Review: Patient education and counseling increase preventive behaviors
ACP J Club. 1998 May-June;128:68. doi:10.7326/ACPJC-1998-128-3-068
Mullen PD, Simons-Morton DG, Ramírez G, et al. A meta-analysis of trials evaluating patient education and counseling for three groups of preventive health behaviors. Patient Educ Couns. 1997 Nov;32:157-73.
To determine, using meta-analysis, the effect of patient education and counseling on preventive behaviors in healthy patients and to determine which approaches were most effective.
Studies were identified by using MEDLINE, Healthline, Dissertation Abstracts, and Psychological Abstracts databases; scanning bibliographies of review papers; and contacting experts.
Published and unpublished studies were selected if they measured the effect of counseling or patient education on a preventive health behavior, studied healthy patients, used a control group, had ≥ 60% follow-up, had ≥ 15 participants in each group at the end of the testing period, reported sufficient data to calculate an effect size, and used intention-to-treat analysis. Studies were excluded if they used medication or included patients with special learning needs.
Data were extracted on intervention characteristics, target behavior addressed, and study characteristics.
74 studies (67 published from 1972 to 1994, 7 unpublished) met the inclusion criteria. Because of the presence of heterogeneity (P < 0.001), studies were analyzed by using the random effects model for 3 behavior groups: smoking (35 studies) and alcohol misuse (4 studies); nutrition (12 studies) and weight control (5 studies); and other behaviors, including contraceptive use (6 studies), breast self-examination (3 studies), exercise (1 study), stressor reduction (2 studies), and injury prevention (6 studies). The overall weighted average effect sizes were 0.61 (95% CI 0.45 to 0.77) for smoking and alcohol misuse, 0.51 (CI 0.20 to 0.82) for nutrition and weight control, and 0.56 (CI 0.34 to 0.77) for other preventive behaviors. Multiple regression analyses showed that for studies on smoking and alcohol misuse, larger effect sizes were associated with nonrandomized designs, a behavioral orientation, the use of media in addition to personal communication, and the use of self-monitoring. In studies on nutrition and weight control, larger effect sizes were associated with the use of several communication channels and supportive materials, ≥ 30-day follow-up, and self-monitoring. In studies on other behaviors, larger effect sizes were associated with low-risk status and < 30-day follow-up. Number of contacts and "essential principles of education" (reinforcement, individualization, relevance, and feedback) were not associated with larger effect sizes in the regression analyses.
Patient education and counseling are effective for increasing preventive behaviors in healthy people. For smoking and alcohol misuse and for nutrition and weight control, self-monitoring leads to more preventive behaviors.
Sources of funding: National Center for Health Services Research and Health Care and Technology Assessment; Henry J. Kaiser Family Foundation; U.S. Office of Disease Prevention and Health Promotion.
For article reprint: Dr. P.D. Mullen, Center for Health Promotion Research and Development, University of Texas School of Public Health, Houston, TX 77030, USA. FAX 713-500-9602.
Mullen and colleagues used meta-analysis to pool all of the available controlled trials of patient education or of counseling for prevention in healthy people. They split the trials into 3 groups: addictive problems that required subtracting a behavior (smoking or alcohol misuse), nonaddictive problems that required subtracting or substituting a behavior (nutrition or weight control), and nonaddictive problems that required adding a behavior (contraception use, breast self-examination, injury prevention, stress reduction, or exercise).
Although grouping diverse behaviors may seem like combining apples and oranges, it addresses 3 very important questions: What is the overall effect of patient counseling across behaviors? Do these effects vary by type of behavior? Are some counseling approaches more successful than others? Mullen and colleagues found that counseling was effective, although it is difficult to determine the effectiveness of counseling across behaviors. Further, counseling was effective for all 3 behavior groups. For specific techniques, the results varied among groups but not in any predictable way. Of note, self-monitoring was effective for the smoking and alcohol misuse group and for the nutrition and weight control group.
So, what can a busy clinician take home from this review? Behavior modification counseling helps patients adopt healthier lifestyles across many behaviors (1). To increase the effectiveness of counseling, patients should be treated by multiple providers who use various methods (1). Even brief counseling can be effective (1). This review suggests that patients who use self-monitoring and track their own behavior (e.g., with a food diary or an exercise log) also have a greater chance of success. One of the most important interventions, which was not addressed in this review, is ensuring that the system of care supports healthy behaviors by using such methods as health promotion posters, easy access to preventive care, and incentives for patients and providers (2).
Scott E. Sherman, MD, MPH
Veterans Affairs Center for the Study of Healthcare Provider BehaviorSepulveda, California, USA