In response to mounting evidence that behaviors, such as cigarette-smoking and consumption of high-fat diets, are risk factors for chronic diseases, several studies target interventions for medically at-risk individuals. Some landmark clinical trials, such as the Multiple Risk Factor Intervention Trial (MRFIT Research Group, 1982), have contributed to our understanding of risk factors in disease. Trials also focus on psychosocial interventions after disease onset to improve treatment adherence and medical outcomes. Other interventions arise from the concept of population-attributable risk, which measures the amount of disease in the population that can be attributed to a given exposure (Marmot, 1994). A large number of people exposed to a small risk might generate more cases than will a small number exposed to a high risk (Rose, 1992), so that when risk is widely distributed in the population, small changes in behavior across an entire population can yield larger improvements in population-attributable risk than would larger changes among a smaller number of highrisk individuals (Marmot, 1994; McKinlay, 1995; Rose, 1992). Both approaches are described below.
Clinical trials such as the Multiple Risk Factor Intervention Trial (MRFIT Research Group, 1982), the Lipid Research Clinics Coronary Primary Prevention Trials (Lipid Research Clinics Program, 1984a,b), and the Lifestyle Heart Trial (Ornish et al., 1990) have provided important contributions to the development of successful interventions and to the current understanding of risk factors for disease. Education and counseling can promote primary prevention measures reducing smoking and choosing a healthy diet. Interventions aimed at secondary prevention behaviors can influence early detection of illness. For instance, willingness to self-examine and participate in screening procedures is important for detection and treatment of cancer. Psychosocial interventions can improve people's coping skills and provide emotional support, thereby improving quality of life and medical outcomes among the chronically ill. The role of behavioral interventions for improving adherence to treatment is discussed below. Interventions addressing behavioral and psychosocial risk factors are also briefly reviewed.
Adherence, the match between a patient's behavior and health care advice (Haynes et al., 1979), mediates the effectiveness of treatment recommendations, the scientific evaluation of treatment protocols, and even public health. For example, when treating bacterial infections, some patients stop taking antibiotics when symptoms stop, but before all the targeted bacteria are eradicated, resulting in relapse for the patient and the development of resistant bacteriological strains. Failure to follow medical recommendations for treatment is a common problem that is not without controversy. The term “adherence” has been increasingly used to replace the previous label of “compliance” to convey the patient's active participation in following a treatment regimen, rather than the patient's submission to a provider's directive (Roter et al., 1998). Between 30% and 70% of patients do not adhere effectively to treatment recommendations. Nonadherence to difficult behavioral recommendations, such as smoking cessation or following a restrictive diet, occurs in more than 80% of patients (National Heart, Lung, and Blood Institute [NHLBI], 1998). The reasons are varied: Providers sometimes fail to describe the treatment regimen clearly, resulting in confusion on the part of the patient. Patients may also not fully appreciate the consequences of nonadherence. Some regimens interfere with daily activities, particularly those requiring multiple doses each day, or those with special instructions regarding meals (e.g., take on empty stomach). Side effects, such as hair loss, can be embarrassing; others, such as dry mouth or gastrointestinal problems, can be uncomfortable. Insurance limits on reimbursement for treatments also can affect adherence. Nonadherence is more than failure to take medications as prescribed or to follow other recommendations for health behavior changes. One survey of oncologists (Hoagland et al., 1983) showed that failure to return for recommended outpatient treatments was the most frequent source of nonadherence. Adherence often depends on the nature of treatment. Therapies that are simple or that produce prompt relief of pain or symptoms typically result in high levels of adherence (Dunbar-Jacob et al., 2000). Adherence is usually poor if therapies last a long time, if they are preventive rather than curative, or if they are complicated. Patients who experience psychological problems or substance abuse are less likely to adhere (NHLBI, 1998).
Renewed attention has been given to non-adherence in recent years, led by concerns about the development of multi-drug-resistant tuberculosis (Cohen, 1997) and HIV (Chesney et al., 1999). Multidisciplinary research efforts are developing new self-report assessments of adherence that show significant relationships with biological outcomes. Electronic medication monitors, which are being used increasingly in research, provide more accurate estimates of adherence to medication regimens and suggest that patients overestimate their own adherence (Cramer et al., 1989) and that provider estimates of adherence are not better than chance (Haubrich et al., 1999).
Effective interventions have been developed to improve cooperation in the acute-care setting. For example, adjunctive nonpharmacologic analgesia involving self-hypnosis has been shown in two randomized trials to reduce pain, anxiety, patient-controlled medication use and episodes of hemodynamic instability and to reduce procedure time by 22% (Lang et al., 1996, 2000).
