DataLink Core

The purpose of this core is to:
  • create a network of diabetes and obesity researchers across 11 health care systems; 
  • create a comprehensive, standardized diabetes registry across these health care systems; and
  • use the registry to conduct surveillance (identify and monitor trends in diabetes incidence and prevalence, in diabetes treatment patterns, and in the incidence and prevalence of diabetes complications).
The SUPREME-DM DataLink is developing a comprehensive, longitudinal clinical registry of over 750,000 insured patients with diabetes mellitus and a similar database of all members without diabetes from 11 integrated health care delivery systems. The registry will cover the period from 2005-2012, drawn from demographic and clinical data elements in electronic health records (EHRs) and other system databases, and calculate and add data on medication adherence. The DataLink Core will use the data infrastructure to identify and monitor trends in Type 2 diabetes incidence and prevalence, and in diabetes treatment patterns and outcomes.
The SUPREME-DM DataLink structure is based on the HMO Research Network’s (HMORN) Virtual Data Warehouse (VDW). All participating sites in the SUPREME-DM project have a VDW, which is a set of standard data tables that allow SAS programs written at one HMORN site to be run against other VDW sites with a minimum of site-specific customization*. It is ‘virtual’ in that it is federated or distributed, rather than consisting of a centrally located store of data.
Four working groups meet regularly to address various aspects of the surveillance aim:

Epidemiologic Methods – This group has developed methods for estimating prevalence and incidence of diabetes and its complications, including standardized methods for identifying numerators and denominators for conducting such analyses.
Pharmacotherapy Surveillance – The purpose of this group is to study anti-hyperglycemic and other drug therapies, including utilization, adherence, and safety and adverse events.
Diabetes Complications – This group has determined the most commonly used definitions for studying diabetes complications. They will examine hypoglycemia, neuropathy, and incidence and prevalence of these and other chronic complications by age stratifications.
Disparities – This group is working to determine the availability of race/ethnicity data  in the (VDW) at the SUPREME-DM sites. For both the overall population and the identified diabetes population at each site, they will examine selected enrollment, demographic, clinical, comorbidity, and utilization variables among those with and without race and ethnicity data to determine if  there may be less missing data among the diabetes population.

*Note that the process of running VDW programs and collating the results is not automated.  Programs are typically distributed via e-mail or by posting them to a secure website.  They must be manually downloaded, approved by the individual site for execution, run by personnel at the sites, and results are then returned manually.Thus, site personnel retain complete control over their local data.

Prevention Core

The purpose of the Prevention Core is to identify opportunities to conduct comparative effectiveness research (CER) in the area of type 2 diabetes prevention.  The CER Prevention project is being conducted to study a group of women who developed gestational diabetes mellitus (GDM) during a recent pregnancy to understand the effectiveness of various communication, counseling, and referral strategies in the six months after delivery in the prevention of type 2 diabetes. The study is designed as an observational CER cohort study among members of three health plans; Kaiser Permanente Southern California, Kaiser Permanente Colorado, and Group Health Cooperative.  The cohort study will describe the frequency of postpartum glucose testing, the communication of postpartum glucose test results, and the frequency of counseling and referrals related to weight management, physical activity, nutrition, and diabetes risk received by women with GDM during the first 6 months postpartum.  The desired outcome is to understand the relative impact of these follow-up approaches on the weight loss at 6 and 12 months postpartum and perception of future risk of developing type 2 diabetes, which in turn may result in the reduction of diabetes risk or the delay in the onset of type 2 diabetes. Data were collected from women in the cohort using a telephone survey, extracted directly from information in the electronic health record (EHR), and by applying natural language processing to information in the EHR.

Natural Language Processing (NLP) Core

The Natural Language Processing (NLP) study is embedded in the Prevention and Treatment studies: two comparative effectiveness research (CER) studies that are part of SUPREME-DM.  The Prevention study focuses on women with recent gestational diabetes mellitus (GDM) during a recent pregnancy to understand the effectiveness of various communication, counseling, and referral strategies on this population. The Treatment study is conducting a trial to evaluate the effectiveness of providing information from the diabetes registry on "early non-adherence" with newly-prescribed medications for patients with cardiovascular disease (CVD) risk factors.
The NLP Core is developing two software applications that will allow this core to electronically assess the use of behavioral counseling in weight loss for women with GDM (Prevention Study) and the use of behavioral counseling regarding medication adherence (Treatment Study) by extracting information from text notes in the electronic health records (EHRs). These software applications will be validated with manual review of the data.

The goal is to use these validated software applications in the conduct of CER studies in the prevention and treatment of diabetes. These NLP-derived measures will be examined as potential mediators or moderators of treatment effectiveness.

Patient-Centered and Self-Management Core

The purpose of this Core is to lead patient survey-related activites, help to identify key terms and phrases in progress notes that denote weight counseling for patients in the Prevention study and medication adherence counseling for patients in the Treatment study, and operationally define medication non-adherence. Survey-related activities include developing, selecting, pilot testing, refining, and translating into Spanish the survey instruments needed for both CER studies. This Core works closely with the DataLink Core, the Prevention Core, and the Treatment Core to measure medication adherence and conceptualize, measure, and extract patient-centered, patient-reported variables.

Treatment Core

The purpose of the Treatment Core is to conduct a study that evaluates the effectiveness of providing information on new prescriptions for uncontrolled risk factors to clinical pharmacists and case managers.

The study is designed for a multi-system, cluster-randomized comparative effectiveness research (CER) intervention using standardized registries to rapidly identify patients who have diabetes mellitus (DM) with uncontrolled cardiovascular disease (CVD) risk factors (A1c, LDL-c and/or SBP). These patients will have been newly prescribed anti-diabetic, antihypertensive, and lipid-lowering (cardio-metabolic) medications. This information will be provided along with tools and strategies to address early non-adherence to clinical outreach staff. This information will guide a proactive telephone contact with patients in the intervention strategy to encourage early and ongoing adherence to their new medication regimen. 

The intervention will assess the degree to which timely provision of new prescription data to care management staff for patients with uncontrolled CVD risk factors (A1c, LDL-c, and/or SBP) leads to improved CVD risk factor levels in diabetes patients with newly initiated cardiometabolic therapy. These data will be compared with “usual” population outreach practices, to assess increased adherence to the newly-initiated cardiometabolic therapies compared to usual population outreach practices.

The SUPREME-DM sites participating in this effort are: Lead site: HealthPartners Institute for Education and Research (HPIER), Minneapolis MN; and four implementation sites: Geisinger Health System, Danville, PA; Group Health Research Institute, Seattle, WA; Kaiser Permanente Northern California, Oakland, CA; and Marshfield Clinic Research Foundation, Marshfield WI. Sites participating in planning, analyses and training are Kaiser Permanente Northwest, Portland, OR; Kaiser Permanente Colorado, Denver, CO; and Michelle Heisler from the University of Michigan. Additional SUPREME-DM sites are involved in the NLP and patient survey activity and analyses of this study.

Desired outcomes are to increase understanding of the efficacy of strategies to address early non-adherence, mediators and moderators or effectivesness, and assess if these strategies lead to improved adherence and clinical outcomes (improved control of HbA1c, BP and LDL), as well as increased advice and education provided to the patient by clinical staff.