1.5[1.1,2.2]

0.016

2.4[1.8,3.3]

<0.001

1.7[1.1,2.6]

0.022

3.0[1.9,4.8]

0.001

1.0[0.7,1.5]

0.852

1.9[1.3,2.6]

0.001

1.2[0.7,2.1]

0.563

1.9[1.2,3.2]

0.010

0.8[0.4,1.7]

0.490

1.9[1.1,3.3]

0.022

0.7[0.6,0.9]

0.008

1.1[1.0,1.3]

0.159

0.6[0.4,1.0]

0.029

1.4[1.0,2.0]

0.079

0.4[0.2,0.7]

0.004

1.2[0.7,2.1]

0.480

1.0

-

1.0 (Ref)

-

1.0 (Ref)

-

1.0 (Ref)

-

1.0 (Ref)

-

1.0 (Ref)

-

§

Smokeless tobacco estimates
Past year suicidal ideation
Past year suicide plan**
With depression
Without
With
Without
depression
depression
depression
OR*
p
OR*
p
OR*
p
OR*
p
¶

Current
Former (quit
1-3 years)
Former (quit
>3 years)
Never

Past year suicide attempt††
With depression
Without
depression
OR*
p
OR*
p

1.6[1.2,2.1]

<0.001

1.7[1.4,2.1]

<0.001

1.6[1.1,2.4]

0.013

1.9[1.2,3.0]

0.005

1.7[1.1,2.8]

0.018

2.5[1.5,4.2]

0.001

1.6[1.1,2.3]

0.020

1.4[1.1,2.0]

0.028

1.2[0.7,2.1]

0.479

1.5[0.8,2.7]

0.162

1.8[0.9,3.8]

0.105

3.1[1.5,6.5]

0.004

1.1[0.9,1.3]

0.487

1.4[1.2,1.7]

<0.001

1.1[0.8,1.4]

0.723

1.3[0.8,2.2]

0.250

1.0[0.7,1.4]

0.829

1.4[0.8,2.7]

0.249

1.0 (Ref)

-

1.0 (Ref)

-

1.0 (Ref)

-

1.0 (Ref)

-

1.0 (Ref)

-

1.0 (Ref)

-

(Cont’d on next page)

87

Table 4.5, (cont’d)
Chewing tobacco estimates
Past year suicidal ideation
Past year suicide plan**
With depression
Without
With
Without
depression
depression
depression
OR*
p
OR*
p
OR*
p
OR*
p
¶

Current
Former (quit
1-3 years)
Former (quit
>3 years)
Never

Past year suicide attempt††
With depression
Without
depression
OR*
p
OR*
p

2.1[1.5,2.9]

<0.001

2.6[2.0,3.5]

<0.001

2.0[1.1,3.6]

0.030

3.1[1.8,5.3]

<0.001

1.8[1.1,3.0]

0.019

4.4[2.4,8.1]

0.001

1.2[0.8,1.7]

0.462

1.5[1.0,2.3]

0.049

1.0[0.6,1.6]

0.935

1.2[0.6,2.4]

0.614

0.9[0.4,1.9]

0.804

2.0[1.0,4.1]

0.058

1.0[0.8,1.2]

0.804

1.3[1.1,1.7]

0.008

0.8[0.6,1.2]

0.296

1.2[0.7,1.9]

0.544

1.1[0.7,1.8]

0.534

1.2[0.6,2.4]

0.681

1.0

-

1.0 (Ref)

-

1.0 (Ref)

-

1.0 (Ref)

-

1.0 (Ref)

-

1.0 (Ref)

-

Past year suicidal ideation¶
With depression
Without
depression
OR*
p
OR*
p

Snuff estimates
Past year suicide plan**
With
Without
depression
depression
OR*
p
OR*
p

Past year suicide attempt††
With depression
Without
depression
OR*
p
OR*
p

1.6[1.2,2.1]
0.001 1.6[1.3,2.0]
<0.001 1.7[1.1,2.5]
0.021
1.9[1.2,2.9]
0.005
1.8[1.1,3.0]
0.018
2.7[1.6,4.5]
0.001
Current
1.6[1.1,2.5]
0.026
1.4[0.9,2.1]
0.123
1.1[0.6,2.2]
0.669
1.6[0.8,2.9]
0.162
2.0[0.9,4.5]
0.110
3.2[1.4,7.5]
0.007
Former (quit
1-3 years)
1.1[0.9,1.4]
0.256 1.5[1.2,1.9]
0.001 1.1[0.8,1.4]
0.556
1.5[0.8,2.7]
0.170
1.1[0.8,1.6]
0.464
1.7[0.7,4.3]
0.216
Former (quit
>3 years)
1.0 (Ref)
1.0 (Ref)
1.0 (Ref)
1.0 (Ref)
1.0 (Ref)
1.0 (Ref)
Never
†
§
* Unadjusted odds ratio with 95% confidence interval Depression was assessed by the NSDUH’s structured interview module;
Smokeless tobacco users had either used chewing tobacco, snuff or both; ¶ Asked to all adults “During the past 12 months, did you
seriously think about trying to kill yourself?”; **Asked to all adults with ideation, “During the past 12 months, did you make any
plans to kill yourself?”; ††Asked to all adults with planning, “During the past 12 months, did you try to kill yourself?”;

88

Table 4.6. Logistic regressions predicting past year suicidal related outcomes by recency of tobacco use stratified by history of
major depression†. Data from the US National Survey on Drug Use and Health (NSDUH), 2008-2010 (n=114,048).
Cigarette smoking estimates
¶
Past year suicidal ideation
Past year suicide plan**
Past year suicide attempt††
With depression
Without
With
Without
With depression
Without
depression
depression
depression
depression
OR*
p
OR*
p
OR*
p
OR*
p
OR*
p
OR*
p
1.7[1.4,2.1]
<0.001
2.4[1.9,2.9]
<0.001
1.4[1.0,2.1]
0.076
3.7[2.4,5.8]
<0.001
1.8[1.1,3.0]
0.017
2.6[1.6,4.3]
0.001
Current
1.3[0.9,1.8]
0.200 1.7[1.2,2.5]
0.008 1.0[0.5,1.8]
0.958
3.1[1.6,6.0]
0.001
0.8[0.4,1.9]
0.652
1.3[0.4,4.0]
0.701
Former (quit
1-3 years)
0.9[0.7,1.1]
0.295 0.9[0.8,1.2]
0.580 0.7[0.4,1.0]
0.067
1.4[0.9,2.3]
0.105
0.5[0.3,1.0]
0.044
0.8[0.4,1.6]
0.541
Former (quit
>3 years)
1.0
1.0 (Ref)
1.0 (Ref)
1.0 (Ref)
1.0 (Ref)
1.0 (Ref)
Never
§
Smokeless tobacco estimates
¶
Past year suicidal ideation
Past year suicide plan**
Past year suicide attempt††
With depression
Without
With
Without
With depression
Without
depression
depression
depression
depression
OR*
p
OR*
p
OR*
p
OR*
p
OR*
p
OR*
p
1.7[1.3,2.3]
0.001
1.7[1.3,2.3]
<0.001
1.4[0.9,2.4]
0.150
2.5[1.6,4.1]
<0.001
2.2[1.3,3.8]
0.005
2.1[1.0,4.4]
0.057
Current
1.4[0.9,2.2]
0.104 2.0[1.4,2.8]
0.001 1.1[0.7,1.9]
0.694
2.3[0.8,6.7]
0.141
1.8[0.7,4.4]
0.213
4.0[1.7,9.5]
0.002
Former (quit
1-3 years)
1.2[1.0,1.4]
0.044 1.2[1.0,1.5]
0.094 1.1[0.8,1.5]
0.436
1.4[0.153]
0.153
1.1[0.7,1.7]
0.793
1.5[0.7,3.0]
0.306
Former (quit
>3 years)
1.0 (Ref)
1.0 (Ref)
1.0 (Ref)
1.0 (Ref)
1.0 (Ref)
1.0 (Ref)
Never
(Cont’d on next page)

89

Table 4.6, (cont’d)
Chewing tobacco estimates
Past year suicidal ideation
Past year suicide plan**
With depression
Without
With
Without
depression
depression
depression
OR*
p
OR*
p
OR*
p
OR*
p
¶

Current
Former (quit
1-3 years)
Former (quit
>3 years)
Never

Past year suicide attempt††
With depression Without depression
OR*

p

2.3[1.4,3.8]

0.002

2.8[1.9,4.1]

<0.001

1.5[0.7,3.0]

0.257

4.6[2.4,8.5]

<0.001

2.8[1.2,6.5]

0.518

1.1[0.7,1.9]

0.644

1.6[0.9,2.8]

0.083

0.9[0.5,1.6]

0.629

1.0[0.4,2.2]

0.993

1.3[0.6,2.6]

0.506

2.6[1.1,6.4]

0.032

1.2[1.0,1.5]

0.067

1.1[0.8,1.4]

0.552

1.0[0.7,1.4]

0.945

0.9[0.5,1.7]

0.763

1.2[0.7,2.0]

0.020

1.0[0.4,2.5]

0.954

1.0

-

1.0 (Ref)

-

1.0 (Ref)

-

1.0 (Ref)

-

1.0 (Ref)

-

1.0 (Ref)

-

¶

Past year suicidal ideation
With depression
Without
depression
OR*
p
OR*
p
1.7[1.3,2.3]

0.001

1.7[1.3,2.3]

0.001

Snuff estimates
Past year suicide plan**
With
Without
depression
depression
OR*
p
OR*
p
1.6[1.0,2.8]

0.061

2.3[1.4,3.7]

0.001

OR*
4.2[1.8,9.8]

p
0.001

Past year suicide attempt††
With depression Without depression
OR*

p

OR*

p

2.3[1.3,4.1]

0.006

2.3[1.1,5.1]

0.034

Current
1.5[1.0,2.4]
0.071 2.0[1.1,3.6]
0.026 1.3[0.7,2.4]
0.407
2.5[0.7,8.6] 0.134
2.0[0.7,5.9]
0.205
4.0[1.4,10]
0.010
Former (quit
1-3 years)
1.3[1.0,1.5]
0.024 1.5[1.1,1.9]
0.005 1.1[0.8,1.5]
0.398
1.8[1.0,3.3] 0.054
1.2[0.8,2.0]
0.331
1.8[0.6,5.4]
0.278
Former (quit
>3 years)
1.0 (Ref)
1.0 (Ref)
1.0 (Ref)
1.0 (Ref)
1.0 (Ref)
1.0 (Ref)
Never
†
* Unadjusted odds ratio with 95% confidence interval; Depression was assessed by asking participants “has a doctor, nurse or other
health professional ever told you that you have major depression?”; § Smokeless tobacco users had either used chewing tobacco,
snuff or both; ¶ Asked to all adults, “During the past 12 months, did you seriously think about trying to kill yourself?”;**Asked to all
adults with ideation, “During the past 12 months, did you make any plans to kill yourself?”; ††Asked to all adults with planning,
“During the past 12 months, did you try to kill yourself?”;

90

Table 4.7. Logistic regressions predicting past year suicidal related outcomes by recency of tobacco use stratified by past year
alcohol use disorders†. Data from the US National Survey on Drug Use and Health (NSDUH), 2008-2010 (n=114,048).
Cigarette smoking estimates
¶
Past year suicidal ideation
Past year suicide plan**
Past year suicide attempt††
With AUD
Without AUD
With AUD
Without AUD
With AUD
Without AUD
OR*
p
OR*
p
OR*
p
OR*
p
OR*
p
OR*
p
1.7[1.3,2.3]
0.001
2.3[2.0,2.6]
<0.001
1.4[0.8,2.5]
0.279
2.5[1.9,3.2]
<0.00
2.8[1.2,6.2]
0.015
2.5[1.7,3.5]
0.001
Current
Former (quit
1-3 years)
Former (quit
>3 years)
Never

