Evidence Report/Technology Assessment

Number 25

Prepared for:
Agency for Healthcare Research and Quality

U.S. Department of Health and Human Services
2101 East Jefferson Street
Rockville, MD 20852
http://www.ahrq.gov

Contract No. 290-97-0011

Prepared by:
Research Triangle Institute, Research Triangle Park, NC (RTI)-University of North Carolina at Chapel Hill, NC (UNC) Evidence-based Practice Center

Alice Ammerman, Dr. P.H., R.D.
Principal Investigator (unc)

Christine Lindquist, Ph.D.
Co-Principal Investigator (RTI)

James Hersey, Ph.D. (RTI)
Anne M. Jackman, M.S.W. (UNC)
Norma I. Gavin, Ph.D. (RTI)
Cristina Garces, B.A. (RTI)
Kathleen N. Lohr, Ph.D. (RTI)
Timothy S. Cary, M.D., M.P.H (UNC).
B. Lynn Whitener, Dr. PH, M.S.L.S. (UNC)

AHRQ Publication No. 01-E029

June 2001

On December 6, 1999, under Public Law 106-129, the Agency for Health Care Policy and Research (AHCPR) was reauthorized and renamed the Agency for Healthcare Research and Quality (AHRQ). The law authorizes AHRQ to continue its research on the cost, quality, and outcomes of health care, and expands its role to improve patient safety and address medical errors.

This report may be used, in whole or in part, as the basis for development of clinical practice guidelines and other quality enhancement tools, or a basis for reimbursement and coverage policies. AHRQ or U.S. Department of Health and Human Services endorsement of such derivative products may not be stated or implied.

The Agency for Healthcare Research and Quality (AHRQ), formerly the Agency for Health Care Policy and Research (AHCPR), through its Evidence-Based Practice Centers (EPCs), sponsors the development of evidence reports and technology assessments to assist public- and private-sector organizations in their efforts to improve the quality of health care in the United States. The reports and assessments provide organizations with comprehensive, science-based information on common, costly medical conditions and new health care technologies. The EPCs systematically review the relevant scientific literature on topics assigned to them by AHRQ and conduct additional analyses when appropriate prior to developing their reports and assessments.

To bring the broadest range of experts into the development of evidence reports and health technology assessments, AHRQ encourages the EPCs to form partnerships and enter into collaborations with other medical and research organizations. The EPCs work with these partner organizations to ensure that the evidence reports and technology assessments they produce will become building blocks for health care quality improvement projects throughout the Nation. The reports undergo peer review prior to their release.

AHRQ expects that the EPC evidence reports and technology assessments will inform individual health plans, providers, and purchasers as well as the health care system as a whole by providing important information to help improve health care quality.

We welcome written comments on this evidence report. They may be sent to: Director, Center for Practice and Technology Assessment, Agency for Healthcare Research and Quality, 6010 Executive Blvd., Suite 300, Rockville, MD 20852.

John M. Eisenberg, M.D.	Douglas B. Kamerow, M.D.
Director	Director, Center for Practice and
Agency for Healthcare Research and Quality	Technology Assessment
	Agency for Healthcare Research and Quality

The authors of this report are responsible for its content. Statements in the report should not be construed as endorsement by the Agency for Healthcare Research and Quality or the U.S. Department of Health and Human Services of a particular drug, device, test, treatment, or other clinical service.

Objectives: This authoritative, systematic review seeks to clarify the existing knowledge base on interventions to alter dietary behavior related to cancer risk and offers directions for future research. Specifically, the review addresses three key questions regarding the efficacy of behavioral interventions for promoting dietary change:

• Is there evidence that one such intervention, alone or in combination, is more effective than another in helping individuals or groups modify their diet to consume more fruits and vegetables and less fat?
• What is the evidence for the efficacy of dietary interventions by population subgroup, particularly by ethnicity and gender?
• What conclusions (if any) can be reached about the cost-effectiveness of these types of interventions?