There have been surprisingly few studies of interventions that might enhance adherence (Shumaker et al., 1998). A recent systematic review of randomized trials of interventions to help patients adhere to medications revealed that successful interventions were those that were multifaceted, including such features as more convenient care, information, counseling, reminders, self-monitoring, reinforcements, and other forms of supervision and attention (Haynes et al., 1996). Relatively few studies have evaluated the benefits of interventions that require permanent lifestyle changes. The difficulties in sustaining the cessation of smoking, weight loss, or initiation of exercise are well recognized (Marlatt and George, 1998). The relapse rates, however, are not uniform for these behaviors. The rate of relapse from treatment of serious obesity is more than 90%, leading to revision of goals in its treatment to more modest but sustainable weight loss (Wadden et al., 1999), but half of those who stop smoking will remain completely abstinent for 2 years (Spiegel et al., 1993).
Addressing Psychosocial Risk Factors
As described in Chapter 2, depression is a risk factor for mortality from multiple causes. Furthermore, “distressed high utilizers” of medical care are substantially more likely to suffer from psychiatric disorders, including major depression, generalized anxiety disorder, and substance abuse (Von Korff et al., 1992). Poor adjustment to illness can increase the cost of medical care by as much as 75% (Browne et al., 1990). These problems make the development of programmatic interventions to provide psychosocial support both humane and expedient. Thus, providing appropriate psychotherapeutic and psychopharmacologic treatment for them not only can improve coping and reduce patient discomfort but also can make the delivery of medical care more efficient. The contributions of clinical behavioral and psychosocial interventions to diabetes, cancer, and heart disease are explored briefly. A recent chapter (Baum, 2000) from an Institute of Medicine (IOM) report provides further discussion of the influence of stress in cancer and cardiovascular disease.
Diabetes Mellitus. To reduce the incidence and severity of complications of diabetes, including vascular, coronary, renal, and neurologic disease, blood sugar must be carefully regulated. Adherence to medication regimens, glucose testing, exercise, and diet influences medical outcomes. Research indicates that coping skills and family stresses influence the management of diabetes (see Glasgow et al., 1999, for a review). Furthermore, depression is a serious co-occurring problem in diabetes (Glasgow et al., 1999; Jacobson, 1996; Lustman et al., 1992) that can affect glycemic control (Lustman et al., 2000). Several reviews and meta-analyses have demonstrated the effectiveness of educational approaches aimed at increasing knowledge, control, and self-efficacy among diabetics (Brown 1990, 1992, 1999; Hampson et al., 2000; Padgett et al., 1988). On the other hand, education did not consistently improve metabolic control (Grey, 2000). Psychosocial interventions (for example, enhancing coping skills and peer support) seem to provide greater success in improving both metabolic outcomes and quality of life (Grey, 2000; Grey et al., 1999). Educational interventions could be more effective when used in combination with behavioral psychosocial interventions (e.g., Brown, 1999, Clement, 1995). However, concerns exist that the beneficial changes might not be sustained long beyond the intervention (Brown, 1992).
Cancer. There is evidence that psychosocial interventions can improve quality of life, psychological adjustment, health status, and survival of cancer patients (see reviews by Andersen, 1992; Blake-Mortimer et al., 1999; Compas et al., 1998; Fawzy et al., 1995; Helgeson and Cohen, 1996; Meyer and Mark, 1995, Montazeri et al., 1998). A meta-analysis of 116 studies on the effects of psychoeducational care provided to adult cancer patients concludes that interventions affect anxiety, depression, and mood (Devine and Westlake, 1995). Another analysis of 45 psychosocial interventions showed statistically significant emotional benefits in adults (Meyer and Mark, 1995). Various interventions have been tested, including teaching specific methods of coping with the stress of cancer (Edgar et al., 1992; Fawzy et al., 1990; Telch and Telch, 1986), providing education and information (Manne et al., 1994), and providing social support and facilitating expression of emotions (Spiegel and Classen, 1995; Spiegel et al., 1981, 1989). Their relative effectiveness has been difficult to assess (Devine and Westlake, 1995; Fawzy, 1999; Meyer and Mark, 1995).
Some evidence supports the effectiveness of psychosocial interventions to improve medical outcomes and prolong survival (for reviews, see Creagan, 1999; Greer, 1999). Spiegel and colleagues (1998) found that psychosocial group treatment in metastatic cancer patients doubled survival time to an average of 18 months, from the point of randomization. A study by Richardson et al. (1990) showed that lymphoma and leukemia patients assigned to 1 of 3 educational and home-visiting supportive interventions had significantly longer survival than did patients allocated to routine care (control). The effect was sustained even when differences in medication adherence were controlled. In a study of 125 patients with metastatic melanoma, quality of life was found to be associated with duration of survival (Butow et al., 1999). A randomized controlled trial of 6 weeks of intensive group therapy aimed at developing active coping among 80 malignant melanoma patients significantly reduced mortality at 6-year follow-up (Fawzy et al., 1993). The mechanisms through which psychosocial interventions exert their effect is unknown, but it has been suggested that depression exacerbates symptoms (Evans et al., 1999) and that psychotherapy augments the immune response (for reviews, see Kiecolt-Glaser and Glaser, 1999; Spiegel et al., 1998). These results should be explored further and confirmed.
