The changing shape of the body mass index distribution curve in the population: implications for public health policy to reduce the prevalence of adult obesity.
View Article:
PubMed Central - PubMed
Affiliation: Department of Medicine (Geriatrics), University of Mississippi Medical Center, Clinical Research Program, 876A Lakeland Dr, Building LK, Jackson, MS 39216, USA. apenman@medicine.umsmed.edu
Many public health practitioners, policy makers, and epidemiologists worldwide have embraced Geoffrey Rose's population-based prevention concept, a strategy of disease prevention that aims to shift the population distribution of a risk factor in a favorable direction by applying interventions to an entire population... We believe that this may not necessarily be the best strategy for the prevention of adult obesity in populations and that the issue deserves closer examination and discussion. Using cross-sectional data from INTERSALT (International Study of Sodium, Potassium, and Blood Pressure), a study of 52 centers in 32 countries with a range of geographic, social, and economic circumstances, Rose et al found strong, statistically significant correlations between the mean values of various cardiovascular disease risk factors and the corresponding prevalences of deviants (extreme values of that risk factor, at or above a certain cut point)... For example, there was a high, statistically significant correlation between the mean value of body mass index (BMI) and the prevalence of overweight; similar high correlations were found between the mean systolic blood pressure (BP) and the prevalence of hypertension, mean weekly alcohol intake and prevalence of heavy drinking, and mean urinary 24-hour sodium excretion and prevalence of high sodium intake... For public health policy, the implication of Rose's statement was that the reduction of disease prevalence requires the application of interventions to all members of the population, not just those in the "upper tail" of the distribution who are at greatest risk; the aim of this approach is to shift downward, or to the left along the X-axis, the entire population distribution of a risk factor... However, we believe there are several problems with this interpretation, at least with respect to the population distribution of BMI... One problem is that Rose compared cross-sectional data from different countries, not from a single country at different times, and interpreted the results to mean that, in a single population over time, a change in the location of the distribution curve similar to the one he observed in data from more than one country would take place... The limited published data from same-population studies that show or discuss changes in the BMI distribution suggest that the population distribution of BMI has become increasingly skewed over time with little or no upward shifting of the entire distribution curve... Finally, cross-sectional data from the Mississippi adult population for the years 1990 through 2003 show that the population distribution of BMI is positively skewed and has become increasingly skewed over time (Figure 2)... The values of many biologic variables are determined by multifactorial processes; if these processes have additive effects, then the values will be normally distributed... However, the growth of living tissues likely proceeds by multiplicative effects, and measures of body size (such as BMI) are more likely to follow a skewed, possibly log-normal, distribution... We believe that the adult population distribution of BMI is more correctly described by a positively skewed distribution and that over time the degree of skewing has increased; that is, there is proportionately much more shifting of the distribution curve at the upper end than the lower (Figures 1B and 1C)... Swinburn and Egger refer to this as the "the runaway weight gain train"... These conclusions are based on limited national and state data, and further analysis is required... This refocusing suggests a return, in part, to the concept of high-risk prevention... Rose hinted at the possibility of this approach for weight reduction in the population, an approach that is especially important considering the J-shaped relationship between body weight (or BMI) and overall mortality. |
Related In:
Results -
Collection
getmorefigures.php?uid=PMC1636707&req=5
Mentions: One problem is that Rose compared cross-sectional data from different countries, not from a single country at different times, and interpreted the results to mean that, in a single population over time, a change in the location of the distribution curve similar to the one he observed in data from more than one country would take place. In other words, a kind of evolution would occur in a population; for example, in response to changing dietary and lifestyle practices, the entire population distribution curve of a risk factor would move upward, or to the right (Figure 1A). However, Rose lacked longitudinal (or time series) data from a single population to support such an evolution of the distribution over time, stating merely that, "It is hard to see how it could fail also to apply to temporal changes within a population. . ." (5). The limited published data from same-population studies that show or discuss changes in the BMI distribution suggest that the population distribution of BMI has become increasingly skewed over time with little or no upward shifting of the entire distribution curve. For example, in the Minnesota Heart Health Program, the greatest increase over time in BMI for both men and women occurred in the upper part of the distribution curve (9). In an adult Norwegian population, the BMI distribution curve shifted to the right over time (10). A graph of National Health and Nutrition Examination Survey III data for 1988 through 1994 showed increasing skewness in the distribution of BMI for all sex–age groups and a greater shift in the upper part of the distribution (11,12). Finally, cross-sectional data from the Mississippi adult population for the years 1990 through 2003 show that the population distribution of BMI is positively skewed and has become increasingly skewed over time (Figure 2) (13). |
View Article: PubMed Central - PubMed
Affiliation: Department of Medicine (Geriatrics), University of Mississippi Medical Center, Clinical Research Program, 876A Lakeland Dr, Building LK, Jackson, MS 39216, USA. apenman@medicine.umsmed.edu
Many public health practitioners, policy makers, and epidemiologists worldwide have embraced Geoffrey Rose's population-based prevention concept, a strategy of disease prevention that aims to shift the population distribution of a risk factor in a favorable direction by applying interventions to an entire population... We believe that this may not necessarily be the best strategy for the prevention of adult obesity in populations and that the issue deserves closer examination and discussion. Using cross-sectional data from INTERSALT (International Study of Sodium, Potassium, and Blood Pressure), a study of 52 centers in 32 countries with a range of geographic, social, and economic circumstances, Rose et al found strong, statistically significant correlations between the mean values of various cardiovascular disease risk factors and the corresponding prevalences of deviants (extreme values of that risk factor, at or above a certain cut point)... For example, there was a high, statistically significant correlation between the mean value of body mass index (BMI) and the prevalence of overweight; similar high correlations were found between the mean systolic blood pressure (BP) and the prevalence of hypertension, mean weekly alcohol intake and prevalence of heavy drinking, and mean urinary 24-hour sodium excretion and prevalence of high sodium intake... For public health policy, the implication of Rose's statement was that the reduction of disease prevalence requires the application of interventions to all members of the population, not just those in the "upper tail" of the distribution who are at greatest risk; the aim of this approach is to shift downward, or to the left along the X-axis, the entire population distribution of a risk factor... However, we believe there are several problems with this interpretation, at least with respect to the population distribution of BMI... One problem is that Rose compared cross-sectional data from different countries, not from a single country at different times, and interpreted the results to mean that, in a single population over time, a change in the location of the distribution curve similar to the one he observed in data from more than one country would take place... The limited published data from same-population studies that show or discuss changes in the BMI distribution suggest that the population distribution of BMI has become increasingly skewed over time with little or no upward shifting of the entire distribution curve... Finally, cross-sectional data from the Mississippi adult population for the years 1990 through 2003 show that the population distribution of BMI is positively skewed and has become increasingly skewed over time (Figure 2)... The values of many biologic variables are determined by multifactorial processes; if these processes have additive effects, then the values will be normally distributed... However, the growth of living tissues likely proceeds by multiplicative effects, and measures of body size (such as BMI) are more likely to follow a skewed, possibly log-normal, distribution... We believe that the adult population distribution of BMI is more correctly described by a positively skewed distribution and that over time the degree of skewing has increased; that is, there is proportionately much more shifting of the distribution curve at the upper end than the lower (Figures 1B and 1C)... Swinburn and Egger refer to this as the "the runaway weight gain train"... These conclusions are based on limited national and state data, and further analysis is required... This refocusing suggests a return, in part, to the concept of high-risk prevention... Rose hinted at the possibility of this approach for weight reduction in the population, an approach that is especially important considering the J-shaped relationship between body weight (or BMI) and overall mortality.