SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgment
  8. Disclosure
  9. REFERENCES

Higher BMI has been associated with a lower risk of suicide in large prospective studies, but the mechanisms for this link require elucidation. In the 2002 and 2004 iterations of the Behavioral Risk Factor Surveillance System (BRFSS), a large, population-based telephone survey of US adults conducted by the Centers for Disease Control and Prevention, participants reported their height, weight, and several potential risk factors for suicide, including alcohol use, mental health, marital status, firearm ownership, and risk-taking behaviors. We assessed whether BMI was associated with these risk factors among 224,247 eligible respondents in 2002 and 275,194 in 2004 after sample-weighted adjustment for age, race, region, smoking, and education. Alcohol-related risk factors tended to be lower with heavier BMI among women, while firearm-related risk factors tended to be higher with heavier BMI among men. Heavier BMI also tended to be associated with unmarried status and poor mental health, especially among women, and with infrequent seat belt use in men and women. No potential risk factors were consistently inversely associated with BMI in both sexes and years. In summary, in these samples of the US population, conventional risk factors for suicide were inconsistently associated with BMI, making them unlikely mediators for the observed relationship of BMI with lower risk of suicide. In some cases, risk factors were actually greater with heavier BMI. Further study of the relationship of BMI and suicide may yield novel modifiable risk factors that could cause or prevent this important cause of death.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgment
  8. Disclosure
  9. REFERENCES

Suicide is public health problem of enormous consequence, accounting for >32,000 deaths in 2004 and representing the 11th leading cause of death in the United States (1). Several important risk factors for suicide are known (2,3,4,5,6), including male sex, white race, depression, disability, life-threatening illness, substance abuse, and firearm ownership.

The epidemic of obesity in the United States has spurred increasing interest in its relationship with depression. The psychosocial burden of obesity is considerable and includes stigmatization in the workplace and in medical settings (7,8). Given the social stigma often associated with obesity, it is unsurprising that obesity has been associated with depressive and other mental health symptoms in studies from around the world (9,10,11), including a few that have documented improvement with weight loss (12,13). At the same time, a few studies have the found the opposite (14,15,16), supporting the “jolly fat” hypothesis first theorized in 1976.

Interestingly, as far back as 1966 (ref. 17), increasingly well-powered and well-designed studies have demonstrated a relationship of heavier BMI with a lower risk of suicide (18,19,20,21,22,23,24), in some cases despite a positive relationship with depression (19). The mechanisms for this association are unknown but speculation has pointed to carbohydrate intake, fatty acid and cholesterol metabolism, insulin sensitivity, and cerebral serotonin and tryptophan levels (18,25,26,27,28). However, the potential relationships of obesity with established risk factors for suicide completion have not been explored.

Given this lack of information, we examined the associations of established risk factors for suicide with heavier BMI (and thus potential mechanisms for the BMI-suicide relationship) in two administrations of the Behavioral Risk Factor Surveillance System (BRFSS), the largest known telephone survey worldwide. BRFSS is a population-based survey administered yearly by the Centers for Disease Control and Prevention and includes extensive self-reported information on height, weight, and several risk factors for suicide.

Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgment
  8. Disclosure
  9. REFERENCES

Survey design

The BRFSS, administered by the Centers for Disease Control and Prevention, is an ongoing data collection program designed to measure behavioral risk factors in the adult population (≥18 years of age) living in households in the 50 states, the District of Columbia, Puerto Rico, Guam, and the Virgin Islands (29). Factors assessed by the BRFSS include tobacco use, health care coverage, HIV/AIDS knowledge and prevention, physical activity, and fruit and vegetable consumption, among others. Data are collected from a random sample of adults (one per household) through a telephone survey conducted by state health personnel or contractors. The questionnaire has three parts: (i) the core component; (ii) optional modules; and (iii) state-added questions. The 2002 and 2004 core and module questionnaires are publicly accessible at http:www.cdc.govbrfssquestionnairesenglish.htm. The 2002 survey had a median CASRO response rate (a conservative calculation that incorporates sampling efficiency and cooperation rate) of 58.3% and a final sample size of 247,964. The corresponding figures for 2004 were 52.7% and 303,822.