1.1[0.7,1.8]

0.619

1.6[1.2,2.0]

0.001

1.2[0.5,3.0]

0.654

1.6[1.0,2.4]

1
0.041

1.4[0.5,4.2]

0.561

1.2[0.8,2.0]

0.402

0.9[0.6,1.3]

0.503

1.1[1.0,1,3]

0.135

0.6[0.3,1.4]

0.270

1.1[0.8,1.5]

0.476

1.0[0.4,2.9]

0.926

0.8[0.5,1.2]

0.207

1.0

-

1.0 (Ref)

-

1.0 (Ref)

-

1.0 (Ref)

-

1.0 (Ref)

-

1.0 (Ref)

-

§

Smokeless tobacco estimates
Past year suicidal ideation
Past year suicide plan**
With AUD
Without AUD
With AUD
Without AUD
OR*
p
OR*
p
OR*
p
OR*
p
¶

Current
Former (quit
1-3 years)
Former (quit
>3 years)
Never

Past year suicide attempt††
With AUD
Without AUD
OR*
p
OR*
p

1.1[0.9,1.4]

0.181

1.2[0.9,1.4]

0.147

1.2[0.8,1.8]

0.273

1.1[0.8,1.6]

0.578

1.3[0.8,2.1]

0.235

1.2[0.8,1.9]

0.358

1.0[0.6,1.5]

0.876

1.2[0.9,1.6]

0.251

1.0[0.5,2.1]

0.954

0.9[0.6,1.5]

0.774

1.0[0.5,2.2]

0.954

2.1[1.0,4.1]

0.037

0.9[0.7,1.1]

0.299

1.3[1.1,1.5]

<0.001

0.9[0.5,1.4]

0.570

1.2[0.8,1.6]

0.328

0.9[0.5,1.3]

0.492

1.1[0.7,1.7]

0.705

1.0 (Ref)

-

1.0 (Ref)

-

1.0 (Ref)

-

1.0 (Ref)

-

1.0 (Ref)

-

1.0 (Ref)

-

(Cont’d on next page)

91

Table 4.7, (cont’d)
Chewing tobacco estimates
Past year suicidal ideation
Past year suicide plan**
With AUD
Without AUD
With AUD
Without AUD
OR*
p
OR*
p
OR*
p
OR*
p
¶

Current
Former (quit
1-3 years)
Former (quit
>3 years)
Never

1.4[1.1,1.9
]
0.7[0.5,1.1
]

0.020

1.5[1.1,2.0]

0.016

0.137

1.2[0.8,1.7]

0.406

0.7[0.6,1.0
]

0.034

1.2[1.0,1.4]

1.0

-

1.0 (Ref)

1.8[1.0,3.3
]
0.8[0.4,1.4
]

0.059

1.2[0.7,1.9]

0.555

1.6[0.8,3.2]

0.160

1.9[1.0,3.7]

0.043

0.406

0.8[0.4,1.4]

0.419

0.7[0.4,1.5]

0.358

1.0[0.5,2.3]

0.937

0.065

0.9[0.5,1.5
]

0.666

0.9[0.6,1.2]

0.326

1.2[0.7,1.9]

0.468

0.9[0.5,1.6]

0.687

-

1.0 (Ref)

-

1.0 (Ref)

-

1.0 (Ref)

-

1.0 (Ref)

-

¶

Past year suicidal ideation
With AUD
Without AUD
OR*
p
OR*
p
Current

1.1[0.9,1.4]

0.231

1.1[0.9,1.4]

Past year suicide attempt††
With AUD
Without AUD
OR*
p
OR*
p

0.423

Snuff estimates
Past year suicide plan**
With AUD
Without AUD
OR*
p
OR*
p
1.2[0.8,1.9
]
1.0[0.4,2.4
]

0.363

1.2[0.8,1.8]

0.443

Past year suicide attempt††
With AUD
Without AUD
OR*
p
OR*
p
1.5[0.9,2.5]

0.162

1.3[0.8,2.1]

0.363

0.799
1.2[0.8,1.8]
0.376
0.935
0.9[0.5,1.7]
0.763
1.3[0.6,2.9]
0.536
2.0[0.9,4.8]
0.099
Former (quit 1.1[0.7,1.6]
1-3 years)
0.898
1.3[1.1,1.5]
0.001 1.0[0.6,1.6
0.967
1.2[0.9,1.8]
0.244
1.0[0.6,1.7]
0.883
1.3[0.7,2.4]
0.372
Former (quit 1.0[0.8,1.3]
]
>3 years)
1.0 (Ref)
1.0 (Ref)
1.0 (Ref)
1.0 (Ref)
1.0 (Ref)
1.0 (Ref)
Never
†
* Unadjusted odds ratio with 95% confidence interval; Alcohol use disorders were either DSM-IV alcohol abuse or alcohol
dependence. Both were assessed using a structured interview asked to all individuals who had drank alcohol in the year prior to
assessment; § Smokeless tobacco users had either used chewing tobacco, snuff or both; ¶ Asked to all adults, “During the past 12
months, did you seriously think about trying to kill yourself?”; ** Asked to all adults with ideation, “During the past 12 months, did
you make any plans to kill yourself?”; ††Asked to all adults with planning, “During the past 12 months, did you try to kill yourself?”;

92

Table 4.8 Mean number of days used tobacco in the past 30 days by past year suicide related outcome status. Data from the US
National Survey on Drug Use and Health (NSDUH), 2008-2010 (n=114,048).
Past year suicidal ideation¶
Past year suicide plan**
Past year suicide attempt††
Yes
No
Yes
No
Yes
No
Weighted
Weighted
Weighted
Weighted
Weighted
Weighted
mean(SE)
mean(SE)
mean(SE)
mean(SE)
mean(SE)
mean(SE)
Cigarettes
23.6(0.3)
22.9(0.1)
23.7(0.6)
22.9(0.1)
23.1(0.8)
22.9(0.1)
Chewing tobacco
12.8(1.9)
14.7(0.6)
15.8(2.6)
14.6(0.6)
13.0(2.5)
14.6(0.6)
Snuff
16.1(1.1)
20.8(0.3)
19.7(1.4)
20.6(0.3)
19.8(1.9)
20.6(0.3)
§
¶
*Standard errors provided in parenthesis Smokeless tobacco users had either used chewing tobacco, snuff or both; Asked to all
adults, “During the past 12 months, did you seriously think about trying to kill yourself?”; ** Asked to all adults with ideation,
“During the past 12 months, did you make any plans to kill yourself?”; ††Asked to all adults with planning, “During the past 12
months, did you try to kill yourself?”;

93

Table 4.9 Significance tests for a possible dose response between past year suicidal related outcomes and number of days
used tobacco in the past 30 days. Data from the US National Survey on Drug Use and Health (NSDUH), 2008-2010*
Past year suicidal ideation†
Past year suicide plan§
Past year suicide attempt¶
Cigarettes Chew
Snuff
Cigarettes
Chew
Snuff
Cigarettes Chew
Snuff
Crude
0.090
0.352
<0.001 (-) 0.204
0.603
0.541
0.752
0.575
0.686
Interview
depression*
*
Yes
0.303
0.669
0.113
0.075
0.096
0.760
0.549
0.073
0.426
No
0.755
0.490
0.001 (-)
0.009 (-)
0.346
0.256
0.050 (-)
0.923
0.946
Self-report
depression†
†

Yes
0.317
0.605
0.018 (-)
0.261
0.941
0.322
0.386
0.041 (-) 0.697
No
0.130
0.189
0.020 (-)
0.313
0.817
0.908
0.144
0.132
0.365
AUD§§
Yes
0.015 (+)
0.537
0.199
0.270
0.044 (+)
0.719
0.216
0.718
0.640
No
0.065
0.533
0.003 (-)
0.123
0.379
0.937
0.791
0.794
0.155
* p-values generated from a series of weighted logistic regressions; † Asked to all adults, “During the past 12 months, did you
seriously think about trying to kill yourself?”; § Asked to all adults with ideation, “During the past 12 months, did you make any
plans to kill yourself?”; ¶ Asked to all adults with planning, “During the past 12 months, did you try to kill yourself?”;**
Depression was assessed by the NSDUH’s structured interview module; †† Depression was assessed by asking participants “has
a doctor, nurse or other health professional ever told you that you have major depression?”; §§Alcohol use disorders were
either DSM-IV alcohol abuse or alcohol dependence;. Both were assessed using a structured interview asked to all individuals
who had smoked cannabis in the year prior to assessment;

94

Table 4.10 Frequency of suicide related outcomes among tobacco users who are either
exclusive cigarette smokers or exclusive smokeless tobacco users. Data from the US National
Survey on Drug Use and Health (NSDUH), 2008-2010
All persons
Ideation
Plan
Attempt
†
†
n
n
%
n
%
n
%†
Cigarette smokers/never SLT
users
Never
35,134 1,374
2.5
372
0.7
183
0.03
Current
29,243 2,245
5.9
754
1.8
459
0.09
Former (quit 1-3 years)
4,681
254
4.1
68
1.2
35
0.05
Former (quit >3 years)
20,731
840
2.8
222
0.7
85
0.02
SLT users/never smokers*
Never
35,134 1,374
2.5
372
0.7
183 0.003
Current
831
29
4.0
9
0.7
4 0.004
Former (quit 1-3 years)
248
8
1.2
4
0.9
2 0.007
Former (quit >3 years)
973
34
2.2
7
0.7
4 0.003
*Smokeless tobacco users had either used chewing tobacco, snuff or both; †Row percentages
p-values for unadjusted logistic regressions:
Cigarette Ideation:
Current=<0.001; Former (quit 1-3 years)=<0.001; Former (quit >3 years)=0.661
Cigarette Plan:
Current=<0.001; Former (quit 1-3 years)=0.001; Former (quit >3 years)=0.937
Cigarette Attempt:
Current=<0.001; Former (quit 1-3 years)=0.020; Former (quit >3 years)=0.087
Smokeless Ideation:
Current=0.807; Former (quit 1-3 years)=0.045; Former (quit >3 years)=0.374
Smokeless Plan:
Current=0.547; Former (quit 1-3 years)=0.561; Former (quit >3 years)=0.459
Smokeless Attempt:
Current=0.911; Former (quit 1-3 years)=N/A ; Former (quit >3 years)=0.823

95

Table 4.11 Frequency of suicide related outcomes among tobacco users who currently use
only one type of tobacco. Data from the US National Survey on Drug Use and Health
(NSDUH), 2008-2010
All persons
Ideation
Plan
Attempt
n
n
%
n
%
n
%
Cigarette smokers/non
current SLT users
Never
36,355 1,416
2.5
383
0.7
189
0.3
Current
37,310 2,885
6.2
946
1.8
570
0.9
Former (quit 1-3 years)
5,923
309
4.3
82
1.3
39
0.4
Former (quit >3 years)
26,016 1,027
2.8
274
0.7
98
0.2
SLT users/non current
smokers
Never
60,546 2,468
2.7
662
0.7
303
0.3
Current
2,684
85
2.6
25
0.6
14
0.4
Former (quit 1-3 years)
949
30
2.3
10
0.7
4
0.2
Former (quit >3 years)
6,799
254
3.1
67
0.8
19
0.3
†
*Smokeless tobacco users had either used chewing tobacco, snuff or both; Row percentages
p-values for unadjusted logistic regressions:
Cigarette Ideation:
Current=<0.001; Former (quit 1-3 years)=<0.001; Former (quit >3 years)=0.039
Cigarette Plan:
Current=<0.001; Former (quit 1-3 years)=0.001; Former (quit >3 years)=0.493
Cigarette Attempt:
Current=<0.001; Former (quit 1-3 years)=0.217; Former (quit >3 years)=0.313
Smokeless Ideation:
Current=0.796; Former (quit 1-3 years)=0.334; Former (quit >3 years)=0.195
Smokeless Plan:
Current=0.420; Former (quit 1-3 years)=0.978; Former (quit >3 years)=0.487
Smokeless Attempt:
Current=0.124; Former (quit 1-3 years)=0.494; Former (quit >3 years)=0.709