Although the potential of psychosocial intervention to slow the progression of cancer is promising, the literature is limited and several reports refute the hypothesis (for example, Cunningham et al., 1998; Gellert et al., 1993; Ilnyckyj et al., 1994; Linn et al., 1982; Morgenstern et al., 1984). One meta-analysis (Meyer and Mark, 1995) showed a small effect of psychosocial interventions on medical measures that was not statistically significant. Carefully designed studies are needed to clarify this issue.
Coronary Disease. Primary prevention can reduce the incidence of coronary disease (Chapter 3), but psychosocial interventions also can affect morbidity and mortality in at-risk patients. As described in Chapter 2, several studies have recently demonstrated that social isolation, depression, and type A personality traits—especially hostility—can mediate medical outcomes for patients with coronary disease (also see Rozanski et al., 1999; Williams and Littman, 1996). Evidence is increasing that psychosocial interventions after the onset of disease are effective supplements to routine cardiac care. One recent meta-analysis of 37 studies (Dusseldorp et al., 1999) found that psychoeducational programs reduced mortality by 34% and decreased recurrence of myocardial infarction by 29%. Another meta-analysis (Linden et al., 1996) of 23 clinical trials on coronary artery disease reported a similar significant reduction in morbidity and mortality with psychosocial interventions, especially during the first 2 years.
The interventions included in the analysis by Linden et al. (1996) were diverse but consistently positive. Powell and Thoresen (1988) found that counseling designed to reduce hostility and impatience typical in type A people reduced mortality among acute myocardial infarction patients who had less serious cardiac disease. Ornish et al. (1990) demonstrated that an intensive program of group support, stress management, moderate exercise, smoking cessation, and strict vegetarian diet resulted in a measurable reversal of coronary artery disease. Blumenthal et al. (1997a) found that stress management in coronary artery disease patients significantly reduced the subsequent risk of a cardiac event.
Many studies support psychosocial interventions, but other evaluations show no significant effects. Black et al. (1998) provided psychiatric evaluation and behavioral therapy to 380 cardiac patients and reported decreases in depression but not in rehospitalization rates. A clinical trial by Jones and West (1996) revealed no benefit from relaxation training and stress management. In contrast to the results of an earlier study that indicated that simply monitoring for psychological distress in cardiac patients reduced mortality (Frasure-Smith and Prince 1985), a follow-up study (Frasure-Smith et al., 1997) could not replicate the results and recommended against implementing such programs into routine care. The discrepancies among studies probably result from methodologic limitations, including small study sizes, varied interventions (some of which may not be behaviorally effective), indefinite clinical endpoints, and lack of intention-to-treat analyses. To address these limitations, a national multicentered clinical trial has been initiated (Enhancing Recovery in Coronary Heart Disease [ENRICHD], 1999), to determine the effects of psychosocial interventions on 3000 patients. Interventions will target depression and social isolation in patients with a recently diagnosed myocardial infarction. Endpoints will include mortality, nonfatal infarctions, cardiovascular hospitalizations, and changes in risk factor profiles (Blumenthal, 1997b; ENRICHD, 2000).
Addressing Behavioral Risk Factors
The primary care physician is in an optimal position to provide advice on healthy behaviors. Many studies have indicated that counseling by a primary care physician can be effective in changing the behaviors of patients but the approaches are varied. Several fundamental characteristics contribute to the effectiveness of these interventions. Recognition of differing patient needs is one fundamental characteristic of practices dedicated to enhancing beneficial behavior change. Some patients need only visual cues as a reminder to ask for help with smoking cessation, to obtain timely mammograms, to exercise more regularly, or to follow up for management of depression (Pronk and O'Connor, 1997; Rogers, 1995). Others respond more favorably to printed materials, coaching via telephone-based counseling, or classes. Some patients cannot change health-related behavior without one-on-one structured education and counseling supplemented by frequent reinforcement from their physicians. Multiple modalities of support are used in the practices that are most heavily committed to encouraging beneficial behavior change and that target individual patients (Oxman et al., 1995). Similarly, multiple methods are necessary to communicate with physicians and other clinical staff to encourage behavior change on their part that reinforces patient behavior change (Green et al., 1988; Greer, 1988). Chart reminders, computerized medical records with automated protocols, and physician and other staff education have all shown promise (Buntinx et al., 1993; Davis et al., 1995).
