Behavioral Science Research in Diabetes: Lifestyle changes related to obesity, eating behavior, and physical activity

As noted above, differences in lifestyle appear to be related to the differential rates of diabetes and obesity across cultures and within our own culture over time (13). These differences in behavior may, in turn, reflect differences in the macroenvironment. Evidence indicating the importance of the environment is seen, for example, in studies comparing Pima Indians, who live in rural Mexico and follow a traditional Pima lifestyle, with Pima Indians living in Arizona,who consume a Westernized diet and are more sedentary(14). Despite the apparent similarity in genetic background of these two Pima communities, the Mexican Pimas have markedly lower rates of obesity and diabetes than the Arizona Pimas. Many other examples of the negative effects of Westernization on eating and exercise, and subsequent risk of obesity and diabetes, are available(15).

Likewise, environmental changes in the U.S. may be leading to the increasing prevalence of obesity(16). It has been suggested that Americans live in an environment rendered unhealthful by their easy access to energy-dense foods and an increasing number of devices (e.g.,television remote controls) that reduce their energy expenditure. Modifying this environment through population-wide changes in eating and physical activity may help prevent obesity.

Currently, most interventions for obesity are conducted at the level of individual patients. Overweight individuals are encouraged to join weight loss programs. Given the epidemic level of obesity in our country, such approaches to the problem may not be the most cost-effective(17). A more global public health approach may be needed.

Environments affect the entire population exposed to them. By affecting policies of companies, government agencies, and other organizations whose decisions influence many people, it may be possible to change the unhealthful environment and thereby change obesity at a population level. In addition, the fact that the current environment is not conducive to healthful eating and activity may explain the poor maintenance seen in most physical activity and nutrition programs. Such educational programs and individual-level treatments will have limited effectiveness when the environment makes it hard to follow the recommendations—i.e., it is hard to follow a healthful diet if grocery stores do not make healthful foods abundantly and consistently available at reasonable prices. Differences in access to healthful foods and opportunities for physical activity may be one of the factors related to the prevalence of obesity in individuals of lower socioeconomic status(16,18). Thus, an important new direction for behavioral research is to study ways to change the macroenvironment and thereby change eating behavior and physical activity.

Correlational data have suggested that environmental factors influence physical activity and eating. For example, Cheadle et al.(18) found strong correlations between fat intake and the percent of local grocery store shelf space devoted to low-fat versus regular milk and meat. Similarly, both the amount of exercise equipment in the home(19) and the density of physical activity facilities in the neighboring community(20) have been associated with adult physical activity levels. The physical activity level of children has been shown to be related to characteristics of their neighborhood environments(21,22).

There have also been small-scale intervention studies suggesting that changing the price or availability of foods in cafeterias or vending machines may influence whether people purchase these items(23,24). Environmental manipulations, including signs to promote the use of stairs rather than elevators (25),have also been effective in changing physical activity patterns. A recent study by Andersen et al. (26)suggests that there may be important differences between ethnic groups in the impact of such environmental manipulations. Signs advocating the use of stairs for cardiovascular health or for weight control increased Caucasians' but not African-Americans' use of stairs. Of particular note are the studies by Ellison et al.(27,28)showing that changes in the food supply in boarding schools are well tolerated and can lead to improvements in physiological risk factors.

Whereas these studies investigating specific manipulations of the environment have tended to show positive effects, there are several large community interventions and school-based approaches that relied mainly on educational programming, which had much more limited effects on eating behavior, eating activity, or obesity(29,30,31). Thus, actually changing the environment may be far more effective than trying to educate the public to deal with an unhealthful environment.

To date, there has been little research on the environmental factors related to eating and physical activity. Such research will require conceptual work to determine what the key variables are and how best to change them; it will also require new methods for measuring these environmental variables and consideration of cultural factors that may influence the results. Environmental variables may be inherently difficult to study because they are ubiquitous; the most important variables may be wide-spread, such as television advertisements, car use, presence of fast-food outlets, and availability of palatable energy-dense foods. People may also resist environmental changes in these domains. Small-scale projects are needed to learn how to intervene on environmental variables.