96

Table 5.1 Descriptive characteristics of current, formers and never smokers. Data from the
Collaborative Psychiatric Epidemiology Surveys (CPES), 2001-2003 (n=10,341).
All
Current smokers
Former smokers
Never smokers
persons
n
%
n
%
n
%
n
%
All persons
10,341
100
2,425
24.5
2,589
27.3
5,327
48.2
Sex
Male
4,507
47.3
1,276
27.6
1,354
30.6
1,877
41.7
Female
5,834
52.7
1,149
21.6
1,235
24.4
3,450
54.0
Age
18-25
1,696
16.5
451
29.1
307
17.3
938
53.6
26-35
2,335
17.6
595
28.2
420
18.0
1,320
53.8
36-45
2,336
20.6
626
28.7
525
21.2
1,185
50.1
46-64
2,870
28.8
632
23.0
888
33.7
1,350
43.3
65+
1,104
16.6
121
8.9
449
26.5
534
43.0
Race
White
4,180
70.7
1,200
25.4
1,219
29.5
1,761
45.2
Asian
2,178
4.3
292
13.5
411
17.8
1,475
68.7
Hispanic
3,119
12.0
632
20.6
767
25.0
1,720
54.4
Black
679
10.9
207
23.6
151
19.8
321
56.6
Other
185
2.1
94
43.8
41
27.2
50
29.0
Study*
NCS-R
5,692
86.8
1,638
25.5
1,569
28.0
2,485
46.5
NLAAS
4,649
13.2
787
17.9
1,020
23.0
2,842
59.1
Income
$0-$19,999
2,563
23.0
711
31.1
571
25.3
1,281
43.6
$20,0003,090
30.0
797
26.1
784
28.3
1,509
45.6
$49,999
$50,0001,735
18.7
398
23.7
453
27.6
884
48.7
$74,999
$75,000 +
2,953
28.4
519
17.9
781
27.7
1,653
54.4
Highest level
of education
<High school
2,159
18.0
619
33.6
522
28.6
1,018
37.7
Less than high
2,717
31.2
835
30.7
662
26.3
1,220
43.0
school
High
2,805
27.6
644
22.3
757
29.1
1,404
48.6
school/GED
Some college
2,660
23.3
327
11.6
648
25.6
1,684
62.8
* NCS-R=National comorbidity survey replication, NLAAS=National Latino and Asian American
Study.

97

Table 5.2 Lifetime diagnosis of mental illness by smoking status. Data from the Collaborative
Psychiatric Epidemiology Surveys (CPES), 2001-2003 (n=10,341).
All
Current smokers
Former smokers
Never smokers
persons
n
n
%
n
%
n
%
Suicidal
ideation
Yes
1,475
529
38.8
387
24.8
559
36.4
No
8,866
1,896
22.4
2,202
27.7
4,768
49.9
Suicidal
planning
Yes
542
235
45.2
124
20.2
183
34.6
No
9,799
2,190
23.5
2,465
27.7
5,144
48.8
Suicidal
attempt
Yes
523
242
50.1
126
22.3
155
27.7
No
9,818
2,183
23.4
2,463
27.5
5,172
49.1
Bipolar I
Yes
104
48
46.1
20
17.7
36
36.3
No
10,237
2,377
24.3
2,569
27.4
5,291
48.3
Bipolar II
Yes
101
58
61.4
20
19.6
23
19.0
No
10,240
2,367
24.2
2,569
27.4
5,291
48.5
Major
Depression
Yes
2,137
630
30.5
558
27.2
949
42.3
No
8,204
1,795
23.3
2,031
27.4
4,378
49.4
Dysthymia
Yes
542
203
38.5
148
29.1
191
32.4
No
9,799
2,222
23.8
2,441
27.3
5,136
48.9
Alcohol use
disorder
Yes
1,326
677
52.1
438
31.9
211
16.0
No
9,015
1,748
21.5
2,151
26.7
5,116
52.9
Substance use
disorder
Yes
828
446
56.0
276
31.3
106
12.7
No
9,513
1,979
21.9
2,313
27.0
5,221
51.1

(Cont’d on the next page)

98

Table 5.2, (cont’d)

All
persons
n

Current smokers
n

Conduct
disorder
Yes
1,612
673
No
8,729
1,752
Post-traumatic
stress disorder
Yes
775
297
No
9,566
2,128
Generalized
anxiety
disorder
Yes
939
311
No
9,402
2,114
Panic
Yes
2,987
963
No
7,354
1,462
*Diagnosis was made using the WMH-CIDI

Former smokers

Never smokers

%

n

%

n

%

45.4
21.1

411
2,178

24.6
27.8

528
4,799

30.0
51.1

40.0
23.4

197
2,392

26.5
27.4

281
5,046

33.5
49.2

35.0
23.6

261
2,328

28.1
27.3

367
4,960

36.9
49.1

32.2
19.9

789
1,800

26.1
24.5

1,235
4,092

41.4
55.6

99

Table 5.3 Social support assessment items and response proportions. Data from the Collaborative Psychiatric
Epidemiology Surveys (CPES), 2001-2003 (n=10,341).
1 (%)
2 (%)
3 (%)
4 (%)
5 (%)
1: How often do you talk on the phone or get together with
14.6
10.4
24.3
30.2
20.6
relatives who do not live with you?
2: How much can you open up to relatives who do not live with
8.5
10.0
18.6
62.9
--you if you need to talk about your worries?
3: How much can you open up to relatives who do not live with
11.1
14.7
26.6
47.6
--you if you need to talk about your serious problems?
4: How often do your relatives make too many demands on
7.7
17.3
34.1
40.9
--you?
5: How often do your relatives argue with you
4.1
13.1
37.0
45.8
--6: How often do you talk on the phone or get together with
14.1
8.7
23.2
32.2
21.8
friends?
7: How much can you rely on your friends for help if you have a
10.1
15.3
28.2
46.5
--serious problem?
8: How much can you open up to your friends if you need to
10.4
14.6
30.5
44.5
--talk about your worries?
9: How often do your friends make too many demands on you?
2.0
11.0
36.8
50.2
--10: How often do your friends argue with you?
1.1
8.6
35.5
54.8
11: When you have a problem or worry, how often do you let
2.9
7.3
20.1
37.9
31.8
your partner worry about it?
12: When you have a problem or worry, how often do you let
9.8
25.9
35.9
19.6
8.8
someone (else) know about it?
Responses were coded as follows:
Items 1, and 6: 1=Less than once a month 2=Once a month 3=A few times a month 4=A few times a week 5=Most
every day; Items 2, 3, 7 and 8: 1=Not at all 2=Little 3=Some 4=A lot; Items 4, 5, 9 and 10 : 1=Often 2=Sometimes
3=Rarely 4=Never; Items 11 and 12: 1=Never 2=Rarely 3=Sometimes 4=Most of the time 5=Always

100

Table 5.4 Social support assessment items and response proportions: exploratory factor analysis sample. Data
from the Collaborative Psychiatric Epidemiology Surveys (CPES), 2001-2003 (n=4,649).
1 (%)
2 (%)
3 (%)
4 (%)
5 (%)
1:How often do you talk on the phone or get together with
14.9
10.0
24.2
30.2
20.9
relatives who do not live with you?
2: How much can you open up to relatives who do not live with
8.0
9.2
17.9
64.9
--you if you need to talk about your worries?
3: How much can you open up to relatives who do not live with
10.7
14.2
26.8
48.4
--you if you need to talk about your serious problems?
4: How often do your relatives make too many demands on
7.4
16.3
35.1
41.2
--you?
5: How often do your relatives argue with you
4.1
12.4
37.2
46.2
--6: How often do you talk on the phone or get together with
13.1
7.8
23.0
32.9
23.2
friends?
7: How much can you rely on your friends for help if you have a
9.0
13.7
27.8
49.6
--serious problem?
8: How much can you open up to your friends if you need to
9.1
13.1
31.0
46.8
--talk about your worries?
9: How often do your friends make too many demands on you?
1.9
10.8
37.3
50.0
--10: How often do your friends argue with you?
1.1
8.3
36.0
54.6
--11: When you have a problem or worry, how often do you let
2.9
7.6
20.6
39.7
29.3
your partner worry about it?
12: When you have a problem or worry, how often do you let
9.0
26.1
36.1
20.3
8.5
someone (else) know about it?
Responses were coded as follows:
Items 1, and 6: 1=Less than once a month 2=Once a month 3=A few times a month 4=A few times a week 5=Most
every day; Items 2, 3, 7 and 8: 1=Not at all 2=Little 3=Some 4=A lot; Items 4, 5, 9 and 10 : 1=Often 2=Sometimes
3=Rarely 4=Never; Items 11 and 12: 1=Never 2=Rarely 3=Sometimes 4=Most of the time 5=Always

101

Table 5.5 Social support assessment items and response proportions: confirmatory factor analysis sample. Data
from the Collaborative Psychiatric Epidemiology Surveys (CPES), 2001-2003 (n=5,692).
1 (%)
2 (%)
3 (%)
4 (%)
5 (%)
1:How often do you talk on the phone or get together with
13.1
12.9
25.1
30.3
18.5
relatives who do not live with you?
2: How much can you open up to relatives who do not live with
11.7
14.8
22.8
50.7
--you if you need to talk about your worries?
3: How much can you open up to relatives who do not live with
13.7
17.8
25.3
43.2
--you if you need to talk about your serious problems?
4: How often do your relatives make too many demands on
9.0
23.6
28.4
39.0
--you?
5: How often do your relatives argue with you
4.2
17.2
35.3
43.3
--6: How often do you talk on the phone or get together with
19.8
14.0
24.2
28.3
13.7
friends?
7: How much can you rely on your friends for help if you have a
16.8
24.7
30.4
28.1
--serious problem?
8: How much can you open up to your friends if you need to
18.0
23.7
27.8
30.6
--talk about your worries?
9: How often do your friends make too many demands on you?
2.5
12.7
33.4
51.5
--10: How often do your friends argue with you?
1.2
10.5
32.3
56.0
--11: When you have a problem or worry, how often do you let
3.3
5.7
17.4
45.2
28.4
your partner worry about it?
12: When you have a problem or worry, how often do you let
15.4
24.7
34.6
15.0
10.2
someone (else) know about it?
Responses were coded as follows:
Items 1, and 6: 1=Less than once a month 2=Once a month 3=A few times a month 4=A few times a week 5=Most
every day; Items 2, 3, 7 and 8: 1=Not at all 2=Little 3=Some 4=A lot; Items 4, 5, 9 and 10 : 1=Often 2=Sometimes
3=Rarely 4=Never; Items 11 and 12: 1=Never 2=Rarely 3=Sometimes 4=Most of the time 5=Always

102

Table 5.6 Estimated associations between cigarette smoking and suicide related
outcomes: unadjusted and adjusted for descriptive characteristics. Data from the
Collaborative Psychiatric Epidemiology Surveys (CPES), 2001-2003 (n=10,341).
Ideation
Plans
Attempts
Unadj.
Multivar.
Unadj.
Multivar.
Unadj.
Multivar.
Smoking
status
Current
Former
Never

2.4[2.0,2.8] 2.2[1.8,2.6] 2.7[2.1,3.5] 2.5[1.9,3.3] 3.8[2.9,4.9] 3.2[2.4,4.2]
1.2[1.0,1.5] 1.3[1.2,1.7] 1.0[0.8,1.4] 1.2[0.9,1.6] 1.4[1.1,1.9] 1.7[1.2,2.3]
Ref
Ref
Ref
Ref
Ref
Ref