A second beneficial approach to behavioral intervention is the organizational leadership to decide to focus on a problem and devote energy and resources to it (Greer, 1988; Hammer and Champy, 1993; Oxman et al., 1995; Patti and Resneck, 1972; Rossi, 1992). A clinical practice that has an enhanced capacity to change patients' health-related behavior has leadership able to relate to the physician staff members and to engender enough emotional, internal, political, and economic support to drive behavior-change efforts toward success (Davis and Taylor-Vaisey, 1997). That presents a major challenge because most clinical practices are organized to deliver acute care rather than to change patients' behavior to prevent illness (Walsh and McPhee, 1992). Engaging busy practices to reach into new health promotion endeavors for which there is little economic reward is challenging, no matter how dedicated the leadership and clinical staff (Fishman et al., 1997). Rising to such a challenge tests the leadership and organizational adaptability of any practice that also must comply with innumerable legal, business, and clinical regulations and requirements. Many variables peculiar to a given practice—such as physician attitudes, local competitive pressures, staff morale, and socioeconomic needs of the patient population—can enhance or inhibit change in the practice toward a greater focus on prevention or other innovation (Crabtree et al., 1998). For example, changing practice patterns to document brief but consistent efforts to encourage smoking cessation initially proved beyond the reach of many good practices (Kottke et al., 1988).
Health care systems and practices in the United States are moving toward use of methods to increase the predictable quality and efficiency of medical care (Berwick, 1989; Carlin et al., 1996; Grimshaw and Russell, 1993, 1994; McDonald, 1976; Miller et al., 1998; Mittman et al., 1992). Current quality improvement models propose a more active and continuous method of identifying problems and testing interventions. This is a change from traditional methods of identifying faulty practices and practitioners by investigating clinical cases that have unsatisfactory outcomes (Balas et al., 1996). Rather than a list of poorly performing health providers, the result of a continuous improvement model can be a testable hypothesis that outlines a series of steps for caring for patients with specific problems that can result in measurable improvement in outcomes or processes (Crabtree et al., 1998; McBride et al., 1993; Solberg et al., 1997).
A simplified continuous-improvement model has four steps: (1) design a guideline with active participation of clinicians; (2) implement the guideline; (3) measure the outcomes; (4) study the outcomes, compared with what was expected, and redesign as needed (Mosser and Sakowski, 1996). Working with two large managed-care organizations, Solberg et al. (1998) conducted an RCT to assess the effectiveness of a process to help primary care clinics develop systems for the delivery of preventive services. Previous research showed that even when external technical assistance succeeded in increasing preventive services, the services declined to baseline when the assistance ended (Magnan et al., 1998). To build practices' internal capacity to initiate and manage change, the IMPROVE (Improving Prevention through Organization, Vision, and Empowerment) trial (Solberg et al., 1998) used organizational development approaches, such as continuous quality improvement and process consultation. The intervention facilitated the formation of continuous improvement teams that instituted prevention processes (Solberg et al., 1995). However, the extent to which patients in the intervention practices are actually receiving more preventive services has not been determined.
Clinical practice guidelines are formal statements that provide guidance to health care practitioners regarding specific clinical circumstances. Ideally, guidelines are based on the best available scientific evidence and clinical judgment. They should lead to the best patient outcomes and should steer clinicians away from unnecessary or extravagant interventions. The appeal of practice guidelines has led to remarkable growth in their development. An editorial in Lancet (Fletcher and Fletcher, 1998) describes beleaguered clinicians faced with more than 2000 sets of guidelines. However, guidelines lack standards of quality and have been developed by fragmented groups that might have different goals, motivations, and capabilities. Furthermore, guidelines are often outdated by the time they are released, often ignore patient preferences (Eddy, 1990), and often emphasize peer consensus rather than outcome evidence.
Many focused interventions to encourage health-related behavior change would benefit from population databases that keep track of patients' medical histories, behaviors, and attitudes. One fundamental factor for practice-based interventions is the availability of a database that defines the population served, accepts searches of health parameters or disease targets, and allows tracking of measurable changes in the defined health behavior or health outcome. An ideal database can link names, addresses, telephone numbers, diagnoses, pharmacy use, and other use of health care visits and educational resources (Redding et al., 1993). An example of a practice-based intervention that requires such a database is improving the diet and exercise patterns of poorly controlled diabetes mellitus patients and tracking their metabolic-outcome measurements for improvement (Thomson O'Brien et al., 1999). However, there has been little systematic research on the benefits of such databases in the United States. Practice databases are available primarily in large, well-organized practices and in staff model health-maintenance organizations whose physicians or other providers are paid salaries. They are not often used in smaller group practices because of the cost and personnel required to maintain them. Their use also raises major legal and ethical issues of privacy and confidentiality that have been the topic of several reviews (Gostin, 1997; Sweeney, 1997; Woodward, 1997).
Need for Research on Practice
Much of the information in this section is based on evidence from uncontrolled trials and one-time interventions in large multispecialty group practices and well-organized staff model health maintenance organizations. Some of the information is based on the opinions of experts. Little of what is known about dissemination is based on well-controlled trials wherein a practice-level intervention is compared with reasonably controlled and parallel practice. Only occasional studies (e.g., Cohen et al., 1999; Cooper et al., 1997) have tried to assess interventions such as screening practices at the level of primary care physicians. Little research funding in the past has been applied to systematic evaluation of fundamental (systemic) changes in clinical practices that might support health-enhancing behavior change in defined populations. Future efforts should test various hypotheses that would encourage experimentation and practice-level interventions.