• Theoretical and epidemiological research. Because this is a new area of research, conceptual work is needed to identify the most important environmental and policy influences on eating and physical activity behaviors. Identifying such influences will require input from scientists, public health experts, marketing researchers, and policy makers. Development of objective psychometrically sound measures of environmental characteristics related to eating and physical activity is also needed, again requiring collaboration across diverse fields. Correlational studies are needed to document associations between environmental and policy variables and behaviors. These studies would generate hypotheses regarding which variables have the most influence on eating and physical activity, and examine the interactions between environmental factors and cultural and socioeconomic differences in the populations.

• Intervention studies. Small focused interventions are needed to examine the effect of specific environmental manipulations on outcomes (i.e.,eating behavior, physical activity, and biological end points). Strategies for changing the home environment (e.g., access to television), other controlled environments (e.g., boarding schools, university dormitories, and summer camp), and the broader neighborhood community should be developed. Changes in the environment to increase access to healthful food and physical activity and to reduce both the financial and behavioral costs of such behaviors should be evaluated. Similarly, it is important to examine ways to decrease access to undesirable behaviors, such as limiting the availability of unhealthy snack foods in school settings or limiting the number of hours of television viewing. It is important to determine whether environmental interventions have similar effects on different ethnic or socioeconomic groups. Studies focusing on the acute effect of environmental change on behavior could be modest in scope; those intended to affect biological end points would need to be more extensive in scope and duration.

As noted earlier, both obesity and physical inactivity are risk factors for diabetes, and the reduction or elimination of such factors appears to be related to prevention and management of this disease. In addition, CHD is the major cause of mortality for individuals with diabetes(11). Both physical activity and weight loss have been shown to reduce blood pressure, improve serum lipid levels, and positively affect other CHD risk factors(11). Thus, development of more effective strategies for adoption and maintenance of healthful eating,physical activity, and body weight would be significant for the prevention and treatment of both diabetes and CHD.

Behavioral approaches to obesity were first introduced during the 1970s and have become increasingly popular in the management of diabetes. Programs that combine diet, exercise, and behavior modification have been shown to be most effective over the short term(11). Currently, a patient entering such a behavioral program will lose 20 lb on average (∼10% of their weight) over the course of 20-26 weeks(32). Alternative approaches to the delivery of the intervention (e.g., through use of computers or correspondence) have been explored(33,34). Although these approaches have typically produced smaller weight loss than face-to-face programs, they may increase the proportion of the population that is willing to participate in weight control interventions.

Longer-term maintenance of weight loss after participation in weight control interventions is less successful. At the 1-year follow-up, patients have typically regained ∼30% of their initial weight loss(32); the few studies with 3-to 5-year follow-up suggest that most patients are back to baseline by this time (35). The most consistent predictors of long-term maintenance of weight loss are increased physical activity and adherence to self-monitoring. Continued treatment contact also appears to improve long-term maintenance of weight loss(36).

There have been a number of studies applying behavioral weight control approaches to type 2 diabetic patients(37). These studies have suggested that improvement in glycemic control and reduction in CHD risk factors are related to the magnitude of weight loss, but even modest weight reduction of ∼10% of body weight appears to improve a patient's glycemic control, blood pressure, lipids, and quality of life(38,39,40). There is some evidence that diabetic patients are less successful in maintaining long-term weight loss than people without diabetes(41), a result perhaps due to metabolic differences between these two groups. Thus, continued efforts are needed to develop behavioral weight loss interventions that will increase the percentage of diabetic patients who are able to lose and maintain weight losses of at least 10% of initial body weight.