Sex
Male
Female

Ref
Ref
Ref
Ref
Ref
Ref
1.4[1.2,1.7] 1.5[1.3,1.8] 1.4[1.1,1.8] 1.5[1.1,1.9] 1.9[1.5,2.4] 2.1[1.6,2.7]

Age
18-25
26-35
36-45
46-64
65+

Ref
0.8[0.7,1.1]
0.9[0.7,1.1]
0.8[0.6,1.0]
0.3[0.2,0.4]

Ref
0.9[0.7,1.1]
0.9[0.7,1.2]
0.8[0.6,1.0]
0.3[0.2,0.4]

Ref
0.9[0.7,1.3]
0.9[0.7,1.3]
0.7[0.5,1.0]
0.2[0.1,0.4]

Ref
0.9[0.6,1.3]
1.0[0.7,1.4]
0.8[0.5,1.1]
0.2[0.1,0.4]

Ref
0.8[0.6,1.2]
0.8[0.6,1.2]
0.6[0.4,0.9]
0.1[0.1,0.2]

Ref
0.9[0.6,1.3]
0.9[0.6,1.3]
0.7[0.5,1.0]
0.1[0.1,0.2]

Ref
0.4[0.4,0.6]
0.6[0.5,0.7]
0.8[0.6,1.0]
1.4[0.9,2.2]

Ref
0.4[0.4,0.6]
0.5[0.4,0.6]
0.7[0.5,0.9]
1.0[0.7,1.6]

Ref
0.5[0.3,0.7]
0.6[0.4,0.7]
0.6[0.4,0.9]
2.2[1.3,3.8]

Ref
0.5[0.3,0.7]
0.5[0.4,0.7]
0.5[0.3,0.8]
1.6[0.9,2.8]

Ref
0.4[0.3,0.6]
0.9[0.7,1.2]
0.8[0.6,1.2]
2.4[1.4,4.2]

Ref
0.5[0.3,0.7]
0.7[0.5,0.9]
0.6[0.4,0.9]
1.5[0.8,2.6]

Race
White
Asian
Hispanic
Black
Other
Income
$0-$19,999
$20,000$49,999
$50,000$74,999
$75,000 +
Highest
level of
education
Less than
high school
High
school/GED
Some
college
College or
more

Ref
Ref
Ref
Ref
Ref
Ref
0.8[0.7,1.0] 0.8[0.7,1.0] 1.1[0.8,1.5] 1.0[0.7,1.4] 0.9[0.7,1.1] 0.9[0.7,1.2]
0.7[0.6,0.9] 0.6[0.5,0.8] 0.7[0.5,1.1] 0.6[0.4,9.4] 0.6[0.4,0.8] 0.6[0.4,0.9]
0.6[0.5,0.8] 0.6[0.4,0.7] 0.6[0.5,0.9] 0.5[0.4,0.8] 0.4[0.3,0.5] 0.5[0.3,0.7]

Ref

Ref

Ref

Ref

Ref

Ref

0.8[0.6,1.0] 0.7[0.5,0.9] 0.9[0.6,1.3] 0.8[0.6,1.2] 0.7[0.5,0.9] 0.6[0.4,0.9]
0.9[0.7,1.2] 0.9[0.7,1.1] 1.0[0.7,1.4] 1.0[0.7,1.5] 0.7[0.5,1.0] 0.8[0.5,1.1]
0.9[0.7,1.1] 1.1[0.8,1.4] 1.1[0.8,1.5] 1.4[0.9,2.1] 0.5[0.3,0.6] 0.6[0.4,1.0]

*Logistic regressions based on weight data.
103

Table 5.7 Estimated associations between cigarette smoking and suicide related
outcomes: unadjusted and adjusted for mental illness (bi-variable). Data from the
Collaborative Psychiatric Epidemiology Surveys (CPES), 2001-2003 (n=10,341).
Ideation
Plans
Attempts
Bipolar I Status
Current
2.3[1.9,2.8]
2.6[2.0,3.4]
2.3[1.9,2.8]
Former
1.2[1.0,1.5]
1.0[0.8,1.4]
1.2[1.0,1.5]
Never
Ref
Ref
Ref
Bipolar II Status
Current
2.3[1.9,2.7]
2.6[2.0,3.3]
2.3[1.9,2.7]
Former
1.2[1.0,1.5]
1.0[0.8,1.4]
1.2[1.0,1.5]
Never
Ref
Ref
Ref
Major Depression
Current
2.2[1.9,2.6]
2.4[1.9,3.1]
2.2[1.9,2.6]
Former
1.2[1.0,1.4]
1.0[0.7,1.3]
1.2[1.0,1.4]
Never
Ref
Ref
Ref
Dysthymia
Current
2.2[1.8,2.6]
2.4[1.9,3.2]
2.2[1.8,2.6]
Former
1.2[1.0,1.4]
1.0[0.7,1.3]
1.2[1.0,1.4]
Never
Ref
Ref
Ref
Alcohol use disorder
Current
1.8[1.5,2.2]
1.9[1.4,2.5]
1.8[1.5,2.2]
Former
1.1[0.9,1.3]
0.9[0.6,1.1]
1.1[0.9,1.3]
Never
Ref
Ref
Ref
Substance use disorder
Current
1.9[1.6,2.3]
1.9[1.4,2.5]
1.9[1.6,2.3]
Former
1.1[0.9,1.4]
0.9[0.6,1.2]
1.1[0.9,1.3]
Never
Ref
Ref
Ref
Conduct disorder
Current
2.0[1.7,2.4]
2.2[1.7,2.9]
2.0[1.7,2.4]
Former
1.2[1.0,1.4]
1.0[0.7,1.3]
1.2[1.0,1.4]
Never
Ref
Ref
Ref
Post-traumatic stress
disorder
Current
2.1[1.8,2.6]
2.3[1.8,3.0]
2.1[1.8,2.6]
Former
1.2[1.0,1.4]
1.0[0.7,1.3]
1.2[1.0,1.4]
Never
Ref
Ref
Ref

(Cont’d on next page)
104

Table 5.7, (cont’d)
Ideation

Plans

Attempts

Generalized anxiety disorder
Former
1.2[1.0,1.4]
1.0[0.7,1.3]
1.2[1.0,1.4]
Never
Ref
Ref
Ref
Panic disorder
Current
2.1[1.7,2.5]
2.3[1.8,2.9]
2.1[1.7,2.5]
Former
1.2[1.0,1.4]
1.0[0.7,1.3]
1.2[1.0,1.4]
Never
Ref
Ref
Ref
*The ORs in this table show the association between cigarette smoking and suicide related
outcomes after adjustment for the respective mental health diagnosis.

105

Table 5.8 Estimated associations between cigarette smoking and suicide related
outcomes: unadjusted and adjusted for mental illness (multi-variable). Data from
the Collaborative Psychiatric Epidemiology Surveys (CPES), 2001-2003 (n=10,341).
Ideation
Plans
Attempts
Unadj. Multivar.
Unadj.
Multivar.
Unadj.
Multivar.
Smoking status
Current

2.4[2.0,2.8]

1.4[1.2,1.8]

2.7[2.1,3.5]

1.4[1.0,1.8]

3.8[2.9,4.9]

2.9[1.5,2.6]

Former

1.2[1.0,1.5]

1.0[0.8,1.2]

1.0[0.8,1.4]

0.8[0.6,1.1]

1.4[1.1,1.9]

1.1[0.8,1.5]

Ref

Ref

Ref

Ref

Ref

Ref

Yes

7.5[4.8,11.8]

5.0[3.1,8.0]

5.8[3.4,10.0]

3.9[2.0,7.4]

7.5[4.8,11.8]

4.1[2.2,7.9]

No

Ref

Ref

Ref

Ref

Ref

Ref

Yes

6.4[4.1,10.0]

3.0[1.8,5.1]

7.4[4.5,12.3]

3.2[1.7,6.0]

6.4[4.1,10.0]

3.5[1.9,6.5]

No

Ref

Ref

Ref

Ref

Ref

Ref

4.5[3.9,5.2]
Ref

3.2[2.7,3.8]
Ref

5.5[4.3,6.8]
Ref

3.8[2.9,5.0]
Ref

4.5[3.9,5.2]
Ref

2.8[2.1,3.7]
Ref

Yes

6.1[4.8,7.6]

1.8[1.3,2.3]

6.4[4.8,8.6]

1.7[1.2,2.4]

6.1[4.8,7.6]

2.2[1.5,3.1]

No

Ref

Ref

Ref

Ref

Ref

Ref

3.2[2.7,3.8]

1.4[1.1,1.9]

3.9[3.1,5.0]

1.3[0.9,1.9]

3.2[2.7,3.8]

1.4[1.0,2.0]

Ref

Ref

Ref

Ref

Ref

Ref

4.0[3.3,4.9]

1.5[1.1,2.1]

5.6[4.3,7.2]

2.2[1.5,3.2]

4.0[3.2,4.9]

1.7[1.2,2.5]

Ref

Ref

Ref

Ref

Ref

Ref

2.7[2.2,3.2]

1.9[1.5,2.3]

3.1[2.3,3.9]

1.7[1.3,2.3]

2.7[2.3,3.3]

1.8[1.4,2.4]

Ref

Ref

Ref

Ref

Ref

Ref

4.7[3.9,5.8]

2.1[1.6,2.6]

6.8[5.3,8.7]

2.8[2.1,3.8]

4.7[3.9,5.8]

2.5[1.8,3.3]

Ref

Ref

Ref

Ref

Ref

Ref

Never
Bipolar I

Bipolar II

Major
Depression
Yes
No
Dysthymia

Alcohol use
disorder
Yes
No
Substance use
disorder
Yes
No
Conduct
disorder
Yes
No
Post-traumatic
stress disorder
Yes
No

(Cont’d on next page)

106

Table 5.8, (cont’d)
Ideation
Unadj. Multivar
.
Generalized
anxiety
disorder
Yes

Plans
Unadj.
Multivar.

Attempts
Unadj.
Multivar.

4.6[3.8,5.5]

1.9[1.5,2.4]

4.6[3.5,5.9]

1.5[1.1,2.1]

4.6[3.8,5.5]

2.2[1.6,3.1]

Ref

Ref

Ref

Ref

Ref

Ref

Yes

3.1[2.7,3.6]

1.6[1.4,2.0]

3.8[3.1,4.8]

1.7[1.4,2.2]

3.1[2.7,3.6]

1.8[1.3,2.4]

No

Ref

Ref

Ref

Ref

Ref

Ref

No
Panic disorder

*The ORs in the multivariable columns of this table show the association between
cigarette smoking and suicide related outcomes after adjustment for the all mental
health diagnoses in this table.

107

Table 5.9 Model fit indices for factor structure of 12 item social support scale. Data from the
Collaborative Psychiatric Epidemiology Surveys (CPES), 2001-2003 (n=4,649).
# of Factors

CFI*

TLI†

RMSEA§

Eigenvalues¶

1

0.749

0.694

0.188

3.733

2

0.893

0.836

0.137

2.032

3

0.972

0.943

0.081

1.270

4

0.982

0.952

0.074

0.979

*CFI=comparative fit index (>0.9=evidence of good fit); †TLI=Tucker-Lewis fit index
(>0.9=evidence of good fit); § RMSEA=root means square error of approximation
(<0.08=evidence of good fit); ¶ Eigenvalues (>1.0=evidence of good fit)

108

Table 5.10 Results from exploratory factor analysis of 12 social support items. Data from the Collaborative Psychiatric
Epidemiology Surveys (CPES), 2001-2003 (n=4,649).
1 Factor
Factor 1
Loading(SE)
1:How often do you talk on the phone or get
together with relatives who do not live with you?
2: How much can you open up to relatives who do
not live with you if you need to talk about your
worries?
3: How much can you open up to relatives who do
not live with you if you need to talk about your
worries?
4: How often do your relatives make too many
demands on you?
5: How often do your relatives argue with you
6: How often do you talk on the phone or get
together with friends?
7: How much can you rely on your friends for help
if you have a serious problem?
8: How much can you open up to your friends if
you need to talk about your worries?
9: How often do your friends make too many
demands on you?
10: How often do your friends argue with you?
11: When you have a problem or worry, how often
do you let your partner worry about it?
12: When you have a problem or worry, how often
do you let someone (else) know about it?

