Who in America Is Trying To Lose Weight?

Methods

The NHIS is an annual, nationally representative survey of about 45 000 households [4]. Each year, the NHIS comprises a core questionnaire that addresses basic demographic and health information such as acute and chronic conditions and doctor visits for each member of the selected households. In addition, one or more supplements containing items of current health interest are administered to randomly selected members of the household. In 1985, and again in 1990, the Health Promotion and Disease Prevention (HPDP) supplements [5, 6] were designed to yield information to assist in monitoring the health status of the nation as we move toward the year 2000.

The HPDP supplements contain, among other items, several questions designed to elicit information on respondents' self-perceived weight status (overweight, underweight, or about right), on those persons who are trying to lose weight, and on the methods Americans use in attempting to lose weight. Additional sociodemographic information on age, race, sex, education, and income is routinely collected in the NHIS.

In both 1985 and 1990, randomly selected adult persons 18 years and older were administered the HPDP questionnaire. In 1985, 33 630 persons (90% of those eligible) responded to the questionnaire. In 1990, 41 104 persons (86.3% of those eligible) were successfully interviewed for the HPDP Current Health Topics supplement. We excluded persons who had been advised by a physician to lose weight due to hypertension or diabetes. Because education was one of our stratification variables, we also limited our analysis to persons 25 years and older. Thus, for the 1985 data we excluded 4823 persons (14.3%) because of age, 3593 persons (10.7%) because they were advised by a physician to lose weight, and 1043 persons (3.1%) because of inadequate sociodemographic information.

After these exclusions, 24 171 persons (71.9%) remained eligible for analysis in 1985. From the 1990 data, 5170 persons (12.6%) were excluded because of age, 2911 persons (7.1%) were excluded because they had been advised by a physician to lose weight, and 1676 persons (4.1%) were excluded because of inadequate sociodemographic information. After these exclusions, 31 347 persons (76.3%) remained eligible for further analysis in 1990.

The NHIS is a multistage survey with clustering and oversampling of selected population groups [4]. Statistical methods appropriate for a complex sample survey were used to estimate the variances and, hence, the standard errors. The computer program SUDAAN [7, 8] was used to conduct these analyses. Data were weighted to reflect the probability of selection and other factors.

Results

In 1990, 45% of Americans considered themselves overweight: 36.7% of men and 52.0% of women (Table 1). Persons of lower socioeconomic status were generally more likely to consider themselves underweight than were those of higher socioeconomic status. Of the racial-ethnic groups considered, black persons were least likely to consider themselves to be overweight and most likely to consider themselves to be about right; however, actual measurements have shown that black persons, especially black women, are more likely to be overweight compared with the general population [1]. Persons with less than a high school education were least likely to see themselves as overweight and were most likely to consider themselves underweight compared with high school graduates and those with more than a high school education. Similarly, those with a family income greater than $20 000 per year were more likely to think of themselves as overweight.

Persons between 45 and 64 years were most likely to think of themselves as overweight, followed by persons 25 to 44 years old. Persons older than 65 years were the least likely to think of themselves as overweight (35.7%). Widowed and never-married persons were least likely to consider themselves overweight (< 37%), whereas currently married persons were most likely to think of themselves as overweight. The self-perceived weight categories had a similar demographic distribution in 1985 (data not shown).

Between 1985 and 1990, few differences were seen in the proportion of people trying to lose weight when viewed in terms of sociodemographic characteristics (Table 2). In both years, about 23% of men and 40% of women were trying to lose weight. Of the three self-perceived weight items examined, the only difference found between 1985 and 1990 was in the self-perception of those persons trying to lose weight. Among those considering themselves underweight, the percentage of those trying to lose weight increased significantly from 0.7% in 1985 to 4.0% in 1990 (P < 0.001).

Few differences were seen between 1985 and 1990 among those trying to lose weight by eating less (Table 3). Smaller proportions of blacks and those with an annual family income less than $20 000 who were trying to lose weight said they were doing so by eating less in 1990 than in 1985. A larger proportion of Hispanics in 1990 than in 1985 said they were trying to lose weight by eating less.

In every sociodemographic category, statistically significant increases were found between 1985 and 1990 in the proportion of those who were trying to lose weight by increasing their physical activity (Table 3). The only instance in which we found a decrease in physical activity between 1985 and 1990 was among persons with a family income less than $20 000.

Of those who were trying to lose weight, several large and statistically significant differences were found between 1985 and 1990 for the first and second best ways to lose weight (Table 4). In 1990, fewer persons than in 1985 thought the best way to lose weight was by not eating before going to bed or by eating fewer calories. However, more persons in 1990 than in 1985 thought the best way to lose weight was by increasing their physical activity or by eating no fat. In 1990, the most frequently chosen response for the best way to lose weight was to eat fewer calories followed by not eating before going to bed. Increasing physical activity was the third most frequently mentioned best way to lose weight.

After respondents were asked for their perceptions of the best way to lose weight, they were asked for their opinion of the second best way to lose weight. A greater proportion of persons in 1990 than in 1985 chose eating no fat, not eating before going to bed, and eating fewer calories as the second best way to lose weight. Fewer persons in 1990 than in 1985 chose increasing physical activity and eating grapefruit before every meal.

Discussion

These data are encouraging because they suggest that, over time, American adults are becoming more educated about strategies for weight loss. People are now more likely to think that eating fewer calories and increasing physical activity are the best methods to institute weight loss. However, the alarming fact that most women and a large proportion of men consider thenselves overweight remains. Also disturbing is the fact that those persons wh1990 were more likely to feel that they were very overweight compared with 1985 responses. This finding suggests an increase in the incidence of obesity or changes over time in the standards people use to assess their weight. In either case, it suggests that many people consider themselves to have a weight problem and that much work is needed in promoting n

It was most surprising to find that, in 1990, a greater proportion of those persons trying to lose weight actually considered themselves underweight. Although this proportion was small (4.0%), it was significantly greater than the 0.7% noted in 1985.

We acknowledge the inherent sex-based differences that are believed to exist with self-reported weight and height measurements (that is, women are more likely to under-report their weight and men to over-report their height) [10]. In an unpublished study by us, however, body mass index (BMI) (kg/m²) was computed from self-reported heights and weights using 1990 HPDP data. The results were similar to those reported by the National Health and Nutrition Examination Survey (NHANES), which are based on actual anthropometry. Like the NHIS, the NHANES is a nationally representative survey. Further, those who perceived themselves as overweight in the 1990 HPDP survey were most likely to fall in the upper BMI percentiles. For these reasons, we feel that the use of self-reported weight data in this study is valid.

We realize that no diet is completely fat-free and therefore assume that those who responded that eating a diet with no fat was effective for weight loss interpreted the question to refer to a diet low in fat or to the practice of not intentionally eating fat. Alternatively, media messages may have led people to believe that they should eat a no fat diet.

Because of the many health-related problems associated with overweight, it remains a public health concern that deserves continued attention and research so that successful weight management will become more widespread.