Trends and Disparities in Provider Diagnosis of Overweight Analysis of NHANES 1999–2004
Rates of overweight and obesity are disproportionately high within minority populations. This study examined the trends in provider diagnosis of overweight from 1999 to 2004 and examined whether there were differences in provider diagnosis based on race/ethnicity. We examined data from 4,071 adults with BMI ≥30 who participated in the National Health and Nutrition Examination Surveys (NHANES) (1999–2004). Provider diagnosis was determined by self-report. From 1999 to 2004, the provider diagnosis of overweight decreased from 71 to 64% (P = 0.003). After controlling for potential confounders, non-Hispanic blacks and Mexican Americans were less likely to report a provider diagnosis of overweight compared to non-Hispanic whites. Odds ratio (OR) (95% confidence interval (CI)) for non-Hispanic blacks was 0.6 (95% CI, 0.4–0.8) and for Mexican Americans was 0.7 (95% CI, 0.4–1.0) compared to non-Hispanic whites. Reasons for this disparity warrant further investigation.
Obesity prevalence is disproportionately higher among minority groups, with 45% of non-Hispanic blacks and 36% of Mexican Americans categorized as obese (1). Race/ethnic disparities in achieving goals of treatment in hypertension and diabetes, potential consequences of obesity, are well recognized (2,3). Studies have not examined whether there are race/ethnic disparities in obesity treatment.
Despite increasing obesity prevalence, there are decreasing trends in the number of patients who report receiving weight loss advice from a health-care provider (4). Prior studies, however, did not have an ethnically diverse population and were limited in assessing whether these trends were similar among minority patients.
Using National Health and Nutrition Examination Survey (NHANES) data, we (i) examined trends in provider diagnosis of overweight (defined as participant report of being told they were overweight by a health-care professional), (ii) determined if there were differences in provider diagnosis based on race/ethnicity, and (iii) examined the association between provider diagnosis and weight-related perceptions.
Methods and Procedures
NHANES is a series of cross-sectional surveys that include a nationally representative sample of the US noninstitutionalized civilian population and over samples non-Hispanic black and Mexican-American populations to provide stable estimates for these groups (5).
The current analyses are limited to adults (≥20 years) in survey years 1999–2000, 2001–2002, and 2003–2004. Of these 15,332 individuals, those with missing BMI data (n = 1,621), missing data in response to the question of provider diagnosis of overweight (n = 2), those who classified themselves as a race/ethnic group other than non-Hispanic white, non-Hispanic black, or Mexican American (n = 283), or those with BMI <30 (n = 9,319) were excluded, leaving 4,107 obese individuals for the current analyses (1,289 in 1999–2000, 1,355 in 2001–2002, and 1,463 in 2003–2004).
We defined “provider diagnosis” of overweight as an affirmative response to the question: “Has a doctor or health professional ever told you that you were overweight?” (5). This question was assessed by interview during NHANES data collection.
Age, sex, race/ethnicity, education level, health insurance status, having a routine health-care provider, and number of provider visits in the past year were assessed by interview.
Height, weight, blood pressure, and fasting glucose were measured using standardized protocols (5). BMI was calculated as weight in kilograms divided by height in meters squared and categorized as: obese class I (BMI of 30–34.9 kg/m2), obese class II (BMI of 35–39.9 kg/m2), and obese class III (BMI of ≥40 kg/m2 (6). Hypertension was defined as a systolic blood pressure ≥140 mm Hg, a diastolic blood pressure ≥90 mm Hg, and/or self-reported antihypertensive medication use. Diabetes was defined as fasting blood glucose of ≥126 mg/dl or self-reported use of antidiabetic medications. Cardiovascular disease was based on participant report of a history of myocardial infarction or stroke.
To assess weight-related perceptions, we analyzed participant responses to the following questions: 1. Do you consider yourself now to be overweight, underweight, or about the right weight? 2. Would you like to weigh more, less, or stay about the same? Due to very few “underweight” responses to question 1 and very few “more” responses to question 2, these responses were set to missing for this analysis.
χ2-Tests compared prevalence estimates of provider diagnosis of overweight among obese individuals across time periods within categories of demographic and health history variables, and within time periods across categories of demographic and health history variables. We tested multiplicative interaction terms between time and demographic variables and between time and health history variables. Independent correlates of provider diagnosis were determined via multiple logistic regression analysis with data pooled across 1999–2004. Based on prior studies (7,8), we included race/ethnicity, survey year, obesity class, gender, education, age, health insurance, having a routine health-care provider, number of provider visits, and comorbid conditions in the models. Multivariable logistic regression assessed race/ethnic differences in weight-related perceptions controlling for age, sex, education, and obesity class. Analyses were conducted using SUDAAN 9.0.1 (RTI International, Research Triangle Park, NC), employing techniques appropriate to the complex survey design of NHANES 1999–2004. P values <0.05 were significant.