This section examines a sampling of studies that are representative of population-based intervention trials in a community, worksite, or school that are focused on changing individual behavior for primary prevention of disease. Given the importance of shifting the population distribution of disease risk, the effectiveness of interventions must be measured among the entire population for whom the intervention is intended, and not only among program participants. In addition, because of the importance of accounting for the influence of secular trends and for other factors not associated with the intervention that could affect behavior change, the studies discussed here included intervention and control conditions alike. Finally, to narrow the field of potential studies, a focus was given to those interventions conducted in the United States that targeted primary prevention of cancer or coronary heart disease, although the committee recognizes that considerable progress has been made using community interventions to address other public health problems.
Several early population-based community studies, including the Minnesota Heart Health Program (MHHP), the Stanford Five City Project (SFCP), and the Pawtucket Heart Health Program (PHHP), tracked changes in morbidity and mortality. For subsequent intervention studies, however, funding did not permit following participants long enough or in sufficient numbers to determine long-term costs and consequences of the interventions for survival, quality of life, or disease incidence. Instead, subsequent population-based intervention research rests on prior evidence linking behavioral outcomes to health benefits, such as reductions in morbidity and mortality (Chapter 3). Thus, for most population-based trials, behavior change is the primary outcome. The behaviors examined include dietary changes, tobacco use, and physical activity.
Large-scale studies. Two early studies targeting cardiovascular disease prevention set the stage for population-based community intervention trials: the North Karelia Project (Puska et al., 1983) and the 1977 Stanford Three Community Study (Farquhar et al., 1977; Fortmann et al., 1981). Although the North Karelia Study was not done in the United States, it is included here because of its importance as a groundbreaking study of community intervention trials. The North Karelia Project grew out of that community's concern about having the highest heart attack risk world-wide (Blackburn, 1983; Keys, 1970; Verschuren et al., 1995). Results of a community-wide intervention implemented in North Karelia were compared with a reference area in eastern Finland. After 10 years, the net effects among middle-aged males included significant reductions in smoking, mean serum cholesterol concentrations, mean systolic blood pressure, and mean diastolic blood pressure; significant declines in mean systolic and diastolic blood pressure were observed among women (Puska et al., 1983). The study set the stage for community-wide intervention studies in the United States, the first of which was the Stanford Three Community Study (SHDPP). Initiated in 1972, that study demonstrated the feasibility and potential effectiveness of mass-media-based educational campaigns combined with intensive instruction of individuals in group or home classes directed at entire communities (Farquhar et al., 1977; Fortmann et al., 1981; Maccoby and Solomon, 1981). Significant reductions in cholesterol and saturated fat were reported at the conclusion of the intervention and were sustained during a 1-year maintenance period (Fortmann et al., 1981).
In the late 1970s, three large community-wide intervention trials were funded by the National Heart, Lung, and Blood Institute: SFCP (Farquhar et al., 1990), MHHP (Luepker et al., 1994), and PHHP (Carleton et al., 1995). All targeted change in risk factors for coronary heart disease (CHD), including high blood pressure, elevated blood cholesterol, cigarette-smoking, and obesity. None was randomized; rather, communities were matched to optimize comparability of study conditions (Murray, 1995). The multiple-risk-factor intervention trials varied in length from 5 to 7 years, and they tracked changes in morbidity and mortality beyond the intervention period. The interventions were aimed at raising public awareness of CHD risk factors through media education. Other objectives were to change risk-related behaviors through public education in schools, worksites, and other community organizations; educate health professionals; and initiate environmental change programs, such as labeling of foods sold in grocery stores and restaurants. For SFCP, significant effects were observed in blood cholesterol, smoking, and systolic and diastolic blood pressure; and decreases in risk—shown in composite risk factor indices— were significantly larger in the intervention than in the comparison communities (Farquhar et al., 1990). At the 3-year follow-up, the possibility was suggested of sustaining at least some observed outcomes, although the magnitude of the long-term effects was small (Winkleby et al., 1996). Fewer significant results were observed in MHHP and PHHP. MHHP reported significant effects for smoking prevalence among women and for physical activity (Luepker et al., 1994). PHHP resulted in smaller increases in body mass index in the intervention communities than in the controls; no other significant results were reported (Carleton et al., 1995).
In 1989, the National Cancer Institute (NCI), building on methods used in cardiovascular disease studies, launched the Community Intervention Trial (COMMIT) for Smoking Cessation (COMMIT, 1991). The trial used 11 matched pairs of communities across North America, and it was designed to test the effectiveness of a multifaceted, 4-year community intervention to encourage smokers, particularly heavy smokers, to achieve and maintain cessation (COMMIT, 1991, 1995a). A significant effect was observed among light-to-moderate smokers, and it appeared to be greater among a less-educated subgroup of participants (COMMIT, 1995a). There was no effect among heavy smokers (COMMIT, 1995a).
