It has been hypothesized that low serum total cholesterol may result in an increased risk of traumatic death, especially suicide. Cohort studies have found low serum total cholesterol to be associated with death due to suicide in most, 1–3 but not all, 4 studies. Low serum total cholesterol has also been associated with an increased risk of depression, 5,6 increased levels of impulsive aggression, 7 and increased risk of acute suicidality in psychiatric inpatients irrespective of age, gender, and nutritional status. 8
One proposed mechanism is that low serum total cholesterol levels lead to lower serotonin concentrations, which in turn may result in depression and ultimately suicide. 9–11 Terao et al12 reported that serum total cholesterol levels in healthy subjects were positively correlated with serotonergic receptor function. Serotonin is believed to play a role in depression in part because depression can be successfully treated with medication that blocks the serotonin transporter. 13 A possible noncausal explanation is that depression results in both lower serum total cholesterol and an increased risk of suicide. 14 It is possible that interleukin-2 causes a decrease in serum total cholesterol, and by suppressing melatonin, increases the risk of depression and suicide. 15
The purpose of the present report is to assess the relation between serum total cholesterol and mortality from suicide among subjects in the Nutrition Canada Survey cohort.
Subjects and Methods
The Nutrition Canada Survey was conducted between September 1970 and December 1972 to provide information on the nutritional status of the Canadian population. Residents of the ten provinces, excluding treaty Aboriginals and persons living in institutions and military camps, were randomly sampled using a three-stage stratified cluster design. A total of 12,795 people responded to the initial invitation to participate (47% response rate); 3,295 unsolicited volunteers also participated. 16 Survey procedures consisted of clinical, dental, and anthropometric examinations; dietary interviews; and blood and urine analysis. Each survey center included a laboratory for the collection and initial processing of blood and urine samples. All samples were then immediately frozen and shipped to a central laboratory for further biochemical analyses. 16
The mortality experience of the cohort was obtained by linking the survey file to Statistics Canada’s National Mortality Database for the years 1970–1993 using the generalized iterative record linkage computer system. 17 No record linkage was attempted for the 2,828 children under 11 years of age at the time of the survey, reducing the cohort from 16,090 to 13,262. We classified deaths as suicide (International Classification of Diseases, 18 8th revision, rubrics 950–959) when suicide was listed as the underlying cause of death.
Participants were followed between the ages of 15 and 84 years. Thus, those younger than 15 years of age at the time of interview began follow-up when they reached 15 years of age, whereas those 85 years of age or older at the time of interview were excluded. The results of this article are based on the mortality experience of both solicited and unsolicited respondents. Exclusion of those 85 years of age or older at baseline (N = 91), pregnant women (N = 902), and those for whom a serum total cholesterol reading was not recorded (N = 715) reduced the number of subjects eligible for analysis to 11,554.
Serum total cholesterol measurements were made with a Technicon colorimeter using the Lieberman-Burchard reaction method. Subjects were not given instructions regarding fasting. We categorized serum total cholesterol levels based on quartile cutpoints (that is, 4.27, 4.97, and 5.77 mmol/liter), as there seemed to be no compelling biological basis for categorizing it otherwise. To determine whether or not our results were overly sensitive to the chosen cutpoints, we made a series of adjustments to these values, in particular to the highest cholesterol quartile, and successively reanalyzed the data.
We conducted a stratified analysis using Mantel-Haenszel estimation for person-time data. 19 The following survey variables were analyzed as potential confounders: age group (15–34, 35–49, 50–69, or 70–84 years of age), sex, smoking status (never, former, or current), total alcohol intake (none, 0.1–10.0, 10.1–25.0, or >25.0 gm per day), history of mental illness, respondent status (solicited or volunteer), and employment status (full-time, homemaker, part-time, student, unemployed, retired, or other). Although highest educational level obtained was recorded, this variable was not examined as a potential confounder because approximately a quarter of the study subjects were 19 years of age or under at baseline. History of mental illness was obtained as part of an interview, administered by trained medical staff, concerning the general medical history of the subjects. Ninety-eight per cent of those reporting a history of mental illness had been either treated by a physician or hospitalized in the past for this condition. Poisson regression was used to assess the possibility of additional confounding beyond age and sex. To explore noncausal explanations of the findings, we conducted three separate subgroup analyses, restricting on years of follow-up, employment status, and history of mental illness.
