August 2011  << Back  

  Can't open the PDF? Click here for help.

 

 RESEARCH ARTICLE

Lack of replication of the association of low serum cholesterol and attempted suicide in another country raises more questions

Jose de Leon, MD

Mental Health Research Center, Eastern State Hospital, Lexington, KY, USA

Paula Mallory, RD

Dietetic Service, Eastern State Hospital, Lexington, KY, USA

Lorraine Maw, MA

Mental Health Research Center, Eastern State Hospital, Lexington, KY, USA

Margaret T. Susce, RN, MLT

Mental Health Research Center, Eastern State Hospital, Lexington, KY, USA

M. Mercedes Perez-Rodriguez, MD

Department of Psychiatry, Mount Sinai School of Medicine, New York, NY, USA

Enrique Baca-Garcia, MD

Department of Psychiatry, Fundacion Jimenez Diaz University Hospital, Autonoma University of Madrid, Madrid, Spain, Department of Psychiatry, Columbia University College of Physicians and Surgeons, New York, NY, USA

BACKGROUND: In 2 Spanish case-control studies, low cholesterol levels in males were consistently associated with suicide attempts.

METHODS: This US study tried to replicate the association between low cholesterol levels and suicide attempts, using a case-control design to study all patients admitted to Eastern State Hospital in Lexington, Kentucky, during a 1-year period. Psychiatric patients who had currently attempted suicide were studied as cases, and psychiatric patients who had not currently attempted suicide served as controls. A fasting serum total cholesterol <160 mg/dL was considered a possible risk factor for suicide. Logistic regression provided an adjusted estimate of the univariate odds ratios (ORs) for confounding factors.

RESULTS: There were 193 current suicide attempters (cases) and 1091 non–current suicide attempters (controls). In the total sample logistic regression model, low cholesterol levels were significantly associated with lower risk of current suicide attempt (OR, 0.60; confidence interval (CI), 0.39 to 0.92) after adjusting for confounding variables. After sex stratification, low cholesterol levels were significant only among men (OR, 0.47; CI, 0.26 to 0.86). This US study did not replicate our prior Spanish findings; to the contrary, low cholesterol levels were not associated with increased suicide risk but with a decreased risk in US men.

CONCLUSIONS: It is possible cholesterol abnormalities and low body mass index may be markers of suicide risk, particularly in some male patients.

KEYWORDS: alcohol-related disorders, body mass index, cholesterol, depression, schizophrenia, suicide

ANNALS OF CLINICAL PSYCHIATRY 2011;23(3):163-170

  INTRODUCTION

Low cholesterol and attempted suicide

More than 30 epidemiologic or clinical studies have shown an association between low total serum cholesterol levels and increased suicide attempts.1,2 The mechanistic hypothesis that explains this association is cholesterol depletion reduces central serotonergic activity,3,4 and low serotonin activity is usually associated with risk of suicide attempt.5

Not all studies have replicated this association; some studies found the opposite6 or negative results.7 Papakostas et al8 have suggested both increases and decreases in neuronal membrane cholesterol and cholesterol content can lead to alterations in serotonergic function.

The lack of replication in all studies of the association between low cholesterol levels and suicide attempts may be explained by the lack of homogeneity among suicide attempters and among controls, and by differences in the characteristics of suicide attempts.9 Regarding controls, the majority of studies have compared groups of suicide attempters with healthy controls without psychiatric disorders. Because the majority of suicide attempters have a co-occurring psychiatric disorder,10 one can argue a study showing differences between suicide attempters and controls with no psychiatric disorders is contaminated by the potential confounding effects of psychiatric disorders on cholesterol levels. Although psychiatric medications—particularly antipsychotics—are associated with high, but not low, cholesterol levels, it cannot be ruled out that these medications may act as confounding factors. Therefore, psychiatric patients without suicide attempts appear to be better controls for suicide studies than are the general population.9

