Exploring metabolic factors and health behaviors in relation to suicide attempts: A case-control study

Author’s Accepted Manuscript Exploring Metabolic Factors and Health Behaviors In Relation To Suicide Attempts: A Case-Control Study Stefan Perera, Rebecca B. Eisen, Meha Bhatt, Brittany B. Dennis, Monica Bawor, Wala ElSheikh, Jane De Jesus, Sumathy Rangarajan, Heather Sholer, Elisabeth Iordan, Pam Mackie, Shofiqul Islam, Mahshid Dehghan, Jennifer Brasch, David Meyre, Russell de Souza, Lehana Thabane, Zainab Samaan

PII: DOI: Reference:

www.elsevier.com/locate/jad

S0165-0327(17)31311-3 https://doi.org/10.1016/j.jad.2017.12.060 JAD9468

To appear in: Journal of Affective Disorders Received date: 28 June 2017 Revised date: 14 November 2017 Accepted date: 30 December 2017 Cite this article as: Stefan Perera, Rebecca B. Eisen, Meha Bhatt, Brittany B. Dennis, Monica Bawor, Wala El-Sheikh, Jane De Jesus, Sumathy Rangarajan, Heather Sholer, Elisabeth Iordan, Pam Mackie, Shofiqul Islam, Mahshid Dehghan, Jennifer Brasch, David Meyre, Russell de Souza, Lehana Thabane and Zainab Samaan, Exploring Metabolic Factors and Health Behaviors In Relation To Suicide Attempts: A Case-Control Study, Journal of Affective Disorders, https://doi.org/10.1016/j.jad.2017.12.060 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Exploring Metabolic Factors and Health Behaviors In Relation To Suicide Attempts: A Case-Control Study Stefan Perera, MSc;1 Rebecca B. Eisen, MSc;2 Meha Bhatt, MSc;1 Brittany B. Dennis, PhD;1,3 Monica Bawor, PhD;2,3 Wala El-Sheikh, BSc;4 Jane DeJesus, BSc, MLT;4 Sumathy Rangarajan, MSc;4 Heather Sholer, RPN;5 Elisabeth Iordan, BScN;5 Pam Mackie,4 Shofiqul Islam, MSc;4 Mahshid Dehghan, PhD;4 Jennifer Brasch, MD;5,6 David Meyre, PhD;1,4 Russell de Souza, ScD, RD;1 Lehana Thabane, PhD;1,4,5,6,7 Zainab Samaan, MBChB, DMMD, PhD, MRCPsych;1,3,4,8,9* 1

Department of Health Research Methods, Evidence and Impact, McMaster University,

1280 Main St. West, Hamilton, ON L8S 4L8, Canada 2

MiNDS Neuroscience Graduate Program, McMaster University, 1280 Main Street W.,

Hamilton, ON L8S 4L8, Canada 3

St. George’s University of London, London, Cranmer Terrace, London SW17 0RE,

United Kingdom 4

Population Health Research Institute, Hamilton General Hospital, 237 Barton Street

East, Hamilton, ON L8L 2X2, Canada 5

St. Joseph’s Healthcare Hamilton, 100 West 5th Street, Hamilton, ON L8N 3K7,

Canada 6

Biostatistics Unit, Centre for Evaluation of Medicine, 25 Main Street West Suite 2000,

Hamilton, ON L8P 1H1, Canada 7

System-Linked Research Unit on Health and Social Service Utilization, McMaster

University, 1280 Main Street West, Hamilton, ON L8S 4L8, Canada

8

Department of Medicine, McMaster University, 1280 Main Street West, Hamilton, ON

L8S 4L8, Canada 9

Peter Boris Centre for Addiction Research, St. Joseph’s Healthcare Hamilton, 100 West

5th Street, Hamilton, ON L8P 3R2, Canada Corresponding Author: Name: Dr. Zainab Samaan Email: [email protected] Phone: 905 522 1155 (ext. 36372) Fax: 905 381-5629 Address: Department of Psychiatry and Behavioural Neuroscience McMaster University 100 West 5th Street, Hamilton, Ontario, L8N 3K7, Canada. ABSTRACT

Background: Suicide attempts are a serious public health concern with devastating global impact, thereby necessitating the development of an adequate prevention strategy. Few known risk factors of suicide attempts are directly modifiable. This study sought to investigate potential associations between health behaviors and suicide attempts, identifying novel opportunities for clinicians to help prevent suicidal behavior. Methods: A case-control study was conducted to compare body weight, serum total cholesterol, physical activity, tobacco use, and dietary food groups among adults who had made a suicide attempt (n=84) to psychiatric inpatients (n=104) and community controls (n=93) without history of suicide attempt. Multivariable binary logistic regression analyses were used to investigate the association between metabolic risk factors and attempted suicide. Results: Psychiatric inpatients who had attempted suicide were less likely to be physically active [moderate/strenuous (OR 0.42, 95% CI 0.19-0.95) and mild (OR 0.35, 95% CI 0.16-0.76)]. Psychiatric inpatients who attempted suicide were more likely to use tobacco (OR 2.25, 95% CI 1.07-4.73). Contrary to prior research, obesity, serum total cholesterol, and diet were not significantly associated with risk of attempted suicide.

2

Limitations: Our study was limited by its cross-sectional design, which precludes the identification of causal or temporal relationships between the risk of attempted suicide and factors such as physical activity and tobacco use. Conclusions: Study results suggest that a history of attempted suicide is associated with a decreased likelihood of being physically active and an increased risk of tobacco use. Further investigation is warranted to understand the role of exercise and tobacco use in suicide intervention and prevention strategies. Keywords: Suicidal Behavior; Suicide Attempt; Obesity; BMI; Physical Activity

INTRODUCTION The World Health Organization cites suicide as one of the 20 most common causes of death, claiming 800,000 lives each year (World Health Organization, 2014). The Centers for Disease Control and Prevention estimates that there are 25 suicide attempts for every death by suicide (Crosby et al., 2011). Suicidal behavior (ideation, attempts, completed suicide) devastates families and communities, along with the individuals who survive suicide attempts. Such individuals commonly experience longterm suffering from psychiatric and medical comorbidity. Many face a future of repeated suicide attempts, hospitalization, poverty, and often, death by suicide (Goldman-Mellor et al., 2014; Pajonk et al., 2005). Several known factors associated with an increased risk of attempted suicide are not directly modifiable, including female sex, age (45-64 and over 85), unanticipated adverse life events, and a family or personal history of previous suicidal behavior (Mann, 2003; Nock et al., 2008). Several health behaviors and metabolic factors such as diet, smoking, physical activity, and body mass index (BMI) may be associated with risk of attempted suicide. Identifying health behaviors that are associated with suicide risk may