We analyzed data from 4,107 obese adults; unweighted proportions were 48.4% non-Hispanic white, 26.0% non-Hispanic black, and 25.6% Mexican American. Non-Hispanic blacks and Mexican Americans were more likely to have missing data for BMI; however, non-Hispanic blacks were more likely to be eligible for the current analyses due to higher prevalence of obesity. In 1999, 71.2% of obese adults reported provider diagnosis of overweight that decreased in subsequent years to 66.7% in 2001 and 63.7% in 2004 (P value for trend = 0.003). In each survey period, non-Hispanic blacks and Mexican Americans were significantly less likely to be diagnosed as overweight compared to non-Hispanic whites (P < 0.001) (see Table 1). Women, having health insurance, comorbid illnesses, and higher number of provider visits, were associated with increased provider diagnosis in univariate analyses. Interactions between time and age (P = 0.003), and between time and health insurance (P = 0.007) were significant such that the decrease in provider diagnosis was greatest for middle-aged individuals and for individuals with insurance. There were no significant interactions between time and gender (P = 0.72) or between time and race/ethnicity (P = 0.18).
Table 1. Prevalence (s.e.) of provider diagnosis of overweight among obese individuals, by demographic and health history characteristics as well as by time point, from 1999 to 2004
After controlling for potential confounders, non-Hispanic blacks and Mexican Americans remained less likely to be diagnosed as overweight when compared to non-Hispanic whites (Table 2). Multivariate odds ratio (OR) for provider diagnosis in non-Hispanic blacks was 0.6 (95% confidence interval (CI), 0.4–0.8) and in Mexican Americans was 0.7 (95% CI, 0.4–1.0). The degree of obesity increased the odds of provider diagnosis with adults in class III obesity having close to a ninefold increase in the odds of provider diagnosis compared to class I obesity (OR = 8.6, 95% CI 5.0–14.9). Diabetes was the only comorbid illness that independently increased the odds of provider diagnosis (OR = 2.9, 95% CI 2.0–4.4).
Table 2. Independent correlates of provider diagnosis via multivariable logistic regression analysis among obese individuals, 1999–2004 pooled analysis
Participants with a provider diagnosis were more likely to perceive themselves as overweight rather than the right weight (95.9% among those reporting a provider diagnosis vs. 84.2% among those not reporting a provider diagnosis; P < 0.001) and preferred to weigh less rather than stay the same weight (96.7% vs. 87.6%, respectively; P < 0.001). After adjustment for potential confounders, non-Hispanic blacks were five times more likely to perceive themselves as being the right weight rather than overweight (OR = 5.0, 95% CI 2.3–10.7) and were four times more likely to prefer to weigh the same rather than weigh less (OR = 3.9, 95% CI 1.9–8.2) when compared to non-Hispanic whites (P < 0.001 for both). Non-Hispanic whites and Mexican Americans did not differ in weight-related perceptions.
Using a nationally representative multiethnic sample with measured height and weight, we have demonstrated decreasing trends in provider diagnosis of overweight from 1999 to 2004. Non-Hispanic blacks and Mexican Americans were less likely to be diagnosed compared to non-Hispanic whites highlighting a potential disparity in obesity management.
Data from the Behavioral Risk Factor Surveillance System demonstrated that between 1994 and 2000 there was a decline (from 44 to 40%) in obese respondents who reported being advised to lose weight by a health-care professional (8). Our findings indicate a similar trend with provider diagnosis suggesting that despite increasing obesity prevalence, intervention by health-care providers is declining. Of concern are the lower rates of diagnosis among non-Hispanic blacks and Mexican Americans. Prior studies have cited time, beliefs that weight loss counseling is ineffective, and lack of patient motivation as barriers to health-care providers addressing obesity (9,10). Given the complexity of addressing obesity and the influence that culturally related factors such as diet may have on obesity, we question whether patient race/ethnicity is an unrecognized barrier in providers addressing obesity.
We additionally found that participants with provider diagnosis were more likely to perceive themselves as overweight and preferred to weigh less. Although it may seem unnecessary to diagnose a patient as overweight or obese, it is possible that adults living in communities with higher obesity prevalence may view obesity as the social norm and not identify it as a health problem. Indeed, non-Hispanic blacks have the highest prevalence of obesity and were more likely to consider themselves the right weight in comparison to other groups, supporting previous findings by Yancey et al. and Burroughs et al. (11,12).
This study has several strengths including a large sample size with a diverse sampling of individuals and the use of measured anthropometrics. Our primary outcome of provider diagnosis is limited, however, because it is self-reported and subject to recall bias and random error. Additionally, the context in which adults were told about their weight and their weight at the time of diagnosis may affect their ability to recall being told. Finally, non-Hispanic blacks and Mexican Americans were more likely to have missing data for BMI and provider diagnosis, and our results may underestimate the problem of decreased provider diagnosis.
In conclusion, our study suggests that provider diagnosis of overweight among obese individuals has decreased between 1999 and 2004; this is despite prior reports of dramatically increasing obesity prevalence during this time. Non-Hispanic blacks and Mexican Americans have lower rates of provider diagnosis. Reasons for this disparity and its impact warrant further investigation.
The authors declared no conflict of interest.