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


High total cholesterol levels reduced the risk for all-cause mortality in adults older than 85 years of age

ACP J Club. 1998 May-June;128:76. doi:10.7326/ACPJC-1998-128-3-076

Source Citation

Weverling-Rijnsburger AW, Blauw GJ, Lagaay AM, et al. Total cholesterol and risk of mortality in the oldest old. Lancet. 1997 Oct 18;350: 1119-23. Erratum. Lancet. 1998 Jan 3;351:70. [PubMed ID: 9343498]



To determine the association between total cholesterol levels and all-cause and cause-specific mortality in persons > 85 years of age.


A community-based cohort study (Leiden 85-plus Study).


Leiden, the Netherlands.


724 adults who were > 85 years of age (median age 89 y, 73% women) at enrollment (1986 to 1988).

Assessment of risk factors

Total cholesterol levels at baseline were categorized into low (< 5 mmol/L [< 193 mg/dL]), moderately high (5.0 to 6.4 mmol/L [193 to 247 mg/dL]), and high (≥ 6.5 mmol/L [≥ 248 mg/dL]). The lowest level was taken as the reference standard (relative risk [RR] 1.0). 28% of the population had low total cholesterol levels, 48% had moderately high levels, and 24% had high levels. Other risk factors were age, sex, hypertension, diabetes mellitus, smoking status, and history of myocardial infarction or cardiovascular disease. These other risk factors and levels of serum albumin were used to adjust all analyses.

Main outcome measures

National and regional databases provided data on all-cause and cause-specific mortality (cardiovascular disease, malignant conditions, respiratory disease, all infections, dementia without psychosis, and external causes). All deaths that occurred in the first year were excluded.

Main results

During follow-up to October 1996, 642 participants (89%) died. Median time of survival was 2.5 years in the lowest category of cholesterol levels, 3.4 years in the moderately high category, and 4.3 years in the highest category (P for trend < 0.001). Each 1 mmol/L (39 mg/dL) increase in cholesterol level was associated with a 15% reduction in the risk for death (RR 0.85, 95% CI 0.79 to 0.92). The 10-year adjusted RR for mortality in the highest cholesterol level group was 0.64 (CI 0.50 to 0.82). Most persons died of cardiovascular disease, and total cholesterol levels were not associated with any differences in cardiovascular mortality among the 3 groups. Decreases in mortality were seen for cancer (P = 0.002) and all infections (P = 0.03) for participants with the highest levels of total cholesterol.


For adults > 85 years of age, high total cholesterol levels were associated with decreased all-cause mortality and mortality from cancer and all infections. Mortality from cardiovascular disease was high but not associated with total cholesterol levels.

Sources of funding: National Institutes of Health and Dutch Ministry of Health and Science.

For article reprint: Dr. A.W. Weverling-Rijnsburger, Leiden University Medical Centre, Department of General Internal Medicine, Box 9600, C1-R45, 2300 RC Leiden, The Netherlands. FAX 31-71-5-248-140.


Although the importance of total blood cholesterol levels for risk for coronary heart disease (CHD) has been well established in middle-aged persons, some uncertainty remains about the strength of this association in very elderly persons. Precise estimation of even small increments in RR for CHD associated with increasing cholesterol levels in elderly persons has public health implications because of the high absolute mortality risks for CHD in this age group. However, most prospective studies that have examined mortality associations in elderly persons have been based on prolonged follow-up of middle-aged persons. Investigators have mainly relied on single measurements of blood cholesterol levels at enrollment to classify participants and thus have underestimated the strength of mortality associations because of regression dilution bias.

Another problem in prospective studies of elderly persons is the influence of comorbid conditions on the relation between risk-factor levels and subsequent mortality. This is most noticeable in short-term follow-up studies where U- and J-shaped associations between cholesterol levels and all-cause mortality have been described in middle age and in old age. The increase in mortality rates in persons with low cholesterol levels probably reflects “reverse causality,” whereby low cholesterol levels are a consequence of comorbid conditions rather than a cause of disease that leads to death.

This issue is particularly relevant to the study by Weverling-Rijnsburger and colleagues, in which excluding deaths that occurred within 1 year or adjusting for albumin levels was not sufficient to correct for the effects of comorbid conditions. 43% of the persons died from cardiovascular disease, 16% from cancer, and 11% from infection. The observed risk associations between cholesterol levels and cardiovascular mortality are not consistent with the literature. Because of the opposing direction of the “true” (e.g., cholesterol levels and CHD) and “reverse” (e.g., cholesterol levels and cancer or infection) causal associations, combining deaths from all causes is not prudent. Apart from the method of correction for regression dilution bias, the chief concern about this report relates to interpretation of the causal associations. Nevertheless, the authors have wisely concluded that the balance of the benefits and harms of cholesterol-lowering therapy in this age group has yet to be determined.

Robert Clarke, MD
Clinical Trial Service UnitOxford, England, UK

Robert Clarke, MD
Clinical Trial Service Unit
Oxford, England, UK