Lester R. Curtin

Prevalence and trends in obesity among US adults, 1999-2008.

CONTEXT The prevalence of obesity increased in the United States between 1976-1980 and 1988-1994 and again between 1988-1994 and 1999-2000. OBJECTIVE To examine trends in obesity from 1999 through 2008 and the current prevalence of obesity and overweight for 2007-2008. DESIGN, SETTING, AND PARTICIPANTS Analysis of height and weight measurements from 5555 adult men and women aged 20 years or older obtained in 2007-2008 as part of the National Health and Nutrition Examination Survey (NHANES), a nationally representative sample of the US population. Data from the NHANES obtained in 2007-2008 were compared with results obtained from 1999 through 2006. MAIN OUTCOME MEASURE Estimates of the prevalence of overweight and obesity in adults. Overweight was defined as a body mass index (BMI) of 25.0 to 29.9. Obesity was defined as a BMI of 30.0 or higher. RESULTS In 2007-2008, the age-adjusted prevalence of obesity was 33.8% (95% confidence interval [CI], 31.6%-36.0%) overall, 32.2% (95% CI, 29.5%-35.0%) among men, and 35.5% (95% CI, 33.2%-37.7%) among women. The corresponding prevalence estimates for overweight and obesity combined (BMI > or = 25) were 68.0% (95% CI, 66.3%-69.8%), 72.3% (95% CI, 70.4%-74.1%), and 64.1% (95% CI, 61.3%-66.9%). Obesity prevalence varied by age group and by racial and ethnic group for both men and women. Over the 10-year period, obesity showed no significant trend among women (adjusted odds ratio [AOR] for 2007-2008 vs 1999-2000, 1.12 [95% CI, 0.89-1.32]). For men, there was a significant linear trend (AOR for 2007-2008 vs 1999-2000, 1.32 [95% CI, 1.12-1.58]); however, the 3 most recent data points did not differ significantly from each other. CONCLUSIONS In 2007-2008, the prevalence of obesity was 32.2% among adult men and 35.5% among adult women. The increases in the prevalence of obesity previously observed do not appear to be continuing at the same rate over the past 10 years, particularly for women and possibly for men.

Prevalence of High Body Mass Index in US Children and Adolescents, 2007-2008

CONTEXT The prevalence of high body mass index (BMI) among children and adolescents in the United States appeared to plateau between 1999 and 2006. OBJECTIVES To provide the most recent estimates of high BMI among children and adolescents and high weight for recumbent length among infants and toddlers and to analyze trends in prevalence between 1999 and 2008. DESIGN, SETTING, AND PARTICIPANTS The National Health and Nutrition Examination Survey 2007-2008, a representative sample of the US population with measured heights and weights on 3281 children and adolescents (2 through 19 years of age) and 719 infants and toddlers (birth to 2 years of age). MAIN OUTCOME MEASURES Prevalence of high weight for recumbent length (> or = 95th percentile of the Centers for Disease Control and Prevention growth charts) among infants and toddlers. Prevalence of high BMI among children and adolescents defined at 3 levels: BMI for age at or above the 97th percentile, at or above the 95th percentile, and at or above the 85th percentile of the BMI-for-age growth charts. Analyses of trends by age, sex, and race/ethnicity from 1999-2000 to 2007-2008. RESULTS In 2007-2008, 9.5% of infants and toddlers (95% confidence interval [CI], 7.3%-11.7%) were at or above the 95th percentile of the weight-for-recumbent-length growth charts. Among children and adolescents aged 2 through 19 years, 11.9% (95% CI, 9.8%-13.9%) were at or above the 97th percentile of the BMI-for-age growth charts; 16.9% (95% CI, 14.1%-19.6%) were at or above the 95th percentile; and 31.7% (95% CI, 29.2%-34.1%) were at or above the 85th percentile of BMI for age. Prevalence estimates differed by age and by race/ethnic group. Trend analyses indicate no significant trend between 1999-2000 and 2007-2008 except at the highest BMI cut point (BMI for age > or = 97th percentile) among all 6- through 19-year-old boys (odds ratio [OR], 1.52; 95% CI, 1.17-2.01) and among non-Hispanic white boys of the same age (OR, 1.87; 95% CI, 1.22-2.94). CONCLUSION No statistically significant linear trends in high weight for recumbent length or high BMI were found over the time periods 1999-2000, 2001-2002, 2003-2004, 2005-2006, and 2007-2008 among girls and boys except among the very heaviest 6- through 19-year-old boys.