2 Factor
Factor 1
Factor 2
Factor 1
Loading(SE) Loading(SE) Loading(SE)

3 Factor
Factor 2
Loading(SE)

Factor 3
Loading(SE)

0.352(0.013)

0.420(0.013)

-0.051(0.016)

0.450(0.018)

0.028(0.018)

-0.028(0.017)

0.724(0.009)

0.821(0.010)

0.009(0.002)

0.841(0.016)

0.007(0.012)

-0.004(0.003)

0.711(0.009)

0.853(0.009)

0.042(0.015)

0.896(0.014)

-0.004(0.007)

0.044(0.016)

-0.370(0.013)

-0.003(0.007)

0.545(0.013)

-0.224(0.025)

0.043(0.013)

0.664(0.014)

-0.379(0.013)

0.013(0.013)

0.580(0.013)

-0.150(0.026)

-0.033(0.016)

0.648(0.015)

0.483(0.011)

0.335(0.016)

-0.386(0.015)

0.022(0.016)

0.517(0.016)

-0.082(0.017)

0.844(0.006)

0.662(0.015)

-0.598(0.015)

0.002(0.008)

0.896(0.012)

0.031(0.013)

0.857(0.006)

0.625(0.015)

-0.621(0.015)

0.003(0.010)

0.909(0.012)

0.003(0.003)

-0.456(0.013)

0.014(0.015)

0.679(0.012)

-0.005(0.006)

-0.237(0.025)

0.628(0.015)

-0.477(0.014)

0.025(0.018)

0.671(0.012)

0.029(0.019)

-0.256(0.023)

0.609(0.015)

0.193(0.019)

0.293(0.020)

0.040(0.021)

0.265(0.025)

0.050(0.025)

0.099(0.024)

0.447(0.012)

0.397(0.014)

-0.251(0.015)

0.251(0.018)

0.299(0.017)

-0.073(0.017)

(Cont’d on next page)

109

Table 5.10,( cont’d)

Factor 1
Loading( SE)
1:How often do you talk on the phone or get
together with relatives who do not live with you?
2: How much can you open up to relatives who do
not live with you if you need to talk about your
worries?
3: How much can you open up to relatives who do
not live with you if you need to talk about your
worries?
4: How often do your relatives make too many
demands on you?
5: How often do your relatives argue with you
6: How often do you talk on the phone or get
together with friends?
7: How much can you rely on your friends for help
if you have a serious problem?
8: How much can you open up to your friends if
you need to talk about your worries?
9: How often do your friends make too many
demands on you?
10: How often do your friends argue with you?
11: When you have a problem or worry, how often
do you let your partner worry about it?
12: When you have a problem or worry, how often
do you let someone (else) know about it?

4 Factor
Factor 2
Factor 3
Loading( SE) Loading( SE)

Factor 4
Loading(
SE)

0.053(0.188)

0.478(0.017)

-0.042(0.020)

-0.052(0.047)

-0.006(0.012)

0.841(0.014)

-0.017(0.009)

0.020(0.014)

-0.007(0.017)

0.879(0.014)

0.036(0.009)

0.032(0.016)

0.011(0.035)

-0.142(0.022)

0.674(0.015)

0.080(0.018)

0.005(0.015)

-0.078(0.021)

0.662(0.014)

0.017(0.009)

3.837(12.729)

0.000(0.001)

0.000(0.001)

0.000(0.004)

0.017(0.061)

0.080(0.017)

-0.015(0.011)

0.786(0.026)

-0.009(0.031)

-0.004(0.004)

0.011(0.009)

0.991(0.020)

-0.005(0.018)

0.046(0.010)

0.649(0.016)

-0.164(0.028)

-0.011(0.037)

0.072(0.015)

0.630(0.016)

-0.174(0.030)

-0.004(0.011)

0.273(0.024)

0.098(0.024)

0.068(0.025)

0.003(0.014)

0.254(0.018)

-0.084(0.017)

0.288(0.017)

110

Table 5.11 Results of a confirmatory factor analysis of a 3 factor solution for 12 social support items. Data from the
Collaborative Psychiatric Epidemiology Surveys (CPES), 2001-2003 (n=5,692).
Factor 1
Factor 2
Factor 3
Loading p-value
Loading
p-value
Loading
pvalue
1: How often do you talk on the phone or get together
with relatives who do not live with you?
2: How much can you open up to relatives who do not
live with you if you need to talk about your worries?
3: How much can you open up to relatives who do not
live with you if you need to talk about your serious
problems?
4: How often do your relatives make too many
demands on you?
5: How often do your relatives argue with you
6: How often do you talk on the phone or get together
with friends?
7: How much can you rely on your friends for help if
you have a serious problem?
8: How much can you open up to your friends if you
need to talk about your worries?
9: How often do your friends make too many demands
on you?
10: How often do your friends argue with you?
11: When you have a problem or worry, how often do
you let your partner worry about it?
12: When you have a problem or worry, how often do
you let someone (else) know about it?

0.522

<0.001

0.918

<0.001

1.000

<0.001

0.347

<0.001

0.179

<0.001

0.630

<0.001

0.952

<0.001

1.000

<0.001

0.377

0.955

<0.001

1.000

<0.001

0.917

<0.001

0.936

<0.001

<0.001

* Loadings for factor 1 are scaled against item 3 (which is fixed at 1). Similarly, loadings for factor 2 are scaled against item 8,
and loadings for factor 3 are scaled against item 5.

111

Table 5.12 Distributions of social support factor scores. Data
from the Collaborative Psychiatric Epidemiology Surveys
(CPES), 2001-2003 (n=10,341).
Factor 1
Mean
Standard
Min
Max
dev.
All persons
0.039
0.686
-1.948
1.318
Smoking status
Current
-0.038
0.705
-1.943
1.318
Former
0.059
0.683
-1.943
1.310
Never
0.068
0.678
-1.948
1.312
Factor 2
Mean
Standard
Min
Max
dev.
All persons
0.082
0.752
-1.944
1.686
Smoking status
Current
0.050
0.778
-1.944
1.598
Former
0.074
0.744
-1.944
1.686
Never
0.103
0.743
-1.944
1.651
Factor 3
Mean
Standard
Min
Max
dev.
All persons
0.021
0.521
-1.923
0.929
Smoking status
Current
-0.053
0.542
-1.893
0.929
Former
0.074
0.509
-1.948
1.312
Never
0.029
0.516
-1.923
0.929

112

Table 5.13 Estimated associations between cigarette smoking and suicide related
outcomes: unadjusted and adjusted for social support (multi-variable). Data from
the Collaborative Psychiatric Epidemiology Surveys (CPES), 2001-2003 (n=10,341).
Ideation
Plans
Attempts
Unadj. Multivar.
Unadj. Multivar.
Unadj. Multivar.
Smoking
status
2.4[2.0,2.8] 2.2[1.8,2.6]
2.7[2.1,3.5] 2.4[1.9,3.1] 3.8[2.9,4.9] 3.3[2.6,4.4]
Current
1.2[1.0,1.5] 1.3[1.1,1.5] 1.0[0.8,1.4] 1.1[0.8,1.4] 1.4[1.1,1.9] 1.5[1.1,2.0]
Former
Ref
Ref
Ref
Ref
Ref
Ref
Never
Social
support*
0.7[0.6,0.8] 0.7[0.6,0.8] 0.6[0.5,0.8] 0.7[0.5,0.8] 0.5[0.4,0.6] 0.6[0.5,0.7]
SS factor
1 score
1.0[0.9,1.1] 1.0[0.9,1.2] 1.0[0.8,1.2] 1.0[0.8,1.2] 1.0[0.8,1.2] 1.0[0.8,1.2]
SS factor
2 score
0.5[0.4,0.6] 0.5[0.5,0.6] 0.4[0.3,0.5] 0.5[0.4,0.6] 0.4[0.3,0.5] 0.5[0.4,0.6]
SS factor
3 score
*Unadjusted models for social support contain all three factor scores. ORs for
social support variables are the estimated OR associated with a one unit increase in
the factor score.

113

Table 5.14 Estimated associations between cigarette smoking and suicide related
outcomes: unadjusted and adjusted for social support, descriptive characteristics and
mental disorders. Data from the Collaborative Psychiatric Epidemiology Surveys
(CPES), 2001-2003 (n=10,341).
Ideation
Plans
Attempts
Unadjusted
Current
2.4[2.0,2.8]
2.7[2.1,3.5]
3.8[2.9,4.9]
Former
1.2[1.0,1.5]
1.0[0.8,1.4]
1.4[1.1,1.9]
Never
Ref
Ref
Ref
Model I (descriptive
characteristics)
Current
2.2[1.8,2.6]
2.5[1.9,3.3]
3.2[2.4,4.2]
Former
1.3[1.2,1.7]
1.2[0.9,1.6]
1.7[1.2,2.3]
Never
Ref
Ref
Ref
Model II (mental
disorders)
Current
1.4[1.2,1.8]
1.4[1.0,1.8]
2.0[1.5,2.7]
Former
1.0[0.8,1.2]
0.8[0.6,1.1]
1.1[0.8,1.5]
Never
Ref
Ref
Ref
Model III (social support)
Current
2.2[1.8,2.6]
2.4[1.9,3.1]
3.3[2.6,4.4]
Former
1.3[1.1,1.5]
1.1[0.8,1.4]
1.5[1.1,2.0]
Never
Ref
Ref
Ref
Model IV (II+III)
Current
1.4[1.2,1.7]
1.3[1.0,1.8]
2.0[1.5,2.6]
Former
1.0[0.9,1.2]
0.8[0.6,1.0]
1.1[0.8,1.5]
Never
Ref
Ref
Ref
Model V (I+II+III)
Current
1.4[1.1,1.7]
1.3[1.0,1.8]
1.8[1.3,2.4]
Former
1.1[0.9,1.3]
0.8[0.6,1.1]
1.2[0.9,1.7]
Never
Ref
Ref
Ref
Model VI (I+II+III)**
Current
1.3[1.1,1.6]
1.3[1.1,1.6]
1.6[1.2,2.1]
Former
1.0[0.8,1.2]
0.8[0.6,1.1]
1.2[0.9,1.7]
Never
Ref
Ref
Ref
*Model I=age, sex, race, income and education
Model II= bipolar I, bipolar II, major depression, dysthymia, alcohol use disorder,
substance use disorder, conduct disorder, post-traumatic stress disorder and
generalized anxiety disorder
Model III= social support scores (f1 f2 and f3)
**Number of disorders instead of separate terms for each disorder as in model II.
114

Table 5.15 Exploratory analysis of subgroup variation in the effects of cigarette
smoking on suicide related outcomes by mental disorder status. Data from the
Collaborative Psychiatric Epidemiology Surveys (CPES), 2001-2003 (n=10,341).
Ideation
Plans
Attempts
OR
p-value
OR
pOR p-value
value
Bipolar I
1.1[0.4,2.9]
0.875 0.4[0.1,1.2] 0.099 0.8[0.2,2.8]
0.756
Bipolar II
1.0[0.3,3.0]
0.974 1.2[0.3,5.2] 0.823 0.6[0.2,2.3]
0.447
Major
0.6[0.4,0.8]
0.004 0.5[0.3,1.1] 0.003 1.1[0.7,1.9]
0.639
depression
Dysthymia
0.6[0.3,1.0]
0.038 0.6[0.3,1.1] 0.103 0.8[0.4,1.6]
0.580
Alcohol use
0.8[0.5,1.4]
0.482 1.0[0.5,2.1] 0.936 0.6[0.3,1.2]
0.185
disorder
Substance
1.0[0.6,1.8]
0.937 1.1[0.5,2.8] 0.773 0.8[0.3,1.7]
0.532
use
disorder
Conduct
1.1[0.7,1.8]
0.550 1.0[0.5,2.8] 0.896 1.1[06,2.1]
0.651
disorder
Post0
0.026 0.4[0.2,0.8] 0.006 0.9[0.5,1.6]
0.691
traumatic
.6[0.4,0.9]
stress
disorder
Generalized
0.8[0.5,1.2]
0.204 0.8[0.5,1.6] 0.577 0.8[0.4,1.4]
0.351
anxiety
disorder
Panic
0.9[0.6,1.2]
0.440 0.7[0.4,1.2] 0.226 1.0[0.6,1.7]
0.977
disorder
Estimates in this table are from a series of regressions containing terms for smoking,
mental disorders and an interaction between smoking and mental disorders. For
example, the “Major depression” estimate of 0.6 is from a model with terms for
smoking, major depression and smoking*major depression. The presented OR is the OR
for the interaction between current smoking and major depression. The inverse
association (OR=0.6) suggests that the effects of cigarette smoking are smaller among
those who have major depression.

