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


Low plasma cholesterol level increased the risk for death in men

ACP J Club. 1992 May-June;116:92. doi:10.7326/ACPJC-1992-116-3-092

Source Citation

Smith GD, Shipley MJ, Marmot MG, Rose G. Plasma cholesterol concentration and mortality: the Whitehall Study. JAMA. 1992 Jan 1;267:70-6.



To examine the association between plasma cholesterol concentration and specific causes of death in middle-aged men.


Cohort followed for ≥ 18 years.


Civil service offices, London, England, from 1967 to 1969.


Cholesterol measurements were available for 17 718 men who were included in the Whitehall Study of 18 403 male civil servants aged 40 through 64 years.

Assessment of risk factors

At entry into the study the men answered questions about age, marital status, employment grade, smoking, and health status. Clinical measurements included plasma cholesterol concentration determined from capillary blood samples.

Main outcome measures

Cardiovascular, cancer, respiratory, and all-cause mortality. Causes of death were obtained from death certificates. Vital status was determined for 99% of the men.

Main results

4022 men (23%) died during the follow-up period. Mortality decreased with lower plasma cholesterol concentration at all levels except for the lowest 5% of cholesterol determinations (age-adjusted relative risk [RR] for a 1 SD decrease in cholesterol level 0.93, 95% CI 0.90 to 0.95). Cardiovascular mortality was consistently related to increasing plasma cholesterol (RR for 1 SD decrease 0.83, CI 0.80 to 0.86). There was no relation with cancer mortality overall, but the group with the lowest 5% of cholesterol levels had the highest mortality rate for 6 of 14 cancer sites examined: rectum, liver, pancreas, lung, bladder, and kidney. The association was similar for smoking-related cancers. There was an inverse relation between cholesterol levels and death from noncardiovascular, noncancer causes (RR 1.12, CI 1.03 to 1.21) and from respiratory disease (RR 1.25, CI 1.11 to 1.40). When deaths during the first 5 and 10 years of the study were excluded, the only inverse associations of cholesterol levels that remained significant were those with all-cause mortality, respiratory disease mortality, and noncardiovascular, non-neoplasm mortality. Being unmarried and having respiratory symptoms, unexplained weight loss, and lower body mass index were all associated with lower cholesterol levels and higher mortality. After adjusting for these factors as well as for smoking and employment grade, no significant inverse association between mortality and cholesterol remained.


Men with the lowest 5% of cholesterol levels may have had health conditions or demographic characteristics that placed them at greater risk for death.

Source of funding: Not stated.

Address for article reprint: Dr. G. Davey Smith, Department of Epidemiology and Population Sciences, London School of Hygiene and Tropical Medicine, Keppel Street, London WC1E 7HT, United Kingdom.


This study offers insights into the causes of the association between low cholesterol levels and elevated noncardiovascular mortality. Among the factors associated with lower mean cholesterol levels are respiratory symptoms, recent unexplained weight loss, low forced expiratory volume at 1 second (FEV1), and low body mass index, factors that themselves are associated with elevated mortality rates. These factors explain much of the association between low cholesterol and elevated noncardiovascular mortality, implying that underlying diseases caused the low cholesterol levels.

This study may overestimate the statistical significance of the association between mortality from specific causes and low cholesterol levels because no adjustment was made for multiple comparisons: Some associations may appear significant simply by chance. The study may also understate some associations, because the cause of death information was obtained from death certificates, which are notoriously unreliable.

One important result of the study is that all-cause mortality, when the first 10 years of deaths are excluded, does not rise with cholesterol level until the highest category of cholesterol (plasma level > 5.92 mmol/L) is reached. Insofar as low cholesterol does not have a causal role in promoting death from other causes, reducing cholesterol at any initial level may be safe, but the survival benefits of cholesterol reduction are much greater at very high levels thanat modest elevations.

Definitive evidence about the consequences of cholesterol reduction must come from clinical trials, not observational studies like this one. The benign results for low cholesterol levels do not contradict findings from clinical trials that diets and drugs that lower cholesterol levels may increase noncardiovascular mortality by unknown mechanisms. As the authors conclude, "establishing the effect of lowered cholesterol levels on mortality risk requires continued analysis and monitoring of results from randomized trials."

Alan M. Garber, MD
Stanford University School of MedicineDepartment of Veterans Affairs Medical CenterPalo Alto, California, USA

Alan M. Garber, MD
Stanford University School of MedicineDepartment of Veterans Affairs Medical Center
Palo Alto, California, USA