Cardiorespiratory Fitness Levels Among US Adults 20–49 Years of Age: Findings From the 1999–2004 National Health and Nutrition Examination Survey

Data from the 1999-2004 National Health and Nutrition Examination Survey were used to describe the distribution of cardiorespiratory fitness and its association with obesity and leisure-time physical activity (LTPA) for adults 20-49 years of age without physical limitations or indications of cardiovascular disease. A sample of 7,437 adults aged 20-49 years were examined at a mobile examination center. Of 4,860 eligible for a submaximal treadmill test, 3,250 completed the test and were included in the analysis. The mean maximal oxygen uptake ( max) was estimated as 44.5, 42.8, and 42.2 mL/kg/minute for men 20-29, 30-39, and 40-49 years of age, respectively. For women, it was 36.5, 35.4, and 34.4 mL/kg/minute for the corresponding age groups. Non-Hispanic black women had lower fitness levels than did non-Hispanic white and Mexican-American women. Regardless of gender or race/ethnicity, people who were obese had a significantly lower estimated maximal oxygen uptake than did nonobese adults. Furthermore, a positive association between fitness level and LTPA participation was observed for both men and women. These results can be used to track future population assessments and to evaluate interventions. The differences in fitness status among population subgroups and by obesity status or LTPA can also be used to develop health policies and targeted educational campaigns.

Smoke-free laws and secondhand smoke exposure in US non-smoking adults, 1999–2002

Objectives: To investigate the relationship between smoke-free law coverage and secondhand smoke (SHS) exposure in the United States non-smoking adult population. Design: We used data from the 1999–2002 National Health and Nutrition Examination Survey, a cross-sectional survey designed to monitor the health and nutritional status of the US population. Serum cotinine levels were available for 5866 non-smoking adults from 57 survey locations. Each location was categorised into one of three groups indicating extensive, limited, and no coverage by a smoke-free law. Main outcome measures: The proportion of adults with SHS exposure, defined as having serum cotinine levels ⩾ 0.05 ng/ml. Results: Among non-smoking adults living in counties with extensive smoke-free law coverage, 12.5% were exposed to SHS, compared with 35.1% with limited coverage, and 45.9% with no law. Adjusting for confounders, men and women residing in counties with extensive coverage had 0.10 (95% confidence interval (CI) 0.06 to 0.16) and 0.19 (95% CI 0.11 to 0.34) times the odds of SHS exposure compared to those residing in counties without a smoke-free law. Conclusions: These results support the scientific evidence suggesting that smoke-free laws are an effective strategy for reducing SHS exposure.

Multiple imputation of missing dual-energy X-ray absorptiometry data in the National Health and Nutrition Examination Survey†

In 1999, dual-energy x-ray absorptiometry (DXA) scans were added to the National Health and Nutrition Examination Survey (NHANES) to provide information on soft tissue composition and bone mineral content. However, in 1999-2004, DXA data were missing in whole or in part for about 21 per cent of the NHANES participants eligible for the DXA examination; and the missingness is associated with important characteristics such as body mass index and age. To handle this missing-data problem, multiple imputation of the missing DXA data was performed. Several features made the project interesting and challenging statistically, including the relationship between missingness on the DXA measures and the values of other variables; the highly multivariate nature of the variables being imputed; the need to transform the DXA variables during the imputation process; the desire to use a large number of non-DXA predictors, many of which had small amounts of missing data themselves, in the imputation models; the use of lower bounds in the imputation procedure; and relationships between the DXA variables and other variables, which helped both in creating and evaluating the imputations. This paper describes the imputation models, methods, and evaluations for this publicly available data resource and demonstrates properties of the imputations via examples of analyses of the data. The analyses suggest that imputation helps to correct biases that occur in estimates based on the data without imputation, and that it helps to increase the precision of estimates as well. Moreover, multiple imputation usually yields larger estimated standard errors than those obtained with single imputation.