Search Strategy: To identify articles publishing the results of behavioral dietary interventions, the reviewers performed literature searches in six databases (MEDLINE, EMBASE, PsycINFO, CINAHL, AGELINE, and AGRICOLA) using a variety of relevant terms.

Selection Criteria: The initial search strategy excluded studies on the basis of date and language of publication, location of the study, whether a free-living population was involved, size of the sample, and other factors. Only studies reporting intake of fruits and vegetables and dietary fat as primary dietary outcomes were included. The researchers reviewed 907 articles and retained 104 (representing 92 independent studies). These articles presented results for behavioral interventions conducted in a wide range of settings.

Data Collection and Analysis: For each article analyzed, a team of two abstractors compiled information regarding the study methodology and results, and the article's quality. Re-review and reconciliation was performed by the Scientific Director. After completion of data abstraction for the 104 eligible articles, summary Evidence Tables were generated that present key details and findings for all eligible articles. Using this data, three increasingly inclusive types of analyses were performed: a meta-analysis, a quantitative analysis of the difference in dietary change between intervention and control groups (a differences-in-deltas approach), and a semiquantitative analysis summarizing the statistical significance of the intervention effect. The report explores the overall efficacy of behavioral interventions on dietary outcomes and considers the influence of specific intervention and population characteristics.

Main Results: In the studies that were reviewed, dietary interventions were consistently associated with an increase in fruit and vegetable consumption (with greater increases noted for fruit intake). More than three-quarters of the studies reviewed reported statistically significant increases in fruit and vegetable intake (as a combined variable). Using the differences-in-deltas approach, the reviewers determined that the average increase in fruit and vegetable intake reported was 0.6 servings per day, and consistent decreases were noted in the intake of total fat and saturated fat. The mean change in total fat intake was estimated as a 7.3 percent reduction in the percentage of calories from fat.

Interventions appeared to be more successful at positively changing dietary behavior in populations at risk of (or diagnosed with) disease than in healthy populations. Several dietary intervention components appear to be particularly promising in modifying dietary change favorable to cancer risk reduction: social support, goal setting, small groups, food-related activities, and the incorporation of family components.

Conclusions: The dissimilarity across studies in terms of outcome measures, study design, analysis strategy, and intervention technique makes it difficult to draw broad conclusions about the efficacy or effectiveness of behavioral dietary interventions. For example, investigators tended to employ more intensive interventions with high-risk populations, making it difficult to discern whether population or intervention characteristics were primarily responsible for increased change in dietary behavior. However, some intervention components may hold promise for future research efforts.

Very few studies were appropriately designed or reported to allow interpretation of evidence for the efficacy of the interventions by population subgroup, particularly low-income or ethnic subgroups. Despite increased Federal agency funding for dietary intervention research in underserved and minority populations, a serious deficit still exists in good-quality, published research designed to determine the relative efficacy of different intervention approaches in these groups.

Furthermore, few studies followed participants for more than a year, and those that did often showed a falling off in the initial dietary behavior change achieved. Thus, more research is needed to determine the longer-term effectiveness of dietary interventions in both the general population and important subgroups and to evaluate programs specifically designed to prevent relapse over time.

Finally, no studies that met the review criteria provided data on the cost-effectiveness of dietary interventions. Comparing the cost-effectiveness of current and innovative intervention approaches will be critical to assessing their broader applicability.

This document is in the public domain and may be used and reprinted without permission, except those copyrighted materials noted, for which further reproduction is prohibited without the specific permission of copyright holders.

Ammerman A, Lindquist C, Hersey J, et al. Efficacy of interventions to modify dietary behavior related to cancer risk. Evidence Report/Technology Assessment No. 25 (Contract No. 290-97-0011 to the Research Triangle Institute-University of North Carolina at Chapel Hill Evidence-based Practice Center), AHRQ Publication No. 01-E029. Rockville (MD): Agency for Healthcare Research and Quality. February 2001.