Behavioral research on physical activity started more recently and has taken a more community-oriented, less clinic-based approach(42,43). The focus has been on developing strategies to increase the proportion of individuals who adopt physical activity and the proportion that will maintain activity for the long term. Frequent contact by phone and print has been found to help promote activity, along with development of personalized messages matched to the participants' readiness to change their behavior(44,45). Several studies have also documented the impact that physicians can have by recommending and “prescribing” exercise to their patients(46,47,48,49). Emphasis has gradually shifted to home-based, rather than clinic-based,physical activity interventions(50,51),and the accumulation of 150 min/week of moderate-intensity physical activity through multiple short bouts of exercise(52) and/or incorporation of lifestyle activity within one's daily routine(53).

Development of effective strategies for long-term maintenance of any behavior change has proven to be difficult. To develop more effective long-term interventions, it is necessary to understand better what motivates continued adherence to lifestyle change. Since research on longterm maintenance requires large sample sizes and long periods of time, developing reliable intermediate end points would allow for evaluation of intervention strategies in a more condensed time frame. Better ways to assess energy intake and expenditure would allow investigators to focus interventions more directly on the behaviors related to body weight.

• Research directly focused on maintenance. Currently, the major problem in the field is the difficulty of ensuring long-term maintenance of behavior change (54). Therefore, concerted efforts are needed to better understand why maintenance is so difficult and how long-term outcomes can be improved. Aspects of maintenance meriting further attention include the following: the nature and frequency of intervention contact, novel methods of promoting self-monitoring and social support, methods of sequencing or matching treatment to relevant individual characteristics, the level of physical activity that should be recommended, and theoretical constructs that may promote long-term adherence. It is recommended that there be a focus on longterm maintenance within the context of existing large-scale trials, such as the Diabetes Prevention Program and the Study of the Health Outcomes of Weight Loss.

• Increased understanding of motivation. Research is needed to understand better what motivates people to initiate changes in diet and physical activity and what sustains engagement in these behaviors for the long term. Recognition of cultural differences in the attitudes about ideal body weight, obesity, and physical activity will be important in this research.

• New approaches to providing treatments. Research is needed that moves beyond traditional clinic-based models and explores other avenues for intervention with the goal of increasing both the number of people who attempt to lose weight and the long-term effects on body weight. New approaches include church-based interventions and interventions in the primary care setting. Since many individuals who change diet or physical activity do so on their own (i.e., without participation in formal programs)(55), it is important to examine ways to facilitate such efforts, such as using computer-based Internet interventions.

To develop approaches to treatment and prevention, it is important to have some understanding of the etiology of obesity and the factors involved in the development of eating and physical activity habits. This understanding is particularly relevant to the growing problem of childhood obesity.

The prevalence of childhood over-weight has increased dramatically over the past 2 decades, nearly doubling by some estimates(56,57). Childhood overweight is particularly common in minority groups, such as African-Americans and Hispanics. With this increased prevalence of childhood obesity has come a startling increase in the prevalence of type 2 diabetes in children and adolescents(2).

Childhood obesity is also significant because of its psychosocial costs(58,59). In a recent study (59),overweight adolescent females, followed for 7 years, were found to complete less schooling, to be less likely to get married, and have lower income than their normal-weight peers.

Obesity in childhood often continues into adulthood. However, the more common path to obesity in adulthood is to be normal weight as a child and gradually gain weight during adulthood(60). Highrisk periods for weight gain include the time period of 25-34 years of age and the time periods surrounding menopause (61). Pregnancy can also be a high-risk time period for a subset of women(62).

Although weight gain and obesity are clearly due to problems in energy balance, it is still unclear whether this is due to high dietary intake, a low level of physical activity, or both. Secular changes in dietary patterns(e.g., increased reliance on fast foods and increased portion sizes) and physical activity (e.g., decreased playtime and increased use of television as a child-care strategy) may contribute to these changes. A better understanding of the process by which children establish their eating and physical activity preferences and the changes that occur with age will facilitate development of more effective approaches for prevention and treatment of obesity.

Research on the etiology of eating and physical activity has pointed out the important influence of parents in the development of children's food preferences, dietary intake, and activity patterns(63). Parent-child relationships in nutrient intake appear to be related more to shared environment than to genetics and appear stronger for mother-child pairs than father-child pairs(63,64).