3

aid the treatment and prevention of suicidal behaviour given that health behaviours are potentially modifiable. This study is designed to identify metabolic factors and health behaviors that are associated with suicide risk to identify novel opportunities for clinicians to help prevent suicide. Suspected risk factors Although current literature is conflicted, the majority of the evidence suggests an association between BMI and suicidal behavior (Perera et al., 2016). Notwithstanding the association of psychiatric illness (Luppino et al., 2010), medical comorbidity (Dixon, 2010), and social stigma (Puhl and Heuer, 2009) with obesity, obese and overweight individuals are shown to be less likely to complete suicide than normal weight individuals; some studies have found an inverse relationship between BMI and completed suicide (McCarthy et al., 2014; Mukamal et al., 2010). However, several other studies have found that obesity increases the risk of attempted suicide (Dong et al., 2006; Wagner et al., 2013). Some studies have suggested that BMI may influence the casefatality of attempted suicide by affecting one’s likelihood of choosing a less lethal method of suicide, or by reducing the case fatality of a specific method of suicide (Batty et al. ; Perera et al., 2016). For example, the use of more lethal methods requiring greater physical exertion, such as hanging or jumping, may be less common among those with a greater BMI. Obesity may be associated with lower risk of completed suicide due to body size and distribution of body fat, resulting in fewer fatalities (and seemingly greater nonlethal suicide attempts) from intentional overdoses. Given that obesity (World Health Organization, 2013) and suicidal behavior (World Health Organization, 2014) both

4

remain significant public health issues, it is vital to clarify the association and direction of effects between them. Cholesterol is an integral part of the central nervous system. Specifically, cholesterol is thought to play an indirect role in brain function through its involvement in cell membrane stability, modification of enzyme function, absorption and transport of fat soluble organic materials, and through its effects on neurotransmission (Zhang, 2011). However, the evidence for serum cholesterol as a biological marker of suicidal behavior is largely inconsistent to date. Some studies have demonstrated a significant association between aberrantly low serum-total cholesterol levels and increased risk of attempted suicide (Diaz-Sastre et al., 2007; Perez-Rodriguez et al., 2008; Zhang, 2011), while others have demonstrated no association (da Graça Cantarelli et al., 2015; Park et al., 2013; Tanskanen et al., 2014). Only a few studies have researched potential relationships between diet and suicidal behavior (Nanri et al., 2013; Zhang et al., 2005). Polyunsaturated fatty acids (PUFA), for instance, are shown to be associated with major depressive disorder (Hibbeln, 2009). However, there is a paucity of research evaluating the effect of seafood, nuts and other foods rich in PUFA among people who have recently attempted suicide. Smoking has been consistently shown to be associated with an increased risk of attempted suicide (Malone et al., 2003). The relationship between smoking and attempted suicide may be confounded given that several of the risk factors between smoking and risk of attempted suicide overlap, such as psychiatric comorbidity, impulsiveness, and lower socio-economic status (Hughes, 2008). Moreover, smoking is associated with negative physical health outcomes which may potentially exacerbate suicide risk.

5

Therefore, it remains important to investigate the association between smoking and attempted suicide in the context of other health behaviors and comparing this risk to both psychiatric and general populations. Our study aimed to assess the association of these metabolic factors and health behaviors among psychiatric inpatients who had made a recent suicide attempt in comparison to psychiatric inpatients and members of the general population who have never attempted suicide. Despite the literature showing a consistent inverse association between BMI and completed suicide, we expect that those who attempted suicide will have greater BMI than those who have never attempted suicide given that some evidence suggests that obesity is associated with an increased risk of attempted suicide (Perera et al., 2016). Although previous research is conflicted, we hypothesized that serum total cholesterol would be inversely associated with risk of attempted suicide given the importance of cholesterol to brain function and based on previous studies that have cited this potential association (Zhang, 2011). Based on previous literature, we anticipated that greater levels of physical activity would be associated with a decreased risk of attempted suicide, and that greater tobacco use would be associated with an increased risk of attempted suicide. METHODS We studied BMI, waist-circumference, serum-total cholesterol, tobacco use, diet, and physical activity. We assessed these factors among psychiatric inpatients (≥ 18 years) who had attempted suicide no more than three months prior to recruitment (case group). We compared the case group to two separate control groups of psychiatric inpatients and community members who had never attempted suicide.

6

Study Sample Data for this study were drawn from the Study of Determinants of Suicide Conventional and Emergent Risk (Samaan et al., 2015b), an age and sex matched casecontrol study designed to explore risk factors of attempted suicide. We completed recruitment and data collection between March 2011 and November 2014 in Hamilton, Ontario, a mid-size Canadian city. Participants were recruited from St. Joseph’s Healthcare and Hamilton Health Sciences Hospitals and the greater Hamilton community. This study was approved by the Hamilton Health Sciences (#10-661) and St. Joseph’s Healthcare Research Ethics Boards (#11-3479). The reporting of results adheres to the Strengthening of Reporting of Observational Studies in Epidemiology (STROBE) guidelines (Von Elm et al., 2007). Inclusion and Exclusion Criteria The study included men and women 18 years or older who could provide written informed consent, communicate in English, and follow study procedures. The cases included psychiatric inpatients who had made a suicide attempt– defined as a selfdirected injury with specific intent to die that necessitated admission to a hospital ward – no more than three months prior to recruitment. The controls consisted of two comparator groups. The first comprised psychiatric inpatients who had no history of suicide attempts and who were admitted to hospital at the same time as the cases. The second control group comprised individuals recruited from the community or non-psychiatric hospital services who had also never attempted suicide. Please see Figure 1 for a summary of the participant selection process.