The BRFSS informs all respondents at the outset that the survey is anonymous and confidential, that it collects no personally identifying information, and that answering any or all questions is entirely voluntary; consent is presumed on the basis of willingness to participate. Our analyses were subjected to ethics review by the Beth Israel Deaconess Medical Center Committee on Clinical Investigations, which provided an exemption from continuing review.

Assessment of BMI

Participants reported their height in feet and inches and their weight in pounds. These were converted to units of BMI in kg/m2. We categorized BMI according to WHO criteria as underweight (<18.5), normal weight (18.5–24.9), overweight (25.0–29.9), class I obesity (30.0–34.9), and class II or higher (≥35.0). We excluded individuals with extreme values of BMI (<15 and >50).

Assessment of potential risk factors for suicide

We evaluated several domains that potential risk factors for suicide encompassed, including risk taking and impulsiveness, mental health, social support, and access to firearms. Specific risk factors included: (i) consumption of five or more drinks on an occasion (or binge drinking) in the preceding 30 days (30); (ii) drinking after having had too much to drink in the last 30 days (31); (iii) average alcohol use above NIAAA recommended limits of two drinks per day for men and one for women; (iv) sunburn in the preceding 12 months (in 2004) (32); (v) use of seatbelts less frequently than nearly always or always (in 2002); (vi) a composite HIV/AIDS risk factor among individuals aged <65 years that included any of the following within the past year: use of intravenous drugs, treatment for a sexually transmitted disease, payment or receipt of money or drugs in exchange for sex, or anal sex without a condom; (vii) number of days in the last 30 that the respondent's mental health was not good; (viii) number of days in the last 30 that the respondent's poor mental or physical health kept the respondent from usual activities, such as self-care, work, or recreation; (ix) marital status other than married or member of unmarried couple; (x) firearms kept in or around one's home; and (xi) the presence of a loaded firearm around one's home. Of these, we viewed alcohol intake, poor mental health and disability, unmarried marital status, and firearm access as relatively established risk factors (33,34,35). We included HIV-related behavior, sunburn, and seatbelt usage as proxies for risk taking and impulsivity, constructs themselves related to suicide (3,36,37,38,39).

Other behavioral characteristics

We grouped state of residence into US Census regions (East, Midwest, South, West, and US territories). Race was categorized into four groups (non-Hispanic white, non-Hispanic black, Hispanic, multiracial/other). Educational attainment was categorized as less than high school, high school graduate, some college, and college graduate. Smoking was categorized as never, former, current less-than-daily, and current daily.

Statistical analyses

We performed regression analyses to determine the prevalence ratios for various risk factors for suicide according to BMI. In all models, we adjusted for age as linear and quadratic terms, race, smoking, education, and region—factors we viewed as likely “upstream” of BMI. Individuals missing these covariates or BMI were excluded, leaving 224,247 participants potentially available for analysis in 2002 and 275,194 in 2004. We then assessed the associations of BMI with individual risk factors in the full populations both adjusted for and stratified by sex. We performed logistic regression in most cases, and analyzed days with poor mental health or limited activities and unmarried marital status, firearm possession, and sunburn (which were common outcomes) with Poisson regression (40). In all cases, we used Intercooled STATA 8.2 for Windows (StataCorp, College Station, TX, 2005) to account for the sampling weights and presented weighted prevalence ratios or β-coefficients (for count variables) with their 95% confidence intervals.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgment
  8. Disclosure
  9. REFERENCES

Selected characteristics of the BRFSS populations in 2002 and 2004 are shown in Table 1. Unmarried marital status and presence of a firearm in the home were the most common risk factors, each prevalent in some 35% of respondents. A minority of participants reported a BMI >30, with a shift toward heavier BMI levels between 2002 and 2004. In 2002, mean BMI levels were 27.2 among white men, 25.8 among white women, 28.0 among black men, and 28.8 among black women. In 2004, the corresponding mean BMI levels were 27.4, 26.0, 27.9, and 29.1.