Young children appear to regulate their energy intake quite accurately. That is, if they are fed a high-calorie preload, they will eat less during a subsequent meal than if they had been fed a low-calorie preload. However, over time, children become less accurate at such regulation(65). These changes may be due to environmental and familial influences. For example, older children are more responsive to the influence of portion size(65). Moreover, mothers who report more control over their children's eating behavior have children who regulate their energy intake less successfully. Recent work suggests that parents shape their children's eating behavior through their feeding practices(66), but also through the foods they offer to their children and through direct modeling.

The strongest predictor of dietary intake is food preference; people tend to eat what they enjoy. Such preferences appear to be in part innate and in part responsive to early feeding experiences(67). Although preference for sodium has been shown to change with repeated exposure to low-sodium foods, it is unclear whether preference for dietary fat can be changed by consuming a low-fat diet (68).

There have been more than 100 studies of physical activity patterns in children and more than 300 studies in adults(21,22). Most are correlational and have examined demographic, psychological/emotional,behavioral, social/cultural, and environmental correlates of physical activity. Significant associations have been found in all domains. Children appear to be the most active segment of the U.S. population, with physical activity levels declining from the age of 6 years on(69). Decreases in physical activity during adolescence are dramatic. Among adults, sedentary behavior is more prevalent for women, the less educated, the poor, and ethnic minorities(70).

Research has shown that obesity aggregates within families. Interventions for childhood obesity that target both the child and the overweight parent appear most successful (71). In a series of studies, Epstein et al.(72) have shown that family-based interventions for overweight children aged 8-12 years produce improvements in obesity that are maintained through 10 years of follow-up.

To date, there have been few studies designed to prevent obesity. Jeffery and French (73) found little benefit over 3 years of a low-intensity intervention for weight gain prevention. In contrast, a more intensive intervention to prevent weight gain and lipid changes during the menopausal transition was quite effective through 5 years of intervention(45).

Research on etiology of obesity is hampered by the inability to accurately assess intake (i.e., total calories, macronutrient intake, and patterns of intake) and physical activity (i.e., overall amount of activity, its intensity, and the patterns of activity). Self-report measures of these parameters are subject to many biases(74). In addition, there have been few theoretically based studies of the etiology of obesity. Environmental, cultural, and policy influences have not been systematically investigated, and most studies have not been designed to be relevant to intervention. Few studies have focused specifically on prevention of weight gain.

• Longitudinal research. Longitudinal research is needed to identify risk and protective factors for childhood overweight and for weight gain during adulthood. These designs should include samples that allow comparisons across racial and ethnic groups. Research designs should include measures of physical activity, energy expenditure, and food intake and should assess aspects of the environment related to eating and exercise behavior (e.g., the portion of meals eaten away from home and the environmental opportunities for activity). For studies of children, particular attention to the family environment is needed; this will allow for examination of how genetic and environmental factors interact to produce overweight or weight gain. Once longitudinal data (with measures at multiple time points) are available, new statistical techniques, such as growth-curve multilevel modeling, can be used to identify predictors that may constitute risk and protective factors and that predict differing growth trajectories. An important longitudinal issue is to determine how and when childhood behaviors carry over to adulthood.

• Research on development of preferences for foods and activities.Research is needed that describes the factors involved in the acquisition of food and physical activity preferences. What are the biological and environmental influences that shape eating and physical activity preferences?How stable are these preferences over time, and what strategies can be used to modify these preferences?

• Intervention studies. Research is needed to determine how to prevent the development of obesity in children and reduce obesity in those children already affected. It is unclear whether there are certain ages that are most conductive to effective intervention (e.g., ages 8-12 years) and what strategies are most effective in modifying diet and activity patterns. Similarly, interventions are needed to prevent the decrease in physical activity that occurs during adolescence and to examine the impact of such interventions on weight change and other health-related parameters in young adults (75). Interventions to prevent weight gain and the development of obesity during adulthood are also needed.