Limitations of follicle-stimulating hormone in assessing menopause status: findings from the National Health and Nutrition Examination Survey (NHANES 1999-2000)*.

Objective: We used data from the National Health and Nutrition Examination Survey (NHANES 1999-2000) to: establish new population-based estimates for follicle-stimulating hormone (FSH) and luteinizing hormone (LH); identify factors associated with FSH; and assess its efficacy in distinguishing among women in the reproductive, menopause transition, and postmenopausal stages. Design: Nationally representative sample of 576 women aged 35 to 60 years examined during NHANES 1999-2000. Results Levels of FSH and LH increased significantly with reproductive stage. (Geometric mean FSH levels for successive stages: reproductive, 7.0 mIU/mL, SE 0.4; menopause transition, 21.9 mIU/mL, SE 3.7; and postmenopause, 45.7 mIU/mL, SE 4.3). There was considerable overlap, however, among distributions of FSH by stage. Only age and reproductive stage were significantly associated with FSH in multivariable analysis. FSH cutoff points between the reproductive and menopause transition stages [FSH = 13 mIU/mL, sensitivity 67.4% (95% CI 50.0-81.1), specificity 88.1% (95% CI 81.1-92.8)] and between the menopause transition and postmenopause stages [FSH = 45 mIU/mL, sensitivity 73.6% (95% CI 60.1-83.7), specificity 70.6% (95% CI 52.4-84.0)] were neither sensitive nor very specific. Conclusions Age and reproductive stage are the most important determinants of FSH levels in US women; however, FSH by itself has limited utility in distinguishing among women in different reproductive stages.

Statistical and Practical Issues in the Design of a National Probability Sample of Births for the Vanguard Study of the National Children's Study

The National Childrens Study is a national household probability sample designed to identify 100,000 children at birth and follow the sampled children for 21 years. Data from the study will support examining numerous hypotheses concerning genetic and environmental effects on the health and development of children. The goals of the study present substantial challenges. For example, the need for preconception, prenatal, and postnatal data requires identifying women in the early stages of pregnancy, the collection of many types of data, and the retention of the children over time. In this paper, we give an overview of the sample design used in a pilot study called the Vanguard Study, and highlight the approaches used to address these challenges. We will also describe the rationale for the sampling choices made at each stage, the unique organizational structure of the NCS and issues we expect to face during implementation.

The 2000 Centers for Disease Control and Prevention growth charts: several insights after 8 years.

This paper explores three issues related to the 2000 Centers for Disease Control and Prevention growth charts. First, it clarifies the methods that were used to create the charts as it has become apparent that the smoothing techniques have been somewhat misunderstood. The techniques included smoothing-selected percentiles between and including the 3rd and 97th percentiles and then approximating these smoothed curves using a procedure to provide the transformation parameters, lambda, mu, and sigma. Only the selected percentiles were used in this process due to small sample sizes beyond these percentiles. Second, given the concern that the infant charts were created with relatively few data points in the first few months of life, it compares the original observed percentiles with percentiles that include newly available US national data for the first few months of life. Third, it discusses the issues that arise if a 99th percentile is extrapolated based on the lambda, mu, and sigma parameters. The 99th percentile of the body mass index-for-age chart has been recommended to identify extremely obese children, yet the 97th percentile is the highest available percentile on the Centers for Disease Control and Prevention growth charts.