Another issue that has not received enough attention in the literature is the crucial role of patients’ sex. Metabolic syndrome11,12 and total cholesterol levels in psychiatric patients both may be influenced by sex.13 In a first case-control study in Spain, we compared serum cholesterol levels in suicide attempters (68 male and 109 female cases) and healthy volunteers (controls) matched by sex, age, and body mass index (BMI).14 We found serum cholesterol levels were significantly lower among suicide attempters than among controls, although the difference remained significant only among men after sex stratification. Matching by sex and BMI allowed us to eliminate some confounding factors regarding low cholesterol levels, but our first Spanish study still had a weakness: the control group had no psychiatric disorders.

Our second case-control study in Spain used healthy and psychiatric controls and controlled for sex, age, low BMI, alcohol and nicotine use, and aggressive and impulsive behaviors.9 Two types of odds ratios (ORs) were calculated to measure the association between attempted suicide and low cholesterol level (defined as serum cholesterol <160 mg/dL). First, uncorrected ORs were calculated. Then, ORs corrected for confounding variables were estimated. The majority of these 2 types of ORs were between 1.5 and 2.0. However, the study was somewhat limited by the relatively small sample size; some of these ORs reached significance, whereas others did not. The sample size became smaller after sex stratification (138 male suicide attempters, 68 male psychiatric controls, and 220 male healthy controls; 279 female suicide attempters, 87 female psychiatric controls, and 138 female healthy controls). After sex stratification, we replicated the finding that low cholesterol was significantly associated with male suicide attempters (OR, 2.0; 95% confidence interval [CI], 0.99 to 4.1) when compared with male healthy controls. This was the only finding replicated in both Spanish studies.

Close attention to the literature suggests, curiously enough, the association between low cholesterol level and suicidal behaviors is more frequent among men than women. Golier et al15 observed men with cholesterol levels above the 25th percentile were less likely to have made a serious suicide attempt than men with low cholesterol levels when age, weight, race, socioeconomic status, alcohol use, and depression were controlled as confounding factors. However, they found no association between cholesterol level and attempted suicide in women. Bocchetta et al16 reported the proportions of men with a history of prior suicide attempt, especially if violent, and of men with a history of completed suicide in a first-degree relative were significantly higher among those men with cholesterol levels in the lowest quartile. They could not find any significant association among women. Soeda et al17 reported low cholesterol increased the likelihood of having a score higher than 3 in the General Health Questionnaire, suggesting poorer mental health, but only in men. Other prior studies on the association between sex and serum cholesterol have been inconclusive; some authors have observed male sex is associated with lower cholesterol levels in several psychiatric disorders.18,19

Low BMI and attempted suicide

Another interesting finding in our second Spanish study9 was low BMI was associated with attempted suicide when male attempters were compared with healthy controls (OR, 3.3; CI, 1.5 to 7.2). However, this is not a new finding. Magnusson et al20 performed a prospective cohort study of more than 1 million conscripted Swedish men. A strong inverse association was found between BMI and suicide. The association was similar when subjects with a mental disorder at baseline were excluded from the analysis. The BMI-suicide association was independent of the length of follow-up, indicating weight loss as a consequence of mental illness did not explain the BMI-suicide association. Magnusson et al20 suggested factors influencing BMI may be causally implicated in the etiology of mental disorders leading to suicide. In a prospective cohort study of US men, risk of death from suicide was strongly inversely related to BMI, but not to height or physical activity. A nonprospective US study did not find this association between BMI and suicide.21 Osler et al22 studied conscripted Danish men born in Copenhagen in 1953. Low BMI, low cognitive test score, and mental disorder at age 18 were associated with an increased risk of suicide. After adjustment, mental disorder remained significantly associated with suicide risk, whereas the estimates for cognitive function and BMI were attenuated, suggesting some of the incidence of low BMI may be explained by mental illness.

