The U.S. Preventive Services Task Force (USPSTF) recently recommended obesity screening and weight loss counseling in primary care settings (1), an important response to the obesity and diabetes epidemics. Unfortunately, clinicians already face daily challenges to executing important preventive service recommendations within the time constraints of primary care (2). We hypothesized that technology can help streamline risk assessment and thus increase clinicians' opportunity for educating patients about their personal risk factors. Our study assessed the feasibility and acceptability of a diabetes-risk questionnaire administered to patients by computer touch screen while they were waiting to see a medical practitioner. A secondary purpose was to determine the validity of self-reported weight in assessing diabetes risk.

There are several ways to measure diabetes risk. At the urban clinic where we did our study, the physicians were familiar with and interested in administering the American Diabetes Association's (ADA's) Diabetes Risk Test (DRT) (3). Preventing or controlling diabetes is a priority for this community health center, which serves a predominantly African American urban community (4,5). The DRT is based on body mass index (BMI), age, family history of diabetes, and level of physical activity (3). It is valid when weight is measured using devices such as electronic scales, spring scales, and balance beam scales (2,6). The validity of DRT results based on self-reported weight immediately before a medical visit has not been assessed. In addition to using the DRT to assess risk for diabetes, some physicians use the results to educate patients about diabetes (7,8). However, DRT's feasibility as part of a large computer-administered health assessment has not been evaluated.

Our study was conducted in a federally funded community health center serving a predominantly African American urban community. We recruited English-speaking women aged 18 to 44 years with a primary care appointment from July through October 2003. For this analysis, pregnant women were excluded because we could not assess their nonpregnant weight. The table on which the computers were set up was between the registration area and the waiting room, which allowed all eligible and willing patients to be screened and to participate in our study. All participants were asked to complete a brief touch-screen risk-assessment questionnaire before seeing their health care providers. A summary of each patient's responses to the questionnaire was printed and given to her to take to her provider, or the summary was given directly to the provider at the patient's request. Participants also received educational brochures on all topics covered in the questionnaire. A pen-and-paper questionnaire was also administered to obtain additional information on factors such as smoking, heart disease, and mental health. Because of confusion about the requirements of the newly implemented Health Insurance Portability and Accountability Act (HIPAA), we were not given patient lists from which to recruit participants. Hence, we were not able to calculate the response rate. The study was approved by the institutional review board at the State University of New York at Albany and the health center's patient board.

Of the 268 women who were not pregnant and who agreed to participate in the study, 231 completed the screening questionnaire and survey sufficiently well for their data to be included in these analyses (self-reported weight was missing for 3 women, and scale-measured weight was missing for 34 women). Of the 231 whose data were used for the study, 214 reported no previous diagnosis of diabetes. Table 1 summarizes the demographic characteristics of the study population. Most women were African American, younger than 35 years of age, and had a high school education or higher. Fewer than 10% were college graduates.

The utility of self-administered, computerized screening questionnaires based on USPSTF recommendations for screening and counseling are being assessed in medical settings. Computerized screening in the waiting room preserves time for health care practitioners to focus on personalized counseling in the examination room. Because computerized screening for diabetes risk is based on self-reported weight, it is vital to know whether self-reported weight would compromise the validity of DRT results. We found that, although only 83% of the women's BMIs were correctly classified through self-reports, the risk scores of 93% of women were correctly classified through the DRT. Only 6 (2.6%) of the 231 study women at risk for diabetes were not identified because of incorrect self-reported weight. In addition, the overall specificity was 97.8%.

Because most participants expected to be weighed by a nurse, we assumed that self-reported weight would be reasonably accurate. This was not the case. Our findings were similar to those of other studies of self-reported height and weight validity, which found that people with a high BMI are more likely than those of normal weight to underestimate their weight regardless of sex, age, race, or ethnicity (7,11-14). Few studies have been done to understand why this bias exists, particularly for overweight females. A cluster of surveys found that overweight people tend to feel greater discomfort with primary care providers than do normal-weight or underweight people (15-19). According to Lawlor and colleagues, another explanation is that overweight women are less likely to weigh themselves, thereby limiting the accuracy of their self-reports (13). Some studies (20,21) found that physicians and nurses are hesitant to weigh women during the medical visit because of the embarrassment and discomfort caused to the patient. And indeed, the physicians in our study specifically asked that weight not be printed on the patient summary. Nevertheless, it is important for health care providers to strike a balance between preserving patients' comfort now and preserving their health for the future by taking the opportunity to counsel them on weight control.

