Using Real-world Data to Assess the Burden of Diabetes in Children and Adolescents in Florida

点击翻译按钮获取中文摘要

ABSTRACT Strong evidence suggests an increasing trend of diabetes onset for both type 1 and type 2 diabetes (T1DM & T2DM) among children and adolescent in the United State (US) and globally since 2000. Significant racial- ethnic and geographic disparities have also been documented in diabetes incidence and in access to care and metabolic control in this vulnerable population. These alarming observations call for a timely surveillance system for the onset of diabetes among children and adolescents, so that tailored public health programs can be developed and health care resources can be efficiently allocated to address the needs of these vulnerable groups in a timely manner. However, diabetes surveillance studies on this particular population are limited in the US. The primary source for current prevalence and incidence estimates of diabetes among children and adolescents in the US for the CDC is from the SEARCH for Diabetes in Youth (SEARCH) study, which relies on an active case reporting system from enrolled clinical centers over ten states. The SEARCH study is important, however, does not include the majority of the population located at the south-east region, where diabetes prevalence was found to be the highest in the country among the adult population (i.e., the “diabetes belt”). Less is known if the prevalence and incidence of diabetes among children and adolescents in this area, including Florida, are also the highest in the country. In this proposed study, we will build a surveillance system in Florida to monitor the prevalence and incidence of diabetes among children and adolescents in an accurate, cost-effective, and timely fashion. As the 3rd-most populous state in the United States with a high level of racial-ethnical and geographic diversities, Florida will serve as an essential piece for the national diabetes surveillance system and provide important data to study the diabetes epidemiology among children and adolescents in the south-east region of the US where diabetes was found the most prevalent in the country. We will leverage our unique real-world data (RWD) source, the OneFlorida network, a large state-wide repository of RWD that contains linked EHRs, claims, vital statistics, and birth records data, covering more than 60% of Floridians across all 67 Florida counties to build this diabetes surveillance system. In particular, we will combine the structured and unstructured data (e.g., physician’s notes) from patients EHRs to develop novel data-driven computable phenotyping (CP) algorithms through modern machine learning methods, so that our estimations on the prevalence and incidence of T1DM and T2DM will be more accurate than prior EHRs-based systems. This study will also generate important data to study the social and behavioral determinants of health among many other potential contributing factors that results in ethnic-racial and geographic disparities of the diabetes, at both individual (e.g., individual health behaviors) and contextual levels (e.g., area-level access to metabolic control programs).