115

APPENDIX B: Educational background/other research experiences

I would like to provide a brief overview of my education and research activities outside
of this dissertation that I have been involved in. My interest in public health and epidemiologist
started while I was an undergraduate at Grand Valley State University. In my second year, I took
completed two independent studies with Drs. Justin Adams (an anthropologist) and Noor
Ghiasvand (a geneticist) where I wrote a review article which I later presented as a poster on
anti-malaria efforts throughout the world.
I came to the department in the Fall of 2009 as a Master’s student, and quickly began
working with Dr. Mat Reeves involved in a project on healthy lifestyles in the US. Specifically,
the goal of this project was to determine temporal changes and regional differences in the
prevalence of adults who live a ‘healthy lifestyle’ in the US—a healthy lifestyle was defined by
not smoking, having a normal body weight, eating a sufficient amount of fruits and vegetables
and getting a recommended amount of exercise. We found that very few adults in the US live a
healthy lifestyle (~5%), that prevalence hasn’t changed much in the past decade and that there
are notable differences by geographic region where the prevalence was highest in the
northeast and western states while lower in the southern and Midwest states. This trend was
independent of race, income, education and age distributions. The abstract and reference for
this paper can be found in appendix 3.
After my first semester, I was approached by Dr. Jim Anthony about joining his research
group as a NIDA-T32 trainee. Dr. Anthony was my professor for two introductory epidemiology
courses, so I had become familiar with his work. This was an excellent fit because at this point

116

my main interest had been on modifiable lifestyle behaviors, and under Jim I began to focus on
a line of research on the epidemiology of cigarette smoking.
My first project as a NIDA fellow focused on cigarette smoking from a global
perspective—specifically—how levels of smoking persistence vary cross-nationally. We found a
considerable amount of cross-national variation in the proportion of ‘ever smokers’ who persist
in smoking, ranging from ~25% in Nigeria to ~80% in China. Interestingly, these differences
were independent of country income level and elapsed time since smoking onset. (Abstract in
appendix 4.)
Other cigarette focused studies, currently under revision, include a case-crossover study
on the propensity that cigarette smoking and smokeless tobacco use predict one another
(appendix 5), a time to event study of the rapid transition from first cigarette to becoming a
daily smoker (appendix 6), a factor analysis on measurement of nicotine dependence in newlyincident smokers (appendix 7) and an analysis on the age-specific incidence rates of cigarette
smoking among young people in the US (appendix 8). While working with Dr. Elizabeth Meier, a
post-doctoral fellow, I was able to modify the methods used in the age-specific incidence of
smoking analysis for an age-specific incidence of prescription opioid use which Dr. Meier
presented at the 2011 College on Problems of Drug Dependence meeting, and later published
(appendix 9).
In addition to my NIDA work, I also worked as a research assistant to Dr. Naomi Breslau
on a project on post-traumatic stress disorder, an area of her expertise. This study asked if
predispositions to PTSD played a different role in whether someone developed PTSD depending
on trauma severity. The general assumption prior to this study was that severe traumatic
117

events (such as rape/sexual assault) that predispositions (such as major depression) play
smaller role in determining whether or not the victim will become a case of PTSD than in less
severe events (accidents, disasters and unexpected death of someone close). Interestingly, we
found little difference in the role of predispositions by trauma severity (appendix 10).
The intersection between my NIDA work with Dr. Anthony focused on cigarette
smoking, and my experiences with Dr. Breslau on mental health played an important role in
leading me to this dissertation topic.

118

APPENDIX C: List of non-dissertation related abstracts

Am J Public Health. 2012 Jul;102(7):1392-8.

Temporal and regional trends in the prevalence of healthy lifestyle
characteristics: United States, 1994-2007.
Troost JP, Rafferty AP, Luo Z, Reeves MJ.
Source
Department of Epidemiology, Michigan State University, East Lansing, MI 48824, USA. jtroost@epi.msu.edu

Abstract
OBJECTIVES:
We examined temporal and regional trends in the prevalence of health lifestyles in the United States.
METHODS:
We used 1994 to 2007 data from the Behavioral Risk Factor Surveillance System to assess 4 healthy lifestyle
characteristics: having a healthy weight, not smoking, consuming fruits and vegetables, and engaging in physical
activity. The concurrent presence of all 4 characteristics was defined as a healthy overall lifestyle. We used logistic
regression to assess temporal and regional trends.
RESULTS:
The percentages of individuals who did not smoke (4% increase) and had a healthy weight (10% decrease) showed
the strongest temporal changes from 1994 to 2007. There was little change in fruit and vegetable consumption or
physical activity. The prevalence of healthy lifestyles increased minimally over time and varied modestly across
regions; in 2007, percentages were higher in the Northeast (6%) and West (6%) than in the South (4%) and
Midwest (4%).
CONCLUSIONS:
Because of the large increases in overweight and the declines in smoking, there was little net change in the
prevalence of healthy lifestyles. Despite regional differences, the prevalence of healthy lifestyles across the United
States remains very low.
doi: 10.2105/AJPH.2011.300326. Epub 2012 May 21.

119

Journal of Epidemiology and Global health. 2012 Sep;2(3):135-144.

An updated global picture of cigarette smoking persistence among adults.
Troost JP, Barondess DA, Storr CL, Wells JE, Al-Hamzawi AO, Andrade LH, Bromet E, Bruffaerts R,
Florescu S, de Girolamo G, de Graaf R, Gureje O, Haro JM, Hu C, Huang Y, Karam AN, Kessler RC, Lepine
JP, Matschinger H, Medina-Mora ME, O’Neil S, Posada-Villa J, Sagar R, Takeshima T, Tomov T, Williams
DR, Anthony JC.
Source
Department of Epidemiology, Michigan State University, East Lansing, MI 48824, USA. jtroost@epi.msu.edu

Abstract
BACKGROUND:
Cross-national variance in smoking prevalence is relatively well documented. The aim of this study is to estimate levels of
smoking persistence across 21 countries with a hypothesized inverse relationship between country income level and smoking
persistence.

METHODS:
Data from the World Health Organization World Mental Health Survey Initiative were used to estimate cross-national
differences in smoking persistence—the proportion of adults who started to smoke and persisted in smoking by the date of the
survey.

RESULTS:
There is large variation in smoking persistence from 25% (Nigeria) to 85% (China), with a random-effects meta-analytic
summary estimate of 55% with considerable cross-national variation. (Cochran’s heterogeneity Q statistic = 6845; p < 0.001).
Meta-regressions indicated that observed differences are not attributable to differences in country’s income level, age
distribution of smokers, or how recent the onset of smoking began within each country.

CONCLUSIONS:
While smoking should remain an important public health issue in any country where smokers are present, this report identifies
several countries with higher levels of smoking persistence (namely, China and India).

120

This work was presented at the 2011 CPDD meeting.
Smokeless tobacco use and onset of cigarette smoking: a case-crossover study

J.P. Troost & J.C. Anthony; Epidemiology, Michigan State University, College of Human
Medicine, East Lansing, MI
AIMS: There is an open question as to whether the onset of smokeless tobacco use is protective of or
associated with an excess risk of tobacco cigarette smoking onset with few studies providing evidence in
either direction. This study seeks to examine the relationship between the smokeless tobacco onset and
tobacco cigarette onset using epidemiological evidence with a case-crossover method.
METHODS: We use a nationally representative sample of persons 12+ using data from the 2004-2008
National Surveys on Drug Use and Health with computer assisted assessment of smokeless tobacco and
tobacco cigarette use. We tested an immediate triggering hypothesis in which a positive (or protective)
association between smokeless tobacco and cigarette onsets would manifest within the first months
after smokeless tobacco onset. Using a within subject design that holds constant individual
susceptibility traits, we compared the rates of smokeless tobacco onsets in the month before tobacco
cigarette onset (the hazard month) to two control months: (1) the month after tobacco cigarette onset
(2) the month before the hazard month. We also stratified by elapsed time since onset to control for
possible recall biases. In a post estimation step we compared other control months.
RESULTS: Among newly-incident tobacco cigarette users, there was no difference in the odds smokeless
tobacco onset in the hazard interval (the month prior to tobacco cigarette use) compared to either
control interval (the month after tobacco cigarette use: RR=1.0, 95% CI: 0.7, 1.5; p=0.8488 or the month
after the month of cigarette onset: RR=1.5, 95% CI: 0.9, 2.3; p=0.079). There were no differences after
stratifying by elapsed time since onset. The use of alternative plausible control months did not alter the
results.
CONCLUSION: At least with respect to an immediate triggering hypothesis, there was no evidence to
suggest a positive or protective association between smokeless tobacco use and tobacco cigarette
smoking. [NIDA support: T32DA021129 (JPT); K05DA15799 (JCA)].