In addition to maintaining a healthy weight, people with type 2 diabetes can control their disease and those without the disease can reduce their risk of getting it through proper diet and exercise. Surprisingly, the study women who reported exercising at least five times per week were more likely than those who do not exercise enough to have risk scores of 5–9, which we categorized as "currently low risk, likely high risk at age 45 years if risk factors do not change."  These women may either exercise frequently for weight control or overestimate their amount of exercise.

Among the African American women in our study, 11.6% reported a previous diagnosis of diabetes, 30.0% were already at high risk for diabetes, and 46.4% will be at high risk by age 45 if no behavior or weight changes are made. Although obesity is one of the leading causes of mortality in the United States, it is a particularly important cause among African American women (22). In 2000, the prevalence of obesity in the United States was higher among African American women (49.7%) than among white women (30.1%) or Hispanic women (39.7%) (23). During the past decade, obesity increased more among African American women (11.5%) than among white women (4.4%) or Hispanic women (7.2%) (23).

Because obesity is increasing rapidly, there is an unprecedented focus on reducing its prevalence in order to achieve concurrent reductions in diabetes and heart disease. From a public health perspective, primary prevention needs to begin with young, at-risk people. The ADA DRT is a valid measure of current diabetes risk (3). However, it is not designed to educate young people who are at risk for diabetes. To address this issue, we divided the DRT low risk classification (3–9) into two: 3–4 indicates low risk and 5–9 indicates currently low risk, likely high risk when 45 or older for women who will be in the high risk group when they are 45 if they do not change their lifestyle. The purpose was to send a brief educational message to those at future risk because diabetes prevention hinges on early education and behavior modification.

These findings have certain limitations. Some women approached for this study refused to participate. Because of the health center's interpretation of HIPAA regulations, we were not permitted access to denominator information; without knowing the number of potential patients, we could not determine the response rate. In addition, we do not have information on women who refused to participate to determine how they may be similar to or different from women who participated. The substantial risk for diabetes for those in our sample is more than the risk for the general population of urban women (23), probably because the study was conducted in a health center where we recruited women consecutively according to their medical appointments. Therefore, the women in our study are likely to have more health problems than women in the general population.

Since height was not measured instrumentally, we used the self-reported height to calculate BMI with both self-reported weight and scale-measured weight. Research suggests that overweight and obese women are more likely than normal-weight women to overestimate height (7). This overestimate is typically small, but can affect estimated BMI and potentially the DRT score. We subtracted first 1 inch and then 2 inches from each woman's height to assess the potential effect of such a bias. The BMI categories and diabetes risk levels were not substantially affected. Bias would not be apparent until everyone overestimated height by 2 inches or more (data not shown). No other components of the DRT were validated in this study. Once screening began, women were asked to participate if they had a medical appointment. This was an arbitrary time in women's medical care, and therefore some providers may not have discussed issues related to diabetes risk because they had discussed the risks recently or the patient's current medical problem precluded such a discussion. Lastly, we used the ADA's DRT to screen women and educate them about diabetes risk. Although other instruments for screening (24) are available, we weighted heavily the request of the health center's physicians that we use the DRT.

Recently, Romera et al discussed concerns related to BMI measurements (25). In their article, they point out that BMI cannot differentiate between body fat and lean mass, thus leading to misclassifications of obesity. However, in our sample of urban women at a community health center, a high BMI most commonly indicates obesity, not the high level of fitness needed to obtain substantial muscle weight.

Changes in the U.S. population can be expected to affect the number of people with diabetes. By 2050, 14% of the population will be African American, and 24% will be Hispanic (26), two groups with a high prevalence of diabetes. Because the number of women at highest risk for diabetes and associated complications is increasing and because obesity and physical inactivity are on the rise, the number of people with diabetes is expected to continue to increase into the 21st century (27-30). 

Physicians should screen and counsel patients about obesity (2) and diabetes risk. Developing methods for effectively addressing obesity in primary care remains a challenge. Using a computerized screening program has several advantages over practitioner inquiry. The approach can save time for practitioners who provide preventive services. Many more questions can be asked by computer than a health care provider has time to ask, and the information gathered can be summarized immediately for both patients and clinicians. Physicians and other health care providers can use this information to focus on personalized counseling instead of risk assessment. For example, patients in our study got brochures about the health behaviors covered in the screening questionnaire as well as a printout of their screening questionnaire answers to give to their health care provider during their visits. Combined, the screening and educational materials may prepare patients for a productive conversation with their provider regarding healthy lifestyle choices.