- Click here for explanation of classification scheme.

To estimate the expected number of years of life lost (YLL) due to overweight and obesity across the life span of an adult.

Data for the YLL calculation was taken from four sources:

• The Third National Health and Nutrition Examination Survey (NHANES III)
• First National Health and Nutrition Epidemiologic Follow-up Study (NHEFS)
• The NHAMES II Mortality Study (NH2MS)
• The life table for the total population of the United States in 1999.

BMI distribution was derived from the NHANES III Survey. Only black and white subjects were used because the sample size of other groups were generally insufficient to allow reasonable precise estimates.

Recruitment

Data from the US Life Tables (1999), NHANES III (1988-1994), First National Health and Nutrition Epidemiologic Follow-up Study (NHANES I and II, 1971-1992) and NHANES II Mortality Study (1976-1992) were used to derive years of life lost estimates for adults aged 18-85 years.

Design

Meta-analysis 

Blinding used

None 

Intervention

None 

Statistical Analysis (method of calculation)

The YLL due to obesity was estimated by combining three types of information: an estimate of the distribution of BMI for each year of adult life; an estimate of the hazard ratio for death given various BMI levels in each year of adult life; and the probability of death during each year of adult life.

BMI distribution was derived from NHANES III. BMI was divided into ordered, non-overlapping categories for staring with less than 17, 17 to <18, 18 to <19 and so on to the highest category, ≥45kg/m². For each race and gender group, the proportion of individuals within each BMI category was estimated. The proportions were calculated using a smoothing procedure based on moving averages.

To estimate the hazard ratio for death at a given BMI level in each year of adult life, data was combined from the First National Health and Nutrition Epidemiologic Follow-up Study and The NHANES II Mortality Study. Cox regression models were used with mortality as the outcome and BMI, BMI², age, age², and their interaction terms as covariates. Potential age, race and gender interactions were accounted for by fitting separate models to each race and gender group. Interactions of BMI and BMI² with age² were not used because prior research indicated that change in the BMI parameters appears to be roughly linear in age. Smoking status was controlled for in all analyses (including former smoking status).

The life table for the total population of the United States in 1999 was used to derive information on the probability of death during each year of life.

Using the above mentioned data, an estimation was made for each age interval, in each of four race/gender groups, of the probability of death within the interval, which was conditional on having lived to the start of that interval and being in the first BMI category. For a person of a given age in a given BMI category, an estimation of the expected age of death was made. BMI=24 was used as a reference category to calculate the YLL.

Timing of Measurements

YLL was calculated based on data from previously mentioned population based surveys. 

Dependent Variables

YLL (the difference between the number of years of life expected if an individual were obese vs. not obese).

Independent Variables

• Age
• BMI.

Control Variables

• Race
• Gender
• Smoking.

 

• Initial N: Four population-based surveys
• Attrition (final N): As above
• Age: 18-85 years
• Ethnicity: Calculations of YLL due to obesity presented for black and white subjects
• Other relevant demographics: None
• Anthropometrics: None
• Location: United States.

 

Key Findings

• Among whites, the overall pattern suggests a J or U shaped association between BMI and YLL. YLL was between one and nine in individuals with BMI's <17 through 19 and YLL was between nine and 13 in individuals with BMI's ≥35.
• White men aged 20 years with BMI's greater than 45 were estimated to have 13 YLL relative to white men aged 20 years with BMI's of 24
• Among white women, the pattern shows less variability across the range of BMI distribution as a function of age
• White women aged 20-30 years with BMI's greater than 45 are estimated to have eight YLL due to obesity
• The overall pattern of findings was different among blacks
• Among black men, estimated YLL due to obesity did not begin for any age group until a BMI of 32 was reached. Obesity appeared to be associated with a slightly increased life expectancy across much of the moderate obesity range.
• Black men aged 20 years with a BMI greater than 45 had 20 YLL. Black women aged 20 years with a BMI greater than 45 had five YLL.

 

 

 

 

• Obesity has a profound effect on life span
• Optimal BMI (the BMI associated with the greatest longevity) for adults aged 18 to 85 years is approximately 23-25kg/m² in whites and 23-30kg/m² in blacks
• In blacks and whites, the risk of increased YLL is greatest at younger ages
• Obesity is associated with only slightly increased life expectance across much of the overweight and moderately obese range for most age groups in blacks. A consistent reduction in expected life expectance was not observed until BMI's of 37-38 for black women and BMI's 32-33 for black men.
• The difference in the pattern on YLL due to obesity between whites and blacks may be due to a number of factors. Distributions in age, health status, socioeconomic status and other variables that can affect mortality rate may differ by race. Confounding variables may differ by race. Body composition may also differ by race, especially in women.

Government:	National Institutes of Health
Not-for-profit	Arthritis Foundation Other non-profit:

Authors note the following limitations:

• Although elevated BMI is the primary index of obesity used in most studies, it is a proxy of adiposity and may not provide the best estimate of the association between obesity and outcomes such as mortality and YLL
• Due to insufficient sample size, the association between obesity and YLL could not be reliably estimated for other racial/ethnic groups
• Small sample size of individuals in the upper BMI categories
• Did not control for the possible confounding effects of subclinical or occult disease.