Increasingly, behavioral research is being conducted on the role of dietary change in cancer risk reduction. To implement the findings of such research, researchers are testing numerous dietary interventions in a variety of populations. Despite these efforts, no clear understanding has emerged regarding which interventions are more efficacious in influencing dietary change and for which groups.

To clarify what is known about the efficacy and effectiveness of behavioral interventions in promoting dietary change, the National Cancer Institute (NCI) -- through the Agency for Healthcare Research and Quality (AHRQ) and its Evidence-based Practice Program -- commissioned the Research Triangle Institute-University of North Carolina at Chapel Hill Evidence-based Practice Center (RTI-UNC EPC) to develop a rigorous evidence report on these issues. To discharge this responsibility, we, as members of the RTI-UNC EPC staff, systematically reviewed and synthesized the findings of 104 articles (representing 92 studies) that report the impact of behavioral interventions on dietary outcomes considered to be relevant to cancer risk: dietary fat intake and consumption of fruits and vegetables. In addition to summarizing the existing knowledge base on behavioral interventions and dietary change, we identified limitations in the current literature and outlined directions for future research. As part of this effort, we empaneled an eight-person Technical Expert Advisory Group (TEAG) that provided advice and assistance throughout the project. In addition, the draft evidence report was subjected to an extensive round of external peer review by experts and clinicians in cancer and nutrition and by potential future users of the report.

In this evidence report, we addressed three key questions regarding the efficacy and effectiveness of behavioral dietary interventions for promoting dietary change:

• Is there evidence that one type of intervention or combination of interventions is more effective than another in helping individuals or groups modify their diets to consume more fruits and vegetables and less fat?
• What is the evidence for the efficacy of dietary interventions by population subgroup, particularly groups defined by ethnicity and sex?
• What conclusions (if any) can be reached about the cost-effectiveness of these types of interventions?

In addressing these questions through our review and secondary analysis of published articles, we considered interventions of all types (i.e., individual dietary counseling, group nutrition classes, social support groups) and in all settings (i.e., school, workplace, health care, environmental modifications, community). We included adults, adolescents, and children who were generally healthy-as well as those who were at elevated risk of or diagnosed with cancer, heart disease, or non-insulin-dependent (type 2) diabetes.

Our primary goal was to explore the overall efficacy and effectiveness of behavioral dietary interventions in increasing fruit and vegetable intake and decreasing dietary fat consumption. In addition, we considered the relative effectiveness of interventions in various population subgroups (such as children versus adults, high-risk versus healthy populations) and we explored the impact of key intervention characteristics (including the use of theory, intervention intensity, intervention setting, the use of social support, and the involvement of families in the intervention) on behavior change outcomes.

Our systematic review of the literature involved a mounting a comprehensive literature identification and screening process, abstracting relevant information from eligible articles, grading the quality of the articles, and generating summary Evidence Tables that present key details and findings for all eligible articles.

To identify studies regarding behavioral dietary interventions, we performed literature searches in six databases (MEDLINE, EMBASE, PsycINFO, CINAHL, AGELINE, and AGRICOLA). Primary Medical Subject Headings (MeSH) terms used in the searches included health behavior, attitude toward health, health promotion, behavior change, food habits, fat-restricted diet, diabetic diet, fruit, vegetables, prevent, counsel, cardiovascular disease, cancer, and neoplasms.

Our initial search strategy excluded studies (a) published before 1975 or in languages other than English; (b) conducted outside of North America, Europe, or Australia; (c) conducted with infants, institutionalized populations, or populations with insulin dependent diabetes mellitus; d) with sample sizes of fewer than 40 subjects at followup; and (e) in which dietary intake was externally controlled. Because the primary dietary outcomes we selected were intake of fruits and vegetables and dietary fat, we also excluded studies that did not report results for these outcomes. We included both randomized, controlled trials and nonrandomized studies that had control or comparison group designs. In all, we reviewed 907 articles and retained 104; collectively, these reported the results of 92 studies, which were used as the unit of analysis in our secondary analyses.

