![[ Visit Client Website ]](images/banner.jpg)
Key Objectives:
This roundtable will include a brief summary of methods and results recently implemented by the Colorado Department of Public Health and Environment to estimate BMI at the census tract geography. After the brief summary, roundtable participants are encouraged to discuss the effectiveness and limitations of this and other small-area estimation techniques, progress towards implementing small-area estimation within their own agencies, and examples that connect the results of these substantial analyses to informing policy.
Brief Summary:
Policy to improve health and well being has shifted from a national and state focus to smaller areas such as zip codes and neighborhoods. The rationale is that there can be considerable variation in health outcomes and potential for more effective health intervention at these smaller geographies. However, collecting health outcome data at the sub-county level is often too costly for most local and state health agencies. Applying model based small-area estimates can be a solution for public health agencies to enhance existing public health surveillance and understanding of the distribution of population health outcomes. Based on similar methods explored by the CDC, the Colorado Department of Public Health and Environment has recently implemented methods using Colorado BRFSS and US Census data from 2011-2013 to estimate the prevalence of overweight and obesity at the census tract level in Colorado. Using a multilevel regression framework, we assessed the variability of BMI between counties and the variability of the relationship of individual-level socio-economic status to BMI between counties to develop census tract level estimates. In order to evaluate the accuracy of the model, estimates were compared with recently collected census tract level BMI data for the City and County of Denver. The goals of this project were to develop and employ a small-area estimate model based on established models and data, and evaluate how well the model predicts BMI. After evaluating the ability of this model to generate “accurate” small-area estimates, we hope to create a reusable framework for calculating additional small-area health outcomes across the state. Finally, simplified messaging and communication about the implications of using small-area estimates must be developed to connect the data with public health professionals in prevention programs who are often not proficient in these advanced methods.