Although not a randomized, controlled intervention trial, the American Stop Smoking Intervention Study (ASSIST) was a large-scale, 7-year demonstration project building on randomized community-wide intervention trials. The intervention was implemented in 17 states through a partnership among NCI, the American Cancer Society, state health departments, and other organizations. The primary goal was to reduce smoking prevalence and cigarette consumption. To assess the results, investigators compared data from ASSIST and non-ASSIST states. Comprehensive tobacco control programs emphasized policy interventions, including indoor air, pollution, youth access, advertising, and tobacco taxes, as well as mass-media interventions and program services such as cessation classes (Manley et al., 1997a,b). Per capita consumption of cigarettes was comparable in ASSIST and non-ASSIST states before the beginning of the 1993 intervention. By 1996, smokers in ASSIST states were smoking 7% fewer cigarettes per capita. The intervention also included guidelines for raising cigarette excise taxes as a means of reducing consumption. Inflation-adjusted cigarette prices were nearly identical in both groups of states before 1993. Although the tobacco industry reduced prices during this period, in 1994 the average price was more than $0.12/pack higher in intervention than in control states (Manley et al., 1997a,b).
Small-scale studies. Several recent community-wide studies have borrowed principles from the early large cardiovascular disease prevention trials, but they have been implemented on a smaller scale and with smaller budgets. It might be difficult for such studies to achieve the necessary intensity and reach to show significant intervention effects. The Bootheel Heart Health Project, for example, was conducted in a six-county area in southeastern Missouri (Brownson et al., 1996). This rural area has the largest African American population in Missouri, and it is characterized by high rates of poverty and low education levels. The intervention was tailored to the community through the participation of local coalitions, each establishing its own priorities for intervention. The researchers conducted population-based cross-sectional surveys before and after the intervention to compare results in communities where there were coalitions against results from communities that did not have coalitions. Physical inactivity decreased and the prevalence of self-reported cholesterol screening increased in communities with active coalitions. Differences observed in self-reported weight gain were in the right direction, although not statistically significant. No differences were found for fruit and vegetable consumption or for smoking prevalence. Similar results were observed in the Heart to Heart Project, which reported decreases in dietary fat consumption and increases in cholesterol screening (Croft et al., 1994; Heath et al., 1995).
A more targeted definition of “community” was used in the Physical Activity for Risk Reduction (PARR) Project, conducted with residents of rental communities administered by the housing authority in Birmingham, Alabama (Lewis et al., 1993). PARR targeted physical inactivity among African Americans of low socioeconomic status who were residents of public housing, and it was evaluated in eight communities randomly assigned to intervention through a staged design (n=6) and control (n=2). Baseline assessments confirmed the low levels of physical activity in the target population. Despite using community residents as interviewers, however, there were substantial problems in obtaining participation from randomly selected households, particularly in the initial survey. Pre- and post-intervention physical activity scores were not significantly different in the intervention and control communities.
In a move toward ensuring greater community input, the Kaiser Family Foundation's Community Health Promotion Grant Program (CHPGP) offered communities substantial flexibility in developing program targets that were responsive to local needs and priorities. This program was designed to foster community health promotion efforts targeting cardiovascular disease, cancer, substance abuse, adolescent pregnancy, and injuries (Tarlov et al., 1987; Wagner et al., 1991). Comparisons among 11 intervention and 11 control communities, however, indicated little evidence of positive changes in the outcomes selected by the intervention communities (Wagner et al., 1991) That project illustrates the challenges of interpreting results when the intervention is not standardized across communities; the lack of consistency in the results was due at least in part to differences in the interventions (Cheadle et al., 1995).
Conclusions. The ability to draw conclusions on the basis of these trials is limited by their designs and methods. Only a few included an adequate number of communities to provide sufficient statistical power. Most studies used random samples for project evaluation, but the response rates varied widely, and few studies had adequate response rates. Most studies used nonvalidated self-report of behaviors as outcome measures. Few studies reported the results of process tracking. The assignment of multiple communities is expensive, and ultimately might require multicenter collaborations, such as that used in the COMMIT (1991) study.