Table 1.  Weighted prevalence of selected characteristics (and 95% confidence intervals) among BRFSS respondents in 2002 and 2004
inline image

Tables 24 show potential risk factors related to suicide according to BMI in 2002 and 2004. In general, the relationships were consistent across the two survey years.

Table 2.  Adjusted weighted prevalence ratios and 95% confidence intervals for alcohol-related risk factors among BRFSS respondents according to BMI
inline image
Table 4.  Adjusted weighted differences in number of days with poor mental health or activity limitation according to BMI
inline image

Table 2 presents prevalence ratios for alcohol-related characteristics according to BMI. Alcohol-related risk behaviors tended to be less frequent with heavier BMI, but this finding was essentially limited to women.

Table 3 presents corresponding prevalence ratios for other characteristics potentially related to suicide. Infrequent seat belt use was consistently higher with heavier BMI in both sexes. Firearm and loaded firearm presence were generally higher with heavier BMI among men but less clearly so among women. The relationships of HIV-related behaviors with BMI were generally weak, with no consistent associations across sexes and years. Unmarried marital status was generally more prevalent with heavier BMI among women but not men. There were weak positive associations of sunburn with greater BMI, albeit slightly stronger in men.

Table 3.  Adjusted weighted prevalence ratios and 95% confidence intervals for risk taking, marital, and firearm-related risk factors among BRFSS respondents according to BMI
inline image

Table 4 presents the number of days that respondents reported poor mental health or activity limitation in the last 30 according to BMI. Extremes of BMI were associated with greater number of days for both variables in both sexes, but overweight and Class I obesity were associated with greater number of days only among women.

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgment
  8. Disclosure
  9. REFERENCES

In this population-based study of some 500,000 American adults, heavier BMI was variably associated with potential risk factors for suicide, with some differences noted by sex. Alcohol-related risk tended to be lower with heavier BMI among women, whereas firearm-related risk tended to be higher among men. Heavier BMI also tended to be associated with unmarried status and poor mental health most notably among women. No potential risk factors were noted to be consistently inversely associated with BMI in both sexes and years.

In general, previous studies support an inverse relationship of BMI with risk of suicide death, but a positive relationship with suicide attempts. Several cohort studies have found an inverse relationship of BMI and risk of suicide (17,21,23), particularly among Scandinavian populations (18,19,20,22) (who have among the highest rates of suicide worldwide). A recent analysis of data from a series of National Health Interview Survey administrations also found an inverse relationship of BMI and suicide albeit without adjustment for smoking (24). It is interesting that the sex and race groups observed elsewhere to have the highest risk of suicide (whites and men) also tended to have lower average levels of BMI (41), but the association of BMI with lower risk of suicide death appears to be consistent in both men and women (19,24) and in racially homogeneous populations.

In contrast, suicide attempts tended to be positively associated with BMI in several samples worldwide (42,43,44). Dong and colleagues also found positive associations of extreme obesity with suicide attempts in two parallel studies (45). However, the relationship of BMI with suicide attempts and depression appears to be modified by sex, much as it was in our findings. For example, in an analysis of the 1992 National Longitudinal Alcohol Epidemiologic Survey, Carpenter and colleagues found that a 10-kg/m2 increment in BMI was associated with increased odds of suicide attempts among women, but significantly decreased odds among men (46). Several other cross-sectional studies have suggested that men and women differ in their relationship between BMI and depressive symptoms (14,16,47).

The paradoxical association of heavier BMI with lower risk of suicide despite mixed or even positive associations with suicide attempts and depressive symptoms suggests that BMI may act “downstream” of suicidal ideation or attempts in preventing suicide (much as male sex may increase completed suicides by promoting use of highly lethal mechanisms like firearms). If this observation is correct, then studies of the mechanisms involved in the BMI-suicide relationship may help to identify potentially modifiable risk factors for suicide late in the causal chain. To date, the most successful of these has been restricted access to highly lethal means of suicide (48,49).