The treatment regimen for individuals with diabetes is extremely complex. The regimen includes both lifestyle components (i.e., diet and physical activity) and typically pharmacological components (i.e., oral medication and insulin). Patients must monitor their blood glucose and use this information to adjust and coordinate eating, physical activity, and medication doses. In addition, many patients with diabetes have other comorbidities, including hypertension and hyperlipidemia, and thus must integrate their diabetes self-care with self-care of these other diseases. The treatment regimen also involves regular visits to physicians or other health care providers and screening and treatment for diabetes-related conditions (e.g., eye screenings and foot care). Tobacco use or excessive alcohol use further complicates the task of behavior change.

It remains unclear how best to integrate these multiple behavior changes. This issue is important within lifestyle behaviors (e.g., is it best to start with diet and add exercise, or start with exercise and add diet, or do both concurrently?) and is made even more complex when medication is added to the treatment regimen. Which behaviors are synergistic and which negatively affect each other remains unclear. What is the optimal role of the health care clinician in sorting out priorities for behavior change?

Previous research has suggested that the combination of diet and physical activity is more effective for long-term weight loss maintenance and improved glycemic control than either intervention alone(11). Physical activity may also act as a catalyst for other behavior changes; individuals who are more active often consume healthier diets and smoke less(76).

The positive impact of the combination of diet plus exercise contrasts with other examples in which the combination of two behavioral goals produces less change in each of the behaviors. In treatment of hypertension, for example,patients who are instructed to follow a low-sodium diet and lose weight are less adherent to either of these changes than when they are introduced separately (77).

Smoking cessation programs have examined the effectiveness of including other lifestyle behaviors within the program to address concerns about weight gain after smoking cessation. Studies combining smoking cessation with a weight loss intervention have been relatively unsuccessful(78), whereas a recent study combining smoking cessation with physical activity appears promising(79). There is also evidence that targeting alcohol consumption and smoking enhances abstinence rates for both behaviors (80).

To date, there has been little research on how best to combine lifestyle and pharmacological treatments to maximize compliance to both regimens and increase overall effectiveness(81). With the increased interest in drug treatment of obesity, such research clearly will be of significance.

Clinically, the major barrier to the combination of lifestyle modification and pharmacological treatment relates to the different specialty groups that focus on each approach. Although physicians feel competent to prescribe medication, they have little or no experience with lifestyle intervention. The lack of time the physician has available for each patient in primary care settings poses another barrier to use of lifestyle intervention. Conversely,the behaviorists who are most expert at lifestyle intervention are not qualified to prescribe drugs.

From a research perspective, a major barrier is the complexity of the study design required for investigation of multiple behavior changes. Such studies require large sample sizes and relatively long duration.

• Research on multiple lifestyle changes. Research is needed to determine which lifestyle interventions act synergistically with each other and which combinations are less effective. Although it is often assumed that modification of multiple lifestyle factors will enhance prevention and treatment efforts for diabetes, research in other areas (e.g., smoking plus weight loss) suggests that targeting multiple behaviors may, in some cases,have a negative impact on treatment. Therefore, it is important to compare the effects of targeting a single lifestyle factor, sequencing of several lifestyle factors, or targeting multiple factors simultaneously. Whether physical activity has a special role as a catalyst for other lifestyle behavior changes deserves specific attention.

• Research combining lifestyle and pharmacological treatment. Research is needed to develop strategies to maximize the effectiveness of the combination of lifestyle and pharmacological treatments. Such research should be multidisciplinary and examine lifestyle plus medication approaches for treatment of obesity, diabetes, and other CHD factors. Questions related to the timing of the two approaches, the impact of patient preference or choice(versus clinicians' choice) regarding these two modalities, and ways to train physicians and other clinicians in the use of lifestyle approaches within clinical settings deserve attention.

Given the strong association between lifestyle behaviors and the prevention and treatment of type 2 diabetes, it is important that greater research attention be directed at issues related to the development of healthful eating and physical activity habits and strategies for modifying unhealthy behaviors. The focus should be on ways to change eating and physical activity behavior both from an individual perspective and a broader environmental perspective.