7

Recruitment and Data Collection Trained research personnel scanned the hospital wards for potential participants who were consecutively recruited based on eligibility. Researchers for inpatient recruitment approached the clinical staff and asked about suitable participants based on study inclusion criteria. Volunteer requests were posted in general hospitals and community settings to recruit community controls. Trained researchers collected data and conducted structured, face-to-face interviews involving validated rating scales and questionnaires. Upon obtaining consent from participants, researchers subsequently collected the data within a single visit by administering the questionnaires in a structured and ordered fashion that took between 30-45 minutes. Participation was done so on a voluntary basis and without compensation. Participants provided socio-demographic factors including age, sex, and socioeconomic status. The Beck Suicide Intent Scale is a 19-item clinical research instrument designed to quantify and assess suicidal intention and was used in this study to ensure that the case group consisted of individuals who attempted suicide with specific intent to die. The Mini-International Neuropsychiatric Interview (M.I.N.I.) is a short structured clinical interview that allows researchers to make a diagnosis of psychiatric disorders according to DSM-IV criteria (Sheehan et al., 1998). The M.I.N.I. was used to keep a record of psychiatric diagnoses in the case and psychiatric control groups. The International Physical Activity Questionnaire, (Hallal and Gomes, 2004) which assesses physical activity during leisure time and at work. However, given that 73.49% of cases and 54.64% of controls were unemployed, we opted to only use the questions pertaining to leisure time and not work. For a more detailed outline of the study procedures, please refer to our published protocol (Samaan et al., 2015a).

8

A food frequency questionnaire was administered to assess participants’ intake of fish and seafood, nuts, fruits, and vegetables over the previous year (Dehghan et al., 2017). These selections were then converted into the average daily intake in grams per day using reference values from the SHARE Food Frequency Questionnaire (SHARE FFQ), (Kelemen et al., 2003) created to be generalizable to a multi-ethnic Canadian population. Research personnel measured participants’ height to the nearest centimetre, weight to the nearest hundred grams, and waist circumference to the nearest millimetre. Laboratory Methods Fasting blood samples were collected and assayed for serum total cholesterol using the Beckman UniCelDxC600 Reagent (Synchron Systems Cholesterol Reagent). After 30 minutes of clotting time, tubes were spun at 1500 x g (3000 rpm) for 15 minutes until blood was well separated. Serum samples were subsequently aliquoted and stored in cryrovials within 2 hours of collection and frozen in liquid nitrogen (- 196°C) at the Clinical Research and Clinical Trials Laboratory, Hamilton. Serum samples were analysed blinded to the case/control status. Statistical Analyses We analysed the prevalence of each potential risk factor, comparing cases to a single comparator group with the psychiatric and community controls combined, and to each of the two control groups separately. This subgroup analysis allowed us to identify factors that had a unique effect on attempted suicide in general. Specifically, we were able to compare to factors that differentiate psychiatric inpatients that attempted suicide from psychiatric inpatients without history of attempted suicide, and psychiatric

9

inpatients who attempted suicide compared to members of the general population without history of attempted suicide. STATA version 13 was used to perform all analyses (STATA, 2013). Sample characteristics are summarized in Table 1 as mean and standard deviation (SD) for continuous variables and as frequency (percent) for categorical variables. We employed an Analysis of Variance (ANOVA) to compare continuous variables across different study groups and chi-square (χ2) tests to compare proportions for categorical variables. Non-parametric equivalent tests and Fisher’s Exact Test were used when statistical assumptions were not met. Binary logistic regression analyses was performed to compute odds ratios [ORs] and accompanying 95% confidence intervals (CIs), and p-values adjusted for age and sex. The level of significance was set alpha = 0.05, and was not adjusted for multiple testing because we have only one hypothesis and primary analyses. Subgroup analyses are exploratory and only help to explore the data further. A multivariable logistic regression model was also designed using all study variables to assess the association between each metabolic variable of interest while controlling for the effects of all other variables. We present the results pictorially for all regression analyses (Figures 2-4). We used unmatched analyses because perfect matching was not feasible. Matched analyses would result in unnecessary exclusion of participant data and reducing sample size. In our primary binary logistic regression analyses, the psychiatric inpatients and community members were combined as a single comparator group, with cases coded as a 1 and controls as a 0. Subgroup analyses were conducted using logistic regression to compare cases to community controls and psychiatric controls separately (again with

10

cases coded as a 1 and each control group coded as 0). The Hosmer-Lemeshow test was used to assess the goodness-of-fit of each logistic regression model. For primary binary logistic regression analyses, we used the rule of thumb of 10 observations (cases) per explanatory variable in the model and therefore had adequate power in our primary analyses, but not in the subgroup analyses which were exploratory (Peduzzi et al., 1996). As a sensitivity analysis, multiple imputation using chained equations (Royston and White, 2011) was performed to adjust for the missing data in the age/sex and multivariable models (please see below primary and subgroup analyses for a sensitivity analysis on missing data). Age, sex, obesity, waist circumference, serum total cholesterol, physical activity, tobacco use, and diet were used to aid in the multiple imputation prediction of missing values.

RESULTS The sample comprised 281 individuals, including 84 case participants and 197 control participants. The control group further consisted of 104 psychiatric inpatients and 93 community members. Figure 1 outlines the number of individuals approached and assessed for eligibility, the number of individuals who were excluded with reasons, and the number of participants included in the final sample. The mean age of cases and controls were 43.00 years (SD = 14.02) and 45.66 years (SD = 15.99) respectively. The case group included 52.4% (n=44) females and the control group included 49.8% females (n=98). The majority of the cases were Caucasian (91.6%), while 72.8% of controls were Caucasian and the second most common ethnicity among controls was African (12.8%).

11

Table 1 displays the demographic characteristics of the study sample. Table 2 provides a per group summary of the primary metabolic variables. Primary Analyses Figure 2 summarizes the odds ratios for individual risk factors adjusted for age and sex and by multivariable adjustment for all risk factors when combining the psychiatric and community controls as a single comparator group. Those who attempted suicide were at a 69% decreased odds of engaging in self-reported mild physical activity (OR 0.31, 95% CI 0.16-0.62, p = 0.001) and at 68% decreased odds of engaging in moderate/strenuous (OR 0.32, 95% CI 0.16-0.64, p = 0.001) physical activity during leisure time compared to those who did not attempt suicide after adjusting for age and sex. Those who attempted suicide had 3.15 times greater odds of being a current user of tobacco products (OR 3.15, 95% CI 1.66-5.95, p < 0.001) and had 2.09 times greater odds of being a former user of tobacco products (OR 2.09 95% CI 1.04-4.19, p = 0.039) than those who had never attempted suicide in age/sex adjusted models. Having previously made a suicide attempt was not significantly associated with obesity and overweightness (BMI > 25), abdominal obesity (wait circumference > 102cm men, > 88 cm women), and serum total cholesterol in age/sex adjusted models (Figure 2, p-values > 0.05). Daily food consumption in each food group (specifically, nuts, fish/seafood, and fruits and vegetables) in grams per day was divided into tertiles and individuals in the upper and middle tertile were compared to those in the lowest tertile for each food group. Individuals who attempted suicide consumed less fish/seafood, nuts, and fruits and vegetables on average; however, only daily consumption of nuts was significantly