A team of trained abstractors completed Data Abstraction Forms, recording information about study methodology, the intervention design, and results. They used Quality Rating Forms to rate the methods, intervention content, and clarity of description for the 92 studies. The Study Director used the abstraction forms and original articles to generate summary Evidence Tables. The Scientific Director and a senior abstractor performed quality control functions; they reviewed information entered into the Evidence Tables and reconciled any discrepancies.

We conducted three sets of secondary analyses:

• a meta-analysis (for the outcome of total fat as a percentage of energy intake);
• a standardized, quantitative analysis of the differences in dietary changes between intervention and comparison or control groups from baseline to follow-up (which we hereafter refer to as the "differences-in-deltas" approach), conducted for the outcomes of combined fruit and vegetable intake, fruit intake, vegetable intake, total fat as a percentage of energy intake, total fat in grams, and saturated fat as a percentage of energy intake; and
• a semiquantitative analysis summarizing the statistical significance of the intervention effects (hereafter, the "summary of significant findings" approach), conducted for the outcomes of combined fruit and vegetable intake, fruit intake, vegetable intake, total fat intake, saturated fat intake, general fat intake scores, specific high-fat foods or cooking practices, and specific low-fat foods or cooking practices.

The primary goal of each of these analytic strategies was to determine the overall effectiveness of dietary interventions in changing dietary behavior. Secondary goals included a determination of the relative effectiveness of different types of interventions for changing dietary behavior in different population subgroups. Together, these analytic approaches represent a spectrum of selectivity, from use of a highly selective class of studies and statistically rigorous methods to assessment of a larger, broadly inclusive set of studies by a simpler analytic technique. Collectively, our unique analytic approach provided some internal validation of the findings, but also pointed to methodologic issues and gaps in the research base of interest to NCI, AHRQ, members of the TEAG, and the clinical and behavioral health community in general.

The currently available literature provided considerable evidence to address our first key question regarding the efficacy or effectiveness of different types or components of interventions in helping individuals or groups modify their dietary intake. These findings are presented below. Very few studies were appropriately designed or reported their findings in a way that permited us to interpret the evidence for the efficacy of interventions by subgroup, particularly low-income or ethnic subgroups. No studies that met our review criteria provided data on the cost-effectiveness of dietary interventions.

Approximately one-third (39/92) of the studies we reviewed reported results of behavioral dietary interventions on fruit and vegetable intake. Based on the small number of studies (14) reporting daily servings of fruits and vegetables as outcomes, and on the high degree of variability across these studies, we concluded that a formal meta-analysis was inappropriate. Therefore, we employed the remaining two analysis strategies in our determination of the impact of interventions on fruit and vegetable intake.

The results of both the differences-in-deltas approach -- based on 17 of 39 studies reporting daily fruit and vegetable intake -- and the summary of significant findings approach -- based on 36 of the 39 studies -- indicated that: (a) dietary interventions were positively associated with changes in fruit and vegetable intake; and (b) when fruit and vegetable intake were measured individually, changes in fruit intake were larger. Although the results from studies that reported significant findings for fruit and vegetable intake varied (depending on the particular outcome measured), the majority of the studies we reviewed reported statistically significant increases in fruit and vegetable intake (either as separate outcomes or combined). Statistically significant increases in fruit and vegetable intake (as a combined variable) were noted in 16 of 22 studies reporting this outcome. Using the differences-in-deltas approach, we determined that the median difference between intervention and control groups in the change in daily servings of fruits and vegetables was +16.6 percentage points. This translates into an average increase in fruit and vegetable intake of 0.6 servings per day.