In summary, the literature suggests low BMI at a young age is associated with increased suicide risk only in men, and low cholesterol may also be more consistently associated with suicide attempts among men than women.

This US study tried to replicate the association between low cholesterol level and attempted suicide, using a case-control design to study all patients admitted to a state hospital in the course of 1 year. Psychiatric patients who had currently attempted suicide (defined as a suicide attempt that led to the patient admission reviewed in this study) were studied as cases, and psychiatric patients who had not currently attempted suicide served as controls.

  METHODS

Subjects

Eastern State Hospital is located in Lexington, Kentucky, and serves as the primary acute psychiatric hospital for central and northern Kentucky, covering approximately one-third of the state. Patients are routinely assessed for recent and past history of attempted suicide, weight, and height by the admissions nursing staff. All cooperating patients are participants in basic laboratory studies, including a study of fasting serum total cholesterol level. Fasting refers to a lack of food beginning the prior evening and lasting until after laboratory tests are performed in the early morning. On the approval of 2 institutional review boards, the charts of all 1622 patients admitted in 1995 were reviewed (no study consent form was used). For those patients having more than one admission, the first admission with a complete data set was used, for a total of 1284 patients. Of the original 1622 patients, 183 (11%) were excluded because no cholesterol level was recorded, another 54 (3%) were excluded because no height or weight was recorded, and 101 had no suicide assessment recorded (6%).

The clinical characteristics of the remaining 1284 patients (79% of 1622), including DSM-IV clinical diagnoses obtained from the charts, are the focus of the statistical analyses described in TABLE 1. As in our first14 and second9 Spanish studies and the prior literature,8,23 a serum cholesterol level <160 mg/dL was considered a possible risk factor for suicide attempt. Based on prior information from the US general population,24 an alternative measure—serum cholesterol cut score lower than the 25th percentile and matched by sex, age, and race—was also explored.

As in our second Spanish study,9 BMI ≤22 was defined as low, and young age (<35) was used to control the association between age and low cholesterol. In the Spanish sample, 2 psychiatric diagnoses—major depression and schizophrenic psychoses—were associated with suicide attempts. These 2 diagnoses were considered potential independent variables in this US sample. As this sample was much larger than the Spanish sample, other psychiatric disorders such as substance use disorder or the association between substance use disorders and mood disorders were considered potentially confounding variables (described in TABLE 1). Drug use disorder was defined as not including nicotine dependence, because it was not usually considered for diagnosis in the chart. Thus, the lack of information on tobacco smoking is a limitation. However, in the Spanish study, smoking was not significantly associated with suicide attempts when using psychiatric controls; it was only significant when using healthy controls.9


TABLE 1

Characteristics of patients admitted to a state hospital

Patient characteristic Males (n = 769) Females (n = 515) Total (N = 1284)
Young age (<35 y), n (%) 310 (40%) 175 (34%) 485 (38%)
Race, n (%)
  White 692 (90%) 459 (89%) 1151 (90%)
  African American 69 (9%) 52 (10%) 121 (9%)
  Other 8 (1%) 4 (1%) 12 (1%)
Suicide attempt, n (%)
  Evera 336 (44%) 253 (49%) 589 (46%)
  Currentb 111 (14%) 82 (16%) 193 (15%)
Cholesterol, n (%)
  <160 ng/dL 177 (23%) 89 (17%) 266 (21%)
  <25th percentile 265 (34%) 108 (21%) 373 (29%)
BMI, n (%)
Low BMI (<22.0 kg/m2), n (%) 203 (26%) 126 (25%) 329 (26%)
DSM-IV diagnosis, n (%)
  Major depression 85 (11%) 68 (13%) 153 (12%)
  Depressive disorder NOS 61 (8%) 51 (10%) 112 (9%)
  Any mood disorder 215 (28%) 208 (40%) 423 (33%)
  Schizophrenia psychosesc 225 (29%) 124 (24%) 349 (27%)
  Any alcohol use disorder 297 (39%) 97 (19%) 394 (31%)
  Any drug use disorder 226 (29%) 91 (18%) 317 (25%)
  Any substance use disorder 402 (52%) 151 (29%) 553 (43%)
  Major depression and SUD 35 (5%) 21 (4%) 56 (4%)
  Depressive disorder NOS and SUD 34 (4%) 18 (4%) 52 (4%)
  Any mood disorder and SUD 92 (12%) 47 (9%) 139 (11%)
Age, y, mean (SD) 38.6 (12.4) 40.6 (13.2) 39.3 (12.8)
Cholesterol, ng/dL, mean (SD) 193.7 (47.4) 197.2 (43.9) 195.1 (46.1)
BMI, kg/m2, mean (SD) 25.9 (5.9) 28.2 (7.9) 26.8 (6.9)
aSuicide attempt that led to a patient admission reviewed in this study.
bAny previous or current suicide attempt.
cSchizophrenia psychoses include DSM-IV schizophrenia and schizoaffective disorder.
BMI: body mass index; NOS: not otherwise specified; SUD: substance use disorder.
Statistical analysis

