Fish et al. Respond

Abstract

We appreciate Levine's response to our article.1 We agree that the reasons individuals are obese are complex and most likely have several different etiologies; for years, research has focused on both individual and neighborhood characteristics. While we understand the concern that reducing our understanding of the epidemic to well-characterized mechanisms may imply simplicity, we respectfully disagree that because of the complexity of the topic, research should not be undertaken nor attempt to delineate this complexity. In our article, we focused on how individuals perceive their neighborhoods, specifically if they perceive them as safe or unsafe. We made no claim that this perception deems a neighborhood as “bad” or “good”; rather, we stated that research should identify how these neighborhood characteristics are being interpreted by and affecting those people living in them. Further, by understanding the potential mechanisms through which neighborhood characteristics translate into individual decisions that affect health, we are more likely to develop policies to mitigate the choices that facilitate worse health. As the health care field moves toward patient-centered medical home models, health care organizations are beginning to recognize that any care provided must focus on the whole patient2 and will likely include a deeper understanding of family dynamics and values, socioeconomic demographics, and potential neighborhood influences (some of which were identified in our original article). Understanding neighborhoods and how residents perceive them may be critical to developing realistic and effective self-care and medical treatment plans. Asking individuals to lose weight to improve their health without understanding the environment in which they live will likely result in inadequate and unrealized care plans that are frustrating to patients and physicians. While physicians may not be able to alter the neighborhoods in which patients live, they certainly can find empathy with the individuals and their social context and develop creative plans to meet the individuals' goals. With respect to Levine's claim that the modeling employed is “futile,” we again respectfully disagree. Freedman asserts caution and does not argue that two-stage least squares (2SLS) models should not be used. He underscores that the “assumptions behind the model…[may] provide the leverage, [rather than] the data fed into the model”3(p711) and that “taking assumptions for granted is what makes statistical techniques into philosophers' stones.”4(p200) For 2SLS models, the main assumptions must be met for the model to be sound.5,6 We outlined each of these assumptions with our affirmations of their soundness and believe that this modeling technique was critical to understanding an important mediator between neighborhood characteristics and obesity. Finally, as Kawachi et al. state, “[b]y incorporating the methods and approaches from economics, geography, sociology (among other disciplines) into public health, research on neighborhood effects is poised to make a quantum leap in causal inference as well as usefulness for policy.”7(p4)