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


Low serum albumin level predicted mortality in elderly persons

ACP J Club. 1995 Mar-April;122:50. doi:10.7326/ACPJC-1995-122-2-050

Source Citation

Corti MC, Guralnik JM, Salive ME, Sorkin JD. Serum albumin level and physical disability as predictors of mortality in older persons. JAMA. 1994 Oct 5;272:1036-42.



To examine the association among serum albumin level, physical disability, and all-cause mortality in elderly men and women living in the community.


Cohort study with mean follow-up of 3.7 years.


3 communities.


4116 persons (mean age 79 y, 64% women) who were ≥ 71 years of age, who had responded to 6 annual telephone and in-person interviews as part of the Established Populations for Epidemiologic Studies of the Elderly study (EPESE), and who consented to having a blood sample taken at the sixth interview between 1987 and 1989.

Assessment of risk factors

Physical disability measured as the presence or absence of mobility disability (inability to walk one half mile or up and down stairs unassisted [modified Rosow-Breslau scale]) and activities of daily living disability (inability to walk across a small room, bathe, or transfer from bed to chair); race; education (years of schooling); smoking status; chronic conditions (heart attack, stroke, cancer, diabetes, or hip fracture); body mass index; and albumin level (< 35 g/L, 35 to 38 g/L, > 38 to 41 g/L, > 41 to 43 g/L, and > 43 g/L). Diagnoses of disabilities and chronic conditions were based on self-reports given during interviews.

Main outcome measures

Total and cause-specific mortality.

Main results

935 participants died during the follow-up period. In both men and women, all-cause mortality rates were highest in those in the lowest albumin level category and decreased with increased albumin levels (P for trend < 0.001). This association was maintained for cause-specific mortality. For men and women with albumin levels < 35 g/L, the adjusted relative risk (RR) for death was 1.9 (95% CI 1.1 to 3.1) and 3.7 (CI 2.5 to 5.5), respectively. Men in the lowest albumin level category were at highest risk for death during the first year only (RR 5.3, CI 2.0 to 13.8), whereas the women in this category had a high risk for death during the first year (RR 4.9, CI 2.1 to 11.5) and stayed at risk during the subsequent years (RR for death ≥ 1 y 3.3, CI 2.1 to 5.2). Within albumin level categories, the risk for death rose as the level of disability increased and, within levels of disability, the risk for death rose as the albumin level decreased. The effect of albumin level on mortality was independent of physical disability and chronic conditions.


Serum albumin levels were an independent risk factor for mortality in elderly persons. Low serum albumin levels were associated with the highest mortality in elderly persons with physical disabilities.

Source of funding: National Institute on Aging.

For article reprint: Dr. M. Corti, National Institute on Aging, 7201 Wisconsin Avenue, Room 3C-309, Bethesda, MD 20892-9205, USA. FAX 301-496-4006.


Clinical epidemiologists will be interested in the study by Corti and colleagues. Although the association between hypoalbuminemia and mortality has been examined before, this study has many strengths common to work coming from the EPESE. The data are from a population-based cohort study in which participants were followed for several years. The investigators had data on serum albumin, functional ability, chronic disease, and both disease-specific and all-cause mortality.

Clinical investigators should also be interested in this study. The finding that serum albumin level is an independent risk factor for mortality should encourage its inclusion as a variable in many studies of elderly persons, especially in clinical trials that use mortality as an end point. Investigators in clinical trials should also consider adding simple measures of disability to data on albumin to create a more powerful measure of frailty and risk for mortality. In particular, trials using interventions such as exercise, diet, and hormone replacement (including that of growth hormone, estrogen, and testosterone) to improve the well-being of frail elderly persons should consider this "frailty index" to either target or stratify participants.

Clinicians caring for elderly persons may find it more difficult to make practical application of this study in their practices. Certainly, this study reinforces the importance of paying attention to and making efforts to improve the nutrition and functional status of elderly patients. Clinicians could also consider using the prognostic information to plan future medical care with their patients. As the authors admit, however, an epidemiologic study "cannot determine if albumin is a secondary marker of risk or a truly independent cause of mortality." Thus, clinicians must await future clinical trials for more specific guidance on how to intervene for patients found to be at high risk by this type of frailty index.

Greg A. Sachs, MD
The University of Chicago Medical CenterChicago, Illinois, USA