Table 1 displays the baseline characteristics of study subjects. Age-adjusted mean daily calorie intake was similar across cholesterol quartiles (data not shown). Cases and person-years are presented by serum total cholesterol, age, and sex, simultaneously in Table 2. During follow-up of the cohort, 27 subjects died from suicide. Relative to those 15–34 years of age, we found an increased risk of suicide for all other age groups, but in particular those 70–84 years of age [rate ratio (RR) = 4.52; 95% confidence interval (CI) = 1.38–14.8]. Males (RR = 3.62; 95% CI = 1.54–8.54) were at an increased risk compared with females.
Table 3 displays the crude, age-adjusted, sex-adjusted, and age- and sex-adjusted estimated RRs of suicide by serum total cholesterol level. Compared with subjects in the highest quartile of serum total cholesterol concentration (>5.77 mmol/liter), those in the lowest quartile (<4.27 mmol/liter) had more than six times the risk of committing suicide (RR = 6.39; 95% CI = 1.27–32.1). Increased RRs of 2.95 and 1.94 were observed for the second (4.27–4.96 mmol/liter) and third (4.97–5.77 mmol/liter) quartiles, respectively. Further adjustment for other potential confounders beyond age and sex did not meaningfully alter the results.
Because of the small number of suicide deaths in the study, when we varied the serum total cholesterol cutpoints there were fluctuations, both up and down, in the point estimates and CI values. The strength of the RR estimates, however, were generally similar to those reported above, and the overall pattern of increasing risk of suicide with decreasing serum total cholesterol level was apparent in each analysis (data not shown).
We found estimates of the RR to be similar or increased in each of the nonreference quartiles when we excluded subjects identified as having a history of mental illness. Compared with those in the highest quartile of serum total cholesterol, those in the lowest were at eight times the risk of suicide (RR = 8.10; 95% CI = 1.10–59.5). Further removal of subjects who were unemployed at baseline did not appreciably alter these results (data not shown). The general pattern of results found in the main analysis of increasing risk of suicide with decreasing serum total cholesterol level was also evident after we excluded the first 5 years of follow-up. Whereas the estimate of the RR in the third quartile was virtually identical to that observed using all of the follow-up time, RR estimates were observed to be higher in the second (RR = 4.00; 95% CI = 0.67–23.9) and lowest (RR = 7.73; 95% CI = 1.29–46.3) quartiles. An analysis restricted to male subjects resulted in RR estimates very similar to those found in the main analysis (data not shown).
Using computerized record linkage, we identified 27 deaths due to suicide in the Canadian Mortality Database among the subjects included in the analysis. The possibility of underascertainment of suicide deaths is always a concern, although it is probably unlikely that ascertainment varied by serum total cholesterol level. Using computerized record linkage for follow-up precludes a definitive answer concerning completeness of mortality tracing. A previous study, 20 however, which used both active follow-up and the computerized record linkage system used by the current study, concluded that the computerized record linkage was comparable with or even superior to active follow-up.