We were unable to explore the relative effectiveness of interventions on many population subgroups because of the minimum cell size requirement we established for specific analyses. (We determined that a minimum of five studies per cell was necessary to conduct a particular analysis.) Nevertheless, our analyses suggested that interventions were more successful at increasing fruit intake among children and vegetable intake among adults. In addition, interventions conducted among higher disease-risk populations were consistently more likely to report statistically significant increases in fruit and vegetable intake than were studies in general populations. For example, all six of the studies conducted in high-risk populations reported significant intervention effects for fruit intake compared with only eight of 14 studies conducted among general-risk populations. We observed a slightly smaller difference (five of seven high-risk studies compared with five of 14 general population studies) for the outcome of vegetable intake.

Among the specific intervention characteristics we explored, several patterns were evident. Studies employing a theoretical basis were more likely to report statistically significant increases in fruits and vegetable intakes than studies that did not utilize theory (14 of 16 studies using theory reported a significant increase in fruit and vegetable intake, while three of six studies that did not indicate the use of theory reported a significant intervention effect for fruit and vegetable intake). In addition, we observed a linear relationship between study quality and the likelihood of reporting significant findings. Also, the use of social support components was associated with more favorable increases in fruit and vegetable intake (using both analytic strategies). For example, all five studies that included a social support component reported a statistically significant increase in fruit intake, compared with nine of 17 studies not using social support that reported a significant effect for fruit; the differences in the proportion of studies reporting a statistically significant effect for vegetable intake was smaller (three of five studies compared with seven of 17 studies). Finally, studies that employed goal setting and interactive activities involving food were more likely to report statistically significant increases in fruit and vegetable intake, although the magnitude of the increases was not notably higher than that in studies not employing such techniques. For example, among the three fruit and vegetable outcomes we explored, the differences in the proportion of studies reporting statistically significant intervention effects between studies that incorporated a goal setting component and studies that did not ranged from 8 percent to 47 percent. We did not have a large enough pool of articles to be able to explore characteristics such as intervention intensity, setting, mode of delivery, use of individual tailoring, or culturally or ethnically specific interventions.

Nearly 80 percent (80/104) of the articles we reviewed reported results for dietary fat, although these outcomes varied tremendously. In determining the impact of behavioral dietary interventions on decreases in fat intake, we used all three analysis strategies. Based on all three techniques, dietary interventions were positively associated with changes in fat consumption. We observed similar decreases in intake of total fat and saturated fat (the two most commonly reported fat outcomes in the studies we reviewed). The median difference between intervention and control groups in the change in total fat intake (as a percentage of total energy intake) was -15.7 percentage points. This translates into a 7.3 percent reduction in the percentage of calories from fat. Among a subset of articles employing biochemical indicators (e.g., measuring changes in blood cholesterol), the decrease in total fat intake was significantly correlated with concomitant decreases in total blood cholesterol (r = 0.763, p = 0.004). The change in saturated fat was not associated with statistically significant decreases in total blood cholesterol.

The large number of studies reporting results for dietary fat enabled us to explore two moderating population characteristics: age and disease risk status. Across the five sets of fat outcomes that we explored using the summary of significant findings approach, studies conducted in high-risk populations were not consistently more likely to report a statistically significant decrease in fat intake. However, the differences-in-deltas analysis indicated that the magnitude of the change in dietary fat, particularly the reduction in saturated fat, was notably higher among interventions conducted in higher disease-risk populations. For example, the median difference in outcome change between intervention and control groups for saturated fat intake among studies conducted with high-risk populations was -29.3 percentage points; among studies conducted with general-risk populations, the median difference-in-deltas was only -14.5 percentage points. Similarly, 16 of the 17 studies analyzed by this method that focused on high-risk populations reported a significant intervention effect for saturated fat intake, compared with only seven of 10 studies conducted in general-risk populations. The pattern of larger effect sizes being observed among the studies focusing on high disease-risk populations was also evident in our meta-analysis. In addition, interventions conducted among children appeared to be more successful at reducing intake of total fat and less successful at reducing intake of saturated fat than interventions conducted among adults, although a very small number of studies measured fat intake among children (six for total fat intake, five for saturated fat intake).