12

associated with a decreased risk of attempted suicide among those in the middle tertile compared to those in the lowest tertile (OR 0.51, 95% CI 0.28-0.92, p = 0.026) in age/sex adjusted models. The highest tertile of nut consumption was not significantly associated with a decreased risk of attempted suicide (OR 0.50, 95% CI 0.22-1.14, p = 0.10). The multivariable binary logistic regression model including all variables (age, sex, obesity/overweight, abdominal obesity, exercise, serum total cholesterol, tobacco use, and dietary variables) demonstrated that those who attempted suicide had 2.25 greater odds of currently using tobacco (OR 2.25, 95% CI 1.07-4.73, p = 0.031). Moreover, the multivariable binary logistic regression demonstrated that those who attempted suicide were at 65% decreased odds of engaging in self-reported mild physical activity during (OR 0.35, 95% CI 0.16-0.76, p = 0.008) and 58% decreased odds of engaging in moderate/strenuous physical activity (OR 0.42, 95% CI 0.19-0.95, p = 0.036) during leisure time as compared to controls who had not attempted suicide (Figure 2). Having previously attempted suicide was not significantly association with BMI, waistcircumference, serum total cholesterol, and each dietary food group in the multivariable logistic regression model. Subgroup Analyses Subgroup analyses (Figures 3 and 4) revealed that the relationship between metabolic risk factors and attempted suicide differed depending on whether cases were compared to community members or psychiatric inpatients. Case status of having attempted suicide remained not associated with BMI, serum-total cholesterol, and waist circumference regardless of the comparator group. The odds of being a current (OR 5.78, 95% CI 2.05-16.31, p= 0.001) and former (OR 3.43, 95% CI 1.30-9.05, p= 0.013) user of

13

tobacco was increased when cases were compared solely to community controls (Figure 3). Having attempted suicide was associated with a decreased odds of engaging in mild (OR 0.12, 95% CI 0.04-0.39, p< 0.001) and moderate/strenuous (OR 0.15, 95% CI 0.050.48, p= 0.001) physical activity when cases were compared to community controls. When compared to community controls, cases were at decreased odds of having moderate (OR 0.35, 95% CI 0.17-0.70, p= 0.003) and high (OR 0.29, 95% CI 0.12-0.72, p= 0.008) consumption of nuts compared to those in the lowest tertile of daily nut consumption in age/sex adjusted models but not after multivariable adjustment. Tobacco use and moderate or strenuous exercise were no longer significantly associated with attempted suicide when psychiatric controls were used as the comparator group (Figure 4). Other than having decreased odds of engaging in mild levels of physical activity (in the age/sex adjusted but not the multivariable model), having attempted suicide was not significantly associated with dietary food groups or any other metabolic factors when cases were compared to psychiatric controls (Figure 4, p-values > 0.05). Sensitivity Analyses Given the comprehensive data collection process, we expected missing data and pre-specified methods of handling missing data. Overall, 270 participants (96%) had their height and weight measured, 264 participants (94%) had their waist-circumference measured, and 263 (93.6%) participants had available serum total cholesterol data. For questionnaires, 263 (93·6%) participants completed the Physical Activity Questionnaire and FFQ, while 273 individuals (97%) self-reported tobacco use. We had complete data for all participants’ age and sex, but in the multivariable logistic regression, only 239 (85%) participants were included in the final model for complete-case analysis. We

14

performed multiple imputation as a sensitivity analysis to determine the impact of the missing data on our primary and subgroup analyses. The results were not changed in the imputed multivariable model as having attempted suicide remained significantly associated with an increased odds of current tobacco use (OR 2.41, 95% CI 1.21-4.79, p= 0.012) and a decreased odds of physical (mild (OR 0.36, 95% CI 0.17-0.74, p= 0.006) moderate/strenuous (OR 0.39, 95% CI 0.18-0.85, p= 0.018)) as compared to the control group. Moreover, having attempted suicide remained unassociated with obesity, waist circumference, serum total cholesterol, and diet (p-values >0.05). In the imputed multivariable subgroup models, having attempted suicide remained associated with an increased odds of tobacco use and decreased odds of engaging in physical activity when cases were compared to community controls. Similarly, having attempted suicide was not significantly associated with any metabolic factors or heath behaviors in an imputed subgroup model comparing cases to psychiatric controls (p-values >0.05).

DISCUSSION This study investigated the relationship between attempted suicide and metabolic factors and health behaviors, including BMI, abdominal obesity, physical activity, serumtotal cholesterol, tobacco use, and diet. While some of these variables have previously been investigated, ours is the first case-control study to comprehensively examine metabolic factors and health behaviors among individuals who have recently attempted suicide in comparison to both psychiatric and community control groups. In both multivariable and age/sex-adjusted models, having attempted suicide was significantly associated with a decreased likelihood of engaging in physical activity

15

during leisure time compared to those who had never attempted suicide. This association held for those who reported mild, moderate, or strenuous activity. Our results are consistent with previous findings of an inverse association between physical activity and suicide risk in a study of inpatient veterans (Davidson et al., 2013). Increased physical activity may decrease suicidal impulses by improving quality of life, producing physical health benefits, and alleviating psychiatric and social disability (Richardson et al., 2014). Physical activity may also decrease perceived burdensomeness – known to be associated with suicidal behavior (Van Orden et al., 2012) – by improving physical health, mood, and sense of accomplishment (Davidson et al., 2013). Exercise programs also provide opportunities for social connectivity, which reduces social isolation and a sense of loneliness. Additionally, strenuous exercise may produce an increased release of endorphins which might stimulate motivation and pleasure centres in the brain (Dishman and O'Connor, 2009). Our subgroup analyses reflects established knowledge that individuals with serious psychiatric illnesses are less physically active than the general population (Richardson et al., 2014). When we compared psychiatric controls with case subjects, we only found significant differences in mild physical activity; the major differences in the association between attempted suicide and physical activity were between the community group and those who had attempted suicide. Our results support previous studies that suggests those with a history of attempted suicide are more likely to smoke (Hughes, 2008). However, considering the high rate of smoking among psychiatric populations (el-Guebaly et al., 2014), smoking may not be a predictor for the risk of suicide attempts in clinical settings. Smoking may

