Improving multiple health risk behaviors in primary care: lessons from the Prescription for Health Common Measures, Better Outcomes (COMBO) study.

BACKGROUND Four health behaviors--smoking, risky drinking, physical inactivity, and unhealthy diets--contribute substantially to health care burden and are common among primary care patients. However, there is insufficient evidence to recommend broadly brief interventions to address all 4 of these in frontline primary care. This study took advantage of a multinetwork initiative to reflect on health behavior outcomes and the challenges of using a common set of measures to assess health behavior-change strategies for multiple health behaviors in routine primary care practice. METHODS Standardized, brief practical health behavior and quality of life measures used across 7 practice-based research networks (PBRNs) with independent primary care interventions in 54 primary care practices between August 2005 and December 2007 were analyzed. Mixed-effects longitudinal models assessed whether intervention patients improved diet, physical activity, smoking, alcohol consumption, and unhealthy days over time. Separate analyses were conducted for each intervention. RESULTS Of 4463 adults, 2199 had follow-up data, and all available data were used in longitudinal analyses. Adjusting for age, race/ethnicity, education, and baseline body mass index where available, diet scores improved significantly in 5 of 7 networks (P < .02). Physical activity improved significantly in 2 networks but declined in one network (P < .024). The likelihood of being a current smoker was reduced in 2 of 5 networks (P < .0001), and average alcoholic drinks per day was reduced in 2 networks (P < .02). Participants reported fewer unhealthy days at follow-up in 3 of 7 networks (P < .01). Details of implementation and the limitations in instrumentation help contextualize these modest outcomes. CONCLUSIONS Although some patients in these 7 PBRNs improved in several health behaviors and quality of life, the strength of evidence for field-ready methods to address multiple health risk behaviors remains elusive. The use of common measures to assess changes in 4 unhealthy behaviors was achieved practically in PBRNs testing diverse strategies to improve behaviors; however, variations in implementation, instrumentation performance, and some features of study design overwhelmed potential cross-PBRN comparisons. For common measures to be useful for comparisons across practices or PBRNs, greater standardization of study designs and careful attention to practicable implementation strategies are necessary.