EVALUATING AGREEMENT BETWEEN PATIENT-REPORTED CLINICAL TRIAL DATA FROM ADOPT PGX AND MEDICARE AND MEDICAID CLAIMS
Author(s)
Jungjun Bae, MS1, Suk-Chan Jang, PharmD, PhD1, Rachel Myers, PhD2, Jason Jiang, Bachelor of Arts1, Marylu Holmes, MPH1, Larisa Cavallari, PharmD1, Elizabeth Harris, BS3, Julie Johnson, PharmD4, Josh Peterson, MD, MPH5, Todd Skaar, PhD6, Paul Richard Dexter, MD6, Simona Volpi, PhD7, Renee Rider, MS, JD, CGC7, Hrishikesh Chakraborty, DrPH8, Erica Elwood, MHA1, Haesuk Park, PhD9;
1University of Florida, Gainesville, FL, USA, 2Duke University, Duke Center for Applied Genomics & Precision Medicine, Durham, NC, USA, 3Duke University, Gainesville, FL, USA, 4The Ohio State University, Gainesville, FL, USA, 5Vanderbilt University Medical Center, Nashville, TN, USA, 6Indiana University School of Medicine, Indianapolis, IN, USA, 7National Human Genome Research Institute, Bethesda, MD, USA, 8Duke University, Durham, NC, USA, 9University of Florida College of Pharmacy, Gainesville, FL, USA
1University of Florida, Gainesville, FL, USA, 2Duke University, Duke Center for Applied Genomics & Precision Medicine, Durham, NC, USA, 3Duke University, Gainesville, FL, USA, 4The Ohio State University, Gainesville, FL, USA, 5Vanderbilt University Medical Center, Nashville, TN, USA, 6Indiana University School of Medicine, Indianapolis, IN, USA, 7National Human Genome Research Institute, Bethesda, MD, USA, 8Duke University, Durham, NC, USA, 9University of Florida College of Pharmacy, Gainesville, FL, USA
OBJECTIVES: A multicenter, randomized pragmatic trial, A Depression and Opioid Pragmatic Trial in Pharmacogenetics (ADOPT PGx; NCT05966129) evaluated the impact of CYP2D6-guided opioid prescribing. To assess the quality of patient-reported data from the trial and its potential integration with administrative claims, we examined agreement between selected patient-reported measures (prescriptions and healthcare utilization) and corresponding variables from Centers for Medicare and Medicaid Services (CMS) Medicare and Medicaid medical and pharmacy claims.
METHODS: Participants in the ADOPT PGx with ≥3 months of chronic pain who used or were considered for tramadol, hydrocodone, or codeine and were linked to CMS claims were included. Patient-reported trial data were compared with linked CMS claims for PGx non-concordance, all-cause emergency department (ED) visits, and hospitalizations. PGx non-concordance was defined using CYP2D6 activity scores for opioids, based on metabolizer phenotype and concomitant inhibitor use. Agreement between trial-reported and claims-based data was assessed using Cohen’s kappa (minimal acceptance threshold of κ≥0.61), and statistical differences were evaluated using McNemar’s test.
RESULTS: Among 398 participants (mean age 62 years; 73% female; 40% Black), 16% were Medicaid beneficiaries, 38% Medicare beneficiaries, and 46% were dual eligible. Agreement between trial-reported outcomes and CMS claims data was weak for PGx non-concordance (κ=0.57) and ED visits (κ=0.54), and minimal for hospitalizations (κ=0.36), all below the threshold for good agreement. Trial data reported a higher proportion of participants with ≥1 ED visit (31% vs. 25%) and hospitalizations (13% vs. 6%), whereas CMS claims captured a higher prevalence of PGx non-concordance (35% vs. 29%) (all p<0.01).
CONCLUSIONS: Agreement between patient-reported trial data and administrative claims ranged from minimal to weak. Trial data reported significantly more ED visits and hospitalizations, whereas CMS claims data captured more PGx non-concordance. These findings underscore the complementary role of administrative claims in improving outcome ascertainment in programmatic PGx research.
METHODS: Participants in the ADOPT PGx with ≥3 months of chronic pain who used or were considered for tramadol, hydrocodone, or codeine and were linked to CMS claims were included. Patient-reported trial data were compared with linked CMS claims for PGx non-concordance, all-cause emergency department (ED) visits, and hospitalizations. PGx non-concordance was defined using CYP2D6 activity scores for opioids, based on metabolizer phenotype and concomitant inhibitor use. Agreement between trial-reported and claims-based data was assessed using Cohen’s kappa (minimal acceptance threshold of κ≥0.61), and statistical differences were evaluated using McNemar’s test.
RESULTS: Among 398 participants (mean age 62 years; 73% female; 40% Black), 16% were Medicaid beneficiaries, 38% Medicare beneficiaries, and 46% were dual eligible. Agreement between trial-reported outcomes and CMS claims data was weak for PGx non-concordance (κ=0.57) and ED visits (κ=0.54), and minimal for hospitalizations (κ=0.36), all below the threshold for good agreement. Trial data reported a higher proportion of participants with ≥1 ED visit (31% vs. 25%) and hospitalizations (13% vs. 6%), whereas CMS claims captured a higher prevalence of PGx non-concordance (35% vs. 29%) (all p<0.01).
CONCLUSIONS: Agreement between patient-reported trial data and administrative claims ranged from minimal to weak. Trial data reported significantly more ED visits and hospitalizations, whereas CMS claims data captured more PGx non-concordance. These findings underscore the complementary role of administrative claims in improving outcome ascertainment in programmatic PGx research.
Conference/Value in Health Info
2026-05, ISPOR 2026, Philadelphia, PA, USA
Value in Health, Volume 29, Issue S6
Code
P61
Topic
Clinical Outcomes
Topic Subcategory
Clinical Outcomes Assessment
Disease
SDC: Systemic Disorders/Conditions (Anesthesia, Auto-Immune Disorders (n.e.c.), Hematological Disorders (non-oncologic), Pain)