16

be a surrogate marker of other substance use, trait impulsivity, and risk-taking behavior which may independently increase the risk of attempting suicide (Mann, 2003). We found no evidence to suggest that those who attempted suicide were more likely to be obese or overweight in our primary or subgroup analyses. Our results are inconsistent with research suggesting that obese and overweight individuals have a decreased risk of completed suicide (McCarthy et al., 2014; Mukamal et al., 2010) and an increased risk of attempted suicide (Dong et al., 2006; Wagner et al., 2013). A number of factors may account for this disparity. Previous research used self-reported measures to collect obesity data (Dong et al., 2006; Wagner et al., 2013). People tend to overestimate their height and underestimate their weight(Wen and Kowaleski-Jones, 2012), which may have biased past self-reported results (Dong et al., 2006; Wagner et al., 2013) Previous studies have also assessed past-year (Phillips et al., 2004) or lifetime suicide attempts (Wagner et al., 2013) in relation to current body weight or have assessed a longitudinal association between a single measure of BMI taken at baseline with suicide attempts that occurred years later (McCarthy et al., 2014; Mukamal et al., 2010). This is a methodological limitation, due to the change in body weight over time, particularly among those with psychiatric and medical illness or major life stressors (Mukamal and Miller, 2010). Measurement of BMI is also an issue, as it does not distinguish between body weight that is carried as muscle or fat, and may be an inappropriate measure of obesity for groups such as elderly people who often experience atypical changes in muscle mass (Mukamal and Miller, 2010). Waist circumference provides an alternative measure of

17

obesity; though our findings suggest that this measure of abdominal obesity was not associated with risk of attempted suicide in the main or subgroup analyses. Past research suggests that low serum-total cholesterol is a suicide risk factor (Diaz-Sastre et al., 2007; Perez-Rodriguez et al., 2008; Zhang, 2011), though our study demonstrated no such association. Moreover, the computed odds ratios suggested a nonsignificant trend towards higher serum-total cholesterol predicting an increased risk of attempted suicide. Our findings are consistent with more recent evidence that high serumtotal cholesterol is a risk factor for attempted suicide (Tanskanen et al., 2014) as well as with other reports that found no significant links between cholesterol and suicide risk (da Graça Cantarelli et al., 2015; Park et al., 2013). Future studies should aim to include more detailed assessments of diet to rule out malnutrition or appetite changes related to depression that may confound this association (Mukamal and Miller, 2010). Low serumtotal cholesterol may be a marker of psychiatric illness such as major depressive disorder rather than of suicidal behavior itself (Zhang, 2011). Consumption of foods rich in polyunsaturated fatty acids (PUFA) (e.g. nuts, fish/seafood) was lower among individuals who attempted suicide compared to community controls, consistent with previous studies (Nanri et al., 2013; Zhang, 2011). Consumption of fruits and vegetables did not differ between community members and psychiatric inpatients. High and moderate consumption of nuts was associated with a decreased risk of attempted suicide when individuals who had attempted suicide were compared to community controls. The FFQ used in the current study was limited to food groups and should be considered as an exploratory analysis that may be followed up by

18

more detailed analyses of diet which may confirm whether lower consumption of fish and nuts among those attempting suicide is related to lower intake of PUFA and suicide risk. Several previous studies investigating BMI, cholesterol, and other potential risk factors of attempted suicide have compared those who have attempted suicide to medically and psychiatrically healthy individuals (Diaz-Sastre et al., 2007; McCarthy et al., 2014; Mukamal et al., 2010; Tanskanen et al., 2014). While widely accepted risk factors of suicidal behavior may dramatically differ between individuals who are suicidal and healthy community members, (Mann, 2003) these differences may not be as strong when individuals who attempted suicide are compared to psychiatric individuals with no history of attempted suicide. This was reflected by our subgroup analyses which revealed that exercise and tobacco use were only significantly associated with attempted suicide when cases were compared to community members but not when they were compared to psychiatric inpatients. Psychiatric patients are the most vulnerable population for suicidal behavior (Mann, 2003), making it imperative to identify risk factors that distinguish those with a psychiatric disorder who attempt suicide from those who do not. Strengths and Limitations This study involved comparator groups from psychiatric and community populations, and assessment of recent suicide attempts requiring hospitalization that were distinct from acts of non-suicidal self-harm. These features helped eliminate bias seen in previous studies and provide more robust results. A sensitivity analysis was conducted to multiply impute missing data, and the results were consistent with our complete-case analyses suggesting minimal bias due to missing data. Our study faced limitations due to its cross-sectional design, which prevented us from identifying any causal relationships

19

between the risk of attempted suicide and factors such as physical activity and tobacco use. A prospective cohort study is however difficult to conduct due to the low event rate of suicide in the population and the difficulty in having all participants exposed to the different risk factors. We had robust measures of BMI, waist-circumference, and serumtotal cholesterol, but our data on physical activity, tobacco use, and diet were selfreported (albeit with validated questionnaires) and therefore prone to recall bias. Moreover, given the high rates of unemployment in both case and control groups, we limited our analyses using the International Physical Activity Questionnaire to items pertaining to physical activity during leisure time, thereby discounting physical activity occurring at work. Future studies should aim for a more comprehensive evaluation of physical activity in the context of suicidal behavior. Future studies should implement repeated measures of physical activity and total cholesterol to account for potential changes in metabolic factors from psychological stress, psychotropic medication use, and incident psychiatric illness. Conclusions Integrating exercise into future intervention and treatment could make a significant difference for people at risk of suicidal behavior. Conversely, several metabolic factors previously thought to affect suicidal behavior – BMI, abdominal obesity, and serum-total cholesterol – demonstrated no significant association with a history of attempted suicide. It is important to clarify why metabolic factors such as BMI may be differentially be association with completed and attempted suicide in future work. Finally, while individuals at risk of attempting suicide tend to use more tobacco products and exercise less, this trend is also seen among non-suicidal psychiatric patients. For this