Statistical analyses were conducted using the Statistical Package for the Social Sciences (SPSS) version 15.0. As in the second Spanish study,9 univariate ORs and 95% CIs were used to determine whether current or past suicide attempts represented the better independent variable. ORs were also used to explore which definition of low cholesterol should be utilized: <160 mg/dL, as in the second Spanish study,9 or <25th percentile of the matched group from the general population.24 In the latter case, data from the US general population24 were used to establish whether the patient ranked lower than the 25th percentile of the group comprising the same sex, age, and race.

Multivariate logistic regression allows univariate ORs to be adjusted for confounding factors. The plan followed the second Spanish study,9 in significant univariate ORs reflecting the association between suicide attempt and low cholesterol level would be adjusted according to the dichotomous independent variables found to be significant. Possible significant variables were low BMI, young age, and clinical diagnosis.9 All clinical diagnoses tested in the statistical models are described in TABLE 1 (bipolar disorder was not significant by itself; it cannot be ruled out this was due to the limited sample size). A backward stepwise procedure was used to select those significant variables in the final logistic regression models. The Hosmer-Lemeshow goodness-of-fit test was used to examine the fitness of the logistic regression models. As in the second Spanish study, logistic regression models were planned in the total sample and then after sex stratification.9 The use of backward stepwise procedures may present problems in selecting between collinear variables, such as low cholesterol defined as <160 mg/dL and low cholesterol defined as <25th percentile of the matched group from the US general population. However, once the backward procedure eliminated one of the variables from the model, the excluded variable no longer influenced the remaining model.

  RESULTS

Sample description

TABLE 1 describes the clinical characteristics of the total sample and the sample after sex stratification. Of the 193 current attempters (cases) and 1091 non–current attempters (controls), 111 were male current attempters (cases) and 658 were male non–current attempters (controls), 82 were female current attempters (cases), and 433 were female non–current attempters (controls).

Univariate OR for the association between low cholesterol level and suicide attempt. In a cross-tabulation analysis, low cholesterol level, defined as <160 mg/dL, was significantly associated with the presence of a current suicide attempt (OR, 0.64; CI, 0.42 to 0.97; χ2 = 4.5; df = 1; P = .034) but was not significantly associated with the presence of ever having had a suicide attempt (χ2 = 1.2; df = 1; P = .27). Thus, the presence or absence of a current suicide attempt was selected as the dependent variable for the logistic regression models.

In the United States, data from the general population allowed an alternative matched definition (cholesterol <25th percentile of the group matched by sex, age, and race). The OR between current suicide attempt and the matched lowest quartile also provided a significant univariate OR (0.62; CI, 0.43 to 0.89; χ2 = 6.7; df = 1; P = .01). As both definitions of low cholesterol provided significant ORs, we decided to include both definitions in the logistic regression model and leave the backward stepwise procedure to select which definition of low cholesterol was more appropriate after adjusting for confounding variables.