In the past 15 years, an increasing number of health promotion studies have been conducted in workplaces and worksites are now considered important channels for delivery of interventions to reduce chronic disease among adult populations (Abrams, 1991; Abrams et al., 1994; Fielding, 1984; Heimendinger et al., 1990; Tilley et al., 1999). The U.S. Department of Health and Human Services conducted two National Surveys of Worksite Health Promotion Activities, one in 1985 and another in 1992 (USDHHS, 1985, 1992). In 1985, 66% of private worksites with 50 or more employees offered health promotion activities. This increased to 81% by 1992 (McGinnis, 1993). Many worksite trials have targeted cancer and cardiovascular disease risk factors either as discrete trials (Byers et al., 1995; Emmons et al., 1999; Glasgow et al., 1995, 1997; Heirich et al., 1993; Jeffery et al., 1993, 1994; Salina et al., 1994; Sorensen et al., 1992, 1996, 1999; Tilley et al., 1999) or within the context of larger community-wide trials (Glasgow et al., 1996; Sorensen et al., 1993). Most of those studies used individual behaviors as the primary outcome. Intervention methods included strategies to incorporate employee input and a variety of activities based on tested behavior change theories. The reported interventions ranged from more intensive group behavioral counseling sessions of varying duration and number and supervised exercise prescriptions to less intense interventions with a wider reach, such as mailed self-help materials and newsletters. Several of the programs achieved statistically significant effects on smoking cessation (Jeffery et al., 1993; Salina et al., 1994; Sorensen et al., 1993, 1996). Jeffery and colleagues (1994) reported that where worksites changed from unrestrictive to restrictive tobacco control policies during the course of the intervention, there were significant reductions in smoking among employees. In the Working Well trial (Sorensen et al., 1996), no trialwide differences in smoking cessation were observed, but one of the four participating study centers reported significant effects for 6-month abstinence rates. That study center was unique in that it integrated an occupational health focus into the health promotion intervention, thereby targeting a key concern of workers in the participating worksites (Sorensen et al., 1995).
The Working Healthy Project (WHP), a multi-risk-factor study that was part of the Working Well trial, showed significant increases in self-reported exercise behavior in the intervention group as compared with controls (Emmons et al., 2000). Dishman and colleagues (1998) reviewed 26 studies of worksite interventions targeting physical activity, including those that did and did not use the worksite as the unit of analysis. The poor scientific quality of the studies precludes judgment about whether such interventions can increase physical activity, and the researchers concluded that there is a need for studies that use valid designs and methods.
Over the past two decades, extensive attention has been paid to health promotion and disease prevention among youth, particularly in schools. Schools provide an established setting in the community for reaching children and their families (Best, 1989; Perry et al., 1989; Stone and Perry, 1990; Stone et al., 1989). Several reviews summarize school-based smoking, physical activity, and nutrition education intervention trials from the 1980s and 1990s (Best, 1989; Contento et al., 1992; Flay et al., 1985; Stone et al., 1998). Some of those trials and analyses are reviewed here.
Reviews of youth smoking-control interventions generally conclude that social influence interventions can curb smoking onset (Best et al., 1988) although recent meta-analyses yielded a somewhat guarded picture of their efficacy. The first (Bruvold, 1993) found that effect sizes were largest for interventions that focus on social reinforcement, moderate for those with either a developmental orientation or a focus on increasing social norms, and small for interventions with a health information focus. A second meta-analysis (Rooney and Murray, 1996) reviewed 90 studies of school-based smoking prevention programs published in 1974–1991. They concluded that the influence of peer or social programs could be improved if they were delivered early in the transition from elementary to middle school (e.g., 6th grade), if same-age peer leaders were used, if they were part of a multicomponent health program, and if booster sessions were included in subsequent years. Although the average effects were small, with a reduction in smoking of as little as 5%, and only 20–30% under optimal conditions, school-based programs showed promise. The Life Skills Training (LST) program, a school-based intervention that teaches personal coping and social skills, has shown promising effects in both immediate and longer-term outcomes (Botvin et al., 1995). Dusenbury and Falco (1997) reported that the results of the 10 published studies of the LST program showed reductions up to 50% to 75% in tobacco, alcohol, and marijuana use at post-test, and a 6-year follow-up of over 4,000 participants indicated a 44% reduction in tobacco use.
Recognition of multilevel influences on smoking in youths has led to multifaceted interventions, including schoolwide media campaigns in combination with individual approaches. Such programs have been effective in reducing smoking prevalence throughout secondary school (Perry et al., 1992). A trial focusing on high-risk youths tested a combined program of mass media and standard school smoking prevention programs. This program was implemented in two schools; two other schools (the controls) had only the school program. At the 2-year followup, prevalence of smoking in the schools was compared; participants in the combined program showed a significantly lower prevalence of smoking than the controls (Flynn et al., 1997).
A recent school-based smoking prevention program (Peterson et al., 2000), The Hutchinson Smoking Prevention Project (HSPP), randomly assigned 40 school districts to experimental or control groups. Students were followed from grade 3 until 2 years after high school. An enhanced social-influence approach to the intervention was used, containing the 15 “essential elements” for school-based tobacco prevention developed by an NCI Advisory Panel (explained in Flay, 1985; Glynn, 1989). No significant differences between the control and experimental groups were evident at grade 12 or 2 years after high school suggesting that the intervention had little, if any, impact. The highly controlled, and well-designed nature of the study, including the high follow-up rates, high compliance with the intervention, the maintenance of the randomization by the school districts, well-matched control and treatment groups, and appropriate statistical analysis, strongly suggest that the failure to achieve change was a result of a failed intervention and not poor methodology. This conclusion implies that future interventions need to take a different approach, critically rethinking the interactions of biological, behavioral, and psychosocial risk factors at social and cultural contexts.