Population-based surveys and their role in public health1

Timely and accurate health information is needed to assess the population’s health status and to determine current and future needs for healthcare services and resources. In particular, detailed information on health and healthcare characteristics of the U.S. population is extremely important to those responsible for the planning, implementation, and assessment of public health programs, as well as the formulation of public health policy. Population-based sample surveys are often used to obtain this needed information. By definition, a population-based sample survey is defined as a study involving a subset (or sample) of individuals selected from a larger population.1 Sample surveys can be based on either probability samples or nonprobability samples. In a probability sample survey, each member of the surveyed population of interest has a known, nonzero probability of selection. Equally important, people are sampled according to a method of random selection. Because of these features, probability samples produce results that can be generalized to the entire sampled population and permit unbiased estimates of characteristics of the population. In addition, the error associated with studying a sample rather than the total population can be estimated. In contrast, surveys based on a nonprobability sample do not have these features and are often referred to as convenience or quota samples. This commentary discusses probability sample surveys. The Centers for Disease Control and Prevention (CDC) conducts a variety of large, population-based, sample surveys that provide critical information about the prevalence and distribution of diseases and disease risk factors; disease etiology; health-related knowledge, attitudes, and beliefs; healthcare utilization; health behaviors; and the implementation of disease prevention and health promotion activities. Surveys also produce important data used in various types of epidemiologic studies. Evidence of the role of probability sample health surveys and how their results have impacted our society is provided by a few key examples. With its direct physical and biochemical measurements, the National Health and Nutrition Examination Survey (NHANES) provides data to assess the health and nutritional status of the U.S. population; monitor the prevalence of major diseases and risk factors; and establish reference standards for child growth and development, blood pressure, cholesterol, and various other biochemical measurements from blood and urine samples. For example, in 1996, the U.S. Food and Drug Administration mandated the fortification of grain and cereal products with folic acid, a vitamin shown to prevent birth defects such as spina bifida and other neural tube defects. Over time, NHANES surveys have measured serum and red blood–cell folate concentrations for women of childbearing age. Comparing results of the 1988–1994 NHANES to the 1999 survey shows substantial increases in folate concentrations among women of childbearing age.2 NHANES surveys have also been useful in monitoring the public health problem of elevated blood-lead levels, which cause a variety of adverse health effects, especially in cognitive development and behavioral problems among young children. The second NHANES (1976–1980) documented that 88% of young children in the United States had elevated blood-lead levels.3 NHANES also showed a decline in blood-lead levels over the survey period that coincided with the removal of lead from gasoline, food containers, plumbing, residential paint, and other consumer products. The survey data documented the success of these primary and secondary prevention efforts. Phase 2 of the third NHANES (1991–1994) showed that by the early 1990s the prevalence of elevated blood-lead levels in young children had plummeted to 4.4%—a 95% reduction from the earlier survey.4 Monitoring personal behaviors associated with health is another important function of surveys. The CDC’s Behavioral Risk Factor Surveillance System survey, for example, is a large telephone survey that includes questions regarding nutritional intake, alcohol and tobacco use, exercise, sexual practices, and healthcare utilization. One analysis from this survey found that from 1991 to 1995, prevalence of heavy alcohol use among pregnant women, at levels that have been shown to cause birth defects and developmental disabilities in the unborn child, had increased substantially.5 This finding boosted interest in public health interventions to reduce alcohol use among pregnant women. Determining personal knowledge, attitudes, and beFrom the National Center for Health Statistics, Centers for Disease Control and Prevention, Hyattsville, Maryland Address correspondence to: Trena M. Ezzati-Rice, MS, 6525 Belcrest Road, Room 915, Hyattsville, MD. E-mail: tme1@cdc.gov. Address reprint requests to: Centers for Disease Control and Prevention, National Immunization Program Resource Center, 1600 Clifton Road NE, Mailstop E-34, Atlanta, Georgia 30333. Fax: (404) 639-8828.