Logistic regression model in the total sample using current suicide attempt as the dependent variable. In the total sample, presence or absence of suicide attempt was the dependent variable in the logistic regression model. The potential independent variables were low cholesterol, defined either as <160 mg/dL or as <25th percentile of the matched sample in the general population; sex; young age; low BMI; and the clinical diagnoses (TABLE 1). The backward stepwise procedure began with all of these variables and progressively eliminated nonsignificant variables. TABLE 2 describes the final model. Low cholesterol (<160 mg/dL) was significantly associated with current suicide attempt (OR, 0.60; CI, 0.39 to 0.92).

As in the second Spanish study,9 logistic regression models were developed after sex stratification (TABLE 2). Low cholesterol (<160 mg/dL) was significantly associated with male sex (OR, 1.4; CI, 1.1 to 1.9; χ2 = 6.2; df = 1; P = .013), indicating stratification by sex is prudent, because sex may influence the frequency of low cholesterol levels.

The male logistic regression model indicated that, after adjusting for confounding variables, low cholesterol (<160 mg/dL) was significantly associated with current suicide attempt (OR, 0.47; CI, 0.26 to 0.86). Low cholesterol (<160 mg/dL) was not significantly associated with current suicide attempt in the female model. In fact, the univariate OR measuring the association between low cholesterol (<160 mg/dL) and current suicide attempt in women was far from significant (OR, 0.88; CI, 0.53 to 1.8; χ2 = .003; df = 1; P = .96).

Variables associated with current suicide attempt in patients with low cholesterol (<160 mg/dL). There were 266 patients with low cholesterol levels (<160 mg/dL). In this sample of 266 patients, low BMI was significantly associated with current suicide attempt (OR, 2.2; CI, 1.0 to 4.7; χ2 = 3.9; df = 1; P = .049). This significant effect was mainly explained by the male sample. In these 175 men, the OR was 4.8 (CI, 1.5 to 15.7; χ2 = 6.7; df = 1; P = .01). In fact, the only variable that was significantly associated with current suicide attempt in these 175 male patients with low cholesterol was low BMI.


TABLE 2

Logistic regression analysis comparing patients with a current suicide attempt vs psychiatric controls

Variables OR 95% CI Wald χ 2 test
(all df = 1)
Totala (N = 1284)
  Low cholesterolb 0.60 0.39 to 0.92 5.5 (P = .02)
  Young agec 1.8 1.3 to 2.5 12.1 (P < .001)
  Alcohol use disorder 1.4 1.0 to 2.0 3.9 (P = .048)
  Depressive disorder NOS and SUD 1.9 1.0 to 3.8 3.9 (P = .05)
  Schizophrenia psychosesd 0.53 0.33 to 0.84 7.2 (P = .007)
  Major depression 3.0 2.0 to 4.4 27.2 (P < .001)
Malese (n = 769)
  Low cholesterolb 0.47 0.26 to 0.86 6.1 (P = .013)
  Young agec 1.7 1.1 to 2.6 5.9 (P = .015)
  Alcohol use disorder 1.8 1.1 to 2.7 6.6 (P = .01)
  Schizophrenia psychosesd 0.44 0.23 to 0.82 6.6 (P = .01)
  Major depression 3.5 2.0 to 6.0 20.6 (P < .001)
Femalesf (n = 515)
  Young agec 2.0 1.2 to 3.2 9.8 (P = .002)
  Depressive disorder NOS 2.0 0.99 to 4.1 3.7 (P = .055)
  Major depression 2.7 1.4 to 4.9 9.8 (P = .002)
aThe logistic regression analysis selected variables using a backward selection procedure, and only significant or close to significant variables (around P = .05) are described.
Hosmer-Lemeshow χ2 = 3.7; df = 8; P = .88.
bLow serum cholesterol: <160 mg/dL.
cYoung age: <35 years.
dSchizophrenia psychoses include DSM-IV schizophrenia and schizoaffective disorder.
eThe logistic regression analysis selected variables using a backward selection procedure, and only significant or close to significant variables are described.
Hosmer-Lemeshow χ2 = 3.8; df = 7; P = .80.
fThe logistic regression analysis selected variables using a backward selection procedure, and only significant or close to significant variables are described.
Hosmer-Lemeshow χ2 = 1.0; df = 3; P = .81.
CI: confidence interval; NOS: not otherwise specified; OR: odds ratio; SUD: substance use disorder.