121

Newly-Incident Smokers:
Daily Smoking Soon After the First Tobacco Cigarette
Jonathan P. Troost, B.S.
James C. Anthony Ph.D.
Both Authors: Department of Epidemiology, College of Human Medicine, Michigan State University,
East Lansing, Michigan 48824
Background: In the United States, every day, as many as 6,000 young people start smoking tobacco
cigarettes. Some transition to daily smoking soon after the onset of smoking. Our goal was to estimate
occurrence and time to daily smoking once smoking starts, using recently gathered epidemiological data
on newly incident smokers.
Methods: Nationally representative community sample surveys were completed each year during 20042008. Standardized assessments disclosed 6,427 newly-incident smokers. We measured the month to
month occurrence of becoming a daily smoker, defined as smoking every day for at least 30 days, among
newly-incident smokers.
Results: An estimated one in nine newly-incident tobacco cigarette smokers transition into daily smoking
within one year after smoking for the first time (11%). There is no marked male-female or urban-rural
variation in this estimate. Older smokers (age 18+) are more likely to transition rapidly into daily
smoking, as compared to adolescent-onset smokers. Exploratory analyses disclosed that adolescentonset Hispanic smokers were less likely to transition rapidly to daily smoking, as compared to
adolescent-onset non-Hispanic Whites. Among non-Hispanics, adult-onset Blacks were more likely to
transition rapidly into daily smoking, as compared to adult-onset Whites.
Conclusions: In the US, smoking a tobacco cigarette for the first time can be a transient experience, but
for at least one in nine (11%), there is a major health-affecting consequence observed within one year
after smoking onset – namely, progression to daily smoking. Risk of rapid transition to daily smoking is
greater when smoking starts at or after age 18, especially for non-Hispanic Blacks.
Key words: Tobacco smoking; epidemiology; daily smoking

122

Measuring the Rapid Transition To Nicotine Dependence: AgeAssociated Biases
Jonathan P. Troost, B.S.
James C. Anthony Ph.D.
Both Authors: Department of Epidemiology, College of Human Medicine, Michigan State University,
East Lansing, Michigan 48824

Introduction: While evidence suggests some smokers experience a rapid transition from first cigarette to
nicotine dependence symptoms, we have not yet established a proper method for measuring nicotine
dependence in its early stages due to studies on nicotine dependence almost exclusively done in
experienced smokers. The goal of this study was to determine the psychometric properties of the
Nicotine Dependence Syndrome Scale plus one item from the Fagerström Test for Nicotine Dependence
in a nationally representative sample of newly incident smokers.
Methods: Data were collected as part of the United State’s National Survey on Drug Use and Health.
3,239 smokers with less than one year since their first cigarette were identified from 2004-2009. The 20item scale was used an in exploratory factor analysis among smokers in even numbered years; odd
numbered year data was used for a confirmatory factor analysis as well as a test of differential item
functioning by age.
Results: Judged by model fit and factor interpretability, a 17-item, single factor solution was selected. All
but one item was retained in the confirmatory analysis. Tests for differential item functioning revealed
that 8 items had significant age associated biases (for example “smoking the same on weekdays as on
weekends” was less often found among adolescents holding constant overall nicotine dependence).
Conclusions: While the NDSS appears to be a useful tool even in the early stages of smoking, ageassociated biases in questions are problematic. Longitudinal research on nicotine dependence traversing
adolescence to adulthood would benefit by either avoiding biased questions and/or developing new
items.

123

Age-Specific Risk of Starting to Smoke Tobacco Cigarettes:
United States, 2004-2008
Jonathan P. Troost, B.S.
James C. Anthony Ph.D.
Both Authors: Department of Epidemiology, College of Human Medicine, Michigan State University,
East Lansing, Michigan 48824

Background: Against a background of generally successful United States tobacco cigarette smoking
prevention and control during the late 20th century, we face continuing burdens of tobacco-attributable
morbidity and mortality. Here, we estimate age-specific risks of starting to smoke tobacco by studying
the early 21st century experiences of adolescents and young adults, and seek systematic replication of a
possible ‘age effect’ where the risk of smoking increases with age and peaks at age 18.
Methods: Standardized field surveys of non-institutionalized community residents age 12-21 years old,
completed each year from 2004 through 2008, with assessment of year at first smoking (n1=86,887
individuals at risk of smoking onset; n2=2,942 newly incident smokers). Newly-incident smoking onset
was ascertained when the year of smoking initiation was the same as the year of survey assessment,
given no prior smoking before that year.
Findings: For each birth cohort traversing adolescence across the years from 2004 through 2008, the risk
of starting to smoke tobacco cigarettes is roughly 1% for 12 year olds, increases with age, peaks at age
18 at 8% (one in 12), and then declines.
Conclusions: The risk of starting to smoke cigarettes in the US remains clinically significant. We can and
should find ways to decrease that risk.
Key words: Tobacco smoking; epidemiology; legal smoking age; smoking initiation

124

Arch Pediatr Adolesc Med. 2012 Sep 1;166(9):803-7.

Extramedical use of prescription pain relievers by youth aged 12 to 21
years in the United States: national estimates by age and by year.
Meier EA, Troost JP, Anthony JC.
Abstract
OBJECTIVE To identify when youth are most likely to start using prescription pain relievers to get high or for other
unapproved indications outside the boundaries of what a prescribing physician might intend (ie, extramedical use).
DESIGN Cross-sectional surveys of adolescent cohorts, 2004 to 2008. SETTING The United States. PARTICIPANTS
Large nationally representative samples of youth in the United States who had been assessed for the 2004 through
2008 National Survey on Drug Use and Health, yielding data from 138 729 participants aged 12 to 21 years. MAIN
OUTCOME MEASURES Estimated age-specific risk of starting extramedical use of prescription pain relievers, year
by year, and confirmation of age at peak risk by tracing the experience of individual cohorts during this period.
RESULTS The estimated peak risk of starting extramedical use of prescription pain relievers occurs in
midadolescence, well before the college years. The age at peak risk is 16 years, when an estimated 2% to 3%
become newly incident users. Smaller risk estimates are observed at age 12 to 14 years and at age 19 to 21 years.
CONCLUSIONS For initiatives to prevent youth from using prescription pain relievers to get high or for other
unapproved indications, a focus on the last year of high school and the post-secondary school years may be too
little too late. Practice-based approaches are needed in addition to public health interventions based on effective
alcohol and tobacco prevention programs during the earlier adolescent years.
doi: 10.1001/archpediatrics.2012.209.

125

Psychol Med. 2013 Feb;43(2):381-90.

Influence of predispositions on post-traumatic stress disorder: does it vary
by trauma severity?
Breslau N, Troost JP, Bohnert K, Luo Z.
Source
Department of Epidemiology and Biostatistics, College of Human Medicine, Michigan State University, East
Lansing, MI 48824, USA. breslau@epi.msu.edu

Abstract
BACKGROUND:
Only a minority of trauma victims (<10%) develops post-traumatic stress disorder (PTSD), suggesting that victims
vary in predispositions to the PTSD response to traumas. It is assumed that the influence of predispositions is
inversely related to trauma severity: when trauma is extreme predispositions are assumed to play a secondary
role. This assumption has not been tested. We estimate the influence of key predispositions on PTSD induced by
an extreme trauma - associated with a high percentage of PTSD - (sexual assault), relative to events of lower
magnitude (accidents, disaster, and unexpected death of someone close).
METHOD:
The National Epidemiologic Survey on Alcohol and Related Conditions (NESARC) is representative of the adult
population of the USA. A total of 34 653 respondents completed the second wave in which lifetime PTSD was
assessed. We conducted three series of multinomial logistic regressions, comparing the influence of six
predispositions on the PTSD effect of sexual assault with each comparison event. Three pre-existing disorders and
three parental history variables were examined.
RESULTS:
Predispositions predicted elevated PTSD risk among victims of sexual assault as they did among victims of
comparison events. We detected no evidence that the influence of predispositions on PTSD risk was significantly
lower when the event was sexual assault, relative to accidents, disasters and unexpected death of someone close.
CONCLUSIONS:
Important predispositions increase the risk of PTSD following sexual assault as much as they do following
accidents, disaster, and unexpected death of someone close. Research on other predispositions and alternative
classifications of event severity would be illuminating.
doi: 10.1017/S0033291712001195. Epub 2012 Jun 18.

126

BIBLIOGRAPHY

127

BIBLIOGRAPHY

Acton, G. S. (2003). Measurement of impulsivity in a hierarchical model of personality traits: implications
for substance use. Subst Use Misuse, 38(1), 67-83.
Alonso, J., Angermeyer, M. C., Bernert, S., Bruffaerts, R., Brugha, T. S., Bryson, H., et al. (2004). Sampling
and methods of the European Study of the Epidemiology of Mental Disorders (ESEMeD) project.
Acta Psychiatr Scand Suppl(420), 8-20.
Bang, S. J., & Commons, K. G. (2011). Age-dependent effects of initial exposure to nicotine on serotonin
neurons. Neuroscience, 179, 1-8.
Barondess, D. A., Meyer, E. M., Boinapally, P. M., Fairman, B., & Anthony, J. C. (2010). Epidemiological
evidence on count processes in the formation of tobacco dependence. Nicotine Tob Res, 12(7),
734-741.
Benowitz, N. L. (2011). Smokeless tobacco as a nicotine delivery device: harm or harm reduction? Clin
Pharmacol Ther, 90(4), 491-493.
Berggren, U., Eriksson, M., Fahlke, C., Blennow, K., & Balldin, J. (2007). Different effects of smoking or
use of smokeless tobacco on platelet MAO-B activity in type 1 alcohol-dependent subjects.
Alcohol Alcohol, 42(3), 267-271.
Berggren, U., Fahlke, C., Eriksson, M., & Balldin, J. (2003). Tobacco use is associated with reduced central
serotonergic neurotransmission in type 1 alcohol-dependent individuals. Alcohol Clin Exp Res,
27(8), 1257-1261.
Berkman, L. F., Glass, T., Brissette, I., & Seeman, T. E. (2000). From social integration to health: Durkheim
in the new millennium. Soc Sci Med, 51(6), 843-857.
Berlin, I., & Anthenelli, R. M. (2001). Monoamine oxidases and tobacco smoking. Int J
Neuropsychopharmacol, 4(1), 33-42.
Berlin, I., Covey, L. S., Donohue, M. C., & Agostiv, V. (2011). Duration of smoking abstinence and suiciderelated outcomes. Nicotine Tob Res, 13(10), 887-893.
Breslau, N., Schultz, L. R., Johnson, E. O., Peterson, E. L., & Davis, G. C. (2005). Smoking and the risk of
suicidal behavior: a prospective study of a community sample. Arch Gen Psychiatry, 62(3), 328334.
Capron, D. W., Allan, N. P., Norr, A. M., Zvolensky, M. J., Schmidt, N. B. (2014) The effect of successful
and unsuccessful smoking cessation on short-term anxiety, depression, and suicidality.
Addicitive Behaviors. 39(4) 782-788
Caris, L.H. & Anthony, J. C. Tobacco and depression: evidence from Chile. Unpublished.
128

Clarke, D. E., Goodwin, R.D., Messias, L. M., Eatson, W. W. (2008). Asthma and suicidal ideation with and
without suicide attempts among adults in the United States: what is the role of cigarette
smoking and mental disorders? Ann Allergy Asthma Immunol. 100(5) 439-46
Clarke, D. E., Eaton, W. W., Petronis, K. R., Ko, J. Y., Chatterjee, A., & Anthony, J. C. (2010). Increased risk
of suicidal ideation in smokers and former smokers compared to never smokers: evidence from
the Baltimore ECA follow-up study. Suicide Life Threat Behav, 40(4), 307-318.
Cross-Cultural Survey Guidelines (2010). from
http://ccsg.isr.umich.edu/pdf/FullGuidelinesMarch2012.pdf
Doll, R., & Peto, R. (1976). Mortality in relation to smoking: 20 years' observations on male British
doctors. Br Med J, 2(6051), 1525-1536.
Doll, R., Peto, R., Boreham, J., & Sutherland, I. (2004). Mortality in relation to smoking: 50 years'
observations on male British doctors. BMJ, 328(7455), 1519.
Durkheim, E. (1951). Suicide, a study in sociology: Glencoe, III.:Free Press.
Fowler, J. S., Logan, J., Wang, G. J., & Volkow, N. D. (2003). Monoamine oxidase and cigarette smoking.
Neurotoxicology, 24(1), 75-82.
Fowler, J. S., Volkow, N. D., Wang, G. J., Pappas, N., Logan, J., MacGregor, R., et al. (1996). Inhibition of
monoamine oxidase B in the brains of smokers. Nature, 379(6567), 733-736.
Fowler, J. S., Volkow, N. D., Wang, G. J., Pappas, N., Logan, J., Shea, C., et al. (1996). Brain monoamine
oxidase A inhibition in cigarette smokers. Proc Natl Acad Sci U S A, 93(24), 14065-14069.
Friberg, L., Cederlof, R., Lorich, U., Lundman, T., & De Faire, U. (1973). Mortality in twins in relation to
smoking habits and alcohol problems. Arch Environ Health, 27(5), 294-304.
Geracioti, T. D., Jr., West, S. A., Baker, D. G., Hill, K. K., Ekhator, N. N., Wortman, M. D., et al. (1999). Low
CSF concentration of a dopamine metabolite in tobacco smokers. Am J Psychiatry, 156(1), 130132.
Grant, B., & Kaplan, K. (2005). Source and Accuracy Statement for the Wave 2 National Epidemiologic
Survey on Alcohol and Related Conditions (NESARC). National Institute on Alcohol Abuse and
Alcoholism: Rockville, Maryland.
Grant, B., Kaplan, K., Shepard, J., & Moore, T. (2003). Source and Accuracy Statement
for Wave 1 of the 2001-2002 National Epidemiologic Survey on Alcohol and
Related Conditions. National Institute on Alcohol Abuse and Alcoholism:
Bethesda MD.
Grucza, R. A., Przybeck, T. R., & Cloninger, C. R. (2007). Prevalence and correlates of binge eating
disorder in a community sample. Compr Psychiatry, 48(2), 124-131.
Harkness, J., Pennell, B. E., Villar, A., Gebler, N., Aguilar-Gaxiola, S., & Bilgen, I. (2008). Translation
procedures and translation assessment in the World Mental Health Survey Initiative.