A review of the literature of school-based physical activity intervention research in 1980s and 1990s (Stone et al., 1998) found that the work was based on multiple theoretical approaches and incorporated simultaneous multicomponent interventions. In general, the studies found significant intervention effects for student knowledge and for psychosocial factors related to physical activity. Significant positive behavior changes were less common, but they were demonstrated in several studies (Dale et al., 1998; Homel et al., 1981; Killen et al., 1988; Leslie et al., 1998; Luepker et al., 1996; McKenzie et al., 1996; Sallis et al., 1999; Tell and Vellar, 1987). Two studies that conducted long-term follow-up found sustained significant differences up to 12 years after the intervention (Luepker et al., 1996; McKenzie et al., 1996; Tell and Vellar, 1987). The more extensive interventions typically had better results (Stone et al., 1998).
Most youth intervention programs to enhance physical activity have been conducted in school environments, typically through the physical education programs in elementary schools. The Child and Adolescent Trial of Cardiovascular Health (CATCH), a multicenter randomized trial for grades 3–5 involving 5,100 students in 96 schools, developed an intensive, teacher-based curriculum for enhancing health behaviors, including physical activity (Luepker et al., 1996). The program demonstrated significant differences in vigorous physical activity between experimental and control schools (Luepker et al., 1996); the differences were maintained three years after the intervention ended in the 5th grade (Stone et al., 1998).
Several school-based trials targeted dietary behaviors and found significant differences in knowledge, attitudes, and behavior change between intervention and control schools. Two exemplary programs are the Class of 1989 Study as part of the Minnesota Heart Health Program for 6th-12th graders (Kelder et al., 1994) and CATCH for 3rd–5th graders (Luepker et al., 1996; Perry et al., 1992). Both studies involved school-based interventions with large samples assessed for a long duration. Both interventions had beneficial effects on diet and eating habits (Nader et al., 1999); however, CATCH did not produce effects on physiological measures related to cardiovascular disease. In a review of interventions to promote healthy dietary behavior in children and adolescents, Perry et al. (1997) concluded that school-based nutrition education programs have been effective in improving aspects of children's eating behaviors, with positive effects also observed in physiological outcomes such as serum cholesterol.
Lessons from Behavioral Intervention Studies
There is clear evidence of efficacy of interventions to establish health-protective or health-enhancing behaviors, such as diet and physical activity; to reduce health-risk behaviors, such as smoking; and to facilitate adaptation to chronic illness, including cancer and heart disease. Yet the behavior changes frequently are difficult to maintain, which poses an important challenge to the field. The limited maintenance of behavioral change seen in initial intervention efforts may be due to the failure to take into account the contextual factors that allow relapse. Advances will require the practical application of new research on the role of contributing contextual factors that include intrapersonal, interpersonal, environmental, and temporal variables. A second challenge is the effective translation of trials to real-world settings. Generalization of effective interventions will require an expansion of the assessment of intervention outcomes delivered in diverse settings. Community-wide and organization-wide interventions have shown varied success. The findings are marred by poor designs and methods. In general, however, the interventions that were more broadly based and multifaceted were more likely to be effective. These challenges are not confined to advances in individual behavior change. As later chapters will reflect, similar challenges apply to all levels of interventions.
Varun is 40. His is a family of four— his mother, wife and son. Lately, his friends have been discussing health insurance, but Varun is not particularly interested. He thinks he is sufficiently covered by his corporate health policy, which provides a total coverage of Rs 2 lakh. He was allowed one dependent and chose his mother, given her age. Is a cover of Rs 2 lakh adequate for a family of four, including a senior, or should Varun revisit his health insurance needs?
Varun is clearly under-insured. His employer-sponsored cover has stagnated and has not been revised. Unfortunately, he doesn’t realise this, and continues to believe that he is well-insured on the health front. But what about coverage for his wife and son? As he grows older, a bigger family, more dependents, rising healthcare costs and longevity will necessitate a reconsideration of his family’s health insurance coverage. He must avoid the circumstances that compel him to meet sudden out-of-pocket expenditures during medical exigencies, and the unnecessary stress that comes with it.
The amount and kind of health cover he needs depends on various parameters, such as Varun’s health condition and his family’s medical history. Moreover, the environment he lives or works in must be considered to account for any health harazds. The prevailing cost of medical treatment in the city may also vary significantly. His cover should be much higher if he lives in a metro. More importantly, a sum assured that seemed sufficient six years ago may be inadequate now. Similarly, over the next few years, healthcare costs may climb higher. So, it is advisable to factor in inflation before deciding. Further, Varun’s preference in terms of hospitals and his income must be considered to decide on the right amount of coverage.
Hospitalisation play a pivotal role in ensuring the health of all family members, from newborns to the elderly. Therefore, it is extremely important that Varun reviews his health cover, at least every three years, or in the event of a change in his family’s situation. Upgrading health insurance in tandem with age, affordability, changing needs and medical inflation is absolutely critical.
Content courtesy: Centre for Investment Education and Learning (CIEL). Contributions by Girija Gadre, Arti Bhargava and Labdhi Mehta.