Practice-level approaches for behavioral counseling and patient health behaviors.

BACKGROUND There is little empirical evidence to show that a practice-level approach that includes identifying patients in need of health behavior advice and linking them to counseling resources either in the practice or in the community results in improvements in patients' behaviors. This study examined whether patients in primary care practices that had practice-level approaches for physical activity and healthy-diet counseling were more likely to have healthier behaviors than patients in practices without practice-level approaches. METHODS A cross-sectional study of 54 primary care practices was conducted from July 2005 to January 2007. Practices were categorized into four groups depending on whether they had both identification tools (health risk assessment, registry) and linking strategies (within practice or to community resources); identification tools but no linking strategies; linking strategies but no identification tools; or neither identification tools nor linking strategies. RESULTS Controlling for patient and practice characteristics, practices that had both identification tools and linking strategies for physical activity counseling were 80% more likely (95% CI=1.25, 2.59) to have patients who reported exercising regularly compared to practices that lacked both. Also, practices that had either identification tools or linking strategies but not both were approximately 50% more likely to have patients who reported exercising regularly. The use of a greater number of practice-level approaches for physical activity counseling was associated with higher odds of patients' reporting exercising regularly (p for trend=0.0002). Use of identification tools and linking strategies for healthy-eating counseling was not associated with patients' reports of healthy diets. CONCLUSIONS This study suggests that practice-level approaches may enable primary care practices to help patients improve physical activity. However, these approaches may have different effects on different behaviors, and merit further research to determine if causal pathways exist and, if so, how they should be applied.