Russ Waitman, PhD, University of Kansas Medical Center

1. Greater Plains Collaborative Reusable Observable Unified Study Environment (GROUSE): Integrating Medicare/Medicaid Claims for Cohort Characterization Russ Waitman, PhD, University of Kansas Medical Center Director of Medical Informatics, Associate Professor, Internal Medicine Greater Plains Collaborative Principal Investigator April 6, 2018 This work is supported in part by NIH grant UL1TR000001 and the PCORI PCORnet Greater Plains Collaborative CDRN

2. GPC Phase 1 Proposal • Our strategy for claims integration was buying the data from CMS and/or piloting Medicaid integration with some states.

3. Philosophy: GPC generally tertiary care sites for their regions…how biased are we…what do we not know? • GPC centers in eight states • less than 20% US population • We want to measure the gain by integrating claims for our 3 cohorts and serve as a basis for future projects that want to reuse this resource (with CMS approval). • Might have desired distributed approach but created a central de-identified resource due to • CMS position when we developed plan in 2015 • Need consistent data management plans at all sites • Timing to obligate money • Cost to manage distribution across all sites

4. Overarching Aims • To understand the development, treatment, and progression of breast cancer • To understand the development, treatment, and progression of amyotrophic lateral sclerosis (ALS) • To understand the development, treatment, progression, and consequences of healthy and unhealthy (overweight and obesity) weight • Serve as a greater national resource to understand the development, treatment, progression, and consequences of acute and chronic disease cared for within the United States healthcare system and in support of quality care for the conditions championed by the Patient Powered Research Networks in PCORnet as well as the other conditions studied by our peer CDRNs.

5. Overarching Analyses • In closing, we will create a de-identified resource that merges CMS claims and GPC site data to characterize the increase in data completeness and comprehensiveness provided through claims integration. We will be conducting three technical and methodological analyses in support of our cohorts: • Quantifying completeness of the health system-derived data repositories • Evaluating the distributions of health and care processes for the patients within the GPC versus the larger Medicare and Medicaid populations in our region to understand how studies of the GPC population generalize to the broader populations in our states. • Using CMS claims data to enhance quality control processes for aggregating health system-derived data and establishing correlations with CMS claims data for health system-derived data to support trial recruitment and observational studies, i.e. validating the use of EMR data for recruitment.

6. Files Requested to Support Aims • 3 years of data for Medicare (2011-2013) and one year Medicaid (2011) which was the freshest available when we started this process for phase 1. • Cost was ~$190,000 for our eight states • Subsequently have purchased 2014-2015 Medicare; in process of buying 2016 and 2012 Medicaid with CTSA funds as GROUSE is part of our CTSA Aims.

7. Data Environment and linkage through “simple” hashing Initially not worried about de-duplication management via GDIT; can address later if desired (note dropped back from PPRN goal)

8. Technical Status • Have 2011-2015 data cross-walked with GPC sites i2b2 data • Currently linking CDMs w/ MED_ADMIN table for Cancer RCR • Currently refreshing crosswalks this Spring for RCR • Aim to purchase 2016 Medicare and 2012 Medicaid this Spring/Summer; refresh crosswalks again in the Fall. • Have access to de-identified CMS files in Oracle format; giant i2b2 of claims, site i2b2/CDMs, and can cut a CDM of claims/EHR on demand (didn’t’ scale well with ~30 billion facts to make the whole CDM but revisiting hardware/software) • VPN access for remote users • Oracle DB and SQL • SAS • R/RStudioServer

9. Current Research • Currently focused on supporting the PCORI Cancer Rapid Cycle Research project: Molecular Testing and Targeted Therapies: Drs. Chrischilles, Carnahan, Schroeder – Iowa; Chapman - South Carolina; Waitman - Kansas • Aim 1: In a cohort of patients with an invasive single primary solid tumor, describe the use of common molecular tumor and, in some cases, germline biomarker testing and associated targeted cancer therapies. • Aim 3: Using the cohort from Aim 1 in sites with linked claims data, assess the completeness of the EHR-derived data for identifying cancer treatments, molecular tests and outcomes. • Dr. Mary Schroeder, Iowa: leads a GPC Breast Cancer Working Group Pilot(s): Oral anti-cancer medications, Drug-drug interactions: tamoxifen and selective serotonin reuptake inhibitors (SSRIs), Chemotherapy: delays and disparity, Cardiac toxicity and cardiovascular events, Upper Extremity Disability (UED) • Recently, 2 NIH CTSA KL2 Awardees will be using GROUSE: • Dr. Andrew Roberts, KUMC & Dr. Aaron Winn, MCW – breast cancer • Have GROUSE as Aim for our CTSA program; scheduling call w/ RESDAC • Planning additional characterization work for obesity with crossover to the NIH Intellectual and Development Disabilities community • Pending NIH R01 submissions so stay tuned • Encourage investigators develop proposals that “reuse” this unique environment.

10. Links to resources • RESDAC to get the data from CMS and costs • https://www.resdac.org/cms-data/request/research-identifiable-files • GPC links • http://gpcnetwork.org/?q=GROUSE - introduction page for investigators • https://informatics.gpcnetwork.org/trac/Project/wiki/GROUSE - technical summary page • https://informatics.gpcnetwork.org/trac/Project/attachment/wiki/GROUSE/ - link direct to attached files with Data Management Plan, Executive Summary, and IRB Protocol