  DISCUSSION

Limitations

Prospective studies or controlled trials would be more appropriate for exploring the effects of low cholesterol in attempted suicide but are much more difficult and expensive to perform. In this naturalistic clinical study, all the assessments reflected usual clinical practice, with their benefits and limitations. There were some missing data, but the sample still had remarkable size and representation. Despite the clinical nature of this study, we were pleasantly surprised some of the clinical characteristics significantly associated with suicide attempt were very similar to those of the second Spanish study, which used a research design. The OR for young age was 1.8 (CI, 1.1 to 2.8) in the Spanish sample vs 1.8 (CI, 1.3 to 2.5) in the US sample. The OR for major depression was 2.7 (CI, 2.0 to 4.4) in the Spanish sample vs 3.0 (CI, 2.0 to 4.4) in the US sample. The OR for schizophrenic psychosis was 0.32 (CI, 0.19 to 0.54) in the Spanish sample vs 0.53 (CI, 0.33 to 0.84) in the US sample. Therefore, these ORs were remarkably similar in the Spanish and US samples, despite different recruitment and assessment methods. The total US sample, which was larger, included in the logistic regression model alcoholism and the association between depression and substance use disorder. These additional results in the US sample are compatible with the literature that suggests alcoholism25,26 and the combination of depression and substance use25-27 are probably important risk factors for suicide attempts. In summary, the robustness of the clinical findings and great similarities in the clinical findings between the Spanish and US samples suggest despite the limitations of the clinical design of the US study, the results may be reliable and trustworthy. In the US sample, there was no information on the type of suicide attempt; violent suicide attempts have been associated with major depression.28 The Spanish and the US studies shared the limitations of not controlling for the effects of medications, and for possible weight or appetite changes associated with psychiatric illness. The US study did not control for smoking, but in the Spanish sample, smoking was not associated with suicide attempts when psychiatric patients served as controls.

Lack of replication of the prior results in Spain

Our first Spanish study14 examined cholesterol levels in current suicide attempters vs matched controls. Our second study9 tested the association between low cholesterol levels (<160 mg/dL) and current or ever suicide attempts. ORs for both current and ever suicide attempts showed a significant association of low cholesterol level with increased risk for suicide attempt. In the US study, low cholesterol level was not associated with an increased risk of attempted suicide but, rather, with a decreased risk in male subjects.

A possible explanation is both increases and decreases in neuronal membrane cholesterol and cholesterol content can lead to alterations in serotonergic function.8 An alternative explanation for the lack of consistency in our Spanish and US studies and in the literature in general is cholesterol levels and BMI may be indirect markers of other lipid or metabolic abnormalities. Other lipids that may be important in suicide include omega-329 or any of the various apolipoproteins.30-33

This study and the literature suggest cholesterol abnormalities and low BMI are more consistently found among male suicide attempters than female attempters. This has been true in both of our studies in Spain and also in the US study. Moreover, the large US study suggests low BMI and low cholesterol may be associated with suicide attempts in men. It is possible in men, low cholesterol and/or low BMI may be markers for some other biologic abnormalities associated with increased risk. We suspect these abnormalities may not be present in all patients but may be markers for subgroups of patients. Thus, only larger samples may identify them. In future studies, we will explore other possible peculiarities that may identify male suicide attempters with unusual metabolic profiles (low BMI and/or low cholesterol level). The recent attempts to conduct comprehensive measures of multiple lipid compounds in blood, called metabolomics,34 may be crucial in exploring the hypothesis that some men at risk for attempted suicide may be characterized by unusual lipid profiles.

