A meta-analysis of trials evaluating patient education and counseling for three groups of preventive health behaviors.

OBJECTIVES To examine the overall effectiveness of patient education and counseling on preventive health behaviors and to examine the effects of various approaches for modifying specific types of behaviors. DATA SOURCES Computerized databases (Medline, Healthline, Dissertation Abstracts, and Psychological Abstracts), bibliographies 1971-1994. Search terms (patient education, patient compliance, and self care) with modifiers (evaluation and specific preventive behaviors). STUDY SELECTION Randomized and non-randomized controlled trials measuring behavior in clinical settings with patients without diagnosed disease. Abstracts and retrieved studies screened by multiple reviewers; 13% of retrieved studies met screening criteria. DATA EXTRACTION Replicated coding by multiple observers. DATA SYNTHESIS Behaviors were grouped based on whether the behavior is addictive and whether the desired change required subtraction of existing behaviors or adding new behaviors. The weighted average effect size from a random effects model for smoking/alcohol studies was 0.61 (CI = 0.45, 0.77), for nutrition/weight, 0.51 (CI = 0.20, 0.82) and for other behaviors, 0.56 (CI = 0.34, 0.77) indicating that the behavioral outcomes for these subgroups were significantly different from zero. Multiple regression models for the three groups indicated that using behavioral techniques, particularly self-monitoring, and using several communication channels, e.g., media plus personal communication, produces larger effects for the smoking/alcohol and nutrition/weight groups. CONCLUSIONS Patient education and counseling contribute to behavior change for primary prevention of disease. Some techniques are more effective than others in changing specific behaviors.