Role of body mass index in the prediction of all cause mortality in over 62,000 men and women. The Italian RIFLE Pooling Project. Risk Factor and Life Expectancy

Abstract

(J Epidemiol Community Health 1998;52:20-26) Despite the large amount of medical literature available on the subject, the relations of body weight, body fatness, and related measures to disease and mortality require further investigation mainly in the search for optimal weight with regard to the best health status and health perspectives. 1-11On the other hand it has taken years for the acceptance of the concept of a J or U shaped relation of body weight and related measures to mortality, 4 8 9 12 and recently the desirable body weight for elderly people seems to be higher than previously claimed.Extensive studies based on population samples with adequate follow up in Italy are limited. For example, the Italian cohorts of the Seven Countries Study could be followed up for 25 years, 13 but they were made only for middle aged men. A larger and more recent analysis, based on the pooling of a few studies, suggested the existence of a parabolic inverse J shaped relation of some parameters of obesity to 10 year mortality, although the step of the left branch of the curve could be reduced by the exclusion of smokers, early deaths, and people carrying serious health conditions at entry. 14The availability of data from the Italian RIFLE (Risk Factors and Life Expectancy) pooling project, 15 including many population samples spread all over the country and large numbers, oVers the possibility to provide more consistent material although based on very simple measures of body fatness.The purpose of this analysis is to describe the relation of body mass index to short-term mortality in a large population sample of men and women in Italy. Methods STUDY POPULATIONThe RIFLE Pooling Project includes epidemiological data from nine diVerent large scale population studies started in Italy between 1978 and 1987 and focused on cardiovascular diseases or other chronic conditions. 15 The nine studies gathered 52 population samples of men and women in the age range of 20 to 69 years. Each sample, except two, were identified by electoral rolls (where all people aged 18 plus are enlisted) in defined geographical areas. Two samples were of occupational origin (large companies with a prevalence of sedentary employees), both located in Rome. The average participation rate was around 70%, while subjects lost to follow up at six years were 2.1%.Only 47 cohorts, however, spread across 13 of the 20 Italian regions, were finally considered because, for the remaining five, a minimum follow up could not be completed.These nine studies proved suitable for pooling because all of them were conducted or coordinated or methodologically advised by the same centre and most of the measurements were made using the same standardisation and quality control procedures.

Concurrent and separate effects of body mass index and waist-to-hip ratio on 24-year mortality in the Population Study of Women in Gothenburg: Evidence of age-dependency

“…Our results are consistent also with previous studies; for instance in an Italian study [14], where the optimal BMI value in the upper age-group of women seemed to be close to 32, and similar results were observed in a study from US [15], the lowest mortality being in the upper quintile (BMI ‡ 29.2). Also in two other US studies, the Buffalo Health Study [5], and the Longitudinal Study of Aging [16], there was no increased risk of death in women with increasing BMI.…”

Body Mass Index and Mortality in a Prospectively Studied Cohort of Scandinavian Women: The Women’s Lifestyle and Health Cohort Study

“…Results from other studies are ambiguous, but generally the effect of excluding early deaths seems to be moderate [3,7,9,[12][13][14]. Several researchers have concluded that the relationship between BMI and all-cause mortality may change as the follow-up time increases [8,15,16].…”

“…Despite extensive epidemiological investigation, the association between overweight and all-cause mortality is still cause for debate [2]. Although most studies have found a J-or U-shaped relationship [3][4][5][6][7][8][9][10][11], with an excess mortality both among those who are very lean and those who are obese, others have not [12][13][14]. Moreover, the association between weight and allcause mortality may vary by age, and there is indeed evidence that the adverse effect of a high body mass index is reduced at older ages [15][16][17][18][19].…”

Longitudinal follow-up of nutritional status and its influencing factors in adults undergoing allogeneic hematopoietic cell transplantation

“…The WHO classification of, for example, normal weight (18.5-25 kg/m 2 ) does not consider the dynamic changes and gender-specific differences in body weight over the entire lifespan, 27 nor does it take into account that when considering allcause mortality, optimal BMI varies by age and sex. 47,48 We therefore recommend that, when the BMI is used in the context of nutritional assessment, it should be employed to compare BMI calculations with age-and gender-specific BMI percentiles of an appropriate reference population.…”