  CONCLUSIONS

This large US clinical study did not replicate the findings from our 2 prior research studies in Spain. Unlike the Spanish studies, low cholesterol was not associated with an increased risk of suicide but, rather, with a decreased risk among US men, when other psychiatric patients served as controls. This lack of replication raises more questions. It is possible cholesterol abnormalities and low BMI may be markers of suicide risk, particularly in some male patients. Future studies will need to explore whether cholesterol abnormalities and/or low BMI may be markers of some other lipid or metabolic abnormality more directly associated with vulnerability to suicide among men.

ACKNOWLEDGEMENT: The authors are grateful to the Eastern State Hospital staff; their work made this study possible. This study was supported by internal resources. Drs. Baca-Garcia and Perez-Rodriguez were supported by grants from the state of Madrid (Comunidad de Madrid CAM 08.5/0003/1999) and the Spanish government (Fondo de Investigación Sanitaria [FIS] PI060092, PI060497, Instituto de Salud Carlos III, CIBERSAM).

DISCLOSURE: The authors report no financial relationship with any company whose products are mentioned in this article or with manufacturers of competing products.

    REFERENCES

  1. Lester D. Serum cholesterol levels and suicide: a meta-analysis. Suicide Life Threat Behav. 2002;32:333–346.
  2. Muldoon MF, Manuck SB, Matthews KA. Lowering cholesterol concentrations and mortality: a quantitative review of primary prevention trials. BMJ. 1990;301:309–314.
  3. Dursun SM, Reveley MA. Low serum cholesterol and serotonin receptor subtypes. Br J Psychiatry. 1993;163:417–418.
  4. Engelberg H. Low serum cholesterol and suicide. Lancet. 1992;339:727–729.
  5. Mann JJ. Neurobiology of suicidal behaviour. Nat Rev Neurosci. 2003;4:819–828.
  6. Brunner J, Bronisch T, Pfister H, et al. High cholesterol, triglycerides, and body-mass index in suicide attempters. Arch Suicide Res. 2006;10:1–9.
  7. Fiedorowicz JG, Coryell WH. Cholesterol and suicide attempts: a prospective study of depressed inpatients. Psychiatry Res. 2007;152:11–20.
  8. Papakostas GI, Petersen T, Mischoulon D, et al. Serum cholesterol and serotonergic function in major depressive disorder. Psychiatry Res. 2003;118:137–145.
  9. Perez-Rodriguez MM, Baca-Garcia E, Diaz-Sastre C, et al. Low serum cholesterol may be associated with suicide attempt history. J Clin Psychiatry. 2008;69:1920–1927.
  10. Arsenault-Lapierre G, Kim C, Turecki G. Psychiatric diagnoses in 3275 suicides: a meta-analysis. BMC Psychiatry. 2004;4:37.
  11. Ford ES, Giles WH, Dietz WH. Prevalence of the metabolic syndrome among US adults: findings from the third National Health and Nutrition Examination Survey. JAMA. 2002;287:356–359.
  12. Susce MT, Villanueva N, Diaz FJ, et al. Obesity and associated complications in patients with severe mental illnesses: a cross sectional survey. J Clin Psychiatry. 2005;66:167–173.
  13. de Leon J, Susce MT, Johnson M, et al. A clinical study of the association of antipsychotics with hyperlipidemia. Schizophr Res. 2007;92:95–102.
  14. Diaz-Sastre C, Baca-Garcia E, Perez-Rodriguez MM, et al. 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. 2007;31:901–905.
