Current issues of ACP Journal Club are published in Annals of Internal Medicine


Violent death was not associated with cholesterol

ACP J Club. 1995 Jan-Feb;122:24. doi:10.7326/ACPJC-1995-122-1-024

Source Citation

Vartiainen E, Puska P, Pekkanen J, et al. Serum cholesterol concentration and mortality from accidents, suicide, and other violent causes. BMJ. 1994 Aug 13;309:445-7.



To determine whether death from accidents, suicides, and other violent causes is associated with serum total cholesterol concentration.


Cohort analytic study of data from 2 adult, randomly selected population samples.


Community study in Finland.


22 432 persons (baseline age range, 25 to 64 y in 1972 or 1977; 11 534 women) from 2 Finnish provinces were surveyed for risk factors for coronary heart disease, including serum cholesterol level. Information on 236 violent deaths was obtained from the national mortality register over the ensuing years.

Assessment of Risk Factors

A self-administered questionnaire comprising questions on socioeconomic and psychosocial factors, including smoking and alcohol use. For univariate analyses, participants were categorized as never-smokers, former smokers, and smokers and were assessed on a 7-point scale for alcohol use. For multivariate analyses, smoking was categorized as smoking or nonsmoking, and alcohol use was categorized as drinking alcohol more or less than once or twice a month. Physical measurements included height, weight, blood pressure, and serum total cholesterol concentrations. Persons were classified into 4 cholesterol categories: < 5.0 mmol/L; 5.0 to 6.49 mmol/L; 6.5 to 7.99 mmol/L; or ≥ 8.00 mmol/L.

Main Outcome Measures

Death from violent causes (i.e., death from causes other than disease) (International Classification of Diseases, 8th Revision, codes 800 to 999).

Main Results

Death from violent causes was not associated with total serum cholesterol concentration. Risk for violent death was associated with smoking among men and women (risk ratio [RR], 2.28 [95% CI, 1.65 to 3.14] and 2.25 [CI, 1.03 to 4.90], respectively) and with alcohol consumption among men (RR, 1.11; CI, 1.03 to 1.19). The association grew stronger with increased smoking and alcohol consumption.


No association was detected between death from violent causes and total serum cholesterol concentration. Smoking and alcohol were associated with violent death.

Sources of funding: Not stated.

For article reprint: Dr. E. Vartiainen, National Public Health Institute, Mannerheimintie 166, FIN-00300, Helsinki, Finland. FAX 358-0-4744-338.


One of the more surprising controversies to emerge in the past 5 years about serum cholesterol is its possible inverse relation to nonillness mortality (death from suicide, accident, or violence) (1). In meta-analyses of primary prevention trials of cholesterol reduction, the association was initially shown as an excess of nonillness mortality. Investigators recently have sought to determine whether epidemiologic evidence indicates that a naturally low serum cholesterol concentration is a risk factor for suicide or accidental death and what mechanisms, if any, could explain this association.

The study by Vartiainen and colleagues is the latest contribution to the debate about this puzzling topic. The data do not support the hypothesis that persons with relatively low serum cholesterol concentrations have elevated nonillness mortality. Several caveats, however, should be recognized to put this new study into the proper context.

This Finnish study is the fourth observational study of cholesterol concentrations and nonillness mortality that had reasonable power to detect an association (i.e., at least 100 suicides and traumatic deaths); however, it is the first not to find a significant, inverse association (1). Second, Finland has one of the highest average cholesterol concentrations of any country; consequently, the cholesterol concentration of the low-cholesterol group in this study was substantially higher than that of low-cholesterol groups in other studies. Finally, the important clinical issue of whether cholesterol reduction increases the risk for death from suicide or trauma can only be resolved in randomized trials of cholesterol lowering.

At this point, research on the association between low or lowered cholesterol concentrations and nonillness mortality is not mature enough to permit conclusions or to discuss clinical implications. That same immaturity, however, makes it an exciting topic to study and follow.

Matthew F. Muldoon, MD, MPH
University of Pittsburgh Pittsburgh, Pennsylvania