We found that subjects with serum total cholesterol concentrations in the lowest quartile were at more than sixfold increased risk of committing suicide in comparison with subjects in the highest quartile. Findings of an inverse association between serum total cholesterol and suicide have also been noted elsewhere. Zureik et al3 found that those with a serum total cholesterol concentration under 4.78 mmol/liter were more than three times as likely to commit suicide (RR = 3.16; 95% CI = 1.38–7.22) than those whose concentration ranged from 4.78 to 6.21 mmol/liter. They also reported that men whose cholesterol concentration decreased by more than 0.13 mmol/liter per year during the first few years of follow-up had more than twice the risk of suicide compared with those whose cholesterol remained relatively stable. Using men who had been screened for the Multiple Risk Factor Intervention Trial, Neaton et al1 found that those with a serum total cholesterol concentration of more than 4.14 mmol/liter were only 60% as likely to die subsequently as a result of suicide (RR = 0.62; 95% CI = 0.46–0.84) as those with a concentration below 4.14 mmol/liter. A Swedish study 2 also observed a strong negative association between cholesterol concentration and mortality from suicide; however, this result did not extend beyond the first 6 years of follow-up. In contrast to the studies cited above, a positive association was reported from a cohort of Japanese-American men living in Hawaii. 4
Evidence from prevention trials has established the efficacy of lowering serum total cholesterol to reduce coronary heart disease risk. These trials, however, have not adequately addressed the effect that lowered cholesterol concentrations have on mortality due to trauma, specifically suicide. Randomized trials have been of limited value in this context for three reasons: they have recorded too few deaths from causes other than ischemic heart disease, their duration has been too short, and treated subjects have not necessarily had low enough cholesterol concentrations to affect suicide risk. 21 Still, a meta-analysis of six primary prevention trials found an increase in deaths unrelated to illness in groups receiving treatment to lower cholesterol concentration relative to controls. 22 On the basis of a review of the deaths in the active-treatment arms, the interpretation of the findings of two of these trials have been challenged 23 but not without rebuttal. 24,25
A number of mechanisms have been proposed to explain the suicide/serum total cholesterol association. Low serum total cholesterol concentration might alter the metabolism of serotonin, leading to depression or to poor control of aggressive impulses and therefore to an increased risk of suicide. 9,10 Several studies in humans and nonhuman primates suggest a specific connection between low or lowered fats or cholesterol levels and low or lowered serotonin activity. 26 Plasma serotonin concentrations are lower in untreated men with persistently low serum total cholesterol concentrations. 9
In a review of serum total cholesterol and mortality, Law et al21 noted that an excess risk was restricted to community-based cohort studies; cohorts of employed men experienced no increased risk. The argument was that community-based cohorts included subjects who were unemployed, and unemployment can have adverse effects on psychological health. 27 As those with a history of mental illness or those who were unemployed represented only a very small proportion of our analytic cohort, this is an unlikely explanation for our findings. A subgroup analysis excluding these groups confirmed this conclusion. Another noncausal explanation is that low serum total cholesterol is a consequence of depression rather than a cause, perhaps simply because depressed people eat less. 14 Our data indicated little difference in age-adjusted mean calories according to serum total cholesterol, however. As well, if low serum total cholesterol is a consequence of depression, we might expect to see the inverse association concentrated in the first few years of follow-up. In fact, however, we found that the association between low serum total cholesterol and suicide was, if anything, even stronger on elimination of the first 5 years of our 23-year follow-up.
An alternate noncausal hypothesis is that increased levels of the cytokine interleukin-2 results in decreased levels of both serum total cholesterol and melatonin; lower melatonin then leads to depression and an increased risk of suicide. 15 It may be that those with low serum total cholesterol are a heterogenous population, composed of those with naturally low levels, those with low levels because of poor diet, and those with low levels because of the effects of interleukin-2. Each would have a different etiology and a different outcome. Because epidemiologic studies have not been able to differentiate between these groups, they have been unable to differentiate adequately between competing hypotheses.
Whether low cholesterol promotes depression, ultimately leading to suicide, or is a consequence of it remains to be determined. This is an increasingly important question, given the ability of new pharmacotherapies to lower serum total cholesterol dramatically. 28
The record linkage portion of this study was conducted by Statistics Canada.
1. Neaton JD, Blackburn H, Jacobs D, Kuller L, Lee DJ, Sherwin R, Shih J, Stamler J, Wentworth D. Serum cholesterol level and mortality findings for men screened in the Multiple Risk Factor Intervention Trial. Multiple Risk Factor Intervention Trial Research Group. Arch Intern Med 1992; 152: 1490–1500.
2. Lindberg G, Rastam L, Gullberg B, Eklund GA. Low serum cholesterol concentration and short term mortality from injuries in men and women. BMJ 1992; 305: 277–279.
3. Zureik M, Courbon D, Ducimetiere P. Serum cholesterol concentration and death from suicide in men: Paris prospective study I. BMJ 1996; 313: 649–651.
4. Iribarren C, Reed DM, Wergowske G, Burchfiel CM, Dwyer JH. Serum cholesterol level and mortality due to suicide and trauma in the Honolulu Heart Program. Arch Intern Med 1995; 155: 695–700.
5. Suarez EC. Relations of trait depression and anxiety to low lipid and lipoprotein concentrations in healthy young adult women. Psychosom Med 1999; 61: 273–279.