20

reason, it remains unclear whether smoking and physical activity habits represent a useful clinical predictor of suicide risk or if they are simply markers of psychiatric disorders. With additional research regarding the causality between smoking, physical activity, and risk of attempted suicide, this line of research will become important to inform future prevention and intervention strategies. CONTRIBUTORS S.P. analysed and interpreted the data and wrote the first draft of manuscript. R.E. and M.B. (third author) contributed to the data analysis and critical revision of the manuscript. M.B. (fifth author) and B.D. assisted with recruitment and critically revised the manuscript. W.E-S. recruited participants and organized data collection. J.D. was the study coordinator and helped with study protocol implementation, data collection and quality assurance. S.R. was the study project manager responsible for the study conduct; with Z.S. she designed the study protocol, provided supervision of study personnel, and coordinated the overall study management. H.S., and E.I. are nurses specializing in psychiatry; they recruited hospitalized study patients and collected data. P.M. was responsible for database management and cleaning. S.I. was responsible for data management. M.D. designed the dietary tool and analysed the dietary data for the DISCOVER study. J.B. provided advice on study procedures including recruitment and the inclusion of psychiatric controls based on her expertise in emergency psychiatry and suicide prevention. D.M. and R.D. critically revised the manuscript. L.T. was responsible for the case-control study design, and overall methodological and statistical aspects of the study. Z.S. conceived the study and was principally responsible for the conduct of the study and obtaining funding. Z.S. also developed the research question, contributed to data analysis and interpretation, and critically revised the manuscript. All authors have reviewed and approved the final manuscript. ROLE OF THE FUNDING SOURCE This work was supported by the Brain and Behavior Research Foundation Young Investigator Grant (#19058). The Brain and Behavior Research Foundation has no role in the design of the study or publication of the results. ACKNOWLEDGEMENTS The authors would like to thank the Brain and Behavior Research Found for funding our work. The authors would like to thank the study participants; without their generous contributions, this study would not be possible. We extend our gratitude to Hamilton Health Sciences and St. Joseph’s Healthcare Hamilton hospitals and managers who facilitated the study and provided helpful feedback throughout the investigation.

21

DECLARATION OF INTEREST Conflicts of interest: none.

References Batty, G.D., Whitley, E., Kivimaki, M., Tynelius, P., Rasmussen, F., Body mass index and attempted suicide: Cohort study of 1,133,019 Swedish men. American Journal of Epidemiology 172, 890-899. Crosby, A., Gfroerer, J., Han, B., Ortega, L., Parks, S.E., 2011. Suicidal thoughts and behaviors among adults aged 18 ears--United States, 2008-2009. US Department of Health and Human Services, Centers for Disease Control and Prevention. CURI HALLAL, P., GOMES VICTORA, C., 2004. Reliability and validity of the international physical activity questionnaire (IPAQ). Med. Sci. Sports Exerc. 36, 556556. da Graça Cantarelli, M., Nardin, P., Buffon, A., Eidt, M.C., Godoy, L.A., Fernandes, B.S., Gonçalves, C.-A., 2015. Serum triglycerides, but not cholesterol or leptin, are decreased in suicide attempters with mood disorders. J. Affect. Disord. 172, 403-409. Davidson, C.L., Babson, K.A., Bonn‐Miller, M.O., Souter, T., Vannoy, S., 2013. The impact of exercise on suicide risk: examining pathways through depression, PTSD, and sleep in an inpatient sample of veterans. Suicide Life Threat. Behav. 43, 279-289. Dehghan, M., Ge, Y., El Sheikh, W., Bawor, M., Rangarajan, S., Dennis, B., Vair, J., Sholer, H., Hutchinson, N., Iordan, E., 2017. Comparability of a short food frequency questionnaire to assess diet quality: the DISCOVER study. International journal of food sciences and nutrition, 1-10.

22

Diaz-Sastre, C., Baca-Garcia, E., Perez-Rodriguez, M.M., Garcia-Resa, E., Ceverino, A., Saiz-Ruiz, J., Oquendo, M.A., De Leon, J., 2007. Low plasma cholesterol levels in suicidal males: a gender-and body mass index-matched case-control study of suicide attempters and nonattempters. Prog. Neuropsychopharmacol. Biol. Psychiatry 31, 901905. Dishman, R.K., O'Connor, P.J., 2009. Lessons in exercise neurobiology: the case of endorphins. Mental Health and Physical Activity 2, 4-9. Dixon, J.B., 2010. The effect of obesity on health outcomes. Molecular and cellular endocrinology 316, 104-108. Dong, C., Li, W., Li, D., Price, R., 2006. Extreme obesity is associated with attempted suicides: results from a family study. Int. J. Obes. 30, 388-390. el-Guebaly, N., Cathcart, J., Currie, S., Brown, D., Gloster, S., 2014. Smoking cessation approaches for persons with mental illness or addictive disorders. Psychiatr. Serv. Goldman-Mellor, S.J., Caspi, A., Harrington, H., Hogan, S., Nada-Raja, S., Poulton, R., Moffitt, T.E., 2014. Suicide attempt in young people: a signal for long-term health care and social needs. JAMA psychiatry 71, 119-127. Hibbeln, J.R., 2009. Depression, Suicide and Deficiencies of Omega–3 Essential Fatty Acids in Modern Diets. Hughes, J.R., 2008. Smoking and suicide: a brief overview. Drug and alcohol dependence 98, 169-178. Kelemen, L.E., Anand, S.S., Vuksan, V., Yi, Q., Teo, K.K., Devanesen, S., Yusuf, S., 2003. Development and evaluation of cultural food frequency questionnaires for South Asians, Chinese, and Europeans in North America. J. Am. Diet. Assoc. 103, 1178-1184.

