Update to Gage and O'Connor's "Nutrition and the Variation in Level and Age Patterns of Mortality" (1994)

Abstract

Update to Gage and O'Connor's "Nutrition and the Variation in Level and Age Patterns of Mortality" (1994) Timothy B. Gage and Kathleen O'Connor Keywords Nutrition, Mortality, Dietary Components, Age Patterns of Mortality. It is widely assumed that nutrition influences the level of mortality. However, the association of nutrition and mortality is not linear; improved nutrition is associated with lower mortality when infectious causes of death dominate, but overnutrition, particularly with respect to the consumption of carbohydrates and fats, is often associated with increased chronic disease mortality. It is also possible that nutrition influences the age patterns of mortality. The age patterns of mortality are known to vary geographically among the world's populations (Coale and Demeny 1966). Historically, these patterns tend to be geographically consistent even as the level of mortality declines. The cause of this human variation is understudied. Preston (1976) argued that the proximate causes of this variation result from varying geographic distributions of causes of death. Variation in nutrition may be responsible for some of this variation, possibly through influences on cause of death. The objective of our 1994 paper was to demonstrate the statistical association of the level of nutrition and dietary components on the overall level of mortality and the age patterns of mortality. For our 1994 paper we collected a worldwide sample of 341 life tables from 96 countries for the period 1952-1975 and matched them with standard national food balance sheet data for each country. One hundred seventy-one tables were for males and 170 were for females; 1 female life table was excluded as defective. Expectation of life at birth in this sample ranged from 31.7 to 75.9 years. The independent variables included total calories and a principal components analysis of diet. In this case the components were interpretable as (1) a meat component, high in protein and fats, (2) carbohydrates, and (3) fats versus proteins. Standard regression analyses indicated that the international variation in nutrition was closely associated with the international variation in mortality. More than 50% of the variation was explained by these regressions, with respect to both level of mortality and the age patterns of mortality. The results indicated that total calories was positively related to expectation of life and was a better indicator than the percentage of caloric requirement. The analysis using dietary [End Page 575] components was similar, except that higher carbohydrate levels were associated with increasing mortality and higher fat versus protein diets were associated with improved mortality, although this effect was attenuated at high and low levels because of significant interaction terms. An analysis of the impact of nutrition on the age patterns of mortality also indicated important associations. r2 values ranged from almost 90% to 20% depending on the parameter of the age patterns of mortality examined (age patterns were parameterized using the Siler model) and total calories versus dietary components. The age patterns of mortality appeared to be more closely associated with the dietary components than with simple calories. Overall, the results indicated that mortality improves as nutrition increases. No increase in mortality was observed even in the best (over?) nourished populations. Of course this might not be due to nutrition but to other phenomena correlated with nutrition, for example, modern medicine. The age patterns of mortality were closely associated with the nutrient composition of the diet, and the nutrients interacted in complex ways. Finally, the results indicated that the association of mortality (both level and age patterns) and nutrition was very strong, as indicated by the r2 levels. Despite the strong effects identified, our 1994 paper has generated few citations to date. Shaw et al. (2005) were unable to demonstrate the nonlinear association of dietary fats and mortality, although dietary fats was defined as "butter" in this study. On the other hand, Miller and Frech (2000) reported results similar to ours using data on all animal fats. Several reviews of diet quality indexes have also referenced the paper (Kant 1996; Seymour et al. 2003). However, the topic has a relatively small literature. Seymour et al. (2003) cited only ten other papers on this topic, only four of which clearly postdate the publication of our...