| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research |
Medical Branch, Medical Corps, Israel Defense Forces, ISRAEL
Address reprint requests to: Dror Mandel, MD, 3 Ha’Emek Street, Ramat Hasharon 47203, ISRAEL. E-mail: mandelrd{at}netvision.net.il
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSREFERENCES |
|---|
Methods: A research questionnaire filled by a representative sample of military personnel upon discharge from service was analyzed. At the same time, weight and height were measured, and BMI was calculated (BMI = weight (Kg)/height2(m2)).
Results: There were 35,951 participants in the study, including 16204 females and 19747 males. There was a positive correlation between BMI and height in men (regression slope = 0.00717, r = 0.015, p = 0.03), while the correlation was negative in women (regression slope = -0.02811, r = -0.05, p < 0.0001). In multiple regression analysis, when BMI was used as the dependent variable and height, gender, ethnic origin, smoking, oral contraceptive use, and level of recreational exercise as the independent variables, only height, gender, and ethnic origin remained significant in the final analysis (R-square 0.0205, p < 0.0001).
Conclusion: In young adults, BMI is affected in a subtle, but opposite manner in males and females. In males, BMI increases with increasing height, while in females, BMI decreases with increasing height.
Key words: BMI, height, men, women, military, ethnicity
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSREFERENCES |
|---|
In Israel, there is mandatory military conscription at the age of 18 years for a period of 1 year and 10 months in females and 3 years in males. At the time of completion of their military service, conscripts are selected at random, using random numbers, and are asked to voluntarily fill out (after obtaining a signed informed consent) a research questionnaire about their medical history and several health related topics, such as usage of oral contraceptives, smoking, and the like. We took advantage of the questionnaires to weigh and measure the subjects. Specifically, we tested the hypotheses that BMI is influenced by height and that the degree of adiposity is not uniformly distributed in the population.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSREFERENCES |
|---|
Data Collection
Trained nurses from the IDF Health Surveillance Section interviewed participants concerning demographic data (subjects’ country of birth and extent of formal education, paternal country of birth and level of education and number of siblings); health behaviors: smoking (active, on a routine basis), alcohol intake (yes/no, on a routine basis), oral contraceptives use (on a routine basis). Exercise habits were self evaluated by the soldier and expressed as a score, in a scale of 1 to 9 (1 = low level of exercise, sedentary lifestyle; 9 = intensive daily exercise). Weight and height were measured by a few, highly trained nurses, after removal of outer garments and shoes, leaving only underwear, using a weight and standing height scale (Sunbeam products Inc, Bridgeview, IL).
Data Analysis
Definitions: BMI was calculated as weight in kilograms divided by height in meters square (kg/m2). Origin was defined according to the birthplace of the subject’s father or paternal grandfather, if the father was Israeli born. Origin was classified as Western (Ashkenazi) origin—subjects born in Europe (excluding Turkey), the Americas, Australia or South Africa; Eastern (Sephardic) origin—subjects born in Asia, Turkey, North Africa or Ethiopia; and Israeli origin—subjects whose father and paternal grandfather were born in Israel.
Statistical Analysis
The Statistical Analysis System (SAS Institute Inc, Cary, NC) was used to analyze the data. Data were analyzed for normality, and regression analysis was used to study the relationship between BMI and height in men and in women separately. General Linear Models were used to study the independent effect of variables such as height, gender, ethnic origin or level of recreational exercise upon BMI. Results are expressed as mean ± SD, or n (%). A p value of <0.05 was considered significant.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSREFERENCES |
|---|
|
|
|
|
DISCUSSION |
|---|
As stated earlier, the BMI provides an index of weight relative to height and is considered in general to be a valid index of adiposity [2]. There are some limitations to this statement: the first one relates to the fact that although BMI correlates strongly with estimates of total body fat measured by hydrodensitometry [4], or by dual energy radiograph absorptiometry [5], it also correlates with fat-free mass (FFM) [2]. During childhood and adolescence, it is recommended to use age and gender-specific percentiles [6, 7], due to the fact that age-related increases in BMI during growth are affected by both increases in the lean component of BMI (FFM/height2) and the fat component of BMI (TBF/height2) [4]. It has been noted that the extent to which each component contributes to the change in BMI depends on the gender, race and age of the individual [8]. In adults, the use of gender-specific percentiles has long been abandoned for many reasons, including the fact that such criteria will change as the body weight distribution of the population changes [9].
If BMI indeed represents an index of adiposity, our findings indicate that taller women, on average, tend to be leaner (or shorter women, fatter), while taller men, on average, tend to be fatter (or shorter men, leaner). While a BMI of >30 is commonly used to define obesity, this criterion may not be applicable to very tall or short persons. From a practical standpoint, a "unigender" definition of obesity based upon a BMI of >30 might slightly overestimate its prevalence in tall women and underestimate it in tall men. We must, however, be cautious with this statement, as it is possible that the excess BMI in tall men may be contributed to by an increase in FFM, or the decrease in BMI in tall women may be as well due to a decrease in FFM, with opposite changes at the other end of the height spectrum. We speculate from our findings that estimates of prevalence of obesity may well vary from population to population depending upon the distribution of heights.
Textbook definitions [10] and various groups of experts [6, 11] recommend a uniform definition for obesity, i.e., a BMI >30 kg/m2, because of extensive studies of the relation between BMI and subsequent mortality and because of practical concerns. The incidence of obesity in childhood is arbitrarily established at a level of 5% (above the 95th percentile for gender and age) [6] and increases to approximately 25% in relatively young American adults (BMI >30 kg/m2), with further increase with increasing age [1]. In the population of our study, a BMI of >30 kg/m2 would correspond to percentile 96.5 in males, and 97 in females. In fact, as long as obesity will not be defined according to outcome criteria, its definition must remain empirical or statistical, and the incidence of obesity in a given population is predetermined by the threshold level chosen.
A limitation of our study relates to the caution that one must apply when extending the results to the general population. Indeed, the sample studied was taken from the military, which in theory may not represent well the general population. Definitions of obesity for the military personnel may be different from the general population, and the low prevalence of obesity might in part be due to higher level of fitness. However, contrary to those of many other countries (in particular the USA), the Israeli army is not made up of volunteers, as the military service is compulsory. Even physically unfit individuals serve in the military (where they perform office duties). Only extreme medical conditions may prevent an individual from serving, and one would not want to include these excluded individuals in "normal" curves. We therefore believe that the sample studied fairly represented the general population. However, in view of the fact that the data presented were obtained on military personnel, further research is needed to confirm our research in the general population.
Ethnicity was found to influence BMI significantly. Of importance is the fact that in multiple regression, the relationship between height and BMI was found to be significant even after introducing ethnicity as a possible confounder.
We conclude that in young adults, BMI is affected in a weak, but statistically significant and opposite manner in males and females. In this age range, when childhood percentiles of BMI are no longer available, it may not be appropriate to use a BMI >30 kg/m2 as a threshold for obesity. We suggest using a BMI of 28.6 kg/m2 in females and 29.1 kg/m2 in males as the upper limit of the normal range (percentile 95), although further research may be needed to confirm these values in the general population.
Received October 10, 2002. Accepted October 21, 2003.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  R. Ghosh and P. Bharati Nutritional Status of Adults among Munda and Pod Populations in a Peri Urban Area of Kolkata City, India Asia Pac J Public Health, June 1, 2006; 18(2): 12 - 20. [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