23

Luppino, F.S., de Wit, L.M., Bouvy, P.F., Stijnen, T., Cuijpers, P., Penninx, B.W., Zitman, F.G., 2010. Overweight, obesity, and depression: a systematic review and metaanalysis of longitudinal studies. Arch. Gen. Psychiatry 67, 220-229. Malone, K.M., Waternaux, C., Haas, G.L., Cooper, T.B., Li, S., Mann, J.J., 2003. Cigarette smoking, suicidal behavior, and serotonin function in major psychiatric disorders. American Journal of Psychiatry 160, 773-779. Mann, J.J., 2003. Neurobiology of suicidal behaviour. Nature Reviews Neuroscience 4, 819-828. McCarthy, J.F., Ilgen, M.A., Austin, K., Blow, F.C., Katz, I.R., 2014. Associations between body mass index and suicide in the veterans affairs health system. Obesity 22, 269-276. Mukamal, K.J., Miller, M., 2010. Invited commentary: body mass index and suicide— untangling an unlikely association. Am. J. Epidemiol., kwq278. Mukamal, K.J., Rimm, E.B., Kawachi, I., O'Reilly, E.J., Calle, E.E., Miller, M., 2010. Body mass index and risk of suicide among one million US adults. Epidemiology 21, 8286. Nanri, A., Mizoue, T., Poudel-Tandukar, K., Noda, M., Kato, M., Kurotani, K., Goto, A., Oba, S., Inoue, M., Tsugane, S., 2013. Dietary patterns and suicide in Japanese adults: the Japan public health center-based prospective study. The British Journal of Psychiatry 203, 422-427. Nock, M.K., Borges, G., Bromet, E.J., Alonso, J., Angermeyer, M., Beautrais, A., Bruffaerts, R., Chiu, W.T., De Girolamo, G., Gluzman, S., 2008. Cross-national

24

prevalence and risk factors for suicidal ideation, plans and attempts. The British Journal of Psychiatry 192, 98-105. Organization, W.H., 2014. Preventing suicide: A global imperative. World Health Organization. Pajonk, F.-G., Ruchholtz, S., Waydhas, C., Schneider-Axmann, T., 2005. Long-term follow-up after severe suicide attempt by multiple blunt trauma. European psychiatry 20, 115-120. Park, S., Yi, K.K., Lim, A., Hong, J., 2013. No association between serum cholesterol and death by suicide in patients with schizophrenia, bipolar affective disorder, or major depressive disorder. Behavioral and Brain Functions 9, 45. Peduzzi, P., Concato, J., Kemper, E., Holford, T.R., Feinstein, A.R., 1996. A simulation study of the number of events per variable in logistic regression analysis. Journal of clinical epidemiology 49, 1373-1379. Perera, S., Eisen, R.B., Dennis, B.B., Bawor, M., Bhatt, M., Bhatnagar, N., Thabane, L., Souza, R., Samaan, Z., 2016. Body Mass Index Is an Important Predictor for Suicide: Results from a Systematic Review and Meta‐Analysis. Suicide and life-threatening behavior. Perez-Rodriguez, M.M., Baca-Garcia, E., Diaz-Sastre, C., Garcia-Resa, E., Ceverino, A., Saiz-Ruiz, J., Oquendo, M.A., De Leon, J., 2008. Low serum cholesterol may be associated with suicide attempt history. The Journal of clinical psychiatry 69, 1920-1927. Phillips, J., Carpenter, K.M., Nunes, E.V., 2004. Suicide risk in depressed methadonemaintained patients: associations with clinical and demographic characteristics. American Journal on Addictions 13, 327-332.

25

Puhl, R.M., Heuer, C.A., 2009. The stigma of obesity: a review and update. Obesity 17, 941-964. Richardson, C.R., Faulkner, G., McDevitt, J., Skrinar, G.S., Hutchinson, D.S., Piette, J.D., 2014. Integrating physical activity into mental health services for persons with serious mental illness. Psychiatr. Serv. Royston, P., White, I.R., 2011. Multiple imputation by chained equations (MICE): implementation in Stata. J Stat Softw 45, 1-20. Samaan, Z., Bawor, M., Dennis, B.B., El-Sheikh, W., DeJesus, J., Rangarajan, S., Vair, J., Sholer, H., Hutchinson, N., Iordan, E., 2015a. Exploring the Determinants of Suicidal Behavior: Conventional and Emergent Risk (DISCOVER): a feasibility study. Pilot and Feasibility Studies 1, 17. Samaan, Z., Bawor, M., Dennis, B.B., El-Sheikh, W., DeJesus, J., Rangarajan, S., Vair, J., Sholer, H., Hutchinson, N., Iordan, E., 2015b. Exploring the Determinants of Suicidal Behavior: Conventional and Emergent Risk (DISCOVER): a feasibility study. Pilot and Feasibility Studies 1, 1. Sheehan, D.V., Lecrubier, Y., Sheehan, K.H., Amorim, P., Janavs, J., Weiller, E., Hergueta, T., Baker, R., Dunbar, G.C., 1998. The Mini-International Neuropsychiatric Interview (MINI): the development and validation of a structured diagnostic psychiatric interview for DSM-IV and ICD-10. J. Clin. Psychiatry 59, 22-33. Tanskanen, A., Vartiainen, E., Tuomilehto, J., Viinamãª, H., Lehtonen, J., Puska, P., 2014. High serum cholesterol and risk of suicide. Am. J. Psychiatry.

26

Van Orden, K.A., Cukrowicz, K.C., Witte, T.K., Joiner Jr, T.E., 2012. Thwarted belongingness and perceived burdensomeness: construct validity and psychometric properties of the Interpersonal Needs Questionnaire. Psychol. Assess. 24, 197. Von Elm, E., Altman, D.G., Egger, M., Pocock, S.J., Gøtzsche, P.C., Vandenbroucke, J.P., Initiative, S., 2007. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. Prev. Med. 45, 247-251. Wagner, B., Klinitzke, G., Brähler, E., Kersting, A., 2013. Extreme obesity is associated with suicidal behavior and suicide attempts in adults: results of a population‐based representative sample. Depress. Anxiety 30, 975-981. Wen, M., Kowaleski-Jones, L., 2012. Sex and ethnic differences in validity of selfreported adult height, weight and body mass index. Ethnicity & disease 22, 72. WHO, 2013. Obesity and overweight—fact sheet No. 311. 2006. Zhang, J., 2011. Epidemiological link between low cholesterol and suicidality: a puzzle never finished. Nutritional neuroscience 14, 268-287. Zhang, J., Li, Y., Torres, M.E., 2005. How does a suicide attempter eat differently from others? Comparison of macronutrient intakes. Nutrition 21, 711-717.

27

Figure 1: Participant Recruitment Flow Diagram

28

Figure 2 : Forest plot summarizing the odds ratios (OR) and confidence intervals (CI) from multivariable and age/sex adjusted logistic regression analyses when treating the psychiatric inpatients and community members without history of suicide attempt as a combined control group. Multivariable analysis included age/sex and all variables listed in the forest plot.