Unlike the pattern observed for fruit and vegetable outcomes, interventions employing a theoretical framework were not consistently more likely to report significant effects (and the magnitude of the intervention effect was actually lower among studies based on theory). Nor was study quality associated with either the likelihood of reporting significant effects or the magnitude of the intervention effect.

Among the specific intervention characteristics we explored, however, several consistent patterns were evident. The use of social support, small groups, and goal setting appeared particularly effective at reducing intake of dietary fat. Greater proportions of studies employing such strategies reported statistically significant findings, and the magnitude of the change in dietary fat (using the differences-in-deltas approach) appeared higher among these studies. For example, among studies so analyzed that incorporated social support components into the intervention, all seven reported a significant intervention effect for total fat intake, and the median difference-in-deltas was -26.7. Among the studies so analyzed that did not report the use of social support, 83 percent (35/42) reported a statistically significant intervention effect for total fat intake, and the median difference in change between intervention and control groups was only -10.4 percentage points. The differences in the impact of goal setting on total fat intake were less pronounced. Among the studies that used goal setting, 18 of 19 reported a significant intervention effect (with the median difference in change being -18.9 percentage points). Among the studies that did not report the use of goal setting, 80 percent (24/30) reported a significant intervention effect (with the median difference in change being -11.0 percentage points). Although studies that involved families in the interventions and used interactive food-related activities were more likely to report significant decreases in fat intake (for example, all 13 of the studies using a family component reported significant decreases in fat intake, but 80 percent -- 29 of 36 -- of the studies not using a family component reported a significant intervention effect for fat intake), the magnitude of the decrease was not higher than that among the studies that did not incorporate these special features. Finally, although very few studies were designed to be culturally or ethnically specific (to the study sample), our results suggest that such studies reported greater decreases in dietary fat (although we did not have a sufficient number of articles to explore the magnitude of this decrease). For example, all five of the interventions designed to be culturally or ethnically specific reported significant decreases in both total and saturated fat intake, compared with 84 percent (37/44) of the studies that were not designed to be culturally or ethnically specific.

Our evidence review lends support to the notion that a wide variety of dietary interventions delivered in many different settings to individuals of different ages, ethnicities, and genders can have a positive impact on dietary behaviors associated with cancer risk reduction. The large proportion of studies showing favorable outcomes in various situations suggests an overall positive effect, although the potential for publication bias may have influenced the likelihood of identifying positive effects of interventions. Similarly, the language limitations of the literature review present an additional source of potential bias.

The lack of similarity across studies in outcome measures, study design, analysis strategy, and intervention technique makes it difficult to draw broad conclusions about the most efficacious behavioral dietary interventions. Nevertheless, our findings offer insight into intervention components that may hold promise for future research efforts. Several dietary intervention components appear to be promising in modifying dietary change. These factors include social support, goal setting, small groups, food-related activities, and the incorporation of family components. Interventions that included "interactions with food," such as cooking or taste testing, seemed particularly promising in increasing fruit and vegetable intake and reducing fat intake.

To gain the most from intervention research, future studies should assess dietary intake at the individual level and should collect detailed process and psychosocial data to help identify determinants of dietary change. With the emergence of new technologies to enhance health communications, research is urgently needed to evaluate the efficacy of these interventions, either relative to or in combination with more traditional approaches to dietary change. Comparing the cost-effectiveness of these different intervention approaches will be critical to assessing their broader applicability.

Few studies we reviewed followed participants for more than a year, and those that did often showed a falling off in the initial dietary behavior change achieved. More research is needed to determine the longer-term effectiveness of dietary interventions and to evaluate programs specifically designed to encourage the maintenance of change and prevent relapse over time.

Priority funding by federal agencies for dietary intervention research among underserved and minority populations has increased the amount of activity in these areas recently. However, the deficit in good quality, published research designed to determine the relative efficacy of different intervention approaches in these high-risk, hard-to-reach populations remains serious.