129

Harrison, L., Martin, S., Eney, T., & Harrington, D. (2007). Comparing drug testing and self-report of drug
use among youths and young adults in the general population Available from
http://www.samhsa.gov/data/nsduh/drugtest.pdf
Heeringa, S., Wells, J. E., Hubbard, F., Mneimneh, Z. N., & Chiu, W. T. (2008). Sample designs and
sampling procedures. In: Kessler RC, Üstün TB, editors. The World Mental Health Survey Initiative
volume one: patterns of mental illness in the WMH Surveys. New York: Cambridge University
Press.
Heeringa, S. G., Wagner, J., Torres, M., Duan, N., Adams, T., & Berglund, P. (2004). Sample designs and
sampling methods for the Collaborative Psychiatric Epidemiology Studies (CPES). Int J Methods
Psychiatr Res, 13(4), 221-240.
Hemenway, D., Solnick, S. J., & Colditz, G. A. (1993). Smoking and suicide among nurses. Am J Public
Health, 83(2), 249-251.
Iwasaki, M., Akechi, T., Uchitomi, Y., & Tsugane, S. (2005). Cigarette smoking and completed suicide
among middle-aged men: a population-based cohort study in Japan. Ann Epidemiol, 15(4), 286292.
Johnston, J. P. (1968). Some observations upon a new inhibitor of monoamine oxidase in brain tissue.
Biochem Pharmacol, 17(7), 1285-1297.
Jokinen, J. Carlborg, A. Martensson, B., Forslund, K. Nordstrom, A.L., Nordstrom, P.. (2007)DST nonsuppression predicts suicide after attempted suicide. Psychiatry Research, 150, 297-303.
Kenny, P. J., File, S. E., & Rattray, M. (2001). Nicotine regulates 5-HT(1A) receptor gene expression in the
cerebral cortex and dorsal hippocampus. Eur J Neurosci, 13(6), 1267-1271.
Kessler, R. C., Berglund, P. A., Borges, G., Castilla-Puentes, R. C., Glantz, M. D., Jaeger, S. A., et al. (2007).
Smoking and suicidal behaviors in the National Comorbidity Survey: Replication. J Nerv Ment Dis,
195(5), 369-377.
Kessler, R. C., Borges, G., Sampson, N., Miller, M., & Nock, M. K. (2009). The association between
smoking and subsequent suicide-related outcomes in the National Comorbidity Survey panel
sample. Mol Psychiatry, 14(12), 1132-1142.
Kessler, R. C., & Ustun, T. B. (2004). The World Mental Health (WMH) Survey Initiative Version of the
World Health Organization (WHO) Composite International Diagnostic Interview (CIDI). Int J
Methods Psychiatr Res, 13(2), 93-121.
Knoll, J., & Magyar, K. (1972). Some puzzling pharmacological effects of monoamine oxidase inhibitors.
Adv Biochem Psychopharmacol, 5, 393-408.
Krishnan-Sarin, S., Reynolds, B., Duhig, A. M., Smith, A., Liss, T., McFetridge, A., et al. (2007). Behavioral
impulsivity predicts treatment outcome in a smoking cessation program for adolescent smokers.
Drug Alcohol Depend, 88(1), 79-82.

130

Lakon, C. M., Hipp, J. R., & Timberlake, D. S. (2010). The social context of adolescent smoking: a systems
perspective. Am J Public Health, 100(7), 1218-1228.
Lance, C. E., & Vandenburg, R. J. (2008). Statistical and Methodological Myths and Urban Legends:
Doctrine, Verity and Fable in Organizational and Social Sciences (1 ed.): Routledge Academic.
Leistikow, B. N., Martin, D. C., & Samuels, S. J. (2000). Injury death excesses in smokers: a 1990-95
United States national cohort study. Inj Prev, 6(4), 277-280.
Lewiecki, E. M., & Miller, S. A. (2013). Suicide, guns, and public policy. Am J Public Health, 103(1), 27-31.
Li, D., Yang, X., Ge, Z., Hao, Y., Wang, Q., Liu, F., et al. (2012). Cigarette smoking and risk of completed
suicide: A meta-analysis of prospective cohort studies. J Psychiatr Res, 46(10), 1257-1266.
Martini, S., Wagner, F.A., Anthony, J.C. (2002). The association of tobacco smoking and depression in
adolescence: evidence in the United States. Substance Use&Misuse, 37(14), 1853-1867.
Mathers, C. D., & Loncar, D. (2006). Projections of global mortality and burden of disease from 2002 to
2030. PLoS Med, 3(11), e442.
McKlveen, J.M., Wilson, J.M., Rubin, R.T., Rhodes, M.E. (2010) Sexually diergic, dose-dependent
hypothalamic-pituitary-adrenal axis responses to nicotine in a dynamic in vitro perfusion
system.Journal of Pharmcological and Toxicological Methods, 61, 311-318
Meier, E. A., Troost, J. P., & Anthony, J. C. (2012). Extramedical Use of Prescription Pain Relievers by
Youth Aged 12 to 21 Years in the United States: National Estimates by Age and by Year. Arch
Pediatr Adolesc Med.
Miller, D. C. (1991). Handbook of research design and social measurement. Newbuty Park, CA: Sage.
Miller, M., Hemenway, D., Bell, N. S., Yore, M. M., & Amoroso, P. J. (2000). Cigarette smoking and
suicide: a prospective study of 300,000 male active-duty Army soldiers. Am J Epidemiol, 151(11),
1060-1063.
Miller, M., Hemenway, D., & Rimm, E. (2000). Cigarettes and suicide: a prospective study of 50,000 men.
Am J Public Health, 90(5), 768-773.
Moscicki, E. K. (1995). Epidemiology of suicide. Int Psychogeriatr, 7(2), 137-148.
NCI. (1992). Smokeless tobacco or health. An international perspective: U.S. Department of Health and
Human Services, Public Health Service, National Institutes of Health.
Newhouse, P. A., Sunderland, T., Tariot, P. N., Blumhardt, C. L., Weingartner, H., Mellow, A., et al.
(1988). Intravenous nicotine in Alzheimer's disease: a pilot study. Psychopharmacology (Berl),
95(2), 171-175.
Norman, T. R., Chamberlain, K. G., & French, M. A. (1987). Platelet monoamine oxidase: low activity in
cigarette smokers. Psychiatry Res, 20(3), 199-205.
131

Oreland, L., Fowler, C. J., & Schalling, D. (1981). Low platelet monoamine oxidase activity in cigarette
smokers. Life Sci, 29(24), 2511-2518.
Paffenbarger, R. S., Jr., Lee, I. M., & Leung, R. (1994). Physical activity and personal characteristics
associated with depression and suicide in American college men. Acta Psychiatr Scand Suppl,
377, 16-22.
Pandey, G. N. (1997). Altered serotonin function in suicide. Evidence from platelet and neuroendocrine
studies. Ann N Y Acad Sci, 836, 182-200.
Pennell, B. E., Bowers, A., Carr, D., Chardoul, S., Cheung, G. Q., Dinkelmann, K., et al. (2004). The
development and implementation of the National Comorbidity Survey Replication, the National
Survey of American Life, and the National Latino and Asian American Survey. Int J Methods
Psychiatr Res, 13(4), 241-269.
Perez, V. W. (2005). The relationship between seriously considering, planning, and attempting suicide in
the youth risk behavior survey. Suicide Life Threat Behav, 35(1), 35-49.
Rantakallio, P., Laara, E., & Koiranen, M. (1995). A 28 year follow up of mortality among women who
smoked during pregnancy. BMJ, 311(7003), 477-480.
Rebholz, C. M., Gu, D., Yang, W., Chen, J., Wu, X., Huang, J. F., et al. (2011). Mortality from suicide and
other external cause injuries in China: a prospective cohort study. BMC Public Health, 11, 56.
Riala, K., Alaraisanen, A., Taanila, A., Hakko, H., Timonen, M., & Rasanen, P. (2007). Regular daily
smoking among 14-year-old adolescents increases the subsequent risk for suicide: the Northern
Finland 1966 Birth Cohort Study. J Clin Psychiatry, 68(5), 775-780.
Rohleder, N. & Kirschbaum, C. (2006). The hypothalamic-pituitary-adrenal (HPA) axis in habitual
smokers. International Journal of Psychophysiology, 59, 236-243
Romano, P. S., Bloom, J., & Syme, S. L. (1991). Smoking, social support, and hassles in an urban AfricanAmerican community. Am J Public Health, 81(11), 1415-1422.
Saccone, N. L., Rice, J. P., Rochberg, N., Goate, A., Reich, T., Shears, S., et al. (1999). Genome screen for
platelet monoamine oxidase (MAO) activity. Am J Med Genet, 88(5), 517-521.
SAMHSA. (2010a). NSDUH: Confidentiality. Retrieved July 2, 2012, from
https://nsduhweb.rti.org/RespWeb/confidentiality.html
SAMHSA. (2010b). NSDUH: Methods. Retrieved June 25, 2012, from
http://www.oas.samhsa.gov/NSDUH/Methods.cfm
Sarason, I. G. (1985). Social Support: Theory, Research and Applications. Boston, MA: Martinus Nijhoff.
Singer, T. P., & Ramsay, R. R. (1995). Flauoprotein structure and mechanism 2. Monoamine oxidases: old
friends hold many surprises. FASEB J, 9(8), 605-610.

132

Smith, A. R., Witte, T. K., Teale, N. E., King, S. L., Bender, T. W., & Joiner, T. E. (2008). Revisiting
impulsivity in suicide: implications for civil liability of third parties. Behav Sci Law, 26(6), 779797.
Smith, G. D., Phillips, A. N., & Neaton, J. D. (1992). Smoking as "independent" risk factor for suicide:
illustration of an artifact from observational epidemiology? Lancet, 340(8821), 709-712.
Tanskanen, A., Tuomilehto, J., Viinamaki, H., Vartiainen, E., Lehtonen, J., & Puska, P. (2000). Smoking and
the risk of suicide. Acta Psychiatr Scand, 101(3), 243-245.
Tverdal, A., Thelle, D., Stensvold, I., Leren, P., & Bjartveit, K. (1993). Mortality in relation to smoking
history: 13 years' follow-up of 68,000 Norwegian men and women 35-49 years. J Clin Epidemiol,
46(5), 475-487.
Waldeck, T. L., & Miller, L. S. (1997). Gender and impulsivity differences in licit substance use. J Subst
Abuse, 9, 269-275.
WHO. Suicide prevention (SURPE). 2012, from
http://www.who.int/mental_health/prevention/suicide/suicideprevent/en
Youdim, M. B., & Riederer, P. (1993). Dopamine metabolism and neurotransmission in primate brain in
relationship to monoamine oxidase A and B inhibition. J Neural Transm Gen Sect, 91(2-3), 181195.

133