6. Horsten M, Wamala SP, Vingerhoets A, Orth-Gomer K. Depressive symptoms, social support, and lipid profile in healthy middle-aged women. Psychosom Med 1997; 59: 521–528.
7. New AS, Sevin EM, Mitropoulou V, Reynolds D, Novotny SL, Callahan A, Trestman RL. Serum cholesterol and impulsivity in personality disorders. Psychiatry Res 1999; 85: 145–150.
8. Papassotiropoulos A, Hawellek B, Frahnert C, Rao GS, Rao ML. The risk of suicidality in psychiatric inpatients increases with low plasma cholesterol. Pharmacopsychiatry 1999; 32: 1–4.
9. Steegmans PH, Fekkes D, Hoes AW, Bak AA, van der Does E, Grobbee DE. Low serum cholesterol concentration and serotonin metabolism in men. BMJ 1996; 312: 221.
10. Engelberg H. Low serum cholesterol and suicide. Lancet 1992; 339: 727–729.
11. Salter M. Low serum cholesterol and suicide. Lancet 1992; 339: 1169–1169.
12. Terao T, Yoshimura R, Ohmori O, Takano T, Takahashi N, Iwata N, Suzuki T, Abe K. Effect of serum cholesterol levels on meta-chlorophenylpiperazine-evoked neuroendocrine responses in healthy subjects. Biol Psychiatry 1997; 41: 974–978.
13. Bligh-Glover W, Kolli TN, Shapiro-Kulnane L, Dilley GE, Friedman L, Balraj E, Rajkowska G, Stockmeier CA. The serotonin transporter in the midbrain of suicide victims with major depression. Biol Psychiatry 2000; 47: 1015–1024.
14. Law M. Having too much evidence (depression, suicide, and low serum cholesterol). BMJ 1996; 313: 651–652.
15. Penttinen J. Hypothesis: low serum cholesterol, suicide, and interleukin-2. Am J Epidemiol 1995; 141: 716–718.
16. Nutrition Canada. Nutrition: A National Priority. A Report from Nutrition Canada
by the Bureau of Nutritional Sciences, Health Protection Branch, Department of National Health and Welfare. Pub. No. H58–36/1973. Ottawa: Information Canada, 1973.
17. Howe G, Lindsay J. A generalized iterative record linkage computer system for use in medical follow-up studies. Comput Biomed 1981; 14: 327–340.
18. National Center for Health Statistics. International Classification of Diseases, 8th Revision, Adapted for Use in the United States. Public Health Service Pub. No. 1693. Washington DC: Public Health Service, U.S. Department of Health, Education and Welfare, 1967.
19. Rothman KJ, Greenland S. Modern Epidemiology. 2nd ed. Philadelphia: Lippincott-Raven Publishers, 1998;269–270.
20. Shannon HS, Jamieson E, Walsh C, Julian JA, Fair ME, Buffet A. Comparison of individual follow-up and computerized linkage using the Canadian Mortality Data Base. Can J Public Health 1989; 80: 54–57.
21. Law MR, Thompson SG, Wald NJ. Assessing possible hazards of reducing serum cholesterol. BMJ 1994; 308: 373–379.
22. Muldoon MF, Manuck SB, Matthews KA. Lowering cholesterol concentrations and mortality: a quantitative review of primary prevention trials. BMJ 1990; 301: 309–314.
23. Wysowski DK, Gross TP. Deaths due to accidents and violence in two recent trials of cholesterol-lowering drugs. Arch Intern Med 1990; 150: 2169–2172.
24. Muldoon MF, Manuck SB, Matthews KA. Does cholesterol lowering increase non-illness-related mortality? Arch Intern Med 1991; 151: 1453–1454.
25. Vasan RS. Lowering cholesterol and death due to accidents, suicides: unresolved issues. Arch Intern Med 1992; 152: 414–417.
26. Golomb BA. Cholesterol and violence: is there a connection? Ann Intern Med 1998; 128: 478–487.
27. Warr P, Jackson P. Factors Influencing the psychological impact of prolonged unemployment and of re-employment. Psychol Med 1999; 15: 795–807.
28. Pitt B, Waters D, Brown WV, van Boven AJ, Schwartz L, Title LM, Eisenberg D. Aggressive lipid-lowering therapy compared with angioplasty in stable coronary artery disease. N Engl J Med 1999; 341: 70–76.