This study used data from NHANES III (1988–1994) and NHANES 1999–2004. NHANES is a multi-stage, stratified probability sample of non-institutionalized U.S. civilians, ages 2 months and older. The survey is conducted by the National Center for Health Statistics (NCHS), Centers for Disease Control and Prevention (CDC) and includes administration of questionnaires, a physical exam, and a battery of laboratory tests. Details of the questionnaires, the exam, and the tests have been described elsewhere [17–20].
Adult participants (≥ 20 years of age) classified as overweight, identified by a measured BMI: 25.0–29.9 kg/m2 (NHANES III, n = 5,653, NHANES 1999–2004, n = 4,645), were included in our sample. Respondents were excluded if they were pregnant or under the age of 20 (children/adolescents). Pregnancy status was determined using self-reported information as well as results of the urine test administered during the medical examination.
Height and weightwere measured by trained examiners during the extensive medical evaluation at the Mobile Examination Center (MEC). Participants were weighed using a scale with a digital display readout, and height was measured using a stadiometer.
Body Mass Index (BMI)was calculated based on these weight and height measurements and calculated as weight (kg)/height (meters) squared. Participants were then classified as obese (≥ 30.0 kg/m2), overweight (25.0–29.9 kg/m2), or normal weight (18.5 – 24.9 kg/m2), which is consistent with the NIH Clinical Guidelines on the Identification, Evaluation and Treatment of Overweight and Obesity in Adults – The Evidence Report .
Weight perceptionwas ascertained asking subjects during the face-to-face interview if they considered themselves to be overweight, underweight, or about the right weight. Less than 0.01% of respondents were missing weight status data, and these individuals were excluded from analyses examining weight distortion. At the time respondents were asked to report their weight status, they were also told that their weight would be measured at a follow-up session.
Poverty Income Ratio
Respondents were asked to select the category of income that most accurately reflected their total combined family income over the last 12 months. Reported income was then used to develop the poverty income ratio (PIR), which is the ratio of income to the family's appropriate poverty threshold. Using the PIR allows income data to be compared across various survey years. Income was categorized, as suggested by NCHS, using the PIR eligibility cut points for the United States Department of Agriculture (USDA) food assistance programs, into the following categories: low (0.000–1.850), middle (1.851–3.500), and high (3.501+) incomes . Individuals with missing PIR data were excluded from all income analyses. This included 9.7% (551) overweight adults from NHANES III and 8.6% (399) overweight adults from NHANES 1999–2004.
Non-age specific estimates were age-standardized to the 2000 Census using the standard population structure as suggested by NCHS. For specific estimates examining age, the following categories were used: 20–34, 35–49, 50–64, and 65+. These categories enable exploration of how weight perception differs across the age spectrum while providing a sufficient number of subjects in each group.
NHANES participants during both survey periods were asked to select the race and ethnicity that they most readily identified with. Responses were then systematically collapsed into the following categories of Non-Hispanic (NH) Whites, Non-Hispanic (NH) Blacks, Mexican-Americans and Other [21, 22]. Only the first three categories are reported on in this paper.
Analysis of the NHANES data was conducted using SAS (version 9, SAS Institute Inc, Cary, NC) and SUDAAN (version 9, Research Triangle Institute, Research Triangle Park, NC) statistical software packages. To account for unequal probabilities of selection and non-response, all our analyses incorporated sample weights in order to provide national estimates of population proportions. Standard errors were calculated in SUDAAN based on the Taylor Series Linearization method . Analyses were conducted separately for each survey period (NHANES III 1988–1994 and NHANES 1999–2004). The study population was stratified by gender, race, age and income. T-tests were used to assess differences between each of the individual groups and the reference population, which was defined as white males or white females, ages 20–34 or a PIR of 0–1.85. T-tests were also used to test differences in proportions between the two survey periods. A p-value < 0.05 was considered statistically significant.