Body Mass Index and Risk of Mortality in Older Adults: Secondary Analysis Using the Longitudinal Study of Aging

Abstract

Body Mass Index (BMI) is one of the most commonly used assessment tools for measuring nutritional status in adults beginning with Andres’ re-analysis of the Metropolitan Life Insurance Company data (1985). There is, however, some controversy over the relative risk of elevated or very low BMI and some controversy over what BMI range would include the lowest level of risk of mortality. Manson, et al., in a study of nurses over 15 years concluded that, among non-smokers who were weight stable, the lowest risk was associated with the lowest BMI category (1995). This study did not include any women over the age of 70. The Nutrition Screening Initiative (NSI) implicitly recognizes increased risk for both elevated and low BMI; the cutoffs for risk on both the Level 1 and Level 2 screening instruments are for BMIs below 24 or above 27 (1993). Allison, et al., using the Longitudinal Study of Aging (LSOA), found that the BMI associated with minimum risk was 27-30 for men and 30-35 for women (1997). However, Allison did not use information about functional limitations present in the subjects at the beginning of the study. Clearly, more work is needed to clarify these issues. In this study, I also used the Longitudinal Study of Aging as a starting point. Mortality, within the 6 year period, was the end-point for a maximum likelihood logit analysis in this study. Covariates included markers for diseases, poverty status, ADLs and IADLs, age in 1984. Dummy variables for BMI categories corresponding to those used in Manson's study were also created. In the analysis, the lowest BMI category, BMI less than 19, was used as the reference category and all other categories were included in the analysis. Our results confirmed that functional limitations, presence of disease and increasing age were associated with increased risk of dying. Poverty status was not. In this analysis, holding all other variables constant, the lowest mortality was associated with the heaviest BMI category and all categories showed lower risk of mortality than did the reference category. This study tends to confirm Allison's work and to raise questions about the applicability of Manson's findings to this older group.