A few potentially important findings related to BMI and suicides are suggested by this study. First, we did not find any potential mechanisms that were consistently lower with heavier BMI in both sexes. This suggests that there are likely to be other mechanisms for the observed inverse relationship of BMI and suicide that require additional study, including metabolic factors related to obesity. For example, low central nervous serotonergic activity has been implicated in impulsivity and suicide (50,51,52), and both dietary cholesterol restriction of cholesterol (26) and insulin resistance (by impairing tryptophan transport) (53) may reduce such activity. Likewise, we encourage suicide surveillance systems, such as the National Violent Death Reporting System (54), to include height and weight as collected data elements in an effort to address the relationships of BMI with impulsivity and other contextual features of suicide.

Also, BMI was associated with higher rates of some factors apt to be associated with higher suicide risk, such as presence of a firearm in or near the home among men and infrequent seat belt use (a marker of risk taking previously linked with heavier BMI (55)). To the degree that these factors have not been accounted for in previous studies of BMI and suicide, the true relationship of BMI and suicide may be even more striking than previously suggested, at least in studies from the United States and similar locations where firearms are a common means of suicide.

Finally, there was some evidence that one potential mechanism for the association of heavier BMI with lower risk could be a lower risk of problem drinking (and perhaps other forms of substance abuse) in women. Indeed, in a nationally representative sample of US adults, heavier BMI tended to be associated positively with several mental health disorders, including panic disorder and major depression, but was inversely associated with substance use disorder (10). Although our results did not confirm a relationship of heavier BMI with less frequent problem drinking in men, these results do leave open the possibility that use of other substances (which we could not assess) is less frequent with heavier BMI and may contribute to its relationship with lower risk of suicide.

The BRFSS is subject to both strengths and limitations. Its strengths include an extremely large and nationally representative sample of adults providing anonymous information on a number of sensitive behaviors not collected in other surveys. In addition, the iterative nature of the BRFSS allowed us to validate our findings in two separate surveys, providing important confirmation of our findings. The range of questions included in the BRFSS is large and diverse, providing insight into a number of characteristics potentially related to suicide.

On the other hand, several limitations should be recognized. First, we did not have information on suicide attempts or ideation, which would have provided useful additional information on the BMI-suicide relationship. The BRFSS also has limited information on alcohol abuse or dependence and no information on use of other substances. Second, the BRFSS is necessarily limited to self-reported information, although it is difficult to determine how more objective information could be collected from a sample of the size and scope of the BRFSS. Nonetheless, information on alcohol consumption in particular appears to be reasonably reliably collected in this survey (56). Likewise, self-reports of height and weight in other surveys tend to rank order individuals validly (57), although with a systematic bias toward underreporting of weight that increases at greater levels of BMI (58). Thus, true BMI is likely one or more units higher than reported by BRFSS participants, particularly those with obesity. Third, the BRFSS is a cross-sectional survey, and hence neither directionality nor causality can be inferred from this study. Longitudinal studies that include measurement of both BMI and risk factors like those included here would be an important next step in clarifying these relationships. Fourth, several risk factors we studied as proxies for risk taking or impulsivity are only indirectly related to suicide risk, such as sunburn, or may be associated with BMI for other reasons, such as discomfort with seatbelt position with greater adiposity.

In conclusion, heavier BMI was not consistently associated with a lower prevalence of risk factors for suicide in this study. Given the strength of the relationship between BMI and suicide, continued research to identify its mechanisms could lead to novel insights into prevention of this important cause of death.

Acknowledgment

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgment
  8. Disclosure
  9. REFERENCES

Both of the authors participated in the conception and design of the study, analysis and/or interpretation of data, critical review and revision of the manuscript, and provision of statistical or content expertise. Dr Mukamal had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

REFERENCES

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgment
  8. Disclosure
  9. REFERENCES