29

Figure 3 Subgroup Logistic Regression Analyses: Forest plot summarizing the odds ratios (OR) and confidence intervals (CI) from multivariable and age/sex adjusted logistic regression analyses using only community members without history of suicide attempt as the control group. Multivariable analysis included age/sex and all variables listed in the forest plot.

30

Figure 4 Subgroup Logistic Regression Analyses: Forest plot summarizing the odds ratios (OR) and confidence intervals (CI) from multivariable and age/sex adjusted logistic regression analyses when only using psychiatric inpatients without history of suicide attempt as the control group. Multivariable analysis included age/sex and all variables listed in the forest plot.

31

Table 1- Descriptive Characteristics of Cases and Controls Cases (n= 84)

Control s (n= 197)

Univariate Differences 1

Psychiatri Communit c Controls y Controls (n=104) (n=93)

Univariate Differences 2

Demographic Information Age (years): mean (SD)

43.00 (14.02)

45.66 (15.99)

Z=1.25 p=0.212

45.04 (14.25)

46.35 (17.78)

F=1.05 df= 2, 278 p=0.353

Sex [n (% female )]

44 (52.38)

98 (49.75)

χ2=0.164 df =1 p = 0.686

52 (50.00)

46 (49.46)

χ2=0.169 df= 2 p = 0.919

Completed Post-Secondary Education or More, n (%)

35 (42.68)

121 (62.69)

χ2=9.39 df =1 p = 0.002

50 (50.00)

71 (76.34)

χ2=23.01 df= 2 p<0.001

Employed n (%)

22 (26.51)

88 (45.36)

χ2=8.63 df =1 p=0.003

33 (32.67)

55 (59.14)

χ2=22.80 df= 2 p<0.001

1

T-tests and chi-square tests were used to compare the cases to the psychiatric/community controls combined. ANOVA and chi-squared tests compare the three different groups: cases, psychiatric controls, and community controls. Analysis of variance tests (ANOVA) were used to compare means for continuous variables. Chi-square tests were used to compare proportions for categorical variables. Fisher’s exact test was used for categorical variables when one or more values in the contingency table were below 5. 2

32

Table 2- Metabolic Factors and Health Behaviours Across Groups Cases (n= 84)

Controls (n= 197)

Univariate Differences1

Psychiatric Controls (n=104)

Community Controls (n=93)

Univariate Differences2

Metabolic Information Body Mass Index (kg/m2): Mean (SD)

27.85 (7.27)

28.40 (6.38)

z=1.07 p=0.284

29.05 (6.56)

27.72 (6.15)

F=1.13 df= 2, 267 p=0.323

Obese (BMI ≥ 30kg/m2): n (%)

23 (28.75)

64 (33.68)

χ2=1.99 df=3 p=0.574

36 (37.11)

28 (30.11)

χ2=4.26 df=6 p=0.641

Overweight (BMI 25-29.99 kg/m2): n (%)

26 (32.50)

68 (35.79)

χ2=1.99 df=3 p=0.574

36 (37.11)

32 (34.41)

χ2=4.26 df=6 p=0.641

Normal Weight (BMI 18.5-24.99 kg/m2): n (%)

28 (35.00)

54 (28.42)

χ2=1.99 df=3 p=0.574

23 (23.71)

31 (33.33)

χ2=4.26 df=6 p=0.641

Underweight (BMI <18.5 kg/m2): n (%)

3 (3.75)

4 (2.11)

χ2=1.99 df=3 p=0.574

2 (2.06)

2 (2.15)

χ2=4.26 df=6 p=0.641

Waist-Circumference Males: (cm): Mean (SD) 90.34 (17.76) Females: 87.5 (21.95)

Males: 95.34 (16.50) Females: 82.37 (18.24)

Z=0.44 p=0.681

Males: 92.49 (18.49) Females: 82.79 (18.35)

Males: 98.32 (13.70) Females: 81.93 (18.31)

F=0.51 df= 2, 261 p=0.600

Serum Total Cholesterol (mmol/L): Mean (SD)

4.78 (1.12)

4.64 (0.96)

Z=-0.46 p=0.649

4.63 (1.02)

4.65 (0.90)

F=0.54 df= 2, 260 p=0.581

Current Tobacco Use: n (%)

35 (43.21)

46 (23.96)

χ2=12.78 df=2 p=0.002

35 (35.35)

11 (11.83)

χ2=28.57 df=4 p<0.001

Former Tobacco Use: n (%)

22 (27.16)

49 (25.52)

χ2=12.78 df=2 p=0.002

26 (26.26)

23 (24.73)

χ2=28.57 df=4 p<0.001

Never Tobacco Use: n (%)

24 (29.63)

97 (50.52)

χ2=12.78 df=2 p=0.002

38 (38.38)

59 (63.44)

χ2=28.57 df=4 p<0.001

30 (49.47)

32 (17.11)

χ2=15.04 df=2 p=0.001

23 (24.47)

9 (9.68)

χ2=22.25 df=4 p<0.001

Tobacco Use

Physical Activity Mainly Sedentary: n (%)

33

Mild Exercise: n (%)

22 (28.95)

77 (41.18)

χ2=15.04 df=2 p=0.001

39 (41.49)

38 (40.86)

χ2=22.25 df=4 p<0.001

Moderate/Strenuous Exercise: n (%)

24 (31.58)

78 (41.71)

χ2=15.04 df=2 p=0.001

32 (34.04)

46 (49.46)

χ2=22.25 df=4 p<0.001

Play sports or exercise during leisure time: n (%)

20 (26.32)

70 (37.43)

χ2=2.97 df=1 p=0.085

28 (29.79)

42 (45.16)

χ2=7.88 df=2 p=0.019

1

T-tests and chi-square tests were used to compare the cases to the psychiatric/community controls combined. ANOVA and chi-squared tests compare the three different groups: cases, psychiatric controls, and community controls. Analysis of variance tests (ANOVA) were used to compare means for continuous variables. Chi-square tests were used to compare proportions for categorical variables. Fisher’s exact test was used for categorical variables when one or more values in the contingency table were below 5. 2

34

Highlights     

Metabolic factors and health behaviors were investigated as suicide risk factors. Novel use of both general population and psychiatric inpatient control groups. Physical activity was associated with decreased risk of suicidal behavior. Tobacco use was associated with increased risk of suicidal behavior. BMI and serum total cholesterol not associated with risk of attempted suicide.

35