  15. Golier JA, Marzuk PM, Leon AC, et al. Low serum cholesterol level and attempted suicide. Am J Psychiatry. 1995;152:419–423.
  16. Bocchetta A, Chillotti C, Carboni G, et al. Association of personal and familial suicide risk with low serum cholesterol concentration in male lithium patients. Acta Psychiatr Scand. 2001;104:37–41.
  17. Soeda S, Terao T, Iwata N, et al. Mental effect of cholesterol in males: protective effect? J Affect Disord. 2006;91:139–144.
  18. Borgherini G, Dorz S, Conforti D, et al. Serum cholesterol and psychological distress in hospitalized depressed patients. Acta Psychiatr Scand. 2002;105:149–152.
  19. New AS, Sevin EM, Mitropoulou V, et al. Serum cholesterol and impulsivity in personality disorders. Psychiatry Res. 1999;85:145–150.
  20. Magnusson PK, Rasmussen F, Lawlor DA, et al. Association of body mass index with suicide mortality: a prospective cohort study of more than one million men. Am J Epidemiol. 2006;163:1–8.
  21. Stack S, Lester D. Body mass and suicide risk. Crisis. 2007;28:46–47.
  22. Osler M, Nybo Andersen AM, Nordentoft M. Impaired childhood development and suicidal behaviour in a cohort of Danish men born in 1953. J Epidemiol Community Health. 2008;62:23–28.
  23. Kim YK, Myint AM. Clinical application of low serum cholesterol as an indicator for suicide risk in major depression. J Affect Disord. 2004;81:161–166.
  24.  National Heart, Lung, and Blood Institute. The Lipid Research Clinics Population Studies data book. Bethesda MD: Department of Health and Human Services; 1980. NIH Publication No. 80–1527.
  25. Sher L. Alcoholism and suicidal behavior: a clinical overview. Acta Psychiatr Scand. 2006;113:13–22.
  26. Sher L. Risk and protective factors for suicide in patients with alcoholism. Scientific World Journal. 2006;6:1405–1411.
  27. Sher L, Stanley BH, Harkavy-Friedman JM, et al. Depressed patients with co-occurring alcohol use disorders: a unique patient population. J Clin Psychiatry. 2008;69:907–915.
  28. Astruc B, Torres S, Jollant F, et al. A history of major depressive disorder influences intent to die in violent suicide attempters. J Clin Psychiatry. 2004;65:690–695.
  29. Zhang J, Li Y, Torres ME. How does a suicide attempter eat differently from others? Comparison of macronutrient intakes. Nutrition. 2005;21:711–717.
  30. Baca-Garcia E, Diaz-Sastre C, García-Resa E, et al. Lack of association between plasma apolipoprotein E and suicide attempts. J Clin Psychiatry. 2004;65:580.
  31. Edgar PF, Hooper AJ, Poa NR, et al. Violent behavior associated with hypocholesterolemia due to a novel APOB gene mutation. Mol Psychiatry. 2007;12:258–263.
  32. Gietl A, Giegling I, Hartmann AM, et al. ABCG1 gene variants in suicidal behavior and aggression-related traits. Eur Neuropsychopharmacol. 2007;17:410–416.
  33. Hwang JP, Yang CH, Hong CJ, et al. Association of APOE genetic polymorphism with cognitive function and suicide history in geriatric depression. Dement Geriatr Cogn Disord. 2006;22:334–338.
  34. Kaddurah-Daouk R, Kristal BS, Weinshilboum RM. Metabolomics: a global biochemical approach to drug response and disease. Annu Rev Pharmacol Toxicol. 2008;48:653–683.

CORRESPONDENCE: Jose de Leon, MD Mental Health Research Center, Eastern State Hospital, 627 West Fourth Street, Lexington, KY 40508 USA, E-MAIL: jdeleon@uky.edu