This study provides new evidence of racial and ethnic differences in health intentions and behaviors for adults with type 2 diabetes. Although there was no difference in the proportions reporting that a healthcare professional recommended increasing exercise across race-gender groups, more African-American men reported an intention to follow the exercise recommendation and a greater proportion reported that they currently were exercising regularly than other race-gender groups. More Hispanic men reported exercising regularly with high physical activity levels than other race-gender groups. Racial/ethnic differences were also evident for diet and weight management. More Hispanic men and women reported that a healthcare professional recommended changes in their diets despite similar BMI levels across groups. While more Caucasian women and Hispanic women reported trying to lose weight, African-American men and Hispanic men had the lowest proportion trying to lose weight. Household income did not impact these racial differences.
This study also highlights the discordance between individuals' impressions of what they are doing regarding healthy behaviors and their actual performance, a disconnect that appears to be more prominent among African-Americans. For example, more African-American men indicated that they were currently exercising regularly than other race-gender groups, but they had the lowest proportion of men with high physical activity levels and the lowest mean IPAQ scores among the men. In contrast, Hispanic men constituted a larger proportion of those who reported that increasing exercise will keep them healthy. They also reported exercising regularly and had the largest proportion with high physical activity.
A number of differences in intentions and approaches to weight loss were seen among the various groups. More Caucasian women and Hispanic women reported trying to lose weight, yet these race-gender groups had the highest proportion of respondents with obesity, indicating that intent did not translate to weight loss. In contrast, African-American men and Hispanic men reported the lowest proportion trying to lose weight. Although there were similar intentions regarding weight loss between the latter two groups, Hispanic men had the highest proportion of physical activity, whereas African-American men had lower proportions of high physical activity levels similar to those of Caucasian and African-American women. It is possible that the choice to pursue a high level of physical activity among Hispanic men is not connected to the desire for weight loss, but driven by other considerations.
There appears to be a willingness among some minority respondents to follow the recommendations of their healthcare providers regarding healthier lifestyle choices. Indeed, African-American women indicated greater intensions to change their existing diets and indicated that they currently exercise more than other groups. This willingness and intention to engage in healthy behaviors needs to be put into action given the health benefits of physical activity, exercise, and weight management for the management of type 2 diabetes . There has been increased public health attention on the need to reduce obesity and increase physical activity because of the epidemic in diabetes of recent years . Until intentions are translated to positive behaviors, there will be little improvement in the twin epidemics of diabetes and obesity. It is hypothesized that since advice is given by the healthcare professionals seen by the SHIELD respondents, there is a need to implement intentions to achieve behaviors. The lack of greater achievement of lifestyle behaviors suggests a need for more access to and use of appropriately designed tools for weight loss (e.g. prescribed diet plans) and exercise (e.g. exercise equipment, aerobic exercise videos), and fitness facilities or exercise programs at community centers or school gymnasiums for effective lifestyle modification among minority individuals. Several studies have indicated that tools and/or alternative methods to education beyond print materials have prompted behavior change. Tailored computer programs for setting goals for nutrition and physical activity which are reviewed at each visit by physicians resulted in increased physical activity and weight loss compared with printed health education materials among overweight patients with T2DM . Step pedometers with and without print materials were effective for increasing physical activity compared with a standard public health recommendation for physical activity or print materials only . Telemedicine, video education and daily text messages via mobile phones have prompted behavior adherence and weight loss compared with print materials and standard education [24–26]. Among stroke patients, an education session did not result in positive diet or physical activity changes . Other studies indicate that education sessions and materials need to be tailored, culturally appropriate with personally relevant information for work and home lives to improve motivation and health behavior change [28–30].
Data from the 2002 Medical Expenditure Panel Survey (MEPS) indicated that 73% of adults with diabetes were told by a health professional to exercise more  which is higher than the approximate 50% of respondents in the present study. Additionally, the MEPS study indicated that the likelihood of receiving medical advice on exercise was less likely in Hispanics but the present study found a similar percentage receiving advice across the race-gender groups. The differences may be due to more current data (2007) in SHIELD and differences in population composition (e.g. age, gender, race). The present study found that <30% of respondents reported exercising regularly and <20% reported high physical activity compared with 39% of adults with diabetes reporting being physically active in the MEPS 2003 survey . The 2000 Behavioral Risk Factor Surveillance System (BRFSS) national survey reported that about 30% of the US adults exercised regularly and 14% did intense physical activity . The MEPS study  indicated that Hispanics with diabetes were potentially more likely than whites (odds ratio = 1.43, 95% CI: 0.98-2.07) to be physically active while the SHIELD study found a greater percentage of Hispanics, especially men, had high physical activity compared with other race groups. For weight control and diet, the BRFSS reported that 38% of US adults were trying to lose weight, which increased to 66% of obese adults, compared with 60-70% of SHIELD T2DM (mostly obese) respondents reporting trying to lose weight. Overall, the SHIELD findings are similar to other national surveys but differences may be due to change over time (2000-2003 to 2007) and different exercise/physical activity questions being utilized in the different surveys.
The present study has limitations that should be considered. The determination of type 2 diabetes and obesity were made based upon self-report rather than clinical or laboratory measures; therefore, there may be misclassification bias. However, all respondents were asked the same questions to assess diabetes and weight status. Medical advice, physical activity, diet, and weight management were also self-reported. However, the IPAQ for physical activity assessment is a validated and well-accepted physical activity instrument. Household panels such as the SHIELD study tend to under-represent the very wealthy and very poor segments of the US population and do not include military or institutionalized individuals. Thus, the study findings should not be generalized to these population segments. The survey was provided in English only, thus potentially excluding individuals who spoke other languages, especially among Hispanic households. Given that the supplemental minority sample that was added in 2007 was not part of the random, stratified sampling method at baseline, these minority respondents may not be representative of the African-American or Hispanic populations in the United States. This study examined health intentions and behaviors by race-gender groups which may not represent individual intent-behavior correlation.